An International Journal of Interdisciplinary Research

Size: px
Start display at page:

Download "An International Journal of Interdisciplinary Research"

Transcription

1 An International Journal of Interdisciplinary Research Vol , 2003 Understanding the Social II: Philosophy of Sociality P Edited by: Raimo Tuomela, Gerhard Preyer, and Georg Peter P Layout and Technical Conception: Georg Peter P Editorial Office: ProtoSociology, Stephan-Heise-Str. 56, Frankfurt am Main, Germany, Phone: (049) , preyer@em.uni-frankfurt.de P Homepage:

2 2 Autor 2003 ProtoSociology Frankfurt am Main Erste Auflage / first published 2003 ISSN Bibliografische Information Der Deutschen Bibliothek Die Deutsche Bibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind im Internet über dnb.ddb.de abrufbar. Alle Rechte vorbehalten. Das Werk einschließlich aller seiner Teile ist urheberrechtlich geschützt. Jede Verwertung außerhalb der engen Grenzen des Urheberrechtsgesetzes ist ohne Zustimmung der Zeitschirft und seines Herausgebers unzulässig und strafbar. Das gilt insbesondere für Vervielfältigungen, Übersetzungen, Mikroverfilmungen und die Einspeisung und Verarbeitung in elektronischen Systemen. Bibliographic information published by Die Deutsche Bibliothek Die Deutsche Bibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at All rights reserved. No part of this publication may be reproduced, stored in a retrievalsystem, or transmitted, in any form or by any means, without the prior permission of Proto- Sociology.

3 Titel 3 Proto Sociology An International Journal of Interdisciplinary Research Double Vol , 2003 Understanding the Social II Philosophy of Sociality Edited by Raimo Tuomela, Gerhard Preyer, and Georg Peter Contents I Membership and Collective Commitments Acting as a Group Member and Collective Commitment... 7 Raimo Tuomela and Maj Tuomela On Collective Identity Kay Mathiesen A Collective s Rational Trust in a Collective s Action Maj Tuomela Social Action in Large Groups Ulrich Baltzer II Collective Intentionality and Social Institutions What do We Mean by We? Stephen P. Turner Collective Intentionality, Complex Economic Behavior, and Valuation John B. Davis Institutions, Collective Goods and Moral Rights Seumas Miller The Micro-Macro Constitution of Power Cristiano Castelfranchi

4 4 Autor III Social Facts and Social Practices Foundations for a Social Ontology Amie L. Thomasson On the Objectivity of Social Facts Antti Saaristo Explaining Practices Petri Ylikoski Some Notes on Ontological Commitment and the Social Sciences Steven Miller IV On Margaret Gilbert s Proposal of Collective Beliefs Two Modes of Collective Belief Christopher McMahon What Really Divides Gilbert and the Rejectionists? K. Brad Wray Why Accept Collective Beliefs? Reply to Gilbert Anthonie Meijers Rejecting Rejectionism Deborah Perron Tollefsen On Social Epistemology and Terrorism in Time of Globalization Goldman s Knowledge in a Social World: Correspondence Truth and the Place of Justification in a Veritistic Social Epistemology Patrick Rysiew Is Terrorism Globalizng? Albert J. Bergesen Contributors Imprint On ProtoSociology Published Journals and Books ProtoSociology: Digital Journals Available Cooperations Announcements

5 Acting as a Group Member and Collective Commitment 5 I Membership and Collective Commitments

6 6 Raimo Tuomela and Maj Tuomela

7 Acting as a Group Member and Collective Commitment 7 Acting as a Group Member and Collective Commitment Raimo Tuomela and Maj Tuomela Abstract In this paper we will study two central social notions, acting as a group member and collective commitment. Our study of the first of these notions is as far as we know the first systematic work on the topic. Acting as a group member is a central notion that obviously must be understood when speaking of the we-perspective, group life, and of social life more generally. Thus, not only philosophy of sociality, philosophy of social science, political and moral philosophy but also the various social sciences need this notion and should benefit from our analyses and arguments. Collective commitment is the other we-perspective notion studied in our paper. We have argued for its importance as representing a kind of social glue needed for group members when thinking and acting as a group. In contrast to some other studies (e.g. Gilbert, 1989, 2000, Castelfranchi, 1995) our most elementary notion of collective commitment is not intrinsically normative but is only instrumentally normative and intention-relative. Thus our treatment covers more ground than the previous accounts do. I Introduction Sometimes groups act as groups, through their members actions. To give some examples, states and nations often enter contracts and act on those contracts. However, when a nation is split into opposing subgroups and the official government or leader of the nation (or state) lacks control over these subgroups, the situation becomes complicated. In this case outside groups (e.g. other nations) have difficulties in dealing with this disintegrating nation as a group. In another example, a group of peace demonstrators gets the permission to march in the streets but is later accused of demolishing property due to some members lack of commitment to the group s rules. These kinds of situations indicate that there is a need to take a closer look at what acting as a group member and what group members collectively committing themselves to something involve and presuppose. We will below elucidate these notions in several different ways and try to show for which situations the various analyses are adequate. The notions of acting as group member,

8 8 Raimo Tuomela and Maj Tuomela the we-mode, and collective commitment (in the sense of the group members persistently having bound themselves to something) are claimed to be crucial concepts for theorizing about social life. There is some previous work on collective commitment (see e.g. Castelfranchi, 1995, Cohen, P. et al., 1997, Gilbert, 1989, 2000, Tuomela, 2000a, 2002a), but there seem to be no serious attempts to analyze the notion of acting as a group member in detail. Acting as a group member intuitively contrasts with acting as a private person (at least with respect to the group concerned). We will argue that an adequate analysis and account of group life requires the use of the notion of acting (or, more broadly, functioning) as a group member. In groups collectively committed to preservation and furthering matters constitutive of their identity, this notion in its standard sense amounts to the same as the notion of we-mode functioning and acting emphasized by Raimo Tuomela (1984, 1995, 2002a, 2002b). Furthermore, collective identification (identification with one s group) amounts to functioning as a group member in its full, group-committal sense (viz. we-mode functioning). In this paper the following broad theses are crucial: (1) Central parts of the social world, including social groups (e.g. task groups, affiliation groups, organizations, states), social practices, and social institutions, conceptually require we-mode thinking and acting (and thus we-mode concepts). (2) We-mode thinking and acting require, and are (conceptually and factually) based on, thinking and acting as a group member. Indeed, we-mode thinking and acting can even be said to amount to thinking and acting fully as a group member. These theses have the obvious corollary thesis (3) Central parts of the social world, including social groups (e.g. task groups, affiliation groups, organizations, states), social practices, and social institutions, conceptually require acting as a group member as their foundation. In this paper we will directly defend only (2). The broad thesis (1) has been defended in the aforementioned works of the first author of this paper. While we do not really defend (1) in this paper much space will be devoted to a conceptual analysis of the notion of thinking and acting as a group member (from here on we will often omit the thinking as a group member

9 Acting as a Group Member and Collective Commitment 9 part, for the sake of simplicity). This analysis will result in several notions of acting as a group member, which all have a role to play in certain contexts. Our second important task in the present paper is to analyze in detail the notion of collective commitment, viz. the result of group members having collectively bound themselves to an idea, person, or action. In its weakest form collective commitment is only instrumental and not properly normative, viz. normative in a moral, legal, or social sense. We will distinguish between several aspects of collective (and also of private) commitment and show how these notions go together with the various notions of thinking and acting as a group member. As claimed, the notion of having an attitude and acting in the we-mode (or group mode) are central for theorizing about social (including societal) phenomena. The concept of we-mode in itself contains the notion of group (and expresses the group-perspective) and is in this sense a holistic notion while ontically it can be regarded as inter-relational, viz. in the sense of applying to interrelated group members. Acting as a group member is a central element of acting in the we-mode. In a way, the we-mode just amounts to thinking and acting as a group member in its fullest sense. For somewhat technical reasons we have, however, chosen to use as our conceptual order to start from (1) the notion of a strong we, incorporating group identity (characterized most centrally in terms of the group s ethos, viz. its constitutive goals, standards, and beliefs), and then to proceed to (2) functioning as a group member. The analysis of (3) we-mode thinking and acting is based on (1) and (2) as well as on collective acceptance and collective commitment. Of these, collective commitment is the central glue of social life. Collective acceptance will be understood in the sense elucidated in some earlier work (most notably Chapter 5 of Tuomela, 2002a). The basic idea is that collective acceptance amounts to coming to hold and holding a shared we-attitude by the participants in question. The we-attitude here is, broadly speaking, either a we-intention (conation) or a we-belief. Roughly, a person has a certain we-attitude ATT if he has the attitude ATT and believes that the others in the group also have or tend to have ATT and that this is mutual knowledge. The notions of we-attitude and collective acceptance can be either in the we-mode or in the I-mode (in spite of the perhaps misleading term we-attitude ). A we-mode we-attitude involves a thick notion of we (entailing We together intend to do X, We believe that p, with the group s commitment to these) while a weattitudein the I-mode only involves a thin notion of we (e.g. we as you and I ).

10 10 Raimo Tuomela and Maj Tuomela We will concentrate on social groups based on voluntary membership. Thus a member cannot in principle be coerced to stay in a group of this kind. We will also assume that there is internal freedom from coercion in the group. That is, the members decision making should not be coerced but must express their intentional agency and free will. Of course, in the context of group life and joint action the participants will have to partially give up their own authority to the group, but this should happen basically without internal coercion. We will also assume that the groups under discussion are free from external coercion in their decisionmaking and group acceptance. A group can be forbidden to exist, forbidden to act (e.g. due to its subversive activities), or it can be coerced to stop acting when engaging e.g. in illegal or asocial activities. Let us consider the main type of a social group under discussion in this paper. Such a social group is composed of members acting either in the standard sense or in the responsibility sense (cases (b) or (c) in Section IV below) when in a group context. Such a group will be assumed will be assumed to have the capacity to act as a group based on its members joint intention, in effect, to promote its ethos, viz. its constitutive goals, values, standards, beliefs, etc. (cf. Tuomela, 1995, Chapter 4, for social groups in this strong sense and Section II below for the ethos). It is thus assumed that a group can commit itself to action, and this requires that its members can collectively commit themselves together to promote (jointly or separately) the ethos. We will also require of these groups that the members will mutually believe or at least be disposed to mutually believe (either de dicto or de re) that they are members of the group be the group a permanent one or only a group of people acting together on a specific occasion (this will cover cases with members promoting the ethos in one of the senses (i)-(iv), to be discussed in Section III). Being a member can then be taken to presuppose disposition to acquire the mutual belief that the members are committed to the group members and indeed to the group. This entails commitment in the we-perspective (in principle with a group-representing thick we as the intensional subject of their group-related attitudes) to all or most of the ingredients of the ethos. (Violators of the constitutive features can then be tolerated to some extent, but wholesale violation is blocked in order to keep the group functional.) We will also be to some extent concerned with weaker kinds of groups when speaking of weak, I-mode acting as a group member (case (a) in Section IV). In these weaker cases it suffices that the members mutually believe or are at least disposed to mutually believe that they are members of the same group.

11 Acting as a Group Member and Collective Commitment 11 To give a quick survey of the topics treated in this paper we give here the titles of the sections to give initial guidance to the reader, although we emphasize that at least up to section VI the sections need to be read consecutively: II A Group s Realm of Concern, Intentional Horizon, and Ethos, III Acting As a Group Member in the Core Sense, IV Varieties of Acting As a Group Member, V The We-mode and the I-mode, VI Collective Commitment, Ethos, and Presupposed Collective Action, VII Why Require Collective Commitment?, VIII The Constitution of Commitment, IX Commitment and Acting as a Group Member, X Conclusion. II A Group s Realm of Concern, Intentional Horizon, and Ethos Let us consider the central notion of a group s realm of concern and relate it to the notions of intentional horizon and ethos. We assume that at least in principle all of these three items are up to the group to determine within a given environmental setting, and this entails that the group is an autonomous one. Accordingly, these items are taken to be based on group acceptance, thus group members collective acceptance in the sense that such and such things are respectively collectively taken to belong to the group s realm of concern, intentional horizon, and ethos. The realm of concern of a group will consist of a class of topics that the group has collectively accepted as its topics of concern, viz. as matters of interest to the group, which, furthermore, are considered in a group context. Accordingly, a topic in this sense can be regarded as a content, or an element of such a content, thought and spoken about in group contexts (as opposed to private ones). (Roughly speaking, a certain topic can be considered either in a group context, public for the group members, or privately, not open to group members.) Thus a broad range of items can in principle be involved topics can be objects, properties, states of affairs, and so on. Topics of concern then are contents and can be contents of attitudes. We will here concentrate on propositional contents only, and will thus deal with propositional attitudes. We take the intentional horizon of a group, g, to consist of the subset of topics in its realm of concern about the elements of which it de facto has some specific attitude about (except for the assumed collective acceptance). For instance, the attitude can be belief, intention, (having a) goal, wish, fear, etc. To take a simple example, our group might have as its topic of concern round (P) objects and their color, assuming for simplicity s

12 12 Raimo Tuomela and Maj Tuomela sake that red (Q) is the only color that can be involved. Our group might believe that there are only round and red objects (x) and non-round and nonred objects but that there are no mixed cases. The intentional horizon in the present linguistic setup consists of those state descriptions in the set {+-P(x) & +-Q(x)} that the group has attitudes about (viz. P(x) & Q(x) and -P(x) & -Q(x)). The intentional horizon is a subset of topics of the realm of concern of g (which allows all the four logically possible state descriptions). The ethos of group g in its strict sense is defined as the set of its constitutive goals, values, beliefs, standards, and norms, which give the group motivating reasons for action, to some degree at least. When there is no clear-cut distinction between the constitutive and non-constitutive goals, beliefs, standards, etc. we can take the ethos in a wider sense to consist of the central goals, beliefs, standards, etc., centrality basically defined in the partly circular terms about the group members collective acceptance of what is central. In our present setup the ethos forms a subset of the intentional horizon of the group. The ethos basically constitutes the identity of the group. To give a simple example, the group might have as its constitutive goal (ethos) to make their village (named a, possibly the only element in the domain and the central object type topic here) wealthy (P) and beautiful (Q). Then, using Goal to represent the goal-attitude, we have Goal(g, P(a) & Q(a)). In general, the ethos can be taken to have the world-to-mind direction of fit of satisfaction, viz. not only constitutive goals, intentions and standards but also constitutive beliefs have the kind of satisfaction condition requiring collective action to make them satisfied (e.g. goals) or to maintain or uphold them (e.g. beliefs). 1 1 Here is a more technical account of the present notions: When considering the notion of topic two dimensions or aspects of the problem must be distinguished. First, there is the question of whether a topic is something purely conceptual (e.g. linguistic) or whether it is something out there in rerum natura. The second aspect concerns whether a topic is something like 1) a collection of a) (real) objects or features (something extra-linguistic) or b) predicates (linguistic things) or whether it is 2) a) a state of affairs (something that obtains or can obtain or be the case) or a b) proposition or sentence which expresses a thought, says something about the world. Let us consider two monadic features or predicates P and Q. Here P and Q (e.g. round and red ) can be taken to represent topics (often in relation to the respective negations -P and -Q). P can be a sortal or noun predicate or it can be a predicate expressing an attribute of an object understood in an ordinary sense. We can here sometimes take the set {P, -P} to represent a topic or, in some other cases especially if P is a sortal predicate (say lion or house ) the topic could be simply P, viz. P-objects. Similarly Q, or sometimes, {Q, -Q} can be taken to represent another topic. This would be a linguistic construal of the notion of a topic representing an aspect or feature of some entity. The corresponding extra-linguistic construal would speak of the sets of objects that our predicates P and Q express (in various circumstances or pos-

13 Acting as a Group Member and Collective Commitment 13 In all, we assumed above an antecedently given realm of concern, analyzed either as a class of predicates or as the class of possible state descriptions formed in terms of them. The intentional horizon is the subclass of the realm of concern about the elements of which the group in actual fact has specific intentional attitudes. Finally, the ethos is the typically much more specific subset of those state descriptions (or, states of affairs extra-linguistically speaking) which are constitutive (or central) for the group and are capable of representing its reasons and (possibly external) motivating states. (Our present technical treatment may require enrichments in some contexts, but for our present purposes the above should suffice.) sible worlds ). Next consider the construal of a topic as a set of contents. In the linguistic construal we are basically dealing with the sets {P(x), -P(x)}, where x ranges over a certain domain of objects and P(x) and -P(x) are open sentences with x technically as a placeholder for names of the objects in that domain of individual objects (the negation -P of P is included as a simple conceptually implied feature). We of course operate similarly with Q. When combining our two predicates we operate with open state descriptions of the kind that the open sentences in the set {+-P(x) & +-Q(x)} represent. Thus the combined topic (combining the predicates P and Q) would correspond to dealing with all the combinations of round, non-round, red, and non-red, objects, putting the matter in extra-linguistic terms. (In addition or in the place of P and Q we might also have simply the class of objects represented by x to be a topic of concern.) We can also say that we deal with all those possible states of affairs (or perhaps elements in them) as the topic, or perhaps exclude some combinations a priori on some grounds. What is central, though, is that a topic in general does not take a specific stand concerning the world and say that the world is precisely such and such (viz. it does not pick out only one state or state description). In our simple case, we propose that the realm of concern of a group then can be regarded as the set of sets {{P, -P}, {Q, -Q}} (or in the simpler version {P, Q}, viz. as the set of the topics the group has) when viewed in the group context. Alternatively, which from our standpoint does not make much difference, it can be regarded as the set {+-P(x) & +- Q(x)} of state descriptions or as the states that they are satisfied by. In a more elaborate treatment one may consider defining the topic of concern of a group in terms of all the Boolean combinations of the predicates (here P and Q) in question. The predicates in question, here P and Q (or in the extra-linguistic treatment, the properties, types of sets of extensions that they express) can concern states, events, or actions (and possibly other ontological kinds). In our later discussion we will largely concentrate on action predicates. Suppose now {+-P(x) & +-Q(x)} represents the antecedently assumed, collectively accepted realm of concern of a group g. Then we say that the intentional horizon of g consists of the subset in its realm of concern about the elements of which it de facto has, some attitude about. For instance, the attitude can be belief, intention, (having a) goal, wish, fear, etc. Our group might believe that there are only round and red objects and non-round and non-red objects. As the intentional horizon in the present setup consists of those state descriptions in {+-P(x) & +-Q(x)} that the group has attitudes about, it is a subset of the realm of concern of g.

14 14 Raimo Tuomela and Maj Tuomela Collective acceptance is basically taken to amount to coming to hold and holding a relevant we-attitude (in this paper most often in the we-mode). The we-attitude can be in the intention family (including also e.g. promises) or in the belief family (actually belief understood as acceptance and not experiential belief); or somewhat more generally, the we-attitude can have the world-to-mind or the mind-to-world direction of fit of satisfaction (see Tuomela, 2000a,b, and especially 2002a, for elaboration). As noted, in the case of a collective accepted ethos the we-attitude will have the world-tomind direction of fit of satisfaction. III Acting as a Group Member in the Core Sense In any group, the members may perform freely chosen actions qua a group member under some provisions. These actions or, more broadly, activities, including mental ones must belong to the topics of concern of the group in group contexts, as opposed to private contexts, and they must be collectively accepted (or at least collectively acceptable) as actions correct for a group member to perform, or they must at least be actions that the group is responsible for (see our analyses below). As indicated, such actions must be (rationally) collectively accepted by (or acceptable to) the group. This acceptance can be either normative, group-binding group acceptance or we-acceptance by the group members or their majority (cf. Tuomela, 1995, Chapter 7, and 2002a, Chapter 5). The acceptance (or the resulting acceptance belief ) here could be the following kind of we-acceptance within the group, g: Ideally, everyone accepts a topic, T, to be a topic of concern for g, and believes that everyone in g so accepts and also believes that this is mutually believed in g. In general, when speaking of collective acceptance we will assume that the participants collectively commit themselves to what they collectively accept. For clarity, we often speak of shared we-mode we-acceptance, in contrast to I-mode we-acceptance (cf. Section V). In the case where actions are under consideration, a topic T i within the action realm of g s concern consists of content-satisfying or content-maintaining actions or activities X i considered in a group context. Let us call their set X i. The class of such sets will be g s action realm of concern. We will below concentrate on X i and the action case. The ethos serves to constitute the group by giving its content, and it can also be said to specify and make possible actions that qualify as actions as a group member. As a consequence it becomes true that an action is action

15 Acting as a Group Member and Collective Commitment 15 as a group member if and only if it is collectively accepted by (or is collectively acceptable to) the group members as an action that promotes, or is at least weakly conducive to, the ethos of the group. With some qualifications for making room for dissidents, this is what our analyses below will elucidate. We will assume as a simplification here that the realm of concern of the group is given and it conceptually precedes the ethos of the group. This (easily liftable) assumption makes it easier to account for the group s changing its ethos and identity and for comparing different (and perhaps competing) groups concerned with the same kinds of topics. Furthermore, the notion of acting in the realm of concern is required for elucidating acting as a group member (cf. especially the responsibility sense (5) below). Our ultimate conceptual explicans here is (rational) collective acceptance (typically in the wemode sense). The realm of concern, the intentional horizon of the group, the ethos, and acting as a group member all ultimately depend on collective acceptance. The general case to be considered is that of a normatively structured group with positions (the unstructured case can be regarded as its special case with no specific positions over and above group membership). We will first classify the types of actions within the action realm of a (structured) group s concern. These actions will all be collectively acceptable and mutually knowable as being at least weakly ethos-promoting (as will be clarified). Here are the four action classes (cf. Tuomela, 2002a,b): (1) positional actions (related to a group position or role), which include (i) actions (tasks) that the position holder in question ought to perform, perhaps in a special way, in certain circumstances and (ii) actions that he may (is permitted to) perform in some circumstances; 2 2 The present class of actions and especially the notion of the right circumstances is discussed in Tuomela (1995), Chapter 5, in detail. There the phrase the right social and normative circumstances for positional action was analyzed in detail. While space does not here permit a fuller discussion, let us draw on an example used in that context. Consider Finland s President Ryti s signing the so-called Ryti-Ribbentrop Pact with Germany in However, Ryti acted in part as a private person and did not satisfy all the position rules concerning a president. He seemingly made a pact with Germany acting as a representative of Finland. While he seems to have made the pact with the approval of the Government this pact was (intentionally) never taken to the Parliament to be ratified by it, and so all the position rules were not satisfied. The Finns did not consider this to be a pact between the two countries while the Germans did. Here we have a real life example of the difficulties involved. We wish to note that in principle also role behavior related to positions can be included in class (1). What were called position-related role actions in Tuomela and Bonnevier-

16 16 Raimo Tuomela and Maj Tuomela (2) actions which constitutive and other group norms and group standards (eg. norms and standards which are not position-specific) require or allow; (3) actions and joint actions that are based on situational intention formation (e.g. agreement making) that has not been codified in the task-right system of g or the group norms of g but which still are consistent with actions in (1) and (2) and are ethos-promoting; (4) freely chosen actions or activities (and possibly joint actions), which are not incompatible with actions in classes (1)-(3); these freely chosen actions belong to the realm of concern of g and are rationally or reasonably collectively accepted by, or acceptable to, the members of g as ethos-promoting actions. Acting as a group member in the core sense is to act intentionally within the group s realm of concern, promoting or at least respecting the ethos, viz. the central, constitutive goals, values, standards, beliefs, norms, elements of the history of the group. This core sense will below be divided into wemode acting and I-mode acting as a group member (see Section V for these notions). The former will be called the standard and the latter the weak sense of acting as a group member. Both are needed for accounting for group life. The basic situation is this: If there is collective acceptance of the ethos, it will be upheld and maintained by the participants by means of actions falling within (1)-(4) (recall the world-to-mind direction of fit). In this sense the ethos is taken by the members to be the group s ethos and to be constitutive of the group. If they jointly intend to maintain and satisfy the ethos, we get collective commitment to it. If they only separately intend to maintain and satisfy it, we get separate intentions and accordingly private commitment to the ethos. The elements of the ethos are constitutive of the identity of the group, or at least of its group-internal or groupjective identity. The group s groupjective identity may also reflect objective features that are not up to the group to determine, e.g. the special nature of the members. These features may still be taken into account in the ethos in terms of the group s mutual belief. Tuomela, 1992, and in Tuomela, 1995, Chapter 8, thus become included here. (For instance, giving public lectures might thus be such extra-duty actions related to a professor s position.) Furthermore, it can be noted that class (4), to be defined below, in addition can be taken to include role actions also investigated in our work (cf. a mother-type s role actions).

17 Acting as a Group Member and Collective Commitment 17 In classes (2), (3), and (4) the actions may also be based on the group s nonconstitutive goals (etc.), thus goals that are not central to the group s identity. Still the actions will at least in an indirect sense be taken to be ethos-promoting. Actions in class (1) are typical positional actions, and subclass (ii) of (1) consists of actions that the holder of a position may choose from. (The taskright system specifying (i) and (ii) may contain rule norms and/or proper social norms in the sense of Tuomela and Bonnevier-Tuomela, 1992, and in improved form in Tuomela, 1995, Chapter 1.) However, actions in classes (2)- (4) can occur also in the positional case. Note that in the case of unstructured groups, class (1) can be empty, when a strict notion of normative position is meant. Action as a group member can be either successful or unsuccessful. What is required is that the group member will intentionally attempt to act in a way that he takes to be within the group s realm of concern without violating the group s ethos. The ethos is at least an underlying or a presupposition reason for the member s action. It need not, however, be a salient motivational reason as especially classes (3) and (4) indicate (cf. also criteria (iii) and (iv) below). The notion of ethos can be understood in a wide sense in which every group can be taken to have an ethos, which may consist just of some basic shared ends or beliefs or something closely analogous. As said, full success in action will not be required. There may thus be failures due to false beliefs about the group s norms and standards, due to lack of skill, or due to environmental obstacles. Acting as a group member in the core sense is equivalent to acting intentionally in an ethos-promoting sense in terms of performing actions falling into one of the classes (1)-(4) or at least attempting so to act. (One cannot non-intentionally act as a group member, although one can non-intentionally conform to the ethos.) The actions in classes (1)-(4) need not (especially in case (4)) be strictly contained in, or specified by, the ethos; it suffices for our present purposes that they at least weakly promote the maintenance and satisfaction of the ethos and are consistent with it. That they promote the ethos is at least a partial motivational reason for these intentionally performed actions. By our wide ethos-promoting (or ethos-respecting ) action concerned either with maintaining the ethos or with the satisfaction of its content we will mean an action satisfying one of the following criteria (i)-(iv): (i) An agent intentionally promotes the ethos by acting with the aim and correct belief that his action will promote the ethos. The agent s motivational reason for performing the action here can be taken to be the mentioned aim in conjunction with the belief.

18 18 Raimo Tuomela and Maj Tuomela Thus, if the ethos involves protection of the environment, the agent may throw the wrap paper of his lunch into a wastebasket with the aim and belief that the ethos thereby gets promoted in the satisfaction sense. On the other hand, the maintenance of the ethos is illustrated by a case in which the agent defends the ethos (possibly under the agent s description of it) in a debate. (ii) The agent intentionally performs a certain action in part for the motivational reason that he (correctly) believes it is what the ethos requires or allows or is probabilified by, but he may lack the aim to truly promote this ethos. This agent is a rule-follower who does not really care about the content of the ethos. For instance, the agent throws the wrap paper of his lunch into a wastebasket believing this is required by the ethos, but not specifically aiming that the ethos thereby gets promoted. The ethos required his action, and obeying it was a part of his reason, but he did not specifically aim at promoting the ethos. The maintenance of the ethos can be promoted by defending it for the reason that he ought to defend it, as it is the ethos he is committed to. (iii) The agent performs a certain action in the wide intentional sense that he does not perform it non-intentionally and performs it in part for the presupposition reason that he at least tacitly correctly believes that the ethos requires or allows or is probabilified by it; but he did not consciously aim at the promotion of the ethos; see Tuomela, 2002a, Chapter 4, for presupposition reasons and routines. The agent might routinely throw the waste paper into the basket or refrain from violating the traffic rules. The agent may in the present case promote the maintenance of the ethos by defending the ethos in part for his tacit (non-salient) reason that he ought to defend it. (iv) The agent conforms to a rule or a standard of action by intentionally acting correctly (e.g. I stop at red signal because it is correct action instead of I stop at red signal because this is what the rule requires ). He acts in part for the reason that it is the right thing to do but not for the reason that his action is required by the ethos. Still, the non-accidentally performed action a kind of pattern-governed action accidentally is (directly or indirectly) required or allowed by, or (probabilistically) conducive to, the ethos of the group. For instance, a child or a similar novice may obey the traffic rule of not crossing a street on red light as a kind of intentional routine or patterngoverned action. (See Tuomela, 2002a, for this kind of intentional patterngoverned action as distinct from non-intentional Sellarsian pattern-gov-

19 Acting as a Group Member and Collective Commitment 19 erned action.) Also case (iii) may be regarded as a strong kind of patterngoverned action. Here the agent on purpose conforms to the traffic rule due to having been trained to do so, but he does not do it for the reason that the traffic rules (assumed to be entailed by or to specify the ethos) require him to stop. He stops in part because he has been conditioned to stop on red light. However, he still takes his action to be the right thing to do. As to the promotion of the maintenance of the ethos, our agent may defend it weakly by just claiming that refraining from crossing on red light is the right thing to do. All our senses (i) - (iv) amount to more than mere accidental compatibility with the ethos. Indeed, if the group members intentionally respect the ethos by their actions in the group context in any of these senses and indeed are correct in their relevant beliefs (concerning the external success conditions of the ethos), then the ethos will be objectively promoted at least to some extent. However, ethos-related goals and intentions need not become satisfied (in cases (iii)-(iv)); and the beliefs and standards involved in the ethos, while they will not be contradicted, will not be actively professed and used as premises in practical and theoretical reasoning. Note that when ethos promotion takes place in the we-mode and concerns (iii) or (iv) the collective commitment to promote the ethos is either presuppositional and thus weak (case (iii)) or opaque and weak in this sense (case (iv)). (A weaker possibility than requiring that each and every individual action as a group member satisfy at least one of the criteria (i)-(iv) would be the more collective requirement that the group promote its ethos through its members actions which on the whole tend to satisfy at least one of (i)-(iv), but we will not in this paper adopt this suggestion.) One may also speak of functioning as a group member, where functioning is broader than acting. Functioning here also includes having propositional attitudes as a group member. We have in effect already encountered this notion when assuming that the group members must have collectively accepted its realm of concern, its intentional horizon, and its ethos, because collective acceptance means coming to hold and holding a relevant we-attitude. While our earlier treatment of the notion of the realm of concern was conducted in terms of contents in a broad sense, in our above characterization of acting as a group member we concentrated on classes of actions. Given the earlier notion of content, we can deal with (propositional) attitudes similarly. Thus, we may speak of attitude contents and actions within the realm of concern of group g, and in the above classification we may include also attitudes in addition to actions whenever needed. The important

20 20 Raimo Tuomela and Maj Tuomela thing to notice here is that those attitudes (say beliefs or hopes) are in the group context (in contrast to a private one) and that they are based on acceptance and thus on something that one can acquire by means of one s intentional action in group contexts (cf. Tuomela, 1992, 2000a). Let us note that the notion of acting as a group member in the case of class (4) occurs within the scope of collective acceptance, viz. a we-attitude of a suitable kind. The circularity involved here is not vicious, as shown elsewhere (see Tuomela, 2002a, Chapters 5 and 6, for discussion and Balzer and Tuomela, 1997, for a logical way to handle the apparent circularity). Before proceeding to our finer analysis of acting and functioning as a group member, some remarks on our central notions of collective acceptance and commitment will have to be made. As claimed, collective acceptance can be regarded as coming to hold and holding a relevant we-attitude, viz. the members collectively accept that p if and only if they accept that p, believe that the others in the group accept that p and also that this is mutually believed in the group. In the case of the ethos it must always have the world-tolanguage direction of fit of satisfaction. (However, e.g. the realm of concern can be regarded as based on a we-belief with the mind-to-world direction of fit.) Collective acceptance involves either collective or private commitment (to the accepted item), because collective acceptance entails the intention to have the accepted content (e.g. goal) as a group content, and this intention in the usual sense involves a commitment, which in turn is entailed by collective acceptance in the ethos case. Said in our more technical terminology, collective acceptance can be in the we-mode or in the I-mode. I-mode collective acceptance amounts to an aggregate of private acceptances, and here a content (e.g. a goal) is not accepted as a group content but, contingently, everybody accepts the content as his content, but possibly for the use of the group. On the level of the group, so to speak, functioning as a group member (viz. functioning as a group member in a collectively committed way for the group) is generally, with some exceptions, we-mode functioning, based on we-mode collective acceptance (see esp. Section VII for arguments). However, there can be also members functioning in an ethos-respecting way but in the I-mode, although all members cannot all the time function only in the I-mode. Accordingly, the kind of collective acceptance that our classes (1) (4), serving to define the core notion of functioning as a group member, are based on, can in principle be either we-mode or I-mode acceptance. The wemode case will basically characterize our standard, we-mode sense and the I- mode case the weak sense of functioning as a group member. When func-

21 Acting as a Group Member and Collective Commitment 21 tioning as a group member is in the I-mode it is based on the agent s private commitment and his private reasons (e.g. I do it because of my private, although possibly group-regarding reasons ) in contrast to the we-mode case that is based on collective commitment and group reasons (e.g. On the basis of our collective commitment, I perform my action at least in part because that is the group norm, or that is what the group requires of me, or that is the right thing to do in our group, etc. ). We will below strongly emphasize the importance of we-mode collective acceptance for groups which can think and act as groups and in fact we require we-mode acceptance in general (at least in the case of a substantial part of the authorized operative members who represent the group) and allow I-mode acceptance only in the case of a small part of the operative members, except in the case of weakly acting as group member which by definition is an I-mode notion (case (a) in Section IV). To recapitulate and expand, (collective) commitment to the ethos is conceptually based on the group members (collective) acceptance and (joint) intention to uphold the collectively accepted ethos, under their own description of the situation. Basically, successful collective acceptance is the group members social acceptance state, which can be analyzed as a we-attitude or, sometimes, as a set of we-attitudes (see Tuomela, 2002a, Chapter 5). We claim that, ontologically, there is one state that can be described as an acceptance state or as a we-attitude. In the ethos case there must be a we-attitude with the world-to-mind direction of fit of satisfaction, viz. a we-attitude in the intention family, because the ethos is meant to be upheld by the members actions. This we-intention involves a commitment to the ethos in the sense of commitment that any intention involves commitment to its content. In principle the above notions of collective acceptance, we-attitude, and we-intention can all be either in the we-mode or in the I-mode or private mode. Assuming that the members indeed have adopted the we-perspective, the shared we-intention here is a joint intention (viz. shared we-mode weintention) rather than an aggregate of private intentions. The we-perspective conceptually requires that the group members be disposed to think of themselves as parties of a we that intends, believes, and acts for the use of the group, and thus the members must be disposed to collectively commit themselves to the contents in question. Under favorable conditions, this becomes manifest as their thinking and acting as a group. Note that in the ethos case the collective commitment is based on their collective attitude (with the world-to-mind direction of fit) to uphold and satisfy the ethos and this amounts to a (we-mode) joint intention. There are also functional argu-

22 22 Raimo Tuomela and Maj Tuomela ments to the effect that full joint intention rather than a conglomeration of private intentions, however dependent, are to be required for full-blown acting as a group member. It need not be required that all group members share the we-mode we-intention and we-commitment, but on functional grounds there must a substantial number of group members who indeed are collectively committed in the full we-mode sense to the ethos (see the more detailed discussion in Section VII). In this paper commitment, because based on intention, in general need be regarded only as non-normative and intention-relative, assuming that proper normativity is social, moral, deontic, or prudential or some other related kind of normativity. However, a commitment is trivially instrumentally normative in the sense e.g. that achieving a goal requires taking certain action. For us, commitment is the state of being bound as a result of the act of committing (binding) oneself. Forming an intention is the central way of committing oneself, the resulting commitment being intention-relative. (We will speak of commitment entailing something and being entailed, and this is to be understood in the same sense as in the case of intentions.) When people collectively accept to perform an action X in the we-mode, they collectively (jointly) we-intend in the we-mode to perform it for the use of the group, being collectively committed to doing what they collectively intend (cf. R. Tuomela s, 2002a, CAT formula as expressing we-mode collective acceptance). In contrast, in the I-mode case they may collectively in an aggregative sense accept to perform an X privately; here they we-intend only in a weak I-mode sense and are privately committed to what they intend. Their aggregative acceptance of action performance consists of each participant s private acceptance of it, either for the group or for himself. We will use the following terminology about commitment: Suppose a group g is committed (or g-committed) to p. In that case typically the group members (or almost all of them or at least of its operative members) are collectively committed qua group members (or, as we also say, we-committed in the we-mode) to p. This we-mode we-commitment involves the we-perspective, thus such a member (e.g. position holder) is committed qua a group member to p for the use and benefit of the group and believes that the others are similarly committed and that this is mutually believed among the participants of the collective commitment. As will be argued, members who thus have collectively committed themselves to p are also socially committed to each other to appropriately participate. Accordingly, on pain of social sanctions effected by the other participants, a we-mode we-commitment ought (at least in an instrumental sense of ought ) to persist as long as the joint

23 Acting as a Group Member and Collective Commitment 23 intention in question does. In the case of normatively structured groups, in order for the group to be committed the operative members or a substantial part of them can be required to be we-committed in the we-mode for the group. The operative members are members who have the authority (generally from the group members) to decide and/or act for the group. However, our formulations below will best suit the egalitarian cases with all members as operative members with their authority derived simply from their membership. A we-commitment can be taken to be in the in the we-mode when it satisfies the Collectivity Condition saying that, in virtue of the participants collective acceptance of a goal or view, etc. as the group s goal or view, it is necessarily the case that the accepted proposition is true or correctly assertable for the group if and only if it is so for any member of the group (cf. e.g. Tuomela, 2002a, Chapter 2). The Collectivity Condition guarantees that the participants stand or fall together with respect to the satisfaction conditions of p. In contrast, I-mode we-commitments are private commitments that do not satisfy the Collectivity Condition. IV Varieties of Acting As a Group Member As said, in addition to the above core notion of acting as a group member characterized in Section III, for some purposes other notions are needed, and we will discuss them below. 3 We will also need a wider notion that can be 3 The general view expressed in this paper bears similarity to the social identity theory and its refinement the self-categorization theory in spite of conceptual and terminological differences. The main concern in social identity theory is the social identity of persons (cf. Turner, 1987, Hogg and Abrams, 1988, 2001, Hogg and Tindale, 2002). This theory defines social identity as those aspects of an individual s self-concept which are based upon social group or category membership together with emotional, evaluative and other psychological correlates, e.g., the self defined as male, European, Londoner, etc. The most distinctive theoretical feature of the self-categorization analysis of group formation and group cohesion is the idea that these depend upon the perception of self and others as a cognitive unit (in contrast to non-members) within the psychological frame of reference, and not upon mutual interpersonal attraction and need satisfaction. (Turner, 1987, p. 64) The central and recently empirically debated hypotheses of this theory (according to Turner, 1987, p. 36) are, firstly, that people are motivated to establish a positively valued distinctiveness for groups with which they identify in contrast to relevant out-groups, and, secondly, that when social identity in terms of some group membership is unsatisfactory, members will attempt to leave that group (psychologically or in reality) to join

24 24 Raimo Tuomela and Maj Tuomela called acting in a group context. This notion will not be much used in the present paper, and thus the following loose characterization will suffice here. Acting in a group context means acting in the public domain of the group, as opposed to acting in the private sphere related to matters that can have some causal significance for the group s functioning. For instance, this kind of action can have consequences for the social atmosphere in, and the cohesion of, the group. The class of actions in group contexts contains as its subclass actions pertaining to the topics of concern of the group., viz. the constitutive topics of interest of the group. While then all acting as a group member (in one of our senses) is acting in a group context, the converse obviously is not true. Here are two specific examples. A group, having finished a job that belongs to its topics of concern, decides to go swimming (or to get some other kind of recreation). This action does not involve action as a group member but it is still acting in a group context and may causally affect the social atmosphere in the group. Our second example is one where the social group is a club and where a club member spontaneously cleans the floor of the clubroom, with the others possibly silent acceptance. Depending perhaps somewhat on what the club is about, this need not be some more positively distinct group and/or make their existing group more positively distinct. According to the self-categorization theory, the group has psychological reality in the sense that there is a specific psychological process, viz., self-categorization or selfgrouping, which corresponds to and underlies the distinctive features of group behavior (Turner, 1987, p. 66). This suggests that acting and functioning as a group member is closely related to taking oneself to have a certain kind of group identity in the sense of the self-categorization theory. The basic reason for acting as a group member and broadly speaking within the weperspective is that social identification with the group under suitable conditions of salience leads group members to act as group members. Most centrally, Turner argues that the cognitive process of depersonalization enables the shift from personal to collective identity: One s unique characteristics fade from awareness and one defines oneself in terms of stereotypical group characteristics (Turner. 1987, Chapter 3). Depersonalization in turn produces acting as a group member, to use our terminology. Group cooperation and positive sentiments towards the group and other members occur because identifying with others leads to a perceived similarity of interests and goals (shared ethos, in our approach). There will also be conformity because of the resulting adoption of shared group norms regulating action (cf. the function of collective commitment in our approach). Thus it can be said that it is a consequence of depersonalization that group members come to see themselves as parts of a group, we. Turner et al. also try to clarify the underlying causes of depersonalization and claim that it is the result of the interaction of three factors: the relative accessibility of a category (where the category roughly corresponds to a non-normative part of the ethos in our sense), its normative fit (corresponding, roughly, to the normative standards in a group ethos), and its comparative fit (reflecting a comparison between us and them, viz. in-group versus out-group).

25 Acting as a Group Member and Collective Commitment 25 acting as a group member, but it is at least acting in a group context. It is also positively causally relevant to group life (and is thus pro-group action). 4 To give our full list of notions of acting as a group member, we begin with a weak notion of acting as a group member. This notion is based on the group members sharing the ethos but being only privately (and not collectively) committed to it. Thus there may be a group of women who grow flowers in the village commons and intend to make their small village look beautiful in this and perhaps other ways. Each of them is only privately committed to making the village beautiful and they mutually know or believe that the others are similarly committed. The participants actions here may but need not be in other ways interdependent. In a weak, I-mode sense we here have action as a group member (based on I-mode collective acceptance). Thus we have this notion: (a) Acting as a group member in a weak sense is acting in the group context in the core sense (involving the action classes (1)-(4)) and promoting the same ethos, E, as the others and by purporting not to violate E. Ethos-promoting action should be understood in the same wide sense as in Section III. Only private (thus I-mode) commitment to E is involved here. (Thus E is not collectively accepted in the we-mode.) Let us next recapitulate in concise form our earlier account of the standard case. Acting as a group member in the standard sense is we-mode acting by means of actions belonging to one of the classes (1)-(4) above and thus it is we-mode acting in the core sense. More exactly, we can give the following analysis: (b) An action X performed by some member or members (or authorized representatives) of group g is acting as a member of g in the standard sense if and only if either the members of g rationally (reasonably) collectively accept for the group with collective commitment that X is an action in one of the senses (1)-(3) or the action X falls in class (4) (and is thus reasonably collectively acceptable as an action within the realm of concern of g and indeed as an action as a group member). This definition emphasizes the collective acceptance aspect involved. We could also have said that the collective acceptance must be in the we-mode. 4 Acting in a group context may also involve subconscious acting e.g. on the basis of ingrained subconscious group beliefs. Yet the acting must be done intentionally, although its belief-related reason remains unconscious at the time of acting. Thus when asked, the agent in question is taken to be disposed to admit that he meant to do what he did (successfully or not) or at least that it was not something he did unintentionally.

26 26 Raimo Tuomela and Maj Tuomela Accordingly, we can say that our notion of weak action as a group member is an I-mode notion and our notion of standard action as a group member is a we-mode notion, while both rely on the mode-independent core notion of acting as a group member. In the case of the standard sense (b) the we-mode is present by definition. The crucial element in a group s functioning as a group is the presence of collective commitment towards the collectively accepted item (and thus the members we-mode commitment and action). Backed further by our later discussion in Section VII concerning what a group s functioning as group involves, we will below also normally assume that in the kinds of acting as a group member to be defined, except in the above weak, I-mode case (a), the members will be collectively committed to acting in ways that at least weakly promote the ethos as a result. To be sure, the content of the collective acceptance in the present context is action X (action that upon collective acceptance will be action as a group member), but this collective acceptance will have to respect (promote) the ethos E (in one of our four senses (i) (iv)). Notice that actually the group s having accepted E as its ethos already entails that E is to be persistently upheld as will be clarified in Section VI. Thus, it is the task of the group members to collectively see to it that E is maintained and satisfied. In the standard case of acting as a group member, the members will have to act intentionally with the right purpose, even if the environment would not cooperate and even if the acting might be routine. One can obviously act within the realm of the group s concern but fail to obey the ethos of the group. For instance, one can even perform treasonable acts against the group and its ethos and be acting as a group member in the responsibility sense to be discussed next. We thus consider adding to the analysis of the standard notion a disjunctive clause that allows for intentional violation of the ethos. This liberalization can be made in view of the fact that the group is collectively committed to satisfying and/or maintaining the ethos and the ethos-spanned sphere of actions. Thus the group is committed to performing actions within (1) - (4) and to actions that accordingly are required for their performance or are otherwise ethos-promoting (these latter kinds of actions can be regarded as falling within class (4)). The group is therefore at least to some extent responsible for any wrongdoings. What we get is the responsibility sense of acting as a group member, given collective commitment among a substantial number of group members. This category is defined by adding the following disjunct to the analysis of the standard sense:

27 Acting as a Group Member and Collective Commitment 27 (5) A group member s intentional action which belongs to the realm of concern of g but which violates the ethos E of g, where the violating action itself is intentionally performed but need not be performed for the reason that it opposes the ethos. (5) then allows for revolutionary action that aims to overthrow or change E, based e.g. on on the agent s want to improve it. Such action can, alternatively, have the violation of E as its anticipated consequence rather than as its aim (as well as, compatibly with (b) just mistaken violation based on incorrect belief). Obviously, this type of action is in the I-mode and is exceptional (thus far from all in the group can at the same time satisfy (5)). Notice that a violator can violate a part of the ethos but still consistently be a group member in the sense of Section I. 5 We now have the following analysis, assuming a substantial amount of collective acceptance of the group s ethos and by implication of the action classes (1)-(4) with collective commitment to maintaining the ethos and thus to performing the actions in these classes: (c) An action X performed by some member or members (or authorized representatives) of group g is acting as a member of g in the responsibility sense if and only if X belongs to one of the classes (1) - (4), being thus action as a group member in the standard sense, or it falls in class (5). We can say that in case (5) g bears some (but not full) responsibility for the violator s action X, the degree of this responsibility depending on the 5 The issue of violators is too broad to be fully discussed in the present paper. Here, however, are some considerations making possible the existence of even wholesale violators, viz. group members rejecting all of the group s ethos and thus at least seemingly contradicting our functional criterion of group membership. To accommodate a wholesale intentional violator we may say this. The group members may falsely (mutually) believe that this person has accepted and is committed to uphold the ethos, at least largely, and thus is a group member. Having later learned from his intentional violating actions that he indeed is a wholesale violator, the group members might let him stay in the group but strictly control his (overt) actions and try to talk him back to proper membership and thus prevent harm from ensuing to the ethos and to other group members. Another possibility is that the wholesale violator aims to completely change the ethos of g. After an initial period of partial acceptance of the ethos in his acting (qualifying him as a group member), he still becomes committed to rejecting the current ethos and hopes to change it for a better one. Even at this later stage the group members might mutually take him to be a group member and also take partial responsibility for his actions in group contexts. There may be many other borderline cases analogous to our two examples but we cannot here discuss the matter further.

28 28 Raimo Tuomela and Maj Tuomela violator s purposes, etc. and on external circumstances. The group s responsibility for X (or the fact that it can be justifiably held as responsible for X) in cases (1) - (4) is based on rational we-mode collective acceptability while in class (5) the action, on the contrary, fails to be collectively acceptable. The essential element here is group members collective commitment to maintaining E, be it purely instrumentally, socially or (quasi-) morally grounded. The only if-part of c) holds only when the condition is understood widely enough. Basically, in the positive cases (1) - (4) the acting concerns the promotion of E while (5) represents the negative cases in which E is intentionally violated. (For a detailed presentation of cases that (5) here in a somewhat generalizing way is taken to represent, see Mäkelä and Tuomela, 2002, and M. Tuomela, 2003.) However, the group is only partly responsible for some member s action when he purports to act against the requirements of his position (group, ethos) and what is collectively acceptable according to the ethos. Such action in effect involves I am not acting as a group member (in the standard sense). Thus the member here is to a large part privately responsible for that action. Nevertheless, the group is responsible for who it accepts as its members, because there is normally (and always in the case of a rational group purporting to act as a group) group-commitment to the ethos (and to the group s identity). Thus the group members ought (at least in an instrumental sense defined with respect to E) collectively to see to it that E is maintained and not violated. So, for instance, the group should dispel members who do not act as ethos-respecting group members and it bears responsibility for membership and for controlling the members actions in a group context. The group is responsible for the action even if not in the fullest and direct sense. The group is responsible to group-external audiences for the member s violation but within the group the wrongdoer bears the primary responsibility. Thus a nation is partially responsible for the actions of its citizens when they violate e.g. a peace treaty with another nation. A nation that is split into subgroups is a poor partner for contracts. The government may lack control, although it ought to be in control when signing a contract with another nation. When a member acts as a group member in one of the above full senses (b) or (c) he can be said to represent the group or act in the name of the group. This is quite analogous to the case where an attorney by his actions represents an agent (viz. what the latter would rationally and justifiably do) in the court. Generally speaking, in such circumstances acting as a group member is related to the case where a group intentionally performs an action

29 Acting as a Group Member and Collective Commitment 29 (see Tuomela, 1989 and 1995, Chapter 5, for a detailed account). We distinguish between two types of cases, differing in generality. We can, firstly, speak of promoting the ethos on a general level. Taking group g to be capable of action qua a group, we assume that g intentionally sees to it that E or brings about E or at least tries to act so. We can also, alternatively and equivalently, say that the action here is describable as g s acting so as to promote its ethos. This is the case when its members jointly (try to) see to it or bring about that the ethos is promoted based on their joint intention We together will see to it that E is promoted. When they together see to it ( stit ) that E is promoted, every group member can be said to have a part or slice in this joint stiting. They may act separately or jointly to this effect and use whatever tools (e.g. hiring agents to do something) that are believed to be useful. Here the members part actions simply are actions qua a group member in the standard sense and by these actions they represent the group. Representation here is based on (justified) attribution their actions are partly attributable to the group (cf. the case where the attorney s actions are attributable to his agent as representing the latter s interests). The totality of the part actions, based on their jointly stiting that E, collectively taken constitutes the group s E-satisfying and E-maintaining action. The group members can be regarded as authorized (viz. collectively accepted) representatives of the group as long as they act as group members in the standard sense. The group acts through them and, conversely, they by their actions represent the group and thus the group s interests (ethos) and consequently their actions represent what the group does (or should E-correctly do) in the situation. The members part actions amount to the full group action but by them the members also represent the group s interests and E-correct action (viz. the totality of E-correct part actions) in a full sense. Leaving this big picture and concentrating on a small part of it, we arrive at our second, more concrete level. Here we may assume that the group performs a specific action, X, mutually believed to promote the ethos or at least not to violate it (recall Section III). In this case there are specific operative members that carry out the task and act so that the action X can be attributed to g on the basis of their together intentionally seeing to it that or bringing about X by their joint or separate action. (Cf. the leaders of the group acting making a deal in the name of the group; alternatively all the members may act as operative members.) The operatives see to it, or bring about, that X and thereby by their actions they strongly represent the group, provided that their actions indeed are ethos-correct and based on collective commitment (giv-

30 30 Raimo Tuomela and Maj Tuomela ing functional unity to the group). Here each participant performed a part or slice of his bringing about X and thus took part in strong representation. Acting as group member in this strong representational sense involving operatives actions amounts to successfully and correctly acting in ways falling within the union of classes (1)-(4). Typically, positional acting such as a professor s giving a course qualifies as strongly representational action performed by an operative member of a group (the university). Thus an action of the group thereby is intentionally generated, for the group saw to it or, perhaps rather, enabled that the professor gave the course. The members of the staff of the university can be said by their actions to strongly represent the university relative to its normatively specified task, viz. providing education to the students. The non-operative members part actions represent the group in the same weaker sense of representation as in the general ethos case, for, despite their role here, the non-operatives acted in the standard sense of acting as a group member. Acting as a group member in the standard sense is group-authorized action and the group can be said to act through the members actions. (If the action had not been successful we might perhaps have called it weakly representative of the group.) Actions by violators are not representative of the group, but they may occasionally be believed by outsiders to be representative; and in any case they signify also to an outsider that the group is responsible for these actions. (This might be called the responsibility sense of representation.) Note that an action as a group member in the strong representation sense is a special case ofan action as a group member in the standard sense (thus mere representation), whereas acting in the standard sense entails representation but not yet strong representation. Furthermore, an action can be acting as a group member in the responsibility sense without being representational action at all. This is because correct action is required in the representational case (but not in the responsibility sense (c)). However, an action representing the group is action as a group member in the responsibility sense. As to the private mode sense (a), it is not implied by any of the other notions (nor, of course, vice versa). Here is a comment related to collective acceptance that defines both the realm of concern of a group, its intentional horizon and ethos (and the ethospromoting action classes (1)-(4) of Section III). We can start with the weakest case where there is only mutual belief that the group collectively accepts E as its ethos and X as action as a group member (or as being at least weakly ethospromoting). This could only be de dicto belief that such and such is the ethos,

31 Acting as a Group Member and Collective Commitment 31 rather than a de re belief of something E that it is the group s ethos. Next, we can objectify the situation and require mutual knowledge of the ethos. Even in the knowledge case there will be borderline or grey situations. In any case we require that a group member has a view (belief) of whether or not an arbitrary X belongs to the ethos-governed or ethos-allowed field (and thus generally speaking is ethos-promoting and correct in the sense of collective acceptability) and also of whether it belongs to the realm of concern of g. His belief may be wrong, except in those very clear-cut cases in which there is mutual knowledge involved both of what the ethos contains and of whether a specific action fits within that area. 6 V The We-mode and the I-mode In this section we will analyze in detail the we-mode (in pre-analytic terms, thinking and acting as a group member) and the I-mode (pre-analytically, thinking and acting as a non-member or privately). Our we-mode and I- mode notions depend on the notion of acting as a group member in the core sense (cf. Section III). We can say that the we-mode is basic the mode of the group perspective. The group perspective gets a central part of its content from forgroupness, which concerns the promotion of the ethos (cf. We intend to uphold E for our group by means of actions in classes (1)-(4), We believe that p for the use of our group ). The other central element is collective commitment, which, especially the involved social commitment, dynamically glues the members to each other and to the ethos of the group. In our analyses in the present section the notion of group can be taken to be a weak one that does not entail that the group can act (either in an objective or in an inter-subjective sense of can ), contrary to the central cases in 6 Baltzer (2003) discusses the important features of anonymity and interchangeability of members in the case of large groups. He also emphasizes that in such groups every member represents the group and every other group member. We agree with his main points. Nevertheless, we wish to emphasize that the notion of acting as a group member of course applies equally well to large as to small groups. Furthermore, the point about representation also applies to small groups. In contrast to large groups, in small groups members tend to have many more private social relations with each other than in large groups. For instance, in the case of a family, these private social relationships are so strong and important that they may be taken to partially define the notion of family (or at least are typically central in the case of particular families).

32 32 Raimo Tuomela and Maj Tuomela the other sections of this paper. However, the members of the group must mutually believe (either in a de dicto sense, viz. under a certain description, or in a de re sense, viz. in the sense of direct acquaintance with, if not all members, at least some typical ones) that they are its members, and they are assumed to have at least approximate mutual knowledge of its ethos (its constitutive features). The group is we for them, but this need not be a conative we, one leading to collective activities, but may be only a non-conative we, one relating to commonly experiencing things of common concern (cf. Schutz, 1967, commented on in Tuomela, 2002c). The notion of functioning as a group member is of course to be understood in the sense of Sections III-IV. The group, g, in question can be either a normatively structured or an unstructured one. In the analyses below we are dealing with a group member who acts as a group member in a relevant sense (or, in the case of the last notion, only in a group context) and we ask in which different modes he could have an attitude (ATT). The attitude ATT need not be one that the ethos of g is concerned with (this applies especially to WM2). We propose the following definitions (cf. Toumela, 2002c): (WM1) Agent A, a member of group g, has a certain attitude ATT with content p in the we-mode relative to group g in a certain situation C if and only if A has ATT with content p and ATT(p) (the attitude with its content) has been collectively accepted with collective commitment (and this is mutually believed) in g as g s attitude, and A is functioning (viz. experiencing, thinking, and/or acting) qua a member of g in the standard sense and is collectively committed in the ATT-way to content p at least in part for g (viz. for the use of g) in C. (WM2) Agent A, a member of group g, has a certain attitude ATT with content p in the weak we-mode relative to group g in a certain situation C if and only if A has ATT with content p and is functioning qua a member of g in the standard sense and is also collectively committed in the ATT-way to content p at least in part for g (viz. for the use of g) in C. (IM1) Agent A, a member of group g, has a certain attitude ATT with content p in the plain I-mode relative to g in a certain situation C if and only if A has ATT with content p and is functioning as a group member of g in sense (a) (the the weakest sense of acting as a group member), and is privately committed in the ATT-way to content p at least in part for himself in C.

33 Acting as a Group Member and Collective Commitment 33 (IM2) (IM3) Agent A, a member of group g, has a certain attitude ATT with content p in the pro-group I-mode relative to g in a certain situation C if and only if A has ATT with content p, is functioning as a member of g in sense (a), and is privately committed in the ATT-way to content p at least in part for g in C. Agent A, a member of group g, has a certain attitude ATT with content p in the instrumental I-mode relative to group g in a certain situation C if and only if A is functioning in the group context related to g (but possibly not acting as a group member in any of our ethos-respecting senses), has ATT with content p and is privately committed in the ATT-way to content p at least in part for himself in C. Action modes can now be accounted for by means of attitude modes and the because of relation ( because in general expressing reason or cause): An action is performed in a certain kind of mode if and only if it is performed because of an attitude had in that same mode. Furthermore, we can take the psychologically effective reasons for action to be contents of attitudes or, in some special cases, the attitudes themselves (cf. Tuomela, 2000c, Searle, 2001, for such a view). Then the present account of an action performed because of an attitude in a certain mode amounts to saying that this action is performed for the reason expressed by the content of that attitude or, in some cases, that attitude with its content such that the reason is in the mode that the attitude is in. In this sense we can speak of e.g. we-mode and I-mode reasons for action. Due to ethos-related forgroupness (viz. the aspect that the content of a weattitude is meant for the use and benefit of the group) the reason for a person s performing an action X (of the right kind) as a group member in the we-mode cases is X is what the group s ethos requires or what serves it. When the members collectively accept such contents for the group they can use them in their inferences and act upon them. In the I-mode cases forgroupness reason is also possible (and even required in case (IM2)) and amounts to advancing the group s goals and interests (is to its benefit, in short). Cases (IM1) and (IM3) also require the presence of a for himself reason (e.g. I do X because it furthers my private interests and goals ). Above in (WM1) and (WM2) we assume that agent A is we-committed in the we-mode to content p (a content within the group s realm of concern) in the ATT way just in case he is collectively committed in the way that is special to the attitude ATT in question. Thus in the case of an intention, for

34 34 Raimo Tuomela and Maj Tuomela instance, he is supposed to respect the direction of fit of the satisfaction condition of intention, viz. world-to-mind direction of fit, while in the case of belief he takes into account the fact that a belief typically has the mind-toworld direction of fit. In (WM1) and (WM2) we-commitment, as defined earlier, entails that A believes that the other group members, or at least a substantial part of the operative members, are collectively committed to p (and believes also that there is mutual belief about this). A is assumed to be one of those members. Note that reflexive collective acceptance of the attitude as the group s attitude is required in (WM1) but not in (WM2) (for discussion, see Balzer and Tuomela, 1997, and Tuomela, 2002a, Chapter 6). In this sense (WM2) is based on cognitively weaker kind of collective acceptance and correspondingly collective commitment. For instance, cases with unreflective agents may fit (WM2); cf. Schutz s examples of primitive weawareness discussed in Tuomela 2002c. Also cases in which there is power structure in the group may satisfy (WM2), viz. there is collective commitment to the attitude but the attitude still is not collectively accepted by the participants as the group s attitude, although we may speak of collective acceptance in a weaker, less cognitive sense which accordingly makes the collective commitment weaker. (The reader should be reminded that collective acceptance of the attitude ATT which need not be ethos-related is not to be confused with the collective acceptance of the realm of concern and of the ethos of the group.) There are other possibilities of weakening (WM1) that are of interest. Thus it might be required that the participants only be privately committed to ATT(p) while still acting as group members in the standard sense (and thus being collectively committed to the ethos). Suppose a group has as its realm of concern to collect small objects of value. Its current ethos might just make it a stamp collectors club ( Our goal is to collect stamps ), while the members would also be privately committed, indeed wecommitted, to collecting coins. This would satisfy the present weakened definiens, and we might call this notion the mixed we-mode (and I-mode) notion and denote it by (WM3). As to the relationships between the I-mode notions, there are a few. The conjunction (IM1) & (IM2) & (IM3) is satisfiable and one can thus simultaneously act in all these three I-mode senses. Case (IM3) assumes that the agent is acting in a group context (and allows that he is possibly acting as a group member in the weak sense (a)). (IM1) obviously entails (IM3), and so does (IM2). (IM1) entails (IM2) in those cases where some forgroupness is involved, and (IM2) entails (IM1) in those contexts where the agent is not acting entirely for the group but also partly for himself. As, in contradistinc-

35 Acting as a Group Member and Collective Commitment 35 tion to (WM1), (WM2), (IM1) and (IM2), (IM3) then does not require that A respects the ethos, this case can accommodate intentional violators of the ethos. Obviously, our notions are just an interesting sample of a variety of modes. For other purposes different notions may be needed. We will later comment on some other aspects of commitment that relate to our above senses of the we-mode and the I-mode. Using concepts to be clarified in Section VIII, we will here only say the following. Being we-committed is taken to involve being socially committed. In the we-mode cases (WM1) and (WM2) being we-committed entails being committed to the others and/or to the group, but does not entail being committed to oneself (in a private sense). In contrast, private commitment in (IM1), (IM2), and (IM3) is to oneself and/or to the others and/or to the group (thus no possibilities are excluded). We may also ask for whom a commitment is made. In the we-mode cases (WM1) and (WM2) the we-commitment is for the group (in addition to being to and, in typical cases also before, the others or the group). In the I-mode cases (IM1), (IM2), and (IM3) the private commitment in principle can be for oneself (in the private sense), for the others, or for the group (where the conjunction of all these is a possibility). In general, forgroupness in our definitions can be taken to range quantitatively from null to full forgroupness. One can speak of the full private mode in the case where there is not relativity to any group. Note that one can act in a kind of quasi I-mode pro-group way in a group context without strictly acting as a group member in the weak sense (one might e.g. accidentally conform to the ethos or do things that indirectly contribute to the ethos). To make the we-mode and the I-mode analyses more analogous and better comparable, we have nevertheless required acting as a group member in a weak sense in (IM1) and (IM2), but not in case (IM3), where almost anything in the group context as opposed to the private context goes. Let us also note that one may give more refined analyses of the notions of commitment involved above, using the distinctions to be made in Section VIII. Thus, in relation to (IM2) a distinction can be made between a group member s being privately committed to himself to act for the group versus his being committed to the others (or committed to the group) to act for the group. The strongest or most groupish I-mode case is (IM2). It is one with private commitment, which is to the group and before the group; and it allows for cases that are only for the group (then the for-himself part shrinks to zero). As to group identification, a group member can identify with the group in various ways. The strongest way to do it is obviously in sense (WM1) and the

36 36 Raimo Tuomela and Maj Tuomela next strongest in the sense (WM2). (IM2) involves a weak element of group (or collective) identification. Clearly, the more we-mode thinking and acting in the group there is the more cohesive it is, the primary factor accounting for cohesion being of course collective commitment (cf. our comments in Section VII). Here are some obvious further points related to our analyses (we draw on Tuomela, 2002c, where a more detailed discussion is available): (i) The I-mode (in any of the senses (IM1), (IM2), or (IM3)) does not entail the we-mode (either in the sense (WM1) or (WM2)). (ii) The we-mode (in sense (WM1) or (WM2)) does not entail the I- mode (in sense (IM1), (IM2), or (IM3)). Thus it can accordingly be said that the we-mode is not reducible to the I- mode. However, note that the we-mode is in the minds and actions of individuals. While the we-mode represents group-level thinking and acting and as a concept is a holistic one, ontically it pertains to individuals and does not postulate supra-individual entities. This entails that the we-mode (involving e.g. agents having thoughts about groups and social structures) does not add anything importantly different to the ontology, although states of we-thinking and I-thinking are ontically different states. To discuss (i) and (ii) briefly, consider an intentional token of an attitude or action, X, that qualifies as acting as a group member. Such an action can be identified either broadly or narrowly. The narrow identification involves the reason that serves to make it intentional. This reason will here be either the we-mode reason or the I-mode reason in the sense discussed in Section I, viz. it is a reason that the agent intentionally has either as a group member or as private person. Suppose X is waving one s hand and indicates the wide identification of its tokens. Suppose there are two different narrow identifications of a certain token of X making it both a personal greeting (I-mode reason) and a token of voting (we-mode reason). If a token is a greeting it is not a voting, and conversely. Thus, the above theses should not be understood in this sense. However, if the present token of hand waving is a token of an I-mode action (greeting) this does not yet entail that it is also a token of a we-mode action (voting), or vice versa, although there happens to be a coincidence in this particular case. This is the sense in which we mean the above theses to be understood. Obviously, there can be mixed we-mode/i-mode cases. For instance, people working in an organization accept to act in the we-mode concerning

37 Acting as a Group Member and Collective Commitment 37 the basic rules of the organization serving its ethos. In the same context they may act in the I-mode and e.g. compete with each other (cf. competing salesmen in the same business company). Let us jot this down as: (iii) Social activities such as group actions and social practices may involve mixtures of we-mode and I-mode actions. The importance of the collective commitment aspect of the we-mode will be argued for in the next section. As to the problem under what conditions the we-mode can be expected and argued to supersede the I-mode, see especially the lengthy discussion in Tuomela (2002c). In this paper we will not consider in detail why the we-mode is needed for the construction of the social world, but some remarks, drawing on Tuomela (2002c) are due. The reasons favoring we-mode thinking and acting can be conceptual and/or constitutive in the sense of showing how the we-mode can serve to constitute new, emergent things, or they can be rational in the sense that instrumental or end rationality favors the we-mode over the I-mode, or they can be factual in some looser sense than strict rationality involves. In our we-mode notions the requirements of functioning as a group member and the presence of forgroupness and collective commitment are central. As seen, forgroupness involves acting so as to produce something for the use and benefit of the group, and it thus concerns what is thought by the group members (or in the case of some non-democratic groups, the group) to be in the interest of the group in the sense of serving at least to some extent the group s and its members constitutive or central goals or interests to the extent the group s and its members goals are compatible with each other (which is not always the case, for here also the I-mode goals of the members must count). We recall that full-blown acting as a group member requires that the ethos of the group be respected at least as far as the reason of the action is concerned (although the intended action itself might fail). This is a kind of cooperation-generating feature. Furthermore, this central feature tends at least to some extent to lead to harmonious social relations between group members. The group regarded as some kind of unity becomes central in people s activities. Because of the cooperative design of organizations and institutions this often requires of the members that their resulting we-mode action, governed by the group s goals and interests, not be merely strategic action vis-à-vis the internal affairs taking place in the group (although it of course may be strategic concerning other groups). However, group goals may

38 38 Raimo Tuomela and Maj Tuomela require coordination and other kinds of strategic action, and mixed cases of we-mode and I-mode activities are possible; and even we-mode competition is possible. In the we-mode case things are being put in the public domain (or rather the group domain) precisely in the sense forgroupness and collective (and the entailed social) commitment do it: Forgroupness licenses the use of collectively accepted ideas in practical inference and action related to the group domain. When intending jointly and therefore being collectively committed the participants give away part of their autonomy to the group, so to speak, and thus to the public domain. We can say that their we-mode actions are out in the open. In other words, we get shareware for the group members. In principle, everybody is assumed to contribute to the group good, to do his fair share, but is also allowed to have his piece of the group cake (products of the joint enterprise), so to speak. No one is the sole author of the we-mode action, even if it is a singular action. A member cannot do fully as he chooses. He is committed not only to himself but also, socially, to the others and/or to the group to perform his share. Note, however, that our we-mode explicates are somewhat idealized and that in actual practice they often are mixed with strategic I-mode thinking and acting also in group contexts. Thus, in realistic cases of the performance of a joint action, say X, people might do all that X requires but they might still harm others and try to satisfy their own private interests perhaps by strategic means (cf. a member secretly plotting against some other members to achieve leadership while still performing his official tasks satisfactorily). This is something relating to the way or manner of doing X but not to the achievement of X itself. Our account (cf. (WM1) and (WM2)) really requires acting for the benefit of the group and this does not allow people to harm each other, at least to a substantial degree. The requirement of acting as a group member for the group guarantees that. One may of course relax the idealized notions (WM1) and (WM2) of the we-mode and require harmonious and cooperative thinking and acting only with respect to X but not with respect to all the manner aspects related to X. However, presently our concern is not so much realism but finding and elucidating central conceptual notions required for describing social life. 7 7 In Tuomela (2002c) the relationship between the we-mode and the I-mode is investigated in functional contexts, such as cooperation, and in institutional contexts. One can indeed find plenty of cases in which we-mode thinking and acting is functionally superior to I-mode thinking and acting (and conversely). R. Tuomela s paper is mainly concerned with comparing the we-mode with (IM2) and (IM3), viz. pro-group I-mode and

39 Acting as a Group Member and Collective Commitment 39 VI Collective Commitment, Ethos, and Presupposed Collective Action This section investigates the relationship between the ethos, collective commitment, and group action. To begin, we consider a social group, g, with a certain ethos (constitutive goals, values, beliefs, standards, norms). Group g s identity is at least in part defined by its collectively accepted ethos, E. That this definition holds is a constitutive principle for the group, and its collective acceptance, at least if rationally made, already gives a kind of weak normative reason for maintaining it. As all of the group s constitutive principles, based merely on collective acceptance, can be changed by the group, so can the group itself in principle change its own identity, thus change E. (We assume here that g is in this sense an autonomous group.) However, as long as it is true that E has been collectively accepted by g s members as its ethos, upholding E is normatively required: It ought to be the case in our group that E is maintained as our ethos (or, if you prefer a weaker statement, On balance, there is normative pro-reason for us to maintain E as our ethos ). This is not a completely tautologous statement. In effect it says that we ought to maintain our group ( us ), but it does not say for how long or in which circumstances. It ought to be the case in our group that E is maintained as our ethos, given that the group accepts E as its ethos. The condition here can get different context-dependent explications. For instance, in a democratic, self-determining group it might amount to given that the overwhelming majority of group members accepts E as the ethos of the group. When speaking of the ethos we are dealing with group survival and the factors that determine it. In typical cases life over death is to be preferred. Here the central element is our collective acceptance to keep the group together, and this collective acceptance may be based on our various reasons and, in some cases, on external circumstances. It is basically a contingent problem how long that collective commitment lasts. The only non-contingent statement that seems possible here is the one that in a rational group it ought to be the case that group g maintains its ethos and thus its identity as long as it accepts E as its ethos (or, in our weaker language, there is normative on-balance reason for maintaining E as long as no reason arises to change or abolish E). The content of as long as is to be determined a posinstrumental I-mode, because of the defined notions they come closest to the we-mode and thus are the hardest to defeat.

40 40 Raimo Tuomela and Maj Tuomela teriori on the grounds of the group members collective acceptance and whatever reasons it is based on (recall the above example using the overwhelming majority principle as a criterion). This kind of commitment to E involves rational persistence and resistance towards change that is not present in e.g. desires and hopes, and the collective aspect makes solo quitting reproachable in principle. We can say that there is intrinsic resistance on conceptual grounds against the group s changing its constitution (identity). Another route to rational resistance to change the group s identity may come from functional considerations related to the group s particular non-constitutive goals and aims. Suppose thus we are a group of musicians forming a quartet and that we plan to arrange a concert on the fiftieth anniversary of Sibelius death in 2007 and maybe have other similar plans. Clearly, then we plan to maintain our group (and thus our ethos E) or there is strong normative reason for maintaining E until that date. Thus, in our example the group will persist and live with the identity involved in E due to collective acceptance and the involved collective commitment. More generally, if collective acceptance of the ethos E is to be rational in the present functional sense, then, other things being equal, the group ought to keep E at least up to the point of time that its plans of action concern. Both the identity argument and the functionality argument bear an analogy to the case in which the group has promised to itself to keep its identity under such and such conditions (e.g. until 2007 in the case of our group of musicians). Then as promising entails undertaking an obligation, the group ought to maintain its ethos and identity at least as long as those conditions obtain. This way of putting the matter indicates that we do not derive the statement that it ought-to-be in the group that E is maintained from the mere natural fact of collective acceptance of E, but rather we derive the ought-to-be from normative (possibly only instrumentally normative) collective acceptance amounting, as it were, to the group s promise to itself: We collectively bind ourselves to be this kind of group, with an identity shaped by E. The above considerations based on the group s identity (and thus, so to speak, on group life versus death) gives a partial justification to the statement It ought-to-be the case in g that E, given that the group has accepted E. It follows that the group members ought to perform relevant world-to-mind activities related to the maintenance and upholding of E is g s ethos (e.g. the Wednesday Lunch Group ought in this way to maintain its ethos Our group meets for lunch every Wednesday ). The group members thus ought

41 Acting as a Group Member and Collective Commitment 41 to maintain E by their actions (and having rationally arrived at its ethos, without further reason, to refrain from changing the ethos e.g. to collecting stamps instead of lunching together). Obviously, given the above E, there is also another ought-to-do norm for the members: they ought to come to the lunch every Wednesday. Thus the statement It ought-to-be the case in g that E is its ethos (or, in the weaker case, There is on-balance normative reason in g to maintain E ) can be taken to entail two different ought-to-do requirements (or in the weaker case give two kinds of on-balance normative action reasons) for the group members. Of course, the members of a free group g could rationally have collectively accepted another ethos instead of E. Thus it is not unconditionally true that g ought to have accepted E as its ethos. Accordingly, the ought-to-be can be changed, and it can be rationally changed if the group finds good reasons to do so group identity need not be that tightly connected to the present ethos. As seen, maintaining or upholding E as the ethos of the group must be clearly distinguished from satisfying it. The group s attitude towards the ethos has world-to-mind direction of fit of satisfaction, and this is the case not only for the goals, standards, and norms involved but also for the constitutive beliefs included in E. The former relies on the group s acceptance: It has accepted E rather than something else as its ethos and continues to accept it. Why it so continues is related not only to the constitutive nature of E serving to shape the identity of the group (the point argued for), but, as seen, also to the group s functionality and instrumental rationality. Thus it is rational for the group to persist in keeping its identity through time in order to make both intra-group interaction and the group s interaction with other (singular as well as collective) agents functional. The group s having adopted E as its ethos requires of a rational group that its members at least to some extent not only try to maintain it but also try to satisfy its content. Thus, given the group s rational acceptance of E, other things being equal, they ought to maintain it and hence resist changing it, and they ought to perform ethos-promoting actions and thus actions conducive to the satisfaction of the ethos. These actions are ones falling into our classes (1) - (4) of section III. Also the normatively required actions by the group members for maintaining the ethos as the group s ethos fall into these classes, especially (1) - (3). However, note that e.g. defending the ethos against attempts to change it is a maintenance action (e.g. maintaining it as the group s ethos that they are to lunch together on Wednesdays) that is to be distinguished from the actions that satisfy the ethos (actually having lunch together on Wednesdays). Of

42 42 Raimo Tuomela and Maj Tuomela course, it is a factual truth that satisfying actions contribute to the maintenance of the ethos. It would be rational for the lunch group to give up its ethos and the group would indeed disintegrate if hardly anybody ever came for lunch on Wednesdays. We recall that, minimally, collective commitment amounts to shared wemode we-commitment: Ideally, everyone in the group is committed to a joint project or action (assumed to satisfy the Collectivity Condition) and believes that the others are similarly committed and that there is mutual belief about this. Here commitment in a single member s case means that he has bound himself, both in his relevant thoughts and actions, to the joint project (joint action) in question and thereby to the others in the group relative to that project. At least in the case of a rational member, he will also have the belief that the others or a substantial part of them are similarly committed. This is we-mode commitment in the sense that the person commits himself as a group member rather than privately. The person in question is disposed to think thoughts of the kind qua members of g we ought to see to it that E and, hence, I ought to participate in our joint action and owe this to the others, viz. I ought to do this, in part, for the reason that the others rightfully expect it from me. The shared (either instrumentally or intrinsically normative) statement we ought to see to it that E is kept as our ethos and is satisfied by us, based on it ought to be the case in our group that E is our ethos, expresses an inter-subjective ought-to-do norm in the group. As seen, the (ought-to-be) norm that E ought to obtain entails concerning its satisfaction that at least rational group members persist in holding it and oppose change, given that the collective acceptance of E has been rational and no important changes in the situation have occurred. This kind of satisfaction must be distinguished from the satisfaction of the content of E. For instance if E = G is a (constitutive) goal of g, then this content is satisfied by the members achieving G for the right reason, viz. precisely that G is the goal of g. In all, both the maintaining of the ethos as the group s ethos and the satisfaction of the ethos require the right kind of obeying action, and such action will be promoting action in the sense clarified in Section III. 8 8 Our basic principles of normative ethos maintenance and change can be schematically put as follows for suitable, generally contingently determined conditions C: (i) g may be either free ( autonomous ) or unfree (ii) a free group g ought to maintain its ethos E, given C (e.g. C might be that the overwhelming majority of the group members collectively accept E as g s ethos); (iii) a free group g may change E, given C (C as in (ii)); (iv) change of E or its substantial parts entails change of identity of g, but the group

43 Acting as a Group Member and Collective Commitment 43 In the case where the group collectively accepts the ethos, E, in the wemode, it is entailed that the group members jointly intend to see to it that it will be maintained and satisfied. Thus the group members are collectively committed to E (viz. its maintenance and especially to its satisfaction). However, there is a difference between the cases where there is mutual knowledge and where there is mere mutual belief about the existence of collective commitment. In the former case the group members must somehow have made public their intentions and commitments to the ethos and what it covers. In the case of merely mutually believed collective commitment, the members may not know that the others are we-committed in the we-mode. Hence they cannot justly sanction each other (when needed), because they cannot be sure that the others indeed failed to do something (as the putative violators might actually not have had the intention in question). If there is mutually known collective commitment the group will be publicly bound in the group to maintain and satisfy the ethos and bear responsibility for these two matters and hence for the group members acting correctly as group members. While with mutually believed collective commitment we get the right kind of collective maintenance and satisfaction behavior, indeed group-binding group action, we do not yet have the publicity and thus objectively justified demands on the members as we would get with mutually known collective commitment. 9 To end this section we consider a weak case of collective action which nevertheless weakly respects the requirements entailed by the ethos of the group and shows that promoting the ethos need not be very demanding, after all, for the group members. To begin with, we note that there can be group action in a weak sense on the basis of shared I-mode commitment to take part in collective action recall our I-mode sense (a) of acting as a group member. In such a case a ought to keep its identity, unless strong reasons for changing it emerge (such as C in (iii)); (v) a free group s commitments not related directly to E may require (ii). Note that it follows from (ii) that a free group g ought not to change E, given C, but the converse is not true. Clause (iv) serves to justify (ii), and similarly do the commitments dealt with in (v). 9 In stronger cases with public multilateral intention expression (non-normative case) we get a better doxastic ground for group action and in the case of the participants having made an agreement to act in a certain way (this includes the case of normative acceptance of a joint plan) we get even a group-binding obligation. Even in this last case we might not get normal positive group action but only omission action. In this sense we might not have full group action based on a joint intention or a shared collective goal.

44 44 Raimo Tuomela and Maj Tuomela participant is not socially committed to the others to participate. Here the participants are not, so to speak, traveling with shared authority in the same boat heading for ethos E (this represents the we-mode case), but rather each of them is sailing towards E in his own boat with intact autonomy (I-mode case). Thus, if we are to have group action qua group action, viz. group action in which necessarily the group acts as a group, joint intention and wemode we-commitment on the part of the participants is required. We will now consider a weaker sense of acting as a group which on the one hand does not require joint intention to promote the ethos or even acting for the reason that the group ethos thereby gets promoted (this is what the I-mode sense (a) of acting as a group member would yield). The group action that we have in mind here does involve collective commitment to the right kind of ethosentailed action but it does not make the ethos a reason for action. This kind of group action consists of our ethos promoting cases (iii) or (iv). Thus, the action need not be more than refraining from action, viz. intentional omission to act in a certain way prohibited by the ethos (this is case (iii) of promoting the ethos of Section III). Considering specifically this case, such action can be called presuppositional we-mode group action where the essential thing is that the ethos functions as a kind of underlying presuppositional reason for the participants action ( collective pattern-governed behavior in the sense of Chapter 3 of Tuomela, 2002, is a case in point). Here the agent can have any kinds of (further or immediate) goals as long as he does not violate the ethos. There need be no common goal here. For example, not violating the traffic rules of Finland is correct societal behavior, and the group members do not strictly act together when driving in the traffic, although they can be said together to promote the traffic rules (part of the ethos) often non-intentionally (viz. not under the right description while still intentionally doing what is right). As another, rather similar example, consider the Flat Earth case the group s ethos includes the belief that the earth is flat. Here we have shared we-mode action as a group member, and such action is presuppositionally constrained (it cannot violate the belief content that the earth is flat). Nevertheless, the actions need not have the same goals. We still have presuppositional we-mode collective action (viz. collective presuppositional non-violation of traffic rules when driving and collective flat-earth respecting activity, respectively). In the case of presuppositional collective action and practice the actions (be they separate or joint actions) that are involved typically respect the ethos only in the sense that while the action in question is intentionally performed (or, in a still weaker case, is not non-intentionally performed) it still does not

45 Acting as a Group Member and Collective Commitment 45 involve the agents intention to maintain or to satisfy the ethos (even if that content or result may come about in a known and even mutually known way). We may say that the ethos is only a presupposition of action and its presuppositional reason. If things start going wrong, and the agent gets social sanctions for his behavior, he may realize that he does not operate according to what is the right thing to do in his group. While presuppositional collective action generally falls short of being group action based on the joint intention to maintain and satisfy the ethos, it may still well be based on collective commitment to act in certain ways that we as theoreticians know to be the right ways according to the ethos. The result of this kind of weaker group action is, nevertheless, that the ethos is promoted and satisfied (albeit not for quite the right reason). For instance, the members preserve order in traffic by presuppositional respect of the traffic rules. As noted, the participants are collectively committed to relevant conformity but not to acting for the reason that the action is required by the ethos. It goes without saying that they need not be paying attention to their being so committed, even if they must of course notice what they are doing under their own descriptions. Notice that we have here more than the ethos getting promoted as merely an unintended consequence of what the people do. We have an underlying structure (e.g. the system of laws of a state) which allows us to say that the group members together promote the ethos, albeit not perhaps intentionally. But had they reflected upon the situation they might have formed the joint intention together intentionally to promote the ethos by their separate actions. There are various mixed possibilities that all serve to create the kind of behavioral order demanded by the ethos but which fall short of being proper we-mode group action based on a joint intention to promote the ethos. In addition to having a mixture of participants acting in the weak I-mode sense (a) of acting as a group member and participants acting in the above presuppositional sense (involving one of the cases (iii) (iv) of promoting the ethos), there can also of course be participants acting in the we-mode for the right ethos-based reason. VII Why Require Collective Commitment? In this section we will try to pull together our different arguments for requiring collective commitment in the analysis of group action. We will below present six interrelated but still different arguments to support the view that

46 46 Raimo Tuomela and Maj Tuomela the group perspective and especially its collective commitment as its essential ingredient is needed to account for group action and group life in general: (1) We-Identity. A group s identity consists mainly of the group s ethos, E, to which the group if rational at all is constitutively committed as a group. Thus the group members or at least a substantial number of them must bind themselves to E and be mutually aware that the others also are similarly committed. This is shared we-mode we-commitment to E. Collective commitment is a kind of glue that keeps the group together as a unit, each member thinking and acting with a we-perspective (the group s perspective) instead of an I-perspective. The we-perspective is here grounded in the shared ethos to which the members are collectively committed. This perspective involves strong collective sociality in the we-mode sense. Thus it involves forgroupness as expressed by group thoughts of the kind We intend to do X or We believe that p and collective commitment to these thoughts; in addition it also includes we-feelings, viz. affective bonds. The group will have to be able to function as a unit, and it accordingly needs an identity, constituted mainly by its ethos E, which gives the constitutive content of the group and serves as the basis of forgroupness, viz. E gives the content of the reasons to think and act for the group. The group members are attached to E through collective commitment. Without a substantial amount of collective commitment towards the ethos, the group accordingly risks losing its identity (or part of it) and falling apart into an aggregate of independent actors furthering the same ethos. Collective commitment functions to guarantee that the identity stays the same, or becomes what the group intends it to become, and serves to unite the members around that identity. Privately committed members may be strongly committed to E, but their commitment does not add to the group s identity but only to their own (social) identity without gluing them to the other members. While the ethos gives the contentual part of the identity of the group, collective commitment per se, independently of the nature of the ethos that it concerns, accounts for the cohesion and causal we-ness in the group. The main work of mutually believed collective commitment is to keep the group members together. This is the basis for a we-perspective that involves shared we-mode attitudes expressible by e.g. We will do X, We believe that p (in contrast to I will do X or I will do X given that you do X, We all intend to do X, etc, and ditto for beliefs). Without collective commitment to E there is not enough reason for the members of g or for outsiders to hold these

47 Acting as a Group Member and Collective Commitment 47 persons as constituting a group (in the sense of a group capable of responsible action as a group) but only reason for holding them as an aggregate of persons privately committed to the same ethos. (Note, however, that groups are not agents and cannot act on their own, although we metaphorically speak of a group s action.) As seen, the group acts as a group (at least in the weak, presuppositional sense of the previous section) when its members act qua group members in a standard sense (cf. above). Case (5) of Section III with intentional incorrectness or unacceptability has relevance only for deciding about retrospective group responsibility for violators. When the violator acts, the group must assume partial responsibility for the action at least vis-à-vis a third party (an external observer): The group should have appropriately controlled its members. 10 Admittedly, aggregates of persons with private commitment to further E may also constitute groups, but in a much weaker sense. Indeed they can only be groups that cannot act as groups, for the latter require the kind of collective unity and persistence guaranteed by collective commitment. The weakest case of a group that can act as a group seems to be one based on shared we-attitudes in the we-mode involving the participants collective commitment to the attitude contents and to each other (cf. Tuomela, 2002b, Chapter 2). When a collectively committed group has some privately committed members, these members are not full-fledged group members. (More exactly, this is so, if it is believed in the group and is out in the open that they are only privately committed. If, in contrast, they are only secretly privately committed they may be falsely believed to be full-fledged members.) As to group action, these members in a weak sense may walk out or do something wrong without less ado than in the collective commitment case. These members are not full-fledged arms and legs in the collective body in question. If, on the other hand, they are mutually believed to be collectively committed, they are treated as full-fledged members and tend to surprise the group when they act according to their own minds. 10 Here is a point concerning outsiders evaluation of group responsibility: Outsiders may not know whether the group members (or, operative members, as the case may be) have collectively committed themselves to E. This is, however, what they must know in order to have a valid claim about group responsibility. However, when the group has promised to perform a certain task as a group or led the outsiders normatively to believe this, the outsiders have some justification to require that the group take responsibility for the group members actions in a group context, be these members in fact committed to E or not.

48 48 Raimo Tuomela and Maj Tuomela (The following observation shows that mutual belief about collective commitment suffices for actual collective commitment, assuming commitment to be a persistent intention to participate: If there is mutual belief among a substantial part of (the operative) group members to the effect that the whole group as a group is committed in the world-to-mind fashion to ethos E, then there is shared we-mode commitment (intention) to contribute to E by those members and hence, assuming successful execution of intention, collective action results. This result requires the additional, but highly reasonable assumption in its antecedent that each member who believes that he is committed also is actually committed.) (2) Responsibility. In order for the group to be able to answer general, both group-external and group-internal, demands for taking responsibility or being responsible, collective commitment must be present. Generally speaking, collective commitment (or, in the non-objective, groupjective case, the mutual belief that there is collective commitment in the group) gives the group members authorization and justified instrumental obligation to control the other group members relevant actions, and thus they can be held responsible for wrongdoings (viz. actions by group members which violate the ethos and what it derivatively covers). A group is responsible for not having controlled its members action at least when there is mutually believed collective commitment. When there are some members who are not collectively committed (and this is mutually believed in the group) the group takes responsibility for having allowed them to take part in group action. The responsibility is based on the group s commitment to see to it (stit) that the ethos is upheld and promoted. Thus, in this case the group members not only ought to but will actually try to collectively see to it that the members participate in upholding the ethos. So in the above kind of case of mutually believed collective commitment by a substantial number of operative members the group is committed, by the observation result mentioned above in the last paragraph of (1), to seeing to it that an action X is performed or that E is upheld by acceptable or correct actions (be those actions joint or separate actions). This mutually believed stit imposes oughts and entails mays or if you prefer, duties and rights of at least an instrumental, E-relative kind on the group members to help, control, pressure, etc. others when needed so that achieving the end result in a joint venture is facilitated. Group activities of this kind are in principle everybody s business. A group member s action is transparent in the sense that it is open for scrutiny by the other members. His action is not his private business.

49 Acting as a Group Member and Collective Commitment 49 (3) Functionality. As in effect seen, collective commitment contributes to the functionality of the group. Without collective commitment the group cannot properly act and function as a group and thus cannot properly fulfill its tasks and purposes. Part of what is involved here over and above what aggregated private commitments give is the social commitment involved in collective commitment: Being committed to each other, the group members can better count on the others performing their tasks, which especially in the case of interdependent and joint actions is central (cf. below). To elaborate, a group cannot function as a group if the members are only privately committed to E. One member may stop acting or act incorrectly if he changes his mind about being privately committed to E. In general, the group needs good reasons to believe that the members are collectively committed in order to dare to act as a group. If all or most of the members are only privately committed to E it would be like an armada of boats without a commander of the armada. Each captain of a member boat is on his own in his attempt to get to the end port E (or to an end promoting it). Those who decide to take on other goals cannot be controlled or sanctioned for that on (even instrumental) social grounds. Given that no extra assumptions concerning the participants interconnection (such as specific agreement or plans to act jointly) are made, merely privately committed members in interdependent I-mode activities have a right to leave, as they have kept their full autonomy to act in this situation. Other participants in the situation may criticize quitters, as they may lose their chance to arrive at their I-mode goals or shared I-mode goal, but they know they have no authority behind their demands to make the quitters stay. On functional grounds we need a group that acts in a coherent way through coherently acting members all of whom participate in seeing to it that they get to the same port. The mutually believed collective commitment gives this kind of stit with at least instrumental normativity to back the activities. The social commitment in the collective commitment is the core of the glue that gives them shared authority over the oughts and mays in question. This brings us to the next argument for collective commitment. (4) Social commitment. As seen, social commitment by itself involves features that can be required for groupness and that in its stronger forms gives a link to moral considerations (avoidance of harm, reproach, etc.). To elaborate, a collective commitment involves social commitment of the members to one another or to the group even when it is only instrumental (in the same sense as in the sentence In order to open the window you ought to go to the window first ). The social commitment here is a commitment to the other

50 50 Raimo Tuomela and Maj Tuomela participants of the joint venture (joint action or acting together to uphold E) or, more generally, to the group. This social commitment is the core of the reproachability feature of collective commitment: a publicly we-committed member who leaves the joint project or intentionally violates the ethos can be criticized by the others, whereas if he had been committed only to himself he would have been less socially criticizable for letting the others down. In the private case he can only be criticized for not being a stable person (a criticism of the rationality of the person). In the purely instrumental collective commitment case a violator can be criticized for soloing instead of acting qua a group member. This collective commitment involves the member s being committed to the others to perform his action. He has given up his full authority over his own action to the group. In the case of quasi-moral and moral commitment, the reproachability may justifiably involve accusations of the person s harming the others by acting in the wrong way. However, such accusations are not warranted if the person in question is only privately committed. Collective commitment entails at least an instrumental ought to see to it together with the others that the ethos is upheld or changed in the direction the group wants, perhaps making it a new group when the change is radical enough. The social aspect of collective commitment gives us the other members partial authority behind that ought, as a participant is committed not only to himself but to the others and to the group. (5) Group goals and beliefs. Groups may have goals and interests that the members do not have and which may even conflict with the members goals and interests (cf. the positional model of group attitudes of Tuomela, 1995, and Section V). When properly acting as group members they will also collectively accept goals and views for the group that they would not adopt or have if only acting privately. For such new goals and beliefs to function properly for the group, collective commitment to them is required. The present argument is not a direct argument for the requirement of collective commitment, but it shows indirectly that also when the ethos of the group does not fully coincide with the private views and ideas of every, or possibly even any one member, collective commitment is a forceful, entailed companion of collective acceptance. (6) We-mode instead of I-mode: Reviewing (1)-(5) we may conclude that the we-mode (and a we-perspective) is often superior to the I-mode (and an I- perspective) in the case of some task performances (here assumed to be mode-independently described). (This argument generalizes the functionality argument (3).)

51 Acting as a Group Member and Collective Commitment 51 Consider thus these two contrasting cases: I accept We together will see to it that X in order to uphold the ethos (group or we-perspective) rather than I will see to it that X (e.g. act towards the satisfaction of the ethos), given that the others or sufficiently many of them also do X. In the we-mode case, we is the intentional subject of attitudes and action, while I is the respective intentional subject in the I-mode case. The we-perspective expresses a voluntarily made non-contingent connection between the group members: qua entering the group they necessarily are connected as a we capable of joint thinking or acting. The we-mode thus conceptually involves the group as an actor in the agents minds so that they necessarily belong to the group and do not only treat it as an element of their environment. (As argued in Tuomela, 2002c, and in (1)-(5), the we-mode is in many group contexts superior to the I-mode, and the former in contrast to the latter requires collective commitment.) In the I-mode case, I may interact with you and act towards the same goal to the extent it is not too costly. However, in the we-mode case the cost argument in general has much less bite. The group member acting qua a group member is supposed not to think of his fate in private terms but only as a member of the group (and thus the group s successes). Collective commitment gives a we-perspective: Our group achieved something. When acting in the we-mode, a participant does not easily leave when it becomes costly for him to stay and participate, because he sees the project as the group s project (as our project) rather than as his own private one. A participant cannot be jumping on and off or in and out and only take part when it suits him. A group does not want to keep members who are there to free-ride on the goodies, but are not going to pay the costs when the going gets tough. VIII The Constitution of Commitment As seen, the social commitment involved in collective commitment (in our account assumed to be a conceptual ingredient in joint intention) is in many ways central. 11 But there are also other constitutive elements in commitment. 11 In the early work on joint intention by Raimo Tuomela the notion of collective or joint commitment is not directly used, although it is entailed (cf. Tuomela, 1984, Tuomela and Miller, 1988). However, basically the same functional role is played by the principles of practical reasoning assumed in this work. A joint intention consists of shared weintentions about which there is mutual belief. A we-intention can be expressed in the form We will do X accepted by each (operative) participant; here X represents a joint

52 52 Raimo Tuomela and Maj Tuomela We will now classify and discuss them briefly and later show the relevance of our classifications to acting as a group member. We will characterize commitaction.) It is required that the we-intenders be disposed to reason in accordance with certain patterns of practical inference, typified by the following schema called (W2): (W2) (i) we will do X. (ii) X cannot be performed by us unless we perform action Z (for instance, in the case of a cooperative action type X, teach agent A, who is one of us, to do somethin related to his performance of actions required of him for X). (iii) We will do Z. (iv) Unless I perform Y, we cannot perform Z. (v) I will do Y (as my contribution to Z). Clearly, (W2) expresses a central element in social commitment. As to the current technical literature on joint commitment, there are not many accounts available. The recent theory by Panzarasa et al. (2002) is worth a short comment. According to these authors joint commitment is taken to amount to persistent joint intention. Joint intention is characterized roughly as follows (we use only verbal descriptions for a two-person case below), where agents A and B jointly intend that a state expressed by a proposition s be attained (there is no mention in these authors treatment of joint attainment). Omitting time considerations, the following analysis is offered A and B jointly intend that s if and only if (i) (a) A intends that s (that the state expressed by s be attained); (b) B intends that s (that the state expressed by s be attained); (ii) A and B mutually believe that (i); (iii) (a) A intends that B will intend that s; (b) B intends that A will intend that s; (iv) A and B mutually believe that (iii). The authors take social commitment to a group, g, to be a primitive notion and define a single agent s social commitment to the group as his singleton group s commitment to the group g. Basically a member A is socially committed to g to making a proposition s true if and only if there is a sequence of actions e such that A is committed to g to performing e, and, in addition, either e is A s plan for achieving s alone, or it is his plan for enabling group g to achieve s, or e is A s plan for achieving s in collaboration with g. Next, a group of persons is defined to have a joint commitment to making s true if and only if they have a joint persistent intention towards s. Furthermore, they have a joint persistent intention towards s if and only if (i) in g it is mutually believed (thus, in the two-person case, A and B mutually believe) that s will be true at a later point of time. (ii) g has (resp. A and B have) the joint intention that s will be true at that later point of time. (iii) each member of g (resp. A and B each) is socially committed to g to fulfilling the joint intention. (iv) in g it is mutually believed (resp. A and B mutually believe) that each member is socially committed to g to fulfilling the joint intention; and (v) it is true (and mutually believed in g) that (ii) will continue to hold until it is mutually believed in g either that s will not be true at the mentioned later point of time, or at least one of the members has no longer the motivation to be part of the group and drops its commitment. Clause (v) seems somewhat circular (what is the motivation?), but we will not here try to improve on the present authors account, which also contains some other, easily repairable formal flaws. Joint intention, as discussed in this paper and in Tuomela (2002b, 2003), involves joint commitment and hence social commitment as well as joint persistence; also the property that the agents must intend that the other one intends to act will hold true (see the proof in Tuomela, 2003). Thus our account entails the features that are central in the Panzarasa et al. account. There is, however, the problem with the latter that it does not

53 Acting as a Group Member and Collective Commitment 53 ment in terms of seven variables. Firstly, obviously the subject of commitment is either a private agent or several agents or a (structured or unstructured) social group. Secondly, one can be committed to a propositional content or, if a more externalist language is preferred, to a state of affairs or end, or to an item in such a content (or state) such as an object (e.g. person) or property. Every such commitment, nevertheless, entails a commitment to act appropriately. For instance, an agent may be committed to good personal relations obtaining in his group, and this entails that he at least in an instrumental sense ought to see to it and, when obeying the ought, he will see to it that that end state obtains. Seeing to it is an action of a general kind. Accordingly, any commitment involves also commitment to action of some kind, standard singular action (e.g. opening the window by the committed person) or to a part of a joint action (e.g. painting the front of the house in the case of some persons acting together to paint a house). Commitment to a belief analogously involves professing the belief and persistently acting on it. We list the action in question as the third constitutive element of commitment (it can be the same action as the one in the content of the commitment, if there is no other content). Fourthly, commitment can be either private (in the sense of involving private authority only) or collective (involving collective authority). A person seem to work unless the proposition expresses joint action (or some closely related togetherness element). Joint intention involves joint action at least in the sense that the participants together will see to it that something is the case (and the participants often will themselves by their own joint action bring about that something). Let now s = the house is painted. Suppose both A and B intend s and intend that the other one intends s, all this under conditions of mutual belief. Here each agent s intention seems to be a single-agent action intention. So here A intends by his actions to bring it about that B intends by his actions to bring about s (and vice versa). But this is not joint action. We are dealing with A s and B s interdependent I-mode intentions but not with their proper we-mode joint intention, and what the realization of these intentions amounts to is that the agents eventually try to bring about s; and if s can indeed be brought about two persons, they may succeed. But the risk of failure is greater than in the case of proper joint action (viz. action as a group). This is because there is better coordination and support in joint action (joint seeing to it that s). But in the Panzarasa et al. account we are (or at least may be) dealing with separate actions. And this of course is no surprise, as the agents lack a proper joint intention. Making the authors joint intention persistent helps a little, but it still fails to make the joint intention and the joint action properly connected. We claim that unless one takes joint intention to be a primitive notion involving relevant actional togetherness (cf. We together will [viz. intend to] see to it that s ) their analysis fails. Note too that the togetherness we claim is missing in their account is built into the notion of we-intention as an attitude, not in the content of the attitude. Thus s by itself may express a joint action or, as well, a non-actional state.

54 54 Raimo Tuomela and Maj Tuomela has a private commitment if he has mentally bound himself in a conative way to a propositional content or an element e.g. an object or property involved in such a content and is disposed to act so as to make the content satisfied. The direction of fit of satisfaction will be world-to-mind. Thus the agent will conate or intend to bring about a state of affairs or to keep a state of affairs as it stands (or something related). The private authority aspect entails to use our terminology below that the agent is committed only to himself and, typically although not necessarily, at least in part for himself. (Thus the authority feature in part contains some of the constitutive elements to be listed below.) An agent s intention to open the window is typically an example of a private commitment and so is his belief that 1+1 = 2. In contrast, a collective commitment at least in its full sense involves an item collectively accepted by some persons as an item to which they as a group are committed. Furthermore, the collective commitment is social in several senses. The persons are committed to the others (and possibly to their group in a more general, holistic sense) and the commitment is a we-commitment involving the mutual belief that the others are similarly committed; more exactly, each participant is committed to the item and believes that the others are and that there is a mutual belief concerning this among them. A collective commitment is for the group and typically (but not necessarily) also public (relative to g). A typical example of a collective commitment is that involved in a full-blown joint intention by some persons to paint a house together or to perform some other joint action, one involving parts or shares for the participants. Our fifth constitutive feature is the mentioned one that the committed person can be committed to himself, to the others, or to the group (cf. Castelfranchi, 1995, Conte and Castelfranchi, 1995, for related distinctions and points). The first two of these components (commitment to oneself and to the others) were already discussed, but let us here make a point about the social sources of commitment related to how violations are sanctioned: (a) The weakest case is purely instrumental. For instance, if A and B have made up the joint plan to paint a house together, it might happen that A had misestimated his skills and ends up losing more time on the job than was estimated in the joint plan and more than he really can afford. Then he just informs B about the situation and is not subject to more sanctioning than perhaps a charge of misestimation. (b) In this case the reaction to violation has the form violating behavior is socially inappropriate or incorrect. There are three sub-cases here:

55 Acting as a Group Member and Collective Commitment 55 (i) The violating behavior concerns the ethos of the group (viz. against the basic goals, values, standards, beliefs, norms, etc. of the group). (ii) The violation concerns the general conventions and customs of the society. (iii) The violation concerns the moral norms and principles internalized by the host society (or at least against what is felt or believed to be morally right or socially appropriate in the society). In all of the b-cases there is also instrumental social commitment (case (a)) involved. This is because of the very nature of joint activities as involving in general only shared control over the outcome. Commitment to one s group involves supporting and promoting the group s ethos and also advancing its less central goals, interests, standards, beliefs, etc. Thus a sports fan may be committed to a sports club and promote its interests and goals. This includes also non-constitutive interests, goals, etc., of the group in addition to its constitutive ones (the ethos). The main conceptual and theoretical difference between a commitment to one s group and a commitment to the members (qua members) of the group is that a group is an entity with a history and an open-ended future. In this sense the group as a conglomeration of its changing members, their interrelations, as well as its ethos and its non-constitutive goals, standards, norms, etc. involves more than the set of its members, qua ethos-maintaining members, at a particular moment. We take commitment to the members of the group to involve basically the latter. The difference between being committed to the other group members and to the group is in part from a social psychological rather than a conceptual point of view also a difference in the personal view of a group member and can vary depending on the case. Some people may tend to think of the others as the entity my group while some other group members may of think of the group as constituted by the members (qua members) of the group. From a conceptual point of view, his being committed commitment to the group entails commitment to its members, and actually also conversely: commitment to the members qua group members presupposes commitment to the group, although this need not be psychologically realized and salient in the members. If I am committed to the group (viewed as an entity) to do my best, I must be committed to the group members collectively taken to do my best. But being bound to the group members qua group members in this collective and perhaps anonymous sense (involving in principle also persons who are not presently members) entails commitment to the group because of the qua-relation.

56 56 Raimo Tuomela and Maj Tuomela Commitment to the group (and commitment to the entailed activities), when psychologically salient, gives additional strength to all other kinds of commitment concerning group matters. This is rather obvious, as commitment to the group entails commitment to group members. But it also involves more, for it contributes to group cohesion (to the bond between the members and the members and the group). Sixthly, an agent can be committed before oneself (or have only himself privately as his audience ), before the others in his group, or before his group. Thus, an agent is committed only before himself if he secretly forms the intention to achieve something (e.g. to help another person or to rob a bank). He is committed before the others if the other members of his group or a subset of them is his (public) audience, and he is committed before the group if he e.g. declares in a newspaper that he will fight for his group (e.g. sports team or country). It is a problem whether I am committed to you to do X strictly entails I am committed to myself to do X. Our suggestion is that the entailment does not hold, although it seems reasonable to accept that the related secondorder statement I am committed to myself to carry out my commitment to you to do X is entailed. The following can, however, seems tenable: If I intend to open the window if and only if I am committed to myself, or the other(s) or to the group to open the window, where the or is an inclusive one. This equivalence lets us move back and forth between the to-person language and the plain action-commitment language. Note that in the case of collective commitment, a social commitment to the group members, including oneself, is entailed. Thus whenever I am committed to you to open the window I will also be committed to myself to open the window, given the assumption of our being collectively committed to seeing to it that I open the window. Accordingly, in the central cases dealt with in this paper, the discussed entailment indeed holds. By using the to-person terminology we get a unified terminological and logical account. Furthermore, social locutions like I promise to you to do X suggest that in the general case the to-person language should be used. Our above equivalence shows that it does not add anything to the non-social case expressed by means of action commitment only. Our main reason for using the to-person notion and terminology is simply that it is needed in the case of full-blown collective commitment. Seventhly, the item which the commitment is about (viz. a state of affairs or an action) or the consequences of its being realized (e.g. a goal which is achieved) can be for the use of the committed agent only, for (all or

57 Acting as a Group Member and Collective Commitment 57 some) other members in his group, or for the whole group, holistically understood. Let us now summarize our account of commitment. We are concerned with a member of group g who is acting as a group member in one of our earlier senses. We have seven variables some of them with non-exclusive values. Using shorthand, they can be written as follows. (1) agent: (a) single, (b) several agents collectively, (c) social group (2) content: (a) propositional content or state (b) non-propositional element, such as object or action, of a propositional content or state of affairs (e.g. ethos, country, swimming) (3) action: (a) single, (b) joint (c) group action (4) authority: (a) private, (b) several members collectively, (c) group (5) agent-object of commitment: (a) oneself, (b) others in g, (c) group (6) audience: (a) before oneself, (b) before others in g, (c) before group g (7) use (of result of commitment realization): (a) for oneself, (b) for others in the group, (c) for the group So, commitment can be symbolized by the following seven-place relation COMM: COMM(agent, content, action, authority, object, audience, use). In plain language this reads: In being we-committed an agent is committed to a certain content, to perform a type of action, with a certain authority, to an agentobject, before a certain audience, and for the use of some person(s) or group. We claim that in the case of commitment in a social situation our seven variables will all apply in the sense that one or more of the values (disjuncts) will obtain. For instance, in (2), (a) and (b) both hold true in many cases, similarly in (3), (4), and (5) both (b) and (c) are often both satisfied. As to (4), one may be both privately and collectively committed (viz. we-committed ) concerning the same action (making (a) and (b) both hold true). On the other hand, e.g. within (3), the conjunction (3)(a) & (3)(b) is trivially inconsistent. Similarly one may study which value combinations concerning different variables are possible. For instance, (3)(c) & (4)(a) is an inconsistent combination. We will not here systematically study all these relationships but only make a few more remarks about interesting possibilities. The commitment content in (2) was assumed to entail an action. This entailment may lead e.g. to conditional action prescriptions, such as Per-

58 58 Raimo Tuomela and Maj Tuomela form X when conditions C obtain, and the entailed actions may be complex ones. What follows concerning (3) is that the agent possibly need not actually act at each point of time when he is committed to something (as the conditions C, etc., for acting may not obtain). As to variable (6), commitment before oneself in the social case may be unilateral commitment. Thus a person may be unilaterally committed to the others without their knowing about it. For instance, I may offer a gift to you, to advance your interests, whether you want it or not. The commitment can of course also be commitment before the others in g and here the others will know about it. For instance, I may promise something to you or express my intention to do something for your benefit (which you accept), and in that sense there is and must be both offer and acceptance. Your acceptance can be just silent acceptance. The last variable (7) is the use variable. In its strong, or rational version, use means not only what one is permitted and, in some cases, required to use, but also the kind of use which will further the purposes and interests respectively of the single user (case (a)), of the group members (case (b)) or of the group (case (c)). As to (c), for the group can be related to various kinds of commitment. In collective we-mode commitment it is a constitutive part of the commitment, for such collective commitment has a collective or jointness content or entails such a content satisfying the Collectivity Condition discussed in Section III. This jointness content basically boils down to our jointly acting together towards something for our group s use (and, generally, benefit). In private commitment the forgroupness feature makes the commitment pro-group. There are some other obvious interpretations and qualifications concerning our variables e.g. in (5) a group can be taken to be committed to itself, to other groups or to some individuals, or to a group of groups (cf. the EU). The combination of (6) (a) and (7) (b) can be understood as a weak, onesided social commitment ((5) (b) or (5) (c)). E.g. a person can commit herself before herself to advance some interest of the group, and this can be regarded as her social commitment to the group to provide the service. IX Commitment and Acting as a Group Member To end this paper, we will reconsider our cases of acting as a group member. We distinguished between several notions or senses: acting qua a group member in a weak, I-mode sense, standard sense, responsibility sense (possi-

59 Acting as a Group Member and Collective Commitment 59 bly with incorrect action), and (strong) representation sense. The task ahead is now to say what kinds of commitment are possible in these cases. Here is our concise and rather obvious account, given our earlier extensive discussion: Acting qua a group member in standard sense. The following can be said about the required commitment in this case: (1) Ideally, there must be collective commitment to E (to uphold E together with others) and thus to intentionally act in accordance with E, viz. to obey E or at least try to obey it. Thus the action classes (1) - (4) of section III are included; these classes also contain unsuccessful intentional attempts to obey E. (In weaker cases at least a substantial number of the operative agents ought to be collectively committed to E and perform actions falling within classes (1) - (4).) (2) Collective commitment involves social commitment: the group members are not only committed to themselves but committed to each other and to the group to uphold E (and to act in ways falling into one of the classes). (3) There is commitment to support E and other, non-constitutive interests, goals, standards, values, beliefs, norms, etc., and accordingly there is commitment to perform actions in classes (1) - (4) in our classification of Section III. (4) In relation to above point (2), the committed actions should be transparent, out in the open, as knowledge of them is gratis shareware in the group. Thus, they cannot be private secrets. They are open for the others to monitor, because the action is part of the group s action or group activities (more generally). Each (operative) group member is committed to the others and to the group when he acts, which means he has given up part of his authority over his own action. His action is not entirely his own business. He is committed to the others to do what he ought to do basically to see to it that E is upheld or to act with this effect, and if there are nonconstitutive goals, they should be collectively observed as well. This entails that he should also monitor what the others are doing. The responsibility is shared for maintaining the ethos by E-congruent actions (and possibly other acceptable group actions) and for the consequences all this has for a third party (audience, external observer). Especially, the group s responsibility to an external audience becomes a costly burden for a group that does not see to it that its members are sincerely collectively committed to E and that does not monitor that incompetent or oppositional members are kept in control, helped or dismissed. We did not know for sure whether they were collectively committed is not an acceptable excuse. They should have found out. Commitment before others is thus recommendable.

60 60 Raimo Tuomela and Maj Tuomela (5) Forgroupness, as discussed in the previous section, is a feature of the collective commitment to E and E-congruent actions (viz. actions falling within the classes (1)-(4) of Section III). (6) Commitment to the group in general (involving its ethos, structure, history, expected future, past, current, and perhaps future members) can be ideally required: it adds strength to the to, for, and before aspects of commitment. (E.g. one can thus be committed to a country, university, or firm.) Acting as a group member in the responsibility sense. The group must be collectively committed to its ethos and to performing actions within classes (1)- (4) of Section III. There is at least an instrumental obligation to do this, and it is strengthened by the social element with at least an instrumental normative component in the social commitment contained in collective commitment, as emphasized many times over in this paper. In this sense we can speak of the group s prospective responsibility here. There is also a corresponding retrospective responsibility in the sense that it bears responsibility also for actions violating the ethos and the ethos-entailed goals, standards, etc. as well as the non-constitutive goals of the group, if there happen to be such. This is what our class (5) with its intentionally performed incorrect or unacceptable actions was meant to capture. In all, the group is responsible both for classes (1) - (4) of Section III (prospective responsibility obligation to perform actions falling into one of these classes) and also retrospectively responsible for (5) (violation). It should have controlled at least what kind of persons are accepted as members (assumed to be we-committed to E). 12 Acting as a group member in the representation sense. Recall the standard sense for what kinds of commitment are required, as they apply here as well. Recall that in the case of acting as a group member in the standard sense mistaken actions are acceptable. In the case of acting as a group member in 12 As to intentional violators (members of g) in the sense of action class 5) of Section III, the following kinds of commitment are possible in their case. There may be private commitment to the topic of concern of g while there is no commitment to the group s ethos (a violator might be a rebel purporting to change the group s ethos). Hence there is no commitment to ethos-congruent action or type. However, a violator may be committed to himself to act in ways related to the topic of concern of g (and perhaps to his modified ethos), but that commitment need not be more than commitment before himself and only for himself. There is possibly no commitment to the group or to the other group members, although a violator may be committed to the group, apart from the aspect of its ethos that he purports to violate.

61 Acting as a Group Member and Collective Commitment 61 the strong representation sense correct actions for the right reason (viz. actions obeying the ethos) are required. Weak sense of acting as a group member. This is the I-mode case with (1) private commitment to E, (2) private commitment to oneself or the others or possibly to the group, (3) private commitment to E and to actions conducive to E, (4) commitment before others or only before oneself, (5) the commitment is (in part) for the group or merely for oneself. Acting as a group member in an organization. Finally, let thus briefly consider a special but typical group, viz. an organization such as a university or a business company. Such an organization has a certain ethos, basic goals, values, standards, etc. These give the content of the organization we may say. Apart from its content, an organization basically consists of a set of positions that are normatively interlocked to each other. Specifically the interrelations between the positions contain power relations and informational relations. (E.g. a position holder can have the power to order the position holder of another position to do something within the ethos-induced domain and different position holders may have, and have the right to have, differing amounts and kinds of knowledge concerning the organization s activities.) When the organization makes decisions, it makes them in virtue of its authorized operatives (typically joint) decisions and when it acts, it performs its actions in virtue of its possibly different authorized operatives actions. The operative members may vary from occasion and task to another, and they may also include non-member operatives on some occasions. Thus when a business company does something X, e.g. decides to start selling a new brand of goods or builds a new house for its own use, the operative members are presumably different and possibly not group members at all. The Hobbesian Author behind the company s actions are the shareholders. They have the power to select a governing board for the company and derivatively it has the power to select the functional position holders from the CEO down to a salesperson and truck driver. We can say that the Author (viz. the group formed of the shareholders) is a core level group in an organization while the functioning operatives (the CEO, various kinds of managers, workers, and what have you) form its hired personnel. Still it is in virtue of their concrete action that concrete actions are in general attributed to the organization, for they represent the organization in the case of almost all issues in the daily life of the organization. Thus the hired personnel can be seen as part of the group in a wide sense. In a business company, with some exceptions such as provided by family businesses, all the position holders are hired ones (the shareholders are not

62 62 Raimo Tuomela and Maj Tuomela position holders in the organization itself). Even though they are hired, when performing company jobs they will act in the codified we-mode that the company charter and other ethos-principles involve. So company functions will in general involve (codified) we-mode activities. This means that there is also a codified social commitment between the position holders: a position holder is often responsible to another one (if they are interlocking positions or are somehow in suitable power relation to each other). In any case a position holder is socially committed to the company to performing his tasks. All the position holders that are hired by a company can be operative members or (when not members) operative agents for some tasks, but hired agents that are not position holders but only temporarily connected to the organization for some tasks are not operative members but only the company s means or tools. As seen, the ultimate core of an organization is constituted on the one hand, by its owners (if it has owners in the legal sense) and, on the other hand, by its positions that constitute the organization. The owners determine (often via suitable operative members) the ethos (and thus the general content) of the organization, and the position holders act to achieve and/or maintain it, we may say. Some of the position holders may in fact have as their task to reformulate the ethos within certain boundaries. What kinds of commitment are there in an organization? There is codified we-mode collective (and social) commitment that involves the positionholders binding themselves qua position holders to the organization to perform certain tasks and indeed to obey the task-right system pertaining to their positions. This codified collective and social commitment is constituted by the obligations and rights involved in the organizational norms. In addition, there will be uncodified we-mode collective and social commitment involved in the freely chosen joint and jointly accountable activities that the position holders choose to do (but which still are promotive of the organizations s ethos). The we-mode activities and commitments that we are speaking of here of course need not be in agreement with what the positionholders might have privately preferred in those situations (maybe they would have privately wanted to spend the day on the beach, etc.). However, the more genuine (viz. in the sense of backed by private preferences and wants) the we-mode activities and commitments are, the better the organization is likely to function in the long run.

63 Acting as a Group Member and Collective Commitment 63 X Conclusion In this paper we have studied in detail two central social notions, acting as a group member and collective commitment. Our study of the first of these notions is as far as we know the first systematic work on the topic. Acting as a group member is a central notion that obviously must be understood when speaking of the we-perspective, group life, and of social life more generally. Thus, not only philosophy of sociality, philosophy of social science, political and moral philosophy but also the various social sciences need this notion and should benefit from our analyses and arguments. Collective commitment is the other we-perspective notion studied in our paper. We have argued for its importance as representing a kind of social glue needed for group members when thinking and acting as a group. In contrast to some other studies (e.g. Gilbert, 1989, 2000, Castelfranchi, 1995) our most elementary notion of collective commitment is not intrinsically normative but is only instrumentally normative and intention-relative. Thus our treatment covers more ground than the previous accounts do.* * We wish to thank Kaarlo Miller for painstakingly checking our manuscript and for providing us with sharp and insightful comments. References Baltzer, U., 2003, Social Action in Large Groups, Protosociology 18, pp Balzer, W. and Tuomela, R., 1997b, A Fixed Point Approach to Collective Attitudes, in Holmström-Hintikka, G. and Tuomela, R. (eds.), Contemporary Action Theory II, Kluwer Academic Publishers, Dordrecht and Boston, pp Castelfranchi, C., 1995, Commitment: From Intentions to Groups and Organizations, in Proceedings of ICMAS 95, MIT Press, Cambridge, Mass., pp Cohen, P. and Levesque, H., and Smith, I.,1997, On Team Formation, in Holmström-Hintikka, G. and Tuomela, R. (eds.), Contemporary Action Theory II, Kluwer Academic Publishers, Dordrecht and Boston, pp Conte, R. and Castelfranchi, C., 1995, Cognitive and Social Action, UCL Press, London Gilbert, M., 1989, On Social Facts, Routledge, London Gilbert, M., 2000, Sociality and Responsibility, Rowman and Littlefield, Lanham Hogg, M. and Abrams, D., 1988, Social Identifications, Routledge, London Hogg, M. and Abrams, D., 2001, Intergroup Relations: Essential Readings, Psychology Press, Hove

64 64 Raimo Tuomela and Maj Tuomela Hogg, M. and Tindale, S., 2002, Blackwell Handbook of Social Psychology: Group Processes, Blackwell, Oxford Mäkelä, P. and Tuomela, R., 2002, Group Action and Group Responsibility, Protosociology XVI, Panzarasa, P., Jennings, N. Norman, T., 2002, Formalizing Collaborative Decisionmaking and Practical Reasoning in Multi-agent Systems, Journal of Logic and Computation 12, Schutz, A., 1967 (orig. 1932), The Phenomenology of the Social World, Northwestern University Press Searle, J., 2001, Rationality in Action, The MIT Press, Cambridge, Mass. Tuomela, M., 2003, A Collective s Rational Trust in a Collective, Protosociology 18, pp Tuomela, R., 1984, A Theory of Social Action, Synthese Library, Reidel Publishing Company, Dordrecht and Boston Tuomela, R., 1989, Collective Action, Supervenience, and Constitution, Synthese 80, Tuomela, R., 1992, Group Beliefs, Synthese 91, Tuomela, R., 1995, The Importance of Us: A Philosophical Study of Basic Social Notions, Stanford University Press Tuomela, R., 2000a, Cooperation: A Philosophical Study, Philosophical Studies Series, Kluwer Academic Publishers, Dordrecht and Boston Tuomela, R., 2000b, Reasons for Action, in Brogaard, B. (ed.), Rationality and Irrationality (Contributions of the Austrian Ludwig Wittgenstein Society 8), Kirchberg. Austrian L. Wittgenstein Society, pp Tuomela, R., 2000c, Belief versus Acceptance, Philosophical Explorations 2, Tuomela, R., 2002a, The Philosophy of Social Practices: A Collective Acceptance View, Cambridge University Press, Cambridge Tuomela, R., 2002b, Joint Intention and Commitment, in Meggle, G. (ed.), Social Facts & Collective Intentionality, in German Library of Sciences, Philosophical Research, vol. 1, Dr. Hänsel-Hohenhausen AG, Frankfurt, pp (earlier announced to be published in Grazer Philosophische Studien) Tuomela, R., 2002c, The We-mode and the I-mode, in F. Schmitt (ed.), Socializing Metaphysics: The Nature of Social Reality, Rowman and Littlefield, Lanham, Md, pp Tuomela, R., 2003, We-Intentions Revisited, Philosophical Studies (forthcoming) Tuomela, R. and Bonnevier-Tuomela, M., 1992, Social Norms, Tasks, and Roles, Reports from the Department of Philosophy, University of Helsinki, N:o 1, 46 pp. Tuomela, R. and Bonnevier-Tuomela, M., 1997, From Social Imitation to Teamwork, in Holmström-Hintikka, G. and Tuomela, R. (eds.), Contemporary Action Theory, Vol. II: Social Action, Kluwer Academic Publishers, Dordrecht and Boston, pp Tuomela, R. and Miller, K. 1988, We-Intentions, Philosophical Studies 53, pp

65 Acting as a Group Member and Collective Commitment 65 Turner, J. C., 1987, Rediscovering the Social Group: A Self-Categorization Theory, Basil Blackwell, Oxford

66 66 Kay Mathiesen On Collective Identity Kay Mathiesen Abstract In this paper, I examine a particularly important kind of social group, what I call a collective. Collectives are distinguished from other social groups by the fact that the members of collectives can think and act in the name of the group; they can collectively plan for its future, work for its success, and grieve at its failure. As a result, collectives have certain person-like properties that other social groups lack. I argue that persons form collectives by taking a shared first person plural perspective based on a shared collective self-concept. This collective self-concept defines the collective, and, along with the causal history of the collective, uniquely identifies it. Over time, collectives persist through changes in membership by maintaining this collective self-concept. Changes in the collective self-concept may be identity preserving, however, if they result from processes that naturally flow from the identity of the collective and which are appropriately caused by (or accepted by) the collective and are connected to previous stages of the collective via collective memories. Introduction The French people, the Jones family, the tennis club, the gay community, the Democratic Party, the Catholic Church these human groups, what I will call collectives, are a ubiquitous and familiar part of our world. We belong to them, leave them, disagree with them, blame them, and praise them. Open any newspaper and you will see references to the beliefs, actions, and concerns of the Navajo, the Palestinians, a Sims 1 community, etc. Collectives not only make their appearance in our everyday life and discourse; they also play a key role in our political, sociological, and psychological discourse and practice. While a number of philosophers have provided accounts of collective action 2 and corporate responsibility, 3 fewer have concerned themselves 1 An on-line role-playing game where people create families of Sims (simulated people), who interact with other Sims and create neighborhoods, etc. ( 2 See e.g., Gilbert (1989, 2000), Tuomela (1995, 2000), Bratman (1999), Kutz (2000), Miller (2001), Velleman (1997), and Searle (1990). 3 See e.g., French (1984).

67 On Collective Identity 67 with the nature of these robust and persistent, but often less formally organized or explicitly rule-bound, collectives. In particular, there is very little work on collective identity. An account of collective identity would allow us to understand how such collectives may be distinguished from each other and how they can persist over time the traditional concerns of philosophers dealing with personal identity. 4 That such an account is sorely needed is clear from the key role that collectives play in our everyday lives and social science discourse. But questions related to collectives also play a key role in a number of philosophical areas, including social ontology, social epistemology, and political and social theory. To give an example of just one such issue from ethics, consider the question of collective responsibility. If we want to say that some collective C was responsible for some action p, then we need to be able to determine that (a) collectives have the capacity to be responsible for their actions, (b) C is a collective, (c) C is the particular collective that is responsible, and (d) the current collective C* existing now at t 2 is the same collective as C that performed the action at t 1. 5 Furthermore, as a result of their special status, collectives have a particular moral importance that other social groups do not. Individuals partake in the hopes, regrets, plans, and successes of the collectives to which they belong; therefore, collectives have an important role in the lives of individuals. Collectives do not simply form the individual or have effects on her quality of life, or the projects that she can carry out; they form one of the ways in which she can be in the world. Being an individual with her own particular plans is only one such way. Another way is to be a member of a collective and to share in plans with others. For this she needs other people and she needs to be able to maintain those traditions, practices, etc. that tie her to these others. As a result collectives have a kind of normative importance that other social groups lack. In this paper, I begin to fill the gap in our knowledge of collectives by providing the beginning of an account of what collectives are and how they persist through time. My main contention in this paper is that individuals can form a collective when they adopt a first person plural perspective defined by a shared collective self-concept, which specifies the collective s history, be- 4 Indeed, some of the more insightful comments about collective identity have come from philosophers comparing it to personal identity (see e.g., Hume (1978), Parfit (1984), and Nozick (1981)). 5 While I do not explicitly discuss collective responsibility in this paper, my account of collective identity would clearly provide answers to all of these questions.

68 68 Kay Mathiesen liefs, aspirations, and other characteristics. In short, collectives are distinguished from other social groups by the fact that the members of collectives conceive of themselves and others as co-members and they have a shared concept of what it means to be one of us. This allows the members of these groups to think and act in the name of the group to plan for its future, to work for its success, and to grieve at its failure. 6 I begin by defending the above view that individuals can form collectives by adopting a first person plural perspective based on a shared collective selfconcept. I then consider and respond to a number of objections to this view. In the second part of the paper, I discuss identity criteria for collectives I provide an account of both synchronic (at a time) and diachronic (over time) identity. In the discussion of synchronic identity, I argue that each collective is defined by its causal history and the content of the collective self-concept. In the discussion of diachronic identity, I argue that collectives maintain their identity over time via their shared self-concept, causal history, and shared collective memory. I. Identifying Collectives As I noted above, collectives (e.g., the Navajo, the Jones family, etc.) are very special sorts of social groups. They have person-like properties, including the capacity to act, to plan, to regret, to hope, to succeed, or to fail. This, at least, is how we think and talk about them. One way to explain this is to argue that when applied to collectives these words do not mean the same thing as when we apply them to persons. But, I am going to assume that we know what we are talking about and that our use of these terms when speaking of collectives is not a mere equivocation or metaphor. The test of this will be whether we can explain how attributing such features to collectives can be appropriate. So, what is distinctive about collectives, as opposed to other social groups, that allows them to have these special person-like features? When one starts to think seriously about collectives the first thing one notices is that certain common accounts of social groups will not work. For instance, although 6 Although I do not specifically discuss their work in this paper, I have gained much by reading the work of Gilbert, Tuomela, Bratman, and Searle on the nature of collective intentions and action. My work would not have been possible without theirs, but it differs from theirs in that I start by focusing on large-scale groups and on collective identity rather than on small-scale groups and collective action.

69 On Collective Identity 69 collectives are composed of persons, they are not simply sets or classes of persons. The tennis club may consist at the moment of all and only the members of the Jones family, but the Jones family and the tennis club are not the same collective. The French people of today share no members with the French people of two hundred years ago and yet we say that the French people have persisted through time. Collectives are also not merely composed of those who socially interact with each other. While members of collectives typically interact, such interaction is neither necessary nor sufficient for the existence of a collective. The members of the Catholic Church have never interacted with most of their fellow members. The members of the gay community may interact with and form alliances with members of the Democratic Party, but they do not thereby form a single collective. I want to suggest a criterion for collective-hood that is similar to Lynn Rudder Baker s (2000) criterion for personhood. According to Baker, a person is an entity that has the capacity for a first person perspective. This first person perspective is a perspective from which one thinks of oneself as an individual facing the world, as a subject distinct from everything else (60). It is this concept of oneself as a subject, according to Baker, that makes persons rational and moral agents. Our ability to think about ourselves allows us to reflect on and evaluate our desires, attitudes, and goals and, thus, allows us to revise, prioritize, and plan. While Baker does not use the term, we can say that beings with a first person perspective have the capacity for autonomy the ability to reflect on and determine their conception of the good and plan of life. 7 The suggestion that I want to make is that, in order for a number of persons to form a collective, they must adopt a first person plural perspective. This perspective is typically expressed by the first person plural pronoun we. 8 To echo Baker, this first person plural perspective is a perspective from which we think of ourselves as a group facing the world, as a collective subject distinct from everything else. 9 And, once we have this perspective, just as I can think about myself reflect on my desires, determine my goals, make plans so 7 For a discussion of this conception of autonomy, see Rawls (1971). 8 For a discussion of how collective intentions are expressed in this we mode see Tuomela (1995). 9 Studies in social identity provide support for the claim that individuals can think in terms of their group membership in this way. These studies show how individuals conception of themselves as group members change their behavior and lead them to reason and act in terms of the group identity, as opposed to their individual identities. See e.g., Brewer and Miller (1996).

70 70 Kay Mathiesen too I can, with others, think about us reflect on our desires, determine our goals, make our plans. 10 However, since we expresses a plural perspective, I recognize that these goals etc. are up to us, not just me. With regard to the collective s goals, intentions, and actions we are collectively, but not individually autonomous. 11 Of course, there are numerous differences between persons and collectives. First, and most importantly for an account of collective identity, is the fact that collectives do not have a single unified consciousness. 12 Each of us knows that I am not the collective, because it is composed of more than just me;, it is us. But who is this us? I do not have a direct first person experience of us, the way I do of myself. 13 Thus, in order to form a collective with others the members of the collective have to share 14 a collective self-concept that defines the collective for the members. This collective self-concept allows members of a collective to think and act in terms of their membership in the collective by giving a particular content to the identity of that collective. It provides a picture of who we are so that the members can have something specific, and in common, in mind when they take the perspective of the collective. Second, since collectives lack a further feature of personhood that Baker discusses, interiority, it is only through communication between members that our collective self-concept and associated beliefs, goals, etc. can be reflected upon, changed, or implemented. Typically, the collective self-concept of robust, long-term collectives will include norms, values, and goals. It may include factual claims about relations to pieces of land, bloodlines, historical events, etc. It is important to note that much of this shared self-concept will not just be in the heads of 10 If one wants to know how we can think in this we mode, I do not have much more to say here than to echo Searle (1995) that it is a natural capacity of ours. I would argue that it is no more (or less) mysterious than how we can think in the I mode. 11 Here I disagree with Margaret Gilbert (1989) who argues that persons form plural subjects via making joint commitments. I do not think such joint commitments are necessary; people can simply take up the first-person plural perspective along with others. Indeed, I think doing so leads to commitments, rather than vice-versa. 12 To put it in Baker s terminology, persons have both the unity of consciousness and the consciousness of unity. Collectives, on the other hand, only have the consciousness of unity (via the conscious states of their members). 13 In this way references to the collective self may differ from those to the individual self. A number of philosophers argue that I cannot be analyzed in favor of any non-indexical terms (see Brook and DeVidi (2001)). 14 The significance of shared here is that there is common knowledge among the members that these properties define the collective identity. (See Lewis (1969), Gilbert (1989), and Tuomela (2001) for discussions of common knowledge)

71 On Collective Identity 71 the individual members. Many aspects of the collective self-concept will be embodied and expressed in texts, symbols, works of art, and architecture. The collective self-concept may also include unstated and non-formulated, but distinctive, ways of life which are expressed through language, slang, distinctive foods, dress, customs, etc. It is often through these practices and traditions that the collective self-concept is enacted, reaffirmed, shared, and transformed. Part of this shared self-concept will include some sense of who counts as a member of the collective. 15 This will typically follow rather naturally from the collective self-concept. So, for example, if our collective self-concept includes an important attachment to a particular piece of land, then the membership criteria will typically include being born in or living on this land. In fact, there is some empirical evidence that members of collectives do share membership criteria that are linked to a conception of who they are as a group. A recent study on Scottish identity found a number of markers of identity (e.g., language, place of birth, family, community involvement, etc.) that were accepted as relevant by wide array of people. 16 They also found particular patterns of identification of self and others, and a shared awareness that such identifications were dependent (in part) on whether other members were willing to accept them. To sum up the criteria for collectivity, some group of persons G form a collective if and only if members of G take the first person plural perspective. 17 Members of G can only take this perspective if they have a shared collective self-concept, including criteria for membership in G. These criteria for collectivity allow for a large range of variation among collectives. What is relevant to membership, for example, can range from merely choosing to join, to passing particular tests, to requirements that one be descended from one of the original members, etc. Further, collectives may vary in their degree of cohesiveness the degree of agreement on the content of the identity among the members may range from near unanimity on specific concepts (more typical in small, tightly knit col- 15 Such membership criteria allow the members to define the boundaries of their group. Collectives have a stronger dependence on the creation of such boundaries than do persons, because there is no natural body of a collective. 16 See Kiely et al (2001). 17 Of course, not all members of the collective will have to take the collective perspective, nor will any one necessarily take it all of the time. Ultimately a more developed account will have something more to say about by whom and when such a perspective must be taken. Here I just leave it at the annoyingly vague some members some of the time.

72 72 Kay Mathiesen lectives) to very general overlapping concepts, with room for multiple interpretations. 18 II. Objections to the First Person Plural Account A number of objections might be made to this account. First, one might be worried that the account is circular. If persons form a collective by thinking that they are part of the collective, then doesn t this require that the collective must exist before they can claim membership in it? But, in fact, there is no problem here. Although the members of most collectives are simply born into them and are taught to see themselves as members of the collective, we can easily see how such collectives could start from scratch. The collective may start off based on some objectively shared characteristic or circumstances. The members may all recognize that each of them individually shares some important property with the others. Think of Marx and Engel s call to the workers of the world to unite. One way to understand the charge to the workers is as a call for them to form a collective, by recognizing that they are members of an oppressed group and using this recognition as the basis for adopting the first person plural perspective. Second, it might be objected that I am assuming that there will be unanimity within the group concerning the collective self-concept. But, most collective identities are contested or there are minorities within groups who accept some of the contents of the collective self-concept, but not all. Clearly, groups will not have some sort of perfect unanimity. I might point out that neither does any biological system work perfectly, but when we are providing a definition, we describe it that way. So, the description of collectivity is similarly idealized in order for us to grasp the essential feature of the shared collective self-concept. This is not to imply that it may not be a good thing, in particular cases, if there is conflict within the collective. Such conflict may lead to positive change and may help the collective better reflect the interests of individual members. 19 In fact, as I will discuss in the section on 18 For a discussion of this idea, see Rawls (1982) on the idea of an overlapping consensus. 19 Of course, a complete account of collective identity would ideally provide a more detailed account of the degree of agreement between members that is necessary in order to form a collective. Here I am simply claiming that there must be some shared collective self-concept. Future work on collectivity would hopefully provide more precise criteria.

73 On Collective Identity 73 diachronic identity I make room for the collective self-concept to change over time. Third, one might ask why I do not base the account of collectives on objective facts about social groups, rather than on possibly false subjective opinions of the members of collective. Indeed, it is important to emphasize that, while the beliefs of the members of the collective constitute and individuate the collective, the contents of these beliefs do not determine what is true about the collective. Our belief that the war we lost was in service of a noble cause may be central to our identity, but nonetheless false. 20 A false belief, however, still has the capacity to bind us together as a collective and to play a crucial role in our goals, values, and projects. The situation is not so different from that of individual s with false beliefs about themselves. To take an extreme case, think of a person (not Napoleon) whose self-concept is that he is Napoleon. The belief that he is Napoleon may form the core of his sense of identity and form the basis for his values, plans goals, etc. Thus, while he is not Napoleon, his belief that he is central to his identity. Fourth, some might object to the claim that membership is based on the features that members believe are important to membership. They might argue that such an account is too subjective and that it could be used to justify an unfair exclusion of persons from the collective. On your account, the objector might say, a person, who contributes to the life of the collective, thinks of himself as a member, and is closely allied to the collective identity, may not meet the collectively held membership criteria and, thus, would not be a member. In the film Sunshine, for example, Sonnenshein, a Hungarian Jew whose family had lived in Hungary for over a hundred years, who had converted to Catholicism, and who had won the Olympic gold medal in fencing for Hungary, is put in a concentration camp and is tortured for maintaining that he is indeed Hungarian. It seems that we would want to uphold Sonnenshein s claim to being Hungarian, even if all of his fellow Hungarians would reject his claim. So, the objector might maintain, whether one is a member of a particular collective is not simply up to the shared beliefs of the members of the collective. I have two responses to this. First, our reaction to the above case may be due to the clear injustice of excluding Sonnenshein from membership. It is important to note, however, that shared membership criteria need not be reasonable or just. We can criticize collectives for having unjust membership 20 This seems to be the situation of American southerners, who argue that the Civil War was in defense of states rights, not slavery.

74 74 Kay Mathiesen criteria, but it does not follow that they fail to be collectives, nor does it follow that their membership criteria do not determine who is a member. This is not to say, however, that it is impossible to criticize a collective s membership criteria. The Sonnenshein case shows that when collectives exist over any period of time they come to take on certain objective properties they will have a particular history and persons will have contributed more or less to the functioning and success of the collective. Thus, there can be a conflict between one s status as a member in terms of these objective features of the collective and one s status in terms of the shared beliefs about membership. In short, while in order to be a collective a group of persons must have a shared collective self-concept, as I noted above, this self-concept does not determine what is true about that collective. 21 Fifth, some might argue that I am emphasizing a non-essential aspect of collectives. Members of collectives may share a first person plural perspective, but this is not what makes some group a collective. Nor, it might be argued, does it provide the key to understanding the beliefs, intentions, or actions of the collective. The objector might argue that whether some number of persons form a collective is completely a matter of whether the behavior of the group can be consistently understood in intentionalistic terms. According to this view, we can see various patterns of behavior among members of a group based on the ways that they interact with each other. If these patterns form a kind of complex functional system that produces adaptive behavior, then these persons form a group with intentions. 22 In response to this objection I would point out that merely showing that one can provide an intentionalistic account of a group s behavior, does not show that this account is correct. It is important that we distinguish groups with a genuine capacity to a make plans, form intentions, and perform actions from those groups that we can describe as-if they were doing so. We can similarly describe the behavior of genes, ants, and rivers in intentionalistic terms, but it follows neither that these entities have intentional states, nor that the explanation is correct in terms of the goals it attributes to the entity. Furthermore, such merely as-if intentions do not 21 I discuss these issues in more detail below in the section on synchronic and diachronic identity. 22 This is reminiscent of Durkheim. Ffor more recent defenses of this approach see Jones (2001) and Tollefsen (2002). According to Tollefsen, we can ascribe to the organization those beliefs [and desires] that an organization ought to have if it wants to achieve its goals efficiently given its function in the environment These attributions are then the basis for explanation and prediction of the actions of organizations (402).

75 On Collective Identity 75 provide the reflective tools necessary for the collective, via its members, to be autonomous in the sense of reflecting on itself and determining its future. As I noted at the beginning of the paper, this is something that collectives do all of the time, and the as-if intentions account cannot explain this capacity of collectives. III. Particular Identity: Synchronic and Diachronic Above I provided conditions for when a number of persons constitute a collective, but this is only the first step in an account of collective identity distinguishing collectives from other kinds of social groups. The second and third steps will allow us to distinguish one collective from another (synchronic identity) and track a collective through time (diachronic identity). Below, I start with an account of synchronic identity; this discussion will lay some of the conceptual groundwork necessary for the account of diachronic identity. However, as we will see below, given the importance of causal history in determining the synchronic identity of a collective, ultimately both aspects of identity are interdependent. A. Synchronic There are a couple of features of collectives that make providing an account of collective identity, even in the synchronic case, particularly tricky. First, providing an account of synchronic identity is more difficult for collectives than for many other entities, because collectives cannot be uniquely identified by their spatio-temporal location. In some cases the location of a collective is determined by the location of its members. 23 However, the very same set of persons may form two distinct collectives think of a running club the members of which are currently all and only the members of the local chapter 23 In some cases the spatio-temporal location of the collective may also be (partly) determined by other objects (e.g., by land, texts, artifacts, etc.) Of course, two distinct collectives may also claim the same texts, pieces of land, etc. For example, there are some concerns at present that the Anglican Church may have a schism due to the proposal by the United States Episcopal Church to approve an openly gay Bishop. If such a schism occurred, then the two new collectives (distinguished from each other by their differing views about the morality of homosexual behavior) will claim the same texts, pieces of land, etc. as physical embodiments of the collective.

76 76 Kay Mathiesen of PETA. 24 Thus, two distinct collectives may occupy the same spatio-temporal location. As a result, we cannot provide identity conditions for collectives that rely on a unique spatio-temporal location. Second, collectives are composed of persons and thus have both a physical and a mental aspect. Thus, we run into the same problems providing an account of collective identity as we do with an account of personal identity. Below, I discuss some of the traditional problems philosophers run into when providing an account of personal identity. I then use the insights gained from this discussion to provide an account of the constitution of the identity of collectives, which will allow us to provide synchronic identity conditions that do not rely on spatio-temporal location. Given that collectives are composed of persons, there is something to be learned about collective identity from accounts of personal identity. Since persons appear to have two components, the mental and the physical, there has been a debate about which is more essential to personal identity some propose a psychological criterion of identity and some a physical criterion. 25 Related to this there are two senses of identity the internal (or subjective) and the external (or objective). There is the person s own subjective sense of identity, the sense that she has of who she is and her sense that she is the same person over time. There is the objective sense of a person s identity, which includes objective facts about the physical continuity of her body, her behavioral traits, her history, etc. Internal and external senses of identity do not always coincide. Someone may be objectively appear to be the same person she has the same body, the same voice, the same habits, talents, and personality traits even if she is in the grip of a serious brain disease that makes her lose her sense of continuity or her sense of a single, separate, personal identity. Collectives, similarly, have both an internal and an external aspect, which, I will argue, jointly determine their identity. The internal criterion for collective identity is provided by the shared collective self-concept described above. Thus, if the same group of people has two distinct collective selfconcepts, then they form two distinct collectives. So, we can distinguish the running club from the PETA chapter, because the members attach different collective self-concepts to these collectives. The internal criterion alone is not sufficient to uniquely identify collectives, however (as we will see below). 24 People for the Ethical Treatment of Animals 25 For arguments for a psychological criterion, see e.g., Locke (1967), and, more recently, Parfit (1984). For arguments for a physical criterion see e.g., Olson (1997).

77 On Collective Identity 77 Thus, a further external criterion is necessary. The external criterion is provided by the causal history of the collective. Each collective will have a particular origin and history, thus, if a current collective does not have the same history as another collective, then it is not the same, no matter how closely they resemble each other in terms of their collective self-concept. This rules out the possibility that we could have two identical collectives simply in virtue of their having the same self-concept. To repeat, the identity of a collective is determined by (1) the shared collective self-concept and (2) the causal history of the collective. There are a couple of peculiar features of collective identity that should be noted. First, given that one of the criteria is based on the beliefs of the members (including beliefs about their history), there may be cases where these beliefs do not reflect the actual causal history of the collective. Indeed, there may be cases where the collective s beliefs about their origin are inaccurate or even completely false. Thus, it could be that we think that we are the descendents of Mildred, and we think that this is what makes us the unique collective that we are, while, in fact, we are NOT the descendents of Mildred (maybe there never was such a person). We are factually wrong about what defines us; it is not that we are the descendents of Mildred, but that we believe that we are (plus whatever is our actual history). 26 Such cases are probably quite common, but they do not provide counter-examples for my account. Second, although the collective self-concept is in some sense subjective, what is consistent with or follows from that self-concept is not. In other words, once members have adopted a shared self-concept, the content of this self-concept has certain objective consequences over which the members may not have complete knowledge or control. Given this, the members of a collective may not necessarily be aware of or agree with what follows from their self-concept. The members, for instance, may fail to know that some action is consistent with their values and goals. Thus, for example, they may fail to acknowledge that their belief in the equality of all human beings is inconsistent with their unjust treatment of those outside of the collective It is an interesting issue how the members of such a collective may react if they learn the truth about their origins. One response would be to say that the collective they thought existed (the descendents of Mildred) never did. Another would be to say that the collective existed, but their tie is to the ideals embodied in the idea of Mildred instead of her actual physical existence. 27 Of course, it may depend on which belief the group finds more important they may choose to give up the belief in blanket equality

78 78 Kay Mathiesen It might be objected that the above account fails to take into account the fact that those outside of the collective are often the ones to determine the identity and membership criteria. American Indians for example, were not a collective by their own lights the members saw themselves not as Indians but as Cherokee, Tohono O Odham, Chumash, etc. 28 And, criteria for inclusion in these groups are often legislated by those outside of the group think of the one drop rule adopted by some States in the U.S. according to which a person is black if he has one drop of African blood in him. This objection makes an important point. Members of such imposed collectives, and those in the society around them, may not believe that they have the power to reject membership in particular collectives. One cannot simply opt out of one s social class or ethnic identity and, if these define important social categories in terms of power and status, then they will have important consequences for the lives of people who inhabit them. Thus, it is often in the best interests of such persons to form a collective on the basis of these categories in order to change the status of their group. Nevertheless, if no one accepted and acted in terms of this externally imposed class or ethnic identity, there would be no collective. So far I have provided an account of what makes each collective the unique particular collective it is. It is important to note, however, that part of the definition of synchronic identity relies on a diachronic notion i.e., the causal history of the collective. In order to trace some history as the history of a particular collective, we will need to be able to track a collective through time. This requires an account of diachronic identity. B. Diachronic I approach the project of giving criteria for diachronic identity for collectives with some trepidation. There are notorious difficulties with providing conditions for the identity of objects over time. The first thing to note about the persistence of collectives over time is that a collective can persist through changes in its members. This makes sense, given the fact that the identity of the collective is not necessarily tied to the identities of the members, but to the distinctive characteristics of the collective itself and its causal history. Thus, the persistence of the collective is dependent on there being persons 28 Similarly blacks, the Irish, etc. did not originally see themselves as forming distinct collectives, but as members of multiple tribes, clans, villages, families, etc.

79 On Collective Identity 79 with appropriate attitudes and beliefs, but not on any particular persons. 29 In this way a collective resembles other composite continuants such as bodies, artifacts, etc. Typically the difficult issues that arise when trying to provide an account of the identity of such composite continuants are explicated by means of the story of the Ship of Theseus. According to the standard story, the Ship of Theseus was repaired over and over again, until all of the parts of the ship were replaced with new ones and no original piece of the ship is left (call the later ship with all new parts, Ship of Theseus*). In this case, most people have the intuition that the Ship of Theseus has persisted through these changes in its parts. In other words, Ship of Theseus is identical to Ship of Theseus*. Yet, if the ship with all of its original parts was completely destroyed and the wood that would have been used to fix it was used to design a new ship (Ship of Theseus**) on the same plan as the Ship of Theseus, most people s intuitions tell them that the Ship of Theseus** is merely a copy and not identical to the Ship of Theseus. Thus, it seems that a certain kind of continuity in this case that parts are changed over time while the old planks are there is necessary for the maintenance of identity. It is difficult, however, to specify precisely what kind of continuity is necessary. If we add another twist on the story and add that the old planks were saved and used to build another Ship of Theseus*** then both Ship of Theseus*** and Ship of Theseus* have some continuity with the Ship of Theseus. One has the same parts as the original ship; the other has functional continuity over time. We may be tempted to say that the Ship of Theseus is identical to Ship of Theseus* and that the Ship of Theseus is identical to Ship of Theseus***, but Ship of Theseus* is not identical to Ship of Theseus***, but then this fails transitivity, which is typically thought to be essential to identity. 30 We can avoid some of these problems by loosening what we will require for a collective at t 2 to be the same collective as one that existed at t 1. We can use a very loose notion of identity according to which sameness is a matter of degree (i.e., entities may be more or less the same as the original), is contextual (i.e., whether some entity is the same as an earlier one will partly depend 29 Some collectives may have a self-concept that includes the mention of particular members, in which case the collective may not be able to persist through changes in membership. Marriage is like this; if one partner dies, and the survivor remarries, this creates a new marriage. Compare two people who form a business partnership, if one partner dies, and the survivor gets a new partner, this will not necessarily create a new business. 30 There are, of course, numerous suggestions for how to deal with these problems, but I will not discuss them here.

80 80 Kay Mathiesen on what the other candidates are), and which allows branching (i.e., two entities at a later time may have equal claim to being the continuant of a single entity). This loosening of the requirements for identity are reasonable in the case of collectives; it is not to be expected that there will be some simple set of necessary and sufficient conditions for the persistence of collective identity over time. While the above loose notion of identity obviously does not capture the notion of logical identity (which requires transitivity), it has affinities with a number of views on personal identity. 31 Derek Parfit, for example, has famously argued that the relation between selves at different times is not that of identity, but that of successor and predecessor. And, according to Robert Nozick, To be something later is to be its closest continuer (1992, 33). 32 On this view whether some entity is you depends on whether there are other closer continuants in existence. We can apply this to collectives so that whether some collective is the same collective as some previous one may partly depend on if there are other possible continuants of that collective. 33 With these provisos in mind, I suggest the following criteria for diachronic collective identity. Some collective C* at one time (t 2 ) is the same collective as C at an earlier time (t 1 ) when it meets the following conditions: 1. Shared Content: The content of the identity at t 2 is part of a chain of overlapping contents that extend to t 1, where, at each point, the changes have taken place in a legitimate* way. (Legitimacy* is determined partly by the shared standards of the members.) 31 Whether or not one thinks this is right for individuals, there are good reasons to think that it is right for collectives indeed both Parfit (1984) and Nozick (1981) use collectives as examples to show that logical identity is not necessary. 32 Nozick (1981) provides the following formal definition of the closest successor : y at t2 is the same person as x at t1, only if, first, y s properties at t2 stem from, grow out of, are causally dependent on x s properties at t1 and second, there is not other z at t2 that stands in a closer (or as close) relationship to x at t1 then y at t2 does (37). 33 To illustrate his closest continuer view Nozick gives a particularly nice example. He asks us to suppose that three of the twenty members of the Vienna Circle, after fleeing Nazi Austria and Germany, settle in Istanbul and continue meeting and having philosophical discussions. Believing that the other seventeen have died, they proclaim themselves the Vienna Circle in exile. But, suppose that they learn that nine members (including the acknowledged central members ) have been meeting in America and continuing their philosophical program there? Then the American group is the real Vienna Circle and the Istanbul group is merely an offshoot. Thus, Nozick concludes, Whether or not a particular group constitutes the Vienna Circle depends on what other groups there actually are (32-33).

81 On Collective Identity Membership criteria: Members at t 2 meet the membership requirements held by members of the collective at t 1, or such membership requirements have changed in ways that are legitimate*. 3. Causal connection: There is some appropriate* causal chain that connects collective at t 1 to the collective at t (Appropriateness* will be partly determined by the shared standards of the members. Typically, appropriate causal chains will include overlapping members and passing on of collective self-concept via stories, texts, cultural traditions, etc.) Common true belief: It is common knowledge among the members that 1-3 are met. The fundamental idea behind the above conditions is that what makes some collective the same over time are (a) the maintenance of the content of its identity and (b) that it is part of an appropriate causal chain that stretches back to the earlier collective. On this view, the shared self-concept may change, but changes must be consistent with core elements and must make sense to the members as a natural outgrowth of their identity. The shared content condition (1) rules out the possibility that two collectives are the same when they have radically different collective self-concepts, which are not the result of some evolution accepted by the members. If, for example, the members of the liberal organization NOW (the National Organization of Women), all abruptly adopted the beliefs of the Eagle Forum (a conservative woman s group), without being able to explain how these new beliefs are appropriate for NOW, they would no longer be the same collective and would be NOW in name only. If, on the other hand, this change in beliefs were the result of a slow process of change that was accepted by the members along the way, then it would be less clear that the conservative NOW at t 2 is not the same collective as the liberal NOW at t I would like to thank Catherine Kentfield for pointing out the importance of this condition to me. 35 The meaning of legitimate and appropriate are left somewhat vague. There are some particular difficulties with trying to specify these conditions more strictly. In particular, what counts as legitimate may widely vary among groups, thus we must be very careful in providing too precise a set of criteria for legitimacy or appropriateness. For example, we would not want to require unanimous or even majority acceptance, because such a requirement may not be consistent with the collective self-concept of the group. Further work on collective identity should provide a more precise picture of the ways in which changes may be legitimate or appropriate. 36 If, on the other hand, NOW became a swimming club, it is unclear that there is any

82 82 Kay Mathiesen Sometimes, but not always, the collective may have explicit criteria for how such changes may be made. Think, for example, of the U.S. constitution s criteria for how the U.S. citizens can change the constitution. The membership criteria condition (2) similarly rules out incongruous and abrupt change. In order for some set of persons to count as members of a continuing collective, there must be some criteria whereby they can be said to count as members of that collective. The membership criteria can and surely will change over time. However, these changes must be in line with the content of the identity of the collective, and must be recognized by the members as legitimate. Thus, just because some person y at t 3 does not meet the criteria that were in place at t 1, does not mean that she is not a member of the same collective, as long as some collectively acceptable principle was used at t 2 to change the membership criteria to include y. The causal connection condition (3) rules out the case where two collectives are the same collective, because one has copied the other. Just because I and my friends want to revive the traditions of Fifth Century Athens and I read all the historical material and we manage to replicate the self-concept, practices, etc., of Fifth Century Athens (assuming this would be possible), it does not mean that my group is the continuation of the collective that existed in Athens in the Fifth Century BCE. Copying is not an appropriate causal connection. There must actually be some chain that connects the current collective to its past stages. Typically, the members will overlap; so that some members at t 2 were members a t 1 and some members at t 3 were members at t I was tempted to add the following memory requirement: Members at t 2 see themselves as carrying on the collective that existed at t 1 and recognize the past stages, including t 1, of the collective as constituting their collective past. This is much like what John Locke (1967) suggested as the criterion of personal identity. A set of collectively shared memories allows members to construct a shared narrative of the collective that provides them with a sense of unity between past and present stages of the collective, as well as a sense of solidarity among the current members of the collective (insofar as it is a story which they all share). Indeed, this may be one of the reasons that members of story that would be able to plausibly link NOW now, with NOW then. 37 This does not necessarily have to be the case, however; there may be a collective whose self-concept includes the belief that the collective skips a generation, so that it is only once all of the previous group have died that new members can join. In such a case, the previous generation could leave texts, stories, traditions, etc., which embody the collective identity.

83 On Collective Identity 83 collectives are often very concerned with that their members know the official history of their collective. Recently, for example, U. S. President Bush was promoting the teaching of American History in U.S. schools. According to Bush, Ignorance of American history and civics weakens our sense of citizenship (Bush, 2002). I have not added this requirement, however, because it is open to some of the same criticisms as was Locke s memory criterion for personal identity. Especially given the fact that collectives beliefs about their past can be wrong and distorted, I do not want to require that in order for some past action of a collective to count as something they did that they must collectively remember this event as part of their history. Even so, it might still be objected that the other criteria for diachronic identity places too much importance on what the members of the collective believe. Those outside of the collective may have more ability to shape and determine the identity of a collective than my account allows. There are limits, for example, to what other people in the society will accept as identity preserving. Say, for instance, that the members of the black collective (assuming there is such a thing) come to accepted the American Association of Physical Anthropologist s (1996) position that races do not really exist. They might come to believe that what black really refers to is a particular way of life, a culture, etc., and that the color of one s skin is irrelevant to whether or not one is black. However, these beliefs of the collective may not be reflected in changes in the categorization practices of the rest of the society. People in society at large might still distinguish real blacks (the ones with particular black physical markers and heredity) from the honorary ones. The view presented here, however, is that those who claim that the real membership criterion is skin color are simply confusing collectives that have formed as a result of socially significant categories (e.g., those social categories that are (or, are treated as) significant within a particular culture) with the social category itself. It may be that skin color is the way in which the society at large categorizes people and labels those in the category black. It does not follow that the black collective must accept these categorization practices as determining their membership criteria. There may also be cases where those outside the collective force changes in the membership criteria which are not accepted by the members of the collective. It is not clear whether this will always result in changes that fail to be identity preserving. It may depend on how consistent the change is, in fact, with the identity of the group. Take the case of clubs, colleges, etc., who are forced to accept minorities or women as members. In these cases, we might want to argue that the changes in membership criteria are legitimate and the

84 84 Kay Mathiesen collectives remain the same, because, given the primary characteristics of the collective, some of the membership criteria held by the members were ad hoc and irrelevant. If the Rotary is a business organization, then why not admit businesswomen? If the purpose of a country club is primarily to facilitate the playing of golf, then why exclude women or minority golfers? In such cases, what is central to the identity of the collective is not changed, thus, we would probably want to say that the membership claims of these individuals are legitimate and the collective remains the same (but with new members). 38 Of course, if all of the original members were so against the change that they refused to recognize the collective as the same, then there would be a strong argument that the previous collective is gone and a new (more just) one has taken its place. Conclusion I have claimed that there is a particularly important kind of social group, what I call a collective, which has person-like properties and has the capacity for autonomy. I have argued that persons can form such collectives when they take on a first person plural perspective based on a shared collective selfconcept. This collective self-concept defines the collective, and, along with its causal history, uniquely identifies it. Over time collectives persist through changes in membership by maintaining this collective self-concept. Changes in the collective self-concept may be identity preserving, however, if they result from processes naturally flowing from the identity of the collective, which are appropriately caused by (or accepted by) the collective. I would like to conclude by noting that I have not claimed that a collective is something over and above its members. The beliefs, thoughts, actions, desires, values, etc. of collectives are those of individuals who, together, act and think in the first person plural mode. Thus, the ultimate moral importance of the capacity for agency on the part of collectives is the fact that it is one of the ways in which individuals express their capacity for agency. In recognizing this fact we may better understand that treating collectives with respect and concern is one of the ways to treat individuals with respect and concern. And, yet, we should never lose sight of the fact that collectives are morally important because the individuals who compose them are morally important; collectives have no independent moral importance. 38 Of course, in these cases we see the exclusion as unjust, but there is no guarantee that the ontology will always line up with the ethics in this way.

85 On Collective Identity 85 References American Association of Physical Anthropologists. AAPA Statement on Biological Aspects of Race. American Journal of Physical Anthropology 101 (1996): Baker, Lynn Rudder. Persons and Bodies: A Constitution View. Cambridge, U.K.: Cambridge University Press, Bratman, Michael. Shared Intention. Faces of Intention: Selected Essays on Intention and Agency. New York: Cambridge University Press, Brewer, Marilynn B., and Norman Miller. Intergroup Relations. Pacific Grove: Brooks/Cole Publishing, Brook, Andrew, and Richard DeVidi, Editors. Self-Reference and Self-Awareness. Advances in Consciousness Research. Philadelphia: John Benjamins Publishing Company, Bush, George. President Introduces History and Civic Education Initiatives. (September 17, 2002). French, Peter. Collective and Corporate Responsibility. New York: Columbia University Press, Gilbert, Margaret. Social Facts. Princeton: Princeton University Press, Sociality and Responsibility: New Essays in Plural Subject Theory. Oxford: Rowman and Littlefield, Hume, David. A Treatise of Human Nature Ed L. A. Selby-Bigge. Oxford: Oxford University Press, Jones, Todd. What CBS Wants: How Groups Can Have (Difficult to Uncover) Beliefs. The Philosophical Forum 32.3 (2001): Kiely, Richard, et al. The Markers and Rules of Scottish National Identity. The Sociological Review (2001): Kutz, Christopher. Complicity: Ethics and Law for a Collective Age. New York: Cambridge University Press, Lewis, David. Convention: A Philosophical Study. Cambridge, Massachusetts: Cambridge University Press, Locke, John. An Essay Concerning Human Understanding. Ed. A. S. Pringle- Pattison. Oxford: Clarendon Press, Miller, Seumas. Social Action: A Teleological Accept. New York: Cambridge University Press, Nozick, Robert. Philosophical Explanations. Cambridge, Massachusetts: Harvard University Press, Olson, Eric. The Human Animal: Personal Identity Without Psychology. Oxford: Oxford University Press, Parfit, David. Reasons and Persons. Oxford: Oxford University Press, Rawls, John. The Idea of an Overlapping Consensus. Oxford Journal of Legal Studies 7.1 (1982): 1-25.

86 86 Kay Mathiesen Searle, John R. Collective Intentions and Actions. Intentions in Communication. eds Philip R. Cohen, Jerry Morgan, and Martha E. Pollack. Cambridge, Massachusetts: The MIT Press, Shoemaker, Sydney. Personal Identity: A Materialists Account. Personal Identity. Editors Sydney Shoemaker and Swinburne. Oxford: Blackwell, Tollefsen, Deborah. Organizations As True Believers. Journal of Social Philosophy 33.3 (2002): Tuomela, Raimo. Cooperation: a Philosophical Study. Boston: Kluwer, The Importance of Us: A Philosophical Study of Basic Social Notions. Stanford: Stanford University Press, Velleman, D. J. How to Share an Intention. Philosophy and Phenomenological Research LVII.1 (1997):

87 A Collective s Rational Trust in a Collectives Action 87 A Collective s Rational Trust in a Collective s Action Maj Tuomela Abstract In this paper, an account of rational social normative trust (RSNTR) and a context for rational trust (Y) will be offered and briefly argued. The account concerns a person s trust in another person that he will perform a specific action. Rational social normative trust is conceived as the trustor s accepting attitude vis-à-vis his dependence on the trustee. This is an attitude that the trustor acquires non-intentionally, because of his belief, due to their relationship of mutual respect, that he is entitled to expect of the trustee, on social or (quasi-) moral normative grounds, that the trustee will intentionally gratify him by his action, and because of his belief that he will indeed do so at least in part in deference to his rights. The trust context Y involves the conditions for a situation where the trustor can rationally consider whether intentional gratification may be expected. Rational trust is distinguished from the more general technical concept of rational reliance by the presence of context Y and the trustor s expectation of the trustee s intentional gratification of the trustor. The trustor s social normative expectation of the trustee s intentional acting with goodwill towards him is the central belief of the trustor involved in rational social normative trust. This belief does not require the trustor s belief of the trustee s genuine attitude of goodwill towards the trustor. The account distinguishes social normative trust from predictive trust by the trustor s socially grounded normative, as opposed to only predictive, expectation of the trustee s intentional gratification of the trustor. Normative trust, and also predictive trust, are more than an evaluation of a person s trustworthiness. In addition to expecting of the trustee and/or expecting that the trustee will intentionally gratify him, the trustor feels comfortable about being dependent on the trustee, and has an accepting attitude regarding his dependent position. Both in normative and predictive trust, the trustor may decide to depend, or to refrain from depending, on the trustee for an action. In the case of normative trust, the trustor genuinely trusts the person he decides to depend on, but in the case of predictive trust, his trusting is comparable to relying on some features of the trustee or the situation. Deciding to trust is to make a bet on someone to act as if one trusted. The account of rational trust (RSNTR) will be applied to a case where the trustee is a collective agent. Criteria for collective agency is then added. Collective agency is discussed, mainly, in the light of Raimo Tuomela s work. When the trustor is a collective, the criteria for collective agency should be satisfied for the trustor as well.

88 88 Maj Tuomela 1. Introduction People often think of trust as general trust towards a close person, trust being a central feature of closeness. Many accounts of trust include some of the following components: expectations or beliefs about intention, ability, persistence, gratification or unharmfulness, and affect. Trust could be conceived of as an attitude that the trustor has towards the trustee in which the trustor, due to their relationship of mutual respect, normatively expects of the trustee, on social or (quasi-)moral grounds, that he will intentionally gratify the trustor within a wide scope of matters (see Hollis, 1998, 160 for normative trust). Such a relationship of mutual respect is often described as one of goodwill and trust. Some scholars stress the result of the trustor s expectation of being gratified by the trustee: the trustor accepts to be vulnerable before the trustee. The trustor s dependence on the trustee s genuine goodwill is the central feature of thick trust (see e.g. Baier, 1994, 98-9). A thinner view of trust is trust based on a prediction of another person s intentionally gratifying action. The prediction can be grounded e.g. in the trustee s personal features, some situational or institutional circumstances, or his instrumental motives. In some of these cases the expected good-willed behavior is less than genuine. In such predictive trust the trustor could be said merely to be depending on some features of the trustee or the situation and not really to be trusting at all. To hold someone as trustworthy is often confused with trusting someone. It is not only that trusting involves an accepting attitude towards being dependent on the trustee that differentiates the two concepts. Holding someone as trustworthy is often the result of an evaluation of a person s character, his motives or some features of the situation and not a consequence of having a relationship of mutual respect with him. Thus, trusting on the basis of considering someone trustworthy is comparable to depending on the features of the person or the situation, just as is the case in predictive trusting (see Loukola, 1999, for discussion on this topic). Some scholars seem to have a view of trust that allows for a prediction of the trustee s unintended gratification of the trustor (cf. Gambetta, 1988, 217). Then trust seems like a mere prediction of the trustee s future action. In some accounts of trust the affective component of trust is central. Important topics of discussion are the trustee s freedom of choice and trust in situations of certainty. Below, I will address these issues and make some distinctions between trust and its close cognates, arguing that an analysis of a person s ratio-

89 A Collective s Rational Trust in a Collectives Action 89 nal 1 social normative trust in another person that the other will perform an action (revised version of M. Tuomela, 2002 and 2003) is a valid analysis of rational genuine trust. Rational genuine trust is here regarded as rational predictive reliance in a specific trust context when coupled with the trustor s socially grounded normative expectation of the trustee s intentional gratification of the trustor due to their relationship of mutual respect. Because of all this the trustors feels comfortable about his dependence on the trustee, and has an accepting attitude vis-à-vis his dependence on him. Reliance is used as a general technical concept that also covers e.g. predictions about the functional effectiveness and safety of a tool for situations of dependence on it, or about a person s performance of an action that does not concern the relier more than to the extent to which he may happen to have interest in evaluating the situation from the outside. The present view of trust involves the trustor s expectation of the trustee s intentional gratification of the trustor, or of the trustee s acting intentionally with goodwill towards him, be it for non-instrumental or instrumental reasons. In the case of social normative trust these reasons also include the trustor s belief of his acknowledged social right to be gratified by the trustee, due to their relationship of mutual respect. When the trustor believes that the trustee s gratification of him merely is an unintended side effect of the trustee s action, or some third person or nobody at all is believed to be intentionally gratified, the more general term reliance seems to be more appropriate. Rational social (socially or (quasi-) morally grounded) normative trust is distinguished from rational predictive trust by the trustor s expectation of the trustee that the trustee will intentionally act with goodwill, and not only an expectation that the trustee will do so. The preposition of is meant to indicate that the expectation is normative (cf. Hollis use of expect of in normative trust in 1998, p.160). In my account of rational social normative trust, the trustor s belief of his (quasi-) moral or social entitlement to be gratified by the trustee is due to his socially grounded belief that the relation- 1 The account (RSNTR) below concerns subjectively rational trust. Trust is an attitude and is thus subjective. I will argue for the necessity and joint sufficiency of the conditions of (RSNTR) and Y for rational social normative trust. In rational intuitive trust, the beliefs of (RSNTR) and Y may be unreflective. When rational intuitive trust is based on a mere disposition to believe, the trustor forms the beliefs when asked about his trusting. It is claimed that the beliefs should be supported by (articulated or intuitive) subjectively rational reasons. The required rationality of the reasons is taken to concern their joint sufficiency and their bearing on the matter, as seen from a subjective point of view, relative to the trustor s resources and his level of optimism. Intuitive reasons may be unreflective or may be dispositions to have reasons.

90 90 Maj Tuomela ship of mutual respect between himself and the trustee involves such rights in the present situation, and the trustor believes that the trustee will act with goodwill at least in part for this reason. The focus is on the relationship and not on existing norms and norm-obedient behavior in general. The trustor s rights are respected due to the relationship between the parties and certain rights come with certain relationships. Predictive trust is not considered to be genuine trust, and can hardly be said to be trust at all, as the trustor merely predicts another person s action involving intentional gratification of the trustor on the basis of some features of the person or the situation. Either the relationship between the persons does not yield rights, on social grounds, to expect gratification or, if it does, the rights do not figure as a partial reason for the trustee s action. However, in both kinds of trust, the trustor feels comfortable being dependent on the trustee, and on the grounds of his beliefs and feelings, he has an accepting attitude concerning his dependent position. Both in normative and predictive trust, the expected goodwill may be more or less genuine, depending on the case. Here is an example of rational social normative trust with the trustor s (A) normative expectation of the trustee s (B) good-willed action out of instrumental motives. A and B are in an exchange situation and they have a relationship of mutual respect. A expects of B to get the goods that he has paid for, not only due to general norms of decency, but due to their relationship. He has generally accepted social rights to get the goods as well as social rights due to their relationship of mutual respect to have the general rights taken into account. However, the expected good-willed action is not believed to be based on genuine goodwill. A believes that B will honor their relationship of mutual respect, not out of B s genuine respect for A, but in order to have business with him in the future. A s genuine trust here is not the thickest kind of genuine trust. Another example in which A s trust is somewhat thicker is when A believes that B has genuine respect for A s rights, but B lacks genuine goodwill towards A in any other respect - B does not genuinely care for A s welfare other than what his social rights will provide. With both normative and predictive trust the trustor may decide to depend, or to refrain from depending, on the trustee for the action. In the first case, the trustor may decide to refrain from depending on the trustee, despite the fact that he genuinely trusts him. Deciding to trust someone is like making a bet on him acting as if one trusted. When we trust someone, we simply non-intentionally have this attitude, and we do not think we will be betrayed. To find a person trustworthy and to trust in him are distinguished from each other in that finding a person trustworthy consists of beliefs, and trust-

91 A Collective s Rational Trust in a Collectives Action 91 ing, in addition to beliefs, of feelings and an accepting attitude about being dependent on him. In this sense predictive trust is a fuller concept than a mere evaluation, although it is weak in the sense that the trustor depends on some qualities of the trustee or the situation. If A has predictive trust in B concerning x and A decides to depend on B for x, A acts as if he genuinely trusts B. A person s trusting behavior may indicate that he has rational social normative trust in another person, that he predicts that he will accidentally fare well, that he predicts that the other will intentionally treat him well for some reason, that he has decided to take a risk, that he has standing beliefs of which situations are safe, that his default-position in life is to trust everyone until he has reasons not to trust, or that he is subjectively non-rational in his trust in some way, e.g. he trusts B for x, doubting that B can perform x. When trust in someone is genuine, it is not considered to be a risk, and might not be thought of at all. When a person starts considering his trust, it could be a sign of distrust (cf. Lagerspetz, 1998, p 2). My ambition is to give an analysis of rational social normative trust, whether it is reflective or unreflective, and whether it involves beliefs and reasons or just dispositions to form such. The analysis of the concept of trust in this paper is based on the above view of normative trust. The account explicates what rational genuine trust consists of, and the defense concerns both the analysis and the view of trust that it expresses. 2. An Account of Rational Social Normative Trust The account of rational social normative trust (RSNTR) for primary trust relations is presented below, starting with the account of the context for rational trust (Y). The following conditions are necessary and jointly sufficient for a context (Y), the trust context, in which A can rationally trust B that he will perform a specific action x: a) A wants B to perform x, an action that A does not intend to perform himself. A thus believes that he is concerned with B s performing x, and believes that he is in that sense dependent on B s action; b) A has a belief about B s belief of the dependence of A; and c) A believes that his dependence is relevant to B. That the dependence is relevant to B means that B is free to allow A s dependence to

92 92 Maj Tuomela play a role in his decision-making, and that B takes A s dependence into account in his actions. A s beliefs in a) - c) may be dispositions to have such beliefs and they are based on (at least) subjectively rational reasons (or dispositions to have such). (RSNTR) For a person A to have rational social normative trust in another person B that he will perform a specific action x in a situation in which context Y is present, the following necessary and jointly sufficient conditions must hold: 1. Intention condition: A expects (or has the disposition to expect) that B intends to perform x. The expectation is based on (at least) subjectively rational reasons or dispositions to have such. 2. Ability condition: A expects (or has the disposition to expect) that it is possible for B to perform x (in that the internal and external opportunities for B s action obtain). The expectation is based on (at least) subjectively rational reasons or dispositions to have such. 3. Goodwill condition: A expects (or has the disposition to expect) that B will intentionally act with goodwill, including goodwill towards A when performing x. A believes (or has the disposition to believe) that, due to their relationship of mutual respect, he is entitled to expect this of B, on social normative or (quasi-) moral grounds, and that B acts at least in part because A has this right. The expectations (or dispositions) are based on (at least) subjectively rational reasons or dispositions to have such. The reasons for A s expectation in 1 may be (partly) the same as here. 4. On the grounds of 1, 2 and 3, A has a positive feeling. This feeling is or at least includes the feeling of being comfortable about his dependence on B for performing x. 5. On the grounds of 1, 2, 3, and 4, A has an accepting attitude vis-à-vis his position of dependence on B for performing x. This attitude is non-intentionally acquired. The rationality of social normative or genuine trust is defined by (RSNTR) 1-5 and the trust context Y. Acting with goodwill for B involves, firstly, his acting with special commitment to his performance of x. Secondly, it involves his acting intentionally so as to advance A s interests, with a description of interests that at least partially coincides with A s description. This is considered to be the same as B acting with a pro-social attitude, or his being

93 A Collective s Rational Trust in a Collectives Action 93 at least one-sidedly socially committed to A to performing the action. Acting with goodwill accounts for the persistence of B s intention and for his acting with persistence and flexibility in a way that at least does not harm A. It should be noted that the belief that B is acting with goodwill does not mean that B is necessarily believed to be good-willed towards A. To intentionally act with goodwill may be based on e.g. respect, genuine caring, but also on prudential or moral reasons. In rational social normative trust, the trustor has rational reasons to normatively expect of the trustee, on social or (quasi-) moral grounds due to their relationship of mutual respect, that he has a we-perspective or a you-perspective, or more specifically, that A has rational reasons to believe that B has such a relation to A that when B performs the action, he ought to take and will take into account their point of view or A s point of view. As this is how A sees the relationship, he normatively expects of B, on social or (quasi-) moral grounds, that B will intentionally act with goodwill towards A and that B will do so at least in part for this reason. A can be wrong about what their relationship is, about the norms of decency connected to various situations, and about B s awareness of the norms. If A believes that B is not willing to act in the right way, A may think that he is mistaken about their relationship, e.g. friendship or some other relationship of mutual respect. This is how I thought you were my friend - situations may start. If there is an agreement between A and B that B will perform x, and A believes that B will not do so, A will probably not believe that he did not make an agreement. Instead, he might believe that there was no sincere agreement. A may think that B is a person with whom A cannot have a sincere agreement-relation. A is entitled to expect of B the performance of x due to social norms pertaining to agreements, but there are no rational reasons for that expectation which are grounded in a relationship of mutual respect. If B is or strives to become A s friend, A takes this as a rational reason to normatively expect of B, on social normative grounds, that he will intentionally act with goodwill towards A in a specific matter that belongs to the area of matters that calls for a friend s support. Of course, A also predicts what B will do on the basis of the friendship. However, A thinks that he has a right to expect of B that he will act with a you- or a we-perspective, and that B shares this view and will conform. A s belief of B s (future) friendship with A could be based on his belief of B s intrinsic caring for A as a person or as a (future) member of their dyad. Intrinsic caring is convincing evidence of a true friendship.

94 94 Maj Tuomela In a situation where A and B are mutually respectful partners in a joint venture or B strives to become such a partner, A takes this as a rational reason to normatively expect of B, on social normative grounds, that he will act with a you- or a we-perspective. The joint venture between these two mutually respectful partners could also be a reason for A s prediction of B s good-willed behavior. However, in the present case A believes that he has a right to expect of B that he will intentionally act with goodwill towards A in the context of or qualifying for the joint project of mutually respectful partners, and that B shares this view and indeed will act in this way, in part for this reason. A may have additional supporting reasons for his belief that B will act with goodwill such as: B s interest in the joint venture and in having a you-perspective are due to his liking A s company or due to the gain that will come out of acting with A; B s interest in having a we-perspective is due to the good atmosphere or the effective team they make. B may have a you- or a we-perspective in an exchange situation with A, respecting A s right to expect of B that he gratifies A, in return for A s gratification of B. On the other hand, when B lacks a we-perspective or he does not respect A, A s social rights to receive what he has paid for are acknowledged in common norms of decency, but A lacks rational reasons to normatively expect of B that he will act as he should. The normative expectation of A should be grounded in A and B s relationship of mutual respect and not only in common norms of decency. People may have temporary relationships of mutual respect, backed by prudential motives to achieve some goals. E.g. A and B honor the other s right to have what should come to him in order to get a good reputation. In another case the mutual respect may be genuine backed by we-ness-values or other-regarding values. Here the motive to act with goodwill may still be only prudential. E.g. A and B genuinely respect each other s rights, but they act with goodwill towards each other in order to achieve their private goals A and B do not genuinely wish to further the other s welfare to a higher extent than giving the other what his right requires. When B s motives for acting with goodwill towards A are mainly prudential and only weakly social in the sense of being other-regarding, A s reasons for his socially grounded normative expectation of B s gratification are weak, and so are his reasons for believing in the stability of B s motives. B may acquire other preferences and do something else. The general social ought may still be present, but the relationship of mutual respect is only weakly socially grounded. Thus, the social ought may become a dead letter in a situation of conflicting preferences where the trustee is not sufficiently social

95 A Collective s Rational Trust in a Collectives Action 95 to be responsive to the trustor s rights from other than prudential reasons. It is the quality (viz., sincerity) of the relationship of mutual respect that is the social ground for A s normative expectation and not only the norms applying to a certain situation and relationship. Thus, genuine trust (social normative trust) may be thick or thin. Cases of rational predictive reliance or prediction which do not qualify as rational social normative trust or predictive trust, are analyzed as in (RSNTR) with the following alternations: One or several of the conditions in Y are lacking or Y concerns B and a third person C, A having the beliefs; (RSNTR) 4 and 5 are lacking; the goodwill-expectation concerns C or nobody at all and no socially grounded normative expectations are required. Rational predictive reliance is comparable to prediction of another person s action. The action may be accidentally beneficial to the relier, who may be dependent on the action. When it is intended to be beneficial to a third person or to the actor himself, and the relier is unconcerned with the action, he is an outsider and has only sufficient interest to evaluate the situation. Secondary trust relations will not be considered here. The trust account (RSNTR), above, concerns rational social normative trust, which is regarded as genuine trust, as opposed to rational predictive trust. The latter kind of trust is analyzed as in (RSNTR) with the exception of the socially or (quasi-) morally grounded normative expectation. In rational predictive trust A only expects that B will perform x, intentionally gratifying A, while still having the beliefs indicated in (RSNTR) 1 3 (without the entitlement belief), but in rational social normative trust A also expects of B that he will do so, partly for the reason that A has a right to be gratified. Thus, in predictive trust A may predict that B will intentionally gratify him by performing x, feel comfortable about being dependent on him, and have an accepting attitude regarding a dependent position, e.g. due to B s moral character, fellow-love, or his fear of legal sanctions. Genuine trust requires more, which is expressed in (RSNTR) 3 by A s socially or (quasi-) morally grounded normative expectation, due to the relationship between A and B, and A s belief that B will act in part for this reason. (See also Hollis, 1998, p. 160 for normative and predictive trust; expecting of and that.) In predictive trust the trustor merely depends on features of the other person or the situation. A does not have a normative expectation of being gratified by B. Either A does not have a relationship of mutual respect with B or A does not think of the matter in terms of B s responsiveness to A s social rights. E.g. A predicts that their friendship will be a reason for B acting with goodwill towards A. Here the difference between social normative trust and

96 96 Maj Tuomela predictive trust is only a matter of A s at least partial focusing on his social rights as a friend, and not only on the friendship as a caring relation. Friends can normatively expect good-willed actions from each other and persons who like each other can predict that the other will treat them well. Liking someone and treating him badly would be inconsistent, but being friends with someone whom you treat badly would be socially or quasi-morally wrong. To hold someone as trustworthy, in a normative or predictive sense, vis-à-vis the performance of a specific action, is to have the beliefs in (RSNTR) 1-3, with or respectively without the normative expectation. For trust we need to add (RSNTR) 4-5 in order to include the affective component and the attitude of acceptance of a vulnerable position before the other. A may decide to trust B for x on the basis of considering B trustworthy. Then A acts as if he trusted B. Trusting someone is not something one decides to do. It is an attitude that one comes to have as being in love. Deciding to trust someone is more like betting on him acting as if one trusted. Trusting is non-intentional either we trust or we do not, and when we trust we do not think we will be betrayed. Both in predictive and social normative trust we may decide to depend on another for an action, or refrain from doing so. When A decides to depend on B for x in predictive trust, A s decision is based on a very thin kind of trust, hardly deserving its name. This kind of trust is merely reliance on the qualities of a person or the situation. When A has genuine (social normative) trust in B, and decides to refrain from depending on B for x, he actually trusts B, even though he chooses to e.g. perform the action himself. 3. A Defense of the Account The analysis draws a line between trust and a broader concept of reliance through the trust context (Y). Y does its work with a view of trust that involves the trustor s (A) expectation of the trustee s (B) intentional gratification of A. When one or several of the conditions for Y are lacking between A and B, A can only consider whether or not to have predictive reliance that B will perform x. If a third person (C) is dependent on B in the sense of Y, A may take into consideration B s intentional gratification of C. If A believes that he is dependent on B s action, but the other clauses of Y are unsatisfied, he may expect B s unintentional gratification of himself as a side effect. In my opinion these cases do not deserve the label of trust, but fit only the more general

97 A Collective s Rational Trust in a Collectives Action 97 notion of predictive reliance. Trust is an attitude that presupposes a relationship between the parties and not merely a person s expectation of another person s accidental gratification. To expect that someone s action will be beneficial to oneself by accident is just part of predicting the consequences of the action. One could argue that the term reliance is not applicable to neutral predictions about the relier s action affecting a third party or nobody at all, as the term indicates some dependence on the expected action. When the action concerns only a third person in the sense of Y a) or nobody at all, the relier is concerned with the action only to the extent to which he bothers to evaluate the situation from an outsider s position, but he is not concerned in the sense of Y a). The use of reliance as a general term can be defended through its wide usage as an expression of one s views of future events and the safety or functional effectiveness of objects, or of trusting a person. Although reliance is sometimes used as a synonym for trust, it will here mainly be a technical term for a prediction-like evaluation. In so far as B s action concerns A in the sense of Y a), but Y b) and/or Y c) are not satisfied, then B s possible gratification of A cannot be believed by A to be intentional. Thus, predictive reliance is the weakest concept here, allowing for only the accidental gratification of the relier. In the example: I trust that the stranger will take care of his mother Y a) is not satisfied between A and B. Y b) is unsatisfied between A and B in I trust that he will save me, although he cannot know that my life is in his hands, and Y c), in I trust that the bus driver will show me the way downtown, as I follow the bus with my car. In the last example, the bus driver may have an awareness of my dependence, but it is not relevant to him. He cannot take it into account in his actions and he thus lacks the freedom to disappoint me. Pettit (1995, 204), whose example this is, does not seem to think that the trustee s lack of freedom of choice disqualifies this example as a case of trust. To Gambetta trust belongs to a context of uncertainty (Gambetta 1988, 218). I agree with him that there is uncertainty, but only in the sense of Y c): the trustee is believed to be taking the trustor s dependence into account in his decision-making and to be free to disappoint the trustor. This is because I agree with Lagerspetz that we do not consider it taking a risk when we actually trust someone (Lagerspetz 1998, 2). Trust involves a belief that we are in good hands. So, with the reservation of the structural uncertainty due to Y c), I agree with Pettit that trust may figure in contexts of certainty (Pettit 1995). The structural uncertainty seems to be important from a third person point-of-view, but the trustor himself feels secure.

98 98 Maj Tuomela The necessity of A s belief of his dependence on B s action (Y a)) is based on an intuition about the meaning of trust that is shared by many scholars. An unconcerned relier is merely predicting the course of action of the other person. In Y a) A s dependence is defined as meaning that A is concerned with B s action in the sense that A wants B to perform x. It is claimed that it would be irrational for A to trust that B will perform x when A believes that he is unaffected by B s performance. Trusting is an investment, and there should be a cost to A if his trust is betrayed. There is no cost if A is unconcerned and mistaken in his reliance. If B s action is unwanted by A, he is not in a context to consider trust. E.g. I trust that you will do what I fear you to do, does not fit the idea of trust. We do not need a notion of dependence that is stronger than A s want as the trust account involves necessary and jointly sufficient conditions. To require that A has a goal that B performs x is more than is needed here, and also brings us to a controversial topic. Even aim-goals require some disposition towards action on the part of the goalholder. A trustor need not have that disposition (cf. Falcone and Castelfranchi, 1998 on a trust account stressing B s goal adoption). Y b) is needed, as B might unintentionally harm A, although he is goodwilled towards A, if B lacks a belief in A s dependence. Y c) is necessary, as argued in the example of the bus. Gambetta also includes the first part of Y c) among the typical features relevant for a context of trust (Gambetta 1988, 219). In the present account trust is seen as a special kind of reliance that arises when the trust context is present and the relier has a positive feeling and an accepting attitude about being dependent on the person relied upon, on the grounds of (RSNTR) 1-3. Reviewing Gambetta s suggestions for additional conditions for a context of trust (ibid, 219), and various examples in which trust is considered, the three necessary clauses of Y seem to be jointly sufficient for a context of rational trust. (RSNTR) 1, the Intention condition, is needed for rational trust, as A must expect that B intends to perform x to be able rationally to trust that B will perform x. In this case A s evaluation of B s confidence in his own ability plays a part. A cannot expect that B intends to act, if he believes that B lacks a belief in his own ability. 1 (RSNTR) 2, the Ability condition is necessary as A also must expect that it is possible for B to perform x, in order to rationally trust that he will perform x. 1 I owe my awareness of the importance of A s belief of B s self-confidence to the work of Castelfranchi and Falcone (1998, Sec. 3.1). However, this is automatically included in (RSNTR) 1 as A s belief of B s intention must include A s belief of B s belief of his ability.

99 A Collective s Rational Trust in a Collectives Action 99 (RSNTR) 3, the Goodwill condition. Without expecting that B will intentionally act with goodwill, including goodwill towards A when performing x, A cannot trust that B will perform x at all, although he could predict that B is likely to perform x. Firstly, if B s commitment to his performance of x did not include a commitment to A when performing x, he could more easily stop intending to perform x when circumstances change. The social commitment is crucial, as it binds B to the performance of the action with a glue made of strong ingredients. In genuine trust (RSNTR), A also believes that he has a right, due to their relationship of mutual respect involving this right, to expect of B his good-willed performance of x, or B s social commitment to performing x, which will be argued below. A also believes that B acts in part for the reason that A has this right. The relationship of mutual respect may be based on e.g. caring, social and moral decency, sincere agreement, or a situation of reciprocity. The one-sided social commitment means that A s acceptance of B s commitment is not required, as in accepting a promise. A simply believes that B ought to be and is socially committed to A, in part because A has this right due to their relationship. Secondly, without goodwill-expectations A cannot believe that B will perform x in a way that avoids harming A. Trust in someone is to believe that he will act without intentional harm. Thirdly, A may think that B could arrange circumstances in a way that excuses him from acting, if A has reasons to believe that B will act without goodwill towards A. A s expectation of B s acting intentionally without ill will is the borderline case for A s expectation of B s good-willed action. The restricted freedom of the trustee is required for rational trust. We do not trust people if we expect that their view of our good differs from our own view, although they are free to advance our interests as they see fit. A goodwill-aspect or some concept of unharmfulness is included in many accounts of trust. In (RSNTR) 4 the positive feeling that accompanies trust is acknowledged. It may vary from warm feelings towards the trustee to feeling comfortable about one s dependence on him for the performance of the action. It is argued that it would be non-rational not to feel good when one has reasons to feel so. In (RSNTR) 5 is established that A is in the trustor s position through his having an accepting attitude vis-à-vis his dependence on B for performance of x. This distinguishes trusting from simply believing and feeling good about believing that someone is trustworthy. However, what ultimately distinguishes a person s trusting someone from his believing in certain features of him and his predicting the person s behavior along with having a comfortable feeling about his position of dependence

100 100 Maj Tuomela and his accepting attitude of such a position, is that in trusting the person he expects good-willed behavior of him and not only that he will act with goodwill, e.g. due to some personal features. The key lies in the relationship of mutual respect. In genuine trust, the relationship has to be such that the trustor believes that he has a socially justified right to expect good-willed behavior of the trustee. He must have rational reasons to believe that their relationship is such that the trustee ought to have, on social or (quasi-) moral grounds, and indeed has, a you-perspective or a we-perspective, e.g. the trustee is or aspires to be a friend, a member of the same group, a respectful party of a joint venture, or just a respectful party in a simple exchange. In these examples the relationship is such that the trustor has a social right to expect varying levels of good-willed behavior of the trustee of various degrees in various situations. Of course, the trustor also needs reasons to believe that the trustee understands their relationship in the same way and acts as he should in part because of the right the trustor has. It is claimed that the social-normative expectation is needed for rational genuine trust, and it is included in the account of rational social normative trust (RSNTR) 3. It should be stressed that the social right is due to the relationship of mutual respect and not only to some norms of decency that apply to the situation in general. Predictive trust need only rely on general norms and the trustee s disposition to be norm-obedient. Then the trustor merely depends on features of the person and the situation. When a person believes that another person will save his life, e.g. due to his moral character or fellow-love, he is simply predicting future events and feels good about being in the hands of the other person. An opposition towards considering predictive trust as trust at all, is widespread in the literature (cf. e.g. Hollis use of the terms normative and predictive trust, 1998, p.160). Of course, every instance of genuine trust also involves a predictive element, but there is more than this to trust. Trust presupposes a relationship of mutual respect in which the trustor believes that he has a social right to expect to be gratified by the trustee and that he will be gratified in part because of this right. When people simply predict what others will do, due to their personal features, warm feelings, group memberships, joint ventures etc., and have an accepting attitude about a dependent position, we have predictive trust. However, some of those reasons may figure also in normative trust. Then the relationship is in focus as well as what the relationship offers as a basis for normative expectations and their satisfaction. Mutual respect secures that the parties have a we-perspective or a you-perspective, and every relationship in which there is a we-per-

101 A Collective s Rational Trust in a Collectives Action 101 spective or a you-perspective is a relationship where the other party is respected. The discussion of normative versus predictive trust constitutes the defense of the necessity of the normative expectation in genuine trust. A typical argument against the present view is that loving/liking and genuine trust accompany each other. It is true that it would be inconsistent to love someone and treat him badly, but there is no social obligation to treat the object of one s love well. When we say that we trust those who love us, we simply acknowledge that we predict that people are not inconsistent (predictive trust ) or that love and respect go together (normative trust). Those who say that they love someone, but in fact do not respect the person, may express their love and want the company of the beloved, but they may harm her in times of crisis. The necessity of the conditions of the trust account has been argued for. The joint sufficiency of these conditions is suggested for the following reasons. There seems to be wide agreement about the relevance of such components of trust as beliefs of intention, ability, persistence, gratification, and affect. What has here been argued to be missing has been added 1) through context Y and the intentionality requirement of the Goodwill condition. This draws a line between trust and a prediction-like reliance, viz. between trust and an evaluation of what others will do that affects others or has some accidental effects on the relier himself; 2) through the fifth condition. This shows the distinction between holding someone to be trustworthy (in a normative or predictive sense) and being in a position of trust or having the attitude of trusting; 3) through the trustor s social-normative expectation in the Goodwill condition and the belief of the trustee s acting in part because of the trustor s right to expect this good-willed action. This normative expectation, grounded in a relationship of mutual respect, supplies the final distinction between genuine trust and predictive trust. The trustor s decision to trust is against the whole idea of trust, although his awareness of having an accepting attitude concerning his dependence when he has sufficient rational reasons for having the beliefs of (RSNTR) may resemble his awareness of his attitude after a decision. However, deciding to trust is like making a bet on someone acting as if one trusted. When we trust in a predictive or normative sense we think we will not be betrayed. On the other hand, when we trust we may decide to depend on the trustee or to refrain from doing so. Predictive trust is, of course, a weaker basis than normative trust for deciding to depend on someone. Predictive trust is reliance on the features of a person or a situation. In a relationship of mutual

102 102 Maj Tuomela respect that justifies a social right to expect intentional gratification, the predictive trustor does not take this right as a partial reason for his trust or he lacks rational reasons for doing so, e.g. due to his belief that the trustee does not recognize the trustor s right or that that right will not affect his action. 4. Trusting a Collective Agent 4.1. General Conditions for and Implications of Collective Agency Before we apply the account of rational trust (RSNTR) to a collective case, we will discuss the special features of collectives as trustees. This will bring us to the topic of conditions for and implications of collective agency. We will also consider what evidence there could be for an outsider to have a justified belief that his trustee is a collective in the sense that it can believe, intend, and act as a collective. The first and most basic problem for a trustor, who wonders whether he can trust a collective to perform an action, is to find out whether the persons in question will act as a true collective or as an aggregate of several persons. In short, the trustee will be acting either as one agent with many hands or as many agents. In both cases the persons may contribute to the achievement of a specific common goal, and in both cases there may be dissidents engaged in counteracting what the others try to achieve. In private action people take responsibility only for what they do, although they may coordinate their actions, so as not to prevent others from acting, or cooperate in some other way. In the first case they could be seen as treating each other as part of the environment. E.g. many persons can participate in digging a large hole in the ground for garbage in the village. People contribute if and when they like. One villager works very hard, and one day he has achieved the goal of his action. The hole is as deep as possible. The other villagers goals are not satisfied by the action of the former, as they all had a shared divided private goal. However, one person s goal-achievement leads to the others abandonment of their goal. On this Weak Collectivity Condition see R. Tuomela and M. Bonnevier-Tuomela (1997, p.24). In collectively intended we-mode action the participants act as a true collective. They are disposed to take responsibility for the others actions, as their goal is a joint goal: when it is satisfied for one person it is necessarily satisfied for all (the Strong Collectivity Condition, p.28). When one person does not perform his part, the goal might not be achieved at all. I will below discuss

103 A Collective s Rational Trust in a Collectives Action 103 various types of we-mode action, which count as a group s acting as a collective or through the actions of its members. Thus, in trusting collectives as agents, the trustor must find reasons for believing that the members of the collective act as one agent, rather than as an aggregate of persons. A good reason for expecting group action is the existence of a reward for the successful action of the group, in which the success requires every group member s correct part-action. A similar reason is the threat of group punishment in the event of failure. When a group sets out to collectively achieve a goal by its members actions, the reward and punishment aspect is, in fact, an in-built part of the endeavor. The achievement of the goal is in itself the reward. However, for outsiders, like trustors, it is not always clear whether groups collectively intend to achieve a goal. In some cases a group may promise as a group to perform an action or by some other means, as a group, lead outsiders to believe that it will act. Then the outsiders, at least, have a reason to believe that it is a group that intends to act. Another problem for the trustor is the difficulty in assessing whether the group is cohesive in its collective intention. Some members may be insincere participants in the collective intention, privately intending to act correctly or incorrectly. The trustor must have a chance to evaluate the cohesiveness or the distribution of sincerity of the group members collective commitment, and the existence and effectiveness of control measures. It seems evident that the greater the number of properly collectively committed members there is, the less control of members is needed, and the more members there are who are disposed to exert control on violators. In addition, a strong commitment to the group itself makes insincere participants of a specific collective commitment to an action disposed to conform under pressure. The trustor is helped in his evaluation of the situation by information about the group members motives vis-à-vis the achievement of the specific goal, as well as, the attachment to the group. Collective believing, intending and acting will be discussed after an example of trust in a collective Trusting the School Class The account of rational trust will now be applied to a case in which the trustee is collective agent B. For the trustor, A, to rationally trust that B will perform x, A first must believe that the context is such that it is rational to consider trust at all. A must have the rational reason-based beliefs of the trust context, Y: a) A wants B to perform x, and believes that he is in that sense

104 104 Maj Tuomela dependent on B; b) A believes that B has a belief of A s dependence,; and c) that it is relevant to B. These could be true of a case where a teacher wants his class to behave well when the headmaster visits the class the following week. The teacher believes that all the pupils are collectively aware of his dependence on the class for behaving well that day, and that they are collectively going to take it into account in their decision making. Having the beliefs of the trust context Y, the teacher believes that the pupils of the class collectively believe that his fate is the hands of the class, and that the class can harm him if it so chooses. He also believes that the pupils collectively intend to take his dependence into account in their decision making. The teacher thinks of the class as a group and not as an aggregate of individuals, as he thinks that they will act as a group, due to the collective reward that he introduces. As one pupil s misbehavior will rob the whole class of the reward, the teacher has good reason to believe that the class will act as a collective and not as an aggregate of persons. This involves that the pupils will keep track of each other in order to ensure that nobody misbehaves. This mutual monitoring, pressuring ( controlling ), and helping is a positive part of acting as a group, for the group itself as well as from the point of view of the trustor. By helping those members of the group who need help in their actions, or by controlling dissidents and unmotivated members, the properly committed and resourceful members can do their utmost for their common cause. In this way even the weak and unmotivated members capacities are used. The internal control system in a group s collective action plays an important part for a trustor who is trying to evaluate the effect of weak achievers and potential dissidents. The group itself gathers the strength of its members and sees to it that the members do what they are supposed to do. A trustor faced with a mere aggregate of persons, each having a private intention of performing an action, must arrive at an attitude of trust towards each person. He cannot count on help for weak achievers or control of violators. Every person is his own master and a person does not have the authority to interfere in another s affairs, although he may try to do so. On the other hand, when there is a collective intention in the group of performing an action, the members are collectively committed to performing the action and they have a mutual belief about this. Accordingly they also believe that they have shared authority over the performance. They believe that they have at least an instrumental right to control what the others do and that they have to let others control what they do. This is due to the mutually believed collective commitment. They are together in this as it is a joint affair. Thus,

105 A Collective s Rational Trust in a Collectives Action 105 collective commitment also involves social commitment. Everyone is not only committed to himself, but to the others in performing his part of the joint project, as part of it. They owe this performance to the others, who have at least an instrumentally based right to count on it. (See R. Tuomela and M. Tuomela, 2003.) According to the account of rational trust (RSNTR) here applied to a case with a collective as a trustee, in order to have rational trust in the pupils that they will collectively behave well on that specific day, the teacher, firstly, must believe, based on rational reasons, that the pupils have a collective intention of behaving well. The teacher has this rational reason-based belief as the pupils know the class will get a trip as a reward for their good behavior. Secondly, the teacher must believe that the group is able to behave well, this belief also being based on rational reasons. In this example, this is the most problematic part. In the class are two boys who seem to be unable to control themselves, although they are sincere about wanting to be good. However, the teacher takes the group s collective ability into account, as the other pupils can be expected to pressure the uneasy boys into doing their best to gain the reward for the class. Thirdly, the teacher needs to rationally believe that the pupils will collectively intentionally act with goodwill, including goodwill towards the teacher, when they act in a well-behaved way in the classroom during the headmaster s visit. This suffices for predictive trust. However, for normative trust, viz. genuine trust, the teacher needs rational reasons for believing that he has a social right to expect good-willed behavior due to their relationship of mutual respect and that the pupils will act accordingly in part because of this right. In this example the teacher could have rational reasons to believe that the teacher and the pupils have a strong group unity that would give him a (quasi-) moral right to expect the pupils good-willed behavior in the present situation. In our example, we will settle for a weaker unity between the parties. The teacher has reasons to believe that the pupils are aware of their strong dependence on the teacher s goodwill for many years to come. He believes that he has a right to expect their good-willed behavior for socially based prudential reasons. They expect his goodwill in exchange for their good-willed behavior. They are all in the same boat in the same way as a couple who maintain their relationship for instrumental reasons. One party has the right to count on the other s good-willed actions, as the latter expects the former party s good-willed actions. This kind of prudentially based social right makes for a thinner trust than a social right based on an other-regarding motive. There is a relationship of mutual respect, but

106 106 Maj Tuomela the respect is based on prudential reasons and not on a genuine wish to take the other party s need into account. In our example, the teacher has a permanent exchange relation with the pupils. If the relation were just a temporary one, an exchange situation would still exist between the parties. He would have a general social right to expect of them good behavior in exchange for the reward. However, the social right would not be grounded in their relationship of mutual respect, but only in common norms of decency. Then he could not have the normative expectation. In this single exchange situation, the teacher would have good reasons to believe that the pupils indeed will collectively intentionally act with goodwill, including goodwill toward him, when they act, as they want to please him in order to get to go on the trip. Due to the goodwill expectation, the teacher would believe that the group is collectively committed to the task of behaving well, with at least one-sided social commitment towards him. The pupils would be committed to the teacher, not because they have promised the teacher to be good, but because they want to please him by being good. This prediction of the pupils action would allow the teacher to have predictive trust. In the present example the pupils motives are prudential, but they recognize the teacher s right to have his part of the deal fulfilled, and they are committed to him in part for this reason. The teacher can have thin genuine trust (social normative trust) in the pupils. As the group is the trustee and not a single person, the social commitment to the trustor is a collective s commitment (see below). The social right, embedded in a relationship of mutual respect, to have such a goodwill belief points at the truly socially relational aspect of trust. Without it the teacher would only have predictive trust. He would only depend on common norms of decency and the prudential motives of the pupils for his goodwill expectations. Without any goodwill expectations the teacher could still predict that the class will behave well due to the reward and that this happens to be favorable to him ( predictive reliance ). In the case of normative trust the teacher believes that the class will be doing him a favor in earning the reward, and that he has a social right to expect this due to their relationship of mutual respect, his right figuring as one of the pupils reasons for acting with goodwill towards him. The trust relationship is grounded in the trustor s dependence on the trustee, and the trustor s belief of the trustee s intentional gratification of the trustor is the core of the cognitive part of his trust. In this example the trustee s motives for accommodating the trustor are instrumental, despite the fact that the trustee acknowledges the trustor s right on the basis of a

107 A Collective s Rational Trust in a Collectives Action 107 relationship of mutual respect. General trust and personal trust require more stable other-regarding motives. When the trustor has the core-belief of rational trust in general in regard to a person, he feels comfortable in his dependence on him, and on these grounds he has an accepting attitude concerning a dependent position towards the trustee in general. To have this accepting attitude is to be in a trustor s position. This marks the difference between believing (and feeling) that someone is trustworthy and trusting. It should also be noted that there is more to general trust and personal trust than the trustor s belief of the trustee s partially non-instrumental motives in gratifying the trustor in a wide scope of matters. As in any kind of genuine trust, genuine general/personal trust cannot be simply reliance on another s personal features (predictive trust ). The parties also have to have a relation in general (and not only in relation to a specific matter) in which the trustor has an acknowledged social right to be treated well, and this right is believed by the trustor to be a partial reason for the trustee s good-willed action. Of course, such social rights are commonly respected by the very same people who have other-regarding motives vis-à-vis the trustor. E.g. friends have a right to be treated well in general by one another, and they also have predominantly non-instrumental motives for respecting this right. In a small community people may have both selfish and other-regarding motives for treating others well. The idea of generalized reciprocity 2 would here give us citizens who act with goodwill towards each other in general, each receiving good-willed treatment in return. They all stand in such relation to each other that grants them the social right to be treated well in a wide scope of matters. They have made their input of good-willed actions to the common pool and they expect to have a share of the output of such actions from the pool when needed. All this could be merely instrumental, but by entering this system they exhibit respect of the others needs and have thus earned the 2 M. Hollis (1998, pp ) discusses generalized reciprocity with an example of blood donation. Each donator maintains the blood bank by his contribution for the use of the community. He refers to R. M. Titmuss (1970) who considers this a gift between strangers. Hollis adds that the strangers are within our network. J. Rawls (1971, p. 302) idea of the veil of ignorance in which people meet to form a just society without knowing beforehand what position they will have, resembles the example with the blood bank. The contributor does not know if he will ever need the blood himself. Rawls (1993, p.50) is clearly speaking of generalized reciprocity in connection to reasonable persons, and he notes that each benefits along with the others. The reasonable indicates a departure from a fully I-rational towards a we-rational attitude. My understanding of Hollis conclusion is that a we-perspective in a system of localized generalized reciprocity could be the key to a trust society.

108 108 Maj Tuomela respect of the others. Newcomers soon come to appreciate the comfort of this general trust society. Persistent freeriders eventually become outcasts Group Action as Collectively Intended Action In the previous section we have presented an example where the trustor believes that the trustee is a group in a sense required by the trust account: it has collective beliefs and a collective intention and ability to act as a group. Below we will offer analyses of collective intending, acting, and believing. A group s intentional normatively group-binding action x, according to Raimo Tuomela s view, ontically amounts to the group members agreement-based joint action, y, which generates x (1995, Chap. 5). R. Tuomela s theory also allows for mutual belief-based joint action to count as a group s action, although it is not a group action in a normative group-binding sense. Put concisely, intentional joint action is collectively intended action that is collectively performed in the distribution sense that every member is performing his part of the action as his part of the collective action under the presupposition that the others will perform their parts. A member s performing his part as part of the collective action involves his disposition to take on tasks which will further the collective achievement of the goal. This includes helping or pressuring the other members to perform their part-actions. In some cases a member s part may be just to be in reserve. The joint action in our example is a (non-normative) group-binding joint action in its weakest sense. There is only a mutual belief-based joint plan or goal among the pupils to behave well. A joint action in its full-blown sense is agreement-based, which can be taken to be equivalent to the members mutually known acceptance of the plan (R. Tuomela, 1995, Chap. 3). This mutual knowledge may be based on an implicit agreement-type situation, in which some persons express their acceptance to act with each other by signaling their acceptance to each other. This is a normatively group-binding action, as the agreement or mutually known acceptance entails a (joint) obligation to act accordingly. In agreement-based or mutually known acceptancebased joint plan-based joint intentions, the persons are jointly committed to act collectively, and in this case the commitment is based on a joint obligation. However, we may also speak of publicly shared joint plans in a weaker quasi-obligating sense, or of non-publicly shared joint plans in a non-obligating sense. (See R. Tuomela s Bulletin Board view, 2000, Chap 2, and especially 2002b.)

109 A Collective s Rational Trust in a Collectives Action 109 Joint action in a less than full-blown public sense (meaning joint action that lacks a mutually known normative acceptance of a joint plan but that involves a publicly shared intention of participating in the joint action) is an action based on an expressed and mutually known intention. Then the collective commitment to act jointly is not based on a joint obligation. Expressing one s intention of taking part in a joint venture is less committing than to accept in a normatively binding sense to take part. However, if some persons lead each other to believe that they will act jointly, e.g., by verbally expressing their intention of doing so, or by implicitly expressing their intention, e.g., by being parties of a joint action being performed, although there is no intrinsic obligation to act, they have created a situation in which justified, or weakly justified social normative expectations typically arise. (See T. Scanlon s Fidelity Principle for related ideas about justified normative expectations (1990, p. 208) 3.) When there is verbal expression of an intention of acting jointly, I regard it as a case of justified social normative expectations, and the joint action as quasi-normatively group-binding. In my view, implicitly expressed intentions of acting jointly cannot be considered as belonging to a sphere of truly publicly shared knowledge, of which specific items can be used as evidence of a person s commitment to a joint venture. If joint action on the basis of such an implicitly expressed joint intention occurs, it is only mutual belief-based collective action. The actors did form a collective intention of acting jointly and they were, indeed, collectively committed to acting together, but the joint plan was shared on the belief level, rather than on the knowledge level, as it relied on a tacit understanding of a situation and not verbal expression. Joint intentions based on mere mutual beliefs fall outside the public realm, as the sharing of the joint plan is too weak to deserve to be called publicly shared. In a case of implicitly expressed intentions of acting jointly, there may or may not be weakly justified normative expectations of the participants actions. This weakest case of joint action cannot be regarded as normatively group-binding action, but merely as group-binding action. However, there is, at least, an instrumental obligation in the case of a mere mutual belief (R. Tuomela, 2002a, 3 Thomas Scanlon s Fidelity principle : If (1) A voluntarily and intentionally leads B to expect that A will do x (unless B consents to A s not doing x; (2) A knows that B wants to be assured of this; (3) A acts with the aim of providing this assurance, and has good reason to believe that he or she has done so; (4) B knows that A has the beliefs and intends as just described; (5) A intends for B to know that A has this knowledge and intent; then, in the absence of some specific justification A must do x unless B consents to x s not being done.

110 110 Maj Tuomela Chap 5). Then, for instrumental reasons I ought to see to it that I perform my part as part of the joint action. My example about the school class is chosen in order to show that a collective s action does not need to be generated by a full-blown joint action. A collective can act as a team by virtue of performing a joint action in its weakest group-binding sense a joint action based on a mutual belief-based shared collective intention of acting jointly. A mutual belief about the existence of the collective intention is typically based on a mutual belief about a shared reason, or on a mutual belief about implicit expressions of such an intention. Everyone is believed to be collectively committed to acting jointly at least for instrumental reasons. In my example the goal cannot be achieved without the part-actions of all the pupils. When they mutually believe that they will act jointly, any normative expectations between the pupils about acting jointly are based on social rather than conceptual grounds (viz. on their understanding that they have made a promise or something analogous). Only the instrumental obligation to act jointly is conceptually based in the sense that the satisfaction of an intention requires certain kind of activity. The group members collective commitment to act is entailed in their collective intention of acting. We have assumed only a mutual belief-based collective intention in our example. It amounts to the members intention of performing their part of the collectively intended action as a part of the collective action under conditions of mutual belief. A collective commitment to perform a joint action entails that each member commit to performing his part of the action as a part of the joint action. Thus each member is committed to doing his share of furthering the performance of the joint action, which means that the members are committed to accepting extra tasks, and to helping and pressuring the others to perform their parts, if needed. From this mutually believed collective commitment it follows that they are disposed, but only instrumentally obligated, to do so. In an agreement-based case there would be a conceptually grounded normative obligation to do so. In our example, however, due to the risk of losing the reward in the case of a violator, one could say that there is also a socially grounded normative expectation of behaving correctly among the pupils. R. Tuomela s analysis of collective belief resembles the account of acting qua a group member through the members collectively accepting the group s ethos (viz. its constitutive and most central goals, values, standards, norms, beliefs etc.) and being collectively committed to upholding it by acting in accordance with it. This account will be presented below. Collective

111 A Collective s Rational Trust in a Collectives Action 111 group-binding belief in the we-mode is analyzed as the group members mutually believed collective acceptance of a proposition as true, with their collective commitment to upholding the belief, e.g. by using it as a premise in their acting and thinking (R. Tuomela, 1995, Chap.7 and 2002a, Chap. 2). We could of course have a case of normatively group-binding collective belief in the we-mode if the collective acceptance is agreementbased. In the example of the school class we need the concept of collective belief to account for the teacher s beliefs of the pupils collective beliefs in the trust context (Y). Also, our teacher considers the pupils to be a group having an intention and thus being collectively committed to act, in this case, in the way he prefers them to act. The entailed collective commitment does important work for him. The two boys who have difficulties in controlling their impulses to misbehave will be duly monitored and pressured by the others to produce the right kind of part-actions. Their classmates are instrumentally obligated and disposed to see to it that also these two boys will perform their parts of behaving well, in order to secure the planned outcome of the collective action. The fact that the teacher believes that the pupils will act as a group takes care of his problem with the two boys. Of course, the situation would be different, if the teacher had reason to believe that the others are unable to control the potential violators. So, the satisfaction of the Ability condition of (RSNTR) is tied to the ability of the group to control deviating behavior. The group s collective commitment to the teacher, involved in his goodwill-belief, is a separate matter to be kept apart from the internal glue which results from the fact that the commitment, involved in the pupils intention of behaving well, is collective. In this example it was assumed that the teacher expected the group s collective one-sided social commitment towards him, due to the group s interest in the promised reward and its respect of his right to have his part of the deal. If the social commitment were based on a promise on the part of the pupils, it would be a two-sided commitment (offered and accepted). This one-sided social commitment expresses the minimum of what is meant by acting with other-directed goodwill. The example above shows us how (RSNTR) is applied in real life cases of collectives as agents to be trusted. The discussion shows that we are able to trust a group as a single agent if the group members act qua group members at least in a strong sense of that notion, involving a collective intention and commitment to perform a collective action. In a collective case of this

112 112 Maj Tuomela kind, the problematic part for the trustor is, of course, finding rational reasons, firstly, to believe that the group members collectively intend to act jointly, and secondly, that they, indeed, are duly collectively committed to act jointly that is, that they are sincere in their collective intentions. The collective commitment entails the group members instrumental obligation and disposition to control the other members, in case of poor ability or motivation. A group member may be insincere in his collective commitment and intend to act as a private person, performing a correct or an incorrect action. His deviant action can be prevented, if he is successfully controlled by the others. So, to summarize, the tasks of the trustor of a collective are to evaluate the sincerity of the group members collective commitment and their ability to control violators. Acting jointly on a shared collective intention is analogous to acting as a group, in the we-mode, which roughly means to act as an agent with many hands in contrast to acting as many agents in interdependence. Acting jointly may be understood to be a group s action also in the mutual belief-based cases, as demonstrated in the example above. Another kind of group action is a group s acting as a team by virtue of the actions of its position-holders, when these act in the we-mode. This means that the positionholder performs his positional task, as his positional task, and not for some private reasons (R. Tuomela, 1995, 2002c). When a collective acts through the action of its position-holder, in order to have rational trust that the group will perform a specific action our trustor s task is to evaluate the conditions of the trust account with respect to the position-holder s action Group Action as Action United by Group Ethos In cases with other notions of a person s acting qua a group member in the we-mode, it is somewhat controversial to classify such an action as a collective s action, since in those cases the group does not act as a team or through its position-holders as formal representatives. Still, it is plausible to claim that a collective acted through the actions of its members, when they acted qua group members in the we-mode (R. Tuomela, 2002c, R. Tuomela and M. Tuomela, 2003). Acting in the we-mode requires collective acceptance, collective commitment, and action for the group (R. Tuomela, 2000, Chap. 2). In the aforementioned, more recent work, R. Tuomela lists two such cases of acting qua a group member. Both of them consist of a group member s action, which qualifies as an action performed qua a group member, as it is an action within the realm of the group s ethos, and per-

113 A Collective s Rational Trust in a Collectives Action 113 formed for the reason that it serves or at least respects the ethos. In the first type of ethos-correct acting, the action is clearly required or allowed. In the second type the action is accepted or acceptable by the group as an ethoscorrect action. (For acting qua a group member see also R. Tuomela, 2002a, Chap. 2 and 6.) The we-mode cases of such ethos-correct actions involve the group member s acting, at least, in part for the group. Acting for the group can be understood as acting for the use of the group, for the benefit of the group, and at least, not against the benefit of the group (R. Tuomela, 2002a, Chap. 5). The requirement of collective acceptance is already satisfied by the fact that the action is in line with the collectively accepted ethos of the group, or accepted as such. The collective commitment requirement is fulfilled through the members collective commitment to the ethos of the group. This entails an obligation to uphold the ethos for each member, by acting in accordance with it and by seeing to it that others also do so. A trustor who wonders whether he can rationally trust that a group will perform an action, in the sense that the action can be attributed to the group through the performance of an ethos-correct action by one of its members, purposes to find rational reasons for forming the beliefs of the trust account. On the grounds of having the required rational reason-based beliefs, he feels comfortable about his dependent position and has an accepting attitude regarding this position. In order to have the rational reason-based beliefs of (RSNTR) and Y, the trustor needs rational reasons for believing that the group member will take his dependence into account in his action and that the member intends and is able to perform the action, and that he will intentionally perform it with goodwill, including goodwill towards him. The trustor must believe that he has a socially grounded right due to their relationship of mutual respect to expect this good-willed action and that the trustee acts in part because the trustor has this right. The special features of this case, which distinguishes it from a case of trust in a single agent, relate to the fact that the group member s choices reflect, or at least, should reflect the group s ethos and the action should be performed for the right reason that it serves to maintain or satisfy the group s ethos. The trustor is confronted with the same type of difficulties as he was in trusting a collective agent that is acting on a shared collective intention. His task is, firstly, to assess whether the expected action qualifies as an action that serves or at least respects the group s ethos and that it is within the realm of concern of the group (the group s topics of concern and group context). If it does not qualify, the group member will act as a private person or act incor-

114 114 Maj Tuomela rectly as a group member. Secondly, he must assure himself of the fact that the action is performed at least in part because the group s ethos is served by it. If it is not, and the group member performs the action, he acts as a private person. If the action is such that the group s ethos requires that it should be performed by the group member, and the trustor has good reason to believe that he will not perform it, then the trustor has reason to believe that the group member will act as a private person, or incorrectly as a group member. All in all, the trustor s task is to evaluate whether the group member will act as a private person or as a group member, in both cases correctly or incorrectly on purpose or by mistake. The group members collective commitment to the ethos of the group involves, at least, an instrumental obligation to uphold the group standards and goals by acting in accordance with them, and by seeing to it that the others also do so. In cases of collective public acceptance of the ethos, there is a normative obligation to maintain the ethos. When the trustor has some doubts about the sincerity of an actor s commitment to the ethos, he can be reassured, if he has rational reasons to believe that the other group members are properly committed, and thus are ready, but also able, to monitor and control that actor s action. However, the group members control of the other members actions could be difficult to exercise when the members actions are not part of a joint action in a narrow sense. After the trustor has good reasons to believe that the group member will act qua a group member, he must turn to the trust account and consider its belief conditions. A trustor may in this way have an attitude of trust that a group will perform an action, in the sense that the action can be attributed to the group, through the action of one of its members, without the group s authorization of the member to act as a formal representative of the group The Violator s Action as Part of the Group Action: an Internal and External View At least three types of group action can be identified depending on the cohesiveness of the group in its performance of the action. Firstly, the fully cohesive group action is an action or many actions performed by the group members, in which every member is sincerely collectively committed to acting in the correct way. Misunderstandings, mistakes and lack of competence may of course lead to incorrect actions, despite the members efforts to help and control. Secondly, the functionally cohesive group action is one or many

115 A Collective s Rational Trust in a Collectives Action 115 actions performed by group members, in which some members are sincerely collectively committed to acting in the correct way, while others, being only privately committed, will act in the right way when they are pressured to conform by the first. Thirdly, there is the weakly cohesive group action in which some of the members intend to act contrary to what is expected of them and the others are either unsuccessful in their control of the violators or do not exercise their control function properly. The action intended by the committed members may happen or be impaired by the violators actions. It is controversial whether this kind of collective activity should be called group action at all. One might want to disagree with a definition of acting qua a group member which allows for deviation in relation to the right kind of intention and commitment as to the general or specific goal of the group. We could have a narrow view of a group s action in which the group acts only through the members who act correctly in the sense of being properly collectively committed, allowing only for unintentional failures. A group member s intentional incorrect actions are his private actions, and these actions are not part of the collective s action. The trustor s task is to consider which members truly constitute the group in a specific action context, and he should take into account the interference of some members private actions on the properly collectively committed group members actions. The group s action is the action performed by the properly collectively committed group members. From the group s internal point of view this narrow view of the group s action could seem right. The group consists of those who are properly collectively committed to the group s ethos and the group acts through its members ethos-correct actions. This we could call the narrow internal view of group agency. In my view, such a narrow notion of a group s action does not give full credit to the concept of a group. A group has dynamics of its own, which distinguishes it from a single agent. In groups with only mutually believed collective commitment to a shared collective goal to be achieved by joint action or by separate actions, there is an instrumental obligation and a disposition, on the part of the members, to see to it that the proper actions are performed. This means that the group members are, at least, instrumentally obligated and disposed to controlling what the others do. This obligation and disposition follow from the collective commitment. In contrast, when the group members collective commitment to act jointly is based on an agreement or public acceptance, or when they are publicly collectively committed to a group goal or standard, the disposition to control others is backed

116 116 Maj Tuomela by a normative obligation to do so. When control is needed, exercised, and is effective, the potential violators are brought into line. They participate in the group s action, although they lack the proper collective commitment. The group action is functionally cohesive, although those who intended to deviate are less than full-blown participants in the group s action. They act qua group members in a weaker sense. According to R. Tuomela, this is group action, because a substantial part of the participants are collectively committed to performing the action and the others also act in the right way even if for the wrong (private) reason. Some members may thus only tacitly accept what the others do, and be only privately committed to performing the action (R. Tuomela, 2000, Chapter 6). Allowing for privately committed members among the collectively committed ones, whether the former have been pressured to conform or not, in my opinion also constitutes group action in a standard sense. The group members themselves could probably accept that some lack collective commitment if they still conform. We could speak of a wide internal view of group agency. The account of rational trust (RSNTR) requires beliefs about intentions, which in the joint action cases should be collective intention-beliefs. Aggregates of persons with parallel private intentions are not groups in any meaningful sense. In the less than full-blown, functionally cohesive case of a group s action, the expected violators, who are effectively controlled by the others, are not parties in the collective intention of the group. A trustor who wonders whether a group intends to perform x is interested in the members collective intention. However, if he has convincing reasons to believe that some expected violators will be effectively controlled, he may form the intention-belief of the trust account, despite the mixture of collective and forced private intentions of the group members. A group of properly collectively committed persons can handle some members who are only privately committed, as long as the latter are successfully pressured to conform in the performance of their actions. Privately committed group members who are pressured to conform may perform their actions properly only as long as the threat of sanctions is effective. When nothing other than social sanctions are applied, they might destroy a joint project if they lose their motivation for being a group member. A trustor, evaluating a group s future action in order to be able to form the intention-belief of the trust account, tries to assess the sincerity of the group members collective commitment to acting in the proper way. In the case of expected violators, he assesses the other members readiness and ability to control the violators. He understands that the unwilling members, who are

117 A Collective s Rational Trust in a Collectives Action 117 pressured to conform, are the weak links in the group, at least with respect to creativity, flexibility, and persistence. The members who are collectively committed may have their own selfish reasons to act with the others, but the commitment is a choice in favor of a group perspective. This means that when the goal is achieved, it is achieved for everyone, and if one member is considered a weak link, he needs to be monitored, and perhaps assisted or pressured. When the control measures are ineffective or the others fail to control the violators, we have a weakly cohesive group action. The violators actions may change the group s action into another kind of group action than that which was intended by the group. The trustor s task is to try to foresee the effect of the violators actions. This is a real practical problem for the trustor. Whether the violators can be said to act qua group members in a wide sense (the external view of group agency) or not, he leaves to theoreticians to decide. To him the group s action to be evaluated, beforehand, is the action of the group members, consisting of proper part-actions, ethos-congruent actions, or incorrect actions. The trustor is an outsider, and naturally has an external view of what constitutes a group s action. As he, in the case of a weakly cohesive group action, has reasons to believe that the violators intended actions are not effectively controlled, he must evaluate the situation vis-à-vis the end result. The action of the group may become something else than that which has been collectively intended by those who belong to the majority of the group. He cannot then have rational reasons to form the intention-belief and the goodwill-belief of the trust account towards the group concerning the expected action. If the trustor believes that the dissidents actions will not impair the group s action, he could have trust in the group that it will perform the expected action. According to R. Tuomela, in the case of dissidents who intentionally oppose what the others do, although they ought to conform, it is the group that acts (R. Tuomela, 1995, Chapter 6). He discusses groups with authorized operative members in which the dissidents are non-operatives. However, in unstructured groups the authorization could be considered to be resolved by satisfying the ethos-requirement. Those who act as the ethos prescribes are authorized to do what they do, and the dissidents are not authorized to act as they do (R. Tuomela and M. Tuomela, 2003). The standard view of collective agency discussed above requires that most of the members are collectively (and some of them privately) committed to performing an action. The wider view of collective agency, which allows for dissidents and which is in line with an external view, will be discussed below.

118 118 Maj Tuomela To claim that people could be said to act qua group members in a wide sense of that concept, when they intentionally deviate from a group s collectively intended goal or ethos is somewhat controversial. When outsiders are justified in considering the violators to be group members and when these are acting in matters within the realm of concern of the group (the group s topics of concern and group context), it seems, however, reasonable to me that the outsiders are justified in considering the violators as acting qua group members, although they intentionally act in an incorrect way. Outgroup persons think of a group s action as constituted of its members actions. The action of a group member is partly attributed to the group, if he acts within the topics of concern of the group and within the group context, whether he acts correctly or incorrectly. Both from an external and internal point of view, the group members task is to monitor and control the actions of its members. If they fail to do so, they have to assume responsibility for an incorrect action and sanction the violator (on group responsibility see P. Mäkelä and R. Tuomela, 2002; R. Tuomela and M. Tuomela, 2003). Especially, when groups present themselves as agents, they cannot escape responsibility for their members unintentionally or intentionally performed incorrect actions. If groups accepted responsibility only for their members correct or unintentionally performed incorrect actions, people would not engage in dealings with groups. Consider a case in which a violator performs an incorrect action on purpose, and the others have tried, but failed to make him conform. The efforts of the group members should convince us that the violator acted qua a group member (although incorrectly) also according to those members who tried to control the violators. This external view of acting qua a group member and of a collective s action pertains to the topic of trust in a collective s action, as a trustor is an outsider who is evaluating whether a specific group will intentionally act in a specific way, and whether it will intentionally act with goodwill toward him. To the trustor confronted by a group whose intentional action he expects, the issue of whose actions constitute the group s action and under what conditions a member s intentional action can be considered a part of the group s intentional action, in the sense that it is attributable to the group, is relevant. It becomes highly relevant to him, if he has a social right to expect of the group that it will act in a specific way. If the violator acted qua a group member, the group is responsible for his action before a third party (R. Tuomela and M. Tuomela, 2003). The trustor, who trusted the group and was betrayed in his trust, or the trustor who did not trust, but had the right to trust and thus to depend on the trustee, could now make claims on the

119 A Collective s Rational Trust in a Collectives Action 119 group for compensation. When the trustor lacks the social right to expect the group s good-willed action, he can have only predictive trust in the group. Then he considers only whether the intentionally good-willed action will come about or not. Of course, he also considers this matter in social normative trust. It was noted above that a trustor can think of an agent as a weakly cohesive group performing an action. From his external point of view he takes a wide view of group agency. The trustor may have reasons to believe that the group will (weakly) collectively perform a specific action, if he thinks that the collectively committed members are able to perform the action, although he believes that the members will be unsuccessful in controlling some intentional violators. Also, as accepted by the standard view of group agency, a trustor may have reason to believe that a functionally cohesive group (with the collectively committed members sufficient control of violators acting with a private commitment) will collectively perform an action. Such a standard view would be compatible with a wide internal view of group agency: All of us performed the action. The external view could be strange to group members. They may want to claim that uncontrolled violators acted as private persons trying to destroy the group s action. I will here attempt to argue for an external view of collective agency, which allows a collective agent to consist of a substantial number of duly collectively committed members, of some privately committed members (pressured to this by the others or not), and of some dissident-members who are committed to counteracting the collective action. As has been presented above, drawing on R. Tuomela s work, the mutually believed collective commitment, which concerns the ethos of the group, amounts to a commitment of each member of the group to see to it, together with the others, that the ethos is maintained or satisfied by actions which further it. This entails that all group members have at least an instrumental obligation and a disposition to do so. The seeing to it entails both helping and controlling others when needed. From this obligation arises the feature of a group s responsibility for what the others do. Public (normative) acceptance involves a normative obligation. In R. Tuomela and M. Tuomela (2003) the social aspect of a collective commitment (social commitment) is argued to add a normative social aspect also to the plain instrumental obligation. The members are committed to each other to upholding the ethos. Drawing on R. Tuomela s work, as has been presented in this paper, collective agency can accept some privately committed potential violators, who have been successfully controlled by the collectively committed members.

120 120 Maj Tuomela An argument for also allowing uncontrolled violators within the collective agent is each agent s obligation to see to it, together with the others, that the collective action will be performed. The members should have controlled the violators actions, or, at least, have seen to it that such persons had been prevented from taking part in the action. They may sanction their dissidents, but before a third party they have to take responsibility for what they did as a group. Here I draw on P. Mäkelä and R. Tuomela (2002) for details, and on R. Tuomela and M. Tuomela (2003) for a discussion on attributing an action to a group in the sense of it being an action qua a group member in a responsibility-sense. A person s action qua a group member in such a sense is a member s action for which the group can justifiably be held at least partly responsible: he acted as the group s ethos prescribes, tried to do so, or should have tried. Accepting that a person may act as a group member in this responsibility-sense, we may also include a dissident s action in the group s action. The group should have controlled his action. This supports the acceptance of an external wide view of group agency as a kind of group acting, thus allowing for weakly cohesive groups actions to count as collectives actions in a weak sense, and of acting qua a group member while intentionally opposing the ethos. An example from life is a group s public demonstrating for peace which involves violent incidents caused by some uncommitted members; or a state agreeing to, but being unable to stop, some rebellious individuals violent attacks against members of other states; or official position-holders of a state publicly opposing the politics that have been accepted by the authorized government. The discussion above of trusting a collective (group) yields two conditions to be added to the trust account (RSNTR) when the trustee is an egalitarian collective. 1. The trustor has rational reasons to believe that the collective is a collective in the sense of its members having the capacity to believe, intend, and act as a collective viz. it satisfies the condition for collective agency: a substantial number of the operative members are collectively committed to the group s ethos (constitutive goals, beliefs, standards etc.) or any collectively intended goal of the group. (This condition basically amounts to a c-group in Tuomela, Chap. 4, 1995.) 2. The trustor has rational reasons to believe that the condition for successful collective agency is satisfied: the members of the collective have the ability to control the actions or neutralize the effect of the actions of (potential) violators. (When the intended group action comes about we have an f-group in the terms of R. Tuomela, 1995.) The second condition could be added to the Ability condition (RSNTR) 2. Fully cohesive and functionally cohesive groups are good candidates for potential trustees. Weakly cohesive groups are

121 A Collective s Rational Trust in a Collectives Action 121 problematic cases, but may become trustees in situations where the violators uncontrolled actions do not impair the end result. In the following section we will reiew the case in which a collective trusts another collective. 5. Collectives as Trustors The account of rational trust could also be applied to a case in which the trustor is a collective agent, and from this we can make an analysis of a collective rationally trusting a collective to perform a specific action. According to the trust account (RSNTR), the trustor has various beliefs and a positive feeling towards the position of being dependent on the trustee, and on these grounds has an accepting attitude towards having a dependent position. It could be generalized that an egalitarian collective is needed that can have collective beliefs and other attitudes. Then the trustor-collective would also have to satisfy the condition for collective agency (cf. above): a substantial number of the operative members are collectively committed to the ethos of the group or any collectively intended goal of the group, and the condition for successful collective agency: the committed members are capable of controlling the actions or neutralizing the effect of the actions of (potential) violators. Then the collective trustor could have collective beliefs and other attitudes as a collective. Accordingly, it could have such attitudes without the (fatal) disturbance of violators, who might succeed in making the group attitudes look internally contradictory. R. Tuomela s analysis of collective belief resembles the account of acting qua a group member through the members collectively accepting the group s ethos, and thus being collectively committed to upholding it by acting in accordance with it. This account was presented above (4. 4). Collective group-binding belief in the we-mode is analyzed as the group members mutually believed collective acceptance of a proposition as true, with their collective commitment to upholding the belief, e.g. by using it as a premise in their acting and thinking (R. Tuomela, 1995, Chap.7 and 2002a, Chap. 2). We could of course have a case of normatively group-binding collective belief in the we-mode if the collective acceptance is agreement-based. The ability of the committed members to control some dissidents is important for practical reasons, although the group could still be said to believe something as a group in a conceptual sense, when a substantial number of the members are properly collectively committed to believe this. Situations in which some dissidents act against what the group is committed to believe

122 122 Maj Tuomela collectively have typical solutions. Take it or leave it is one. This is what often happens when a small subgroup of a group stretches the patience of the larger group. First the subgroup may conform to, but not obey, the ethos of the group. Later it may gain strength and start to proclaim in words and by action that the ethos of the group is a view that is opposite to the ethos. When the group rids itself of the opposing members these might form a new competing group with an ethos that is in opposition to the ethos of their former group. The collective acceptance-belief model is well suited for collectives as trustors. Everyone in the group ought to believe that x in an acceptance sense of believing, as long as this is the collectively accepted belief for the group. This means that everyone accepts x as the group s belief (view) and hence accept that they will act accordingly for this reason within the realm of the group s concern, and everyone believes that there is a mutual belief about this in the group. If this holds true, then everyone ought to be, and is collectively committed to believe x. There is at least a mutual belief vis-à-vis the instrumental obligation and the collective commitment, due to the collective acceptance of the belief, which need not be public (R.Tuomela, 1995, Chap. 7; 2002a, Chap. 5). Dissidents are those individuals who are typically mistakenly assumed to be among those who accept the belief. Then the mutual belief is not true and the dissidents are not among those who mutually believe. Dissidents could be acknowledged and tolerated because the others believe that they can control them and eventually persuade them to fall in line (R. Tuomela and M. Tuomela, 2003). In all collective commitments, however instrumental the obligation to be committed may be, there is a social aspect the commitment to the others in the group (social commitment). We owe it to each other that we act as we should, and that we see to it that our ethos is maintained by suitable actions. The instrumental obligation is thus strengthened by at least an instrumentally justified normative social expectation. Dissidents are not only reproachable for disturbing the joint project, but for letting the others down, as the others have at least an instrumentally justified social right to count on their part performances. In a case with a mere mutually believed collective commitment some dissidents may, of course, be mistakenly assumed to be collectively committed. Then the others are not justified in their reproach (R. Tuomela and M. Tuomela, 2003). When the trustor is a collective agent capable of having collective acceptance beliefs, we may use the trust account through Y a) to (RSNTR) 3. As (RSNTR) 4 involves the trustor s feeling of being comfortable with his de-

123 A Collective s Rational Trust in a Collectives Action 123 pendence on the trustee, we would have to assume that a collective can have feelings as a collective, which seems strange. However, a collective could have an acceptance-feeling in the acceptance-belief sense. E.g. I ought to accept and do accept feeling x as it is the group s collectively accepted feeling or attitude for the group, and ought to accept and do accept acting in accordance with it for this reason, within the realm of the group s concern, believing that everyone else in the group ought to accept and does accept it, and that there is a mutual belief about this in the group. When this holds for a substantial number of the members, we get a group feeling in the same way as a group belief. Acceptance-feelings could, of course, be supported by shared we-feelings. Thus normative group feelings may involve the genuine feeling of each group member, just as normative group beliefs may involve shared we-beliefs the members genuine beliefs. (RSNTR) 5 is concerned with having an accepting attitude vis-à-vis the dependent position. For a collective trustor we would follow the same lines as above, in the case of a feeling. Instead of a person s having an accepting attitude about a dependent position, the group would collectively accept that it has this accepting attitude concerning this position. This collective acceptance is mutually believed, but it is not required that the acceptance be public. A collective acceptance with this kind of content could well be the case on the grounds that the group members have collectively accepted all the beliefs of RSNTR as the group s beliefs, and the feeling as the group s feeling or attitude. To summarize: when an egalitarian collective is a trustor it has to satisfy the condition for collective agency. It must be capable of having collective beliefs and other attitudes that serve as premises for the members actions. So, this condition must be added to the trust account (RSNTR). The members of the trusting collective must have the capacity to have beliefs and other attitudes, to intend, and to act as a collective viz. they satisfy the condition for collective agency: a substantial number of the operative members are collectively committed to the group s ethos (constitutive goals, beliefs, standards etc.) and any other collectively intended goals of the group. For the group to appear uniform in front of others it has to satisfy the condition for successful collective agency: the members of the group must have the ability to control the actions or neutralize the effect of the actions of (potential) violators.

124 124 Maj Tuomela 6. Conclusion The present analysis offers a new approach to the topic of trust by presenting the following elements: Necessary and jointly sufficient conditions for rational social normative trust and a context for rational trust. Criteria for a distinction between trust and a broader concept of prediction-like reliance, between trusting and considering someone trustworthy, and between social normative trust (genuine trust) and predictive trust. The restricted freedom of the trustee. Acting with goodwill as a basket for the elements of commitment, persistence, and gratification, an idea which allows for a variety of motives (e.g. prudential motives), and reasons (e.g., agreements, proper social norms 4, traits, interests, and relations). A discussion of collective agency (a collective commitment to group ethos or collective goals) and a collective s rational trust in a collective s action (a collective commitment to group ethos or collective goals and an ability to control dissidents).* * I wish to thank Prof. Cristiano Castelfranchi, Prof. Robert Audi, Dr. Mark Siebel, as well as my Finnish colleagues, Pekka Mäkelä, Kaarlo Miller, Dr. Petri Ylikoski, and Prof. Raimo Tuomela for comments on my work. The funding of the Finnish Academy and the Swedish Cultural Foundation in Finland has made this work possible. References Baier, A., 1994, Moral Predjudices: Essays on Ethics, Harvard University Press, MA Castelfranchi, C. and Falcone, R., 1998, Principles of Trust for MAS: Cognitive Anatomy, Social Importance, and Quantification, Proceedings of the International Conference on Multi-Agent Systems (ICMAS 98), Paris, July, Gambetta, D., 1988, Can we trust trust, in Gambetta, D. (ed.) Trust - Making and Breaking Cooperative Relations, New York: Basil Blackwell, Inc., Hollis, M., 1998, Trust within Reason, Cambridge: Cambridge University Press Lagerspetz, O., 1998, Trust: The Tacit Demand, Dordrecht: Kluwer Academic Publishers 4 Proper social norms are unarticulated conventions and group-specific norms (R. Tuomela and M. Bonnevier-Tuomela, 1992 and 1998, and R. Tuomela 1995, 16-28).

125 A Collective s Rational Trust in a Collectives Action 125 Loukola, O., 1999, Who do you trust? Combining Morality and Rationality, Reports from the Department of Philosophy No 4, University of Helsinki Mäkelä, P. and Tuomela, R., 2002, Group Action and Group Responsibility, Protosociology XVI, Pettit, P., 1995, The Cunning of Trust, Philosophy and Public Affairs, Vol. 24, Rawls, J., 1971, A Theory of Justice, Mass.: Harvard University Press Rawls, J., 1993, Political Liberalism, New York: Columbia University Press Scanlon, T., 1990, Promises and Practices in Philosophy and Public Affairs Nr 19, Titmuss, R. M., 1970, The Gift Relationship: From Human Blood to Social Policy, London: Allen & Unwin Tuomela, M., 2000a, The Components of Rational Trust and the Power of Social Pressure, Licentiate thesis at the Department of Social Psychology, University of Helsinki Tuomela, M., 2000b, The Mental Components of Rational Trust, Proceedings of the Autonomous Agents 2000 Workshop on Fraud, Deception, and Trust, Barcelona, June 2000, Tuomela, M., 2000c Rational Trust, in Brogaard, B. (ed.) Rationality and Irrationality (Contributions of the Austrian Wittgenstein Society, 8), Kirchberg: Austrian L. Wittgenstein Society, Tuomela, M., 2002, On Rational Trust, in Meggle, G. (ed.) Social Facts & Collective Intentionality, Frankfurt am Main: Dr. Hänsel-Hohenhausen AG, Tuomela, M., 2003, The Components of Rational Trust, forthcoming in Sintonen, M., Ylikoski, P., Miller, K. (eds.) Realism in Action: Essays in the Philosophy of the Social Sciences, Dordrecht: Kluwer Academic Publishers, Tuomela, M. and Hofmann, S., 2003, Simulating Rational Social Normative Trust, Predictive Trust, and Predictive Reliance Among Agents, forthcoming in van den Hoven (ed.) Ethics and Information Technology, Vol. 5 Tuomela, R., 1995, The Importance of Us: A Philosophical Study of Basic Social Notions, Stanford: Stanford University Press Tuomela, R., 2000, Cooperation: A Philosophical Study, Philosophical Studies Series 82, Kluwer Academic Publishers, Dordrecht Tuomela, R., 2002a, The Philosophy of Social Practices: A Collective Acceptance View, Cambridge university press Tuomela, R., 2002 b, Joint Intention and Commitment, in Meggle, G. (ed.) Social Facts & Collective Intentionality, Frankfurt am Main: Dr. Hänsel-Hohenhausen AG, Tuomela, R., 2002 c, The We-mode and the I-mode, in Schmitt, F. (ed.) Socializing Metaphysics: The Nature of Social Reality, Rowman and Littlefield, Lanham, Tuomela, R. and Bonnevier-Tuomela, M., 1992, Social Norms, Tasks, and Roles, Reports from the Department of Philosophy, University of Helsinki, No 1

126 126 Maj Tuomela Tuomela, R. and Bonnevier-Tuomela, M., 1997, From Social Imitation to Teamwork, in Holmström-Hintikka, G. and Tuomela R. (eds.) Contemporary Action Theory, Vol. II: Social Action, Kluwer Academic Publishers, Dordrecht and Boston, 1-47 Tuomela, R. and Tuomela-Bonnevier, M., 1998, Norms and Agreement, in Attwool, E. and Comanducci, P. (eds.) Sources of Law and Legislation, Vol. III, Stuttgart: Franz Steiner Verlag, Tuomela, R. and Tuomela, M., 2003, Acting as a Group Member and Collective Commitment, Protosociology Vol. 18, 7-65

127 Social Action in Large Groups 127 Social Action in Large Groups Ulrich Baltzer Abstract Large Groups are not constituted simply by adding further members to small groups. There is a qualitative difference between the social actions which take place in small communities and those in large ones. Large communities are irreducibly characterized by anonymity, i.e., the members of large groups don t know of most of the other members as individual. Therefore, social action in large groups is based on a sign process: each member of a large group is understood as a representative of the other anonymous members of the group as well as a sign for the group as a whole. Once upon the time the philosopher Charles S. Peirce told his readers that man is a sign. 1 If you look at the people surrounding you or at yourself, this opinion seems to be quite implausible. But after reading this paper you should be convinced that Peirce was right because being a sign marks an essential difference between large and small communities. There are large social groups only if the members of these groups are treated as signs this is not the case with the members of small groups. (Examples for large groups are nation-states and large companies, examples for small groups are families.) In other words: There is a qualitative difference, not only a quantitative one, between the social actions which take place in small communities and those in large ones. Large Groups are not constituted simply by adding further members to small groups. The result of this paper has been stated already. But, is its premiss correct? Are there really respectable theories of social action which maintain that large communities differ from small ones only in the number of its members? There are such theories: contractarianism could be offered as an example as well as methodological individualism in social philosophy. One can see from two opposite perspectives that contractarianism does not make any difference between small and large groups. The first perspective is bottom-up: it starts with small communities and grows into the large. The 1 Cf. Peirce, Ch. S Some Consequences of Four Incapacities. In Writings of Charles S. Peirce. A Chronological Edition. Vol. 2: , ed. by E. C. Moore et al , here p Bloomington: Indiana UP 1984.

128 128 Ulrich Baltzer second is top-down; its method is the reverse of the former perspective. The founding fathers of contractarianism, Hobbes and Locke, can be classified as using the first perspective. The leading picture of both philosophers seems to be an ordinary contract between two or three parties. Hobbes conceives the transition from the state of nature to the creation of the society as a reciprocal contract between two parties that favour a third one, viz. the sovereign. Hobbes is basically governed by the idea of a contract between three persons. 2 Two of them agree in the social contract to relinquish their respective natural right of everyman to every thing. 3 The third person is favoured by this agreement without being party of the contract. She can now make use of her natural right to everything unhindered by the rights of the parties of the contract. Therefore, she becomes the sole and absolute souvereign who has the live and limb of the other persons at her disposal. It is essential for this paper that Hobbes does not make any difference between the social contract which results in the situation in which the souvereign rules about two persons and those agreements that concern two thousand or two hundred millions. That is, larger communities arise from smaller ones simply by the fact that more persons commit themselves. There is no third party in Locke s conception. According to him, there are two parties who mutually promise in an explicit way to keep the rules and laws that shall be in force in their community. Locke is less interested in the founding of societies, rather he analyzes the integration of persons in an existing society. Such an integration requires that the person who wants to be integrated must explicitly promise to keep the laws of the integrating society. 4 In turn, these laws have legal force in the society because every member has explicitly recognized them in the face of the other members. As it is the case with Hobbes conception there is only a quantitative difference between small social groups and large ones. A large community arises from a smaller 2 Of course, Hobbes state of war in the state of nature would not be given if there were only three persons. According to Hobbes, the persons in the state of nature have to use force as a precaution. But this does not establish a stable order, because even the actually most powerful person can t be sure that there isn t any further person that has more power than she herself. If there is a constellation of only three persons there would be a (temporarily) stable social order without any enduring state of war because of the mutually known strength, means, and cunning of the three persons. 3 Hobbes, Th Leviathan, ed. with an Introduction by J. C. A. Gaskin, I, 14, 4, p. 87. Oxford/ New York: Oxford UP Cf. Locke, J The Second Treatise of Government, ed. with an Introduction and Notes by P. Laslett, ch. VIII, 95, 113, 117. Cambridge: Cambridge UP

129 Social Action in Large Groups 129 one by the fact that further persons explicitly commit themselves to keep the laws of the community. Let us come now onto the second perspective, the top-down approach. The most prominent examples of this perspective are Kant s conception of the social contract and its development by John Rawls. Contrary to the approaches of Hobbes and Locke, which rest on a prudent voluntariness, the necessity of the social contract results from the concept of law itself according to Kant. Kant conceives of the law as the epitome of the conditions under which the acts of will of one person can be combined with those of another person in a manner that agrees with the universal rule of freedom. 5 Law provides the exterior freedom for our actions, that is, without law there were no room for actions that are governed by practical reason. In other words: practical reason requires law. Therefore, the social contract is founded in the end on practical reason. Each man is a priori subject to the social contract simply by the fact that man is a being who has practical reason according to Kant. The validity of the social contract does not rest on utility or voluntary acts as it is the case with Hobbes and Locke s approaches. Kant s conception of the social contract begins with the most extended community, viz. mankind. Each man wants a priori the state of law because each member of mankind has the faculty of practical reason. Each concrete group that has less members than mankind derives the validity of its laws and rules from the most extended community. Kant s approach leads from the maximally conceivable community via subtraction of members to smaller groups. This subtraction is feasible because Kant bases his foundation of law onto practical reason with which each single person is individually endowed. There must not be a qualitative difference on this continuous way from large groups down to small ones, otherwise the concrete norms which are valid in small communities would be independent of the basic principles of law that are derived from practical reason. Rawls conception of the original contract 6 can be assigned also to the topdown approach although he does not begin with mankind as the first step of his analysis. Nonetheless, the subject of his theory are large communities, viz. nation-states. On the one hand, small groups are certainly not his target when he proposes the veil of ignorance as means for agreeing on the prin- 5 Cf. Kant, I Die Metaphysik der Sitten. Rechtslehre. In Kant s gesammelte Schriften, Abteilung I: Werke, ed. by Preußische Akademie der Wissenschaften, vol. 6, p Berlin: Reimer [and other publishers] 1900 seq. 6 Cf. Rawls, J A Theory of Justice, ch. 3. Cambridge (Ma.): Belknap.

130 130 Ulrich Baltzer ciples of just institutions. On the other hand, there are no considerations that would set a lower limit of the size of the community whose institutions are to be formed. Therefore, Rawls veil of ignorance could also be used in negotiating just rules in a very small community. We have reached a first preliminary result: This paper does not argue with a chimera; there are really theories which assume that there is only a quantitative difference between small and large groups. And these theories are quite important in the field of political philosophy. Small communities differ not only from large communities by the number of the respective members. In what is the qualitative difference between small and large groups founded? According to Margaret Gilbert the essential difference consists in the anonymity that is the typical feature of large groups. 7 Anonymity in large groups means not only that one knows a lot of members slightly, rather one doesn t know of most of the other members as individual. Who does know all Germans, who does even know all employees of the own university? In small communities matters are quite different. In these communities you know all members, maybe only sligthly. Everbody knows which person is a member and which not. Each member of a small group can distinguish members from persons who are not members of the group. If the small community isn t too secretive this distinction can also be made easily by outsiders. It is clear from inside as well as from outside who is member of the small group and who is not. However, in large communities there is nobody neither member nor outsider who know of each member. Is it a consequence from this fact that large groups can t be delimited? Is there only one large community that can be delimited with complete justification, viz. mankind, that contains every man without exception? Therefore, was Kant right in choosing mankind as the starting point of his approach? In our daily life we take it for granted that there are large communities which contain less members than mankind. Otherwise, there were no hostility to foreigners, everbody would be domestic. It would be incomprehensible that people could be proud of their large company. The shame of some Germans concerning the atrocities of the Third Reich would be curious a shame that may refer to actions which were carried out before the own birth. 7 Cf. Gilbert, M Reconsidering the Actual Contract Theory of Political Obligation. In Sociality and Responsibility, Lanham: Rowman & Littlefield; see also Gilbert, M On Social Facts, p. 212 seq. Princeton (NJ): Princeton UP.

131 Social Action in Large Groups 131 There seems to be no problem in our daily life to delimit large communities although they are characterized by anonymity. The problem of delimitation arises from the following premiss: There must be at least a single authority or an individual person who knows every member in order to determine the boundary of a community. But we can drop this premiss without hesitation. It is sufficient for the delimitation if we are able to find out in the individual case whether a formerly unknown person is a member of the respective group or not. One possible method to find out the membership of a unknown person is to ask a third person who knows something about the former person. Such an interconnection of knowledge needs even in the case of large communities astonishingly few intermediate connections. This finding is underlined by the fact that every arbitrary indivdual is related to any prominent figure on average over three up to five intermediate relatives. Another well known method to find out the membership of a person consists in the examination of certain features of the unknown person. E. g. passport controls at airports and border crossing points use this method. Therefore, the anonymity in large communities does not vanish totally concerning the members of the community but it disappears in each individual case that is examined in detail. Anonymity is here merely an epistemic hindrance that can t be removed in its entirety but in each individual case when it is necessary to do so. This kind of removable anonymity can t be the irreducible distinguishing feature of large groups sought after. A more promising candidate is the kind of anonymity which Martin Heidegger sketches in those passages of Being and Time that analyze the forms of our daily social life. A lot of things which we are normally doing are subservient to the One ( das Man ): One does not do that!, One has to go to the seaside in the holidays., etc. What us deem right or wrong, what we strive for or avoid, what we do or refrain from doing is to a large extent or more likely: totally subject to the decision and ruling of the One. Heidegger stresses the anonymity of the One. According to him those other persons who determine the everyday life of a single person as the One are not certain other persons. Any person can stand for them. 8 The One is the epitome of the claims and imperatives imposed on oneself by the nameless and exchangeable others. The impression that the claims of the One are claims on the part of the others is, according to Heidegger, a widely used strategy in our daily life because it is very comfortable. In this way the individual is able to get out of the responsibility for his actions and opinions and 8 Cf. Heidegger M Sein und Zeit, 27, p Tübingen: Niemeyer

132 132 Ulrich Baltzer to push it onto the others. With that it becomes invisible that the individual person is part of these others, that he himself helps to establish the claims of the One, which in turn are represented as obligations forced on the individual. Heidegger calls this disassociation from the others that denies the own membership the distantiality ( Abständigkeit ). Besides the distantiality the One is characterized by the averageness ( Durchschnittlichkeit ) and the leveling down ( Einebnung ). 9 Especially in highly routinized episodes of our daily life, that are central for our social life because of their frequent occurrence, the One unfolds its power owing to its inconspicuousness and its naturalness. In these episodes everybody does what one normally is doing in such situations without noticing it as a ruling. It seems to be simply natural to act or feel in the respective way. Fashions of every stripe are a good example for this work of the One. How strange does one feel himself, which mocking and disparaging looks one harvests if one is wearing clothes that didn t attract attention at all 15 years ago? How many people who swear now by the black colour of their car had the likewise firm conviction 30 years ago that no car can be black except hearses and cabs? And because one likes certain things, is striving for them and is doing them, there are no other things except these. Everyone of us knows the phenomenon that one is standing at the electrical shop in front of 35 different toasters and one can t find any square model because the round models are fashionable. Everything that does not go with the mainstream will be dismissed as outdated or odd, or will be actively fighted as awkwardness, or will even be punished as disruption of the public order. The One is the uniform average, is the normative demand on part of the others. Who is really the One? The who is not this one, not that one, not oneself, not some people, and not the sum of them all. The who is the neuter, the one 10 On the one hand, the One is extremely real and effective as we saw above with examples. On the other hand, the One can just as little be linked to an individual person or an individual group of persons as one can find the statistically computed average family with its 1.6 children in the real world. Contrary to the anonymity that concerns the delimitation of large groups, the One is a kind of anonymity that can t be derived from features of individual persons or small communities. At a certain size of the community the neuter the One comes into being and with that there is a kind of ano- 9 Cf. Heidegger 1927, 27, p. 126 seq. 10 Heidegger, M Being and Time. Transl. by J. Macquarrie and E. Robinson, 27, p. 164 New York: Harper & Row (italics in the original).

133 Social Action in Large Groups 133 nymity that has no counterpart in small communities. Therefore, considerations which are valid for individuals or small groups can t be transferred without breaking to large social groups. And, of course, this is true vice versa. Although Heidegger presents some deep insights into the structure of the One, his Being and Time gives no support in elaborating a theory of social action in large groups. This has two reasons. The first reason has been discussed in the debate about the question which status has the One in Being and Time. 11 Without entering the details of this debate one result can be fixed: Heidegger assesses the One in two opposing ways. On the one hand, Heidegger describes the One as an existential ( Existenzial ) of human existence ( Dasein ). 12 Therefore, even on the deepest ontological level the human existence would not be what it is without the One. The One is not an avoidable aberration; it is inevitable. On the other hand, the passages of Being and Time that are devoted to the One are written in a disparaging tone and the One assembles all features that are the antithesis to the authenticity and naturalness sought by Heidegger. The One unites all the characteristics that have to be got rid of by the insight into the ontological structure of human existence. It is only consequential that Heidegger sets the social action under the dictate of the One against the authentical form of human relationships which is based on the care that is freeing the other ( vorspringend-befreiende Fürsorge ) instead of submitting him to the One. 13 One person is caring in this way for another person by giving the latter the 11 Cf. Löwith, K Das Individuum in der Rolle des Mitmenschen. In Mensch und Menschenwelt. Beiträge zur Anthropologie. Sämtliche Schriften Bd. 1, Stuttgart: Metzler 1981; Lévinas, E Totalité et infini. Essai sur l extériorité. Den Haag: Nijhoff; Theunissen, M Der Andere. Studien zur Sozialontologie der Gegenwart. Berlin: de Gruyter; Tugendhat, E Selbstbewußtsein und Selbstbestimmung. Sprachanalytische Interpretationen, 230 seq. Frankfurt a. Main: Suhrkamp; Dreyfus, H. L Being-inthe-World. A Commentary on Heidegger s Being and Time, p. 154 seq. Cambridge: MIT; Demmerling, Ch Hermeneutik der Alltäglichkeit und In-der-Welt-sein ( 25-38). In: Martin Heidegger. Sein und Zeit (Klassiker Auslegen 25), ed by Th. Rentsch, , esp. p. 95 seq., Berlin: Akademie. 12 Cf. Heidegger 1927, 27, p Cf. Heidegger 1927, 26, p Heidegger distinguishes two kinds of care. The care which was mentioned already in the main text ( vorspringend-befreiende Fürsorge ) is contrasted with a second type of care that replaces the care of the respective person by that one of another person ( einspringend-beherrschende Fürsorge ). The latter form is inauthentic, because it is a leaping in for the other, a taking care of their possibilities for them. The other comes under the rule of the caring person because some difficulties are taken away from the person cared for, but in that she looses some freedom of action, too. According to Heidegger, a typical example is the welfare aid because the own care of a person who counts as a hardship case is replaced by the care of the institution.

134 134 Ulrich Baltzer opportunity not only to see about which she herself has to care but also to devote herself to this fundamental care. That is, this kind of care takes place in face-to-face interactions of two persons. Therefore, Heidegger tries to cure large communites of their lack of authenticity that hurts him by reverting to relationships between two persons. If the anonymity of the One is a specific and irreducible feature of large groups then this remedy is simply the call for the disbanding of large communities. This means obviously that a philosophical theory of large communities has to declare itself bankrupt. There is a second, more decisive reason why Heidegger s analysis of the One is no help for a philosophical theory of social action in large groups. In a passage of Being and Time Heidegger characterizes the One as the Nobody ( das Niemand ). 14 Although the One is ruling the averge everyday social life it is a nobody according to Heidegger: the power of the One can t be bound to any concrete person or small group. But, is that correct? Do the demands and imperatives of the One not have a concrete face? The neighbour is looking at me disapprovingly if I don t wipe clean the staircase, isn t he? Don t the mocking remarks of my colleagues make me buy a new suitcase? Certainly, the One is neither my neighbour nor my colleagues. But they give the One a face and a voice; they are the representatives of the One. The disapproval of my neighbour could be easier borne if it could be understood as coming from the private preferences of my neighbour instead of stemmig from the imperatives of the One. Although Heidegger talks explicitly about representatives in discussing the distantiality of the One he does not deepen this thought. From Heiddeger s point of view there is merely the alternative between the concreteness of the other human being in a face-to-face interaction and the totally disembodied nobody of the One. Heidegger doesn t realize the possibility that the concrete partner in the interaction can be the representative of the One which in turn is not a concrete person. Understanding the partner in the interaction as representative of the One does not mean to replace a person who is temporarily absent by another person who is present. The One doesn t give orders as a foreman gives orders instead of the boss. The foreman is a representative of the boss in the sense of replacing one object by another object that belongs to the same category as the former one. The boss could give orders himself if he were present. It is impossible for the One to be present in this sense. My partner in the face-toface interaction who gives the One a concrete body is no substitute for a bodily object. The One which is represented by the concrete person can t be 14 Cf. Heidegger 1927, 27, p. 128.

135 Social Action in Large Groups 135 present without the relationships of signs. It can t appear without being represented. It is never present in its entirety, it is always indicated by signs. That means the other way round that my partner in the interaction is not merely understood as a concrete person but as a human being that counts besides his concrete features as a representamen for a being that isn t concrete at all. Nobody is able to see a large community in its entirety. One knows only several members who refer as members to the large fraction of the unknown, anonymous members. In the case of large communities this reference to the anonymous members can t be transformed into a reference to certain kown persons because there are too many members. And everybody takes this into account. This makes the difference between large and small groups. Certainly, there are also persons in small communities who are understood as signs, e. g., my wife reminds me of my mother in law. But this kind of sign is only an indirect way to come into contact with somebody with whom I could also be connected in a direct way. This direct connection lacks large communities because I can refer to large groups only with the help of signs. A large group is nowhere in its entirety, it is only present in tokens that are taken as refering to an absent entity. This absence has two forms. On the one hand, the large group is the absent entity. In this case the absent entity is categorically different from the person who is the representamen of the group. The group is present in its members, but only insofar as each member is a sign of all the other members taken as a single entity. This sign process can be got under way by each member but it can t be finished because most of the other members are not even known of. On the other hand the absent entity is categorically the same as the refering token. This is the case if the sign process is understood as connecting a certain member with another absent, anonymous member of the same group. The entity refered to is now the individual member, not the group as it was the case before. Now, we can replace the refering entity with the entity refered to. Both are likewise individual members and in being a member they are only numerically different. From this point of view it is true that man is man. This dimension of the sign process allows the delimitation of large communities. The membership is a sign process that refers to anonymous persons, but it does not refer to anyone at all. Each of these anonymous persons has one feature: (s)he is a member of the respective large community. That does not imply that any of the members has voluntarily chosen or accepted to be a member. A baby who is born in Germany and who has German parents is a German citizen according to German law this baby need not have an idea what it means to be citizen.

136 136 Ulrich Baltzer Small communities are, of course, not simply a set of persons. The members of small groups are also understood as members and refer as such to the other members of the group and to the group itself. But these other members are known without exception as well as the network of the relationships that constitutes the group is known in its entirety. Contrary to large groups there is no need to use members as a sort of variable that represent other members who can t be known. The anonymity makes a sign process inevitable in large communities which forbids to adopt theories that were derived from the analysis of small communities and vice versa. The sign process marks the qualitative gap between large and small groups. Large groups differ from small groups not only by the number of their members. Therefore, an adequate theory of large communities can t be designed without including elements of semiotics. The lack of these elements in contractarianism seems to be for me one more reason not to trust this school of thought in the field of political and social philosophy.

137 What do We Mean by We? 137 II Collective Intentionality and Social Institutions

138 138 Stephen P. Turner

139 What do We Mean by We? 139 What do We Mean by We? Stephen P. Turner Abstract The analytic philosophy form of the problem of collective intentionality originated with the claim that individual statements of the form I intend x cannot add up to a we intend x statement. Analytic philosophers from Wilfrid Sellars on have pursued a strategy that construes these sentences as individual tellings of statements whose form is collective. The point of the strategy is to avoid the problematic idea of a real collective subject. This approach creates unusual epistemic problems. Although telling of collective intentions is parallel to the expression of individual intention, one can be deceived about them. I suggest that none of the supposed evidence could solve the problem of deception, because there is no fact of the relevant kind to be deceived about. I also argue that this strategy is unnecessary. Statements like Joe Namath s guarantee of victory in the Superbowl are model non-collective statements which are interchangeable with many supposed collective statements. Yet, no novel mode of telling and nothing epistemically anomalous is required by this statement. The statement is merely an individual statement conditional on a variety of facts, which happen to include facts about other people, whose only commitments are epistemic. Sellars s problem structure is then itself critiqued to suggest that it confuses a grammatical problem with a factual-theoretical problem about the reality of collectivities and the cognitive character of intention attributions, and further confuses collective intentionality with a problem about the nature of morality. Concepts of collective will, group intention, objective mind, and the like have a checkered past. Much of this past is marked by a dialectic between two classes of claims. One class involves the idea that there are explananda, such as morality, which require an extraordinary explanans, such as a group will, and that these explanations necessarily involve collective rather than individual intentions or commitments. The second class of claims, exemplified in Weber s methodological writings, suggests that the problematic explananda are misdescribed when they are described in collective terms, that notions like the will of the people are fictions, and that ideas of group belief, group mind, and group agency are only metaphors. A key concern of Weber s was the question of whether there is any explanatory value added by appeals to collective concepts. He denied that there was. In what follows, I will apply some of these concerns to the current analytic philosophy form of the

140 140 Stephen P. Turner problem, and suggest that the concerns not only continue to be relevant, but reveal epistemic issues that undermine the whole project. The concept of collective intentionality as it has been reborn in analytic philosophy was designed to avoid the problems posed by the idea of a collective agent with a group mind. The main attraction of the reborn concept is in connection with a newly emerging orthodoxy about norms. The orthodoxy holds, roughly, that norms are (and may be explained as) rational solutions to coordination problems, but not fully explained in this way. To make a rational solution normative requires that it be the subject of a collective agreement. Because actual norms are not the subjects of actual agreements (and because in the nature of the case some optimally rational solutions must be tacit) the agreement, such as making a piece of paper count as a valuable exchange commodity must be understood as a matter of collective intention. Here the explanatory function of assertions about collective intentions is to make the solution normative. In what follows I will focus on the narrow question of whether there is any fact of the matter about particular supposed collective intentions. This question, in turn, I will treat entirely in connection with the problem of the added explanatory value of appeals to collective intentionality. I will take it for granted that an argument that the appeals are explanatorily superfluous because there is no fact that they are needed to explain is sufficient for abandoning the concept as an explanatory concept. There may be non-explanatory concepts of collective intentionality that might have some role in ethics, and about that topic I will say nothing directly. Rather that undertaking a detailed discussion of the particular analyses and arguments in the literature, or indeed of surveying them, I will consider what I take to be a basic structural feature of arguments for the existence of collective intentions, which I will call the anomalization strategy, and consider one issue that this strategy is designed to respond to, the epistemic problem of the grounds for claims about collective intentions. This will require me to characterize in broad terms a number of topics on which there are very diverse positions. I will, however, examine in more detail the text that was the starting point of the analytic rehabilitation of the concept of collective intention, Wilfred Sellars s paper, Imperatives, Intentions, and the Logic of Ought (1963), which deals in a particularly explicit way with the epistemic issues raised by the concept.

141 What do We Mean by We? 141 The Analytic Problem The central intuitions behind the analytic approach are grammatical. There are collective intentional objects, such as plans, or goals, such as the victory of a team in a competition, which seem to require collective subjects, especially when they are the object of intentions expressed by will and shall. 1 I will win the Superbowl cannot mean what it says: only teams win football games. To say that there are no genuine collective intenders thus creates a problem. If all the intenders are individual, and this means the same thing as all intentions are individual, then apparent collective intention statements need to be analyzed into individual intention statements. Thus it is a problem within analytic philosophy. One approach to this problem would be to think that collective statements ought to be analyzable into an enumerative list of statements of the individual intentions of the members of a group. Typical examples of this kind of reduction fail, especially (and perhaps only) when the objects of the intentions are themselves collective. I will play Beethoven s Fifth beautifully said truly by all the members of an orchestra allows for a lot of discordance. The orchestra will play it beautifully does not follow from these individual intentions to play well, nor is it equivalent to a list of the statements of individuals. Consider a more straightforward example: We, the Dallas Cowboys, intend to win the Superbowl. In this case the individual analysis seems to be excluded because only teams win Superbowls. I will win the Superbowl is of doubtful intelligibility, and it becomes no more intelligible if repeated by each member of the team. Yet it would be very strange to claim that winning the Superbowl could not be the object of any intention, individual, collective, or otherwise, and if it cannot be the intention of an individual it must be the team that has the intention. So it appears we are forced on grammatical grounds to concede that something needs to correspond to the intention to win that is not reducible to a series of ordinary individual intentions. 2 The problem then becomes one of characterizing these intentions without locating them in an actual group mind. 1 A simple, standard statement of the problem is found in Velleman (1997, 29-30). 2 One may quibble with many of the examples here: part of an individual s playing beautifully is playing non-discordantly, and that is a causal rather than a grammatical fact. So the meaning of this result is not clear, a point to which I will return below in connection with the problem of collective states and collective goods. But it is clear that having the orchestra itself reach the state of non-discordant playing is not the appropriate direct object of a single individual intention, at least in the same sense that the bassoon player s own playing is the appropriate object of her intentions.

142 142 Stephen P. Turner Collective Intentions as a Form of Individual Intention The novel approach to the problem of conceiving collective intentions without a group mind in a way that has now become standard is owed to Sellars: He classified intentional statements into subsets, and made we-intentional statements into a subset of the class of statements of intentions that might truly be made by individuals, rather than the expressions of a mysterious collective being. This appears to avoid the epistemic problem. By classifying these commitment-statements as statements of individual intention which happen to have a we or collective form, they have the same status as individual intentional statements, although they happen to be collective in form. But even for Sellars, possessing the form alone cannot be enough to make a statement genuinely collective. Something in the statements in the we subset refers to other people, the others covered by the we. So there is an added element, which makes these statements epistemically distinct from other expressions of intentions. For simplicity, let me call this relation real co-intending or S. We-statements of intentions may refer to others in two distinct ways or senses. In one the reference is to the enumerated individual intentions of the members of the we. A claim like we disapprove of women smoking could simply be a descriptive generalization about individual commitments of some set of persons defined by the we. If the reference to others is simply enumerative, there would be nothing added by the concept of collective, because there would be nothing explanatory that was added by the we to a list of Joe disapproves, Bill disapproves, Sally disapproves sentences. The second way, the collective sense of we disapprove of women smoking, is distinct from this and signifies group disapproval, a collective fact. This is the sense that concerns Sellars. In what sense are they distinct? Sellars claims that we-intentional sentences like we disapprove of smoking are irreducible to any set of individual statements of disapproval. One aspect of the difference between collective and enumerated disapproval statements is that it is possible to say we disapprove of smoking, but I don t without contradiction (1963, 198). The first clause of the previous sentence is a collective intentional statement, which concerns the group s intentions or attitudes; the second clause refers to the speaker s own attitude. An enumerated statement would contradict a Bill disapproves statement because the statement people in this group disapprove of women smoking is a generalization about the attitudes of individuals.

143 What do We Mean by We? 143 Sellars s formulation is a way of capturing, in the familiar form of grammatical intuitions, something that is fundamental to the whole notion of collective intentionality. The distinction is familiar from Rousseau: between the volonté generale and the volonté des tous. The latter is enumerative. The former is collective. The volonté generale not only explains some things that the volonté des tous cannot, it may very well conflict with it, and differs from the sum of the individual wills. This implies also that it is epistemically distinct from them that there is a fact of the matter about the volonté generale that is different than the facts of the matter of the individual wills that sum up to the volonté de tous. The underlying distinctions here, in one form or another, run through the entire literature, often in concealed form. 3 An example is the distinction between legality and legitimacy. Legitimacy is an explanatory notion that refers back to the individual beliefs and actions of citizens or subjects. The individuals in question may have a variety of beliefs in the legitimacy of the order in question: some may obey entirely out of self-interest; others may believe in the rightness of the order on theological grounds; others in mixtures of various kinds of beliefs; alternatively they may all subscribe to a single legitimating creed and think uniformly. The fact that they hold the beliefs, and their practical consequence of probable obedience, is an explanation which explains, sociologically, why (and how) a regime is taken to be legal by those ruled by it. It does not, however, explain why it is legal. There is a separate question of legality real legality, so to speak which is not a sociological, but rather a normative question. We can call these facts about the beliefs, intentions, and probable actions of individuals qua individuals the Weberian facts. They can be described in such a way that they contain no reference to anything collective other than in the content of the beliefs of individuals, nor do they make claims about the relations between the intentions of the individuals, such as the claims that they are part of a collective intention. What then, about legality? Ordinary questions about legality, for example, of a command, are answered internally by reference to the rules of the legal 3 The issue of explanatory significance is sometimes obscured by the form that the discussion takes, and sometimes irrelevant to it. Much of the discussion is designed to provide an intuitively plausible definition of some concepts, such as shared co-operative activity, that is taken to be already reasonably well-understood. Definitions may attempt to account for the supposed added explanatory force; other definitions do not. The cases I will be concerned with are, like Sellars, those that purport to deal with added explanatory significance.

144 144 Stephen P. Turner system itself. But the legal system cannot provide the basis for the claim to its own legality, without circularity, for example by stating a law defining itself as legal, since there is an issue of the legality of the law doing the defining that cannot be resolved without reference to itself. In deciding whether a regime is legal, we are thus faced with a choice to say that there is a fact of the matter about basic legality, perhaps a normative fact of some kind, or to accept that there is nothing to legality but legitimacy that legality is an internal, constitutive fiction of the legal system and no more. If we say that legality is a matter of a collective intention that the law of a particular regime is legally binding, we don t avoid this dilemma, we just restate it: is collective intentionality an explanatory notion that goes beyond legitimacy, or is it just another fiction? In the case of the law, the source of the problem is obvious if one describes the legal system normatively, in terms of real legality or real bindingness (or some other feature of the law, such as the conclusive affirmative character of the rule specifying the laws as in Hart 1961, 92) rather than legitimacy (something along the lines of believed to be binding or valid widely enough to continue to be the only reasonably elaborate justification of state action employed in a region ), one has created an explanatory problem that appeals to legitimacy cannot solve. The fact that some members of a group believe something is binding or uniquely valid neither makes it genuinely binding on them or uniquely valid for them. But if the category legal is a normative one, so is the description legal regime. If we describe the same facts non-normatively, legitimacy does the explanatory job: it enables us to account for the outcomes, such as the de facto monopoly of what people in a region believe to be legitimate. This kind of explanation has ordinarily been resisted. If we start with legality, and assume that this is what needs to be explained i.e. that this is the fact of the matter (as the philosophy of law does), the explanation is not satisfactory. A typical set of claims goes like this. The bindingness of the law depends on the consent of the persons bound, on acceptance, or more minimally recognition. Individual consent is not enough the law needs to be binding even on those who decide to break it so there needs to be some notion of collective consent which allows for those members of the collective who do not consent also to be bound. We may leave aside the problem of circularity in the use of the notion of consent, which is in some sense a legal concept in the first place, and focus on what may be thought of as an epistemic problem: what is the relation between the facts of consent and legality? As a practical matter, what

145 What do We Mean by We? 145 counts as the law is not a matter of theory, or group will, but a matter of who speaks for the law. Whether someone a judge or ruler does this legally is a legal question decided by other people who also speak in the name of the law. According to older legal theory, the law was the command of the sovereign, and thus based on his will, but the sovereign was conceived to represent something more than himself. Later, it was the will of the people. This was a model theory of collective intentionality: legality was a matter of collective will. But at every step, it involved fictions, abstractions of a normative rather than a empirical kind, and theories about what was and was not legal that were distinctly local, held by very few people ( a few judges and law scholars perhaps) and subject to multiple interpretations. 4 There are of course plenty of facts as well as fictions: the institutions of the law, courts, police, and jails, are indeed there. The normative fact of legality is a matter of the element of will, or recognition. But it too is largely fictional. What counts as recognition is acquiescence to a set of claims and actions made in the name of law by states and judges, which is to say the facts that come under the heading of legitimacy. Thus there is a large gap between two kinds of facts: the actual beliefs of individuals about the law and the categorically different phenomenon of legality itself, which is resolved by recourse to a fiction, collective consent, which cannot be identified with the actual beliefs of individuals except by appeal to other fictions or conventions, such as the use of elections as a surrogate for collective consent. Why is this relevant? In one sense, the law is the best case of collective willing available to us: laws are written, explicit, knowable in a familiar sense at the term, and univocal. If we encounter problems here, the situation with supposed collective tacit agreements, plans, commitments, and the like that are also unspoken, belong to no authority, and are assumed to be in common is going to be significantly worse. No one denies the fact of laws, at least in the sense of written documents explicitly stating rules. Whether there is a fact at all about such things as collective will is a different matter. 4 The devolution of the will theory of the law to the less demanding notion of a locally accepted role of recognition (Hart 1961, 112) has a history that is not irrelevant here. Axel Hägerström produced a devastating critique of will theories of the law (1953, 1-256) that is acknowledged and to some extend incorporated in Hart (1961, 76, 78). Legal positivism is based in part on an acknowledgment of the validity of this critique, which cuts off the possibility of grounding the law in facts about intention. Hägerström s critique of Kelsen s claim that the legal system has an independent reality, however, is relevant here, for it shows the problems of circularity clearly (1953, 267)

146 146 Stephen P. Turner The legal case suggests that it is a fictional construction, but that it rests on materials of individual belief. What are the Facts of Collective Intention? If a statement about collective intentions is not a generalization of the Sam disapproves, Sam believes in its legality kind, how does it relate to empirical reality? As the history of political appropriations of Rousseau shows, this is not an idle question. The problem derives from the intrinsic structure of Sellars s approach, and holds for any form of the claim that there is collective agency without collective agents. If there is no real collective agent, there is also no real collective speaker who can express these intentions. But one can say that there are speakers who can express these intentions, or whose statements can be assessed as expressions of collective intentions. This is Sellars s approach: individuals are entitled to express we-intentions because they are their own individual intentions, but happen to have the form of we statements. Merely possessing the form, however, cannot be enough. Normally, claims about group intentions, according to Sellars, are related to some ordinary facts about other people: the fewer the people in the group who believe p or intend X do A, the less defensible becomes the statement that the group believes that p or intends that X do A (1963, 203). But less defensible is a weak epistemic term. Neither introspection with respect to one s own intentions nor sociological inquiry about the (enumerated) intentions of other individuals, nor the two together, suffice to warrant claims about group or shared intentions. So what would? In general the problem is this: an individual intention, even one in a collective or we form, needs to have some additional property, call it S, that makes it collective or shared. The fact that an individual intends in a collective mode is not enough to count the intention as a collective intention. And this poses an epistemic problem. If we take intentional statements to be themselves epistemically unproblematic, we are still left with the property S, which needs its own grounding. What would that grounding be? What sort of facts might there be about S? One answer to this, to which I will return, is that there is no deep problem about S. If we can know something about the intentions of others, we can know about such things as their likenesses to one another, their complementarity, and their mutual consistency. We can also know such things as whether universal intentions, such as those expressed by an ought, are

147 What do We Mean by We? 147 shared (cf. Sellars 1963, 204), or whether an intention is a slice of a collective intention (Tuomela forthcoming), or part of a pool of wills dedicated, as one, to a goal (Gilbert 1990, 7). Is this a reasonable conclusion? These terms pool, share, slice, and the like, are of course metaphors, terms of art even sameness and complementarity are terms that are not to be found in the folk psychology of intention. So it appears that we are in a situation with S that is not unlike, and perhaps is parallel to the situation with legality in this respect: If one assumes they state a fact, they define a particular explanatory problem; but there may be no good reason to make this assumption, to describe in this way, or to accept the consequences of doing so. Sellars makes two moves that relate to this question. He acknowledges that we may be deceived about whether there are shared universal intentions in a given case; he also asserts that a person who shares none of the intentions of the group would scarcely be said to be one of us (1963, 203) which seems to imply that sharing intentions is a feature of groups and a logical condition of an us. He also makes the important point that the actual existence of shared universal intentions is [not] a condition of participating in moral discourse (1963, 205). Getting to the point of agreement is a task; intersubjectivity of intention, S or real co-intending, is not a given, but an achievement. It should be observed that this reasoning has now gone very far beyond ordinary language and grammatical intuitions. Not only do we not have a term in folk psychology for talking about S, we have no epistemic guideposts for talking about truth with respect to S, the state of achieving S, and so forth. But we do have some indications of what sort of fact it might be. Sellars has a theory about the capacity for arriving at S, a capacity which Sellars calls a novel form of consciousness, and it is a kind of social theory. The factual precondition for the form of consciousness is the internalization of a concept of a group. A concept of the group is internalized as a concept of us, and this produces a change in consciousness which is or permits, or in Sellars s mysterious term becomes a form of intending (1963, 203). Let me observe the peculiar appeal here to the social science concept of internalization. Internalization a term which metaphorically describes a process which, empirically, means nothing more than responding in a way that shows that the subject has acquired attitudes that are more or less consonant with the group. I note this because it is an early indication of the basic epistemic problem of Sellars s text: a gap between the epistemic base and the terms he employs. Nothing about the process of internalization as it is ordi-

148 148 Stephen P. Turner narily studied for example by testing people s attitudes in response to a questionnaire warrants claims about shared intentions or novel forms of consciousness. If there is, as there appears to be, a huge gap between this sort of evidence and the idea of genuinely shared intentions, why does Sellars suppose that there are such things in the first place, or that there are facts about them? His grounds for thinking that there are such things as shared intentions are logical, not empirical. Sellars characterizes the logical relation between the claims as follows: the truth of Group G intends that people do A in C is a necessary condition of the genuineness of People ought to do A in C said by a member of the group (1963, 204). Genuineness is, as we have seen, analogous to sincerity with respect to individual intentions. But Sellars adds that genuineness is a more complicated matter than the candor or absence of self-deception of an individual. One can know that he intends that people do A in C, and yet be deceived about the group s intention. The group has shared intentions by virtue of the fact that its members intend [in the relevant] mode But the fact that its members intend in this mode does not guarantee that in point of fact there are shared intentions (1963, 205). In point of fact implies that there is some sort of fact (what I have labeled S) about group intention to be deceived about, and thus genuineness cannot be guaranteed by anything in the intention itself. How does the epistemic situation differ from the situation with individual intentions? In the first place, we have no privileged direct, incorrigible access to the basis for them, as there might be for an individual s own intentions. Even a member of the group can be deceived about the group s intention (1963, 205). But to make sense of error in a way consistent with his analysis, Sellars must devise a novel way of talking about it. So he characterizes error in terms of a new concept: the genuineness of the statement. Genuineness takes up much of the epistemic role of sincerity in relation to individual intentions. Sellars gives a few hints about genuineness, which he seems to link to awareness of groupness, as when he says, one who does not intend in the we mode, i.e. has no sense of belonging to the group cannot have more than a truncated understanding of ought (1963, 205). So there can be truncated or incomplete understandings of ought that are not genuine, and they are not genuine because the individual intender has no sense of belonging to a group. But this does not get us very far, because the same epistemic issues arise with the distinctions on which the reasoning rests. What is a sense of belonging to a group and what is it based on? Can it too be mistaken? What epistemic claims can it be said to warrant? It may be

149 What do We Mean by We? 149 observed that if there is no fact of the matter about group intentions other than what is disclosed by (possibly mistaken) tellings, we are worse off with the notion of wrongness and genuineness than we are in the case of individual intentions. In the individual case, at least, what people do is related, though perhaps in complex ways, to the truth of their expressions of intention. Their actions indicate whether they mean what they say. And we have elaborate conceptual resources, including the folk theory of mind and a complex language for dealing with such things as weakness of will that allow us to deploy this evidence, even if in the end it is in some sense also insufficient to do such things as remove all the conflicts between plausible ascriptions of intention. Matters differ in a significant way with respect to S, and indeed with collective intentionality generally. The fact that legal theory relies on fictional constructions of collective will is revealing. Perhaps there is an alternative. And the philosophical tradition itself contains a developed conception of the same material which treats it as something akin to fictions. Philosophers as diverse as G. H. Mead and Slavoj Zizek who have talked about such things as the generalized other or the Lacanian big other have treated them as psychological constructions of individuals that are causally produced in interaction, but not the sorts of things that could be taken as veridical facts. One might think that there was some sort of knowledge of intentions that would allow intentions to be treated in such a way that they could be categorized and assessed for their co-intentionality. This is the thought behind the idea that there is no problem for claims about co-intentionality beyond the ordinary problems we have with individual intentions. But of course knowledge of intentions is not of this kind. It comes in the form of typifications, as in Weber, or is a matter of placing an action under a description, in the phrase of G. E. M. Anscombe. Correct description or typification is nonexclusive: there is no single right description. But the use of the notion of intention in claims about co-intentionality seems to require something more than a non-exclusive description. Moreover, the fact that one can describe the actions of others in a way that makes them co-intentional does not serve to exclude the possibility that one can describe them in a way that doesn t make them co-intentional yet allows for the same collective result. This is precisely the possibility that the sociological, causal concept of legitimacy exemplifies. Legality is not what is explained, but something under a different description, a legitimate order, which is factually indistinguishable from a legal order. What is missing from the description is the (possibly non-factual) notion of legality.

150 150 Stephen P. Turner How does the problem of description bear on collective intentions? Sellars s article begins with the sentence My purpose in this paper is to explore the logic of ought with a view to determining the relation between it and other key terms of practical discourse (1963, 159). The starting point of this account is a concept that is supposed to be already more or less understood and to exist in a more or less stable relationship to other terms, just as the starting point of Hart is the concept of law. One can of course pursue an ontological strategy like this: ontologize such terms provisionally and attempt to provide a theory that grounds them. But how does this kind of ontologization of these concepts relate to ontology in the sense discussed here the claim that the concepts do some necessary explanatory work? Although we may begin, in both cases, with a presumption that there is some necessary role here, and that there would be some sort of damage to the web of belief if these concepts were omitted from our ontology, in the case of explanatory questions we can shed this assumption quickly. There is no a priori reason to think that an explanation can be given under these particular descriptions. Perhaps, as in the case of legality, it is impossible to avoid circularity in explaining them. But this would just mean that in order to explain one needs to place the facts under a different description. It does not mean that whatever odd explanation the facts under these descriptions requires needs to be taken seriously as an explanatory necessity, and ontologized. The epistemic problems with S suggest that collective intentionality is dispensable. It is doubtful that there is a way to get from the data that justify a sense of group membership, a sense that someone is one of us, or of a conviction that we understand others, to claims about S. Sellars does not entertain the thought that there is no S, and therefore no such thing as collective intention: he simply asserts that some intentions are shared that are other than simple individual intentions. The we-statements as analyzed by Sellars, however, require S, that is to say they require a relation between intentions. Something has to give way here. The simplest solution is to abandon the formulation of the problem. The Problem Restated Despite the distinguished philosophical lineage of this problem, it is worth considering whether Sellars simply confused the matter by thinking a) that the enumeration analysis and the collective intentionality analysis were the relevant alternatives for accounting for these statements, b) that the onto-

151 What do We Mean by We? 151 logical model of collective intenders, however problematic it was as a scientific theory (or as ontology), might not nevertheless provide the right analysis of a large class of collective statements, c) that the problem was soluble or even properly understood as a problem of linguistic analysis, d) that the problem of collective intentionality had anything to do with the problem of ought. In the first part of this article I have argued with respect to a) that the fundamental problem is explanatory necessity, and that the burden of proof for the collective intentionality argument is to show that there is some fact of the matter that cannot be accounted for or understood in terms of an explanation that does not appeal to S. I claimed that the gap between the relevant evidence and the fact is large. I suggested that the supposed explanatory problem is an artifact of description, and that we have alternative descriptions that do not require collective intention explanations. In what follows, I will give some grounds for b), c), and d). The relation between the claims is this: many of the puzzle cases that motivate the collective intentionality approach depend on thinking that the collective agent language in question is not metaphorical, so to rehabilitate this language would serve to remove this motivation, and also result in a simpler analysis. But I do not propose to rehabilitate notions of real collective agents as a true explanatory theory, or to ontologize them, but to treat the appeal to them as a plausible expression of a commonplace cognitive situation, i.e. an intelligible, commonplace error. Rather than attack each of these items issues in detail, I will outline an alternative account, and show why it is more plausible than (and avoids the epistemically problematic results of) the anomalization strategy that Sellars pioneered. To see why it is more plausible, however, it is necessary to restate the basic issues. The problem of collective intentionality is a result of three conflicting groups of considerations, each of which has some strong associated intuitions. One is an intuition of analytic philosophy itself that commonplace expressions need to make sense in terms of a relatively simple analysis, and that the ubiquity of an expression is grounds for thinking that it shouldn t be explained in terms of non-ubiquitous facts. We locutions are, as Gilbert insists, pervasive (1996, 19). This seems to indicate that they cannot be simply erroneous, and more generally that they cannot be understood to be based on some sort of theory or ideology or cultural tradition that is itself local. It also seems to imply that the usages cannot be merely analogical extensions of normal use. Each of these approaches to we statements would make analysis too complicated. The second intuition is that the theory of real social wholes, group minds, and the like is false, or worse, metaphysical rub-

152 152 Stephen P. Turner bish, and consequently the explanation of the ubiquitous we usages cannot be found there. The third intuition is epistemic: that the only facts of the matter are what I have called the Weberian ones, that is to say warranted typifications of the actions and beliefs that are the basics for attributions of intentions. And this means that if there is a property S, it has to be accessible in an ordinary way, through human interpretation. We come, through experience of what people say and do, to be comfortable with them, able to explain their actions purposively, predict them reasonably well, and so forth. But nothing in this experience provides grounds for claims about collective intentions in the relevant sense that we have internalized something that entitles us to tell collective intentions, in the sense that it grounds the claims that an S relation obtains between the intentions of others. Where does the conflict with the Weberian intuition arise? Weber s restatement of the problem of legal validity is designed to be factual. We can see that others take an order to be legitimate, that they act in a way that amounts to sufficient obedience for the order to hold a monopoly of legitimacy, and so forth. What we can t derive from these facts is the legality of the order, just as we cannot derive facts about the supposed relation of co-intentionality or S from the same kinds of facts. In each case, we have actions that amount to some recognizable collective outcome: a legal order that is legitimate in the sense that it is believed to be legitimate. And in each case there is an explanation that does not involve or require an explanation appealing to collective intention. I have suggested that the key conflict that the anomalization strategy tries to deal with is between the surface grammar of we-intentional statements, which is parallel to that of I-intentional statements, and the epistemic constraint I have stated in terms of the Weberian evidence available to the speaker, but without accepting the grammatically easiest solution: to treat the we as an agent parallel to the agent making I-intention statements, meaning a real, person-like thing with a mind capable of forming intentions. The anomalization strategy must save the surface grammar by making we statements into expressions of commitments parallel to I disapprove of women smoking. It does so by creating a novel grammatical category in which to place the we-expression, a category that does not get tangled up in the problem of group minds. In the case of Sellars, the novel category is telling. The problem this tries to avoid is the ontologization of the subject, the subject of b) and c). I will suggest that ordinary speech may well include entrenched metaphors which ontologize social objects, and that indeed there are good pre-linguistic reasons for expecting the particular kinds of metaphors that are

153 What do We Mean by We? 153 at stake in this discussion. The right analysis of these usages, I will suggest, is that they are false, but true metaphorically, not that their ubiquity makes them valid. I will handle d) more casually. The problem of morality, of the logical status of oughts, and of the explanatory role of the normative is beyond the scope of this article. However, I will suggest that the problem with ought that Sellars constructed is not an explanatory problem for an ethnographic understanding of morality, and that the approach conflicts with the basic data of ethnography, and indeed with the history of our own moral ideas. The Namath Analysis: A Non-Anomalization Strategy When Joe Namath was asked whether the New York Jets would be victorious over the Baltimore Colts in Super Bowl II, he replied I guarantee it. The promise was shocking, but neither unintelligible nor uninterpretable. Like almost all promises, whether they involve individual or collective outcomes, Namath s guarantee was implicitly conditional on various contingencies that the world would not end, that the game would be played, and, more generally, that the usual sorts of conditions that promissors are held to be accountable for would hold. His guarantee can be understood as follows if my teammates do what I reasonably believe they will do, I promise that I will do well enough for my team to win. At no point in making this promise would Namath have needed to go beyond the Weberian evidence about his fellow players, or make any intentional claim about the collective object the team. Namath was not making a claim about the relations between the intentions of the players, nor claiming to be a spokesperson, telling the collective intentions of the team. There need have been nothing collective about the intentions at all. He could have been as behaviorist as Quine about his teammates and opponents, and merely made a commitment with respect to his own performance, or even less his guarantee would have been sufficiently grounded by any evidence that warranted a prediction that the Jets would win. His promise did not need to be an anomalous promise, in an anomalous collective mode. The sports fans of the day took it, in any case, as his promise, a personal promise, as indeed it was. One may notice some pervasive ambiguities about claims like Namath s. Surface grammar is often misleading. What appears to be a shall statement (Namath could have said we shall win ) may be something less a prediction, to which the speaker s only commitment is epistemic. The same

154 154 Stephen P. Turner statement could have been intelligibly made by a bookie, or a commentator. If it had been, it would have been clear that the commitment to it was epistemic, and the response to a question like how can you guarantee it would be to cite the grounds of the belief. The fulfillment of the prediction may as a matter of causal fact require that various people have particular intentions, though it may not. It could be that the statement is an entirely non-intentional prediction about an object individuated as a team. Or it could be a statement about an object individuated as a team and endowed by the prediction-maker with quasi-intentional properties, such as response to feedback that ascribed, analogically, intentions : such sports phrases as the Cowboys always find a way to win come to mind. One may ask whether there would have been any difference if Namath had said we guarantee it. It would be different if the we was enumerative if every member of the team also guaranteed it. The grounds for that claim would be that each member had in fact done so. These sentences are related causally in such a way that there is a striking phenomenon of interchangeability between them. Such sentences as the team shall win, I guarantee it, we shall win and we intend to win spoken by a team owner or representative all amount to more or less the same thing, in the sense that none of them is likely to be true without the others also being true. This puts such statements as we intend to win in a somewhat different light: The burden is to show that they are not interchangeable for causal reasons, which is to show that there is a causally relevant fact that makes them non-interchangeable. If there is no such fact, there is no special mystery about what we mean by we. If collective intentional statements could be justified by showing that S is causally relevant, this burden could be met. But this returns us to the epistemic form of the problem. If we are unable to find a way to warrant factual claims about S in the first place, we are going to be unable to show that it is causally relevant. Representation: Who Has a Right to Tell The motivation for Sellars s choice of the notion of telling can best be understood if we see it in the larger historical context of the the problem of moral commands or presumptions. The shall nots of the Decalogue are deliverances of God. The epistemic problem is to ascertain whether they are in fact His deliverances, which in this case turns on the question of whether Moses received them from God and spoke correctly for God. When shalls

155 What do We Mean by We? 155 become disembodied, as in Kant, this story breaks down. When Sellars reembodies shalls as we shalls, he inverts the epistemic problem: what was formerly part of the essence of the prescription, the impersonal, becomes so to speak part of the accident; the individual intentions become the essence. The core epistemic problem is the individual s intentions, which he can tell; the secondary problem is the relation between these and the intentions of others, which as we have seen is deeply muddled. To make the parallel between collective and individual intentional telling succeed, it is necessary to claim that genuine expressions in the we-mode are direct expressions of commitments, or primitive rather than derived from some set of ideas about entitlements to speak for. So part of the burden of this analysis is to make S similarly primitive, that is to say independent of theories of various kinds. If we return to the prehistory of these ideas about prescription, it is evident that they were not handled in a theory-free way. Even in the case where a person speaks for an us, when there is a challenge to the authority of a person to make a claim, it is resolved not by an inquiry into the genuineness of the relation of the person to the group, but into the authority of the person to speak for a group, authority that is typically grounded in facts about a person and theories that use these facts to justify claims to authority. There is nothing primitive about such questions of genuineness, and the fact that in political contexts we use other terms, such as legality, points to the reason. One needs a fairly elaborate theory of the nature of the authority relation, the nature of the collectivity, and so forth in order to justify a claim to represent something as legal. Similarly, one needs a theory of representation to justify claims to speak for a group. And here we arrive at a conflict. Theories of representation depend on collective notions the nation, the court, the family, and all the other sorts of things other than persons that one could claim to speak for. When the Pope speaks for the Church, it is an act of a specific kind, constituted by an elaborate theology. When the Queen uses we as the Queen, she is speaking not only as a representative in some abstract anomalized sense of telling, but is a representative in terms of a specific theory and set of rituals and associated beliefs that makes her a representative authorized to do the telling, and the telling itself is not some sort of grammatically unusual act, but a perfectly ordinary linguistic act done by a person believed to have special powers. The argument from the ubiquity of these claims implies that they must be in some sense true, and deserving of a proper analysis a general account for a general linguistic phenomenon rather than an explanatory social theoretical account that treats them as local

156 156 Stephen P. Turner ideological formations, that is to say as part of the content of the beliefs of agents rather than of some sort of general, analyzable relationship. But the beliefs involved in actual claims to speak for include a quite different set of issues. These issues are not simply grammatical. Two things need to be said about these non-grammatical issues. One involves what might be called the ideational, the other the sub-ideational. By ideational I mean to capture the following: many examples of collective claims are not based, as Sellars and others (notably Searle) often imply, on a group sense in some sort of raw, preconceptual sense, but on a fully developed set of ideas about the group a theory, if you wish, about the existence of nations, races, and so on. The fact that these groups and the sense of group consciousness that is connected with them is ideational, that it needs to be taught (and that the teaching precedes the sense ), that there are endless variations on the theories that justify even such a notion as nationhood, suggests rather strongly that these objects are objects by virtue of the beliefs that surround them social constructions, in a banal sense and that analyzing the claims made about them, and in the light of theories about them, ought to proceed not in terms of some sort of alternative theory invented by an analyst, but in terms of the beliefs themselves. Beliefs about the special properties of the persons speaking are not merely ornamental they are the justification for the we and the use of the collective voice, without which the usages are unintelligible. Similarly, the idea that someone can speak for society, indeed the idea of society, is an idea with a history, and not some sort of natural feature of social life. We are more used to the idea of someone in authority speaking for the thing they have authority in relation to, but these sorts of speaking for relations have an ideational and typically theological background as well, for example in such ideas as the head of the family speaking for the family. If we language is pervasive, in short, it is because the various ideas that warrant it are commonplaces. But this does not mean that they are anything but extended metaphors, or that they warrant a general analysis either. The problem of representation and the need for backing by theories of representation puts a different light on Sellars s analysis of the basis of we statements in feelings of groupness. This is simply another theory of representation, that is to say an account that purports to authorize someone speaking for others, for collective entities, and so forth. But it presupposes relevant content: it appeals to ideas about collectivity (hidden in the notion of group sense, in his case) that are historical, questionable, and perhaps mythological, rather than to intrinsic non-local features of language.

157 What do We Mean by We? 157 The Intentional Bias The non-ideational or sub-ideational the level signified in its vestigial form by the term sense in Sellars s group sense is nevertheless relevant to an understanding of the ubiquity of collective language. Begin with some cognitive science banalities. We are probably disposed, at some very basic, pre-linguistic, cognitive level, to the following: to think of the world as being made up of agents, that is to say anthromorphically; to seek agents behind results that could have been intended by agents, and to objectify the agents, imagine them, and identify them with something in the world. This is a disposition that experience validates much of the time. Many of the entities with which we deal are causal systems endowed with feedback. By the same token, many of the facts and objects with which we deal could have been intended (or produced by an entity with feedback). Could have been intended is a category that is very capacious, if one allows for the intentions of God and considers that people do not start out with a strong a priori sense of the difference between that which is caused by intenders and that which is merely caused. When we tell a child who has hurt herself by running into a chair that it is bad, and punish it on their behalf, we enact a little ritual that depends on these banalities. A hurt is a could have been intended fact; a chair could be a causal system endowed with feedback. Only a more advanced knowledge of the world than that of the young child would allow us to comfortably say which is which. If a chair can be so readily personifed, or intentionalized, it is hardly surprising that collective objects are, or that we have many elaborate entrenched agent-centered metaphors for talking about the putative causes of possibly intended outcomes. This is one way in which we might arrive at a sense that there is collective agency, and therefore real collective intenders. But there are others. I observed my own children learning collective concepts, and it was notable that they first employed terms such as team observationally rather than functionally, as in we are on the same team, we are wearing the same color socks. This is suggestive, as is the fact that collective identities so often rest on external identifiers, such as emblems. Personifying the notion of team is a natural next step. Perhaps the fact that collectivities are personified, or more typically analogized to the body, is simply a result of the unsurprising fact that we choose to understand and communicate about the world in terms of personification or bodies because it is easy we are all embodied and are persons: And it is an easy way to speak about abstractions: seeing the Church as the body of Christ or the nation as a mother, for example. These considerations allow us to deal with

158 158 Stephen P. Turner a number of standard examples in the literature. Velleman gives the case of the Dean who asks the department what it intends to do about a vacancy (1977, 29). This is a query in the same class as a query about what the church is going to do about something an entrenched metaphor with a meaning that is sufficiently fixed by virtue of the fact that we have a theory of representation, of what it would be to speak for the department or the Church, that we have an idea of who can give an answer to it and therefore what it means. This gives us an alternative answer to the problem of ubiquity. We are cognitively disposed to see the church, or the nation, or society as something that has a purpose, and has a reality because it has a purpose. 5 But it is not a disposition that we should succumb to to the extent of ontologizing collective purposes. Ontologically, we can place these statements on a par with those of an adult who personifies a chair that they stumbled into as a bad chair. Nevertheless people do make serious claims about collective purposes. What is the status of these claims? One approach is to validate them by drawing a parallel between them and mental concepts, and suggest that there is a folk sociology that parallels folk psychology. If a case can be made for intentions as part and parcel of an indispensable folk theory, it is reasoned, the same case can be made for collective intentions. This parallel, however, is defective. Although it does appear that intentional language is more or less universal (Needham 1972) 6, there is no universal folk sociology. Indeed social concepts vary so radically from totemism, such historically specific ideas as the nation, to a bewildering array of ideas of kinship that what is more impressively consistent are the sources of social metaphors, such as the body and kinship. There is every reason to say that these claims are false or metaphorical (i.e. literally false but with some true implications), and that they have precisely the same cognitive status as claims about God s will, speaking for God, and so forth. It is only a prejudice of analysis that pervasive usages 5 The argument from ubiquity of course cuts both ways. The fact that discussions of the relation I have called S requires elaborate metaphors about the pooling of wills and slicing of intentions indicates that these are not ubiquitous. So the contrast between the collective intentionality analysis and this one is not between one resting on ubiquitous usages and another resting on metaphors, but between one resting on unusual metaphors expressing epistemically dubious facts and one involving ubiquitous metaphors rooted in a commonplace cognitive situation. 6 But see Fingarette for the interesting argument that intention is absent from Confucian Chinese (1972, 37-56). Fingarette also gives a depiction of Confucian morality that is consistent with the comments made here about morality, where he characterizes the Confucian view of community as a holy rite (1972, 1).

159 What do We Mean by We? 159 of this sort cannot be dismissed as false, and that their pervasiveness means that they require a grammatical analysis. In short, there are good cognitive science reasons to expect us to personify society. Knowing them, however, gives us reasons for doubting that the usage points to something explanatory, especially when it appears in such a problematic form as a statement about a collective intention which is pervasively interchangeable with many other claims that lack the problematic implications of the concept. Explaining Obligation: Analytic Theory, Social Theory, or Circular Definition? Consider Gilbert s summary of her analytic argument : social groups are plural subjects; plural subjects are constituted by joint commitments, which immediately generate obligations (1996, 368). In what sense is this an analytic argument? Terms like constitute and generate are systematically ambiguous between produce and define, so it is an important question. If it is a matter of definitions of these terms, the argument is analytic only in the sense that it is circular: obligations are obligations if and only if they result from joint commitments. Presumably this is not what is meant. If we take constitute and generate as causal terms, or something akin to causal terms, this formulation is a rudimentary social theory: a story about how joint commitments simultaneously produce both obligations and social groups. Whether, as a social theory, it is plausible, is a question that is only intermittently addressed in this body of thought, but it is nevertheless never too far away. Obligations, after all, are the subject of explanatory attempts, even theories, in the social sciences, and these attempts compete with the joint commitments explanation, though it is not always clear how, since they are typically directed at other well-known aspects of what is usually understood as morality and normativity that the collective intentionality account ignores. Let me give a simple example of this. The classical discussions of the origins of morality were concerned with primitive morality, in which people believed in the existence of impersonal forces which would punish those who violated norms, such as taboos. Even in much more advanced societies, such as the Potlach communities described in Marcel Mauss s The Gift (1967 [1925]), the obligation to reciprocate was a matter of the existence of impersonal forces that were produced by the act of giving a gift which obliged the

160 160 Stephen P. Turner recipient to give something back. Suffice it to say that an explicit feature of these systems of morality was the insistence on their impersonality, which is to say that they are not understood by their adherents as a matter of commitment, collective or otherwise. Promising, similarly, is historically and ethnographically connected with oath-taking, itself a paradigm case of invoking impersonal mechanical forces of punishment which will operate as a result of a failure to fulfill the terms of the oath. More generally, folk accounts of morality are bound up closely with distinctive beliefs, often about God, about magic, and so forth. These beliefs are the subject of epistemic commitment to beliefs about impersonal forces, rather than the normative commitment to a common intention that is central to this account of morality. Suppose that we take the joint commitments theory as an empirically meaningful account of morality, rather than as a closed and circular scheme of definitions which makes the account analytically true (as would be the case if we tried to save the argument by claiming that concepts of tabu are not really moral ). It would be possible to claim that the impersonality of moral forces that is such an impressive feature of these moral systems is simply an error on the part of the participants, and that what they are conceiving as impersonal forces are merely the reified and misrecognized products of their own joint commitments. Durkheim famously does something analogous to this when he suggests that the worship of God in primitive society is a misdirected form of the worship of society. But Durkheim had a more general account of the nature of the collective mind that would sustain the idea that individuals with their cognitive limitations would fail to grasp the reality of the collective consciousness and misrepresent it to themselves. In the hands of the thinker who has to say well of course moral systems are the result of joint commitments, after all, even the Potlach system would only work if people were jointly (normatively rather than merely epistemically) committed to it matters are different. This now seems to be a claim impervious to evidence, true only by virtue of the meaning of the terms, and thus nonexplanatory. At best it is a definition of morality, and a tendentious one at that. Let me make one final comment. The history of the social sciences is a history of emancipation from the intellectual propensity to intentionalize social phenomenon part of the process that Weber called the disenchantment of the world. If we wish to understand prices in a market economy, for example, we look for a visible hand, and finding none, imagine an invisible hand which we can then explain in terms of the individual decisions of buy-

161 What do We Mean by We? 161 ers and sellers which produce the result that an invisible hand would have produced without anyone collective or not having had an intention to produce those prices. One of the great tropes of nineteenth century statistics was the image of a powerful ruler who mandated the exact number of suicides every year, as though by law. The point, of course, was that there could have been no such ruler but the result, namely a predictable number of suicides, nevertheless occurred, as a result of individual decisions, with no collective agency, and no command. From the point of view of this massive body of intellectual development, the idea of collective agency is an archaism and reversion. To accept such an explanation today is to say that the usual approaches to explaining obligation have failed, and that we need a novel explanation of a different, anomalous, kind one which appeals to explainers that are outside the normal facts of the matter. Surely that point has not arrived. Bibliography Bratman, Michael E Shared Cooperative Activity. Philosophical Review 101: Fingarette, Herbert Confucius: The Secular as Sacred. New York: Harper & Row. Gilbert, Margaret Walking Together: A Paradigmatic Social Phenomenon. Midwest Studies in Philosophy VI, Notre Dame, IN: University of Notre Dame Press. Reprinted in Living Together: Rationality, Sociality, and Obligation, 1996, Lanham, MA: Rowman and Littlefield Publishers, Inc. Gilbert, Margaret Living Together: Rationality, Sociality, and Obligation.. Lanham, MA: Rowman and Littlefield Publishers, Inc. Graham, Keith Practical Reasoning in a Social World: How We Act Together. Cambridge: Cambridge University Press. Hägerström, Axel Inquiries into the Nature of Law and Morals, edited by Karl Olivecrona, translated by C. D. Broad. Uppsala: Almqvist & Wiksells. Hägerström, Axel Philosophy and Religion. London: George Allen & Unwin. Hare, R. M. 1970[1963]. Universal Prescriptivism, reprinted in Readings in Ethical Theory, 2nd ed., edited by Wilfrid Sellars and John Hospers. New York: Appleton-Century-Crofts. Hart, H. L. A The Concept of Law. Oxford: Clarendon Press. Kantorowicz, Ernest H The King s Two Bodies: A Study in Mediaeval Political Theology. Princeton, NJ: Princeton University Press. Kelsen, Hans. 1961[1945]. General Theory of Law and State. New York: Russell & Russell.

162 162 Stephen P. Turner Mauss, Marcel. 1967[1925] The Gift: Forms and Functions of Exchange in Archaic Societies. Translated by Ian Cunnison. New York: W. W. Norton & Co. Originally published in L Année Sociologique, n.s. 1: Needham, Robert Belief, Language, and Experience. Chicago: University of Chicago Press. Searle, John Collective Intentionality and Action. In Intentions in Communication, edited by Phillip R. Cohen, Jerry Morgan, and Martha Pollack. Cambridge, MA: MIT Press: Searle, John The Construction of Social Reality. New York: The Free Press. Sellars, Wilfrid Imperatives, Intentions, and the Logic of Ought. In Morality and the Language of Conduct, edited by Hector-Neri Castañeda and George Nakhnikian. Detroit, MI: Wayne State University Press. Tuomela, Raimo. forthcoming. We-Intentions Revisited. Turner, Stephen P Tradition and Cognitive Science: Oakeshott s Undoing of the Kantian Mind. Philosophy of the Social Sciences 33: Weber, Max. 1978[1968]. Economy and Society: An Outline of Interpretive Sociology, 3 vols., edited by Guenther Roth and Claus Wittich. Berkeley and Los Angeles, CA: University of California Press. Velleman, J. David How to Share an Intention. Philosophy and Phenomenological Research 57:

163 Collective Intentionality, Complex Economic Behavior, and Valuation 163 Collective Intentionality, Complex Economic Behavior, and Valuation John B. Davis Abstract We think, therefore we are. [Shaftesbury, 1963 [1900], vol. 2, p. 275] This paper argues that collective intentionality analysis (principally as drawn from the work of Raimo Tuomela) provides a theoretical framework, complementary to traditional instrumental rationality analysis, that allows us to explain economic behavior as complex. Economic behavior may be regarded as complex if it cannot be reduced to a single explanatory framework. Contemporary mainstream economics, in its reliance on instrumental rationality as the exclusive basis for explaining economic behavior, does not offer an account of economic behavior as complex. Coupling collective intentionality analysis with instrumental rationality analysis, however, makes such an account possible, since collective intentionality analysis arguably presupposes a distinct form of rationality, here labeled a deontological or principle-based rationality. In this paper I depart from the standard view of the individual in economics as an atomistic being to consider the individual as a socially embedded being. There are of course many different ways of understanding individuals as socially embedded; the conception I employ, however, is based on collective intentionality analysis, particularly as formulated by Raimo Tuomela. There is an advantage to economic analysis in doing this. Whereas other views of social embeddedness are holistic, and reason mostly in terms of social entities, collective intentionality analysis is explicitly an account of individuals, albeit in a particular kind of social setting. This makes it possible to compare the understanding of economic behavior that emerges from a collective intentionality analysis of individuals with the understanding of economic behavior associated with the standard rationality view of individuals as atomistic beings. Further, as an account of individuals, collective intentionality analysis also offers a way of understanding the seemingly paradoxical idea that individuals can be socially embedded and yet remain distinct beings.

164 164 John B. Davis The basic idea derives from our understanding of first person plural intentions, or we-intentions. Only individuals form such intentions, just as only individuals form first person singular intentions, or I-intentions, but weintentions effectively embed social relationships in individuals. This contrasts with holist accounts of social embeddedness that rather run the risk of eliminating individuals when they embed individuals in social relationships. Collective intentionality analysis thus allows us to both talk about socially embedded individuals specifically as individuals, and compare their behavior to that of atomistic individuals. Finally, since individuals form both kinds of intentions, combining accounts of behavior understood in collective intentionality terms what I characterize as deontologically rational behavior with accounts of behavior understood in instrumentally rational terms, offers foundations for a complete account of individual economic behavior. I suggest that economic behavior in such accounts should be considered complex. Determining the extent to which individuals are deontologically rational rather than instrumentally rational in economic life seems to be in part an empirical question of the extent to which individuals are active in social settings in which they express themselves in we-intention terms. In the discussion here, I restrict my attention to smaller, relatively cohesive, institutionally well-structured social settings social groups on the assumption that shared intentions are more likely to have specific consequences for individual behavior in these sorts of circumstances than in larger, more diffuse, loosely organized social settings. 1 My argument for this assumption is not that smaller social groups more effectively monitor or discipline individual action this would reduce deontologically rational to instrumentally rational behavior but rather that smaller groups have stronger prospects of producing determinate outcomes, and this reinforces individuals commitment to their shared intentions. Compare, for example, the need an individual feels in a place of employment to act upon relatively well-defined intentions shared with other employees ( we need do our respective jobs to meet our production deadline ) versus the lesser need an individual feels in larger, more loosely organized social settings to act upon vague intentions that may only be weakly shared by others ( we need to do something about inflation ). Focusing on mid-sized social groups, then, essentially operationalizes collective 1 A similar argument regarding the economic consequences of social proximity was made by Adam Smith, though he relied on sympathy rather than shared intentions as the underlying motive force.

165 Collective Intentionality, Complex Economic Behavior, and Valuation 165 intentionality analysis for economics by emphasizing the kinds of behavioral consequences that have been the subject of atomistic individual rationality analysis. Section 7.1 briefly discusses social groups. Section 7.2 then reviews Tuomela s contribution to collective intentionality analysis. In section 7.3 I turn to how the socially embedded individual conception explained in collective intentionality terms involves a view of individual economic behavior distinct from that involved in standard rationality theory. A different view of normative reasoning associated with a collective intentionality analysis of the socially embedded individual conception in discussed in Section 7.4. Finally, section 7.5 offers concluding remarks. 7.1 Social Groups Social groups that are relatively cohesive and well-structured have been extensively investigated in sociology, anthropology, and social psychology. A social group may be characterized as (1) a plurality of individuals tied to one another by (2) some principle of membership that implies (3) a system of individual rights and obligations. A social group is not the same thing as a social category (such as income class, ethnicity, gender, etc.), which researchers employ to classify or group individuals according to a set of characteristics which the researcher selects. Groups, rather, are collections of individuals whose shared characteristics derive from their interaction with one another. Of the characteristics of social groups surveyed in the recent literature, I emphasize the following as particularly relevant to the analysis of mid-sized social groups in economic life: that individuals engage in repeated interaction, that they define themselves as members of a group, that they are defined by others as belonging to the group, that they share and observe group rules and norms, and that they participate in a set of interlocking roles that are central to how the group functions (Cartwright and Zander, 1968, p. 48). One advantage of construing groups in these terms is that it is allows us to say that groups need not operate on a face-to-face basis. Much sociological, anthropological, philosophical, and social psychological research focuses on small groups that do exhibit regular face-to-face contact. Indeed, in the limit a relationship between two people can be seen as a kind of group if those individuals see themselves as being in some type of repeated interaction with one another and observe rules and norms which determine roles for them in the relationship. Margaret Gilbert uses as one of her main examples the idea

166 166 John B. Davis of two people taking a walk together (Gilbert, 1989). Were taking a walk together a regular interaction between two individuals, on the understanding here they would constitute a group. More long-lasting relationships of all kinds between two individuals, then, would also qualify as instances of groups. In economic life, however, groups are generally seen as being larger, somewhat more impersonal, and not infrequently involve limited or even no face-to-face contact (firms, unions, cartels, governments, etc.), and this is the sort of case I focus upon in order to concentrate on individuals social embeddedness. The two-person limiting case kind of group, however, would still be worth investigating in economics to explain recurring market interactions between individuals, where trust relationships are better modeled along the lines of group behavior than in standard atomistic individual terms. But I do not address this sort of case here, in order to avoid needing to include sympathy or empathy as elements or factors in the analysis. 7.2 Tuomela s Analysis The philosophical literature on collective or shared intentionality distinguishes we-intentions corresponding to use of we language from I-intentions corresponding to use of I language. We-intentions are explained as a structure of mutually reinforcing, reciprocal attitudes shared by individuals in a social group. Important contributors have been Bratman (1993, 1999), Gilbert (1989, 1996), Searle (1990, 1995), and Tuomela (1991, 1995). Others, such as Etizioni (1988), have also distinguished I and we thinking, without employing the idea that shared intentionality can be described as a structure of mutually reinforcing, reciprocal attitudes. An advantage of Tuomela s work is its specifically individual focus. Though he sometimes informally refers to intentions being shared, he also emphasizes that this is not meant to literally imply that we-intentions exist in society apart from individual weintentions. Rather, a we-intention is defined as an individual s attribution of an intention to the members of a group to which the individual belongs, based on that individual both having that we-intention and also believing that we-intention is held by other individuals in the same group. That is, I can only use we language that pertains to you and I, if I think that you would similarly apply it to you and I. Thus, expressing we-intentions is a matter of whether there exists a set reciprocal attitudes, not whether there is an actual sharing of attitudes. Indeed, in the limit an individual could have a we-intention that no other individuals have, if that individual were

167 Collective Intentionality, Complex Economic Behavior, and Valuation 167 simply mistaken about others we-intentions. Thus, a we-intention is not a supra-individual group intention separate from the attributions individuals make to groups, and when people use expressions such as, the intentions of the group, this is just a shorthand device for referring to a collection of individual we-intentions on the part of individuals in the group. Tuomela s analysis of what he regards as the 167standard case is as follows. An individual expressing a we-intention assumes that it is mutually believed that the we-intention is held by other group members. Consider the case in which an individual s we-intention is rooted in an attitude ( fear ), which the individual believes other group members also attribute to the group. For an individual A who is a member of a group G, A we-fears that X iff A fears that X and believes that it is feared in G that X and that it is mutually believed in G that X is feared in G (Tuomela, 1995, p. 38). X is mutually believed if not only do I believe others believe X, but they also believe that I believe X. 2 On this basis, A might suppose that group G has some intention reflecting G s fear of X (say, whether the group will avoid some danger). Of course A can only surmise that others in G have the same fear and also that the fear of X is mutually believed by members of the group. The strongest case using the idea that X is mutually believed (a shared belief) would involve saying that the fear that X is iteratively believed by everyone. But Tuomela allows mutual to have strong and weak interpretations, because groups themselves have strong and weak criteria for supposing their members share a belief, attitude, or intention. The main point is that we-attitudes are a group attitude not in the sense that a group over and above its members has an attitude towards something, but in the sense that individuals generally in a group have some such attitude that they express in we terms. Thus saying that they generally have a we-attitude depends not just on the mutual belief condition, but on both conditions which when combined provide us with a reason to suppose that individual members of a group are justified in saying what they (that is, we ) intend. 3 2 Gilbert explains we language in a similar way: A person X s full-blooded use of we in Shall we do A? with respect to Y, Z, and himself, is appropriate if and only if it expresses his recognition of the fact that he and the others are jointly ready to share in doing A in relevant circumstances (1989, p. 199). Gilbert holds that individuals use of we language constitutes a plural subject (1989, pp. 199ff). 3 Tuomela draws on an account of mutual belief that has become fairly standard among philosophers which relies on the idea of a hierarchical set of beliefs iterated across individuals (Tuomela, 1995, pp. 41ff). See Shwayder (1965, p. 257) and Lewis (1969, pp. 52ff) for early formulations.

168 168 John B. Davis Tuomela uses this framework to distinguish between rules and norms, and I refer the reader to his work for a fuller account. Rules are the product of an explicit or implicit agreement brought about by some authority, and used to determine a distribution of tasks and activities to individuals. Rules may be formal and written, such as laws, statutes, regulations, charters, by-laws, etc., or they may be informal agreements between individuals, sometimes orally established and sometimes silently agreed to. In contrast, in the case of norms a network of mutual beliefs substitutes for actual agreements between individuals in determining distributions of tasks and activities across individuals. As with we-intentions generally, mutual beliefs are beliefs reciprocally established between individuals, such that each believes that others have the same belief, and each also believes that others think the same about the others, and so on in a structure of reinforcing, mutually held beliefs. Rules and norms are both understood to have motivational force, meaning that they constitute reasons for action on the part of the individuals who accept them. Indeed rules and norms are typically framed as ought principles, and impose requirements on individuals as members of groups in the form of specific prescriptions for individual action. Formally, individual A feels obliged to do X, because A is a member of the group with a we-intention whose consequence is a rule or norm to the effect, we believe members of the group should do X. But rules and norms are different in virtue of the different means by which they enforce a distribution of tasks and activities among individuals (Tuomela, 1995, pp ). The prescriptive force of rules derives from there being sanctions that apply, whether formal/legal or informal, to those individuals who do not observe them. In contrast to rules, sanctioning with norms takes the form of approval or disapproval on the part of others. Because norms are internalized by individuals, in that they themselves accept them as reasons for acting, individuals apply others potential disapproval to themselves, as when feeling shame or embarrassment. In Tuomela s framework, then, it can be said that rules are the basis for institutions, and norms are the basis for social values. While it is true that many institutions also involve norms, as relatively settled social arrangements institutions generally place greater reliance on rules. In contrast, since social values are rarely rooted in agreements, even informally, they usually place little weight on rules. Rather, social values reflect systems of mutual belief about individuals interaction with one another. Thus when individuals create and/or change institutions, they adopt new rules, and produce new we-attitudes that define group action within an institutional framework that can be characterized in terms of agreements and corresponding sanctions.

169 Collective Intentionality, Complex Economic Behavior, and Valuation 169 When individuals develop and/or influence social values, they adopt new norms, and produce new we-attitudes that define group action within a social value framework based on their mutual beliefs and systems of approval and disapproval. In both frameworks, rules/institutions and norms/social values, we-intentions are the foundation for understanding group action. Individuals thus influence institutions and social values as members of groups, and group action is the intermediate link between individual action and supra-individual institutions and social values missing from mainstream accounts of individuals influence on institutions and social values. Finally, it is important to emphasize that rules and norms can create different types of obligations sometimes implicitly, sometimes explicitly for individuals in terms of how different tasks, rights, and positions apply to different individuals in groups. Tuomela characterizes an individual s position within a particular group in terms of that individual s tasks and rights within that group. An individual s tasks and rights are then further distinguished according to whether they flow from rules or norms operating within the group, that is, whether they are rule-based tasks and rights or norm-based tasks and rights. In contrast, across groups, individuals social positions are understood in terms of the whole array of actions that individuals are required and permitted to do across various economic and social settings. These social positions assign individuals a variety of different tasks whose performance is in each instance protected by rights, where these tasksrights combinations may themselves exist within established modes of implementation that are also understood in tasks-rights terms. The overall framework thus explains individual rights and duties within and across groups in terms of task-rights pairs that ultimately have we-attitudes in groups as their foundation. 7.3 A Revised View of Individual Economic Behavior The behavior of atomistic individuals is understood in instrumentally rational terms, because individual objective functions are defined solely in terms of individuals own preferences. With no basis for action other than their own preferences, and putting aside that they might act out of habit or behave irrationally, atomistic individuals can do nothing other than seek to satisfy their own preferences. In contrast, when we treat individuals as socially embedded, we no longer say that individuals act only on their own preferences, because they also act in accordance with those rules and norms which func-

170 170 John B. Davis tion as ought principles what I have termed a deontologically rational or perhaps a rationally principled type of behavior. But proponents of the atomistic individual conception understand rule-following and norm-observance in instrumentally rational terms. Are ought principles operating in social groups then better explained in instrumentally rational terms? Is instrumental rationality a sufficient explanation of individual behavior? There seem to be three objections to saying that something other than instrumentally rational behavior is involved here. I respond to each objection. First Objection One way in which to argue that observing rules and norms which have the force of ought principles is still fully within the compass of instrumentally rationality is to maintain that the individual becomes subject to constraints additional to those usually assumed in standard constrained optimization analysis, namely, constraints associated with group membership. Though these additional social group constraints further narrow individuals choice sets, individuals would still maximize preferences, suggesting that socially embedded individuals are not significantly different from atomistic individuals. This argument, however, ignores what is involved in saying that individuals observe rules and norms on account of their sharing intentions with others regarding those rules and norms. Shared intentions are those intentions which individuals ascribe to the groups of which they are members. But as intentions, they must stem from individual objective functions rather than constitute constraints on those objective functions. That is, shared intentions are like our ordinary intentions in expressing what individuals choose to do rather than what they are limited to doing. It is true that individuals in groups are more constrained in their behavior as compared to when they act outside of groups. But this type of constraint has an intended aspect, and is consequently different from the usual sort of constraint that is entirely external to the individual. Second Objection A second argument for explaining shared intentions in instrumentally rational terms accepts that shared intentions stem from individual objective functions, but argues that they express individual we-preferences, just as ordinary

171 Collective Intentionality, Complex Economic Behavior, and Valuation 171 individual intentions express an individual s own preferences (or I-preferences). We-preferences have been analyzed by Sugden, and characterized as team preferences (Sugden, 2000; also cf. Bacharach, 1999). Sugden explicitly rejects collective intentionality analysis as developed by Tuomela, Gilbert, and others, on the grounds that it assumes individuals are bound by obligations or ought principles, which he regards as inconsistent with an account of instrumentally rational behavior (Sugden, 2000, pp ). To preserve the latter, he reasons, the former has to go. This implies that rules and norms are things that members of teams prefer to observe rather than believe they ought to observe. Moreover, if we-intentions are really the product of wepreferences, then it seems that it is no longer necessary to say that individuals in teams (or groups) need to be treated as socially embedded, since the obligations or ought principles they observe are what they prefer. Sugden essentially draws this conclusion when he argues that the existence question regarding whether teams and other groups exist (and therefore can act as agents) is independent of the theory of instrumental rationality enlarged to include we-preferences. Were groups thought to be agents over and above their members, there clearly would be a stronger case for saying that their obligations and ought principles were not always preferred by their members. Sugden s argument, accordingly, depends on supposing that we-preferences do not really impose obligations or ought principles on individuals. Why is it, then, that Tuomela and other proponents of collective intentionality analysis claim that this is a necessary dimension of we-intentions? The answer lies in their analysis of shared intentions as sets of reciprocal attitudes across individuals in groups. Though shared intentions are indeed individual intentions, unlike team preferences, which represent only what an individual independently prefers for the team, an individual s shared intention is one element in a set of reciprocal attitudes. Thus, when individuals ascribe intentions to groups of which they are members, this represents not what they prefer to ascribe to the group, but rather what they believe to be the group s intention based on what they believe that they and other group members believe to be the group s intention. On this view, shared intentions imply ought principles, because individuals share an intention over which they have very limited influence. Indeed, this combination of sharing an intention and having it stand over oneself helps explain the particular quality of ought principles as binding precepts that individuals nonetheless embrace. Preferences, by comparison, have but one master, namely, the individual.

172 172 John B. Davis It is true that team preferences do have a shared aspect to them. But absent a set of feedback connections between individuals, such as Tuomela describes for shared intentions, the shared nature of team preferences is simply the result of an accidental alignment of individuals we-preferences about teams of which they are members. Team members may happen to share preferences about the teams they are on. Yet if these preferences regarding the team begin to diverge, there is nothing in the interaction between individuals that brings about an adjustment in individuals preferences regarding the team. Indeed, there are many examples of teams in the real world which operate on the basis of Sugden s team preferences, and as a result break down, simply because individuals are driven by what they prefer rather than by what they believe obligates them. The problem, basically, is that, with we-preferences, just as with ordinary individual preferences, de gustibus non est disputandum. That is, individuals retain their atomistic status, and the teams of which they are members do not exist as teams in the customary sense of the term. Third Objection These conclusions, however, suggests a third argument regarding how instrumental rationality might be sustained vis-à-vis collective intentionality analysis. Suppose that we treat rules and norms as conventions understood as coordination equilibria (Lewis, 1969). Then using evolutionary game theory, individuals can be seen as instrumentally rational players who seek the best possible response to one another s individual strategies (a Nash equilibrium), and rules and norms can be explained as endogenously determined sets of reciprocal expectations. This would allow for a feedback/adjustment process, as operates in collective intentionality theory, but it would not explain this process in terms of ought principles. Rather, following Hume s view of conventions, individuals find it in their interest to conform to rules and norms to which they expect others will conform. There are different ways of explaining why individuals would find this in their interest. Hume relied on sentiments of approval and disapproval, and indeed used this as the basis for his theory of justice. Since a system of justice implies ought obligations, this game theoretic/instrumental rationality framework can also be argued to explain the ought content of rules and norms, whether in moral or pragmatic terms. But in contrast to collective intentionality analysis, ought principles in this instance derive from what individuals find to be in their interest.

173 Collective Intentionality, Complex Economic Behavior, and Valuation 173 In collective intentionality analysis, ought principles derive from shared intentions, and shared intentions are explained in terms of individuals use of first person plural we language. Barring cases of deceit, first person plural we language cannot be explained in terms of first person singular I language, unless one denies elemental differences in human language, and engages in a reductionist sort of reasoning that has no support among linguists. In collective intentionality analysis, the reason that shared intention implies ought principles is that they require a commitment on the part of the individual absent in the case of ordinary intentions that can be expressed in first person singular terms. Thus it seems clear that ought principles that derive from collective intentionality analysis are not reducible to ought principles that might emerge from a Humean framework. But this does not imply that the latter involves an unacceptable account of ought principles, or that this account should be eliminated to produce one of ought principles cast exclusively in shared intention and commitment terms. Rather it seems that both reductionist arguments should be rejected, because neither goes through, and because both are part of the view that thinking about individuals and society can be explained in terms of two inalterably opposed intellectual traditions: methodological individualism and methodological holism. Indeed, both accounts of ought principles arguably have real world foundations. Just as there are teams that operate (often poorly) in terms of individual wepreferences, so there are ought principles based on instrumentally rational behavior. Just as there are social groups that operate (usually more successfully) in terms of we-intentions, so there are ought principles based on individual commitment. My position is that individual behavior is complex in being rooted in both types of intentions. The challenge economists consequently face is in determining both the mix of types of behavior associated with different kinds of intentions, and in properly ascribing each kind of behavior to the correct real world circumstances. Much mainstream economics, because of its adherence to the atomistic conception of the individual, imperialistically imposes instrumental rationality arguments on social settings where it does not apply. In using the wrong explanation in the wrong circumstances, mainstream economists impose thin institutional explanations that overlook how the functioning of some social groups and institutional structures depends upon ought principles stronger than can be explained in instrumentally rational terms. The holist economics tradition, in contrast, has at times been equally imperialistic, though in reverse direction, in using social whole-type explanations in circumstances for which they do not apply. These thick institu-

174 174 John B. Davis tional explanations overlook the extent to which individuals are relatively free of shared intentional experience, as well as individuals need for navigating across social structures. I do not attempt here to set forth a specific account of individual economic behavior as complex. There are a variety of different ways in which the two can be related. 4 But consider an example. Suppose an employee in a business is assigned a set of rule and norm-based tasks associated with doing a particular job. If one rule is to invoice customers by the end of the month, and the norm for how this is to be done is to include in the invoice a complete description of all purchases made by those customers, the individual assigned these tasks is likely still free to perform them in a variety of ways (inquire as to customer satisfaction, pursue follow-up orders, institute new record-keeping practices, etc.). How well individuals do their assigned jobs, then, can be a matter of the extent to which they also act on their own preferences regarding the way a job is best done. They consequently act in an instrumentally rational way when already behaving in a rationally principled manner. 7.4 Normative Reasoning and the Conception of the Individual as Socially Embedded Deontologically rational behavior need not be normative in raising explicitly ethical issues, but it can be. Moreover, a specifically normative deontologically rational behavior might take on a variety of value forms according to the range of values operating in social groups. Thus in contrast to the more narrow normative framework standardly associated with instrumentally rationality analysis, valuation in collective intentionality analysis is complex and multi-dimensional. Following Amartya Sen, we might refer to this enlarged normative framework as a deontic-value inclusive consequentialist framework (Sen, 2001, p. 64). As he explains it, It is neither that the good comes first, and then rights and duties, nor that rights and duties congeal first followed by the good, but that they are linked concepts that demand simultaneous consideration. While considerations of freedoms, rights and duties are not the only ones that matter 4 For one example of how such an explanation might be produced, see Minkler (1999), where a commitment function is added to a standard utility function representation of individual behavior. The individual is said to engage in a two-step iterative procedure with the first step corresponding to a response to group requirements and the second step corresponding to an instrumentally rational maximization of utility.

175 Collective Intentionality, Complex Economic Behavior, and Valuation 175 (for example, well-being does too), they are nevertheless part of the contentions that we have reason to take into account in deciding on what would best or acceptable to do. The issue surely is simultaneity (Sen, 2001, p. 61). Here I address how normative values might arise and operate not just in social groups but in organizations and institutions generally, or, as it has recently been expressed, whether we may treat values as partly endogenous to the economic system, and economic systems and their performance as partly functions of people s values (Ben-Ner and Putterman, 1998, p. xvii; emphasis in original). I begin by contrasting the standard view of how normative values operate in organizations made up of atomistic individuals. Essentially following Hume, the standard account attempts to explain how conventions that lack a normative character in themselves can nonetheless come to acquire the status of moral norms. Moral Sentiments in Organizations and Institutions Hume took a system of justice to be a set of conventions that arise when individuals come to expect one another to behave in regular and predictable ways (Hume, 1888 [1739]). Individuals abide by a system of justice, because they find it in their interest to conform to its rules when they expect others to conform to them as well. The idea that such rules are conventional comes from supposing that there are different possible rules of justice, and those that actually come about reflect a history of contingent interactions between people. Nothing a priori moral underlies actual systems of justice, making them for Hume not partly but entirely endogenous to the economic system. But why, then, should such rules be thought normative in nature? Why should they be thought to be anything more than simply persistent regularities in social behavior akin to other regularities that no one believes have normative content? Hume s view, based on eighteenth century Scottish school psychology of sympathy, was that conformity with such regularities evokes sentiments of approval, and failing to conform with them evokes sentiments of disapproval. When these sentiments become widely shared and become attached to an idea of the general good, he believed they may then be characterized as a moral approval and disapproval. A system of justice, then, is ultimately nothing more than a relatively settled set of conventional expectations between individuals chiefly concerned with their own interest that is reinforced by sentiments of approval and disapproval.

176 176 John B. Davis This conception has been modified and re-developed in recent years by Lewis and Sugden. For Lewis, conventions are coordination equilibria (Lewis, 1969). Coordination equilibria can be explained in game-theoretic terms with players acting on individual strategies to achieve a common expectation regarding which individual strategies offer the best reply to one another (that is, they are Nash equilibria). Hume s psychology of sympathy is replaced by the characterization of individuals in terms of strategies, but any norms that emerge are still conventional and entirely endogenous to the economic system. Sugden similarly explains conventions in terms of individuals expectations of one another conforming to regularities in behavior, but adds a concern individuals are said to have over incurring others resentment as an emotion underlying conformity to conventions (Sugden, 1986, 1989). When this emotion operates widely to reinforce individuals adherence to conventions, Sugden suggests that normative expectations obtain among them (Sugden, 1998). But against this it might be said that the emotion of resentment deserves the label normative as little as Hume s approval and disapproval deserves the label moral. Sugden argues in reply that this criticism misses the point behind providing a Humean naturalistic analysis of values. In such an analysis, the definition of a moral sentiment has to be naturalistic; one cannot then object that some of the sentiments allowed by the definition are not really moral (Sudgen, 1998, p. 84). In my view, such a response is question-begging. The claim that moral sentiments are just that, namely, somehow moral, needs a stronger defense than the suggestion that it should be possible to explain moral values naturalistically, and that therefore there must exist such things as moral sentiments. Indeed, making this sort of argument seems to involve exactly what G.E. Moore famously labeled the naturalistic fallacy (Moore, 1903). Note also, that behind the program of producing a naturalistic account of normative values is closely associated with the aim of producing a positivist interpretation of moral life. Sudgen is explicit about this connection, asserting that economists trained in a positivist tradition must seek to explain normative values without assuming the existence of moral facts (Sugden, 1998, p. 76). A moral fact is a fact about something being right or wrong, good or bad, etc. For example, one might say it is a fact specifically a moral fact that it is wrong to needlessly harm another person. To deny that moral facts exist is to say there is nothing in society or nature that can be described in normative terms as a matter of fact. Normative values, rather, are subjective judgments, and must accordingly be explained in terms of some sort of moral sensibility people exercise and impose on the world. Two obvious

177 Collective Intentionality, Complex Economic Behavior, and Valuation 177 problems arise with this sort of approach. First, the idea that human society can or should be described naturalistically is highly questionable, and has not surprisingly been long contested in the history of social science. Sugden and others in this tradition generally do not explain why a natural science approach to social science might be plausible, and I am skeptical that any good arguments exist on this score. Second, this approach creates a very strong problem for making the transition from is to ought (thus Moore s naturalistic fallacy). Sugden addresses this problem by simply insisting on a re-definition of moral in naturalistic terms. Whether this is a reasonable re-definition of moral, however, depends on whether the account of moral behavior that emerges on these terms captures what we ordinarily think is bound up with the normative. What is it, then, that is most characteristic of the normative domain? We can begin to answer this question by emphasizing the implied content of the moral ought. When ought appears in an expression in a normative way, it indicates the presence of a moral obligation. If individuals do something because they believe they morally ought to, they do so out of a sense of moral obligation operating upon them. There are many ways of understanding what this sense of ought involves, but following Kant (Kant 1948 [1785]), I take the minimum essential idea to be a matter of doing something because it is required irrespective of one s inclinations or desires. But then the idea of acting out of a sense of obligation is not what is involved in acting on a moral sentiment. If one is motivated to respect a norm or convention, because one fears others resentment or disapproval for failing to do so, it is not a sense of obligation but an inclination that operates as one s motivation. Rather fear of others resentment or disapproval has replaced acting out of a sense of obligation. The closest Sugden comes to referring to a sense of obligation, then, is when he argues that one of the virtues of his analysis is that it allows us to consider cases in which normative expectations and self-interest pull in opposite directions cases in which individuals follow conventions even though this is contrary to self-interest (Sudgen, 1998, p. 83). But this is not evidence of acting out of a sense of obligation, since one might well be inclined to observe a convention that was contrary to self-interest. Thus if we take the idea of acting out of a sense of obligation to underlie what is involved in moral ought thinking, a naturalistic, moral sentiments type of approach does not capture what we ordinarily think is bound up with the normative. This gives us good reason to conclude that the Lewis-Sugden type development of Hume regarding how values operate in organizations and institutions as conventions is not an account of distinctively normative

178 178 John B. Davis expectations. This in turn suggests that a naturalistic approach probably cannot explain how normative values arise and operate in organizations and institutions. Thus, since the Humean tradition derives from its starting point from the notion that individuals are naturalistically described as isolated from one another, and generally acting in their own interest, let us rather begin at a different starting point by describing individuals as embedded in social groups in the collective intentionality sense, and then ask how normative values might arise and operate in organizations and institutions. Two questions can be addressed. (1) Does this alternative strategy enable us to talk about what is most characteristic of the normative domain, namely, a sense of obligation that individuals have when they use ought language? (2) Does this strategy provide us an account of the variety and range of different types of relationships between normative values in social life along the lines of Sen s deontic-value inclusive consequentialist framework? Question 1: Moral Obligation in Organizations and Institutions? The first question can be answered by explaining what must be involved in saying that socially embedded individuals, understood in a collective intentionality sense, have we-intentions as well as I-intentions. The Humean framework, by taking individuals as fundamentally isolated from one another, operates exclusively with I-intentions. As previously argued, the use of we language generally creates obligations for individuals and not just in the moral sense. The collective intentionality framework, particularly as developed by Tuomela, shares with the Lewis-Sudgen account of convention the idea that individual expectations are established within a system of reciprocal expectations between individuals. But Tuomela s account is different in that individual expectations have as their object reciprocal sets of we-intentions rather than have as their object the I-intentions implicitly involved in the Humean framework. We-intentions, as previously argued, create obligations for individuals, because the successful use of we by an individual needs to conform to how others use that same we. Outside of a requirement of using language correctly, this obligation does not exist for the use of I. Of course some obligations which individuals recognize are pragmatic, and consequently do not have moral content. But on the interpretation here collective intentionality analysis is not naturalistic, and certainly not motivated by positivistic aims. Thus it is as reasonable to suppose that moral facts exist as to suppose that they do not. From this it follows that some of the obligations

179 Collective Intentionality, Complex Economic Behavior, and Valuation 179 individuals observe are indeed moral in nature. Though the dividing line between pragmatic and moral obligations may often be difficult to draw, and though it may change over time, it seems there are many clear cases of each, and thus fair to say that individuals who form we-intentions and use we language often operate under a sense of moral obligation. So a collective intentionality framework, by operating with a conception of socially embedded individuals rather than atomistic ones, makes it possible to include a sense of moral obligation alongside individual inclination as a form of individual motivation. Turning to the second question above, then, what does the collective intentionality framework and the conception of individuals as socially embedded tell us about the range and variety of normative values in social life and the relationships between them? Question 2: An Expanded Normative Domain? The emphasis on moral obligation thus far has rested on looking at moral obligation as something that particular individuals recognize. But a fuller characterization of the concept needs to see these obligations not just from the point of view of the individuals who have them, but also from the point of view of the individuals to whom they may apply. This suggests that a concept of moral obligation which relies on an externalist conception of the individual, where this is a matter of understanding individuals in terms of their relations to one another, in contrast to a concept of moral obligation which relies on an internalist conception of the individual, where this is a matter of understanding individuals in terms of properties that apply to them independently of their relations to one another (Davis, 2003). An example of the latter is the Pareto efficiency standard, which employs an internalist conception of the individual to explain normative recommendations that judge states of affairs according to whether one person is better off ceteris paribus all other individuals. Externalist-individual normative concepts, it can be argued, just because they emphasize relationships between individuals, generally require that we give attention to a range of normative concepts that go beyond whatever particular normative concepts (say, regarding what is good) might constitute a particular individual s moral view. Thus to give any kind of detailed explanation of the moral obligations that one has to others, one typically also needs to have an understanding of others and one s own rights. But systems of rights are themselves generally embedded in broad social commitments to

180 180 John B. Davis such ideals as freedom, equality, fairness, human dignity, community, justice, etc. Thus employing an externalist-individual type of normative concept typically commits one to examining of an entire range of accompanying normative concerns. This also means that the connections between different normative concerns cannot generally be mapped out with any high degree of precision, making moral questions complex and often ultimately undecidable. The idea of an expanded normative domain may be linked to a collective intentionality understanding of individuals social embeddedness in terms of their involvements in social groups in the following way. Social groups generally have goals that help define them. Thus their members we-intentions often concern consequentialist type of moral reasoning, as when something is regarded as right on account of its helping bring about some outcome desired by the group. But this sort of consequentialist moral reasoning, when it is expressed in we-intention terms, also has independent concepts of moral obligation associated with it. Thus a particular individual using we language in regard to what potential good consequences a group wants to bring about operates both with an idea that what is right is a matter of bringing about the relevant outcome, and also the idea that what is right is a matter of observing obligations upon oneself understood in we-intentions terms. This latter sense of right may draw in turn upon other ideals such as justice, equality, and dignity. This is one way of talking about a deontic-value inclusive consequential reasoning, in which, [i]t is neither that the good comes first, and then rights and duties, nor that rights and duties congeal first followed by the good, but that they are linked concepts that demand simultaneous consideration (Ibid., pp. 64, 61). 7.5 Concluding Remarks This paper does not attempt to explain how instrumentally rational and deontologically rational economic behavior are coordinated. It does suggest, however, that this may in part depend upon the extent to which individuals are active in social group settings in which their behavior has an economic character, that is, where production, exchange, and consumption activities are engaged in. The standard view on the part of proponents of the atomistic individual conception is that behavior in groups can always in principle be decomposed into the behavior of instrumentally rational individuals. But this view has not stood up to scrutiny (cf. Kincaid, 1996), and in any case

181 Collective Intentionality, Complex Economic Behavior, and Valuation 181 such arguments beg the central issue here, namely, that individuals act on we-intentions as well as on I-intentions. Thus the need to explain behavior as complex remains on the agenda. In closing I merely suggest a set of considerations that could figure in the way in which this issue might be addressed. One possible view is that one of the two spheres somehow determines the boundaries of the other. Thus social groups might establish certain domains of activity in which individuals would act in an instrumentally rational fashion. As in the example at the end of the third section, the practices in a business firm, or in a department of one, might be to collectively delegate to particular individuals the responsibility of acting as they would find rational, subject to their observing the boundaries placed on that domain of activity by the group. Alternatively, instrumentally rational individuals might delegate domains of activity in which group considerations are regarded as primary. For example, were heads of households instrumentally rational in the market, they might nonetheless treat the household as a sphere in which customary relationships reflecting we-intentions would prevail. But this general model one sphere determining the boundaries of the other also suggests another model in which behavior in one sphere invades the boundaries and undermines the behavior of the other. For example, individuals may express we-intentions deceitfully, and act in ways that are contrary to them. Alternatively, groups may seek to impose rules and norms on individuals where mutual beliefs are absent. One reason that instrumental rationality theory has been attractive in economics is that having a single model of analysis makes possible a high degree of logical and mathematical determinacy in economic explanation. But the consequences of achieving this precision are that certain types of behavior go unexplained, and possibly that the activity of individuals that is meant to be explained is misrepresented. Collective intentionality analysis constitutes one framework in which these risks might be avoided. The implication of this paper is that a larger framework including that analysis which presupposes that economic behavior is complex is more likely to offer a more adequate account of economic behavior on both counts. Thanks for comments on a previous version of this paper go to Jelle de Boer, Alain Marciano, Jochen Runde, John Searle, and Raimo Tuomela. All errors are mine.

182 182 John B. Davis References Bacharach, M. (1999) Interactive team reasoning: a contribution to the theory of cooperation, Research in Economics, 53: Ben-Ner, A. and Putterman, L. (1998) Values and institutions in economic analysis, in A. Ben-Ner and L. Putterman, eds., Economics, Values, and Organization, Cambridge: Cambridge University Press, pp Bratman, M. (1993) Shared Intention, Ethics, 104: Bratman, M. (1999) Faces of Intention, Cambridge: Cambridge University Press. Cartwright, D. and Zander, A. (1968) Group Dynamics Research and Theory, New York: Harper and Row. Davis, J. (2003) The Theory of the Individual in Economics: Identity and Value, London: Routledge Etzioni, A. (1988) The Moral Dimension: Toward A New Economics, New York: Free Press. Gilbert, M. (1989) On Social Facts, London: Routledge. Gilbert, M. (1996) Living Together: Rationality, Sociality, and Obligation, Lanham, MD: Rowman & Littlefield. Hume, D. (1888 [1739]) A Treatise of Human Nature, L.Selby-Bigge, (ed.), Oxford: Clarendon. Kant, I. (1948 [1785]) Groundwork of the metaphysic of morals, trans. H. Paton, New York: Harper & Row. Lewis, D. (1969) Convention: A Philosophical Study, Cambridge, MA: Harvard University Press. Minkler, L. (1999) The Problem with Utility: Toward a Non-Consequentialist/ Utility Theory Synthesis, Review of Social Economy, 52 (1): Moore, G. (1903) Principia Ethica, Cambridge: Cambridge University Press. Searle, J. (1990) Collective Intentions and Actions, in Intentions in Communication, P. Cohen, J. Morgan, and M.E. Pollack (eds.), Cambridge, Mass.: MIT Press. Searle, J. (1995) The Construction of Social Reality, New York: Free Press. Shaftesbury, A. (1900) Characteristics of Men, Manners, Opinions, Times, J. Robertson (ed.), reprint ed., Gloucester, MA: Peter Smith, Sen, A. (2001) Symposium on Amartya Sen s Philosophy: 4 Reply, Economics and Philosophy, 17 (1), pp Shwayder, D. (1965) The Stratification of Behaviour, London: Routledge. Sugden, R. (1986) The Economics of Rights, Co-operation and Welfare, Oxford: Blackwell. Sugden, R. (1989) Spontaneous Order, Journal of Economic Perspectives, 3: Sugden, R. (1998) Normative expectations, in A. Ben-Ner and L. Putterman, eds., Economics, Values, and Organization, Cambridge: Cambridge University Press, pp

183 Collective Intentionality, Complex Economic Behavior, and Valuation 183 Sugden, R. (2000) Team Preferences, Economics and Philosophy, 16 (2), pp Tuomela, R. (1991) We Will Do It: An Analysis of Group Intentions, Philosophy and Phenomenological Research, 51: Tuomela, R. (1995) The Importance of Us: A Philosophical Study of Basic Social Notions, Stanford: Stanford University Press.

184 184 Seumas Miller Institutions, Collective Goods and Moral Rights Seumas Miller Abstract In this paper I offer a teleological account of social institutions. Specifically, I argue that: (a) social institutions have as their defining purposes or ends the provision of collective goods, and; (b) participants in social institutions have moral rights to such collective goods, and the moral rights in question are individual, and jointly held, moral rights. Introduction In this paper I offer a teleological account of social institutions. Specifically, I argue that: (a) social institutions have as their defining purposes or ends the provision of collective goods, and; (b) participants in social institutions have moral rights to such collective goods, and the moral rights in question are individual, and jointly held, moral rights. Social institutions are often organisations. Moreover, many institutions are systems of organisations. For example, capitalism is a particular kind of economic institution, and in modern times capitalism consists in large part in specific organisational forms including multi-national corporations organised into a system. Nevertheless, some institutions are not organisations, or system of organisations, and do not require organisations. For example, the English language is an institution, but not an organisation. Moreover, it would be possible for a language to exist independently of any organisations specifically concerned with language. However, in this paper my concern is only with social institutions that are also organisations. 1 Such institutions consist of individual persons who occupy roles defined in terms of tasks, and rules regulating the performance of those tasks. On my 1 Note also that uses of the term institution in such expressions as the institution of government, are often ambiguous. Sometimes what is meant is a particular token eg the current government in Australia, and sometimes a type eg governments in general. Restricting the notion of an institution to organisations is helpful in this regard; the term organisation almost always refers to a token.

185 Institutions, Collective Goods and Moral Rights 185 teleological account, normatively speaking, these roles are related to one another in terms of their contribution to the goal(s) or end(s) or function(s) of the institution, and also (usually) hierarchically. 2 The relationship between the constitutive roles of an institution can be referred to as the structure of the institution. Aside from the explicitly stated, or defined, tasks and rules, there is an important implicit and informal dimension of an institution roughly describable as institutional culture. This notion comprises the attitudes, values, norms, and the ethos or spirit which pervades an institution. Culture in this sense determines much of the activity of the members of that institution, or at least the manner in which that activity is undertaken. So while the explicitly determined rules and tasks may say nothing about being secretive or sticking by one s mates come what may or having a hostile or negative attitude to particular social groups, these attitudes and practices might in fact be pervasive; they might be part of the culture. It is important to state a number of things at the outset. First, this is a normative account of social institutions, not a descriptive account; it is an account of what the institution ought to be about, not what it has been or is about. Second, the end or function of an institution is a human good, objectively speaking; it is not a mere aim or want. Third, the notion of good in question is relatively wide, and includes both ethical and aesthetic goods. However, I maintain that the principal kind of good aimed at by human institutions are moral goods. (I use the terms moral and ethical interchangeably). Accordingly, I will for the most part speak of moral goods, or human goods, or simply goods. Fourth, the end or function of an institution is a collective good. Naturally, whether or not a descriptive theory of an institution is warranted depends on empirical facts. Moreover, the falsity of a given descriptive theory of an institution would put pressure on the acceptability of any normative theory of that institution. So a normative theory of an institution is in part dependent on a correct empirical account of that institution. For example, if it turned out that no instance of some institution of a particular kind, eg university, in fact involved to any extent the pursuit of the human good proposed in some normative theory of that kind of institution (say, knowledge) then this would make it implausible to claim that institutions of this kind, nevertheless, in general ought to aim at that good. That said, there 2 The term function is multiply ambiguous. Here I use it to mean end or purpose ie. a state of affairs that is aimed at by institutional actors, albeit not always consciously.

186 186 Seumas Miller is logical space between a correct descriptive account of an institution and a correct normative theory of that institution. This logical space is the distance between the actual state of the institution and the state that the institution ought to be in. 3 It is important to note that on my teleological account an organisation, or system of organisations, whose members pursued an end that they wrongly believed to be a human good could, nevertheless, be an institution. In this connection consider the institution of cannibalism. However, such an institution would be deficient qua institution; deficient in that the end pursued ought not as a matter of objective truth to have been pursued. Indeed, an organisation, or system of organisations, that was morally abhorrent to many of its participants could, nevertheless, be an institution. In this connection consider slavery. Although many slaves abhorred slavery, nevertheless, they participated in the institution of slavery because coerced into doing so. However, once again such an institution would be deficient qua institution; deficient in that the means used (buying and selling of humans) in the service of economic ends ought not as a matter of objective truth to have been used. While the structure, function and culture of an institution provide a framework within which individuals act, they do not fully determine the actions of individuals. There are a number of reasons for this. For one thing, rules, norms and ends cannot cover every contingency that might arise; for another, rules, norms and so on, themselves need to be interpreted and applied. Moreover, changing circumstances and unforeseeable problems make it desirable to vest individuals with discretionary powers to rethink and adjust old rules, norms, and ends, and sometimes elaborate new ones. 4 Inevitably, the individuals who occupy institutional roles are possessed of varying degrees of discretionary power in relation to their actions. These 3 In fact there is a fourfold connected set of normative theory types corresponding to four senses of the term institution : there is the theory of institutions in general ie. universities, police services, governments etc. ; the theory of a particular type of institution eg universities in general; the theory of a particular instance or token of an institution type eg the University of Oxford, and; the theory of a particular instance or toke of an institution type at a particular period of time eg the University of Oxford during the period of the Second World War. My teleological theory or theories can accommodate all four senses. 4 On the other hand, radical revision, especially in relation to proper institutional purposes, might destroy an institution. Some think the contemporary western university is under threat by virtue of a process of radical revision of academic purposes in favour of economic ones.

187 Institutions, Collective Goods and Moral Rights 187 discretionary powers are of different kinds and operate at different levels. For example, senior and middle level public servants have discretion in the way they implement policies, in their allocations of priorities and resources, and in the methods and criteria of evaluation of programs. Indeed, senior public servants often exercise discretion in relation to the formulation of policies. Consider that Gordon Chase, the New York Health Services Administrator, conceived, developed and implemented, the methadone program in New York in the early 1970 s, notwithstanding political opposition to it. 5 Lower echelon public servants also have discretionary powers. Police officers have to interpret rules and regulations, customs officers have the discretionary power to stop and search one passenger rather than another, and so on. Traditionally, members of the so-called professions, such as doctors, lawyers, members of the clergy, engineers, and academics, have enjoyed a very high degree of individual autonomy, notwithstanding their membership of, and regulation by professional associations. In recent times they have increasingly been housed in large bureaucratic organisations in which their professional autonomy has evidently diminished somewhat. I have been emphasising the discretionary powers and the relative autonomy of certain categories of individual institutional actors. However, it is not only the individual actions of institutional actors that are not fully determined by structure, function and culture. Many joint or cooperative actions that take place in institutions are not determined by structure, function or culture. For example, a senior public servant might put together a team of like minded people and they might pursue a specific agenda which is not one determined by the prevailing institutional structure, function or culture, and is even in part inconsistent with them. It should also be noted that legitimate individual or collective discretionary activity undertaken within an institution is typically facilitated by a rational internal structure -including role structure by rational policy and decision making procedures, and by a rational institutional culture. By rational, it is here meant internally consistent, as well as rational in the light of the institution s purposes. Arguably, the police culture in many contemporary police services is not rational in this sense. In particular, a culture of secrecy and solidarity amongst street cops is from the standpoint of this rationality, as opposed to the rationality of some self-interested factions within police 5 D.P. Warwick Ethics of Administrative Discretion in (ed.) J.L. Fleishman et al Public Duties: The Moral Obligations of Public Officials (Boston: Harvard University Press, 1981) p.93.

188 188 Seumas Miller organisations inconsistent with a hierarchical organisational structure preoccupied with accountability. Accordingly, it is likely that many (individual and collective) discretionary judgments will be ones that do not contribute to the realisation of the institution s purposes, even if they do facilitate the narrow self-interest of individuals and factional elements. Aside from the internal dimensions of an institution, there are its external relationships, including its relationships to other institutions. In particular, there is the extent of the independence of an institution from other institutions, including government. One thinks here of the separation of powers between the legislative, executive and judicial institutions in the United States of America and elsewhere. It should be noted that, strictly speaking, independence is not the same thing as autonomy, but is rather a necessary condition for it. An institution possessed of independence from other institutions might still lack autonomy, if it lacked the kinds of rational internal structure and culture noted above. Indeed, internal conflicts can paralyse an institution to the point where it becomes incapable of pursuing its institutional purposes. A university, for example, might enjoy independence from outside institutions, including government, but might be paralysed by internal conflict between staff and students. Such conflict might take the form (in part) of ongoing demonstrations that disrupt classes, and thereby prevent lectures and tutorials being held. Joint Actions and Organisations The underlying conception of social action that informs my account of institutions is not only a teleological model, it is also a cooperation model. 6 Such a model stands in contrast with social theories that emphasise social conflict and social power. In some social and political theorising there is a tendency to become fixated with the power dimension of social action in particular. On this kind of view, social action is taken to be action principally driven by socio-political forces, including national, racial, gender, cultural, ethnic, economic class and institutional forces. We are all members of social groups and social organisations and we are somehow supposedly controlled, or at least causally 6 This conception is elaborated in detail in Seumas Miller Social Action: A Teleological Account (Cambridge University Press, 2001).

189 Institutions, Collective Goods and Moral Rights 189 determined, by social forces permeating these groups and organisations. Moreover, these groups and organisations are locked in conflict with one another. Thus males are supposedly principally engaged in attempting to dominate females, capitalists to dominate workers, whites to dominate blacks, and so on. Human action is understood only, or at least principally, in terms of socio-political power struggles. One such influential contemporary social power theorist is Foucault. Thus Foucault: Power is everywhere; not because it embraces everything, but because it comes from everywhere. 7 The appropriate response here is not to deny the power dimension of human action its importance is a matter of common knowledge but rather to reaffirm a different and contrasting, but nevertheless very fundamental feature of such action, namely social cooperation. There are many kinds of human action which are cooperative. There are joint actions, such as playing in an orchestra or building a house. There are conventions, including linguistic conventions, that enable the collective end of communication. There are institutions and institutional actions of various types. Take an institution of education. This institution has as a collective end the provision of a range of intellectual skills and the acquisition of certain kinds of knowledge. People learn to read, write and count, and they acquire a body of knowledge. So an educational institution serves the collective end of education. Moreover an educational institution relies on interdependence of action, and hence cooperation. The students work at learning on condition the teachers work at teaching, and vice-versa. Enough has been said by way of demonstrating what is in fact rather obvious; joint action, and cooperation in the service of collective ends more generally, are pervasive features of human life in all societies. The central concept in all this is that of joint action. 8 Joint actions consist of the individual actions of a number of agents directed to the realisation of a collective end. A collective end notwithstanding its name is a species of individual end; it is an end possessed by each of the individuals involved in the joint action. However it is an end, which is not realised by the action of any one of the individuals; the actions of all or most realise the end. Examples of joint action are two people lifting a table together, and two men jointly pushing a car. So joint actions can be analysed in terms of individualist notions. 7 Michel Foucault History of Sexuality (vol.1; an Introduction) trans. R. Hurley (New York: Pantheon, 1978) p.93 8 See Miller op. cit. Chapter 2.

190 190 Seumas Miller Moreover, these individualist notions do not necessarily bring with them a normative dimension, other than the minimal normativity of instrumental rationality. By the term, instrumental rationality I mean the rationality of undertaking an action which is the means to one s end; I ought to do x for the reason that x is the means to my end, e. Accordingly, I conclude that joint actions do not necessarily have a normative dimension (other than in this minimal sense). We can further distinguish between different, albeit connected and overlapping, generic kinds of joint action and activity, including communicative, economic, educative, sexual and religious activity. The repetition over time, and duplication in space, of any one of these different kinds of generic joint activity can give rise to a more or less connected, and more or less continuous, stretch of joint activity which I will term a sphere of activity. So the ongoing series of economic transactions across Germany constitutes a sphere of activity. Spheres of activity (by stipulative definition) are regulated by conventions. They thereby take on different specific forms according to the specific conventions which structure them. Often, though not necessarily, they are also regulated by explicit rules, including laws. And they are also often but again, not necessarily regulated by social norms. Here I distinguish between rules, conventions and social norms. 9 A rule is a (typically explicitly formulated) directive issued by some authority or other to undertake a certain course of action in certain circumstances. So laws are a species of rule. A convention is a regularity in action performed by a set of individual agents and directed to a collective end. More precisely, conventions are joint procedures, where a procedure is (roughly speaking) a habit; the automatic repetition of an action in a recurring situation. So a joint procedure is a procedure which is jointly followed because it realises a collective end. Moreover, in my view conventions qua conventions are normatively neutral other than in the minimal sense involved in instrumental rationality. I maintain that conventions are normatively neutral notwithstanding the fact that agents can be held to have failed to conform to any given convention. On my account a conventional regularity in action does not constitute a standard of behaviour to which agents ought to conform. Naturally, an agent by not conforming might fail to realise the collective end that he has; the end which is realised by conformity to the convention. However, that would be a mere failure of instrumental rationality. 9 ibid. Chapters 3 and 4.

191 Institutions, Collective Goods and Moral Rights 191 So conventional actions are not necessarily the content of directives issued by authorities. Nor are they necessarily explicitly formulated anywhere. Thus driving on the left hand side is a convention in Australia, though not in Germany; it is a regularity in action among Australians, which secures the collective end of avoiding car collisions. As it happens, this convention is also a law, though it might not have been. Social norms are regularities that are also norms; agents believe that they have a duty to conform or that they otherwise ought to conform. Such norms include ones respecting and enforcing rights. Here the ought is not that of mere instrumental rationality; it is not simply a matter of believing that I ought to conform because it serves my purpose. Some conventions and all rules are also norms in this strong sense. For example, the convention and the law to drive on the left is a norm; people feel that they ought to conform. This strong (and wide) sense of ought includes but is not exhausted by the so called moral ought. Organisations consist of an (embodied) formal structure of interlocking roles. 10 These roles can be defined in terms of tasks, procedures (in my above sense) and conventions. Moreover, unlike social groups, organisations are individuated by the kind of activity that they undertake, and also by their characteristic functions or ends. So we have governments, universities, business corporations, armies, and so on. Perhaps governments have as an end or goal the ordering and leading of societies, universities the end of discovering and disseminating knowledge, and so on. A further defining feature of organisations is that organisational action typically consists in, what I have elsewhere termed, a layered structure of joint actions. 11 Suppose a number of actions are severally necessary 12 and jointly sufficient to achieve some collective end. Call these actions level two actions. Suppose, in addition, that each of these level two actions, is itself at least in part a joint action whose component actions are severally necessary and jointly sufficient for the performance of the level two action in question. Call these component actions, level one actions. So the collective end of the level one actions is the performance of the level two action. One illustration of the notion of a layered structure of joint actions is an army fighting 10 ibid. Chapter ibid. p.173f. 12 Here I am simplifying for the sake of clarity. For what I say here is not strictly correct, at least in the case of many actions performed by members of organisations. Rather, typically some threshold set of actions is necessary to achieve the end; moreover the boundaries of this set are vague.

192 192 Seumas Miller a battle. The individual members of the mortar squad jointly operate the mortar in the first phase. The set of tanks jointly move forward firing in the second phase. Finally, the set of foot soldiers jointly position themselves to hold the ground vacated by the retreating enemy force. The actions of each of the individual foot soldiers, mortar squad members and individual members of the tank team are level one actions. However, the infantry and the cavalry (the tanks) and the mortar squad jointly act to defeat the enemy. Each of these actions is a level two action. It is important to note that on my (stipulative) definition of organisations they are, qua organisations, non-normative entities. In this respect they are analogous to conventions, as I have defined conventions. So being an organisation is not of itself something that is either ethically or aesthetically good or bad, any more than being a convention is in itself ethically or aesthetically good or bad. 13 I can consistently hold this while maintaining that organisations, as well as conventions, are a pervasive and necessary feature of human life, being indispensable instruments for realising collective ends. Collective ends are a species of individual end; but merely being an end is in itself neither (say) morally good nor morally bad, any more than being an intention or a belief are in themselves morally good or morally bad. While my definition of an organisation does not include any reference to a normative dimension, most organisations do as a matter of contingent fact possess a normative dimension. As was the case with conventions, this normative dimension will be possessed (especially, though not exclusively) by virtue of the particular moral/immoral ends (goods) that an organisation serves, as well as by virtue of the particular moral (or immoral) activities that it undertakes. 14 Further, most organisations possess a normative dimension by virtue (in part) of the norms governing the constitutive organisational roles. More specifically, most organisations consist of a hierarchical role structure in which the tasks and procedures that define the individual roles are governed by norms. It is not simply that an employee in fact undertakes a particular set of 13 I am attempting to use the terms good and bad in a relatively wide sense so as to embrace a range of objective goods, as opposed to things that are merely believed to be good, or simply wanted or aimed at. Such objective goods are principally ethical or moral goods. I use the terms ethical and moral interchangeably. But they also include aesthetic goods. 14 Here and elsewhere in this paper I do not bother to distinguish objective from subjectively felt moral considerations. Suffice it to say that social norms necessarily embody subjectively felt moral considerations, but not necessarily objectively correct ones. See Miller op.cit. Chapter 4.

193 Institutions, Collective Goods and Moral Rights 193 tasks, or tends to comply with the directives of his employer. Rather the employee undertakes those tasks, and obeys the directives of his employer, by virtue of the social and other norms governing the employee s (and employer s) roles. The employee believes that someone in his position has a duty to perform the tasks in question, and believes that the employer has a right to issue directives to his employees. Moreover, this belief is shared and acted upon by the other employees (and employers), and is held by each conditionally on its being held by the others. In short, the behaviour of the employees and employers is governed by social norms. It should also be noted that the social norms governing the roles and role structures of organisations are both formal and informal. If formal, then they are typically enshrined in explicit rules, regulations and laws, including laws of contract. For example, an employee not only believes that he ought to undertake certain tasks and not others, but these tasks are explicitly set forth in his contract of employment. Informal social norms in large part comprise the culture of an organisation. Organisations with the above detailed normative dimension are social institutions. 15 So institutions are often organisations, and many systems of organisations are also institutions. My concern in this paper is with social institutions that are also organisations, or organisations of organisations. I am offering a teleological cooperation model according to which the collective end of social institutions is a collective good. In the next section I turn to a consideration of the collective goods provided by institutions; and specifically the relationship between such goods and the moral rights and duties of institutional actors. Joint Rights to Collective Goods In what follows I am assuming a particular notion of moral rights, and a distinction between human rights and institutional rights. By definition, human rights are moral rights. But not all institutional rights are moral rights. With respect to moral rights I make the following points. First, moral rights generate concomitant duties on others eg A s right to life generates a duty on the part of B not to kill A. Second, human rights, but not necessarily institutional moral rights, are justifiably enforceable eg A has a right not to 15 For an influential related account of institutions see Anthony Giddens The Constitution of Society: Outline of the Theory of Structuration (Polity Press, 1984)

194 194 Seumas Miller assaulted by B, and if B assaults, or attempts to assault, A, then B can legitimately be prevented from assaulting A by means of coercion. Third, bearers of human rights, in particular, do not necessarily have to assert a given human right in order for the right to be violated eg. an infant may have a right to life even thought it does not have the ability to assert it (or for that matter to waive it). With respect to correlative duties a threefold distinction made by Henry Shue is relevant. Shue distinguishes between three sorts of duties that correlate with human rights or at least what he calls basic rights. 16 These are the duties to: (i) avoid depriving; (ii) protect from deprivation; (iii) aid the deprived. Consider (ii) above, the duty to protect from deprivation. Take police institutions. Police are typically engaged in protecting someone from being deprived of their right to life, liberty or property. Moreover, a distinctive feature of policing is the use of, or threatened use of, coercive force. 17 Here the institution of the police is different from other institutions that are either not principally concerned with protecting moral rights, or that do not necessarily rely on coercion in the service of moral rights. Take welfare institutions. There is a human right to a subsistence living, and aiding the deprived (to use Shue s terminology) is a fundamental purpose of welfare institutions. However, aiding the deprived does not necessarily or routinely involve the use of, or threat of the use of, coercive force. Thus welfare institutions are different in kind from policing institutions. Now consider business. Many business organisations do not have the protection of human rights or the aiding of the deprived as a primary purpose; nor should they. On the other hand, moral rights are an important side constraint on business activity. Business organisations should not engage in activities that violate the human rights of workers or other members of the community. So some institutions have as their principal end or function the protection of human rights or the aiding of those deprived of their human rights; but many institutions do not. On the other hand, human rights constitute an important side constraint on the activities of all institutions. Notice that when the protection of a human right or the aiding of those deprived of a human right is the end or function of an institution it is a 16 Henry Shue Basic Rights (Princeton: Princeton University Press, 1996) p See Seumas Miller and John Blackler Ethical Issues in Policing: Contemporary Problems and Perspectives (London: Ashgate, 2003) Chapter 1.

195 Institutions, Collective Goods and Moral Rights 195 collective end. Each member of a community has an individual human right to (say) welfare, if they need it. However, it is only when a certain threshold of need exists that the establishment of an institution takes place; a welfare institution is not created because a single person s right to a subsistence living has not been realised. When such a threshold of need exists, what is required is collective or joint action on the part of many persons. Accordingly, a cooperative enterprise or institution is established that has as a collective end the provision of welfare to the many needy by means of the joint activity of members of the welfare institution. So much for the relationship between social institutions and human rights per se. However, there are a range of moral rights that might be termed institutional moral rights. These are moral rights that depend in part on rights generating properties possessed by human beings qua human beings, but also in part on membership of a community or of a morally legitimate institution, or occupancy of a morally legitimate institutional role. Such institutional moral rights include the right to vote and to stand for political office, the right of legislators to enact legislation, of judges to make binding judgments, of police to arrest offenders, and of patients to sue doctors for negligence. Here we need to distinguish between: (a) institutional rights that embody human rights in institutional settings, and therefore depend in part on rights generating properties that human beings possess as human beings (these are institutional moral rights), and; (b) institutional rights that do not embody human rights in institutional settings (these are not necessarily institutional moral rights, but rather mere institutional rights). The right to vote and the right to stand for office embody the human right to autonomy in the institutional setting of the state; hence to make a law to exclude certain people from having a vote or standing for office, as happened under apartheid in South Africa, is to violate a moral right. But the right to make the next move in a game of chess, or to move a pawn one space forward, but not (say) three spaces sidewards, is entirely dependent on the rules of chess; if the rules had been different eg each player must make two consecutive moves or pawns can move sidewards, then the rights that players have would be entirely different. In other words these rights that chess players have are mere institutional rights; they depend entirely on the rules of the institution of the game of chess. 18 Likewise, (legally enshrined) parking rights, including reserved 18 See John Searle The Construction of Social Reality (London: Penguin, 1995) for an account of rights as mere institutional rights ie rights that are the upshot of constitutive

196 196 Seumas Miller spaces, one hour parking spaces etc. in universities are mere institutional rights, as opposed to institutional moral rights. Let us now focus on institutional moral rights. There are at least two species of institutional (moral) rights. There are individual institutional (moral) rights and there are joint moral rights. Joint moral rights are moral rights that attach to individual persons, but do so jointly. For example, in the context of some institution of property rights the joint owners of a piece of land might have a joint right to exclude would-be trespassers. Take another example, the right to political secession. Arguably, the Kurds in Iraq have a right to secede. But, if this is a right, it is not a right that some Kurdish person has as an individual, even an individual qua member of the Kurdish people. After all, an individual person cannot secede. The right of the Kurds to secede if it exists is a right which attaches to the individual members of the Kurdish social group, but does so jointly. At this point a number of questions present themselves. Firstly, precisely what are joint rights? Secondly, what sort of good are joint rights the rights to? Are they rights to collective goods? Let me now try and answer some of these questions, beginning with the definition of joint rights. 19 Roughly speaking, two or more agents have a joint moral right to some good, if they each have an individual moral right to that good, if no-one else has a moral right to that good, and if the individual right of each is dependent on the individual rights of the others. Let us assume that participation in the institution of the government of Canada is a good. Let us further assume that: each Canadian citizen has a moral right to participate in political institutions in Canada; non-canadians do not have a right to such political participation in Canada; and the right to political participation of each Canadian is dependent on the possession of the right to political participation in Canada of all the other Canadians. In that case Canadians have a joint moral right. Such joint rights need to be distinguished from universal individual human rights. Take the right to life as an example of a universal individual human right. Each human being has an individual human right to life. However, since my possession of the right to life is wholly dependent on rules. 19 The material in this section draws heavily from my book on social action viz. Miller Social Action op.cit. Chapter 7. That chapter in turn relies heavily on an earlier article of mine, Collective Rights in Public Affairs Quarterly vol.13 no

197 Institutions, Collective Goods and Moral Rights 197 properties I possess as an individual, it is not the case that my possession of the right to life is dependent on your possession of that right. 20 However, notice that joint rights can be based in part on properties individuals possess as individuals. The right to participate in political instititutions is based in part on membership of a political and legal community, and in part on possession of the property or right of autonomy. So much for joint rights per se. Let us now consider collective goods. 21 Here there are a number of distinctions to be made. As Raz points out, there are necessarily collective goods and ones that are merely contingently collective. 22 A right of access to a water supply might only be contingently collective. This would be so if when the water supply is cut off, everybody s supply is cut off. But under a different system selective cutting off is possible. By contrast a tolerant society is necessarily a collective good. The tolerance of the society is not something that could be channelled to certain individuals only. Following Reaume, in respect of necessarily collective goods, we can further distinguish between those that an agent can choose not to enjoy himself, and those that he cannot choose not to enjoy. 23 Perhaps clean streets are of 20 Joint rights also need to be distinguished from conditional individual rights. By mutual consent I might have a right to fish in your river on condition that you have a right to hunt in my woods. However, neither I nor you have a joint right; rather we each have a conditional individual right. For one thing, the content of my right brings with it no essential reference to the content of yours. For another, I can unilaterally extinguish your right, as you can mine. 21 There is a further distinction that is typically made in respect of collective goods, namely, the distinction between collective goods that are non-rival in consumption and ones that are rival. A non-rival good is such that the consumption by one person does not reduce the consumption by another. The former are referred to as public goods. It might be held that a good which is not non-rival in consumption must be an individual good. However, I do not accept this. There is the possibility of a good which cannot be produced other than by joint effort, and which is therefore reasonably referred to as a collective good. For example, the Great Wall of China. Secondly, there might be goods which cannot be enjoyed by one person without being enjoyed by others, but which are nevertheless scarce. For example, various kinds of relational goods, such as camaraderie within a circle of friends. Perhaps, beyond a certain optimal point, the larger the circle of friends the less possible camaraderie within he group is. If so, at points above the threshold each new addition to the circle diminishes the enjoyment of available camaraderie. 22 Joseph Raz Rights-based Moralities in (ed.) J. Waldron Theories of Rights (Oxford: Oxford University Press, 1984) p.187. See also his Morality of Freedom (Oxford: Clarendon Press, 1986) chapter Denise Reaume Individuals, Groups and Rights to Public Goods University of Toronto Law Journal vol

198 198 Seumas Miller the former kind and a law abiding society of the latter. A recluse could not be prevented from enjoying clean streets or a law abiding society, but he could choose not to enjoy clean streets by never going out. On the other hand even by staying at home he cannot choose not to enjoy a law abiding society. Let us now to explore the connection between joint rights and collective goods. First we need to get a clearer picture of joint action or activity. Here I rely on the account of joint action provided in the second section of this paper. Sometimes the end realised in joint action is not merely a collective end, it is also a collective good. If so, then a joint right may well be generated. It is easy to see why these agents, and not other agents, would have a right to such a good; they are the ones responsible for its existence, or continued existence. In this connection consider the managers and workers in a factory that produces cars that are sold. Managers and workers in the factory but not necessarily others have a joint right to be remunerated from the sales of the cars that they jointly produced. 24 It is also clear that if one participating agent has a right to the good, then other things being equal so do the others. That is, there is interdependence of rights with respect to the good. Let us look a little closer at this joint right. Firstly, what is the basis of this right? Perhaps the basis of this joint right is simply the fact that each of the participants performed some individual action and the result of their actions was the coming into existence of some good. This is not quite correct. I do not have a joint right in virtue of performing an action which is describable independently of the actions of the other agents. For example, I do not have a right to use a house we jointly built simply in virtue of the fact that I built one of the walls. Rather I have a right in virtue of having contributed to the building of the house by building that wall. My wall building considered by itself would not give rise to a right in respect of the house. For there might not be a house, but only my wall. Secondly, what is the relationship, if any, between joint rights and collective goods? The good is a collective end to which the participants possess a joint right. Accordingly, even if the good considered in itself is not a collective good, each of the individual (jointly held) rights to it is a right to it qua collective good. For example, two mining companies might work together to enable the buying of land and the mining of ore on the land. Assume that the 24 Managers and workers might in effect have waived this joint right by virtue of their employment contracts. But this does not affect the substantive point in relation to their prior joint moral right.

199 Institutions, Collective Goods and Moral Rights 199 deposit is too large and too inaccessible for it to be profitable for one company to go it alone. By prior agreement the two companies might each possess an individual right to sell 50% of the ore produced. But this agreement is something additional to the joint right to the land bought and the ore produced, and indeed presupposes this joint right. Imagine that, contra expectations, the ore produced could not be parceled out in the manner envisaged in the agreement. Perhaps there turns out to be only one large buyer who will not accept two separate suppliers. If so, each individual mining company could still claim an individual (jointly held) right to the ore produced, and therefore legitimately insist on making some different agreement in relation to selling it (or perhaps no agreement). Thus far my discussion of jointly held rights to jointly produced goods has implicitly assumed that the good produced is not constituted by the actions which produce it. But there are joint actions the collective end of which just is those joint actions themselves. And of these, some are cases in which the collective end is a good. Thus the members of an orchestra might have as a (collective) end in itself, simply that they play music together. Here the good aimed at is just the joint activity that realises the good. By contrast, the members of the city garbage collection agency might have as a (collective) end in itself that the city streets are clean, but not have their joint activity of street sweeping and garbage collection as an end in itself. Here the good aimed at the clean streets is not constituted by the joint activity that brings it about. Let us now consider goods that consist in the joint activity undertaken to produce them. Reaume has given consideration to such goods. 25 She claims that such goods are necessarily collective in the sense of goods such that an agent that contributes to their production cannot be excluded from their enjoyment, either by another agent or by the agent himself. She gives as an example friendship. If Fred enjoys the friendship of Mary then Mary enjoys the friendship of Fred. More important for our purposes here are examples such as participation in cultural and educational institutions. Let us call such goods participatory goods. As Reaume points out, examples such as cultural and educational participation are problematic as instances of goods in respect of which persons have joint rights, since it is by no means clear that there is a right to such goods. 26 For such cases involve the production of a benefit which can only be enjoyed 25 Reaume op.cit. 26 ibid.p.9.

200 200 Seumas Miller by those who contribute to the production. Here there are no merely instrumental individual actions performed in the service of a logically separable collective end; rather the joint action is itself the end. But in that case there is no point in claiming a right to enjoy the good, having participated in the activity; having participated, a person has necessarily enjoyed the good. The production of participatory goods as distinct from the enjoyment of these goods, having participated in their production involves moral rights. Specifically, there are rights in relation to participation, and these are conceivably of two sorts. First, there are rights to participate. These can be held against those who are already participating, or against outsiders seeking to prevent an individual from participating, or indeed everyone from participating. Second, there are rights to enforce participation. 27 Are either of these kinds of right joint rights? Rights to participate are joint rights. I have a right not to be prevented from participating, and I have this right jointly with you. Rights to enforce participation are more problematic. There are some joint rights to enforce participation. For example, the members of the community may have the joint right to enforce participation in the war effort. If we assume that the war effort is a good then the war effort is a participatory good. Presumably the war effort is only an instrumental, and not an intrinsic good. Perhaps, parents, and even the members of a community, have a joint right to enforce the participation of children in educational institutions. If so, then this is a joint right to enforce participation in a participatory good that is also an intrinsic good. What of other participatory goods such as culture? Intellectual inquiry, artistic and spiritual expression, appear to be human rights in the sense that they are possessed by virtue of distinctively human properties; they are not possessed by groups or individuals qua members of groups. Moreover, they are rights to participate, and also rights not to be forced to participate. In denying that such rights are collective I do not mean to imply that they are necessarily individual rights, at least in the sense of rights a solitary individual can possess and enjoy simply in virtue of properties he or she possesses as an individual. Rather many such rights are a species of joint right. They are rights individuals possess in virtue of needs they have to form relations, or engage in activities, with other individuals Or rights to enforce non-participation, as in the case of social taboos, such as incest. Here there is a joint right to prevent members of the group from participating in a joint action (incest) which realises a collective end an end which is either intrinsically bad, or at least has bad consequences. 28 It might be argued that these cultural rights are social, as opposed to merely interper-

201 Institutions, Collective Goods and Moral Rights 201 Importantly, for our purposes here, such joint rights in relation to participatory goods are not collective rights in the sense of being rights possessed in virtue of membership of a social group. They are basic human rights to form relations and engage in joint activities with other humans by mutual consent. Consider the attempt of the Eastern European communist regimes to prevent inter-cultural intellectual interaction between academics from universities in Eastern Europe and academics in universities in the West. Consider, also the policy of the government of the Taliban in Afghanistan specifically, the Department for the Promotion of Virtue and Prevention of Vice 29 not only to force women not to show their faces in public they must wear veils but also to restrict their individual freedom to form friendships, especially friendships with males who are not members of their family, or who are not approved of by their fathers or husbands. Here we need to distinguish the exercise of basic interpersonal human rights from the organisations and institutions which might claim to embody them. In some cases such claims seem false. Consider, for example, the socalled cultural imperialism of large western media and entertainment corporations. These corporations are able to ensure that the cultural material produced by themselves is dominant in domestic, international and inter-cultural communication. But it is by no means clear that this material should be considered to be an expression of the culture of a people, as opposed to the consumer product of a few organisations. In other cases an institution might seek to impose its culture on individuals. Such imposition looks to be in violation of human rights. Notwithstanding the above, there appear to be some special cases in which imposition of participatory goods on individuals might be justifiable. The most obvious of these is the imposition of education on children. Other cases sonal. I reject this. The right to communicate is at least in the first instance the right of an individual mind to communicate to another individual mind. Artistic expression can, of course, take a collective form, as in the case of an orchestra. But here first point the right of the individuals to play music together on the basis of individual mutual consent is not dependent on membership of some prior social or cultural group. Indeed, it might be that a musical band sets out to constitute itself of individuals from diverse social and cultural groups for the purposes of generating trans-cultural music. Second, the right of group audiences is logically dependent on the right of an individual person to be an audience. Again, the right to friendship is at least in the first instance a right held between two individuals. 29 See Ahmed Rashid Taliban:The Story of the Afghan Warlords (London: Pan Books, 2001) Chapter 8.

202 202 Seumas Miller might be ones involving institutions which have been overpowered by external, or indeed internal, coercion, and are in need of special protection in the short term in order to rebuild themselves. Perhaps the Russian Orthodox Church is an example of this. Such imposition is at least in principle permissible in virtue of the fact that human rights, including human rights in relation to participatory goods, are not absolute. We have discussed two kinds of joint rights in respect of collective goods that have been jointly produced participatory and non-participatory. In these cases the joint right of participants is at least in part grounded in the fact that they are the ones who produced the collective good. Consider the managers and workers in a factory or the members of an orchestra. Earlier we saw that there are at least some joint rights to collective goods that have not been jointly produced by those that possess the joint right. 30 Such collective 30 Are rights to collective goods that have not been jointly produced necessarily joint rights? I think not. If a good is contingently collective but not jointly produced there seems to be no reason to think a priori that the right to it must be joint. This is because an individual would have a right to the good, even if there were no other individuals. After all, the existence of the good does not depend on the efforts of other individuals.(or in the case where a good was produced by individuals, but not jointly, a single individual could produce the good, the first individual might have a right to it, but the second individual person who could produce the good might not have a right to it in virtue of not possessing whatever properties generate the right. Consider a situation in which there are two agents A and B. A has right to a preventative herbal cure for a life threatening disease only A has. B has the ability to easily prepare the herbal cure, but B does not have the right to the cure since B does not have the disease.) Moreover, if others do exist this does not affect the possibility of any individual s enjoyment of his or her right. Hene there seems no impediment to the exercise of individual rights. What of necessarily collective goods that an agent cannot be excluded from by other agents? Here the same argument holds. Again there is apparently no impediment to the exercise of individual rights. The case of necessarily collective goods that an agent cannot be excluded from either by himself, or by other agents, is more problematic. Could there be an individual right as opposed to a joint right to such a collective good? The point here is that everyone has a right to the good, but if I choose not to exercise my right then you cannot exercise yours. Consider the case of lights on in the dormitory. Either lights are on or they are off. Moreover, everyone views lights on as a good, and an indeed a right. However, some want to exercise their right at times when others do not. Here we seem forced to conclude that there are no individual rights, since if they existed they could not in principle be exercised. Are there any joint rights? Once again there seems to be no joint right, since it is difficult to see what such a joint right would be a joint right to given there is no right that would overrule those who wanted to lights off at a given time. Perhaps the group has a right not to have the lights cut by someone outside the group, if the group can agree. That is, there is a conditional joint right in relation to the lights. The condition is that the group reach agreement. But surely this is better described as a set of individual rights conditional on agreement. After all, given the agreement, I now have a right to lights on, independently of you.

203 Institutions, Collective Goods and Moral Rights 203 goods include welfare assistance provided by welfare institutions and security provided by police institutions. The joint right to welfare assistance and security is grounded in individual human rights. Collective Ends and Deontic Properties I have argued that social institutions are in part defined in terms of their collective ends or functions. I have further argued that when such collective ends are also collective goods then this gives rise to joint moral rights to those goods. That is, functional properties (collective ends) give rise to deontic properties (joint rights) under certain conditions especially the condition that the collective end in question is also a human good. Let us now give more give more detailed attention to deontic properties that attach to institutional roles. Such deontic properties comprise both rights and duties. The sources of deontic properties, such as rights and duties, include the following ones. Firstly, as we have just seen, collective ends that are also collective goods generate joint moral rights under certain conditions. So the normative dimension of institutions involves more than deontic properties, and some deontic properties are presupposed by this normative dimension. Specifically, it involves the ethical or moral worth of the ends or purposes of institutions ie. the kinds of human goods that institutions produce or fail to produce. 31 Accordingly, the definition of an institution will typically include a description of the human good that it purports to produce. For example, governments should provide leadership, universities purport to produce knowledge and understanding, language enables the communication of truths, marriages facilitate the caring and moral development of children, economic systems ought to produce material well-being, and so on. Secondly, explicitly formulated directives issued by properly constituted authorities can give rise to duties to perform certain kind of actions under certain circumstances eg. a military officer instructs a subordinate to under- 31 I have used the terms ethical, moral and human good rather loosely. Arguably, there are human goods eg aesthetic goods, that are not also moral goods. At any rate, I am committed to human collective goods as the ends of social institutions. As it happens, I believe most, if not all such goods, are ethical goods in a broad sense of the term ethical (though not necessarily moral goods in a narrow sense of that term.) I do not have the space here to sort out this issue; hence the conflation.

204 204 Seumas Miller take a military task. That said, I do not accept the proposition that general policies or habits necessarily generate standards of conduct as opposed to mere regularities in behaviour and therefore deontic properties. Thirdly, promises generate obligations to perform the promised actions. This is relevant to institutional actors in so far as they undertake oaths of office and the like. Moreover, part of the moral and/or legal basis of some institutions, such as corporations and governments, might consist in part on contracts or quasi-contracts eg. the so-called social contract between an elected government and its citizens. It should be noted that since promise making is itself a social institution in need of explanation, it cannot be used as the key explanatory device in the explanation of social institutions in general. Fourth, various natural properties that persons have eg human needs, generate moral rights, namely, human rights and correlative duties. Typically, human rights act as side constraints on the actions of institutional actors. However, sometimes but certainly not always the end or function of an institution is the realisation of human rights. For example, police organisations have as one of their basic purpose to provide for the human right to security of life and of person. This multiplicity of sources of deontic properties suggests a pluralist explanatory account of the deontic properties of institutional actors. I am happy to endorse a pluralism of sorts. I am committed to a teleological account of social institutions. On such an account participants in an institution pursue a collective good and this gives rise to deontic properties. However, this (normative) teleological explanation of deontic properties needs to be supplemented by recourse to the other above-mentioned sources of deontic properties eg promises, human rights. Some theorists have argued that the individual or collective imposition of functional properties on persons or objects can in and of itself generate deontic properties. 32 I suggest that it is a non sequitur to move from individual or collective imposition of functional properties to deontic proper- 32 Searle Construction of Social Reality op.cit.. Searle in a later paper ( Reply to Raimo Tuomela Philosophy and Phenomenological Research vol.lvii 1997) claims that the point of the deontic power is to enable the performance of the function. But the function of money viz. exchange can be performed by shells (without deontic status), and the function of surgeons by non-accredited persons with surgery skills. That is, it is false that deontic powers are necessary for the performance of these functions. See also Raimo Tuomela s (with W Balzer) Collective Acceptance and Collective Social Notions Synthese no

205 Institutions, Collective Goods and Moral Rights 205 ties. 33 For it is perfectly possible for a person or an object to have a set of functions and be treated as having those functions without any associated (non-instrumental) deontology. For example, a set of individuals might use a certain sort of relatively rare shell as a medium of exchange, and do so notwithstanding the fact that no-one had any desire to possess these shells independent of the fact that they could be used as a medium of exchange. In this scenario all that is required is that each exchanges shells for goods, and goods for shells, intends to continue to do so, and believes that all the others do and intend likewise. Of course it would add greatly to the stability of this arrangement if these shells were somehow authorised as an official medium of exchange, and if a (rule constituted) system of rights and duties in relation to the exchange of these shells was introduced and enforced. However, such a deontological structure is not a necessary feature of the system of exchange. Moreover, even where a deontological framework has been adopted, the relevant deontic powers do not subsume, take the place of, provide the basis for, or go hand in hand with, the substantive functions of the objects or persons in question. Here by substantive functions I mean the functions which are central and necessary to the object or person being the kind of thing that it is. Consider an incompetent surgeon who is incapable of performing a successful operation on anybody, and who largely avoids doing so, or when he absolutely has to, always ensures that he is part of a team comprised of other competent surgeons and nurses who actually do the work. By virtue of being a fully accredited surgeon this person has a set of deontic powers, including the right to perform surgery, and others have deontic powers in relation to him, including the right that he performs operations competently and with due diligence. Moreover, these deontic powers are maintained in part by (say) the Royal College of Surgeons, his colleagues and the community. However, the surgeon simply does not possess the substantive functions of a surgeon. The deontology is there but the substantive functionality is not. Accordingly, it is simply false to claim that he is a surgeon. If someone cannot perform, and knows nothing about, surgery he cannot be a surgeon, irrespective of whether he has the highest professional qualification there is, is treated as if he were a surgeon, and indeed is widely believed to be the finest surgeon in the land. It might be thought that this kind of example suits my argument by virtue of the physical character of the role of surgeon. But my point can be made in relation to non-physical roles and objects. Consider a philosophical article 33 For an extended treatment of this issue see my Social Action op.cit. Chapter 6.

206 206 Seumas Miller published in a leading journal. This is an institutional fact. Now recall the recent notorious Sokal hoax. Sokal, a physicist decided to write a piece of gibberish and submit it to a learned journal specialising in postmodernist writings. The journal accepted his piece, and published it. So the piece of writing was institutionally certified as being a very high quality philosophical article. Moreover, many who read it believed this to be the case. In reality, however, it was nothing of the kind; it was gibberish and intentionally created by Sokal as such. So the article possessed the relevant collectively imposed deontic properties. However, this piece was not a philosophical article; for it lacked the intellectual properties necessary for it to be so. Accordingly, the collective imposition of deontic properties was not sufficient to generate the relevant institutional fact. Institutional authorisation and collective intentionality cannot turn entirely incompetent surgeons into real surgeons; nor can they turn gibberish into real philosophical articles. Naturally, whether or not deontic properties, including rights, are respected might ultimately turn on the attitude of the community. So general community acceptance of deontic properties, including rights in a weak sense of acceptance is necessary for the maintenance of the deontological framework in that community. So much is trivially true. However, are there other cases of deontic properties especially those involved in authority relations which might be ontologically dependent on collective intentionality. Consider Peter Sellars in the movie, Being There. Sellars plays the role of a gardener who for various reasons begins to be treated by the staff of the President, and ultimately by everyone, as if he were the President of the USA. Eventually, he might even have run for office and be elected. Unfortunately, he has no understanding of the political system, or of relevant policies, and has no leadership qualities whatsoever. Nevertheless, it seems to be the case that the gardener is, or has become the President by virtue of collective acceptance. Institutional authorities are vulnerable to a degree that other institutions, such as language, are not. As Searle points out, the communist government of Russia turned out to have clay feet. 34 Once people chose not to obey its directives, it was finished; it simply ceased to function or exist as a government. However, it is difficult to see how the English language could go out of existence in such spectacular fashion; for it depends on millions of often 34 Searle Construction of Social Reality op.cit. p.91.

207 Institutions, Collective Goods and Moral Rights 207 disconnected communicative interactions between millions of different people. There is a reason for the vulnerability of institutional authorities. In the special case of institutional authorities deontic properties are ontologically dependent on collective acceptance; no collective acceptance, no deontic properties. The point here is not simply that (say) rulers cannot exercise their right to rule if their right to rule is not collectively accepted. Rather a ruler does not even possess a right to rule unless she is able to exercise authority over her subjects. This seems to be a general feature of the deontic properties of those in authority.

208 208 Cristiano Castelfranchi The Micro-Macro Constitution of Power 1 Cristiano Castelfranchi Abstract Our focus is the dialectic relationship between personal, social, collective, and institutional powers; that is the Proteus-like nature of power; how power produces power, how one form of power founds another form of it. Even the magic, count as, performative power of institutional acts is given from the institution to the lay-agent, but hidden is given to the institution by the acceptance and conformity of the mass of people. We provide an ontology of personal powers, deriving from them (plus the interdependencies relations) the most important forms of power at the interpersonal level ( comparative power, power-over, rewarding power, power of influencing, negotiation power, collective power, deontic power ). In the second part, we discuss a more institutional notion of power, the process of empowerment and its relation with permission. Premise How important power is in human relations, organizations and societies does not need many words to be argued. Consequently, power is also a fundamental topic in the social sciences, in particular in sociological theory (starting from classics like Marx, Pareto, Weber), in political science, in organization sciences, in anthropology; but also in economics (negotiation power, position power, theory of oligopolies and monopolies, etc.). On the contrary, Cognitive Science and, in particular, models of artificial intelligence and interaction have systematically ignored this perspective (Castelfranchi, 1991). This is probably due to the fact that the prevailing perspective in AI was individualism and mentalism (Gasser, 1991) that looks far from this social 1 A preliminary version of this work All I know About Power (and Something More) has been presented as an invited talk at 2nd SARA Workshop in Lisbon February 11, I would like to thank Helder Coelho and the other participants for their comments. This work has been developed within the ALFEBIITE European Project: A Logical Framework For Ethical Behaviour Between Infohabitants In The Information Trading Economy Of The Universal Information Ecosystem FET-IST ; and the Italian MIUR- Cofin2001 Project Agenti software e commercio elettronico. I m also very grateful to Raimo Tuomela and to Maria Miceli for their remarks.

209 The Micro-Macro Constitution of Power 209 phenomenon. We will on the contrary show that for a good theory of power also an individualistic and cognitive perspective (bottom-up approach) is necessary, although it is also fundamental to model social relations both in an objective perspective (no awareness, shared knowledge, etc.), as a collective construct and convention, and in a structural, emergent, and macro-level perspective. The focus of this work is the complex and dialectic relationship between personal, social, and institutional powers. This distinction is not new, it is even obvious or commonsensical. For example, within the theory of law, Hohfeld (1913) explicitly distinguished between legal power (of/for doing a legal action) and the physical power needed to carry out the act necessary for the exercise of a legal power. 2 However, first, this multi-layered nature of power is not acknowledged in all the disciplines (for instance sociology typically rejects the idea of a connection between the personal powers and the social ones; see 10); second, in our view there is not a good analytic theory of the personal and interpersonal layers of power and of their not so obvious relationships with the power that we prefer to call institutional (where the legal one is just a sub-case based not simply on conventions, traditions, and social norms but on a system of laws 3 ). Forexample,Hohfeld s term physical power is a bit reductive, since in fact one notonly needs strength, skills, and a working body, but the appropriate action plan and the relevant information, i.e. knowledge, and also the required mental abilities and conditions (motives, awareness, confidence, and so on). We have developed in the past years an extended theory of personal and interpersonal forms of power and on such a basis of the relationships of dependence and autonomy (Castelfranchi, 1995a; Falcone and Castelfranchi, 2003). We will present here this ontology and theory, and we will also attempt to connect it to the theory of institutional actions and of institutional powers and empowerment. The research program of our group has been in fact the modeling of the intermediate layer and the transition between individual mind and action, and the collective and institutional reality, while modeling the cognitive mediators of macro-social and institutional action, and the socio-cognitive 2 He also distinguished between the legal power and the permission to carry out those acts (see later). 3 When considering the institutional aspects, power has been mainly addressed from a normative and legal point of view (authority, responsibility, authorization, etc.). See in particular Jones and Sergot (1996).

210 210 Cristiano Castelfranchi character of pre-institutional, merely interpersonal, informal forms of deontic notions (Conte and Castelfranchi, 1994; Castelfranchi, 1998, 2000). We claim that understanding how collective or macro-social phenomena are implemented in and work through the actions (and thus the mind, but not necessarily the consciousness) of the individual agents is crucial and provides part of their theoretical foundations. Our issues here are: how institutional action and power require and use personal powers; how social pre-institutional powers are built upon and result from personal powers and the objectively emerging interpersonal dependence relationships; how social powers and institutional powers feed back to the personal layer and augment personal powers; how deontic notions (like permission) are related to those levels of power; different kinds and levels of empowerment and their relationships; how individuals (unconsciously) empower institutions (and their representatives) and institutions give power to them. In a sense our hidden issue beyond the inter-layer (institutional-personal) relation is the polymorphic nature of power: how one form of power changes into another. Notice that our categories apply to abstract agents, independent of granularity. This is why it just looks like an individualistic approach. In principle, one might recursively apply the notions of personal powers or the dependence network to collective agents (like groups, organizations, or nations) or to positional entities like roles (endowed with powers and depending on each other Conte and Sichman, 2002). However, we will exemplify the basic level as relative to individual agents (i.e., not composed of autonomous agents). Let us first build our basic personal and social notions, with their dialectics, and then get to the institutional dimension. In the first part, we adopt the bottom-up stance, providing an ontology of individual and personal powers, discussing some paradoxes related to these notions, and deriving from them (plus the interferences and interdependencies among agents) the most important forms of social power at the social interpersonal level. In the second part, we discuss a more institutional notion of power, special powers given by role and representation. We also discuss the complicated circular relationship existing between the power of

211 The Micro-Macro Constitution of Power 211 the institution and the powers of the individual playing its institutional role. All this is far from being complete and formalized (and even fully coherent); it is also a non-experimental (simulative) study. However, we believe that it might be inspiring both: for formal studies and for MAS systems, and for simulation studies: in particular, for modeling the dynamic aspects of the dialectic between the micro and the macro level that cannot be accomplished just with speculative or formal means. 1 The Bottom Up Perspective. Starting from the nonsocial/personal sources of power Contrary to commonsense expectations and to social sciences definitions, Power is not an intrinsically social notion 4, and does not pertain only to the theory of society or at least of social action. Of course, it expresses a very important societal and interactive phenomenon, but it basically refers to the relationship between an agent, its goals and its ability and resources, and is grounded in the general theory of action. The basic notion of power is for us the notion of Power-of (personal power, Latin potest). To have power is a notion relative to a goal. First of all, power is power of doing or achieving something (Parisi and Castefranchi, 1978); second, the meaning that we ascribe to X has the Power-of N is as follows: If X has/had the goal of N, X would be Able and in Condition of N (where N is either doing action A or achieving goal G ). To be able and in condition constitute the basic powers 5, but also the conditional clause is very important because it might imply two crucial facts: that X does not really or fully have the Power-of N if she 6 has no discretion about the use of this power (of the basic powers), i.e. if she cannot (lack of power!) N whenever she wants and likes. that a particular condition (for cognitive agents) for doing something is the possibility of choosing and intending it. 4 There is a correction of this for autonomous agents (see Note 10), but it is better to be clearcutting here in opposing to current sociological stereotype on power. 5 To be interpreted in a broad sense: physical but also mental and social skills. 6 X will be a she while the other agents Y, Z,.. will be he.

212 212 Cristiano Castelfranchi This provides a layered structure of having the power of with some sort of basic powers and meta-power. Let s sketch along these lines our notions of personal (non -social) power to later build on it while following the tradition of Hobbes, but rather far from current sociopolitical view various kinds of social power. This concept that an agent is capable and in condition to do a given action A and/or to achieve a given goal G realizing a given state p has very important non-social consequences. For example, the agent s decision about what candidate intention or active goal to choose as an actual intention to be pursued is based on this; more precisely it is based on the beliefs the agent has about one s own power relative to that goal and action/plan (see Bratman, 1988; Cohen and Levesque, 1990; Castelfranchi, 1996; and 3.). What we claim and want to show is how very important aspects and types of social power are based on and derive from the agent s individual Powerof. In fact, the Dependence of an agent on another (Conte & Castelfranchi, 1994; Sichman et al. 1994) is the result of the lack of power of the first agent relative to a given action/goal, while the other agent has such a power (Castelfranchi, 1991). This makes the former agent dependent on the latter, irrespective of any awareness or decision. We need two notions of power: one is objective i.e. from the point of view of an observer (say the scientist), the other is subjective, i.e. what the agent itself believes and feels about its Power-of (Sichman, 1995). Among cognitive agents (like BDI agents) this is important since there are paradoxical relationships between the two levels. However, we do not want a definition that necessarily includes the agent s awareness in the basic notion of power. This will not be general and applicable for example to reactive or behavior-based agents. 1.1 A Basic Ontology of Individual Powers Let us shortly provide a basic vocabulary or ontology for a vocabulary (for an extended introduction and argument see Parisi and Castelfranchi, 1978, Castelfranchi, 1991). The notion of an a-social, individual Power-of reaching one s own goals is based on

213 The Micro-Macro Constitution of Power 213 a) internal powers or abilities, capacities, action repertoire (skills), knowledge, motivation, etc.; b) external powers : resources, favoring conditions. The agent is endowed with a certain action in her repertoire, she CANDO such an action; and she CAN, by successfully executing such an action, achieve her goal-state p. 7 More precisely: Power-of X A (action) X has the necessary skills (in her repertoire) i.e. she is able to perform action A; she is also in condition to do it, since the necessary external conditions and opportunities hold. Sometimes, the predicate CANDO or CAN are used (including or not external opportunities). Consider the definition in van Linder and al. (1996). Here CAN includes not only the agent s ability but also opportunities; but for our need it is to strong that the definition includes the predicate KNOW i.e. that X also knows to have such a power. As we said this would make the notion of power applicable only to cognitive agents. Notice that this notion of Power-of is context sensitive and dependent, since it incorporates to be in condition to do A, i.e. the external opportunities, means and conditions. To be complete one should specify that X has the Power-of A/p in context C and at time T (which should be an interval, from t to t ). Power-of X p (goal/result) There is an action A in X s repertoire such that (Power-of X A) & this action produces p, & p is the possible goal of some agent. Thus: (Power-of X A) > (Power-of X p) Vice versa, the Power-of p presupposes the Power-of doing some action (at least one) that produces p. Subjective Power-of X A (or Subjective-CAN X A), where X knows to be able and to have the opportunity of/to... Analogously it exists the Subjective Power-of X p. These are the basic powers; the basic constitution of an agent s power. 7 Several logics have been proposed and could be used to formalize these notions: modal, action and ability logics; both in philosophy (for example, Pörn, Balzer),and in AI (for example, Werner, Meyer, Cohen-Levesque, etc).

214 214 Cristiano Castelfranchi One might also distinguish between direct and indirect power. This seems important for some social issues (in particular we will find this distinction necessary for a good notion of empowerment ). In fact one agent could be able to make another agent do something (perform an action or realize a goal) while she is not personally, directly able to do so. 8 As we will see, if an agent X has not (directly/personally) the Power-of doing A and achieving p, but she has the Power-of inducing or obliging (in some way) another agent to do A or to achieve p, then the former gets the power she lacks: she exploits the power of the other, she in fact expand her powers, and acquires by sociality the Power-of A/p. Lack of Power X A/p The notion of Lack of Power is important since the whole theory of Dependence is based on it. When X has not the Power-of A/p, she either lacks some skill, or some resource (is resource-dependent) or some opportunity (for ex. necessary access to knowledge). She is not able to reach some of her goals. It would be possible to measure such a lack of power (and consequently possible Dependence; see later). For example: how many of her goals can X reach by herself? Subjective-Lack of Power X A X believes that he has not the ability or the opportunities of/to (see 3.) 1.2 Beyond Basic Powers: the Intention and Deliberation Components Skills, abilities, know how, and external resources and opportunities represent the basic power, in its Internal and External attribution. But there is 8 It is possible to express this for example with the bring it about that predicate (Pörn, 1970; Jones and Sergot, 1996; Santos et al., 1997; see note 52, with a different symbol). The fact that X makes Y realize p usually implies that X brings it about that p. ExEyp ==> Exp One might have on the contrary also a directly/personally bringing about, where Edir xp ==> Exp since (Edir xp) would be one of the ways for (Exp); but not vice versa: Exp =/=> Edir xp ExEyp ==> Not Edir xp Edir x p ==> Not Ex Ey p

215 The Micro-Macro Constitution of Power 215 a very important part which has to do with the teleological character of action and agenthood, and with the possibility of intending and deliberating. Lack of Power because of Goal-Balance A very important form of Lack of Power (and conversely a fundamental constituent of having the Power-of) is the lack of power due to goal-balance and preferences in decision. If X in context C cannot choose goal G because there are strongly coercive and prevailing urgencies (like fear, or obligations, or hunger), it is not really in condition to pursue G. She does not fully have the Power-of p in those conditions, since she is not free to pursue p. This is important in particular at the social layer, since acting on goal balance (via incentives or arguments) is one of the main ways of conditioning the power of others and for acquiring Power-over them. Power Possess in the true/full power As we saw, the notion of power is relative to a goal or an action A (that presupposes a goal); however, an agent endowed with the power of realizing p (of performing A) doesn t necessarily have that goal. In this case her power is useless for her, and is not even full and complete, since having a goal is a condition for pursuing and realizing it. Thus, X lacks the motivational part of Power-of. Moreover, there is another fundamental part, related to the goal: the possibility and capability to decide about it and to pursue it at one s own discretion. Without this very core of power, power is not complete. We might distinguish between two parts or notions of power: the possession of power (Barnes, 1988) and the basic, possessed power. The possession of power presupposes and incorporates the other 9. The true and full power is the power possession. That is the meta-power consisting in disposing of, deciding about its actual employment; in having control over the resources, action capabilities and conditions. This includes the capacity to decide (motivation, intentionality, awareness of power (see 3.), decision, etc.), social autonomy 10, non-dependence as for 9 Obviously, the possess of nothing is nothing; but, also, having something without deciding about its use, without disposing of it is not having it, is having nothing (as for power). 10 So eventually it happens to be true that the full notion of power for agents is intrinsically and unavoidably social. But not because it applies only to sets of agents (Barnes, 1988) or because it is synonym of influence (Weber, Dahl, etc. see 10.), but because it presupposes autonomy from the others, non-heterogeneous-determination, endogenous control.

216 216 Cristiano Castelfranchi decision (discretion) (Falcone and Castelfranchi, 2000). Without this kernel (control, discretion) power is not true power, since it is incomplete for a true agent. Fig.1 The power of an agent X is of the agent that disposes of it, controls it for its own purposes; this can be X herself or another agent. (see 5. on power circulation, etc.). Let us now build social power relations on such a simple basis, starting from the objective level. 2 Towards Sociality: From Personal Powers to Social Dependence Sociality presuppose a common world, hence interference (Castelfranchi, 1998): the action of one agent can favor (positive interference) or hamper/compromise the goals of another agent (negative interference). Since agents have limited personal powers, and compete for achieving their goals, they need social powers. They also compete for resources (both material and social) and for having the necessary power for their goals. Dependence Dependence is an objective social relation between at least two agents. It is objective in the sense that it holds independently of the involved agents awareness and wants. It is the combination of lack of Power-of of one agent (relative to one of its own goals) and the corresponding Power-of of the other agent. It is the result of some interference between the two. Dependence X Y A p = Lack of Power X A/p & Power-of Y A/p that is X needs Y s action or resource to realize p.

217 The Micro-Macro Constitution of Power Dependence Relations among Agents Dependence constitutes the basis of all social interaction, the reason for Society; it motivates cooperation in its different kinds. There are several typical dependence patterns. In (Castelfranchi, 1991) the OR-Dependence, a disjunctive composition of dependence relations, and the AND-dependence, a conjunction of dependence relations, were distinguished. To give a flavor of those distinctions let us just detail the case of a two-way dependence between agents (bilateral dependence). There are two possible kinds of bilateral dependence: Reciprocal Dependence, in which A depends on B as for his goal p, and B depends on A as for her own goal q. They need each other s action, but for two different personal goals. This is the basis of a pervasive and fundamental form of human (and possibly artificial) interaction: Social Exchange, in which A performs an action useful-for/required by B for q, in order to obtain an action by B useful for his personal goal p. A and B are not co-interested in the fulfillment of the goal of the other. Mutual Dependence, in which A depends on B as for his goal p, and B depends on A as for the same goal p. They have a common goal, and they depend on each other as for this shared goal. When this situation is known by A and B, it becomes the basis of true Cooperation. A and B are co-interested in the success of the goal of the other (instrumental to p). A helps B to pursue her own goal, and vice versa. The Dependence network determines and predicts partnerships and coalitions formation (Ketchpel, 1995; Shechory and Kraus, 1995), competition, cooperation, exchange, functional structure in organizations, rational and effective communication, and negotiation power( 4.3). Dependence has a quantity or degree. This quantity derives from: a) the value of the goal p for X and its necessity; b) the degree of power of Y as for p; c) the number of goals for which X depends on Y; d) the number of alternatives (OR-Dep links) at X s disposal as for reaching p; (see 4.3) See also Emerson (1962) the Dependence of an actor A upon actor B is (1) directly proportional to A s motivational investment in goals mediated by B, and (2) inversely proportional to the availability of those goals to A outside of the A-B relation.. The

218 218 Cristiano Castelfranchi e) in case of Reciprocal Dependence there is also a problem of Differential Power: is X more dependent on Y than Y is dependent on X? Another very interesting notion (deriving from Durkheim) is that of cohesion defined as the degree of interdependence (see also Emerson, 1962); the maximum degree would occur when given a group of N agents (each with a single personal goal) each agent in the group has N-1 agents depending on her. This is also a very balanced situation. But as we will see there are very unbalanced situations of interdependence. We will not develop here the quantification of Dependence (and power). 3 The Paradox of Power and Dependence Among Cognitive Agents At a first sight, since dependence is objective and independent of subjective awareness, and the agent can know or simply believe it, the situation is rather clear, like in the following figure: Fig.2 However, this is too simple; there are in fact strange effect of one s beliefs on one s own power and thus on one s dependence relationships: if X is a cognitive agent and believes that she does not have some power, if X does not know that she has that power, she really lacks it because of this misbelief. So, this belief is a self-realizing assertion. importance of Dependence relations is well established both in social theory (Emerson; Thibaut and Kelley; the tradition of social networks (see the very good special issue by David Miller, 1992 and 1992b; Gould,1993; Markovsky et al. 1993) and in MAS and social simulation (Luck; Yu; Ossowsky; and in particular: Sichman, 1995; Conte and Sichman,1995; Conte and Sichman, 2002).

219 The Micro-Macro Constitution of Power 219 Not (Subj-Power-of X A/p) > Not (Power-of X A/p) To fully have some power one should be aware of it (see Figure 1.). Correspondingly, if X believes that she is dependent on Y (provided that his belief about the power of Y is correct) it is really dependent on Y, because of this belief. Notice that the produced Dependence is an Objective one, because - as we will argue he really needs the action of the other agent. This depends on a very special and important feature of Cognitive Agents: the fact that they base their goals and decisions on what they believe (Bratman, 1988; Cohen and Levesque, 1990; Castelfranchi, 1996). More precisely in order to decide to do something, to intend something, and then to pursue (and to achieve) a goal, they have to believe to be able to do what is needed, or at least not to believe to be unable ( let s try ). If they do not believe so (i.e., if they believe they don t have the Power-of) they cannot 12 and will not do the action, pursue (and achieve) the goal. Thus, if they do not believe they can, they in fact cannot! This is not true for any kind of purposive system or agent. In fact more simple goal-oriented or even goal-directed architectures are conceivable, where the selection of the goal and the presence of the appropriate (associated) action in the repertoire (ability) is enough for executing it, and having the Power-of that action/goal. 13 Simple animals and simple animats do not need to believe to be able in order to be able. In cognitive agents one of the pre-conditions for the execution of the action is paradoxically that the agent believes to be able to do it. From what we argued it also follows that: if a cognitive agent believes to be dependent on another it is so. Or better this is true at least for the first belief component of believing to be dependent: (Bel x (Not (Power-of x A/p))) & (Bel x (Power-of y A/p)) If it is true that (Power-of y A/p) then the first belief -although perhaps initially false makes X actually dependent. In a sense, we should include -for 12 Notice that this requires another definition of CANNOT: it is not simply equal to not having A in his action repertoire. 13 Thus, for example in a logic of action and agency while it is correct to define the CAN operator as Knows x Able x A (van Linder et al. 1996) if we want to model BDI or in general cognitive agents, this cannot be the general notion of CAN, and a weaker definition is needed for non-cognitive agents, where CAN just means that the agent has such an action in its repertoire and the opportunities/conditions for that action hold.

220 220 Cristiano Castelfranchi cognitive agents the fact that X knows that she is able to do A, among the pre-condition for the actions execution; if X ignores that she is able to do A, a crucial condition for performing that action is lacking. 14 There seems to be a contradiction between on the one side including mental and motivational processes among the internal powers that give X the Power-of A/p, and on the other side claiming that Power is a potential notion, not necessarily on act ( 10.); X has the power even while not doing A. It is true that to be executive X s power must converge with X s intention ( If X would Intend to do A, she could (would be able and in condition to) ). But we also need a potential notion that does not imply X s actual intention to do A. As we saw, what is enough for X s Power-of A/p is that X can intend to do A, she is able (mental skills) and in condition (not necessarily lacking the possibility of choosing A/p: goal-balance lack of power). 4 Towards Sociality: From Personal Powers (plus Social Dependence) to Social Powers There are several forms of social power among humans and animals (and several of them will be important also in virtual and artificial societies). We will consider only four forms: comparative value; Power-over and its species of incentive-power and rewarding-power; influencing power and its species command power and negotiation power. 4.1 Social Comparison and Hierarchy of Value X s Power-of A/p already starts to be social for the mere fact of being compared with Y s Power-of A/p. If this power is quantifiable and quantifie 15 X s power-of A/p can be greater than that of Y s. And in general X s powers (in a given domain) could be better than Y s powers. This contributes to X s market value. This also determines when believed by others X s status in the symbolic hierarchy of his society ( who is the best as for ), i.e., his social 14 The mere belief is sufficient for attempting but is a cause of failure. 15 We didn t introduce the idea of more or less Power-of A/p. A degree of power can be established on several dimensions. For example, the probability of X s action to achieve p, or the probability of X s ability to be successfully employed, or the probability, etc. But also a possible degree of achievement of p when p is a scalable goal; or the number of different skills/resources X has for p; etc.

221 The Micro-Macro Constitution of Power 221 reputation and image. This evaluation determines whether X will be preferred or not as a partner in social interactions (Miceli and Castelfranchi, 1989). X s reputation, image, value-position is a fundamental social power and greatly improves X s personal power. Position and reputation are capital ; they reflect power and give power. There is a hard competition for this form of power among social agents. Negotiation power ( 4.3) is in a sense a more complex and rich form and an application of this power (although it can be defined also objectively, while we defined this power in terms of agents beliefs). 4.2 Power-over (Rewarding and Incentive Power) Power-over is due to Dependence. More precisely it is just a facet of the social dependence relation; it is Dependence from the perspective of the other agent. In fact, Lack of Power X A/p & Power-of Y A/p implies Y s capability (power) of letting X realize her goal p (Positive powerexercise), or of preventing her from reaching it (Negative power-exercise). In other words, when X is Dependent on Y for satisfying her goal p, Y gets some Power-over goal p of X, thanks to his Power-of performing A. One might prefer to call the power of Y over X not the mere objective fact that X depends on Y, but the fact that Y is aware of this (and thus can use it). The awareness of this relation gives him such power (cf. also 3.). This is a reasonable terminological choice for cognitive agents social theory. We will adopt this stronger (objective + subjective) 16 notion of power-over in this paper. 17 In this perspective one should define Power-over as follows: 16 The merely objective form would be (O-Power-Over y x a p) = (DEP x y a p); the merely subjective form would be (S-Power-Over y x a p) = (Believe y (DEP x y a p)). This would have an illusory form: (S-Power-Over y x a p) = (Believe y (DEP x y a p)) & (Not (DEP x y a p)) (but see note 19); and a realistic form that we have just preferred. 17 However, we should not forget that Y in fact can, is in condition of favoring or damaging X even when he ignores or does not want this relation. Moreover, a complete theory would be needed relative to bilateral unawareness, X s awareness, Y s awareness, X s mere belief, Y s belief.

222 222 Cristiano Castelfranchi (Power-Over Y X A/p) = (KNOW Y (DEP X Y A/p)) 18 Both the ingredients of the definition are necessary: if there were only Y s Belief about X s Dependence, there would be not a power of Y over X, but only a subjective delusion of power; if, on the contrary, there were only the objective Dependence of X on Y, without Y s awareness, Y would have just an accidental Power-of damaging or favoring X s goals, but in fact he would not dispose of such a power. To dispose of it, Y must know it (as we saw this is a general problem; 3.). 19 When Y uses this Power-over X s goal in order to influence X, to induce her to do something, we call his power Incentive-Power since Y uses it for providing an incentive to X (incentives are for influencing, for motivating) or Rewarding-power (if and when Y uses it for giving X positive or negative reinforcements, frustrations or satisfaction, for learning, for example in training X). We will see how an agent will use this Incentive/Reward power for influencing the other and acquire new powers. Frequently, what really matters is not Y s Power-over X, but X s belief about Y s power (note 16 and 19): the show is more important than the substance (signs and exhibition of power; marketing; politics; etc.). Power-over Relations Mirroring Dependence Relations Since Dependence relations are the direct basis of Power-over relations, we can derive from a Dependence Network a corresponding Power-over Net among the agents involved. The properties of Dep relationships turn into the properties of the Power-over relationships. For example, the more X Depends on Y as for p, the more Y gets Power-over X; the more X depends on 18 This is a merely static relation (like Dependence): the power that y has over x (more precisely, over x s goals). It is not the exercise of this power, which would be an action of y toward x relative to and based on such a power 19 It is in fact sufficient that X believes to be dependent on Y, and that Y believes the same or believes that X believs so. Also because X s belief is self-fulfilling, as we know. Notice that apart from this trick (of restricting Power-over to the conscious one, the full form) it would be impossible to have social situations without power relations: any social reality implies power relationships. In fact, sociality consists of and is based on a common world i.e. interference (Castelfranchi, 1998) and interference produces Dependence and the other side of dependence relationships are power relationships (Emerson, 1962; Miller, 1992a, b).

223 The Micro-Macro Constitution of Power 223 Y for many goals, the more Y gets in general Power-over X. If Dependence has a quantity, Power-over has a related quantity. Bilateral Dependence relations generate bilateral Power-over relations. In exchange, for example, we have reciprocal Reward/Incentive powers: X can satisfy or sanction Y, and vice versa. It is also possible to derive other Power-over relations from a given Powerover relation. For example, Power-over has a sort of restricted Transitivity : if Y has Power-over X s goal p thanks to his action Ay, and X has power-over Z s goal q thanks to her action Ax, and p is that X does Ax ; then Y has power-over Z s goal q (since Y can prevent or induce the action of X needed by Z) (see also 8.2). 4.3 Negotiation Power Zlotkin and Rosenschein (Zlotkin and Rosenschein, 1993) observed that, intuitively, in multi-agent systems, agents differ with regard to the following dimensions: a) the Power-of deciding whether to form coalitions and with whom; b) the social alternatives each has got; c) the value each agent s collaboration has for the other agents. As these authors conclude, an agent scoring high on all these dimensions could augment the price for its share. The authors intuition is obviously right (it is the very classical notion of power in market; Miller 1992), and we will endeavor to do justice to it in our framework. Intuition (a) is correct. Some agents find themselves in a privileged position with regard to their capacity and their efficacy to negotiate, induce others to join them in coalition, and obtain a greater share of the global utility. This is related to points (b) and (c). Intuition (b) is also correct. Some agents have more options than others. They can choose whether and with whom to exchange or cooperate. In our terms, suppose X is dependent either on Y or on Z; and both Y and Z are co-dependent on X. From X s point of view, this is an advantageous situation, because she can choose the more convenient partner to interact with. Finally, intuition (c) is correct as well. As we said, both Y and Z depend on X. In other terms, X has no rivals, and her abilities are valued twice as much as those of Y and Z on the market.

224 224 Cristiano Castelfranchi In other words, the price at which X can sell her performance is higher. So, the authors conclusion is also correct. This is also why agents move from one market to another: to improve their negotiation power. While entering a new market, X can completely change her value and power: in Market M, she was for example dependent and helpless. Besides, she lacked the abilities required to fulfill the others needs. Consequently, she could find no partners for cooperation or exchange; in Market N, she is perhaps highly requested: many agents are co-dependent on her (everybody needs action Ax), and she has got many alternative partners for exchange or cooperation (several agents can perform the action she is in need of). Therefore, from the market in which they are inserted, and precisely from their dependence relations, agents derive their power of negotiation. More precisely, such a power is grounded upon a specific pattern combining codependence and OR-dependence in a bilateral-dependence structure. (For a formal treatment of the negotiation power and agent value, see (Castelfranchi and Conte, 1996)). By co-dependence we mean that a set of agents Agj depend on the same X which is a contended for social resource. In other words, X happens to find herself at the center of a dependence network. The agent X is in a relationship of OR-dependence with regard to a set of agents Agj when she depends on any of them for a given goal of his. This equals saying that each member of Agj has at least one action sufficient for X s goal to be achieved. Any member of the set Agj is sufficient but unnecessary for X s goal. OR-dependence provides the dependent agent with a number of alternative ways to achieve her goal, among which she shall choose the most convenient. The number of alternatives amounts to the number of agents contained in the set Agj. If we assume one given agent X as the temporary center of a dependence network and compare her (possibly null) social appeal with her (possibly null) need for social resource, we may derive her value on the social market, that is to say, her capacity to establish relationships of exchange, cooperation, coalition, etc.. More generally, we will derive her power of negotiation, or her exchange value. This property is supposed to be a quantifiable dimension along which different agents are compared. Intuitively, this notion of the power of negotiation corresponds to how well the agents can sell on their market, how useful they are precisely for those agents whose abilities they need.

225 The Micro-Macro Constitution of Power From Power-over to Power of Influencing (and from Dependence to the Goal of Influencing) Power-over is important not only because it gives the agents the opportunity to help (cooperate, exchange) or fight against each other; it is very important also as the main basis and instrument of their Power of Influencing others. This is the most typical notion of power considered in the social sciences. They are interested in the different reasons and ways (authority, sanctions, prestige, etc.) an agent Y can induce another to do something (see 10.) In fact, autonomous agents in M-A worlds do not limit themselves just to Coordinate their actions (plans, goals), trying to avoid interferences, damages, delays, and wastes, and to take advantage of each other. They are much more ambitious and able: they try to change the goals (and thus the behaviors) of the agents present in the same world. This is the most important motivation to social action; this is, for example, the main motivation for communication (more important than information for coordination). For example, if agent X Depends on agent Y as for her goal p, X is not satisfied with just waiting for Y s doing the expected action, or with hoping that Y does not perform some dangerous action (remember that X can depend on the execution or the non-execution of a certain action: in other terms, the omission of an action is an action). X will actively try to obtain from Y the desired action or to prevent Y from executing it. In other words, X will have the Goal of Influencing Y to do or not to do something. Given that an autonomous cognitive agent s behaviors depend on his goals, X has the goal that Y Wants or does not Want to do something. Thus, from the Dep of X on Y we derive a Goal of X of influencing Y (Castelfranchi, 1991): (DEP X Y A/p) => (GOAL X (INTEND Y (DO Y A))); 20 and for this goal she intends to do something in order to induce such an intention in Y. But the goal of Influencing is nothing without the Power of Influencing, i.e. the power to induce someone to do (or not to do) something, or better (among cognitive agents) the power to induce someone to have a certain goal, and possibly to intend something. 20 We use Cohen and Levesque notation integrated with the Dependence predicate. This means that X has the intention of doing A since and until she believes that this will induce Y to have the goal.

226 226 Cristiano Castelfranchi (Let s notice that, paradoxically, if X depends on Y, X has the Goal of Influencing Y, but Y has the Power of Influencing X! 21 ) We define X s Power of Influencing Y about goal p (INFL-POWER X Y Ax p), the fact that X has (thanks to her action Ax) the Power-of making Y pursue goal p. Since Y is a cognitive agent, to pursue p Y must intend p. Thus the influencing power is the Power-of making Y intend/pursue p. 22 That is: X has the Power-of influencing Y if she CANDO such an act that makes Y have p as a goal of his own. This action generally implies making Y believe something which is somehow related to p. For instance, an act of that sort might be a threat ( If you don t pursue p, I will thwart your goal q ). It is important to notice that the Power of influencing is not a power of Control over Y s action. In fact X can induce Y to do A, but to have a full control over the behavior of Y, it should be true that Y does A if and only if X exercises her power (see 10, on Porn s theory). A relevant concept related to the influencing power (following Weber s, Emerson s, etc. tradition) is that of resistance to power. Resistance of Y is due to the costs Y incurs (abandoned goals; wasted resources; risks; etc.) for doing what X wants he does. Perhaps one should consider as part of Y s resistance also the resistance to accept the beliefs that X uses to persuade him, when influence is obtained through communication. For surely the power of influencing and inducing Y to do something presupposes Y s possible resistance and entails the power of overcoming it. A special, relevant case of Power of influencing is what we may call Command Power: the power of influencing with some normative base; based on some subordination, norm, authority, role, hierarchy (see 9), previous agreement and contracts (Castelfranchi, 1995): Y has the obligation (and by giving a command to him X creates or instantiates such an obligation) to do what she asks him to do. There are many causes or bases of X s Power of Influencing Y, like prestige, authority, hypnosis or psychoanalysis; however, the most basic and relevant case lies in X s Power-over Y and this transition is very important for social theory. 21 This is a very frequent human tragedy, for example in love, but also in business, politics, etc. 22 In (Castelfranchi, Miceli, Cesta, 1992) we defined it as follows: (INFL-POWER X Y A p) =def (CANDO X A) & ((DONE-BY X A) => (EVENTU- ALLY (GOAL Y p))). Obviously this is the objective version of such a power; also the subjective is possible. If Influencing Power is based on Power-over, it is necessarily subjective since we gave a subjective version of the Power-over.

227 The Micro-Macro Constitution of Power 227 By using her Power-over Y relative to Y s goal q (threatening or promising q) 23, X can induce Y to do or not to do something (better, to have the goal to do it). Of course, many additional conditions are required: a) that Y believes in X s Power-over his goal; b) that the value for Y of his goal q is greater than the cost of adopting or pursuing X s goal p; c) that X puts as a condition for favoring or not damaging Y s goal q, the fact that Y adopts her goal that p. Notice that: (INFL-POWER X Y Ax p) & (Power-of Y Ay/p) > (Powerof X p) That is: if X has goal that p, and has -through his action Ax the power of influencing Y to pursue p; and Y is in fact able to realize p through his action Ay; then X has the Power-of p (although obviously an indirect one) (like, in the logic of bringing it about, EyExp > Eyp Pörn, 1970 notes 8 and 52). Since X s Influencing-Power on Y can be due to her Power-Over, i.e. to some Power-of q, we observe both a multiplication of powers (X gets new Powers-of) a circulation of powers (Y s Power-of p becomes X s Power-of p) ( 5.), and a transformation of powers (X s Power-of q becomes X s Power-of p). Power transformation and multiplication are very important. Power tends to be like money: it can be spent for everything independently of its original use value ; X spends her Power-of q for obtaining p. This is one of the bases of Hobbes pessimistic view in the Leviathan about a general inclination in all mankind, a perpetual and restless striving of power after power, that ceaseth only in death. And Hobbes built this view starting from power at the action/personal layer. Other bases of the Influencing power are not due to Power-over (rewards, incentives) but just to spontaneous adhesion and submission or benevolence (like in a baby-mother interaction), or imitation/contagion. 23 To be sure for having the Power of influencing Y, it is a sufficient condition that Y believes that X has some Power-over his goals (the toy guns in robberies).

228 228 Cristiano Castelfranchi For example, if X knows that Y imitates her, and does whatever X does, X can on purpose induce Y to do A, without using Y s dependence on her as reward. For example, If Y is Benevolent (in Cohen and Levesque sense) towards X, X has got an Influencing Power on Y; X is able and in condition to induce Y to do A or pursue p, independently of any specific Power-over Y s goals. 24 In sum, in this section we saw how all the most important forms of social power completely or mainly emerge from and build upon the agents personal powers Fig. 3 5 Circulation, multiplication, and transformation of Power The main function of pro-social or positive sociality is the multiplication of the power of the participating agents. 25 Unlike Huberman and colleagues 24 Paradoxically, this non-reward based power of influencing gives X rewarding Power-over Y. In fact if X can when she likes induce Y to do A, Y is no longer in condition to pursue his own goals when he likes; thus he is dependent on X for being free of pursuing his own goals, thus X has Power-over Y. 25 It seems that the less the individual Self-Sufficiency (number of self-realisable goal, on number of needs) the more sociality is useful and could multiply the powers. (But the function is complex, because we need agents high power of (capability, resources), and low Self-Sufficiency ). In other terms, the more the individuals are dependent on each other, the more sociality multiplies their power. This is one of the reasons why division of labour and specialisation are so productive.

229 The Micro-Macro Constitution of Power 229 (Huberman and Hogg, 1994) we do not assume that the greatest advantage of (cooperative) sociality is to speed up the search for solutions to common problems, or to find better solutions to them, but rather to multiply individual powers: Any agent, while remaining limited in its capabilities, skills and resources, finds the number of goals it can pursue and achieve increased by virtue of its use of others skills and resources. In a sense, any agent s limits of power, and its differences from others in the kind of power it is endowed with, turn into an advantage: although not omnipotent, the agent is allowed to overcome its cognitive, and practical limits through sociality. But, within this general phenomenon we should distinguish two very different kinds of Power improvement. Let us call circulation of powers the first, and complex power construction the second. Power Circulation or Exchange, and Power Conferring Suppose X has goal G that p, and the Power-of producing q (action Ax with effect q); Y has the goal G that q, and the Power-of producing p. In other words, if X had goal p Y could satisfy it; the reverse for Y. X and Y are Dependent on each other ( 2.1). Positive sociality (goal adoption in form of help, exchange, etc.) allows the use of the power of Y for the goal of X (and/or vice versa). One agent can use the power of another agent for his own goals. The result is a multiplication of each individual Power-of; of the average Power-of, and of the total number of goals that are reached (or could be reached) in the community-population. The ability or resource already exists for some agent, but either it is not used for a certain result (goal of some agent), or it is not used for the goal of a certain agent. More agents will reach their own goals just using the potential power of other agents. (As we observed above, this also implies some transformation of the agents powers). This is the market mechanism, but there is also another facet of the phenomenon. Conferring Power. While non-opposing, letting, permitting, acknowledging rights, etc. X gives power to Y, creates (in part) Y s power. While adhering to Y s request, by obeying, submitting, (i.e. using her own powers for Y s goals) again X gives power to Y, creates (in part) his power. We call this inter-

230 230 Cristiano Castelfranchi personal conferring of power. 26 This effect of multiplication, transfer or conferring of power to the other is quite different from the next one. Co-Powers This is the case of the creation of a completely new power that no agent already personally would have, made of the combination of more elementary individual powers. In fact, some goals require complex actions, i.e. some complementary actions organized in a suitable way (in sequence and/or in parallel); in other words, a multi-agent plan is needed. Suppose that there is no agent capable of executing all such actions (or of executing them in the right temporal-spatial pattern: for example, simultaneously). In this case, only cooperation (agents sharing the same end), or, more in general, the combination of the individual powers in the same plan for the same end, allows the fulfillment of this end. Such goals could even be invented, created in a combinatorial way by the agents, given that the combination of the effects of their actions is open. No agent perhaps had such a goal before the social interaction: new means create new ends. This power is the power of the Collective: the collective agent (group, organization) can have more power than the trivial sum of the powers of its members. New results can be achieved thanks to coordinated collective action 27. For example no agent can have the personal Power-of playing ring-a-ring-o - roses. This is Werner s notion of CoCAN (Werner, 1990). If an agent can control a collective power, he acquires an enormous power (Barnes, 1998). 5.1 Collective Power Let s call Collective Power-of A/p the power of a team, a jointly intending group of agents (Gilbert, 1999; Tuomela, 1988, 1998), of successfully performing action A, and achieving goal p. Which are the relationships between the power of a collective and the powers of its members? Is the collective 26 This obviously applies also between groups, organizations, institutions, etc. 27 Also quantity can be important; there might be some threshold, some minimal necessary effort for achieving p.

231 The Micro-Macro Constitution of Power 231 power the resultant of the individual powers? The answer is yes, but as we just saw it is a bit complicated by the co-power phenomena, where individually no agent is able to bring it about that p although collectively (by cooperation) they can; only the joint coordinated effort can succeed. So we can claim that: a necessary condition for a team T having the Powerof A/p is that either at least one of its members individually has the Power-of A/p, or a sub-group st of T has the Co-Power-of A/p. However, this is not a sufficient condition. That an agent X, member of team T, has the Power-of A/p is not enough for T having this power. Additional conditions are necessary, at least that: i) X is when required willing (spontaneously or by influence and coercion) to use her power; ii) the other members of T will not or cannot interfere, and they will (in fact or intentionally) coordinate with X, for example, they have the intention that X performs A (Grosz and Kraus, 1996: intention that ; Castelfranchi, 1998 delegation ). 28 In other terms a collective power (or a we-power) is a power that the collective has only if the agents are willing and coordinated, e.g. are jointly intending. If X is a loyal and obedient member of T, always ready to obey or adopt T s goals, all his powers are under T s control and become collective powers. Groups usually regulate by norms and rules which powers of each member should be at disposal, under the control of T. The collective power of a group tends to concentrate in what we call the power possessors (Barnes, 1988), that is in a few individuals or sub-groups that decide about its use, precisely because to be effective it needs some orchestration. They have discretion about, they dispose of and control the employment of the collective power, so they really have it. In fact we know that there is a special crucial nucleus of power as capacity and 28 Since our notion of Power-of is context sensitive and incorporates to be in condition to do A, i.e. the external opportunities, means and conditions, one might consider the non negative interference of the others as one of those conditions and include point (ii) in the very notion of X having the Power-of A/p in the context/situation of team T. Analogously, one might be parsimonious also on condition (i) while claiming that in fact the intention to do A/achieve p is one of the internal conditions of an effective Power-of A/p. But as we saw in 3. this is too strong.

232 232 Cristiano Castelfranchi this nucleus is the possibility and meta-capability of deciding about its use ( 1.2). 6 The Dialectic Between Individual Power-of and Social Power There is either a virtuous or a vicious circle between Power-of (personal power) (i.e. being able and in condition to achieve goals) and Social Power. Any lack of personal power (lack of abilities, competence, knowledge, controlled resources) reduces the various forms of social power, and the probability of having goal-adoption relationships able to increase that power. Conversely, the greater our personal power, our value, the greater the possibility to acquire social power and -through this to increase our personal power: i. Social Power derives from Power-of ; ii. Social Power increases Power-of Fig. 4 As we saw there are several forms of social power, and this dialectic (figure 4) is generally true for all of them. As the agents compete for their private/personal Power-of they compete even more for social power (image/status, incentive power, influencing power, negotiation power), especially because several forms of it are intrinsically comparative and hierarchical. As already noticed, for example, the Power of Influencing is generally based on the Power-over, but the Power of Influencing is also a new basis for a new level of Power-over. From X s Power of Influencing Y we can derive a new power of X over Y s goal. In fact, the Power of Influencing Y is the power to induce Y to do or not to do something, to pursue or not pursue a certain goal, to have a certain intention; and

233 The Micro-Macro Constitution of Power 233 to have an active goal, to pursue it, to choose it as an intention is one of the main conditions to achieve it, thus X can favor or prevent the achievement of Y s goal p (Power-over) (or of another alternative goal q) just by inducing Y to pursue or not to pursue p (Power of Influencing)! Fig. 5 How Power-over gives Power-of It is pretty clear how X s Power-of (combined with Y s lack of power) gives X some Power-over Y; but let us see better the reverse: how X s Power-over Y gives X some Power-of. Social interaction is aimed at satisfying some goal, by acquiring or maintaining Power-of achieving goals. Pro-social interaction (exchange, cooperation, etc.) is empowering: an agent tries to achieve a goal through the action and then the Power-of of another agent (reliance/delegation see 8.4) or at least thanks to the permission, the lack of opposition and obstacles from another agent. When X depends on Y she lacks some power (cannot achieve p) that Y has; if X succeeds in obtaining (through induction, coercion, exchange, cooperation, or whatever) that Y does what she needs, X will achieve p. In other words, X s ability to induce Y to do A (to realize p) becomes X s ability to achieve p (through Y); Y s power becomes X s power; more precisely X thanks to social power and interaction acquires the Power-of p. X s power is greater than in isolation. If for whatever reason X can induce Y to do whatever she wants, X acquires (can exploit) all the powers of Y s. We have shown how Y s social goal-adoption gives power to X. But how does X obtain Y s adoption? How does X induce Y to do A? Through some form of social power; for example thanks to the fact that Y too depends on X, and X uses such a dependence for promoting exchanges via some promise or

234 234 Cristiano Castelfranchi threat. Thus X s Power-over Y gives X Power-of Influencing Y (to do A). But what gives X such Power-over Y? As we know, Power-over is just based on Power-of and lack of power. Thus iii. Power-of produces Power-of through the social medium (Fig. 4). But also iv. Social Power produces Social Power through the acquisition of Power-of (Fig. 4). Value (evaluation) and Signs of Power It is important to emphasize that we gave a simplified and objectivist view of these relationships. Actually, what really gives agents Social Power are mostly the beliefs of the other cognitive agents (Barnes, 1998), and in general the signs of Power-of, not the real Power-of per se. We mean: For social comparison, social value and hierarchies, this is obvious since the subjective perspective, the belief, is in the definition of the notion of value and of comparison; as we said we refer to a hierarchy of images, of evaluations, just mental constructions (Miceli and Castelfranchi, 1989). Also in order to use our Power-over (which is objective and independent of the awareness of the other) as Incentive Power, it is necessary that the other believes that we have Power-over his goals. Moreover, this is not only necessary: it is sufficient! Our real power might be a bluff. Also our Power of Influencing, based either on our Incentive (Promise/Threat) power, or on some prestige, authority, etc. depends on what the other believes and perceives about us (Barnes, 1998). This is not true only among cognitive agents (driven by their explicit beliefs and evaluations of others), this is true also among simple animals or behavior-based agents. There can be hierarchies among them, and influencing power, based on the individual power (for ex. strength, aggressiveness, etc.), but in fact this means that there is some sign for learning or for attributing a power to the other (the size for strength, for example); thus in fact it is not real power (strength) but its sign, the basis of the acquired social power.

235 The Micro-Macro Constitution of Power 235 Institutional Empowerment as Independent of Individual Power-of Statements (i) and (ii) are less true for Institutional power, which is in a sense artificial and arbitrary, not naturally based on or derived from the qualities or the resources of the individual. It is just conventional. Thus the individual can be empowered by the institution, receiving a lot of power from it, independently of his own Power-of. We will see later that this is not completely and always true (there is a dialectic also between the power of the institution -and the institutional empowerment and the personal power of the role-player); however, it remains important that in that case the source of the power the individual has over others, or of his influencing power, is not his own power. (see 8. and 9.) 7 The Dynamics of Dependence and Power It is worth noticing that our notion of social power is intrinsically non-static, non-hierarchical, non-dyadic. We adopted some sort of context-dependent emergentist perspective: power emerges from personal capacity, resources, goals, and preferences, and from the presence of network of agents endowed with different capacities, resources, goals, and preferences. This notion is non hierarchical for such bottom-up, distributed, peer relations, and pre-institutional view. All this makes it non-static because the definition allows for a very strong dynamics in dependence and power relationships. Dependence and power positions and relations are continuously changing and Dependence dynamics determines Power dynamics. (We will address this point in detail in the next section.) Our notion is also only seemingly dyadic. Rewarding-Power is in fact a relation between two abstract agents (then individuals, groups, organizations, etc.), however: the power existing between them cannot be understood by simply considering their relation without reference to their wider social context (Wartenberg, 1992) Our degree of dependence of X on Y is related to the M-A context in which they are situated: their market. Changing the M-A context in which X and Y are situated, the power relation between them could change dramatically. The power rela-

236 236 Cristiano Castelfranchi tion between them is only relative to a certain M-A context 7.1 Dynamics Dependence (and power) relation change very simply and continuously in an open and evolving world: Exogenous Dynamics Agents may abandon a given world or population or market ; the disappearance of an agent can change the dependence relations among several others. New agents enter a given world or population or market by immigration or birth; the appearance of a new agent can change the dependence relations among several others. Through social adoption and alliances Agents acquire (and lose) indirect-powers Dependence (and power) relations can be created, maintained or eliminated on purpose. For example, X in order to induce Y to exchange with him, can create Y s dependence on him, by activating or creating in Y a new goal he is not able to achieve, either by persuasion (communication), or by acting on Y s body, or by modifying Y s world or representation of the world. For example X could destroy some physical condition necessary for some action/goal of Y, that Y is not able to restore. X can also make Y dependent just by inducing in Y the belief to be dependent. (In this way X increases her Powerover Y, and her negotiation power; see 3.5). In partnership formation the agent s Dependence increases. Usually, when X chooses Y as a partner she has some alternatives (that affect its degree of dependence and the strength of Y s Power-over X); but after this choice and agreement usually some of those opportunities (offers) disappear (because of deadlines or formation of other partnerships). Moreover, in decision, in negotiation, in agreement some resource is consumed, there are costs; if after that transaction X changes her mind and abandons Y, those costs are wasted. For those reasons, X, in choosing Y, is making herself more dependent on Y for that goal Additional reasons for this reinforcing effect are: acquaintance and increased trust; specialization and its effects (Durkeim); routine behavior, etc.

237 The Micro-Macro Constitution of Power 237 Endogenous Dynamics Agents can acquire (by learning) or lose capabilities, know how, expertise. Agents change their goals; since dependence is goal-relative, a new goal can create new dependence relations; the loss of a given goal means the elimination of dependence; after an agent has satisfied a goal (thanks to another agent) he is no longer dependent on the latter as for that goal. Agents acquire or lose resources (because of their activity, competition, exchange, etc.); new resources imply new powers and possible dependence of others; lost resources imply lost power and social power (fewer possible agents depending on us). As a consequence of these two factors and of their combination Dependence Networks are very dynamic structures Struggling for Power Part of such a dynamics is due to competition and conflicts. As we pointed out along all the paper, there are several levels and kinds of struggling for power. Agents compete at the basic level for resources, knowledge, etc. that give them personal power (Power-of); they compete for establishing dependencies relations and on such a basis for negotiation power, alliances and partnerships that improve their power; they compete for their comparative value and consequent position (status) in the social hierarchy, that increases their power; they compete for influence; they compete for power positions in organizations and institutions and for having authority roles (Hobbes view). This competition can be both objective and even unaware (X tries to achieve goal p to have a given power and in doing so she negatively interferes with Y s goal q, damaging Y, i.e. reducing his power). But when the agents are aware of a competing situation they could initiate conflicting and agonistic behavior. In other words, they can plan additional actions for preventing others from achieving their goals, or for handicapping, damaging, eliminating others. Thus war among agents is like the dynamic establishment and transformation of alliances and cooperation a very dynamic factor for the continuous change of agents dependence relations and powers. 30 Sichman, 1995; Sichman et al. 1994; Veneziano et al., 1996.

238 238 Cristiano Castelfranchi 8 Empowerment Empowerment is i) a state transition, ii) a (social) action, iii) a relationship. It is a relationship and a social action between two agents: Y (the empowering agent) and X (the empowered agent). Empowerment in fact means that an agent X comes to have more power than before, and this power is given to her by another agent Y. Thanks to Y s action (that we call empowering action ) X acquires the additional power P (state transition). Thus empowerment is a relation between Y, X, an additional power P of X s, Y s action Ay of bringing about that X passes from a set of powers S to a larger set S (S + P ). (EMPOWER Y, Ay, X, Ax/G, P ): by doing Ay Y empowers X as for the action Ax and its goal G; X acquires the Power-of G. We can reasonably restrict the notion of empowering to non-accidental bringing about that ; i.e. to intentional actions aimed at augmenting in a specific way X s powers. 31 This is the broad and weak notion of empowering : to make X able and in condition to do Ax (if she likes). The Webster NewWorld Dictionary s definition is: a) to give power to, authorize; b) to enable. (The former is in fact a sub-case of the latter, which is probably too broad; see later.) At the preinstitutional, interpersonal layer, any power in its abstract sense, i.e. any internal or external resource of X or condition necessary for a successful action (performing the action Ax and achieving its goal G), can be an object of an empowerment relation and act. X is dependent on Y as for a given resource or condition r (let s put them together in an abstract notion of resource r that designates whatever is necessary for Ax). X lacks r and thus lacks power P for successfully performing Ax. Y acts in order to make X no longer dependent on him, and disposing of r. After this X is endowed with power P (she has the necessary resource r). 32 Y can empower X by providing her knowledge (for example the solution of the examination test; a password; the enemy s plan; the phone number of a 31 We leave the non-intentional cases in a broader notion and set of causing that X gets greater power that for example includes wrong moves in competitions and wars that give advantages to the enemy, and that for sure we would not like to call empowerment. 32 X will not necessarily perform Ax; she simply can.

239 The Micro-Macro Constitution of Power 239 man; etc.), know how, skills (by training and teaching), tools and means (money, a gun, a car, etc.), external practical conditions (by removing obstacles, for example by leaving open the door of the prison where X is imprisoned; by opening the sea for letting X pass through; etc.). To be sure, for a narrow and more precise notion of empowering this seems too broad: occasionally and contingently helping X (so that she can realize her goal G) is not enough for empowering (it is just permitting or helping). After Y s action, X should be able to achieve G by herself, without Y s intervention, directly and autonomously. We mean that if, thanks to Y s direct and active help, X achieves her goal there is no empowerment. Except in the case that Y s help is at X s disposal (for example thanks to a promise, or a hierarchical position), where X has the power of inducing Y to do what she needs when she needs it. For a true empowerment, X has to acquire thanks to Y s action the Power-of autonomously producing/achieving G: the resources and all the actions in the plan for realizing G must be under her control. 33 In some cases, when Y provides r to X he loses r, viz. he can no longer use, enjoy r (money, car, or space, when Y moves off in order to let X pass through). In other cases, Y does not lose r while letting X accede to it (skills, knowledge, etc.). However, in any case, Y loses something: his (at least occasional) Power-over X, X s dependence on him as for r, Ax, and G. Not necessarily X receives this power from Y and Y passes this power to X (transfer or transmission of power). Neither is it in fact necessary that Y has the power P that he provides to X (for example, Y might give X the access to a room, or to a meeting, where he is not permitted to accede; for example, a trainer or a teacher could not fully have the abilities that he develops in the student), nor is it necessary in case he has P that he loses P while passing it to X. The most important distinction between different forms of empowerment seems that between practical and deontic empowerment. Deontic power means here that from the normative point of view X is in condition to perform Ax; i.e., she is permitted, authorized, entitled to do Ax, or has acknowl- 33 One would probably like to exclude from the notion of empowerment the simple acquisition of one-action or occasional power. Perhaps we do not like to conceive any action that puts X in condition of as an empowerment, and we would prefer to refer to more stable powers. This means as we just saw that Y gives X a power that she controls and can spend if and when she likes, and perhaps several times. In case of external conditions, resources, and tools this means that X has a stable access to r, and in a sense possesses it and controls it (that is obvious for internal r). The distinction is commonsensical but not so strong. Even official and institutional authorization can in fact be just for one time.

240 240 Cristiano Castelfranchi edged rights to do Ax. Obviously, she can be normatively in condition to do Ax (without incurring oppositions, violations or punishments) without being practically capable or in condition to do Ax. That is, she can lack the practical power for acting. Vice versa, X might be perfectly able and in condition to perform Ax, but she does not have permission, consensus, authorization, title or whatever is normatively needed for an acceptable and accepted action. The two forms of power are rather (but not completely) independent. Y can empower X in both ways. Let us study the relationships between these two forms while analyzing the relationships between permission and empowerment at the interpersonal, pre-institutional level, where the notion of permission is much weaker than in law. 8.1 Permission Vs. Empowerment Empowering is not the same as providing permission, and vice versa. On the one hand, at a pre-institutional level, there are other ways and forms of empowerment that do not include or use permitting (for example, teaching and training in the use of guns and war techniques). On the other hand, permitting is just a specific and partial case of empowering. In our view, pre-legal, pragmatic permission (Castelfranchi, 1997) provides in fact power, a specific form of it. But, for example, it does not provide skills, knowledge, tools, and other possible bases of power. It provides deontic power: the possibility of doing the action without conflict, without violating another s (entitled) expectations, without upsetting him and eliciting opposition and persecution (retaliation and punishment). 34 When the conflict with the other and his opposition also creates a practical obstacle to the use of the needed resource or condition r for action Ax, the deontic lack of power determines a material lack of power: X is not able and in condition to do action Ax (and achieve her goal G) since she depends on Y for successfully performing Ax (and has no alternative plans to Ax). Sometimes, on the contrary, Y s permission is materially superfluous; X is able and in condition to do Ax in any case; she does not depend on Y for 34 That become legal violations, legal persecution and sanctions, legal obstacles,.. when the other who should permit is the authority.

241 The Micro-Macro Constitution of Power 241 performing Ax. X simply has a broader goal G (G + g) where g is not disappointing Y, not having Y s protests or retaliation or conflict. In our vocabulary X is only weakly dependent on Y : she depends on Y for the broad goal G but not for G that she is able and in condition to achieve. In the latter case (weak dependence Jennings and Mandami, 1992), permission provides only what we call the deontic conditions, the deontic Powerof; in the former case (strong dependence) permitting while removing the deontic obstacle also provides the material power for Ax/G. Y could prevent X from doing Ax (i.e. Y can negatively interfere, has Power-over X s goal; X depends on Y s not opposing, and not creating obstacles) 35. Y decides not to hamper X, i.e. he lets/permits X to achieve G. Thanks to Y s abstaining from interfering, X acquires the complete Power-of G. Fig. 6 In conclusion, when Y gives permission to X as for action Ax, he can either give X the Power-of (performing) Ax or not. But Y empowers X in both cases. Empowerment is goal-relative; either Y is empowering X relative to performing action Ax (a sub-goal for G, and thus for G), or is empowering X for G a higher-goal of G. 36 In a celebrated example of Makinson (cited in Jones and Sergot, 1996), a priest is paradoxically empowered to marry a couple that did not promise to bring up their children in the same religion, but he is not permitted to do so given the specific instructions of the ecclesiastic authorities. If we analyze this paradoxical situation more carefully, we can notice that the priest is empowered to marry people and that couple too (the marriage would be valid), but 35 A threat of possible protests, retaliations, or sanctions of Y after X s action is an obstacle for X in the sense of lack of power from goal-balance: X cannot do Ax because of serious expected consequences ( 1.2). 36 Only a sufficiently analytic theory of goals and goal directed actions can deal with deontic problems.

242 242 Cristiano Castelfranchi since he is not permitted to do so he does not have the Power-of marrying that couple without violating a prescription of the authorities! So he is weakly dependent on the authorities: he does not depend on them for the goal of marrying, but for the goal of marring without violating instructions. Permission would provide him this larger power, would empower him with regard to the broader goal. Permission (in its appropriate pragmatic conditions) empowers. 8.2 Transmitting Powers In some cases in empowerment Y s personal power, Power-of, is transferred to X. How is this possible? Via resource transfer: Without a given resource that Y controls (by regulating the access to it), X cannot achieve his goal G. Y makes such a resource accessible to X. Now X is able to achieve her own goal (a special kind of resource is authorization, normative permission; another is information) Via education and training: Y trains X so that X becomes via learning independently able to achieve G (Notice that not necessarily Y himself is able to do A; thus it is not necessarily a transfer of power). The strange triangle: Y uses his Power-of to create X s Reward and Influencing Power. Y makes Z believe that if Z does not do what X asks, he will use his Power-over Z s goals. Now X has the Power-over Z s goals and on such a basis the power of influencing Z, which are in fact based on Y s Power-of and Power-over. But it is as if X had such a Power-of and Power-over. Consider for example the case where Y pays Z for obeying X. Notice that in our terminology X is not empowered of, does not receive Y s Power-of (for example, of beating Z, or of paying Z), but she receives the Power of influencing Z; since she can do this by herself, after Y s action. This works through a threat or promise If you (do not) do what X expects, I will do A (which is bad/good for you), which is based on Y s adoption of X s goal. In such a way Y empowers X.

243 The Micro-Macro Constitution of Power 243 Fig Sub-institutional and Institutional Empowerment As we said, several of these notions and mechanisms are applicable also at the institutional level. Within an organization, or between organizations, precisely the same mechanisms work. Organization O can empower its member M or the organization O, via (material, cognitive or deontic) resource transfer, training, or triangulation. We call this sub-institutional empowerment because the definitional notions do not require anything special and specifically institutional or organizational. We would like to call institutional action, power, empowerment, the level where something special and new happens. In particular, we have in mind the conventional, artificial, and performative value of institutionalized action, when a normal act of agent X count as (Searle, 1969, 1995; Tuomela, 1999, ; Jones & Sergot, 1996) a conventional or institutional action, and acquires special effects which are not natural effects of that action out of an institutional context (see 9.) (Goldman s conventional generation ). In other words, power (delegation, action, etc.) at the institutional level uses natural (personal and interpersonal) mechanisms and special mechanisms. Since true institutional empowerment works through the collective acceptance and acknowledgment of the institutional role and of the count-as effect of the action, it necessarily requires communication and signs of the empowerment. People must know and recognize the fact that X has been empowered. Thus at that layer, empowering in part means giving to X the apparent features, the signs for exerting the power. Those are part of the necessary resources for successfully performing the institutional action, i.e. for 37 For a good discussion of those issues, see Hindriks, 2003.

244 244 Cristiano Castelfranchi the action Ax of X counting-as the institutional action Ai. Thus, institutions provide special and recognizable rooms for tribunals, special dresses, hats, ritual procedures, magic words (formulas), etc. 8.4 Reliance, Delegation, and Empowerment Does any delegation of a task to X entail an empowerment of X? Does any empowerment presuppose Y s reliance upon X? The general notion of Delegation/Reliance In Delegation or Reliance Y needs or likes an action of X and includes it in his own plan: he relies on X. He plans to achieve p through the activity of X. So, he is constructing a MA plan and X has a share in this plan: X s delegated task is either a state-goal or an action-goal. Unilateral Weak Delegation In Unilateral Delegation there is neither bilateral awareness of the delegation, nor agreement: X is not aware of the fact that Y is exploiting her action. One can even delegate some task to an object or tool, relying on it for some support and result (Castelfranchi, 1998). In the weakest and passive form of unilateral delegation Y is just exploiting the autonomous behavior of X; he does not cause or elicit it. For example I count on and exploit the action of another agent to stop the bus at my bus stop. Delegation by induction In this stronger form of delegation Y is himself eliciting or inducing X s behavior in order to exploit it. Depending on the reactive or deliberative character of X, the induction is either based on some stimulus or on beliefs and complex types of influence. Delegation by acceptance (strong delegation) This Delegation is based on X s awareness of Y s intention to exploit her action; normally it is based on X s adopting Y s goal (Social Goal-Adoption), possibly after some negotiation (request, offer, etc.) concluded by some agreement and social commitment. It is very important to notice two differences between weak and strong delegation.

245 The Micro-Macro Constitution of Power 245 In weak Delegation Y and X do not necessarily have a common goal apart from the goal that X performs the delegated action (X could exploit some side effect of X intentional action); while in strong Delegation they necessarily have some common goal beyond the execution, since X has adopted some goal of Y. There is an agreement. In strong delegation X comes to have a goal (the delegated goal), while in the very weak form X happens to autonomously have such a goal. A presupposition of Y relying on an action of X for the realization of some of his goals, is that Y believes that X is able and in condition to perform Ax (or will acquire the needed powers). Sometimes, while delegating an action to X, Y has to empower X because he believes that X lacks some of the necessary r for Ax. In such cases part of the delegation act and plan is Y s supporting X or transferring to X some resources. Thus, Y s Delegation to X of action Ax frequently entails Y s empowerment of X as for Ax. We cannot send our son to buy some food in a market without providing him with some money; we cannot ask or order X to do something for us without implying a permission to her to do that action. When an agent Y delegates to another agent X the realization of some goal or the execution of some task, he should at the same time intend to permit X to do what he has delegated, i.e. he intends not to create obstacles (Grosz and Kraus, 1996; Castelfranchi, 2003). The institutional notion of delegation (which is also its most usual meaning) is richer than the interpersonal one. It implies the transfer of the task. It also entails some normative content of the actions: usually the delegated task was a task of Y, and there is some obligation to do it; Y has the authority and role for delegating a task to X; when delegated, X inherits the obligation and is explicitly or implicitly authorized to do Ax on behalf of Y and his organization; she will not use only her personal resources, but also resources of the organization, and she receives them and the authorization to use them; and so on. Thus, any institutional delegation is a strong form of delegation (an explicit or explicit contract ), and entails some empowerment. At the true institutional layer of delegation empowerment becomes necessary and definitional. If an institution wants to act through the actions of some member, it must endow him with institutional (count-as) powers and the related signs of it.

246 246 Cristiano Castelfranchi In conclusion, not all delegations from Y to X entail empowerment from Y to X (except at the institutional layer); not all empowerments from Y to X entail a delegation from Y to X: that is Y s reliance on X for some original goal of Y. Empowerment can be just due to goal adoption, i.e. to Y deciding of letting or making X realize her own goal. True Institutional Empowerment (the Count-As empowerment) is a strange process compared with interpersonal empowerment because actually at a deeper level of analysis it is not a simple bilateral process and transfer. The compliance of a third party is strictly necessary: the public, the people involved in the institution. The efficacy of the conventional institutional act in fact presupposes a tacit agreement or consensus of people in front of it. People (P) must recognize X s act as a special one and acting on such a basis; actually is this that gives the act its special effect. If X s action Ax counts as action Ai of the institution Ist, people must act as if Ai has happened. It is a sort of selfrealizing expectation: since and until people expect that Ax counts as Ai, it counts as Ai. They must (conditionally) believe or at least accept (Meijers, 2002) that this is true and that the others believe/accept as they do and will act accordingly. The effectiveness of the count-as effect passes through the minds and the consequential behavior of people. While endowing X with this special power the institution is prescribing people to accept this and to act on such a basis. Thanks to P compliance with Ist, and its delegation and empowerment X is really empowered; in fact by both Ist and P. P obviously do not recognize this role; they are simply believed to acknowledge what already exists, but in fact they are creating it thanks to this acknowledgment. Any Count-as effect (convention) and any true institutional empowerment is due to a collective acceptance of the fact, and to a diffuse or to collective intention of acting accordingly (Tuomela, 1999, 2002). 9 The Top Down Perspective. Institutional Power and Empowerment Institutional actions, actions in a role, on behalf of, are special actions endowed with special conventional or count-as effects. For example, the action of marrying a couple requires some ritual conditions. The performer

247 The Micro-Macro Constitution of Power 247 must be a priest (actually in his function, and conscious, etc.) and in order to be valid (effective) the action must be performed following certain constitutive rules; for example by saying specific words like I proclaim you husband and wife (in the Italian rite). Performing this action in those specific conditions actually produces the marrying effect. This is one kind of special, count as effect. As Searle (1995) see also Tuomela, 1999, 2002; and Jones and Sergot, 1996 have theorized and formalized, the action A performed by X in that context or institution counts as action A, and by bringing it about that p, X brings it about that q (let us call this: performative effect). Consider now another example: X can be a member of a group/organization in an official role, acting in the quality/role of, and on behalf of 38, and this means that when X performs a given action in her role the organization or the group had performed it. X s action counts as group action (Carmo and Pacheco, 2000). This is another kind of special effect ( representative effect ). We will now see how special actions require special powers, but not only. In order for an agent to be able to bring it about the institutional, symbolic action he must be able to perform the vehicle action, the material action that supports or implements such a symbolic action. X must personally have such a power. If the priest is physically unable to do what he has to do (say he is aphasic) he cannot marry the couple; if the policeman is not able to move his arms, he cannot regulate the traffic; if a manager is not able to sign a contract, he cannot sign a contract for his corporation. Every execution of action A implies the execution of the supporting/ vehicle action A which counts as A. Not vice versa: not any performance of action A produces and implies A (contextual conditions are critical). 39 In sum, powers for count-as-actions are based on/exploit personal powers. However, this is not the whole story. For effectively performing a counting as A, additional powers are necessary (powers conferred by the institu- 38 Jones and Sergot s to speak for. 39 Analogously, X s action as a member of O, in a quality, in the role implies some X s action. On this we disagree with Carmo & Pacheco (2000) assumptions. However, probably the solution of the problem is in the complex structure of the action. For surely X has to personally perform some action, however the action he is performing in a role is just the vehicle, the support of the complete action of bringing about that Ai. Same actions (or better behaviors) in different conditions produce and are aimed at producing different specific effects and are in fact different actions.

248 248 Cristiano Castelfranchi tion; Searle, 1995; see also Hindriks, 2002). Personal powers are necessary but not sufficient. Fig. 8 In organizations agents are fungible (they can be replaced) precisely because the power for their institutional actions is provided by the organization itself and by the fact of being its members in that role. However, obviously they are fungible on condition that they have the required personal competence and qualities needed for implementing and vehicling the organizational actions. Thus fungibility is limited: not anybody can replace anybody and can play any role. In sum, institutional powers cannot be reduced to the pre-institutional powers of the individuals or to their sum or collective effect (see 5.); there is something qualitatively new that only institutions give to the individual; however, institutional power presupposes and uses individual powers. The distinction shown in Fig. 8 is important and clear but rather simplistic, in fact additional, institutional powers are frequently necessary or useful even for the practical effect of the action not for its performative effect. We mean that the institution empowers also for the successful performance of the supporting action A (especially when this is a social action, based on influencing others). For ex. to arrest requires that X blocks, handcuffs and brings Y to prison. This is not enough to really institutionally/legally arrest and not simply kidnap somebody (X must be a policemen, acting not for private purposes, etc.); thus additional symbolic powers are needed for the symbolic action. But this is not my point here. My point is: does X personally, physically have the Power-of blocking, handcuffing, bringing Y in prison? Perhaps Y is much bigger and stronger than X and X would never be able to block him and imprison him. X succeeds only because Y knows (believes) that she is a po-

249 The Micro-Macro Constitution of Power 249 liceman (this is why Y does not resist and obeys). To be recognized and entitled as a policeman gives X even the power to successfully perform the supporting/practical action. Thus there is in this case a double empowerment by the institution/role (compare Fig. 9 with Fig. 8), Fig. 9 and in some cases the required personal qualities and competence are close to disappear: the symbolic signs of the role do all the job. Such an institutional over-shaping should not make us forget the fact that we stressed above: the institution exploits the personal powers of X (of its agent) and it is also the power of its agents that gives power to the institution. Let us consider this very interesting process, which historically is a forerunner of the other. 9.1 Bottom Up Empowerment: From Agents to their Institution Originally policemen were just thugs, bravos chosen for their personal strength, cruelty, and bravery. The repressive and enforcing power of a wicked squired was due to the personal powers of his thugs, that put their Power-of at his service; not very differently from Mafia. The greater the recognized authority and the normative base of power, the less it is necessary to use the personal strength of agents. But it is not accidental the fact that even in our modern mental hospitals male-nurses were selected for their size and strength! However, this story where authority gradually supplements and replaces strength holds mainly for physical qualities, but not for others. For example, teachers continue to be selected for their personal culture, intelli-

250 250 Cristiano Castelfranchi gence, etc. that they use for the public institutional purposes, although when they reject a student they use their special conventional institutional powers. Institution acquires, buys the powers that its agents put at its disposal. Fig. 10 Again we have a dialectic and circular relation in power dynamics. The case of policemen is the most clear and representative. For several occasions and functions the institution takes its force from the exercise of the individual power: in some circumstances, agents must be able to use their physical force, or their bravery, or their intelligence; and in such a way the institution succeeds in repressing, controlling, discovering, etc. For several occasions and functions the agents take their force from their institutional role. For example, because people know that if one escapes from or eliminates one single agent, there will be another, another and another who will exercise their institutional power (collective power). One does not surrender to the individual but to the institution, or better to its representative. Representative means precisely this: that X s act count as the act of the institution, X s act has special institutional effects, and X acts on behalf of the institution A clear definition od acting on behalf of is needed. We think that this requires the combination of our notion of delegation/reliance (in particular strong reliance based on adoption) plus Jones&Sergot speaks for notion.

251 The Micro-Macro Constitution of Power 251 Normative base of institutional empowerment The Top Down, institutional empowerment is a normative or deontic empowerment; it is not only due to the number and the collective force of the group (threat beyond my personal forces). The empowerment consists in the fact that there is a normative delegation and a normative symbolic power. There is a norm that prescribes the addressees of the action of the agent to treat him as a representative of Ist (for example to obey him); there is a norm that entitles the agent and authorizes him to act on behalf of (i.e. his act counts as an act of Ist s; and he acts for the delegated purposes and in the interest of Ist); there is a norm establishing the artificial-performative effects of its actions; there are norms establishing the right procedures and rituals for these special actions. Thus X is normatively empowered (in terms of authorization, resources, signs, and special effects) by Ist. People Empowering the Institution (the Leviathan) Of course, the more the power of the institution is based on authority and norm obedience, the many indeed will be the people dominated by such an institution. It is the institution s subjects, or its clients, or in general lay people not its representatives those who give it such a power; which has an externality-based and conventional nature. We mean that, on the one hand, it would be impossible to establish a large and stable dominion (like a state) only on violence and continuous control. On the other hand, the fact that one respects the authority, decides not to oppose and rebel to it, the fact that one surrenders to its symbolic force, etc. makes it an authority ; the fact that one accepts the conventional and artificial effects of the institutional actions, gives them such an effect. And this behavior is spreading, and self-confirming: since one does A, others do the same, and vice versa. 41 The more people 41 What is the relation between Collective Power ( 5.1) and Institutional Power? Rather foundational. Institutional power is based on and is a form of Collective Power. In fact only a Collective can create it by collective acceptance and acting. It is a Collective power not because a great number of people is needed (we can have conventions between two agents), but because it is a co-power; only together we can create a conventional meaning, act, use; it is the outcomes of our compliance if not of our we-intending (Tuomela, 1988, 1998).

252 252 Cristiano Castelfranchi acknowledge such an authority and follow the institutional prescriptions the more the institution becomes strong and has Power-of, Power-over and Power of influencing the individuals. In a sense, it is the arrested guy who by surrendering not to the agent s private strength but to his institutional force gives to the policeman (by giving to the institution, and vice-versa) such a force. In a sense, he is surrendering to his own alienated force/power (Marx). From Institutional to Social and Personal Powers: The Abuse of Authority Like for social power, Institutional power can increase social and personal powers. For example, the Abuse of Authority is precisely this: the institutional power used for private, personal ends and interests; thus for augmenting personal powers. It works through social power. Since people worry about my institutional power or are credulous about the institutional character of my request and are obedient, I can obtain from them some action that they would not do just for me as a lay person. I got thanks to my institutional role an influencing power and I use it for satisfying some goal of mine that I would not be able and in condition to realize. Thus I acquire new Power-of relative to my private goals. 10 Some Close Comparisons The framework we presented is in disagreement with some important theoretical traditions about power in the social sciences. Both in social and political theory the dominant tradition provides a basic notion of power which is immediately relational and influence-based. Such a basic notion is what we call influencing power (that we in great part found upon and derive from, the social incentive power (dependence), in its turn derived from the individual Power-of in a multi-agent environment). Since the classical Weber s definition (1922), power is conceived only within a social relation and as the possibility (based on whatever base) of imposing even against an opposition our own will; that is in terms of deliberated influence. Basically, power is conceived as the capacity of Ego of obtaining from Alter a given behavior/action that Alter would not perform/choose without Ego s intervention (Dahl, 1957; Crozier, 1964, 1971). This definition is cor-

253 The Micro-Macro Constitution of Power 253 rect 42, but it is not the basic notion of power, even at the interactional social layer. It should not be mixed up for example with Alter s dependence on Ego for goal G and Ego s consequent Power-over Alter s goal i.e. Ego s Power-of rewarding (or not) Alter by making him achieve G. Using this for inducing Alter to do something is just one use of it. Friedberg Friedberg founds the social, relational power upon the inter-dependence relationships (Friedberg, 1993). We do agree about this foundation, but we are able to derive inter-dependence relationships, to let them emerge from personal features of interfering agents! Moreover, we do not see power only in bilateral dependence relationships and in possible lack of exchange balance. We do not accept to put any dependence based interaction (help, cooperation, exploitation, etc.) under the category of exchange 43. This is misleading and economically biased. Here social power is identified with negotiation power. To be sure, both those criticisms are addressed in fact to all the important Social Networks tradition (Miller, 1992a,b): how and from what the dependence relationships (network) come, is not clear enough; and in fact they reduce dependence to exchange and power to negotiation power. It is also arbitrary to propose as the primitive and foundational atom what in fact already is a complex molecule : inter-dependence is crucial, but it is the outcome of X s dependence on Y that not necessarily is combined with Y s dependence on X, and makes its own important predictions. It is false that only the unbalance of inter-dependence gives power on Y to X; Y s dependence provides X power in any case: she is fact able and in condition to harm or favor Y, and she might (although inferior) use this power for example because of her ignorance of Y s Power-over him Except that the purposive aspect should be explicit; Ego s influence cannot be merely accidental; Ego must be able to obtain on purpose Alter s behavior. 43 Friedberg even claims that there is no relation without exchange that for us is a meaningless petitio principi either merely tautological (due to a too vague, abstract and counterintuitive notion of exchange ) or clearly false (what about spontaneous help, assistance of a complertely helpless person just for pity, pityful actions towards a dead body, etc.). 44 One might introduce an argument about the intrinsic unbalanced-dependence nature of power only at the subjective level (what the agent believes) and by carefully developing the theory of goal-balance lack of power. If the agent either ignores Y s Power-over her or does not care of this (lower value of the goal), in both cases it is necessarily true that (suibjectively) Y s Power-over her is unbalanced and inferior. If X were aware of Y s

254 254 Cristiano Castelfranchi Moreover, all this reasoning and argument are just based on a negative, restrictive view of power as the capability to harm Y, to frustrate Y s goals, as Power-of threatening and coerce. This is a very partial, biased, arbitrary view of power. Dependence provides the power for both: either blocking, harming Y, or favoring him, by letting or making him realize his goal. For sure X can exercise this beneficial power towards Y rather independently of any consideration about Y s Power-over her. Emerson Emerson (Emerson, 1962) defines power as follows: The power of an actor A over an actor B is the amount of resistance on the part of B which can be potentially overcome by A (p.32). Also he remains very close to Weber s notion 45, and the definition clearly refers to inducing B to do (or not to do) something, to force him, influencing or commanding. However, Emerson also provides a theory of Power-Dependence relations. Our theory of dependence relationships and of their link with power is in part developed on the basis of the social psychological literature of the 60s, in particular Thibault and Kelly (1959), Homans, Blau, and a rather strong convergence can be noticed with Emerson s view. Both our notion of Dependence and how we derive power from dependence are very close to Emerson s theory. However, it is important to understand major and quite relevant differences. First of all, Emerson does not account for the relationship between social power and other more basic and general notions of power, like personal power based on skills and resources. 46 On the contrary following Hobbes and others we start from individual power (Power-of), derive dependence social relationships and power from it, and model a feedback loop from social power again to Power-of. We search for a real micro-foundation, and our theory of power seems more general and complete. As for the social Dependence theory Emerson s definition of Dependence is very good ( A depends upon B if he aspires to goals or gratifications whose Power-over her and care about it, she would not exercise her power but she would have it; or better, she would have part of the conditions for a full Power-over Y, but not all of them, since she would not be in condition to prefer and intend to do the action (lack of power due to goal-balance). 45 Emerson is right in remarking how Weber constructs a typology rather than an organized theory of power (note 4). However, Emerson remains victim of Weber s rather restrictive and very sociological view of power as influence, command and control. 46 He even claims that Personal traits, skills or possessions have no place in a general theory (p.32)

255 The Micro-Macro Constitution of Power 255 achievement is facilitated by appropriate actions on B s part. p. 32), but it is not derived from a more basic theory of interference, of individual power and lack of power, and of agents goals, action repertoire, conditions, etc. Moreover, we do not claim that those relationships are necessarily mutual (or better bilateral) 47. More importantly, we believe in a theory of objective dependence relationships as distinct from what the agents understand or intend. This theory is only implicit in Emerson, while such an objective basis looks rather important for a theory of unconscious constraints on social behaviors, of interests as distinguished from goals, etc. (Conte and Castelfranchi, 1994). As for Power Emerson s definition and understanding of dependence in terms to grant or deny, to facilitate or hinder, the other s gratification (p.32) is very important; and his reference to reward power (French and Raven, 1959) too; but it is a pity that he does not realize that this form of social power (what we call Power over or reward power) is different from the other kind of social power that we call power of influencing. The former view of power (to overcome resistance) is not the same as the latter (power of facilitating or hindering B s gratification). This power (reward power) can be (and usually is) used for overcoming B s resistance and inducing B to do something; and vice versa usually the power of influencing the other is based upon and exploits his dependence and our reward power over him. But neither this relationship is necessary, nor are these two forms of social power just one and unique form. On the one hand as we saw reward power can be used not only for influencing or forcing B to do something, but also in any form of help and goal adoption where A takes into account B s dependence and uses his Powerover the goals of the other. On the other hand, we have claimed that not all our power of influencing is bases on our power of rewarding (threats and promises). A can induce B to do something by exploiting contagion or imitation, or by activating a new goal, etc Of course, from a formal point of view one can always postulate a dependence relation of zero value or strength. But from a substantial point of view it is false that dependence relations are necessarily mutual. Moreover, Emerson does not distinguish qualitatively between mutual and reciprocal dependence relations; which for us is a fundamental distinction. 48 Only if one adopts a much broader view of Dependence (with a different definition) and of autonomy see (Falcone and Castelfranchi, 2003) one can also consider this form of influencing power as exploiting some dependence of the other agent, but in a different way.

256 256 Cristiano Castelfranchi Another very important difference with Emerson s theory is that we do not believe that a tendency to balance or balancing operation exists. Agents do not attempt to balance their relationships. This is only the emergent result of the indefatigable activity of individuals and groups aimed not at balancing relationships but at not being subject and possibly having more negotiation and influencing power than another. When an agent is inferior in a power relationship s/he generally acts in order to improve her/his position, but not in order to equate their positions. If this happens it is just a (very unstable and dynamic) balance resulting from such a reciprocal activity. This equilibrium is simply apparent, is just an illusion (the picture taken by the observer in special instants, just because s/he is interested in noticing such a balance). Balance is not even a tendency. The real tendency is to continuously prevail, having the best possible position. When an agent is in a balanced dependence and power relation s/he is not necessarily satisfied and quiet. S/he searches more negotiation power (for getting to a better deal) or a greater influencing power (for inducing the other to do whatever s/he needs the other does). What really could be observed is a permanent instability and competition (remember Hobbes sentence). Power can be accumulated and stored Contrary to Friedberg s claim, power can be accumulated and stored; this issue deserves a special point. Even if power were merely relational, relations can be maintained without being currently exploited, and accumulated for future possible uses. Moreover, this merely relational view of power makes its source mysterious. On the contrary social power derives from objective dependence relationships, and those relations emerge from personal powers of agents in a common world, i.e. in potential interference conditions (Castelfranchi, 1998). Power is like position energy (there is not only energy in action, kinetic). For example, people in debt, people I can, at a given moment, induce to do something for me just to discharge their debt, is a potential power, is an accumulation of power. In this sense a Mafia boss can be more or less powerful; in this sense good relationships are a capital that can be invested or not. Analogously, the Power-of threats is a power that you do not need to spend. I have the force for harming you (for example, I have my army, missiles, etc.) but I do not attack you; I just threaten you to do so, and this is enough for inducing you to do what I want. I don t spend my resources, and my threatening capital remains untouched; it can be even augmented, since I can induce you to aggress or to threaten somebody else on my behalf.

257 The Micro-Macro Constitution of Power 257 Power is a relational but also a potential notion. X can do A, it is not actually doing A. In the perspective of a theory of power accumulation a very important issue is: when X uses its power will she lose it? It depends. As for the basic pre-social power: a) Is this power basically based on a consumable resource r, so that if X uses it X wastes it and has less power? b) Or is this resource increased in quantity or value by its use? (like skills, competence, etc.)? c) Or does r remain untouched (like a given piece of information, say a telephone number)? 49 As for social, relational, power, in particular the Power-over or incentive power which is based upon Dependence: If X adopts and satisfies Y s goal, is her Power-over Y finished or augmented? It depends: if Y had just an occasional goal G, he is no longer dependent on X, and X loses her power over Y (as for G); the same holds if r is finished, or if X passes to Y r that can be used by him alone (this time or several times): Y becomes autonomous as for r. If G s satisfaction reproduces Y s goal G (even with greater value) X consolidates her Power-over Y. The same holds if Y trusts X more or reduced his alternatives for the previous choice (market specialization, loyalty). Barnes Also Barnes view is intrinsically social (although with a different, knowledge-based approach). He claims that power is a social phenomenon only relative to sets of interacting individuals. Moreover, there cannot be more power than is known/assumed to exist. (Barnes, 1988). On the one hand, it is rather strange that he does not realize that his (good) notion of power as generalized action capacity, and of possession of power as 49 This is in fact one of the components of power and dependence dynamics, in addition to agents changing their desires and intentions, and agents acquiring or losing skills, competence, knowledge; and in addition to agents leaving or joining a given population or context.

258 258 Cristiano Castelfranchi discretion about the use of the former capacity, equally applies to individuals, collectives, organizations, institutions, etc., and there is no intrinsic reason for restricting it to sets of interacting individuals. On the other hand, we disagree about the exclusive foundation of social power on knowledge sharing and distribution. Power seems just ascribed, given by the others beliefs, in a sense merely conventional. It is not clear what is the role of objective capacities beyond reputation; it is true that the others attitude gives power, but not all the power, not even all the social power. Pörn Dealing with Pörn s approach to power would require a much longer and careful discussion. He distinguishes between two principal forms of power: influence and normative power. Influence is analyzed in terms of control (see below), while normative power in terms of the power of making the action punishable. 50 We are not interested here in his notion of normative power 51 that is not our notion of institutional power. Let us see the other kind of power. As for what we call interpersonal power relations he proposes a much more sophisticated and systematic analysis of all possible influence positions/relations by combining simple influence types (production, counter-production, toleration, counter-toleration, prevention, counter-prevention, leave, counter-leave). All this construction should be retrieved and expanded into a larger theory. The problem in my limited view is that he builds too much while following a formal reasoning upon basic notions and operators that are not so conceptually clear and expressive. In general, since he and his milieu were mainly interested in normative theory, the cognitive foundations are rather weak or unclear (implicit). For example, (like in all the logic of action of the bringing it about that to see to it ) the teleonomic, intentional nature of the action is non explicit; in such a way a lot of relevant points are put aside (for example, the real notion of successful action and the distinction between the intended and motivating results and any other (expected or accidental) outcome of the action). Without an explicit theory of goals we cannot understand for what and about what the power is; also the social one. 50 He studies also some relations between the two: how influence relations can be subject to norms; how normative relations can be subject to influence. 51 Which looks in fact just like the other side of our deontic power (from the point of view of the other agent) but more oriented towards legal theory.

259 The Micro-Macro Constitution of Power 259 Our notions are far from being formal and systematic (exploring all possible positions and their logical relationships), but we believe that they are more cognitively grounded and more fine-grained at the level of the theory of individual action and of social interaction. For example, it seems too simplistic to derive a notion of control and of exercising control as a form of influence relation just from the combination of action operators not so clearly defined: DiDjp = i exercises control over j s doing p (p.17). Given the interpretation of Dip 52, we even ignore whether this influence on j is accidental or not. Apart from this, since the basic model of action is too poor DiDjp might represent very different things: that i provokes j s action, or that i simply acts in such a way that j s independent action be successful (produces a give result). 53 But while the latter represents a case of i s (accidental?) help and of j s dependence on i; the former represents a real influencing the other to do, a real induction. We characterized these as two very different (although related) forms of social power. Moreover, exercising control presupposes having control; however, having control over j does not mean only being able to provoke his behavior but also being able to block or prevent it. Thus DiDjp is not enough for representing the exercise of i control (power) over j s doing p, since i induces j to do but perhaps would not be able to prevent an autonomous Djp. The lack of any theory of the mental representations of the agents may be crucial for power; for example we cannot say that: X can bring it about that p if and only if either X brings it about that p or X forbears to bring it about that p 54. This seems to presuppose not only opportunity but awareness of them which has not been explicitly included in the notions of opportunity and of can. 55 More generally, no theory of social influence can be made without an explicit theory of acting on the decision and intentions of the subject and of his possible understanding and adopting the goal of the influencer. 52 Dip = It follows from what i does that p ; i sees to it p ; i brings it about that p ; i acts in such a way that p ; it is a thing done by i that p ; i does p. (p.2). 53 Balzer (1992) for example gives a similar definition of exerting power over a person, but not ambiguous under this respect (and involving beliefs and intentions). 54 While it is correct to us that to forbear presupposes that X can. 55 I also disagree with Pörn s parafrasing X can bring it about that p with It can be (the case) that X brings about that p. X s power of doing is much more that it can be that X does, the former implies the latter but cointains the idea that thanks to internal features (skills, information, strenght, decision ability) (that Pörn in fact takes into account) including discretion and willingness, and external conditions it can happen that ; that is also different from it can be which is an assertion about what is credible.

260 260 Cristiano Castelfranchi For example, without this it is impossible to distinguish between manipulation and general influence (Balzer, 1992; Parisi and Castelfranchi, 1978). This is enough for giving an idea of the kind of divergences and of the needed for an analytical work of comparison. 11 Conclusions Let us stress just a few points. We (in part) found social power on non-social, personal power (resources, capabilities, etc.), and in this perspective we strongly anchor the notion of power to a goal-directed view of action and agenthood. There is power (or not) only relative to some goal personal at the personal layer, collective or institutional at the other layers. We are not claiming that the individualistic approach is sufficient for the theory of power. We are just claiming that it is necessary. Having a theory of personal power, partially deriving from it the theory of interpersonal powers (comparative, rewarding, and influencing), and connecting it with the theory of institutional power, increases our theoretical understanding and predictions. Our micro-macro cognitive-based approach to social phenomena (Conte and Castelfranchi, 1994) is not aimed at reduction. We simply claim that the attempt to found sociological concepts in a completely autonomous way, without any explicit relationship with the micro-level notions, and refusing to look at the obvious links between the individual and the sociological level, is not a heuristic move 56. It does not make sociological theory stronger. The problem is not that of reducing sociology to psychology, or unilaterally founding sociological constructs on psychological ones, the problem is accounting for the bilateral grounding and the bilateral influence between the micro and the macro layers. Moreover, it is important to consider that what appears as a (partially) individualistic (psychological) foundation is in fact an abstract, agent based foundation. If collective entities (like groups, team, organizations) can be conceived as abstract, complex or high level agents, all the theory that we have just exposed for individuals can be abstracted and applied both to individual agents and to abstract agents. In other words, among groups, organizations, nations we find Power-of, dependence relations, Power-of rewarding, Power of influencing and all the form of social power that we have already called interpersonal but that are in fact inter-agents. The 56 How and why many languages use the same word power?

261 The Micro-Macro Constitution of Power 261 same abstract theory applies to different layers of organization of action and sociality. The real topic of this paper has been how power produces power, and, more precisely, how one form of power founds another form of it, and how one kind of power transforms into another one. This is the Proteus-like nature of power. This is also one of the reasons why power does not at all tend to balance (except for the incessant human competition and ambition) but per se tends to concentration and inequality. 57 We also applied this dialectic view of transforming power into power and of empowering to the relationship between individuals and institutions. Even the magic, count as, performative power of institutional acts is given from the institution to the lay-agent, but hidden is given to the institution by the acceptance and conformity of the mass of people. This mutual (unaware) empowering is far from being fully and explicitly modeled. Let us also remark how not all social (and societal) reality is acceptance - based, a collective construction; the conventional result of some explicit and organizational, or diffused and tacit agreement and pact. Part of social reality is merely emerging and self-organizing in an objective way; it is given, independent of human awareness, decision and even acceptance. (Conte and Castelfranchi, 1994; Castelfranchi, 2001). This is also the reason why we spent enough time on the notion of Dependence and on the Dependence network. Obviously, when agents realize that they are dependent on others, when in general they realize their objective interference, interests, etc. relationships (Cognitive Emergence; Castelfranchi, 1998b) consequences are important, as we just saw for example about the Power-over and the Power of Influencing. A lot of points remain to be clarified, a lot of relationships between the different forms and layers of power must be explored. We just wish to have persuaded the reader that: i) a micro-foundation of the social theory is useful and necessary, also because it is not individualistic but applies recursively at different level of organization; ii) that mental ingredients are much needed in this foundation at the individual, social, and institutional level: many forms of powers just depend on or are due to what people believe or accept. 57 This is conversely and unfortunately true also for serious lacks of power like sickness, stupidity, ignorance, emargination, poorness, that tend to reproduce and maintain themselves, and to increment each other reciprocally.

262 262 Cristiano Castelfranchi References Balzer, W A Theory of Power in Small Groups. In H. Westmeyer (ed.) The Structuralist Program in Psychology. Bern. Hogrefe & Huber Pub Barnes, B The Nature of Power, Cambridge, Polity Press. Bratman, M. E Intentions, plans, and practical reason. Harvard University Press, Cambridge MA. Carmo, J. and Pacheco, O Deontic and Action Logics for Collective Agency and Roles. In R. Demolombe and R. Hilpinen (eds.) Proceedings of the Fifht International Workshop on Deontic Logic in Computer Science (DEON 00), ONERA-DGA, Touluse, Castelfranchi, C Social Power a Missed Point in DAI, MA and HCI. In Decentralized AI-1990, Y. Demazau and J.P. Mueller (Eds), Elsevier, North Holland, Castelfranchi, C. 1995a Guarantees for Autonomy in Cognitive Agent Architecture. In M. Woolridge & N. Jennings (eds.) Agent Theories, Architectures, and Languages. Heidelberg, Springer & Verlag, Castelfranchi, C., 1995b Social Commitment: from individual intentions to groups and organizations, In ICMAS 95 First International Conference on Multi-Agent Systems, AAAI-MIT Press, Castelfranchi, C Reasons: Belief Support and Goal Dynamics. Mathware & Soft Computing, 3., Castelfranchi, C Practical Permission : Dependence, Power, and Social Commitment. In: Bell, J., Huang, Z., Parsons, S. (eds.) Practical Reasoning and Rationality. Workshop Proceedings. Castelfranchi, C Modelling Social Action for AI Agents. Artificial Intelligence, 103, 1998, Castelfranchi, C. 1998b. Simulating with Cognitive Agents: The Importance of Cognitive Emergence. In J. Sichman, R. Conte, N. Gilbert (eds.) Multi-Agent Systems and Agent-Based Simulation. Springer, Berlin LNAI 1534, Castelfranchi, C Through the Agents Minds: Cognitive Mediators of Social Action. In Mind & Society, Torino, Rosembergh, Castelfranchi, C The theory of social functions. Challenges for multi-agentbased social simulation and multi-agent learning. Cognitive Systems, Elsevier. Castelfranchi, C Grounding We-intentions in Individual Social Attitudes. In M. Sintonen, P. Ylikoski, K. Miller (eds.) Realism in action: Essays in the philosophy of social sciences. Dordrecht, Kluwer. Castelfranchi, C., and Conte, R The Dynamics Of Dependence Networks And Power Relations In Open Multiagent Systems. In Proceedings of COOP 96, 15-23, INRIA, Sophia-Antipolis, Castelfranchi, C., Miceli, M., & Cesta, A Dependence relations among au-

263 The Micro-Macro Constitution of Power 263 tonomous agents. In E. Werner & Y. Demazeau (Eds.), Decentralized A.I. 3, Amsterdam: Elsevier Science Publishers B Cohen, P. R., and Levesque, H. J Intention is choice with commitment. Artificial Intelligence, 42, Conte, R., and Castelfranchi, C Cognitive and Social Action. London, UCL Press. Conte, R., Miceli, M., & Castelfranchi, C Limits and levels of cooperation: Disentangling various forms of social interaction. In Y. Demazeau & J.P. Mueller (Eds.) Decentralized AI 2, Amsterdam, Elsevier. Conte, R. and Sichman, J.S DEPNET: How to benefit from social dependence. Journal of Mathematical Sociology. Conte, R. and Sichman, J.S Dependence Graphs: Dependence Within and Between Groups In Computational and Mathematical Organization Theory, 8(2), Crozier, M Le phénomène bureaucratique, Paris. Le Seuil. Crozier, M L acteur et le système, Paris, Le Seuil. Dahl, R On the Concept of Power. Behavioral Science, 2, Emerson, R.M Power-Dependence Relations. American Sociological Review. Vol.27, 1, Falcone, R., and Castelfranchi, C., 2000, Delegation-based conflicts between client and contractor, Journal of Intelligent Automation and Soft Computing, TSI Press, Vol.6, N 3, Falcone R., and Castelfranchi C. 2003, Adjustable Social Autonomy, in The Open Agent Society Jeremy Pitt (Ed.),John Wiley & Sons, Ltd. (in press). French J.R. and Raven, B The Bases of Social Power, In Studies in Social Power, D. Cartwright (Ed.) Ann Harbor, Michigan, Institute for Social Research. Friedberg, E.1993 Le povoir et la règle. Dynamique de l action organisée, Edition du Seuil, Paris. Gasser, L Social conceptions of knowledge and action: DAI foundations and open systems semantics. Artificial Intelligence 47: Gilbert, M What is it for us to intend? (Invited talk) CollInt-1, Munich, June, Gould, R.V Collective action and network structure, American Sociological Review, 58, Grosz, B.J., and Kraus, S Collaborative Plans for Complex Group Action. Artificial Intelligence, 86, Hindriks, F.A Social groups, collective intentionality, and the anti-hegelian scepticism. In M. Sintonen, P. Ylikoski, K. Miller (eds.) Realism in action: Essays in the philosophy of social sciences. Dordrecht, Kluwer. Hohfeld, W.N Fundamental Legal Conceptions as Applied in Juridical Reasoning, and other Legal Essays. Yale Univ. Press, New Haven Connecticut, 1966 (Reprint of the original edition of 1913).

264 264 Cristiano Castelfranchi Huberman, B.A & Hogg, T Communities of practice. Performance and evolution. Palo Alto, CA: Xerox Palo Alto Research Center. Jennings, N.R. & E.H. Mandami Using joint responsibility to coordinate collaborative problem solving in dynamic environments. Proceedings of the 10th National Conference on Artificial Intelligence, San Mateo, CA: Kaufmann Jones, A.J. and Sergot, M A formal characterization of institutionalized power. Journal of the Interest Group in Pure and Applied Logic, vol. 4, no.3, Ketchpel, S Coalition formation among autonomous agents. In C. Castelfranchi & J.P. Muller (eds.), From Reaction to Cognition, (Proceedings of MAAMAW 93, Neuchatel, CH: University of Neuchatel, August 24-27).Heidelberg, Springer, LNAI 957. Markovsky, B., Skvoretz, J., Willer, D., Lovaglia, M.J., J. Erger 1993, The seeds of weak power: An extension of network exchange theory, American Sociological Review, 58, Meijers, A Collective Agents and Cognitive Attitudes. Protosociology, Vol. 16. xx xx Miceli, M., Castelfranchi, C., A Cognitive Approach to Values, Journal for the Theory of Social Behaviour, Vol.19, no.2, June 1989, Miller, D. (ed.) 1992a The Location of Power in Exchange Networks. Special Issue of Social Networks, 14. Miller, D. 1992b Predicting power in exchange networks: a brief history and introduction to the issues. In Miller, D. (ed.) 1992a, Parisi, D. and Castefranchi, C Potere. Rassegna Italiana di Sociologia. 2, Pörn, I. 1970, The Logic of Power. Oxford, Basil Blackwell. Santos, F., Jones, A.J., and Carmo, J Responsibility for Action in Organizations: A Formal Model. G. Holmstrom-Hintikka and R. Tuomela (eds.) Contemporary Action Theory, II (Social Action), Synthese Library, 267, Kluwer, Searle, J. R Speech Acts. Cambridge University Press, Cambridge. Searle J.R., 1995, The construction of social reality. N.Y. The Free Press. Shechory, O. and Kraus, S Coalition formation among autonomous agents: strategies and complexity. In C. Castelfranchi & J.P. Muller (eds.), From Reaction to Cognition (Pre-Proceedings of MAAMAW 93, Neuchatel, CH: University of Neuchatel, August 24-27) Springer, LNAI 957. Sichman, J.S Du Raisonnement Social Chez les Agents. Une Approche Fondée sur la Théorie de la Dépendence, PhD. Thesis, LIFIA, Grénoble, France. Sichman, J.S., Conte, R., Castelfranchi, C., & Demazeau, Y A social reasoning mechanism based on dependence networks. In A G. Cohn (Ed.), Proceedings of the 11th. European Conference on Artificial Intelligence. Baffins Lane, England: John Wiley & Sons, pp Thibaut, J. and Kelley, H.H The Social Psychology of Groups. New York, Wiley

265 The Micro-Macro Constitution of Power 265 & Sons. Tuomela, R. and Miller, K We-Intentions. Philosophical Studies, 53, Tuomela, R Collective and Joint Intention (Rosselli Foundation conference Cognitive Theory of Social Action Turin, June 1998). In Mind & Society, Torino, Rosembergh, II, Tuomela, R. and Balzer, W., 1999 Collective Acceptance and Collective Social Notions, Synthese 117, Tuomela, R The philosophy of social practices. A collective acceptance view. Cambridge, Cambridge Univ. Press. van Linder, B. van der Hoek, W. and Meyer, J.J. Ch Formalising Motivational Attitudes: On Wishes, Goals and Commitments. In J. Bell and Z. Huang (eds) Practical Reasoning and Rationality. Proceedings of the DRUMS II Workshop, Windsor, UK, 1996, Veneziano, V. Conte, R. e Castelfranchi, C Formation of partnerships: The mediatory role of dependence networks. Mitteilungen aus den Arbeitskreisen, MASSIM-96, Heft nr. 53, Ulm 5-6 März 1996, Wartenberg, T.E. (ed.) 1992 Rethinking Power. N.Y., State University of New York Press. Weber, M. 1922, Wirtschaft und Gesellschaft. Tubingen, Mohr. Werner, E. 1990, What can agents do together? A semantics for reasoning about cooperative ability. Proceedings of the 9th European Conference on Artificial Intelligence. London: Pitman. Zlotkin, G. & Rosenschein, J One, Two, Many: Coalitions in Multi-Agent Systems, in Pre-Proceedings of MAAMAW 93, Neuchatel, CH: University of Neuchatel, August

266 266 Cristiano Castelfranchi

267 Foundations for a Social Ontology 267 III Social Facts and Social Practices

268 268 Amie L. Thomasson

269 Foundations for a Social Ontology 269 Foundations for a Social Ontology Amie L. Thomasson Abstract The existence of a social world raises both the metaphysical puzzle: how can there be a reality of facts and objects that are genuinely created by human intentionality? and the epistemological puzzle: how can such a product of human intentionality include objective facts available for investigation and discovery by the social sciences? I argue that Searle s story about the creation of social facts in The Construction of Social Reality is too narrow to fully solve either side of the puzzle. By acknowledging different forms of rules for constructing social reality paralleling rules for creating make-believe truths, we can build a more comprehensive social ontology and allow for a more appropriate range of discovery for the social sciences. Nonetheless, I argue that despite the parallels between methods for constructing make-believe and social facts, it would be mistaken to treat talk about social reality as involving a mere pretense to refer. The social sciences study political entities such as nations and laws, economic facts concerning money, debt, inflation and recession, and social facts concerning class structures, race and gender relations. In each case, it is fairly clear that the facts and objects studied by the social sciences are not part of the independent fabric of the world in the way that the facts and objects studied by the natural sciences are instead, in some sense, these facts and things depend on human intentionality, and are creations of our beliefs and way of life. Yet many are skeptical that human beliefs, intentions, and practices can be genuinely creative, since normally (as metaphysicians are fond of pointing out) beliefs that something is the case do not make it so. So a first puzzle that arises for those who would pursue social ontology is whether there can be a reality of facts (that Hawaii is a state, that the Euro is a valid currency), and even a whole range of objects (laws, nations, corporations, money, property) that are genuinely created by human intentionality, and if so, by what means such creation is possible. Even if we can tell a story to explain how human beliefs and practices may be ontologically creative, a more difficult puzzle arises for the would-be social ontologist. For we think of the social sciences as reporting objective facts about these social entities, and discovering facts, and even kinds of things, that

270 270 Amie L. Thomasson were previously unknown. But the very idea that social reality is created by human intentionality is in tension with the view that it can it include objective facts available for investigation and discovery by the social sciences. Most people are accustomed to thinking of creations of human intentionality (if they accept that there are any) on the paradigm of imaginary objects as purely subjective, personal creations, inaccessible to study by others and having exactly those features we think of them as having. No investigations into them would then be appropriate: they would be impossible for people other than the creator, and unnecessary for the creator, since she automatically knows all there is to know of them. So the second puzzle is an epistemological one: How could the social sciences be thought to provide objective knowledge of, or make any new discoveries about social facts and objects created by human intentionality? In this paper I will attempt to make progress on both sides of the puzzle by offering an account of some ways in which social reality may be created that can do justice to range of entities studied in the social sciences, and to the idea that the social sciences may be involved in the discovery of facts and even kinds of facts previously unknown. 1. Searle s Social Ontology John Searle may be credited with bringing something like the above puzzles into recent discussion, motivating his study of social reality by posing the question How can there be an objective reality that exists in part by human agreement? (1995, 2). His work thus provides a natural place to start in seeking a solution to the puzzles of social reality, so I will begin with a brief overview of his account. As we shall see, however, to fully solve either side of the puzzle we will need a broader account of the methods for constructing social reality than Searle offers. According to Searle, accounting for the construction of social reality requires exactly three elements. The assignment of function, collective intentionality, and constitutive rules (1995, 13). The first two, on Searle s picture, are essential to the construction of social facts generally, while the last enters into the construction of institutional facts more particularly. Social reality is constructed by our collectively imposing functions on brute physical reality. In the simplest cases, we assign agentive functions to natural objects (as when we assign some trees the function of being lumber) or create objects that we collectively assign a certain function (as when we create a bench)

271 Foundations for a Social Ontology 271 (1995, 20) in these cases mere social facts are created. More interesting cases arise when institutional facts are created by collectively imposing upon some entity a new function that it could not perform solely in virtue of its physical features, as, for example when we impose upon a piece of paper the function of serving as a valid unit of exchange (money). The latter functions are status functions : functions the object could not perform solely in virtue of its physical properties (1995, 41). In order to impose status functions, we must collectively accept constitutive rules, rules that stipulate that a certain x counts as y in the relevant context C (1995, 44). Thus, for example, there is a constitutive rule that certain pieces of paper issued by the bureau of printing and engraving count as $20 bills, which endows them with the status function of serving as a legal unit of exchange. This marks the difference between merely social facts and properly institutional facts: something s being a screwdriver or a bathtub is merely a social fact, since these things could fulfill their functions of turning screws or holding water just on the basis of their physical constitution; in contrast, something s being a dollar bill, a citizen, or a touchdown is an institutional fact involving functions beyond those that the object s physical properties endow it with, functions that can only be acquired by collective acceptance of constitutive rules stipulating what counts as being a dollar bill, citizen, or touchdown. This story begins to solve the ontological side of the puzzle by showing a method whereby human intentionality can overlay new social and institutional facts on brute physical reality. Searle also solves part of the epistemological side of the puzzle by suggesting that although the relevant facts created are ontologically subjective in the sense that they depend on mental states, nonetheless claims about such facts may be epistemically objective in the sense that their truth or falsity is independent of the attitudes, feelings, and points of view of the makers and hearers of the judgement (1995, 8). I will suggest, however, that while Searle deserves credit both for raising the beginnings of the puzzles and partially solving them, the story he provides about the method of constructing social reality is too narrow to fully resolve either the deeper ontological or metaphysical puzzles of the social world. We can see this by examining two features of social reality that Searle suggests follow from his construction story: 1) the logical priority of brute facts over social facts (1995, 34-5, 56-7) 1 and 2) the self-referentiality of social concepts (1995, 32-4). The idea that social reality, and in particular institutional 1 Although the initial section heading is The Logical Priority of Brute Facts over Institutional Facts, he claims later down that this applies to social facts in general (1995, 35).

272 272 Amie L. Thomasson reality, at bottom depends on brute physical reality seems entirely appropriate, but by the logical priority of brute facts over social facts, Searle seems to have a much more specific thesis in mind namely, that no social objects are created at all, since any purported social or institutional object is really just a material object that is assigned a new function: Such material objects as are involved in institutional reality, e.g., bits of paper, are objects like any others, but the imposition of status-functions on these objects creates a level of description of the object where it is an institutional object, e.g., a twenty dollar bill. The object is no different; rather, a new status with an accompanying function has been assigned to an old object. (1995, 57). This is a result Searle seems to find welcome, since it limits the ontological impact of his study by ensuring that he is not really countenancing any additional objects at all; so-called social objects are merely material objects with new social facts attached to them. And indeed this does seem to follow from Searle s account of how social and institutional reality is constructed, if we take his account as describing the only possible methods for such construction. For on that account, as we have seen, social reality is constructed by means of assigning functions to preexisting objects. In the case of institutional reality, this must be done by collectively imposing a status function (y) on some object (x) that pre-exists this imposition. Although functions may be applied to institutional objects themselves (as a citizen may be made president), at some initial point, this process must bottom out in phenomena whose existence is not a matter of human agreement (1995, 55). In the case of institutional facts, in the formula x counts as y in context C, there must be an initial x term that is ultimately a brute object, which we then determine to count as a y, which we may in turn count as a z, and so on. In fact, although most of his favorite examples concern institutional facts, Searle explicitly takes this feature to hold not just for institutional facts, but for all social facts (1995, 31). And again as long as one accepts that his story of the construction of social reality is the story of how social facts are constructed, it is easy to see that logical priority would hold for mere social facts as well, for whether or not the relevant function is a status-function, it ultimately has to be imposed on some physical object. Thus, e.g. whether the function is being money or being a screwdriver, there must be some (ultimately physical) object to be assigned the function of serving as a unit of exchange, or of turning screws.

273 Foundations for a Social Ontology 273 But however plausible it is for the examples Searle chooses, only a small subclass of those social and institutional entities we concern ourselves with in daily life and study in the social sciences can really be understood as material objects overlaid with new functions. 2 Entities such as the U.S. Constitution, General Motors, or Calvinist doctrine obviously have close relations to individual material objects, but in no case is there some particular material object (with additional social properties) with which we can identify the entity in question they seem to be abstract in the sense of not being identifiable with any particular material object. If we are to take these commitments at face value, rather than attempting to find some way to paraphrase all talk about them into talk about concrete individuals, we will have to admit that new social objects, not just facts, can be created by the cooperation of collective intentionality and brute reality. 3 In the case of the U.S. Constitution, for example, even if the original document in the Archives were destroyed, we certainly would not thereby declare that the United States was a nation without a constitution; a fire may destroy an original document of historical interest, but not the U.S. Constitution itself. In fact, it seems that the social and institutional world is becoming increasingly abstract, as paper messages are replaced by correspondence and fax, web-sites replacing billboard advertising, and so on. Even money is increasingly becoming abstract no particular pieces of paper in a vault, no particular entries in a particular book, no particular magnetic traces on a computer disk, can be that material entity which has the social property of being the money in my bank account, for any of these can cease to be without my deposit evaporating. This makes it increasingly important that a comprehensive social ontology include abstract social objects as well as concrete ones. In short, while he provides a story of how some social and institutional facts may be constructed, it seems that Searle has not provided a complete enough story to explain how human beliefs and practices can create not just new social and institutional facts about familiar material objects, but entire and grand structures of new entities such as religions, nations, militaries, economies, and so on, on the basis of the fundamental physical world. To account for this, we will need to broaden our view of the methods by means 2 David-Hillel Ruben (1997, 447) makes the similar point that Searle s ontology cannot cover such entities as nations, labor unions, and classes. 3 Ruben (1985) argues that social entities such as nations cannot be reduced to individuals (or groups of them), and Gilbert (1996, chapters 6 and 7) offers similar arguments that facts about social groups cannot be reduced to facts about individual members.

274 274 Amie L. Thomasson of which social reality can be constructed. In section 2 below I will return to suggest what other methods of construction should be countenanced to overcome this problem. A second feature that falls out of Searle s analysis of the construction of social reality is the so-called self-referentiality of social concepts. Since, on his analysis, social reality is created by our collectively imposing functions on objects, social concepts are all self-referential, in the sense that, for any social concept F, x is F implies x is used as F, or regarded as F, or believed to be F... (1995, 32). Thus, as Searle writes, Part of being a cocktail party is being thought to be a cocktail party; part of being a war is being thought to be a war (1995, 34). This seems to have the consequence that all social facts created by collective intentionality are epistemically transparent in the sense that their existence logically entails that they are believed to exist, at least by those people whose collective intentions create them. As Searle writes For these sorts of facts, it seems to be almost a logical truth that you cannot fool all the people all the time. (1995, 32). But actually the situation is more complicated, since Searle s original form of a constitutive rule x counts as y in C is ambiguous: is the x term supposed to refer to an individual entity, simply assigning it a new function, or is it supposed to refer to anything of a particular kind? If it always has to be applied to an individual entity, then it follows that every social fact is believed to exist (at least by the people whose collective intentionality establish the fact). But this seems clearly false even for such prime examples of Searle s as money, and would leave no room for discovery of any social facts, except those regarding other societies than one s own. For, as Searle himself notes, a particular social fact (e.g. that a particular piece of paper is a dollar bill) may exist unknown by anyone if the note has fallen through the cracks of the floor in the Bureau of Printing and Engraving (1995, 32). So it seems that in some cases (but not others) we should take the x as referring to a type of thing rather than an individual token thing. As a result, as he puts it, this self-referentiality applies to some social concepts (like being a cocktail party) on a token-by-token basis any individual event must be regarded as a cocktail party to be a cocktail party while other social concepts (like being money) are only self-referential by type (for an object to be money requires only that everyone accepts that things of this sort count as dollar bills). Searle takes self-referentiality to apply not just for institutional facts (such as being a dollar bill), but social facts generally (such as being a cocktail party), providing a way of distinguishing these from natural facts. Self-

275 Foundations for a Social Ontology 275 referentiality is a peculiar logical feature that distinguishes social concepts from such natural concepts as mountain or molecule (1995, 33). Again, as long as we assume that Searle s account of the construction of social reality by means of imposition of function is a the only way social reality is created, self-referentiality does seem to follow. 4 For according to Searle all functions are observer-relative, and so (regardless of whether or not the object could perform the function on its own), in order for some object to have ( be assigned ) the relevant function F, it must be created to achieve F, or accepted as being good-for F, or used for F, etc. Thus screwdrivers and cocktail parties, no less than money and citizens, must be used, regarded, accepted as being such in order to exist. 5 But the idea that all social concepts are self-referential entails that there cannot be social facts of any kind whose existence members of that society do not know about for if there are social facts of a given kind F, people must accept that certain things (or things of certain sorts) are F (and, since their collective acceptance makes it so, they must collectively be right about what things or sorts of things are F). But this severely limits the role the social sciences can play in expanding human knowledge many of the discoveries of greatest moment in the social sciences are of things such as economic cycles, class systems, and power structures, that are capable of existing even if no one believes that anything of the kind exists, or even if no one entertains the relevant concept at all or has prior beliefs about anything of the kind. 6 Call a kind F of social entities epistemically opaque if things of that kind are capable of existing even if no one believes that anything of kind F exists, and conceptually opaque if things of that kind are capable of existing even if no one has any F-regarding beliefs whatsoever. 4 At least this seems to be true for all of the constructed social facts created by collective intentionality. But although facts about collective intentionality itself ought, by Searle s lights, to count as social (1995, 26), they do not obviously exhibit either self-referentiality or a logical priority of brute facts over social facts. 5 This does not, of course, mean that people must accept that, e.g., money is created through the methods described, only that people must accept that this stuff counts as money (even if they are deceived in thinking it is money not merely in virtue of being believed to be, but in virtue of intrinsic value, divine command, or some such) (Searle 1995, 47). 6 Searle acknowledges that social facts can exist without anyone knowing the truth about their method of construction (i.e. that they exist only because we believe them to) (1995, 21-2, 47). This sort of ignorance, however, must not be confused with much more substantive forms of ignorance discussed above: that social facts involving Fs may exist without anyone believing that there is anything of the kind F, or having any F-regarding beliefs.

276 276 Amie L. Thomasson Recessions 7, for example, seem to be epistemically and conceptually opaque. They certainly depend on collective intentionality (and thus qualify as social facts by Searle s criterion), for they depend on collective acceptance of certain monetary systems. But a given economic state can be a recession even if no one thinks it is, and even if no one regards anything as a recession or any conditions as sufficient for counting as a recession. The concept of a recession is recent the first recorded use of the term was in a 1929 article in the Economist but there could have been recessions long before then. Contrary to Searle s general claim, seeming to be a recession is not logically prior to being a recession. Racism is another social phenomenon that depends on certain beliefs (e.g. about the abilities and appropriate roles of people of differing races) and practices (such as differential legal, housing, or employment practices). But something or someone can be racist without anyone regarding anything as racist racism clearly existed long before anyone took any activity or pattern of behavior to be racist. Many of the power structures pointed out by political scientists and sociologists i.e. those involving the economic power of a company in a small community, community-enforced gender roles, or a class structure can exist without anyone having any beliefs about power structures of that kind. 8 Thus it seems we need to broaden the story of the methods that may be used to construct social reality not only to account for the full range of social entities studied by the social sciences, but also to solve the deeper epistemological side of the puzzle regarding the social world: how the socially constructed world could not only be epistemically objective, but could also include epistemically and conceptually opaque kinds of entities that are previously unknown and even unconceived of, and may be discovered by the social sciences. 9 7 The example of recessions was raised in a paper by Mark Lance during an Author Meets Critics session on Searle s The Construction of Social Reality at the 1997 Pacific Division APA. 8 Although Searle spends a great deal of time discussing the place of power in social reality, here again he discusses only those powers with which a person becomes endowed in virtue of collective intentional acts accepting that the person has the relevant power (to drive legally, to rule the country...) (1995, 104). But not all forms of social power are created through explicitly endowing someone or something with power; powers, too, may arise as unnoticed byproducts of existing social institutions. 9 Some may be tempted to the view that, as social phenomena, such things don t really exist until we have the concepts for them. But the point here is precisely that, while some social concepts (such as money) require intentional states involving that concept in order for things of that kind to exist, others (such as recessions) do not. The idea that recessions

277 Foundations for a Social Ontology 277 While Searle s basic picture of social reality as jointly dependent on collective intentionality and brute reality seems exactly right, the limitations of his view may be seen in part as based in artificially limiting the kind or nature of dependence involved on either side. 10 For, from acknowledging that social and institutional facts in general depend on brute facts and could not exist without the independent physical world, it does not follow that each social fact or object depends rigidly on a particular brute object or that each social object is identical with some material object with a new function. Corporations, laws and governments all seem to depend on the physical world for their existence, and are created by real and intentional acts of writing, voting, etc. Yet none of these abstract social entities is identifiable with some particular physical object or brute fact. The second limitation in Searle s theory likewise seems to derive from an overly narrow view of dependence. In fact, his claim that all social concepts are self-referential may derive in part from a simple mistake about what dependence on intentionality amounts to. According to Searle, all social features are observer-relative features; features which exist relative to the intentionality of observers, users, etc. ; i.e. which depend on intentionality (1995, 9). Yet he seems to assume that, if a feature F is dependent on intentionality, then F must be recognized or acknowledged in order to exist. As he writes: it is a logical consequence of the account of the distinction [between observerrelative and intrinsic features] as I have so far given it that for any observerrelative feature F, seeming to be F is logically prior to being F, because appropriately understood seeming to be F [being thought or believed to be F] is a necessary condition of being F (1995, 13). But being dependent on being recognized is not a logical consequence of being dependent on intentionality. For something can be dependent on intentionality without depending on its being believed to exist, or even its being considered at all; it may depend instead on something else s being thought of, or merely on the existence of certain agents with typical intentional abilities and racism do not exist until we have concepts for them, on reflection, has no more plausibility than the idea that electrons did not exist until we developed the concept for them. Something falling under the concept of electron could (and did) exist long before scientists discovered them, and beliefs and practices that fall under the concept of racism could (and did) exist long before the concept or word was known. Our concepts might change our understanding of the world; developing the concept of racism explicitly might even lead to changes in society; but that does not entail that the entities or practices described cannot exist unless we have the concepts to describe them. 10 For details on varieties of possible forms of dependence, see Chapter 2 of my (1999).

278 278 Amie L. Thomasson and practices. Not all social entities arise through human agreement about them; some social entities (such as recessions) may arise as byproducts of our collective beliefs, practices, and existing institutions, without depending on their being recognized or accepted for what they are. 11 So while Searle provides an insightful story about one method via which a world of social facts may be created about which social scientists may make objectively true judgments, the story he gives about their creation is far too narrow to account for either the full range of objects studied by the social sciences, or the full potential of the social sciences to engage in a process of discovery. We need a more complete story to track the variety of ways in which social entities can be created in order to remove the ambiguities of Searle s story, account for the creation of abstract social objects as well as concrete social facts, and explain the possibility of unknown social kinds awaiting discovery by the social sciences. 12 The remaining sections develop the beginnings of such a story. 2. Construction of Social Entities Following Searle (1995, 26), we can begin by noting that the simplest, most basic social facts on which all other social entities depend are facts of collective intentionality: That a group of people collectively believes or desires something, accepts a certain principle, etc. Given the basic facts of collective intentionality and the surrounding world in which collective intentions occur, the rest of the social world emerges. Certain sorts of social facts and objects are intentionally created (these I will refer to as constructed, since this implies intention in creating them); others are the unintended byproducts of collective intentions and/or intentionally created social constructions (these I will refer to as generated ). I will discuss methods of intentional construction in this section, and return to discuss generation in section This parallels to some extent Adrian Haddock s (2002, 10-11) diagnosis of Skinner s mistake in discussing actions, as accepting that actions must be intentional, but not that there may nonetheless be many true descriptions of actions under which they were not intended. 12 The intersections of different varieties of dependence provide several dimensions of variation in the types of object that exhibit this joint dependence. For an outline of the kinds of object that may be jointly dependent on intentionality and material reality, see Chapters 8 and 10 of my (1999).

279 Foundations for a Social Ontology 279 As I mentioned at the outset, imagination provides the usual paradigm of human intentionality being creative. The usual stereotype has it that socalled creations of the imagination could not be the objects of objective knowledge or discovery. But those who, like Kendall Walton, have worked more seriously with the idea of make-believe, studying public childhood games of make-believe (rather than private fantasy) and adult forms of makebelieve involving representational works of art, have shown the variety of ways in which fictional truths (truths in the world of a game of make-believe or... work of art (Walton 1990, 35)) may be generated, and how these may be independent of the individual beliefs and imaginings of participants in games of pretense (Walton 1990, 42). Thus starting from the make-believe case can, I think, provide us with some insight about the different forms the creation of social facts can take, and how these can be (at least to some extent) objective matters for discovery, independent of the beliefs of the participants. The games of make-believe to be considered are public games of imagination involving external props perhaps dolls, trucks, tree stumps or mud pies in children s games, or representational works of art in adult games. The parallels between creating fictional truths in such games and creating social facts are many and striking. In both cases, there is a joint dependence on collective intentionality and also on certain external physical objects and facts (e.g. props and objects of imaginings in the make-believe case, pieces of paper or rivers in the social case). In both cases, the role of the mind-independent world in creating these facts gives them a relative independence from human beliefs and intentions. And in both cases, there are various sorts of rule and method that can generate fictional or social facts, and fictional or social objects from these two bases of collective intentionality and external reality. But in both cases, although the rules of the game (Walton s principles of generation and Searle s constitutive rules ) must be at least implicitly understood and accepted in order to do their work, they may or may not be explicitly stipulated. They may simply be embodied in background knowledge and practices as we, say, become competent players of children s games, appreciators of art, or members of society and need not be something the participants explicitly have in mind or can verbally articulate (Walton, 1990, 38). Thus although, with Searle and Walton, I will continue below to speak in terms of collective acceptance of rules of certain forms (in order to make the underlying logical structures clear), that is not to imply that the participants must actually be able to articulate such rules they may be embodied in collective reactions and practices (e.g. involving what we do

280 280 Amie L. Thomasson or do not accept as money) rather than formally stipulated and explicitly accepted. The relevant rules can take various forms. The simplest rules in games of make-believe involve de re demands, that is, they require that we imagine something of a particular object: of a certain child (say) we are to imagine that she is the queen, or of a particular stump, we are to imagine that it is a bear. Similarly, the simplest cases for constructing social facts and objects involve de re reference to a particular object, and collectively accepting that it counts as having a certain social status (i.e. that this person counts as our leader, this river counts as the boundary of our territory, etc.) That is, more formally, individual social facts (that this is the boundary of our territory, that this person is our leader) and individual concrete social objects (boundaries, leaders) may be constructed by (formal or informal, tacit or explicit) rules of the form: Singular Rules: 1. (Of a) We collectively accept: Sa (where S names a social feature) Call constitutive rules of this form Singular Rules. Although outsiders and even certain individuals within the relevant society may be wrong about the fact that this individual is S, or may discover it (as children in the society will as they grow up), such individual social facts and objects cannot exist without being collectively believed to exist within the relevant society; they are epistemically transparent. These facts and objects may be the objects of social scientific discovery only for social scientists in a different context (from a different culture or time). But this is certainly not the only method for creating fictional truths using collective intentionality and extant objects, nor is it the only (or even most typical) process for creating social facts. As Walton notes, games of makebelieve may also be driven by certain general conditional rules, say, that if there is a stump at a certain place, one is to imagine that there is a bear there (1990, 40) so that, in the context of the game of make-believe, stumps count as bears. 13 Similarly, most typically, social facts in modern societies are created 13 Walton s distinction (1990, 39-40) between categorical and conditional rules does not line up evenly with the above distinctions between singular, universal, and existential rules. Nonetheless, his example of a conditional rule, that (all) stumps count as bears, is the inspiration for isolating universal rules as described above. His categorical rules simply prescribe that certain things be imagined to be the case, while conditional rules demand that certain things be imagined to be the case on the condition that other (real-world) facts hold.

281 Foundations for a Social Ontology 281 by the acceptance of constitutive rules that stipulate that if anything meets certain conditions, then it counts as having a particular social feature; i.e. any bills of this type issued by the bureau of printing and engraving count as twenty dollar bills. I will call these universal rules. Universal Rules: 2. For all x, we collectively accept that (if x meets all conditions in C, then Sx) Distinguishing singular rules from universal rules removes the ambiguity of Searle s formula x counts as y in context C, and makes evident the different epistemic relation members of the relevant society have to social facts of each type. For when fictional truths are created by universal rules, discovery of token facts in the game is possible. Thus, as Walton points out, the children of the example above may genuinely discover that there is (fictionally) a bear hidden in the thicket (1990, 40). Similarly in the social case, token social facts (that this is a $20 bill) and objects (the $20 bill) created via universal rules may exist without anyone knowing of their particular existence (Searle 1995, 32-3), since no one may know (or notice) that a particular thing fulfills the requirements of C. Indeed formulating the rule in this way, as a universal conditional rule, makes it clear that even for social facts directly constructed by this method, people may not know that anything meeting the relevant conditions exists, and so may not know that there is anything of the relevant social kind. Thus, for example, we could collectively accept that whoever pulls the sword from the stone counts as king, without knowing that someone has done so and thus become king. The conditions specified in C may be of many types they may involve other social concepts, they may be deferential rules appealing to authorities, and they may even appeal to future states of the world. The actual rule accepted for money, e.g. probably does not directly specify the relevant conditions on properties and means of production, but rather is something more like this: we accept that if an object meets the government standards, it counts money. The fact that the rules primarily accepted may be deferential to detailed constraints stipulated by others makes evident another potential source of some forms of ignorance and error regarding social facts in one s own society, even though these are at bottom created by collective acceptance. Moreover, where social concepts application conditions appeal to future states of the world, some ignorance at least at a time is not accidental but essential. Thus, e.g. we accept that an action counts as scoring the winning goal if it counts as scoring a goal in a situation that eliminates a

282 282 Amie L. Thomasson tie, and there are no further goals scored during that game. But whether or not a goal scored with two minutes to go is the winning goal is something no one can know until later. 14 The fact that universal laws appeal to conditions in the world, where we may not know if those conditions obtain, explains how it can easily be the case that many or even all people may be ignorant of the existence of facts of certain types, even when the existence of such facts depends on the collective acceptance of constitutive rules regarding them. This method of creation does not, however, provide for the existence of conceptually opaque kinds of social facts: Facts of a kind S for which no one has any S-regarding beliefs in fact still, in this case, there must be collective acceptance of conditions sufficient for being S, even if it is unknown whether these conditions are met. So these cases still cannot make sense of the possibility of social science discovering new kinds of social object (recessions, racism, etc.) that had previously been unthought of. (I will return to this in Section 5.) Moreover, it can only yield further concrete social objects and facts, for it merely assigns a new social status to material objects, and so still cannot account for the creation of abstract social objects. There is, however, a third method for creating fictional truths beyond adopting the singular and universal rules so far described. Consider the games of make-believe involved in discussion of the content of a work of literary fiction. In those cases, we are not typically asked to imagine something of a particular extant person, or of people meeting certain qualifications. Instead, we are simply required to imagine (de dicto) that there was such and such a person engaged in such and such activities Here, seemingly, the rules of the make-believe (that we pretend that what the book says is true) introduce a new fictional character rather than just introducing new, fictional, facts about extant people or things of certain kinds. Similarly, there are rules for creating new social objects, of the form: Existential Rules: 3. We collectively accept that (if all conditions C obtain, then there is some x such that Sx) Call these existential rules, since, unlike the others, they introduce the existential quantifier into the intensional context of collective acceptance, 14 The example is (modified) from Haddock (2002), which provides a detailed treatment of cases in which the application conditions for a concept depend on the occurrence of future events. Haddock suggests that there is a harmless sense in which such facts only come into existence later, since these involve the original events in a mere Cambridge change.

283 Foundations for a Social Ontology 283 and thus, unlike the others, seem to ensure the creation of new social objects, not just the application of new social statuses to extant physical objects. For example, we collectively accept that, if a majority of members of congress vote to approve a certain bill, then a law is created. Here again, as in the second case, the kinds of facts created may be epistemically opaque, since there is some relative independence of the existence of members of the social kind S from people s belief in their existence, as everyone may lack knowledge that the relevant conditions C obtain, and thus fail to know of the existence of one or all things of kind S. This accounts for the obscurity of many institutional facts in our own society for we collectively accept, say, that any act passed by congress counts as a law, but most of those acts passed remain unknown to most members of the relevant society whose base-line collective acceptance empowers congress to create laws, and so there may be laws few people know of despite the fact that the very existence of laws depends on a certain form of collective acceptance. But despite these forms of relative epistemic independence, facts of these kinds remain conceptually transparent; indeed certain facts about the nature of the kinds of social entities constructed by means of the last two kinds of rules must be known. 15 For each such social kind S, necessarily there is something that is S only if some constitutive rule is collectively accepted that lays out sufficient conditions for something to be S (or for there to be an S). Since those rules establish the relevant conditions, they must be correct. Thus nothing of the kind S can exist without there being S-regarding beliefs (indeed without members of the relevant society collectively knowing of certain sufficient conditions for something to be S, or for there to be an S). So none of these forms of construction can fully solve the epistemological side of the puzzle and account for the ability of social science to discover sorts of entity no one had conceived of before. Recognizing all three methods described above for creating social entities does, however, enable us to acknowledge the existence of abstract as well as concrete social entities, and thereby to resolve the deeper metaphysical puzzle by showing how this kind of constitutive rule enables new kinds of entities (laws, governments, corporations ) to be created, rather than just adding new facts and labels to extant material entities. 15 For further details on the relation between constructed social kinds and our knowledge of them, see my (forthcoming).

284 284 Amie L. Thomasson 3. Are We Just Pretending About the Social? But some might doubt that the ontological puzzle has been solved in the way I planned. Many philosophers including Walton himself (1990, 42) would say that the so-called fictional truths are no truths at all; speaking of fictional truths is just a way of speaking of what sorts of pretense are licensed by the game of make-believe. Similarly, most (including Walton) would say that really there are no fictional objects or facts generated by the principles of generation accepted in games of make-believe; it is just licensed in the relevant game to assert that there are such facts or objects. (So, e.g., there is really no Tom Sawyer created by accepting the rule that what the book says is fictionally-true. Instead, as Walton would say, the Twain books authorize a certain game of make-believe such that one who says There once was a boy called Tom Sawyer makes it fictional of herself that she speaks truly.) 16 As a result, the strong parallels identified above might lead some to suspect that talk about social facts and objects, too, is really just pretenseful talk; we pretend that things have acquired a new property (say, being a licensed driver or a president) when in fact there are really no such properties; or we pretend that a new entity has been created (a law or a corporation) when really there are no such things. This would explain the giddiness Searle describes feeling when contemplating the structure of facts built up by human agreement in the social world, and his sense that there is an element of magic, a conjuring trick, a sleight of hand in the creation of institutional facts Surely when you get down to brass tacks, these are not real facts (1995, 45). Perhaps human intentionality, after all, can create no new facts or objects, but only make it true that, in our pretense, we accept certain claims indeed such an idea seems to be built into the use of the term fictitious names in the U.S. for the legal names of corporations. Perhaps, then, all talk about the social world can be understood as pretenseful, so that the real solution to the metaphysical puzzle of the social world is to deny that we need any story of how these entities can be created, since there are really no such entities (though we 16 As I have argued elsewhere (1999, ), although the pretense reading may be suitable for internal discourse about what happens according to the story or in the world of the fiction, difficulties arise in applying the pretense account to external talk about fictional characters as characters, appearing in literary works, created by authors, etc. These problems, I argue, give us reason for accepting that there are fictional characters despite the role of pretense in internal discourse about them. So I do not personally fully endorse the account of fiction given by Walton and cited above.

285 Foundations for a Social Ontology 285 are all co-implicated in a pretense that there are); we need no social ontology at all, as we can provide instead a fictionalist account of the social. I think many would find this a welcome result, with an admirably spare ontology. But I think such a move would be mistaken, for there is a crucial disanalogy between the case of make-believe and the case of social reality. 17 If children in a game of make-believe agree that stumps count as bears, then although it may be fictional that a particular stump is a bear, that alone cannot make it true simpliciter that the stump is a bear, since terms like bear also have uses outside of the make-believe situation, which set up application conditions for them (i.e. that the thing designated be of the same species as all or most things in the group of entities ostended as bears) that are not met by stumps. Similarly, if a work of literary fiction claims that there is a man who is twelve feet tall, that alone cannot make it the case, since the conditions necessary for there to be a twelve-foot tall man cannot be met merely by people accepting that it is so. The same does not hold true, however, for many social kinds. If two tribes accept that a particular river counts as the boundary between their territories, it is the boundary for that is all it takes for it to be a boundary. If two parties agree that signing a statement on a piece of paper counts as making a deal between them (one that can t be broken merely by destroying the paper, and thus something not identical with the paper itself), then they have made a deal, for that is all it takes to make a deal. Most terms for constructed social kinds have no other use (outside of the system of social rules) establishing deep application conditions that cannot be met merely through social agreement. On the contrary, most such terms are such that, as a matter of meaning and rules of use, things falling under that concept exist if (and only if) the relevant constitutive rule is collectively accepted and the conditions stipulated therein hold. 18 Thus for example, if one believes that a certain society collectively accepts that two unmarried people who knowingly and willingly take prescribed vows before a justice of the peace and two witnesses count as married (and 17 That is, between the case of internal make-believe or fictional discourse and discourse about social reality. There are, however, important parallels between external fictional discourse and discourse about social reality. See note 19 below. 18 This helps to answer more fully another puzzle Searle (1995, 94) raises: What sorts of facts can and cannot be created by collective agreement? Searle answers only that facts involving certain kinds of powers can be created just when collective recognition of a certain power is constitutive of having it (1995, 96), but he does not provide further criteria to distinguish which cases those include or why.

286 286 Amie L. Thomasson are thereby endowed with all the relevant legal rights and social status), and believes that some people fulfill these conditions, it makes little sense to ask with Searle (in his toughest metaphysical mood ), is making certain noises in a ceremony really getting married? (1995, 45), or to deny that there are marriages, or to hold that all talk about marriages (like talk about dragons) must be pretenseful or mistaken. Since those conditions are jointly logically sufficient for the existence of the relevant social entity, accepting that there is a collectively accepted constitutive rule and that the conditions stipulated therein hold, but denying that there is the relevant social entity just shows a misunderstanding of the conditions for there to be such a social entity. 19 Thus the general form of an answer to the original ontological question of how it is possible for mere human agreement and intentionality to create social facts and entities such as marriages, corporations, and laws, on top of the brute facts of the world turns out to be relatively simple: Because all that many social concepts require for their application is that certain constitutive rules are collectively accepted, and (in the case of singular rules) that they are applied to an extant object; or (in the case of universal or existential rules) that the conditions they stipulate as sufficient for a thing to be/there to be something of the social kind, hold. For many constructive social concepts (unlike, e.g., natural kind concepts), given that, the existence of something of the relevant social kind is logically ensured; nothing more is needed. 4. Generation of Social Entities While we can agree with Searle that social concepts are alike in requiring, as one of their conditions of application, the existence of collective intentions, in order to found a more complete social ontology and account for possibilities of discovery in the social sciences we must recognize that social concepts are nonetheless extremely diverse in their application conditions, including the ways those conditions appeal to collective intentions. 20 Thus as we have seen, for some social concepts S, collective agreement that things meeting 19 As I discuss elsewhere (2001, 325-6) (roughly following Schiffer (1996)), external discourse about fictional characters has a similar feature: According to the meaning and rules of use of terms like fictional character, it is logically sufficient for there to be a fictional character (not a person), that someone tell a story in the right kind of society mentioning the relevant name and attributing its referent some characteristic. 20 Haddock (2002, 22) draws out some of the diversity in application conditions that social concepts, action concepts, and other concepts may possess.

287 Foundations for a Social Ontology 287 certain conditions count as S, combined with the fact that those conditions hold, is necessary and sufficient for the creation of things of that kind. Nonetheless, it is important to note that for other social concepts, collective agreement about what it takes to be of that kind is not at all necessary for there to be things of the kind. The concepts discussed so far have been mainly those that participants in a social world use in constructing (conceptually transparent) entities in their own society; as a result the participants must have some knowledge of the existence or nature of the constructed objects or kinds. But things are quite different for social concepts defined from the outside by those reflecting on or analyzing social practices, beliefs, and institutions. These are conceptually opaque: they may require the existence of certain forms of collective intentionality for the existence of members of the kind, without requiring any sort of collective belief or agreement about things of that kind. As mentioned above, the most basic social facts are those that simply involve collective beliefs and intentions. Even these alone (before one gets to the direct construction of further social entities) can lead to the inadvertent generation of social facts of kinds unintended by the participants and that may in principle exist unrecognized by anyone. 21 Collective beliefs that (say) people of other races are inferior and collective practices in systematically denying jobs, housing and other benefits to people based on race are themselves sufficient to ensure that that society has the additional social feature of being racist, without the need for anyone to accept any constitutive rules regarding racism. A society with widespread collective beliefs that supernatural occurrences can be brought about by performing certain rituals, automatically is a society that has the additional social feature of being superstitious, without the need for anyone to accept any constitutive rules about what counts as superstitious. In each case, the mere existence of the relevant collective intentions, perhaps in the relevant context, is logically sufficient to ensure that the relevant social fact of kind S obtains, without the need for 21 The difference between constructing social entities and generating them (perhaps unintentionally) by means of collective intentionality is parallel to the difference Alvin Goldman (1970, 18) marks between the conventional generation versus simple generation of acts from basic intentional bodily actions. Whereas in the conventional case, the generation of one act (signaling a left turn) from another (sticking one s left arm out) relies on the acceptance of a rule that doing the latter counts as doing the former, in the case of simple generation, doing an act of the relevant kind in the relevant circumstances (jumping 6 foot 3 inches) automatically ensures the performance of another act (outjumping one s opponent) (Goldman 1970, 26-27).

288 288 Amie L. Thomasson anyone to have any thoughts about facts of kind S at all. (These may be required for defining the word S, but not for facts of kind S to exist). So some (epistemically and conceptually) opaque social facts may be generated just through basic collective beliefs, desires, etc. But there are even more possibilities for previously un-thought-of kinds of social facts that can be generated based on the directly constructed social world. Once a social world is constructed, there will be all sorts of interesting patterns and causal relations within it economic cycles, patterns of human settlement and property use, of human behavior, of distribution of goods and status, and so on, many of which the participants in the relevant society may lack any concept of or beliefs about (each being interested solely in his or her own transactions). While these relevant patterns and relations could not exist without the social world and the collective intentions that construct it, they can exist quite well without anyone having any thoughts about or recognition of those patterns as such. Social scientists may then discover such patterns and invent labels for them (perhaps with definitions that stipulate the conditions to be fulfilled for a label such as recession to apply). But the features discerned may nonetheless exist independently of the beliefs of social scientists, and may count as genuine discoveries, despite their dependence (at some level) on the intentional states of the members of the participating society. Thus, for example, money paradigmatically exists as a constructed, conceptually transparent social kind, as do companies, stocks, and all the various trade transactions that contribute to an economy. But once established, there may be all sorts of interesting patterns in the resulting system that may be discovered and studied, and terms describing them may be stipulatively defined. If an economist studying economic cycles defines recession as two or more consecutive quarters of negative growth, then the existence of recessions certainly depends on collective intentionality (since there could not be economies at all without monetary systems and trade transactions, all of which require collective intentionality), but requires no beliefs of anyone s about recessions (not even the beliefs of the defining economist) at all, for according to the conditions stipulated, there is a recession when there are two or more consecutive quarters of negative growth, regardless of whether or not any economist has defined the term recession. In the case of generated as constructed social entities, whatever the conditions are that are criterial for being of a certain social kind, we have reason to say that there really are things of that kind (that we are not merely pretending) as long as we have reason to think that those conditions are fulfilled. As in the case of marriages above, if one grasps the concept of a recession and

289 Foundations for a Social Ontology 289 knows that the relevant conditions are sometimes fulfilled, it makes little sense to ask whether there really are recessions. Accepting that social facts and entities may be generated as by-products of collective intentionality itself and of (perhaps a multitude of) constructed social facts enables us to accept the dependence of the social world on collective intentionality and brute reality without giving up the idea that many social facts, objects, and kinds may not be known or even conceived of by participants, but remain in need of discovery by the social sciences, and thus to see the way to solving the remaining corner of the epistemological puzzle of social reality. Whether or not the story developed above provides a complete account of ways social reality can come into existence, I hope that it provides the basis for explaining how it is possible for there to be a wide range of social facts and objects built up on the basis of the independent physical world and collective intentionality, and how the social world can nonetheless be subject to investigations and discoveries by the social sciences. I also hope to have provided at least some initial reasons for taking such creations ontologically seriously, rather than taking talk about them to be mere pretenseful talk. If that much has been achieved, we will have at least laid the foundations for a social ontology. References Gilbert, M Living Together: Rationality, Sociality, and Obligation. Lanham, Md.: Rowman & Littlefield Publishers. Goldman, A A Theory of Human Action. Princeton: Princeton University Press. Haddock, A Rewriting the Past: Retrospective Description and its Consequences. Philosophy of the Social Sciences 32 (1): Hornsby, J Collectives and Intentionality. Philosophy and Phenomenological Research 57 (2): Hund, J Searle s The Construction of Social Reality. Philosophy of the Social Sciences 28 (1): Ruben, D. H The Metaphysics of the Social World. London: Routledge. Ruben, D. H John Searle s The Construction of Social Reality, Philosophy and Phenomenological Research 57 (2): Schiffer, S Language-Created Language-Independent Entities. Philosophical Topics 24 (1): Searle, J The Construction of Social Reality. New York: The Free Press.

290 290 Amie L. Thomasson Thomasson, A. L Fiction and Metaphysics. Cambridge: Cambridge University Press. Thomasson, A. L Ontological Minimalism. American Philosophical Quarterly 38 (4): Thomasson, A. L. (forthcoming). Realism and Human Kinds. Philosophy and Phenomenological Research. Walton, K Mimesis as Make-Believe. Cambridge, Massachusetts: Harvard University Press.

291 On the Objectivity of Social Facts 291 On the Objectivity of Social Facts Antti Saaristo Abstract It is a commonplace that social facts are objective in the sense that we cannot change them at will. A further platitude is that in another sense social facts are not objective, since they are fundamentally dependent on human practices. This paper presents a conceptual framework for analysing these seemingly contradictory intuitions. I argue that although John Searle s distinction between epistemic and ontological objectivity takes us in the right direction, Searle s discussion is nonetheless insufficient for explaining what it is in the nature of social facts that gives rise to the opposing intuitions. I argue that a Durkheimian account, especially as developed by Uskali Mäki, can fare better. Finally, I show how the Durkheimian account serves as the conceptual basis for distinguishing between methodological individualism and different forms of methodological holism. 1. Introduction Are social facts objective? Obviously this is a philosophical question par excellence, since in order to answer it one must begin by performing the most philosophical of all routines. In other words, one must lean back, light one s pipe, think 1 for a while and then, looking from beneath one s knitted brows, pronounce with great seriousness: Well, it all depends on what you mean by objective. To get a clear grasp of the problem, let us look at a simple example. Since in what follows I will concentrate largely on Searle (1995), let us also use Searle s favourite example of a social fact, namely the fact that the pieces of paper in my pocket are money. 2 The monetary status of these pieces of paper 1 I gratefully acknowledge the support I have received when writing down my thoughts on this topic intellectually from J. McKenzie Alexander, Nancy Cartwright, Peter Dietsch, Steven Lukes, Eleonora Montuschi and Raimo Tuomela, and financially from the Helsingin Sanomat Centenary Foundation and London School of Economics. 2 There are all sorts of social facts. Some are more objective than others. Within the limits of this paper I of course cannot discuss the distinguishing features of all the different social facts. Instead my goal is to explicate the core elements of the objectivity of paradigmatic social facts. To use Hacking s (1997) terminology, I am in the business of explicat-

292 292 Antti Saaristo is quite objective. Regardless of how much I hope that also other pieces of paper in my possession would be money, which pieces are money and which pieces are not is beyond my control. In this sense the fact seems to be objective. However, the pieces of paper that are money do not have their monetary status in virtue of their physical properties. It is not the physical structure of the pieces of paper qua physical entities that causes them to function as media of exchange. Rather, as for example Searle (1995) and Tuomela (1995, 2002) emphasise, the reality of the status is due to the fact that people in general think of them as money and accept them as media of exchange, as money. In other words, the pieces of paper function as media of exchange in virtue of the social status assigned to the pieces, and the assignment and constitution of this status is due to collective acceptance (see Searle 1995, Tuomela 1995, 2002). Remove humans and their practices, and social facts and social reality disappear as well. In such a situation the pieces of paper would not be money anymore, even though they would continue to have all the physical properties they had when they were money. This seems to suggest that the fact that the pieces of paper in my pocket are money is not objective after all. The following definition captures the standard common-sense notion of objectivity: (1) A fact is objective iff it obtains independently of us. 3 The fact that the pieces of paper in my pocket are money cannot be objective in the sense of Definition (1), since, as was explained, the monetary status of the pieces obviously depends on human practices and, hence, the obtainment of the fact indeed depends on us. But Searle s example of money also shows that any notion of objectivity that simply implies that social facts are not objective cannot be fully satisfactory. An acceptable and interesting theory of the objectivity of social facts must be able to account for the intuition that there is something perfectly objective in social facts. After all, I ing the definitive logic of social facts, not of classifying different possible empirical scenarios if the present theory is to be applied in empirical contexts, all sorts of relaxations may have to be done, depending on the particular context (see Tuomela 1995, 2000, 2002). 3 This definition is analogous to some standard textbook definitions of metaphysical scientific realism. Hence, even though I have chosen to conduct my discussion in terms of objectivity, the arguments of this essay are translatable into a discussion on the prospects of scientific realism in the social sciences (for a similar line of thought, see Mäki 1996).

293 On the Objectivity of Social Facts 293 have been calling the thesis that the pieces of paper in my pocket are money a fact. An adequate theory of the objectivity of social facts must explain why I am justified to do so. In fact, Searle thinks, that this ambiguity concerning the objectivity of social facts captures the core puzzle in understanding social reality. In short, the fundamental problem is to explain how it is possible that there are portions of the real world, objective facts in the world, that are only facts by human agreement (Searle 1995, 1). This means that in order to understand properly the objectivity of social facts we cannot be satisfied with Definition (1). The next section examines Searle s attempt to explicate the exact nature of the objectivity of social facts. I argue that even though Searle s account succeeds in capturing certain essential features of social facts (and of social reality in general), it nonetheless fails to offer the best possible framework for discussing the objectivity of social facts. Section 3 elucidates the reasons for my dissatisfaction with Searle s view, and Sections 4 and 5 represent an alternative framework. Section 6 ties up loose ends by explicating the different possible ways of interpreting the new framework, offering also a basis for addressing the problem of holism in social science more fruitfully than has traditionally been done. Note that even though I am dissatisfied with Searle s discussion on the objectivity of facts, the framework I offer as a substitute is based largely on Searle s theory of the construction of the social world. Hence my criticism of Searle is primarily that his characterisations of facts cannot do justice to the richness of his own philosophical theory regarding the essence of social facts. 2. Searle on the Objectivity of Social Facts The starting point of Searle s (1995, 7 ff.) discussion on the objectivity of social facts is the idea that a universal notion of objectivity, such as the one expressed by Definition (1), cannot possibly be rich enough for social facts. Rather, objectivity has several different senses, two of which are crucial here. According to Searle, in order to capture both of the mentioned intuitions we must distinguish between ontological and epistemic objectivity. Ontological objectivity and subjectivity, Searle tells us, are predicates of entities and types of entities (Searle 1995, 8). An ontologically subjective entity is dependent on a perceiver or a mental state. Searle s example is pain. Brute natural objects, such as mountains (Searle s example), are ontologically

294 294 Antti Saaristo objective entities. I will conduct my discussion mainly in terms of facts rather than entities, and by ontologically objective facts I mean facts whose obtainment is independent of human activity (see Section 4 below). Epistemic objectivity and subjectivity, then, are properties of statements, not of entities. An epistemically subjective statement is such that its truth or falsity cannot be settled objectively, because the truth or falsity is not a simple matter of fact but depends on certain attitudes, feelings, and point of view of the makers and hearers of the judgment (Searle 1995, 8). Typical examples of such subjective statements are statements that are based on the taste and attitudes of the maker of the statement. Searle s example is the statement that Rembrandt is a better artist than Rubens. Epistemically objective statements are such that their truth-values are not dependent on attitudes or tastes. Searle s example is the statement that Rembrandt lived in Amsterdam during the year When in what follows I occasionally talk about epistemically objective (or subjective) facts, I use the expression as a shorthand of this notion of the nature of the sentences describing the facts (see also Section 4 below). To the first approximation, Searle s conceptual apparatus appears to meet the abovementioned requirements very well. The ontological notion of objectivity captures the common-sense intuition embodied in Definition (1), allowing us to conclude that social facts are not ontologically objective. The problem with Definition (1), however, was that it could not account for the other intuition we have, namely, that social facts appear to be objective in the sense that their obtainment is immune to our personal attitudes, desires, feelings etc. This, of course, is not a problem for Searle since he can say that although social facts are ontologically subjective, the facts (or, rather, the statements describing them) are nonetheless epistemically objective. Thus Searle s solution manages to avoid the obvious problem with Definition (1), and by doing so it also succeeds in doing justice to the core puzzle of social reality. Notwithstanding these indisputable successes, I nonetheless think that we can give a better account of the objectivity of social facts than the one given by Searle. However, I should emphasise that I do not think that there are strictly speaking any flaws in Searle s argumentation. Rather, his approach is incomplete since it fails to capture some crucial features of social facts. The next section explains what limitations I see in Searle s account.

295 On the Objectivity of Social Facts Shortcomings of Searle s Account There are at least three major (interconnected) shortcomings in Searle s account. (i) It is not clear how the two distinctions are related to each other (and indeed to the structure of social facts), and this poses problems for attempts to construct a Searlean taxonomy of facts. (ii) The account fails to highlight the distinctive character of social facts. (iii) The account fails to increase our understanding as to why social facts are ontologically subjective but epistemically objective. The last two problems are the most important problems here, but let me start with the first problem, since it sheds light on the two more serious problems Searle s Taxonomy Searle seems to think that the two distinctions ontological and epistemic objectivity v. subjectivity enable us to construct a taxonomy of facts. In other words, Searle appears to think that we can classify facts to those that are (a) both ontologically and epistemically objective, (b) ontologically and epistemically subjective, (c) ontologically objective but epistemically subjective and (d) ontologically subjective but epistemically objective. It is not clear that this way of classifying facts will result in anything interesting, largely because the epistemic distinction concerns, as Searle (1995, 8) emphasises, predicates of judgements, whereas the ontological distinction deals with predicates of entities. It is not obvious that the marriage of these two distinctions, operating with completely different categories, will reveal anything valuable about different kinds of facts in general and about social facts in particular. After all, neither of the distinctions is defined in terms of facts. And indeed Searle is led to make some unintuitive assertions about different kinds of facts. The class (a) is clear enough, but what should we think about, say, the class (c). Again, Searle s intuition is respectable enough: For example, the statement Mt. Everest is more beautiful than Mt. Whitney is about ontologically objective entities, but makes a subjective judgment about them (Searle 1995, 8-9). The problem is, though, that it is not clear whether this teaches us anything new about facts. The fact involved seems to be simply that the maker of the statement prefers Mt. Everest to Mt. Whitney in aesthetic terms. Hence the entities most intimately associated with this example preferences, mental states are by Searle s own definition ontologically subjective entities, even though this was supposed to be an

296 296 Antti Saaristo example of ontologically objective entities (what mountains of course are). I do not think Searle says anything false here, but I also fail to see why this kind of classification should be of great importance when addressing the objectivity of different kinds of facts. Similarly for the class (b): Supposedly we should find here value judgements about entities whose mode of existence depends on mental states. An example might be the following: I dislike the pain I am now feeling. Again, it is not clear whether we learn anything new about facts by learning to classify statements like this as belonging to the class (b). The facts involved seem to be simply the two ontologically subjective facts that I am in pain and that I find it unpleasant. Searle s point seems to be to emphasise that statements about pain are epistemically objective in the sense that their truth-values are independent of my attitudes, whereas my dislike is an attitude and hence the same cannot be said about it. However, the truth-value does depend on my feelings (i.e. on whether I do feel pain or not) and hence Searle s (1995, 8) explicit definition seems to render it epistemically subjective. Similarly, when we move beyond first appearances the epistemic status of the judgements about my preferences becomes problematic as well. Preferences are by Searle s definition ontologically subjective, but in what sense can we say that the judgements about them are epistemically subjective? Surely the truthvalues are dependent on my tastes etc., but nonetheless there seems to be something objective in them, since many of my preferences are just given to me in introspection or action, and I cannot change them at will. It seems to me that learning to place facts such as the ones discussed here to the class (b) raises more questions than it answers. Finally, even if we sidestep these problems with epistemic objectivity and subjectivity, there are still other confusing issues left. Searle defines the epistemic objectivity and subjectivity of a given fact in terms of whether the truth conditions of a sentence describing the fact depend on the person uttering the statement or not. Hence the same statement can be epistemically subjective or objective depending on who actually utters it. For example, the statement John Searle likes beer is epistemically subjective when uttered by Searle, but epistemically objective when uttered by someone else. As such this feature of Searle s classificatory system can be seen as an advantage, since it allows for a clear distinction between the first and third-person approaches to the mental facts cognitive psychology deals with (see Dennett 1991 for the importance of this feature and Sections 4 and 5 below for what this implies for the present discussion). However, the price Searle s theory pays for this triumph is the serious problem described in Section 3.2. Also the problem

297 On the Objectivity of Social Facts 297 discussed in 3.3 can be seen as another setback caused by the features that are at play here, since by shifting the focus from the facts themselves onto their descriptions, the ultimate structure of the facts remains, of course, unaccounted for Social Facts and Mental Facts Searle s book The Construction of Social Reality (1995) is about the distinguishing features of social reality. Hence, one would expect the book to explain what, exactly, in the objectivity of social facts is different from the objectivity of other kinds of facts, and what, exactly, in the structure of social facts explains the distinctive kind of objectivity common to social facts. However, despite the obvious importance these questions have for Searle s overall project, Searle s discussion of objectivity fails to answer them. This section explains why Searle s account fails in the first task. The second failure is analysed in Section 3.3. A peculiar implication of Searle s characterisations is that the statements I am in pain and The pieces of paper in my pocket are money have exactly the same status! These statements are about entities whose mode of existence is ontologically subjective, since the existence of the entities depends on us. However, the statements reporting the facts are epistemically objective, since each statement is made true by the existence of an actual fact [albeit ontologically subjective] that is not dependent on any stance, attitudes, or opinions of observers (Searle 1995, 9). 5 4 In Searle s defence I should add that conceivably Searle is not even aiming at any kind of taxonomy of facts, but simply at an analysis of different possible ways of applying the concept of objectivity (recall, however, that Searle himself formulated the core task of social ontology (and of his 1995 book) in terms of understanding the specific nature of social facts). Be that as it may, I think a theory that would succeed in doing justice to the features highlighted by Searle s discussion on objectivity while simultaneously presenting a classification of facts on the basis of their fundamental structure would in any case be preferable to Searle s account. In the sections below I seek to offer such a theory. 5 This way of arguing for the epistemic objectivity of social facts seems to be in tension with what Searle says about social facts. For Searle, social facts are based on collective acceptance; social facts are only facts by human agreement (Searle 1995, 1). Now surely human agreement is tied to human attitudes. However, Searle can hold that a social fact is epistemically objective due to its independence of the attitudes of the person observing the fact here and now, even if the fact is not independent of attitudes in general. In Sections 4 and 5 this distinction is explicated and its importance is made clear. Here it suffices to say that the fact that this confusion remains unresolved is one more reason why

298 298 Antti Saaristo The ultimate goal of Searle s discussion is to highlight the distinctive features of social facts, but his discussion in terms of ontological and epistemic objectivity fails, in the end, to distinguish between social facts and the class of all too familiar mental facts. Surely this is a shortcoming in a theory aiming to explicate the distinguishing features of social reality in general and social facts in particular. Indeed, given the context of Searle s discussion, this ambiguity in Searle s line of thought forms a sufficient reason for opting for a more resourceful framework What Makes Social Facts Special? Philosophers who accept the impossibility of private languages and advocate a communalist theory of language, meaning and mental content (see, for example, Williams 1999) might want to argue that actually the fact that Searle s classifications treat the sentences I am in pain and The pieces of paper in my pocket are money as having the same epistemic status and referring to entities of the same ontological status should be seen as an advantage of Searle s theory. This would be inappropriate for two reasons. First, Searle wants to see an essential difference between mental and social facts, and so the fact that his theory fails to reflect what he thinks this difference is must surely be seen as a shortcoming of the theory. Second, the message of the communalists is that the structure of certain facts is, contrary to pretheoretical intuitions, irreducibly social rather than individualistic. Hence a satisfactory classificatory system of different kinds of facts should reveal what it is in the structure of social facts that sets them apart from individualistic facts, and then the communalist could argue why she thinks we should see the facts under scrutiny as social facts, rather than as individualistic mental facts. Searle s framework is too conceptually deprived to function as such a tool for conceptual clarification. Within the limits of this paper I of course cannot take a stand regarding the problems of rule-following, communalism etc., but I wanted to mention them briefly in order to draw attention to the third major shortcoming of Searle s discussion on the objectivity of (social) facts (in Section 6 these problems will also return in a more substantial manner). In short, Searle s analysis we cannot be completely satisfied with Searle s account. The other example is problematic too. No doubt pain is independent of attitudes, but it seems to be dependent on feelings, which was earlier (Searle 1995, 8) said to suffice for epistemic subjectivity.

299 On the Objectivity of Social Facts 299 of the objectivity of different types of facts fails to penetrate the facts themselves and to explain which features of social facts actually make the sentences describing the facts epistemically objective despite the ontological subjectivity of the entities involved. It is Searle s concentration on the uses of the concept of objectivity at the expense of analysing the structure of the distinctive features of different facts that makes Searle s account unsuitable as an analytic tool for shedding light on many disputes in the philosophy of the human sciences, such as the quarrel over the communalist and individualist theories of meaning and mental content. What is needed is an account of the objectivity of social facts that can explain why the judgements about both individualistic mental facts and social facts are epistemically objective despite the ontological subjectivity of the facts, while also explicating precisely what differentiates social facts from individualistic mental facts (if they indeed are different). This is something Searle s framework cannot offer. 4. Towards a Durkheimian Account of the Objectivity of Social Facts Let us return to Definition (1) that captured the common-sense notion of objectivity: (1) A fact is objective iff it obtains independently of us. We saw that this definition must be modified, since as it stands it renders social facts obviously subjective and thus fails to capture the aspect of social facts that Searle called epistemic objectivity. Searle s account, which in the end turned out to be seriously restricted (even though not strictly speaking incorrect or false), was based on the insight that perhaps there are different senses of objectivity. Searle aimed to capture these different senses by pointing out that we can understand independence in Definition (1) either as ontological independence or as epistemic independence. Thus, the Searlean picture replaces Definition (1) with the following definitions. (2) A fact is ontologically objective iff its obtainment is independent of us (if it could obtain independently of humans). (3) A statement is epistemically objective iff the truth-value of the statement is independent of the attitudes etc. of the maker of the state-

300 300 Antti Saaristo ment (if a representation of the fact is made true or false by something independent of the attitudes of the person constructing the representation, be the fact represented ontologically objective or subjective). A fundamental reason why Searle s account led to the (interconnected) problems discussed in 3.1, 3.2 and 3.3 is that (2) does not actually add anything to (1), and (3) does not talk primarily about facts anymore, but about sentences. Hence, we need new candidates to replace (1), (2) and (3) that are about features of facts. At this point it is time to turn to the locus classicus of the philosophy of social facts, i.e., to Emile Durkheim s The Rules of Sociological Method (Durkheim 1895/1982). In short, Durkheim seeks to capture the objectivity of social facts with his notorious maxim that we must treat social facts as things. Naturally, this astonishing principle leads us to ask what does it mean to say that social facts are to be treated as things. As Durkheim explains, the claim is not that social facts are material things, but that they are things just as are material things (Durkheim 1895/ 1982, 35). For Durkheim, things include all that which the mind cannot understand without going outside itself (Durkheim 1895/1982, 36). In other words, a description or representation of a thing must satisfy Definition (3). Hence, in Durkheim s picture the defining feature of a thing is that the descriptions of it are epistemically objective in Searle s sense. Consequently, where Searle s notion of epistemic objectivity captures a feature of neither entities nor facts, but of sentences and representations, also for Durkheim [t]o treat facts of a certain order as things is therefore not to place them in this or that category of reality; it is to observe towards them a certain attitude of mind (Durkheim 1895/1982, 36). However, precisely because the maxim to treat social facts as things does not go further than the Searlean definition (3) and hence points out only what the representations of social facts have in common with the representations of many other facts, being thus completely silent about the distinctive features of social facts, Durkheim cannot leave his discussion on the objectivity of social facts here. But it is not clear how Durkheim could proceed coherently. Durkheim s dilemma is that on the one hand he thinks that in the name of naturalism we must accept that society comprises only individuals (Durkheim 1895/1982: 39), but on the other he is also committed to the view that the study of social facts is an autonomous discipline, the objects of which are not studied by non-social human sciences or natural sciences. For Durkheim, the characterising feature of social facts is that regardless of which

301 On the Objectivity of Social Facts 301 individual perspective they are observed from, social facts are necessarily treated or observed as things, whereas, e.g., individualistic psychological facts are treated as things only when observed from the third-person perspective (see Sections 3.1 and 5). In Durkheim s view the crucial question for the possibility of sui generis social science, then, is this: What is it in the nature of social facts that makes them things, i.e., epistemically objective, in this fullblown sense and thus to differ qualitatively from their constitutive parts, even when in the strict ontological sense society consists of nothing but individuals? As we know, Durkheim s answer is that the epistemically objective status of social facts is due to the fundamental structure of social facts: What constitutes social facts are the beliefs, tendencies and practices of the group [of individuals] taken collectively (Durkheim 1895/1982, 54, emphasis added). This suggests that in order to do justice to Searle s line of reasoning about ontological and epistemic objectivity so that we also manage to capture the distinctive features of social facts, we should concentrate on the notion of us in Definition (1), rather than on the notion of independence as Searle does. We can understand us here in two quite different senses. First, us might be taken to mean something like humans in general or, in some cases, the members of the social group in question. 6 However, us may also be taken to mean any particular individual. Consequently, the Durkheimian replacement of Definition (1) would explicate two different senses of objectivity with the following pair of definitions. (4) A fact is objective 1 iff it obtains independently of human attitudes and actions. (5) A fact is objective 2 iff it obtains independently of the attitudes and actions of any particular individual. Now Definition (4) captures the common-sense intuition of objectivity as expressed by Definitions (1) and (2), and hence also Searle s notion of ontological objectivity. Obviously social facts are not objective 1 in the sense of 6 From the present perspective of explicating the structure of social facts the difference between humans in general and the members of the social group in question is not crucial. What matters is that both formulations bring in intersubjectivity. However, for some other fundamental problems of the philosophy of social science the difference can be crucial (e.g., when tackling the challenge of cultural relativism) and should even if I cannot do it in this paper accordingly be explicated in the definitions of objectivity. I thank Steven Lukes for highlighting the importance of this point to me.

302 302 Antti Saaristo Definition (4). However, what we required was a notion of objectivity that captures also the intuition about the epistemic objectivity of social facts, as highlighted by Searle s example of money and as aspired after by Definition (3). Objectivity 2 in the sense of Definition (5) can offer that and, moreover, it points out why social facts are epistemically objective in the sense of Definition (3), and does it in a way that sets social facts clearly apart from facts about individuals that are objective neither in the sense of Definition (4) nor in the sense of (5) a feature emphasised both by Searle and Durkheim. Furthermore, Definition (5) can provide all this without confusing social facts with brute natural facts that are objective in the sense of Definition (4) (and, hence, also in the sense of (1) and (2)), which, obviously, is another requirement an acceptable view must meet. It could be argued that it is to some extent misleading to call the account of social facts defended in this paper Durkheimian, since Durkheim s view, as Steven Lukes (1982, 7-8) emphasises, in essence includes metaphysically highly dubious notions such as irreducible macro-level causal forces, and I certainly do not wish to commit myself to the existence of such forces. 7 However, I believe that the importance I lay upon Definition (5) justifies the label Durkheimian. In the following section I will nonetheless leave Durkheim on the background and explicate the distinctive nature of social facts in terms of Uskali Mäki s (1996) discussion of the modes of existence of economic entities. All the same, it should be clear that Section 5 is basically a clarification of the Durkheimian lines of reasoning of this section. 5. Explicating the Objectivity of Social Facts This section presents a classification of different kinds of facts, basing on the internal structure of the facts themselves. I have largely adopted the termi- 7 One might wonder why I do not mention Durkheim s notorious notion of a collective mind as the example of Durkheim s ontologically unacceptable notions. The reason for this is that arguably that notion is of great importance, and the naturalisation of it is indeed the key for understanding social ontology (see Section 6.3 below). Of course also Durkheim s macro-causation may turn out to be indispensable in social scientific explanations, but then, as Lukes (1982, 16) insists, we need a naturalistic micro-theory to back up the macro-explanations (perhaps along the lines of Jackson and Pettit s (e.g., 1992) theory of programme explanations). Be that as it may, in the name of naturalism the relevance of macro-level causal relations must be tied to the pragmatics of social scientific explanations, whereas a naturalised notion of a collective mind may play a more substantial role.

303 On the Objectivity of Social Facts 303 nology of Uskali Mäki (1996), whose line of thought my presentation for the most part agrees with. However, the aim of Mäki s discussion is to examine the prospects of scientific realism in economics, and he states his definitions in terms of the modes of existence of physical, economic and psychological objects. The present discussion differs from this in two ways. First, I generalise the account to apply to social sciences and social reality in general and, second, I continue to talk about facts rather than objects. Hence, in what follows the definition of independent facts is based on Mäki s (1996, 432) definition of independent existence, the definition of external facts on Mäki s (1996, 432) external existence and the definition of objective facts on Mäki s (1996, 433) objective existence. I should also add that Mäki is not committed to the idea that the account would be in any sense Durkheimian. Let us call facts that are objective in the strong sense of Definitions (1), (2) and (4) independent facts. A fact is an independent fact (it obtains independently) iff it obtains independently of the human mind. 8 Brute natural facts are independent facts, whereas social (and psychological) facts are not. We can define the ontological version of methodological individualism in the philosophy of social science in terms of independent facts as follows. (OI) Social facts do not obtain independently. On the basis of Section 4 above it is easy to see that Durkheim, who is generally seen as the ultimate methodological holist, nonetheless accepts (OI). Of course the acceptance of (OI) leaves room for many anti-reductionist forms of methodological holism, since (OI) simply states that social facts are neither based on any non-material Cartesian substance nor are they parts of the natural world that could survive independently of human activity. Within the limits of this paper I, of course, cannot discuss the problem of methodological holism (see, however, Section 6 below), but since the fact that even Durkheim accepts (OI) should soothe some of the metaphysical worries philosophers tend to have concerning methodological holism I think this small point is well worth making. 8 The fact that independence appears both in the definiens and in the definiendum should not confuse us here. After all, this is not a semantic analysis of the concept, but rather an explication of independently of what a fact must obtain if the fact is said to obtain independently. This remark applies mutatis mutandis also to the definitions of external and objective facts below.

304 304 Antti Saaristo Let us call the category of facts that are epistemically objective in the sense of satisfying Definition (3), but nonetheless intersubjective 9 in the sense of Definition (5), i.e., of excluding individualistic facts, the category of external facts. A fact is an external fact (it obtains externally) iff it obtains externally to or independently of any individual human mind. As a stronger notion independence clearly implies externality, and thus independent brute natural facts are also external facts. However, also social facts in Durkheim s sense, i.e., facts that depend on human attitudes taken collectively, can be external facts. It is exactly the externality of social facts that explains their ontological subjectivity and epistemic objectivity. The pieces of paper in my pocket are money because they are collectively accepted as money. Hence their status as money is not an independent fact. Despite its non-independence, however, the monetary status of the pieces is quite immune to changes in my attitudes; the fact that the pieces are money is a fact external to and independent of me (or of any particular individual for that matter). However, it is possible that when we are analysing social facts we sometimes must take into account also facts that are neither independent nor external. For example, there might be a society in which only those pieces of paper are money that are pronounced to be money by one particular individual (a bit like English bank notes that include the signature of the chief cashier of the Bank of England). The acceptance of a single individual is of course dependent on that individual, and hence it is not an external fact. So would the involvement of individual facts like this destroy the objectivity of social facts? Not really, since also in this case the fundamental fact that renders the non-external fact socially significant is itself an external fact, since the non-external fact is socially important only in so far as people collectively accept its role (perhaps unwillingly or tacitly by just going along with a perceived custom). It is the external fact of collective acceptance that makes the non-external fact socially significant. The social power inherent in the acceptance of the single individual comes from the underlying external fact, and hence even in this case the fact that certain pieces of paper are money is still ultimately an external fact. Even in a society where only those pieces of paper are money that the king declares to be money the king s declaration can as- 9 This of course suggests that Searle s solution to call social facts ontologically subjective can be misleading in some contexts, since the essence of social facts seems to be precisely their intersubjectivity as opposed to subjectivity.

305 On the Objectivity of Social Facts 305 sign the status of money to the pieces of paper only in so far as the members of the society accept (perhaps tacitly) this procedure as the correct way of introducing new tokens of money. Social status requires collective acceptance. However, even if we grant that in order to have social significance a nonexternal fact presupposes an underlying external fact, it remains true that there are situations in which social explanations must make use of non-external facts such as the acceptance of a particular individual. Hence, to complete 10 the framework in which the objectivity of different kinds of facts can be discussed, let us also examine in what sense non-external facts can be objective. The key insight here, I think, is Dennett s (e.g., 1991, 71) insistence that the perspective of scientific objectivity and in Dennett s view the perspective of all science is the third-person point of view. Dennett thinks that if the human sciences are to be mature sciences they must be able to represent individual facts from the objective third-person point of view. Let us follow Dennett and call facts that are capable for being so represented objective facts. A fact is an objective fact (it obtains objectively) relative to a given representation iff it obtains unconstituted by that particular representation. Once again, Durkheim is in full agreement: The facts of individual psychology themselves [ ] must be considered in this light [i.e., as objective facts]. Indeed, although by definition they are internal to ourselves, the consciousness that we have them reveals to us neither their inmost character nor their origin. [ ] This is precisely why during this century an objective psychology has been founded whose fundamental rule is to study mental facts from the outside, namely as things. (Durkheim 1895/1982, ) In other words, whereas material and social facts are essentially objective in the manner described above, facts of individual psychology are objective only in so far as we represent them from the third-person perspective (to use Dennett s terminology), i.e., in so far as we treat them as things (in Durkheim s terminology). As the final point of this section I should add that the definition of external facts given above is not perfectly adequate, since strictly speaking social facts as external facts are not completely independent of any particular individual. External facts depend on individuals qua group members, but the impact of each individual alone is quite negligible if the impact were not negligible, 10 Actually, this completes only the account inspired by Mäki s argumentation the next section introduces a one more category of facts.

306 306 Antti Saaristo then the fact in question would not be an external fact, but, by definition, (presentable as) a mere objective fact. 6. Are Social Facts Sui Generis? The last remaining clarification that needs to be done before the framework for understanding the objectivity of social facts can be regarded as complete is to explicate the different ways of understanding the status of external social facts otherwise we have not moved beyond mystical emergentism, which holds that social facts are emergent features of groups of individuals. I call the different accounts to be discussed below the Rational Choice account, the Wittgensteinian account and the Durkheimian account. The explications of this section pin down the core of the problem of methodological individualism versus methodological holism (recall that all sensible holists accept the ontological thesis (OI)) The Rational Choice Account The first possibility of understanding the nature of social facts as external facts is to say that they do not form a sui generis category of facts in any substantial sense. Social facts as external facts are seen essentially as aggregates of individual psychological facts. In the present context paradigmatic representatives of this approach are D. H. Mellor (1982) and Peter Abell (2000). 11 For Mellor (1982, 69) there is no more to the sociology of groups than the psychological states which make people co-operate. Similarly, Abell defines the Rational Choice approach to social theory so that rational choice theory invites us to understand individual actors [ ] as acting, or more likely interacting, in a manner such that they can be deemed to be doing the best they can for themselves, given their objectives, resources, and circumstances, as they see them (Abell 2000, 223). External social facts, then, are significant only in the sense that when aiming to maximise one s utility function part of the circumstances a rational agent must take into account is the behaviour of others. Sociality reduces to strategic interaction. Abell argues that the Rational Choice account can be seen as an essentially 11 See also the classical formulations of methodological individualism by Hayek, Popper, Watkins and others in O Neill 1973.

307 On the Objectivity of Social Facts 307 Weberian position since for Weber action can be social only in the sense that it takes account of the behaviour of others and is thereby oriented in its course (Abell 2000, 223). According to this definition, there is no room for external facts in any irreducible sense; they are nothing over and above collections of facts about individuals, which a rational agent must take into account just as she must take into account facts (aggregate and otherwise) about the physical environment. For pragmatic reasons we may want to talk about social facts, but this talk does not capture anything substantial that would not be captured by talk about a number of individualistic facts. Hence the Rational Choice account is an individualistic account par excellence. Social facts as external facts are not sui generis. In addition to Rational Choice Theory, views that tend to subscribe to this understanding of the status of external social facts include theories that find their inspiration in modern evolutionary biology and psychology. According to this picture, to put it roughly, typically human forms of behaviour (including norms and social practices) are based on genetically encoded, evolved behavioural tendencies or mental modules, and the role environment (social and physical) plays is only to trigger these internal capacities into action. Methodological individualism is compromised here no more than in Rational Choice Theory; what really count in social explanations are objective facts about individuals. To sum up, an advocate of this reading of external facts would hold that there are only physical (independent) facts and individualistic psychological (objective) facts. In addition to these, we may for pragmatic reasons wish to talk about aggregates of objective psychological facts as external social facts. However, just like a collection of physical facts does not form a new basic category in comparison to individual physical facts, a social fact as an aggregate of facts about individuals does not belong to any category substantially different from psychological facts. By the same token, though, this view implies eliminativism about social facts no more than modern physics implies that macroscopic physical entities do not exist. According to this view, social facts as external facts are real and consist of systems of interlocking beliefs and intentions of individuals.

308 308 Antti Saaristo 6.2. The Wittgensteinian 12 Account The philosophers who think that participation in social practices is what makes human form of life, including individual intentional agency, possible, take a substantial step away from the methodological individualism of the Rational Choice approach. 13 For them social practices are conceptually prior to individual psychological facts, and hence external social facts are sui generis and clearly apart from mere individualistic facts. For the Wittgensteinians external facts are substantial facts rather than mere aggregates of psychological facts constructed artificially for the practical purposes of explanations. Indeed, they think that the very substance of social practices is what makes meaningful individual thoughts and actions (including language) possible. The core idea here is the argument made famous by Kripke (1982). In Kripke s view only social practices can give rise to the normativity inherent in meaning and, hence, in intentional agency (including mental contents). If agents simply followed their internal voice be it the voice of reason as in Rational Choice Theory or the echoes of the human evolutionary history as in socio-biology there could not be a clear (indeed: objective!) distinction between when an agent thinks she has got something (e.g., the meaning of a term) right and when she really is correct. Of course even adherents of the Rational Choice account admit this in the case of familiar social institutions such as money (epistemic objectivity), but the radical thesis of the Wittgensteinians is that the same applies to contentful thoughts, meanings and intentionality in general. The idea is that just as a person s use of the pieces of paper in her pocket as money is made possible only by participating in social practices where the pieces of paper are accepted as media of exchange (and thus as money), also to be able to intend and refer with one s thoughts and sentences one has to participate in social practices that constitute the meaning of the signs one uses in one s language and thought (see, e.g., Esfeld 1999 and Williams 1999). Just as the pieces of paper cannot be money in virtue of their independent physical properties, also a sign (linguistic or mental) cannot represent or refer to anything (i.e., be intentional, about something) in virtue of its indepen- 12 I have labelled this account Wittgensteinian, since most adherents of this position locate its foundations in Wittgenstein s later philosophy (especially Wittgenstein 1953). As with Durkheim, I remain uncommitted regarding exegetical issues. 13 Examples of Wittgensteinian philosophers in this sense and whose discussions proceed in terms relevant to the present paper include Barnes (1983), Bloor (1996, 1997), Kusch (1999), McDowell (1998), Pettit (1993) and Williams (1999).

309 On the Objectivity of Social Facts 309 dent properties (see, e.g., Haugeland 1990). Something needs to be added. The Wittgensteinians think that the analogy between money and meaning does not stop here. Indeed, just as my individual psychological acceptance cannot be enough for the pieces of paper in my pocket to be money, likewise an individualistic solution cannot suffice for the normativity required for intentionality and objective meanings (Haugeland 1990, 414). What justifies my claims to be in the possession of money or to understanding (or even to being an intentional agent) is the successful participation in social practices and language games. In this manner, the Wittgensteinians think, thoughts and language really are meaningful (intentional, about something) only within social practices. This has been seen as the very feature that distinguishes methodological holists from individualists: For the individualist, concepts are individually held, mental entities. But for the non-individualist, these are irreducibly social. Even if it is accepted by the individualist that beliefs often involve social concepts, he cannot dispel the view that they exist only as internal states of individuals. The non-individualist contends that, whether or not they are attitudes of individuals, these have a constitutive relation with intrinsically social practices. (Bhargava 1992, 12.) Hence, the Wittgensteinians think that the classification of facts as represented in Section 5 is essentially correct, but when the focus of attention is on the human sciences, the crucial category is that of external facts, as opposed to the concentration on psychological facts, which is essential to the Rational Choice approach. It is the obtainment of external facts that makes individualistic psychological facts contentful mental states possible. According to the Wittgensteinian picture, the conceptual order is not captured by the received view that independent natural facts give rise to objective psychological facts, which in turn generate external social facts. Rather, independent natural facts generate external social practices (such as language) that form the bedrock of the human form of life, and it is only in virtue of these practices that objective psychological facts are possible The Durkheimian Account Another way of moving away from the methodological individualism of the Rational Choice approach without rejecting the ontological thesis (OI) is to

310 310 Antti Saaristo concentrate on collective agency. The idea is that truly collective action, including collective acceptance that constitutes social facts, is based on a mode of agency different from individual agency. Durkheim held notoriously that social facts are constituted by collective consciousness, and hence I have labelled the views subscribing to this understanding of external social facts as Durkheimian accounts. This view is obviously based on an explicit rejection of the aggregative view advocated by the Rational Choice approaches. As Durkheim puts it (to use emotions as an example), [a]n outburst of collective emotion in gathering does not merely express the sum total of what individual feelings share in common, but is something of a very different order (Durkheim 1895/1982, 56, emphasis added). This is of course the very line of reasoning in Durkheim s thought that has caused a lot of agitation among modern social theorists. Lukes captures this problem well: When writing of social facts as external to indviduals he [Durkheim] usually meant external to any given individual, but often suggested (especially to critical readers) that he meant external to all individuals in a given society or group : hence, the often repeated charge against him that he hypostasised or reified society, a charge which is by no means unfounded (Lukes 1982, 4). However, I have already argued that Durkheim accepts ontological individualism in the sense of (OI). Hence the only consistent way of understanding the strong externality Lukes talks about is the collective consciousness view sketched above, where the collective consciousness does not belong to a new holistic entity, but rather consists of a psychology different in kind from the psychology associated with individual action. In other words, Lukes notion of external to all individuals in a given society must be seen as the claim that, pace the Rational Choice approach, social facts are not aggregates of individual-mode psychological facts. Of course, a modern theorist cannot be content with vague appeals to collective consciousnesses, and hence the challenge for modern Durkheimians is, as Lukes (1982) constantly emphasises, to provide a naturalistic explication of the micro-foundations of Durkheim s concepts. This agenda is explicitly accepted by the social psychologists that subscribe to the so-called social identity approach (e.g., Hogg and Abrams 1988). In particular those social identity theorists who study crowd behaviour (e.g., Drury and Reicher 2000) could accept the above quote from Durkheim as it stands. However, in this paper I am interested in different ways of under-

311 On the Objectivity of Social Facts 311 standing the objectivity of social facts. Consequently, instead of discussing the attempts of the social identity theorists to empirically justify the Durkheimian account, I concentrate on explicating the view concerning the status of external facts the account implies. The Durkheimian picture can be seen either as a further development of the Wittgensteinian view, or as an alternative to it. Philosophers who see it as a development of the Wittgensteinian position, include Barnes (2000) and, to some extent, Bloor (1996), Kusch (1999), and possibly Tuomela (especially 2002). 14 The motivation for their Durkheimian position is that they want to resist the anti-naturalist conclusion of some Wittgensteinians (notably McDowell 1998, perhaps also Wittgenstein himself), according to which social practices that form the bedrock of the human form of life are so fundamental that we must accept them as primitive notions and refuse to analyse them further. As the slogan goes, we should not even try to dig below the bedrock. There are two problems with McDowell s quietism. First, there is the thoroughgoing naturalism of Barnes, Bloor and Kusch. They think that we should aim at a unified scientific worldview, where the emergence of social practices is not seen as an unanalysable brute fact but as a natural process, the conditions of which can be explicated. Second, without a Durkheimian notion of collective agency that is essentially different from individual agency, the whole picture is at risk of being viciously circular. On the one hand, the acceptance of naturalism in the sense of (OI) appears to force Barnes, Bloor and Kusch to say that social facts (including social practices) are based on individual agency. On the other, their Wittgensteinianism drives them to hold that individual agency is based on social practices. This sounds undeniably circular, 15 for they need a notion of pre-intentional collective agency that 14 A parallel view in the philosophy of mind is defended by, for example, Esfeld (1999) and Haugeland (1990). Haugeland calls this position, in which social practices are the bottom line, neo-pragmatism, conformism, left-wing (i.e., socialist) phenomenology, or simply third base (Haugeland 1990, 404). Haugeland does not mention Durkheim, but he seems (my utter ignorance about baseball prevents me for saying this for sure; see Haugeland 1990, 403) to share the view that the position is a further development of a Wittgensteinian conception. Be that as it may, I believe that the interpretation of Barnes and Kusch given below is compatible with the view defended by Haugeland. For an attempt to combine these parallel themes in social ontology and in the philosophy of mind, see Kusch s (1999, Part II) insightful discussion. 15 It seems to me that this kind of circularity is indeed a problem for Kusch 1997, whereas Kusch 1999 can be seen as the disentanglement of the problem.

312 312 Antti Saaristo makes stable social practices possible, which in turn provide the preconditions of intentional individual agency. Kusch (1999) does not have an explicit concept of collective agency, but he comes close to the Wittgensteinian collective agency view when he argues that for social practices to be possible, individuals participating in them must be equipped with an imitation device, a sanctioning device and an adjustment device (Kusch 1999, , see also Haugeland 1990, 404). Such agents are fundamentally social, since they aim at acting in the same way as other agents (imitation), sanctioning those who break ranks and, moreover, seek to adjust themselves in accordance with the feedback they receive. All this is supposed to be pre-linguistic and non-conscious. Behaviour shaped by such devices (or dispositions) is obviously very different from the standard understanding of individual action in the sense of rational choice. Hence I think that Kusch s account of the necessary conditions of social practices (which, in turn, make individual agency possible) amounts to implicitly assuming collective agency. Barnes (2000) has a discussion on the preconditions of social practices very similar to that of Haugeland and Kusch. Also Barnes thinks that social practices presuppose imitation, sanctioning and adjustment much as Haugeland and Kusch describe. The main difference seems to be simply that Barnes explicitly concludes that [i]t is necessary to postulate that non-independent individuals do this work (Barnes 2000, 54), and this non-independence is best conceptualised in terms of collective agency (Barnes 2000, 56). Barnes emphasises in particular that, in contrast to the Rational Choice accounts, his notion of collective agency of non-independent individuals amounts to the view that the collective agency required for social practices is not representable as mere strategic interaction. However, some Durkheimians do not accept the Wittgensteinian version of anti-individualism. For these philosophers intentionality is a primitive notion and not something that is made possible by social practices. Gilbert s (1989) theory of social facts and collective intentionality is explicitly both Durkheimian and anti-wittgensteinian and, as Bloor (1996, see also Haugeland 1990) convincingly demonstrates, so is Searle s (1995) theory, although Searle does not state it in these terms. Tuomela (especially his 2002) must be placed somewhere in between, since he argues that the Wittgensteinian view in the sense of Section 6.2 above is a contingent truth about the actual world, but not a conceptually necessary condition for all possible meaningful action (including contentful thinking), as the mentioned Wittgensteinians (with the exception of Pettit, who shares Tuomela s view) seem to think.

313 On the Objectivity of Social Facts 313 Despite the fundamental disagreement concerning the ultimate sources of intentionality (social or innate), in aspects relevant to the present paper the non-wittgensteinian version is very similar to the Wittgensteinian version of Durkheimianism. The connecting idea is that the construction and maintenance of sufficiently stable social practices presuppose a stronger conception of togetherness than the aggregative notion of the Rational Choice account. The advocates of the Durkheimian view have aimed to justify their position, for example, by appealing to our intuitions on what it is like to act together (Gilbert 1989) or to the view that sees social practices essentially as collectively rational solutions to collective action dilemmas (such as the Prisoners Dilemma) that cannot be so resolved by individual-mode intentional action (Tuomela 2000, 2002). The central idea here is that collective action and social facts are based on collective we-mode intentionality so that when acting together, the intentions of individuals are (irreducibly) of the form we will do X, and the agents derive their individual-mode intentions (their individual roles in the collective action) of the form in order for us to do X, I will have to do Y, hence I do Y from the previously formed we-mode intention (Searle 1995). Or as Tuomela (2002, 26) puts it, an agent x has a reason-based (weak) we-intention p (where p is a collective action type) iff (i) x intends p (ii) x believes that everybody in the relevant collection of agents intends p (iii) x believes that there is a mutual belief in the collection that everybody intends p (iv) (i) at least partly because of (ii) and (iii). According to this account, agents form a social group to the extent that they share a we-intention (Gilbert 1989, Tuomela 1995). 16 The clause (iv) sets the account apart from rational choice models by acknowledging purely social reasons (or causes) for action. In the case of our most fundamental practices 16 Gilbert (1989) thinks that this constitutes one more reason why we cannot accept the Rational Choice approach, since the adherents of it tend to say that the aggregate that counts is the sum of individuals in the given social group and hence they end up presupposing some social facts their approach cannot account for. In Gilbert s view the Rational Choice approach either falls back to the ontological anti-individualism in the sense of rejecting (OI), or fails to address the philosophical questions about the ultimate nature of social facts.

314 314 Antti Saaristo (such as those constituting language), the Wittgensteinian version of the collective agency view can read the notions of intending and believing in Tuomela s analysis in the same non-normative, purely functional sense in which we attribute intentions and beliefs to non-linguistic animals, whereas the non-wittgensteinians read them as full-blown propositional attitudes. Similarly, for the Wittgensteinians the connection expressed by (iv) is a mere causal connection instantiated in the manner described by Kusch s three devices and Barnes collective agency (constituting thus a normative practice), whereas for the non-wittgensteinians it is also an inherently normative reason-relation. Fortunately for the present purpose of explicating the different senses of the objectivity of social facts this dispute does not have to be solved, since despite their differences, the Durkheimian views are also in agreement regarding certain important issues. In particular, all the Durkheimians are committed to a fourth category of facts. This is the category of full-blown social facts and it can be defined as follows: A fact is a social fact iff it is constituted by collective acceptance based on collective agency (or collective intentionality). All social facts are external facts, but all external facts are not social facts (in particular, the external facts captured by the Rational Choice approach cannot be social facts in this sense). Social facts are also quite objective from any individual perspective, since individual perspectives do not even participate in their construction and reproduction, for the facts are based on collective agency. Where the Wittgensteinian and anti-wittgensteinian versions of the present Durkheimian account differ from each other is in the way they understand the conceptual order of facts of different kind. The Wittgensteinians think that there are natural (i) independent facts that give rise to non-intentional collective agency in the sense of agents equipped with social devices (Kusch) or dispositions (Haugeland). Collective agency, in turn, constitutes (ii) social facts (essentially, social practices). According to the Wittgensteinians, all this can be explained within the framework of evolutionary biology and other causal explanations of the natural sciences. The social practices, then, constitute by their existence the normative framework that makes psychological (iii) objective facts possible, opening thus the door for normative human sciences that aim to understand human action. The psychological facts in place, we can finally have (iv) external facts in the sense the Rational Choice approach sees them, i.e., as aggregates of objective interlocking psychological facts.

315 On the Objectivity of Social Facts 315 The anti-wittgensteinian view does not imply a similar distinction between the natural and the human sciences, or between explanation and understanding. However, the problem also the anti-wittgensteinian Durkheimians see in most attempts for naturalistic explanations of the social world is the unquestioned commitment to individualism. The anti-wittgensteinians hold that the ontologically fundamental level is that of natural (i) independent facts. But these natural facts are then seen to have evolved to give rise to two modes of intentional agency that are irreducible to one another: we-mode agency and individual-mode agency. These modes are both biologically primitive features of individual agents, and as such (ii) objective facts. Action based on we-mode intentionality constitutes truly (iiia) social facts whereas individual-mode actions constitute (iiib) external facts in the sense of the Rational Choice approach. Note finally that by accepting the non-normative and non-conceptual (or causal-functional) reading of Tuomela s analysis of we-intentions (and collective acceptance) the Wittgensteinians can count as social facts also facts that are so naturalistic that they do not involve explicit construction but are nonetheless socially produced and maintained (Tuomela has mentioned a we-fear as an example of a social fact of this kind). In the anti-wittgensteinian understanding of social facts these kinds of facts come out as nonsocial facts, which may be seen as a shortcoming of the account. References Abell, Peter Sociological Theory and Rational Choice Theory. In The Blackwell Companion to Social Theory, edited by B. S. Turner, Oxford: Blackwell. Barnes, B Social Life as a Bootstrapped Induction. Sociology 17 (4): Barnes, B Understanding Agency: Social Theory and Responsible Action. London: Sage Publications. Bhargava, R Individualism in Social Science: Forms and Limits of a Methodology. Oxford: Clarendon Press Bloor, D Idealism and the Sociology of Knowledge. Edinburgh Working Papers in Sociology 5: Bloor, D Wittgenstein, Rules and Institutions. London: Routledge. Dennett, D. C Consciousness Explained. London: Allen Lane. Drury, J., and S. Reicher Collective Action and Psychological Change: The Emergence of New Social Identities. British Journal of Social Psychology 39:

316 316 Antti Saaristo Durkheim, E. 1895/1982. The Rules of Sociological Method. In The Rules of Sociological Method and Selected Texts on Sociology and its Method, edited by S. Lukes and translated by W. D. Halls, London: Macmillan. Esfeld, M Rule-Following and the Ontology of the Mind. In Metaphysics in the Post-Metaphysical Age. Papers of the 22 nd International Wittgenstein Symposium (Contributions of the Austrian Ludwig Wittgenstein Society VII(1)), edited by U. Meixner and P. Simons, Kirchberg: Austrian Ludwig Wittgenstein Society. Gilbert, M On Social Facts. London: Routledge. Hacking, I John Searle s Building Blocks. History of the Human Sciences 10: Haugeland, J The Intentionality All-Stars. Philosophical Perspectives 4 (Action Theory and Philosophy of Mind): Hogg, M. A., and D. Abrams Social Identifications: A Social Psychology of Intergroup Relations and Group Processes. London: Routledge. Jackson, F., and P. Pettit Structural Explanation in Social Theory. In Reduction, Explanation, and Realism, edited by D. Charles and K. Lennon, Oxford: Clarendon Press. Kripke, S. A Wittgenstein on Rules and Private Language. Oxford: Basil Blackwell. Kusch, M The Sociophilosophy of Folk Psychology. Studies in History and Philosophy of Science 28: Kusch, M Psychological Knowledge: A Social History and Philosophy. London: Routledge. Lukes, S Introduction. In The Rules of Sociological Method and Selected Texts on Sociology and its Method, edited by S. Lukes, London: Macmillan. Mäki, U Scientific Realism and Some Peculiarities of Economics. In Realism and Anti-realism in the Philosophy of Science: Beijing International Conference 1992 (Boston Studies in the Philosophy of Science 169), edited by R. S. Cohen, R. Hilpinen and Qiu Renzong, Dordrecht: Kluwer. McDowell, J Mind, Value, and Reality. Cambridge, Mass. and London: Harvard University Press. Mellor, D. H The Reduction of Society. Philosophy 57: O Neill, J. (ed.) Modes of Individualism and Collectivism. London: Heinemann. Pettit, Philip The Common Mind: An Essay on Psychology, Society, and Politics. Oxford: Oxford University Press. Searle, J. R The Construction of Social Reality. London: Penguin. Tuomela, R The Importance of Us: A Philosophical Study of Basic Social Notions. Stanford: Stanford University Press. Tuomela, R Cooperation: A Philosophical Study. Dordrecht: Kluwer. Tuomela, R The Philosophy of Social Practices: A Collective Acceptance View. Cambridge: Cambridge University Press.

317 Explaining Practices 317 Explaining Practices Petri Ylikoski Abstract This paper discusses Stephen Turner s recent critique of theories of social practices. It shows that his arguments are valid against common explanatory uses of these concepts, but not against practices in general. There are plenty of legitimate non-explanatory uses for practice concepts. The paper also suggests that Turner s main arguments derive from two principles that have much wider application than practice theories. Consequently, they should be considered as general constraints on every social theory. 1 The Practice Turn Social practices have become popular in contemporary philosophy and social theory. (Schatzki et al. 2001, Tuomela 2002, Stern 2003) They promise to provide a way to overcome some old and entrenched dichotomies and conceptual problems. For example, old conceptual problems related to rules and their application, structure and agency, or the nature of cultural differences are expected to vanish when they are rethought in terms of practices. In contrast to older conceptions, practices seem to provide a more dynamical and flexible way to grasp social reality. Practices may be central, or even fundamental, concepts for social theory but their identity remains elusive. The cover of recent anthology about the practice turn (Schatzki et al. 2001) claims that the practice theory is said to stand opposed to prominent modes of thought such as individualism, intellectualism, structuralism, systems theory, and many strains of humanism and post-structuralism. However, a closer look at the papers show that there is not much common ground between different authors. The aims of their theories are quite different and sometimes even incompatible. The only shared element seems to be the belief in the theoretical importance of the concept of practice. One of the most elusive features of the practice theories is related to explanation. Practices are often adopted as a replacement for earlier explanatory concepts (like structure, mentality, interest, or norm) and practice theories

318 318 Petri Ylikoski are clearly used to answer explanation-seeking questions. However, the nature of these new explanations remains hard to pin down. There is surprisingly little discussion about the explanations provided by practice theories. Over the last decade Stephen P. Turner (1994, 2002, 2003) has tried to disturb this silence by arguing that there is something seriously wrong with the explanations provided in terms of practices. He provocatively claims that the concept of practices is deeply flawed (Turner 1994, 11). He suggests that the notion of practice should be given up and replaced with a less problematic notion of habit. If he is right, much of the 20 th century social theory is in deep trouble. In this paper, I will discuss Turner s arguments against practices and assess their scope. I will argue that the proper scope of his arguments is both broader and narrower than the theories of practices. The scope is broader because the problematic pattern of explanatory inference found by Turner is also found in the theories that practice theories are intended to replace. On the other hand, I argue that there are ways to use the concept of practice that escape Turner s arguments. I will also argue that it is possible to improve his arguments. 2 Turner s Practices For Turner, the meaning of the term practices is extremely broad. It names a whole family of concepts. Among the members of this family are concepts like paradigm, ideology, shared presupposition, conceptual framework, form of life, habitus, and tacit knowledge. This list looks very heterogeneous, but according to Turner these various concepts share an important feature that justifies their inclusion under the same general concept. The crucial common denominator is the commitment to some kind of non-public collective objects (hereafter NCOs) tacitly occupying the minds of the members of the relevant social group. These tacit common possessions are thought to be analogical to public things like articulated beliefs, attitudes and rules. Furthermore, they are assumed to have the same causal powers as their explicit counterparts. The role of NCOs in these theories is essentially explanatory, and this assumed explanatory power is something Turner wishes to challenge. Turner finds these collective non-public analogues to explicit things like rules, beliefs, commitments, or norms to be ontologically mysterious. The following extract from Michael Polanyi, who has influenced many practice theorists, shows clearly the problem:

319 Explaining Practices 319 I shall take as my clue for this investigation the well-known fact that the aim of a skilful performance is achieved by the observance of a set of rules which are not known as such to the person following them (Polanyi 1958, 49.) Polanyi appeals here to rules that operate unbeknownst to the person following them. Tacit knowledge is by its nature such that it cannot be fully transformed to explicit rules, but it is still assumed to have many of the features of its explicit counterpart. The analogy is strange, and it creates a host of conceptual problems for practices and their conceptual kin. If tacit knowledge is like explicit knowledge, why it cannot be transformed to one? Is rule-following without rules and relevant intentions to follow the rule a coherent idea at all? Furthermore, this analogical solution only seems to shift the explanatory problem to the next level: if explicit rules cannot account for the practical wisdom, is there any reason to suppose that implicit rules would do better? Some of these problems are specific to Polanyi s proposal, but Turner claims that all NCOs share a similar set of basic problems. If the analogy is taken strictly, the NCOs cannot solve the theoretical problem they are supposed to solve, and if the analogy is loosened, the NCOs just become mystical and their assumed explanatory powers illusion. However, Turner s principal arguments against NCOs in The Social Theory of Practices are not in fact based on their inherent ontological mysteriousness. The trouble lies rather with the ways in which they are inferred and assumptions about their transmission or reproduction. According to Turner, these problems are fatal to practice theories. His positive suggestion is that we give up the idea of sharing. Instead of shared NCOs we should have individual habits and various public processes of habituation. This should not be seen as a conceptual replacement but as an elimination of one of the central elements in most prominent practice theories. As Turner himself points out, habits are simply the part of the phenomenon described by the term practices that remains when the idea of people sharing some NCOs is eliminated (Turner 2002, 23). However, the consequences of the change are significant, since it completely rules out some popular patterns of explanation. For example, it inverts the usual explanation of tradition. One can no longer say that rituals are performed because people share a common framework. The rituals are now regarded as behavioral technologies that produce a certain (rather superficial) uniformity of individual habits (Turner 2002, 12). Turner suggests that practice concepts can only be used as instruments of understanding or some sort of sensitizing concepts. It might be helpful, for

320 320 Petri Ylikoski example in understanding people from other culture, to assume that they are holding certain tacit premises that are different from ours. However, the important thing is that these postulated entities are only our instruments, not real things. People might behave as if they hold those premises, but the fact is that we do not have any reason to assume that they really have them (Turner 1994, 37, 78-79). To the contrary, such a realist assumption leads to many conceptual and theoretical problems. As a consequence, Turner is suggesting an instrumentalist attitude towards these concepts. However, his reasons are not typical empiricist arguments against realism, but considerations that derive from a position that could be described as causal realism. In order to keep our realism consistent, we have to be instrumentalists about NCOs. Before going to specific arguments, I want to make a couple of comments about the scope of Turner s arguments. The first thing to notice is that the rhetoric of Turner (1994) gives an impression of being against social practices in general, but in fact he is only against some common explanatory patterns that are related to concepts like practices. This limits the attack significantly; these concepts are often used in non-explanatory manner. This point can be illustrated by considering Pierre Bourdieu s concept of habitus. The application of Turner s arguments is straightforward in the case of Bourdieu. For Bourdieu, habitus is an explanatory concept and in applying it he commits many of the fallacies Turner identifies. For example, the perfect reproduction of habitus from generation to generation seems utterly mystical. However, the concepts of habitus and practice have clearly different roles in Bourdieu s thinking. Habitus and its conceptual derivatives, like the class habitus, are theoretical concepts that are used to explain observed regularities of behavior and practices. (Bourdieu 1990, ) In these contexts, Bourdieu s concept of practice seems to be purely descriptive. The practices are the explananda, not the explanantia. These practices are observable and public things and nothing in Turner s arguments implies that they would be somehow illegitimate objects of explanation. It might be that Bourdieu s intention to explain the differences in practices in terms of differences in habitus leads him to describe these differences in a biased manner, but this does not make the differences non-existent. Consequently, there seems to be room for a theoretically innocent and non-ambitious use of the concept of practices. This example points to a general problem with Turner s choice of terminology: his arguments are really directed against certain widely used ways of explaining practices, not against practices themselves. The talk of the concept of practices is deeply flawed is highly misleading. It is important to

321 Explaining Practices 321 recognize this for two reasons. First, Turner identifies some general explanatory failures that can also be committed by authors who are not employing the concept of practices. In fact, the best examples of the failure can be found from the time before the recent practice turn. Explanations in terms of shared norms, values, mentalities, or worldviews face exactly the same problems. A good example is functionalist sociology with its internalized norms. The real problem is with theoretical explanations presupposing strange nonpublic collectively shared objects, not with practices per se. Second, some authors have turned to concepts like practices in order to escape the explanatory problems Turner identifies. There is a number of ways of using concept like practices that do not commit the explanatory fallacies identified by Turner. Naturally, his arguments do not apply to these cases. Furthermore, there is nothing in Turner s arguments that show that practices are somehow illegitimate objects of empirical analysis or explananda for social theories. To the contrary, they seem to underline the fact that they have not yet been explained properly. The final important observation is that Turner should not be interpreted as arguing for methodological individualism. His suggestion that we replace the concept of practice with the concept of habit has lead many to assume that the argument is one more version of the individualist case against holistic concepts. (For example, Barnes 2001.) This reading confuses two things. No matter how individualistic, or adequate, Turner s concept of habit is, his main arguments are independent of it. He is not claiming that explanatory uses of the concept of practices are mystical because they are nonindividualistic. The trouble lies elsewhere. Consequently, his arguments should be evaluated without assimilating them to the messy issue of individualism. 3 What Is Explained by Practice Theories? Let us take a look at explanations in terms of practices that Turner has in mind. A typical example is an explanation of a cultural difference between two groups. The difference can be in the manners, practices, values, beliefs, or habits of the groups. Quite often these differences become visible as difficulties in communication or understanding between the members of two groups. Of course, the differences do not have to be related to communication, any consistent observable difference will do. The crucial thing is that the explanandum is an observable difference between two groups. This dif-

322 322 Petri Ylikoski ference is often described using vocabulary that has already been influenced by some theoretical considerations. The next step is an inference to the best explanation that postulates some theoretical objects (NCOs) and assumes that the members of the group share the same objects in their minds. The people in the group A share x, and the people in the group B share y. The guiding idea is that the sharing does the crucial explanatory job: the differences between groups are accounted for by the fact that the group A possess x and the group B possess y. People in a given group believe or behave in a certain manner because they all share the same NCOs. This explanation seems to promise some economics of scale: there are probably many differences between the groups, but the well-chosen NCOs seem to explain them all in unified manner. For example, consider two social groups of scientists. The groups probably differ in their theoretical beliefs, experimental techniques, methodological preferences, and epistemic values. The explanation of all these differences separately does not seem very attractive option. The differences seem to be more easily accountable in terms of their different paradigms, traditions or conceptual schemes. One difference in terms of NCO seems to promise explain all, or at least most, of the observable differences. The same explanatory pattern seems to provide also another explanation: the explanation for the similarity within the group. If people behave or believe similarly within a group, maybe it is because they have a similar Weltanschauung, conceptual framework, habitus, or tradition. For example, the relative lack of difficulties in communication or understanding between the members of a group could be accounted for in this manner. The persistence or continuity of the ways of doing things within the group could be explained similarly. So we have two kinds of explanation: explanations of differences between groups and explanations of similarity within a group. Turner has three kinds of problems with these explanations. First, he has ontological problems with the nature of the postulated NCOs. They seem to be utterly mystical analogical objects: non-explicit premises, goals that are not consciously intended, rules or norms unknown to the agent, or implicit values that have causal powers to guide how people having them think or behave. These issues were discussed in the previous section, so we can proceed to two other problems that are more directly linked to explanation. It is important to keep in mind that these problems are independent of each other in the sense that even if the NCOs were shown to be legitimate theoretical objects in the social sciences, the two problems with explanation could still have a bite.

323 Explaining Practices 323 The second problem concerns the intended explanandum. Turner suspects that the way in which the explanandum is identified is somehow already biased towards the NCO doing the explanatory work. The relationship between the two might be too close and the explanandum an artefact produced by the theoretical concepts used in its description. The third problem concerns the idea of sharing the same NCO and is related to a more general constraint on any social theory. I will discuss the first worry in this section and the second in the next section. So, in which way the explanandum of a typical practice explanation could be a fake or a misapprehension? (Turner 2002, 162) To begin with, Turner points out that the objects of explanation in explanations appealing to practices are not usually well defined and that there is a temptation to adjust the explanandum to the available means of explanation. The questionable trick seems to be the following: concepts like practices are used to create unity within the phenomenon to be explained already in the phase of describing the explanandum. After this molding of the phenomenon to be described, the explanation in terms of one s favourite NCOs seems to be the only plausible explanation. This raises the suspicion that the explanandum is biased or a fake. If the differences to be explained were given a different description, the assumed virtues of the proposed explanation might just go away. However, the suspicious circle also goes to another direction: x and y are first inferred from the observed differences between A and B and then used to explain these same differences. Again, the explanans does not seem to be distinct from the explanandum, raising question whether this is an explanation at all. The circle seems to be too small to be virtuous. Turner also makes the claim that the identification of practice is dependent on a cultural perspective and he suggests that the differences between practices cannot be described naturalistically. He supports these claims by the observation that practices are identifiable only through a contrast and the identity of the practice seems to depend on the chosen contrast. (Turner 1994, ) For example, the identity of Finnish culture would be different when compared to Swedish and when it is compared to Russian culture. Depending on the expectations raised by the contrast, different characteristic features suggest themselves as defining features of the culture. If practices are observer-relative in this sense, then they are not fully objective. As Turner puts the matter himself: [ ] what is revealed about our practices is a matter of the contingencies of available comparisons, and that a practice as such, practices as an object, is inaccessible (Turner 1994, 37.)

324 324 Petri Ylikoski Turner suggests that this explains why the identity of a practice is so difficult to pin down and why the explanations in terms of practices always seem to lack something essential. They are not real causal objects. The arguments above look a bit too sketchy to raise confidence. I think Turner is after something important, but the reasons for the explanatory failure should be made clearer. The revised argument goes as follows. The explananda we have discussed thus far are explanations of differences between groups and similarities within a group. The problem is that these two do not constitute two separate explananda in the cases under consideration. To the contrary, the two explanations are intimately connected in two ways. On the one hand, the explanations referring to shared NCOs can explain differences between two groups only if they can also explain the similarities within the group. If explanations of similarity within the group fail, the explanations of differences between groups will fail too. I will call this the problem of explanatory dependence. On the other hand, the identity of the explanandum for the explanation of similarity is dependent on the contrast provided by the outside comparison. I will call this the problem of nonautonomous identification. The recognition of these two problems changes our perception of the practice explanations. Originally, they seemed to be primarily explanations of differences and the explanation of similarities was just a bonus. Now we can see that this is a highly misleading picture. It is true that the explanandum arises from the comparison of the two groups, but the explanation itself is a different matter. It should be able to handle the similarities before it can have any success in explaining the differences. It is the explanation of the differences that is the bonus. But this bonus is only available if the explanation of the similarities within the group works. To see the reasons for this failure, let us consider a principle that should be acceptable to everyone: The similarities within the group should be identifiable and explainable without any reference to outside comparisons. The principle should be obvious, since its denial would lead to absurd results. However, the explanations under our consideration do not respect it. The neat inference to the best explanation that starts from the differences between two groups leads to an explanatory setting where the principle is violated.

325 Explaining Practices 325 The problem of non-autonomous identification arises because the differences and similarities are second-order properties. Obviously nothing is different expect in contrast to something else. The problem is that the same can be true of the similarities we are trying to explain: some of the similarities within the group are only visible when the group is contrasted with something different. As long as we concentrate on practices within one group we can only see the differences between individual practices. However, when compared to another group these activities start to have some sort of unity: individual practices are more similar to individual practices within the group than to the individual practices within the other group. This is the basis for Turner s worry that the explanandum of practice explanations is somehow observer relative or non-objective. So the problem is not so much with the explanation of differences, as he seems to suggest, but with the fake similarities. If the similarities (within the group) to be explained cannot be identified without outside comparison, they are not real. And if they are not real, there is nothing to explain. Consequently, we do not have explanation for the differences between the groups either. The explanatory dependence between the two guarantees this. The above line of reasoning leads to the conclusion that practice theories should be able to explain first-order or primary properties of their explanandum phenomena. Since these properties are identified without the comparison to other groups, the explanatory factors inferred from the differences between two groups do not have any upper hand. In fact their origin may be a burden, as many of Turner s examples show. When explaining the primary properties of the practices, the real challenge is often to account for the stability and change. The explananda like these are difficult for concepts like form of life, tradition, conceptual scheme, paradigm, Weltanschauung, and culture. They cannot be used to explain the persistence, continuity or change of the practices they describe. The explanatory use always presupposes that the extension to new situations somehow takes care of itself in some routine manner. But the problem is that this is just an assumption: the extension in time and the changes do not proceed automatically. Other explanatory resources will be needed for this. The NCOs turn out to be excess theoretical baggage that brings conceptual problems without providing any help in real explanatory challenges. If the above arguments are valid, we have serious reasons to doubt the explanatory powers of explanations appealing to NCOs inferred from the differences between groups. These concepts are too shallow to be really explanatory. They are at most placeholders for real explanations. However, the

326 326 Petri Ylikoski above line of reasoning does not show that their non-explanatory uses are somehow illegitimate. It is also important to notice that the argument is general: its force is not limited to the recent practice theories. 4 A Cognitive Constraint for Social Theories Turner s second, and in my mind more crucial, argument is related to the problems in conceiving a mechanism for the transmission of NCOs from person to person. It is generally accepted that in the communication of the explicit counterparts of the NCOs there is a tendency to the transformation of the communicated message. In other words, the fidelity of the communication is not perfect. However, the theories building on shared NCOs seem to presuppose that they can be transmitted more or less intact. This is in fact imperative from the point of view of the explanation. People should have the same tacit things in their heads, since without the sameness of NCOs the intended explanation will not go through. This requirement puts high demands on the mechanisms of acquisition and Turner argues that this challenge cannot be met: [ ] no account of the acquisition of practices that makes sense causally supports the idea that the same internal thing, the same practice, is reproduced in another person (Turner 1994, 13.) His claim is basically that no plausible theory of learning or communication is able to support the demands of the practice theories. The supporters of these theories are forced to appeal metaphors like reproduction instead of real theories of acquisition. Of course, it is possible that there could be some still unknown or undiscovered psychological processes through which the practices could be transmitted. However, this seems highly unlikely. If the NCOs are anything like their non-tacit counterparts, the relative infidelity of the transmissions is to be expected. What would happen if we were to relax the requirements a bit and assume that the fidelity of the transmission is just an idealized description of the process? The theoretical edifice would be then accommodated to the realities of learning and communication. Instead of sharing the exactly the same thing, the people would have only similar NCOs in their heads. There would be some variation between the members of the group, but the members would still be more similar to each other than to the members of other groups. According to Turner, this is not possible: once we accept the virtual

327 Explaining Practices 327 necessity of variation, the whole model of sharing and reproduction of NCOs falls apart. All we have is the process of individual habituation: If the idea of preserving sameness is removed from the other cases of transmission, they collapse into the case of the acquisition of habits, habits which may vary in internal structure and produce performances that are externally the same, and habits which may produce performances that vary slightly but are sufficiently intelligible or sufficiently predictable for the purposes at hand (Turner 1994, 77.) With individual habits, however much they are disciplined, the whole explanatory project of building on shared NCOs collapses. The sameness of external performance is not a result of sameness of internal structure. Nothing in the process of acquisition of individual habits guarantees the sameness of the internal stuff. To the contrary, it is to be expected that there is variation in the ways in which the habits are both acquired and implemented. In this new landscape of individual habits, the idea of sharing tacit mental objects has no purpose, nor explanatory power. The similarities of external behavior and the possibility of successful communication are explained by other means: [ ] by performing in certain ways, people acquire habits which lead them to continue to perform, more or less, in the same ways. The observances, so to speak, cause individual habits, not some sort of collectively shared single habit called a practice or a way of life, which one may possess or fail to possess. If this is so, the collective or public facts about traditions or cultural systems of meaning begin and end with the observances or public objects themselves. Everything else is individual there is no collective tacit fact of the matter at all (Turner 1994: 100.) It is important to understand that this is not an empiricist argument to the effect that we cannot observe the invisible parts of the practices or prove their existence. The argument is not about the ontology of NCOs, but about the plausibility of the mechanisms of their transmission. The problem is that there is no plausible way of communicating the same NCO from one person to another. The explicit counterparts of NCOs face the constant problem of transformation during the communication, and there is no reason to suspect that the NCOs wouldn t be similar in this respect. Otherwise they would be truly magical. It is also noteworthy that Turner does not demand that practice theories should contain a full-fledged psychological theory of learning. He is only

328 328 Petri Ylikoski asking the theories to be consistent with some plausible theory about transmission (Turner 2002, 7). Of course, it follows that if a theory of practices is not compatible with what is known about communication and learning, then the burden of proof is on the proponents of the theory. They should provide an outline of a plausible theory of transmission in order to get their theory taken seriously as an explanatory account. When the argument is expressed in these terms, it is easy to see that it is not limited to the practice theories. It is based on a general constraint on any social theory. The requirement that a theory should be compatible with a plausible account of learning is justified. Furthermore, it is far from trivial. The same problem of mechanisms of transmission is also faced by other contemporary theories in the human sciences. While discussing the theories of cultural evolution based on the idea of meme, Scott Atran notes: The real mystery is how any group of people manages an effectice degree of common understanding given that transformation of ideas during transmission is the rule rather than exception (Atran 2002, 241.) Atran is here basically appealing to the same general constraint as Turner. He also seems to have the same basic message: social theorists should pay more attention to the theories of cognition. (See Turner 2002, 2003.) It is debatable how deep into cognitive sciences the social scientists should immerse and how fruitful it would be, but at least their general requirement of compatibility should be observed by all social theorists. 5 Forms of Life and Interests I have argued above that there are legitimate, but non-explanatory, ways to use practice concepts. As an example of this, let us now consider in more detail Wittgenstein s notion of form of life (Lebensformen) as it is used in the sociology of scientific knowledge (SSK). The sociologists of scientific knowledge, like Barry Barnes and David Bloor, have given a central role to this concept in their theoretical work (Bloor 1983; Barnes, Bloor and Henry 1996; see also Collins 1985, 2001; Collins and Kusch 1998). Turner discusses this concept extensively (Turner 1994, 68-75) and it is interesting to see whether his arguments have any bite on the use of the concept in SSK as a number of authors have suggested (Lynch 1997, Pickering 1997, Stern 2003). I will not be interested in questioning whether the concept of form of life is used in SSK in the same sense as Wittgenstein used it. (See Friedman 1998.)

329 Explaining Practices 329 I am only interested in seeing whether the current use of this concept by these sociologists of scientific knowledge can escape Turner s criticism. My basic claim is that their use of this concept is innocent in a sense that they do not commit to the explanatory fallacies identified by Turner. If there are problems with their application of Wittgenstein s ideas, these problems must lie elsewhere. What saves the SSK from Turner s arguments is the finitist account of meaning. (For finitism, see Barnes, Bloor and Henry 1996, Bloor 1997.) Barnes, Bloor, and Henry explain: To take the scientist to be a participant in a form of life is to insist [ ] on the relationship as facilitative. The individual, by virtue of training, participated in a tradition, and has available what it provides. In making use of what is available, in whatever way, he or she contributes to the continuation of the tradition. The tradition does not make participants use it in this way or that; rather, in using it in this way or that they constitute the tradition. This conception of participation needs to be extended to all the elements of tradition, even the most apparently constraining ones: laws, rules, norms, meanings, and definitions. Participants are those who share the elements of the tradition, not those who deploy them in one way rather than other, still less those constrained or determined by them (Barnes, Bloor & Henry 1996, 116.) The important thing in this passage from our point of view is that the form of life (or tradition) is treated as a set of available resources that is a public, non-tacit, object. The members of the form of life share a number of things, but these things are safely public objects like training and exemplars. It is exactly the same thing Turner wanted to emphasize in his own account. Secondly, the form of life has no explanatory role, to the contrary, it is something to be explained. Again, this is in line with Turner s own theory. 1 1 This does not mean that Turner is in complete agreement with Barnes and Bloor. It is not clear whether Turner would accept the finitist account of meaning. Furthermore, there are some differences in terminology and conceptual focus. For example, Barnes has recently noted that Turner s arguments are not against truly social practices, but against shared individual practices. The real social practices are properties of groups, not their members. However, the examples Turner discusses are cases where the members of the group share some tacit individual objects. In this sense they are not social enough (Barnes 2001, ) Social practices in Barnes sense are activities that are needed to coordinate the enactment of individual practices, or habits if this terminology is preferred. It is clear that there is room for such things in Turner s conceptual apparatus, so he is not arguing for habit-individualism that completely eliminates the social from practices. Furthermore, both Barnes and Turner agree that practices in this sense are not proper explanatory factors, but things to be explained.

330 330 Petri Ylikoski One important consequence of the finitist account of meaning is that it robs meanings, rules, norms and the like of their ability to determine their future extension. Their future use is open-ended and dependent on other causal factors. No matter whether they are public or tacit, these things do not have explanatory power. The practices do not self-reproduce, and if they are reproduced it must be understood by reference to contingencies external to practices. (Barnes, 2001, 27.) Since the NCOs are usually assumed to be analogical to rules, norms or the like, the consistent finitist would not have any inclination to postulate such things. The inference to the best explanation cannot lead to things one considers non-explanatory. Consequently, the NCOs that Turner opposes lose much of their theoretical appeal and explanatory power. However, this does not mean that these concepts lose their function. They still have many important descriptive uses. The replacement with individual concepts like habit suggested by Turner would only mean impoverished vocabulary. The finitist need not change his terminology to avoid explanatory fallacies identified by Turner. For them practices and ways of doing things are always things to be explained not the proper explanatory factors. I want to emphasize that explanatory questions in the context of practices are fully legitimate. It will not do, as Pickering (1997, 329) seems to suggest, that one just describes how the practices unfold. Here the finitist account proposed by Barnes and Bloor is still the most promising alternative. It emphasizes the role of knowledgeable, informed and goal-directed action in the enactment of social practices. (Barnes, 2001, 21.) In order to answer how- and why-questions about the extension, continuation and changes of practices, concepts like aim, goal, and interest are needed. Contrary to the claims by the advocates of practices like Pickering (1995) and Rouse (2002), practices do not replace these concepts, but rather provide means to describe their objects of explanation. There is no practice turn away from these objects, if they are understood correctly. One only needs to remember that goals and interests are not unmoved movers that can explain everything as Pickering (1995, 54) suggests. The shortcomings of some earlier explanations do not suggest giving up all explanatory aspirations, but more careful thinking about the intended explanandum and the explanatory mechanisms. When used properly, these concepts can still be used fruitfully. One just have to keeps the explanatory questions small enough. (See Ylikoski 2001.)

331 Explaining Practices Concluding Remarks In this paper I have argued for a number of theses about Stephen Turner s critique of social practices. First, I showed that the his arguments are not directed against practice theories themselves, but against their certain explanatory uses. There are still many perfectly legitimate non-explanatory uses for these concepts. As an example I used the concept of form of life as it is used in the sociology of scientific knowledge. I also argued that the application of his arguments is wider than practices theories. In this regard his rhetoric is highly misleading. In a fundamental sense his arguments are not about social practices. They are really about much of the 20 th century social theory. Behind his arguments are two principles that every social theory should respect. The first principle says that the similarities to be explained should be describable without reference to outside comparisons. The second principle demands that any social theory should be compatible with a plausible theory of learning and communication. These two principles have a much wider scope of application than theories of social practices. References Atran, S In Gods We Trust. The Evolutionary Landscape of Religion. Oxford: Oxford University Press. Barnes, B., D. Bloor and J. Henry 1996: Scientific Knowledge. A Sociological Analysis. London: Athlone. Barnes, B Practice as collective action. In The Practice Turn in Contemporary Theory edited by T. R. Schatzki, K. Knorr-Cetina and E. von Savigny, London: Routledge. Bloor, D Wittgenstein. A Social Theory of Knowledge. London: Macmillan. Bloor, D Wittgenstein, Rules and Institutions. London: Routledge. Bourdieu, P The Logic of Practice. Cambridge: Polity Press. Collins, H. M Changing Order. London: Sage. Collins, H,. M. & Kusch, M. 1998: The Shape of Actions. What Humans and Machines Can Do. Cambridge, MA: The MIT Press. Collins, H. M What Is Tacit Knowledge?. In The Practice Turn in Contemporary Theory edited by T. R. Schatzki, K. Knorr-Cetina and E. von Savigny, London: Routledge. Friedman, M On the Sociology of Scientific Knowledge and its Philosophical Agenda. Studies in the History and Philosophy of Science 29: Lynch, M Theorizing Practice. Human Studies 20:

332 332 Petri Ylikoski Pickering, A The Mangle of Practice. Chicago: The University of Chicago Press. Pickering, A Time and a Theory of the Visible. Human Studies 20: Polanyi, M Personal Knowledge. London: Routledge & Kegan Paul. Rouse, J How Scientific Practices Matter. Chicago: The University of Chicago Press. Stern, D. G The Practical Turn. In The Blackwell Guide to the Philosophy of the Social Sciences edited by S. P. Turner and P. A. Roth, Oxford: Blackwell. Tuomela, R The Philosophy of Social Practices. Cambridge: Cambridge University Press. Turner, S. P The Social Theory of Practices. Cambridge: Polity Press. Turner, S. P Brains/Practices/Relativism. Chicago: The University of Chicago Press. Turner, S. P Tradition and Cognitive Science. Oakeshott s Undoing of the Kantian Mind. Philosophy of the Social Sciences 33: Ylikoski, P Understanding Interests and Causal Explanation. PhD. Thesis. University of Helsinki. Available at ylikoski/

333 Some Notes on Ontological Commitment and the Social Sciences 333 Some Notes on Ontological Commitment and the Social Sciences Steven Miller Abstract The analysis is a reflection of Alston s contention (in response to Putnam s conceptual relativity) that much of what we assert or believe carries no real ontological commitment. I place these remarks in the context of the social sciences, and attempt to show for the special sciences the condition holds true. The major reason for this is that our notions of collective intentionality (Searle) or weakly conceiver-independent (Moser) reality depends on the successful blocking of a regress problem. Because this problem has not been sufficiently addressed, what passes as an ontological commitment is in effect a clever semantic equivalence. In a fairly recent attempt to argue for his alethic realistic conception of truth, Alston (1996) advances some interesting comments on the nature of what, broadly, could be construed as a type of ontological commitment. The comments come in the form of trying to come to terms with Putnam s (1981, 1987, 1988) arguments for conceptual relativity. While Alston argues that the realist notion he proposes is consistent with how one would go about assessing truth values within a given conceptual scheme, he acknowledges that the ontological critique presented by Putnam is not automatically refuted by the ontological side of realism. This is not a major concern for alethic realism, Alston argues (p. 180) since a realist conception of truth does not commit one to a particular ontological stance. I am not convinced that arguing for this type (alethic) realism completely satisfies Putnam s criticism, since if we even neglect the ontological side, it still lessons the possibility that between differing conceptual schemes the assignment of truth values to statements may still differ (see Lynch, 1998 for a purported solution of this). But my concern is not about the nature of conceptual schemes, per se, but rather the possible implications for a related area of how and why Alston says what he does about ontological relativity. More specifically, Alston in trying to defuse the force of Putnam s relativism, recasts the ontological as follows (1996:178): To spell this out further, even though the form of what I am saying when I say, in the usual way, that the leaves on the tree are red appears to carry

334 334 Steven Miller with it a commitment to a substance-property ontology, I need not intend it in that way, and it is not usually so intended. In most of what people say or think, they are only concerned to commit themselves to the more specific aspects of what is being said or thought leaving the choice of one ontological framework rather than another up for grabs. It is as if we were always asserting a large disjunction containing all the ways in which the specific empirical content of the leaves on the tree are red can appear in alternative ontologies. We, so to say, represent that disjunction by explicitly formulating one disjunct is substituted for it, so long as the same empirical content is preserved. Thus, on Alston s view we can assume a type of natural onotological attitude by simply acknowledging that it is possible (logically) to split what ontological commitments one may have into a part that one is concerned with (or attending to) and another part which is not attended to. Whether the latter is even accessible to a person given the former is assumed but not demonstrated by Alston. For example, even if my cognitive facilities are properly functioning (e.g., Platinga, 1993), there is no good reason to assume that alternative ontologies are even a real option for me. Thus, the leaves on the tree are red may be simply as far as I take any ontological concern. It is not necessarily that I do (even if I am capable of doing so) first list my ontological alternatives and then choose to focus on one of them; perhaps, I do not even realize this is a possibility for me. Alternatively, even if I do realize such a possibility, its specification may be so vague that choosing it is simply irrelevant. In other words, if the leaves on the tree are red is even perceived by me to be a (possible) substance-property ontology, this choice may be so far removed from what I want to assert, that it is simply irrelevant to me. A little later on the same page, Alston says further: Most of what we believe, and are concerned to assert, carries no real ontological commitment; despite appearances it is neutral on ontological issues, including the ones that, according to Putnam, there is no objective way of deciding. Our neutrality on ontological issues, given the beliefs we wish to assert, does not erase the possibility of ontological commitment (if one so chooses) but does raise some interesting issues. One is the implication suggested by the term real above. The question is under what circumstances, indeed, become real. Thus, if my cognitive intent is not to do so (although I could if pressed sufficiently), would the content of my non-commitment be affected in such a way that it would impact my real ontological commitment? In other words, if I declare, I am now aware of my belief that the

335 Some Notes on Ontological Commitment and the Social Sciences 335 leaves on the tree are red and, furthermore, I believe in the existence of a substance-property ontology of my belief, is this now sufficient to make my ontological commitment real. Let s suppose it is. But, then, we are forced into saying that the reality of any ontological commitment (or at least one of this sort) is a type of illocutionary-act (Austin, 1962) that requires an overt stating of ( up-front, if you will) the corresponding ontological commitment. What is it, then, that does carry a real ontological commitment? Alston s response to Putnam is in the context of trying to show that Putnam s ontological relativity must, at best, be relativized to a narrow conception of ontology, while still leaving the minimal claims of alethic realism untouched in terms of conceptual schemes. While I agree with Alston s emphasis on conceptual schemes not implying indeterminateness for realism, Alston s comments on the ontological side do raise some deeper issues that may be worth a second look. If we return to the original quote, Alston is saying that most of us are not really concerned about ontological commitment and that, at best, if pressed, we most likely would acknowledge a disjunction of a given assertion with the (vague) possibility of further commitment. This may possibly be so, but then Alston goes on to say,... we don t care which other disjunct is substituted for it, so long as the same empirical content is preserved. What I wish to claim is that the preservation of the same empirical content cannot be determined with clarity in a large class of examples, and because it cannot, Alston may be correct (despite what he says here) that what we normally attend to carries with it no ontological commitment. But this lack of ontological commitment, I believe (contra Alston) is not due to a possible, if unperceived, equal empirical content, but because the empirical content cannot be assessed in any meaningful way. If so, Alston may be closer to Putnam (on the conceptual relativity issue) then he imagines. What I have in mind here are the kinds of entities (or concepts) that are prominent in some of the social sciences. I do this for two reasons. First, there is still a modicum of unease about what types of ontological commitments, if any, do socially constructed objects carry. I am not interested, however, in reviewing the many facets this debate has taken (Archer, et al., 1998; Haslanger, 1995; Searle, 1995) but rather only to note, following Moser (1993), that things can be strongly or weakly conceiving-independent. Many of our concepts in the social sciences are at least (perhaps at most) weakly conceiving-independent in the sense that they only partially depend on having had to be initially thought about. They retain a realist interpre-

336 336 Steven Miller tation in that they do not require us to constantly think about thinking about them. However, even on a minimal realist construal, their ontological status is problematic not so much due to their psychological status but due to vagueness of how else they could be viewed in terms of (some type of) equivalent empirical content. If, according to Alston, there is no ontological problem as long as the empirical content is the same, how is such an equivalence to be meaningfully expressed in the social sciences? By meaningfully expressed, I mean something like Nozick s (2001) recent discussion about the property of being objective as a (partial) function of a transformation that is invariant. Invariance means that under specified suitable transformations the empirical content of the ontological alternative remains secure. However, I am suggesting that for a class of terms in the social sciences suitable transformations are not available in some meaningful sense. The result is that the concerns expressed by Alston are relevant but for different reasons, i.e., ontological concerns are not attended to because the ontological alternatives cannot be meaningfully incorporated into one s understanding the empirical content is not the same. An example from the social sciences may illustrate this ontological ambiguity. Let us look at the central term social structure. The following seem to be fairly representative (Parpora, 1998:334): 1. Patterns of aggregate behavior that are stable over time. 2. Lawlike regularities that govern the behavior of social facts. 3. Systems of human relationships among social positions. 4. Collective rules and resources that structure behavior. Do these definitions of social structure carry any real ontological commitments, as Alston would ask? To determine this, we need to sort out when and under what circumstances this could possibly occur. Let s assume the four definitions are understood by a person, even though this understanding may not be grasped by others. And let s further assume these definitions constitute the exhaustive set within a language (L) for understanding what a social structure is. One option, then, for a type of ontological commitment would be a Tarski-like recursive analysis whereby (1), for instance, implies (2), (3) and (4) disjunctively, or perhaps just any given definition, e.g., (2) as (2) / (2). Neither would be what Alston is probably referring to, although the first above could have some parallels to his idea of a disjunct. Now if I take (1) for instance and agree, with Alston, that most of our assertions carry no real ontological commitment, what can I make of (1)?

337 Some Notes on Ontological Commitment and the Social Sciences 337 Certainly on a minimalist realist construal (1) may be potentially assessed as having a truth value, but this alone tells us little of its ontological status. If it has none, then (1) Social Structures df Patterns of aggregate behavior that are stable over time remains in an ambiguous conceptual suspension, where its empirical content (determined in the social sciences by empirical methods) gives us no description of its actual ontological status. Intensionality and Extensionality Writers such as Rosenberg (1995) here correctly pointed out that the problem of intensionality is particularly acute for the social sciences in conceptualizing them as scientific enterprises; and certainly much of this accrues to the very ambiguity and vagueness of the terms employed. Nevertheless, there remains a certain type of ontological unease for the types of statements in 1-4 that goes beyond the terminological issue. I would suggest, however, that such unease is unwarranted and, from at least one perspective, can address Alston s concern about a lack of ontological commitment. The extension issue, to begin with, has a weak and strong aspect. For 1-4 above, any pair seems to be extensionally equivalent at least in weak sense, and perhaps all pairs are also in a strong sense. The strong sense may be questionable in some case like (2) where law like would need to be given a clearer description. However, let s assume there is, following Kirkham s (1997) terminology, an extensional equivalence between any pair of definitions for 1-4. If any pair, 1-4, are extensionally equivalent, then by definition they are not intensionally equivalent their informational content is different. The historical struggle in the social sciences has in large part been to establish such extensional equivalences empiricially, with less then complete success. But even if complete success were possible for this project, it would be selfdefeating because the attempt to rid the social sciences of their intensional baggage would have the opposite effect: the intensional distinctions then, in turn, become ever more salient. Conversely, if it could be shown that any pair were intensionally equivalent, then extensional equivalence would follow. Moreover, such an analysis would have the advantage of covering possible worlds (Lewis, 1973:84-91) and, therefore, giving additional force for modal applications, especially necessarily. Such a situation could cash out in having both the extensional and intensional projects of the social sciences being simultaneously met. This situation, in turn, might strongly suggest, in relation to Alston s concerns for

338 338 Steven Miller ontological commitment, that there could be both a content which carries a commitment and an empirically equivalent content, whatever its disjunctive nature may be, e.g. (1) and (3) or (4), etc. However, establishing what constitutes such intensional equivalence for social science terms is problematic. Kirkham (1997:10) defines a case of essential equivalence as one that has the following structure: Necessarily (x / y), which if we substitute any given definition of social structure would be, for example, Necessarily (systems of human relationships among social positions (3) / collective rules and resources that structure behavior (4). However, it is not clear how any intensional equivalence is to be settled upon, other than a purely stipulative one. Likewise, even if some type of equivalence can be claimed for the actual/possible worlds, it is easy to produce a possible counter example, thus invalidating the claim of any type of essential equivalence. For example, collective rules and resources that structure behavior (4) may be possible without the creation of social positions (3). If such is the case for a large class of terms and claims in the social sciences, then the extionsonality project cannot be fulfilled as a function of the intensionality one. If, on the other hand, it is only the extensionality project that is germane to the social sciences, formulating a credible view of at least minimally necessary and/or sufficient conditions is required. This project is also full of pitfalls even if one resorts to claims of empirical adequacy. This can be quickly seen when we view (1)-(4) as definitions of a social structure. Commitment Possibilities The considerations just mentioned are, of course, known within the philosophy of the social sciences. What I think is somewhat new, however, are the implications for the possibility of any meaningful type of ontological commitment for the social sciences. By meaningful I mean some type of minimal intersubjective consensus by relevant observers of what sort or type of ontology would be a sufficient (at least) indicator of what it is that is there? If a key concept in the social sciences such as social structure does commit us to anything, what could it possibly be? Alston sees it in two ways: that the things we talk about and believe carry no real ontological commitment, or if we really press someone on the issue, there may be some (at least one) class of things serving as a disjunct to the class of assertions actually made. The ontological side of the disjunct must only have the property that it retains the same empirical content of the original assertion. There are a number of

339 Some Notes on Ontological Commitment and the Social Sciences 339 positions intertwined here, and although I am in favor of Alston s overall analysis, some comments are needed. Then I will return to the issue of the social sciences. There is the matter of discerning when and how an assertion carries no real ontological commitment. I believe Alston has in mind locutionary speech acts so that in his related position some form of empirical equivalency is at least potentially possible. When he says, then, that most of our beliefs and assertions carry no real ontological commitment, he must have in mind a standard or criterion by which they do not do so. But if they are something more than performative speech acts, it is unclear what is at stake here. Put differently, there are really two issues confronting us. The first is the blanket statement that most of our beliefs and assertions are ontologically-free. The second is that they appear to be but upon closer examination they do have such a commitment but only on the alternation and empirical equivalency stipulations. There is also a third possibility that many (most) of our speech acts do not have ontological commitment because once we try to see what such commitment might consist of by way of alternation/empirical equivalency, we see that the implication to the assertion or belief is not possible, or perhaps too vague or ambiguous to be of any real use. On the first possibility if I say a social structure is, for instance, [the] collective rules and resources that structure behavior (4), it is the kind of assertion that simply carries no ontological commitment; and this is so irrespective of whether or not the terms are vague or ambiguous. Here, neither extensionality nor intensionality considerations figure into the equation. Upon the next reading, however, (4) does have some type of commitment either on the alternate [i.e., (3) or (1) or (2) or whatever] or empirically equivalency criterion, or perhaps both. This type of situation would be, of course, at least the one favored by the social science enterprise. Indeed, but mistakenly, empirical equivalence is defined as being operationally defined; that is, social structure under (4) is whatever we purportedly measure it as being. The Alston conception is not about this type of equivalency, of course, but (minimally) a semantic equivalence where empirical is broadly construed. But here, even if the alternation stipulation is granted, the empirical equivalence notion cannot be sustained since even on semantical grounds there is no way of assuring that (4), above, commits us to any other [(1), (3), etc.] definition. Moreover, a possible move to establish such equivalence by abstracting the content of (4), or (4) and any other definition, may lead us to the possibility of an ontological commitment but one that would be unsuitable for the purposes of the social sciences. For example, taking (4)

340 340 Steven Miller and arguing for almost any version of metaphysical/ontological classification would, even on a minimal construal, lead us into swampyconceptual terrain. Should (4) be defined in nominalist or metaphysical realist terms, or as bundles, substrata or substances, or even as genuine propositions? It would even be difficult to come to some consensus if in (4) collective rules, resources and structure[d] behavior are kinds or have properties, or exemplify themselves in certain ways, and so forth. While Alston s alternation principle might stand, the empirical equivalence one can only if its abstractness is conceded. But then if it is, it does not handle in a meaningful way the commitment problem for the social sciences: what bundle(s) cover collective rules or social structure itself? Returning to the third issue mentioned above, I believe Alston s real point is that many of our assertions and beliefs carry no actual ontological commitment simply because their construal by traditional ontological and metaphysical classification systems makes virtually little or no sense. And, perhaps, this is Alston s real intent in that he can show by whatever interpretation given above (or at least my interpretation of what he is saying) that what is really at stake in assertions and beliefs is that they can only be shown to exhibit the minimal criteria for alethic realism. Are we any clearer on what type(s) of ontological commitments are made in the social sciences by central terms such as social class? Actually, many of the issues related to what constitutes the social ontological have been worked out nicely by Searle (1995, 1998) using his earlier work on intentionality (1983) and the nature of the mind (1998). Searle s work won t be reviewed here, but such features as his distinction between observer independent and observer-dependent notions of reality are of importance here and in general. Basically, Searle holds that by way of the mind-intentionality connection, forms of collective intentionality are historically developed eventually resulting in a variety of forms of institutional reality. Such representations of the social world are ontologically substantive, often exemplified by how constitutive rules come about and serve as genuine ontological types. That such ideas of social reality are, indeed, real is rather obvious. For this perspective, and related to the social sciences, constructs such as social structure do exist, even if their empirical content is not strictly equivalent. Thus, contra Alston, claims of these kinds do carry some type of ontological commitment. Searle, indeed, makes a persuasive case for the social being real. However, there is one issue in all of this that needs some additional clarification. It is the one assumption that is often quickly alluded to in ontologi-

341 Some Notes on Ontological Commitment and the Social Sciences 341 cal discussions of the social but one that is not pushed for enough. It is one also, then, that gives rise to Alston s concerns that our assertions and beliefs carry no real or important ontological commitments. Searle (1998:117) himself puts it nicely as follows: Notice that the intentionality that creates observer-relative phenomena is not itself observer relative. The fact that this object is a chair depends on, among other things, our attitudes, but those attitudes are not themselves observer-relative phenomena through the exercise of our intentionality, that intentionality does not depend on any further intentionality. Once we have an attitude, it does not matter whether anyone else thinks we have that attitude. In a similar way, Moser (1993:21-22) conceptualizes the observer-relative issue as this: Realists can use a distinction between strong and weak conceiving-independence. Something, X is strongly conceiving-independent with respect to its having feature F if and only if X s having F does not depend, logically or causally, on someone s conceiving of something. If, for example, the earth s magnetic field exists independently of anyone s conceiving of something, then the earth s magnetic field exists with strong conceivingindependence. Something, X, is weakly conceiving-independent with respect to its having feature F if and only if X s having F does not depend, logically or causally, on someone s conceiving of X s having F. Moser (p. 21) puts this process less formally by saying that one s conceiving of Socrates is weakly conceiving because it does not require one s conceiving of one s conceiving of Socrates. Searle, likewise, situates the possibility of a collective intentionality as ultimately a primitive. Now, if the possibility of regress or circularity must be stopped at an intentionality which in turn cannot be intended, and if this is defined as the basis of a weakly conceiving ontology, then the commitment to such an ontology is not only observerrelative but also non-observer, non-relative. It is the fulfillment of both conditions simultaneously that raises paradoxical, if not contradictory, issues for an ontology in the social sciences. There is, of course, as Searle rightly argues, a social reality made up of social and institutional facts, but it is a reality that cannot possibly be supported by the foundationalism upon which it ultimately depends. This lack of support does not lie in the fact that weakly conceiving-independent socially produced entities can disappear as conditions (e.g., col-

342 342 Steven Miller lective intentionality) wavers or changes this inevitably happens rather, the support required is based on an implicit metaphysical principle that suggests observer-relative reality has (but cannot have) a bedrock intentionality. The result is that any attempt to clearly specify what an ontological commitment is for social science statements can certainly be made but the equivalency is at best arbitrary. It is probably in this sense that Alston s claim is correct: many of our beliefs and assertions carry no real ontological commitment. However, for the social sciences the implication is stronger; they do but doing so doesn t (ontologically) make it so. Thus if I say that a social structure = systems of human relationships among social positions (3), there cannot (e.g., Putnam) be any independent confirmation of the right-hand side of the equation outside the conceptual system that is being employed. Or somewhat more precisely, a type of confirmation is possible but its very plausibility is contingent upon, first, a collective intentionality that itself cannot be justified, and, secondly, a methodological apparatus which makes it appear that there is a unique ontological referent. This later condition is fulfilled by equating empirically equivalent with operationally equivalent in which a measurement technique is substituted for an ontological assumption. For instance in (3) above, my operational definition of social positions not only creates the ontological category but ensures its existence by the measurement technique I employ to measure it. Moreover, it does not matter if in my attempt to hypothesize a relationship between social positions and some other indicator, such as performance on a so-called standardized test, that the predicted relationship may be greater or weaker than I hypothesized. This simply accounts for a range of possible values that is already a part of my measurement system; it does not weaken or strengthen the case of this (or any) definition of social structure being independently justified as an ontological category. Perhaps, if one were to engage in a type of semantic ascent here, it would be possible to argue that there is an essence within the methodology itself [e.g., Bernoulli s Theorem (Hacking, 2001)] that when correctly applied does justify constructs such as social positions and test achievement. However, this is not necessarily the case since the constructs themselves should have (ontologically) an independent reality. Their variation by way of a more general principle is simply (still) a contingent matter. Along these lines, Moser (1995:29) has succinctly stated a nice distinction pertinent here as, ontology construed as what there is, one might argue, does not recapitulate semantics. For the case at hand, however, I would like to reverse it saying that for many beliefs and assertions in the social sciences

343 Some Notes on Ontological Commitment and the Social Sciences 343 the ontology they proclaim is nothing more than a recapitulation of its semantics. If no ontological commitment is made, nothing is lost because nothing is really there. Or more accurately, what is claimed to be there, is there only because we say and agree that it is. Such considerations do lend themselves to a possible anti-realist stance but need not if one accepts the weakly conceiving-independent thesis. A minimally realist ontological position would be consistent with this and certainly a weak realist epistemology would be consistent. The key, again, appears to be how a weakly conceiving independent-reality is configured in terms of regress-circularity stopping moves. If observer-relative intentionality is stipulated as primitive, then a weakly conceiving-independent reality can be claimed, even though such an ontological commitment is largely meaningless. Its meaninglessness stems from the fact that intentional reflexiveness is only in the direction of the (collective) intention as intention to the reality so described. This overlooks the need for the intentionality to be directed, so to speak, in the other direction: how otherwise could I recognize and justify the (collective) intention that is being ontologically committed to? But if this other direction is acknowledged, then the regress problem re-emerges. I conclude that when Alston s claim is applied to the social sciences it is essentially correct. However, the reason why these types of beliefs and assertions have no real ontological significance is because of the taking of intentionality as a primitive. While I cannot pursue it here, I believe the regress problem can be solved within Searle s own theory of Mind. Namely, if intentionality is a naturally occurring aspect of what we call Mind, and if the Mind is the result of our evolutionary history (especially the development of the brain), what becomes collective intentionality gains or has whatever conceiving-independent status it has because of this, not in spite of it. References Alston, W.P. (1996). A Realist Conception of Truth. Ithaca, NY: Cornell University Press. Archer, M., Bhaskar, Collier, A., Lawson, T., & Norrie, A. (1998). Critical Realism: Essential Readings. New York: Routledge. Austin, J.L. (1962). How to Do Things with Words. Cambridge, MA: Harvard University Press. Hacking, I. (2001). An Introduction to Probability and Inductive Logic. Cambridge, MA: Cambridge University Press.

344 344 Steven Miller Haslanger, S. (1995). Ontology and Social Construction. Philosophical Topics, 23(2), Kirkham, R.L. (1997). Theories of Truth: A Critical Introduction. Cambridge, MA: The MIT Press. Lewis, D. (1973). Counterfactuals. Cambridge, MA: Harvard University Press. Lynch, M.P. (1998). Truth in Context: An Essay on Pluralism and Objectivity. Cambridge, MA: MIT Press. Moser, P.K. (1993). Philosophy after Objectivity: Making Sense in Perspective. Nozick, R. (2001). Invariances: The Structure of the Objective World. Cambridge, MA: The Belknap Press. Platinga, A. (1993). Warrant and Proper Function. New York: Oxford University Press. Porpora, D.V. (1989). Four Concepts of Social Structure. Journal for the Theory of Social Behavior, 19(2): Putnam, H. (1981). Reason, Truth, and History. Cambridge, MA: Cambridge University Press.. (1987). The Many Faces of Realism. LaSalle, IL: Open Court.. (1988). Representation and Reality. Cambridge, MA: The MIT Press. Rosenberg, A. (1995). Philosophy of Social Science (2 nd ed.). Boulder, CO: Westview Press. Searle, J. (1995). The Construction of Social Reality. New York: Free Press. Searle, J.R. (1983). Intentionality: An Essay in the Philosophy of Mind. Cambridge, MA: Cambridge University Press.

345 Two Modes of Collective Belief 345 IV On Margaret Gilbert s Proposal of Collective Beliefs

346 346 Christopher McMahon

347 Two Modes of Collective Belief 347 Two Modes of Collective Belief Christopher McMahon Abstract Margaret Gilbert has defended the view that there is such a thing as genuine collective belief, in contrast to mere collective acceptance. I argue that even if she is right, we need to distinguish two modes of collective belief. On one, a group s believing something as a body is a matter of its relating to a proposition, as a body, in the same way that an individual who has formed a belief on some matter relates to the proposition believed. On the other, a group s believing something as a body is a matter of its relating to a proposition, as a body, in the same way that in individual who is forming a belief on some matter relates to the proposition believed. As part of her general theory of plural subjects, Margaret Gilbert has offered an account of collective belief. Central to this account is the rejection of the summative view. This holds that it is appropriate to speak of collective belief if and only if all the members of some group believe a particular proposition. According to Gilbert, the summative view is mistaken in both the claims it makes. The fact that all the members of some group believe a particular proposition is not a sufficient basis for ascribing a collective belief to the group, and belief by all the members is not necessary either. Indeed, we can sometimes speak of collective belief even if none of the members of a group believe the proposition the group is said to believe. Gilbert s view is that, the members of a population, P, collectively believe that p if and only if they are jointly committed to believe that p as a body. 1 Joint commitment is the element that is missing from the summative account, and for Gilbert, a group can be jointly committed to the instantiation of some mental property as a body when none of the associated individuals instantiates that property. Recent discussion of Gilbert s view has focused on whether what she has labeled collective belief would be better characterized as collective acceptance. Those who favor an acceptance interpretation Gilbert calls them rejectionists cite a number of differences between belief and accep- 1 Gilbert, Margaret Belief and Acceptance as Features of Groups. Protosociology 16: 42.

348 348 Christopher McMahon tance. 2 Perhaps the most important are that acceptance is under the control of the will in a way that belief is not, and that belief must aim at truth, while acceptance is typically guided by considerations of utility. A paradigm of acceptance in the sense that contrasts with belief is accepting something for the purposes of argument to see what follows from it, without undertaking any commitment to embrace what follows from it. One important reason that acceptance can seem to be the appropriate label for what Gilbert calls collective belief is that joint commitment, in her sense, brings with it obligations, and failure to discharge these obligations is open to rebuke. Talk of obligations seems to be appropriate only in connection with action and the will. And if the sustaining of a personal attribute by a group is a matter of the performance of certain actions by the members, it is natural to suppose that the attribute itself is best classified as one that is under the control of the will. Gilbert has replied that although a collective belief can be said to be constituted by the collective will, understood as a joint commitment among the members to perform certain actions, it does not follow that a collective belief is willed by the collective. 3 She notes that there is room in her account for a notion of collective acceptance as well, and presumably in this case we can speak of something that is willed by the collective. 4 In this essay, I shall not be directly concerned with these issues. Instead I shall try to show that even if we accept Gilbert s claim that what she is talking about is appropriately labeled belief rather than acceptance, we need to make a distinction between two different ways that a group can be said to believe something as a body. The distinction is based on the contrast between the epistemic operation of belief formation and the downstream behavior occasioned by a belief after it has been formed. The failure to note this distinction is responsible, I think, for a certain amount of confusion in Gilbert s account of collective belief. I begin by exploring the evidence for the two different modes of collective belief in Gilbert s writings. Then I offer an explicit account of the contrast in terms of a distinction between two forms of rational cooperation. The discussion draws on my book Collective Rationality and Collective Reasoning. 5 2 See Meijers, Anthonie Collective Agents and Cognitive Attitudes. Protosociology 16: In What is Wrong with Rejectionists (forthcoming) Gerhard Preyer defends collective belief against rejectionist arguments. 3 Belief and Acceptance as Features of Groups: Belief and Acceptance as Features of Groups: New York: Cambridge University Press.

349 Two Modes of Collective Belief 349 The focus of the investigation will be Gilbert s treatment of collective belief in On Social Facts. 6 Chapter V of this book contains her most detailed consideration of examples of collective belief, and the two different phenomena that I want to distinguish are displayed in these examples. Gilbert s discussion in this work uses the phrase joint acceptance. But, as she has noted, there is a perfectly legitimate sense of acceptance in which it involves belief, rather than the utility-guided action that writers who contrast acceptance with belief have emphasized. 7 In describing Gilbert s view, I shall follow her in speaking of acceptance, but I shall distinguish between a discretionary sense of acceptance and an epistemic sense of acceptance. I. Gilbert s View A useful place to begin expounding Gilbert s view is the following statement by her: What I want to stress is that, as I understand it, the behavior that results from collective belief is driven by the concept of belief and the concept of X-ing-as-a-body. It is as if the participants ask themselves, What do I need to do to make it the case as best I can that I and these others together believe that p as a body? and then act accordingly. I have suggested that the answer given by our everyday understanding for the simple interpersonal case is that, among other things, in reasoning together we say things that entail p rather than not-p, we do not deny p without preamble, and so on. More briefly, we attempt as best we can to make it true that the body we constitute relates to p in the way any individual who believes that p relates to p. 8 Gilbert speaks here of actions that individuals might perform, but as we have seen, she does not think that this prevents us from ascribing belief, in contrast to some state of the will, to the group. Let us consider in a bit more detail how it might be appropriate to employ the concept of belief, in contrast to acceptance, in connection with the phenomena Gilbert describes. I have said that I shall distinguish two forms of acceptance, discretionary acceptance and epistemic acceptance. The contrast turns on the extent to London: Routledge, In what follows, the parenthetical page references are to this book. 7 See, for example, Gilbert, Margaret. More on Collective Belief. In her Living Together: Rationality, Sociality, and Obligation, Lanham MD: Rowman and Littlefield. 8 More on Collective Belief: 356-7

350 350 Christopher McMahon which what we accept is up to us. Epistemic acceptance is forced by the evidence; discretionary acceptance is not. When we consider joint acceptance, this contrast has implications for the contributions that the members of a group make to the group s accepting a proposition. Joint epistemic acceptance involves contributions that are themselves forced by the evidence; joint discretionary acceptance involves contributions that are under the control of the will. Acceptance in the sense intended by those who distinguish belief and acceptance is a form of discretionary acceptance. Some of Gilbert s remarks in On Social Facts suggest that mere acceptance is what she has in mind. Thus she says that one who participates in joint acceptance of p thereby accepts an obligation to do what he can to bring it about that any joint endeavors...be conducted on the assumption that p is true (p. 306, emphasis added). I believe, however, that it is possible to move closer to the condition that Gilbert specifies to relate as a body to some proposition in the way an individual who believes it relates to it while remaining within the domain of discretionary acceptance. Gilbert has offered a number of glosses on the notion of collective belief, including belief of a group. To the extent that this suggests a discrete center of consciousness, many will find it objectionable. But suppose we employ instead the notion of the position of the group. To speak of some view as the position of the group is to imply that the group has made a decision to defend it. This decision need not, however, be based on a disinterested attempt to determine what the evidence supports. Often, the situation can be most accurately described as one in which the group first recognizes that its interests would be served if a certain view were true, and then undertakes to defend this view. Corporations often seem to proceed in this way. It was for many years the position of tobacco companies that smoking does not cause cancer. It seems now to be the position of oil companies that drilling in the Arctic National Wildlife Refuge will not do any environmental damage. But more than in the case of accepting something for the sake of argument, that some view is the position of the group means that the group attempts, as best it can, to relate to the view in the way an individual relates to something she believes. In particular, the members undertake to defend p as true. We should be clear about why it is appropriate to classify cases of this sort as involving discretionary, in contrast to epistemic, acceptance. Once it has been decided what the group s position will be, the members cooperate to make it the case that the group relates to the proposition in question in the way an individual relates to something she believes. But in the examples just

351 Two Modes of Collective Belief 351 considered, the decision concerning what view to accept is not the result of a good faith effort to determine what the relevant evidence supports. It is driven by considerations concerning what it would be in the group s interests to have accepted by others. The thinking underlying a group s decision concerning what its position will be need not be self-consciously calculating; it may involve some form of collective self-deception. But the will enters into the adoption of the group s view in a way that is incompatible with proper epistemic functioning. As has been noted, this is ultimately an observation about what is going on in the individuals who constitute the group. The force of these points may become clearer once it is noticed that we can speak of the position of a group, and of a collective effort to defend a view as true, even where there is no pretence that the decision concerning which view to defend was guided by the evidence. In a debating contest, one side may be assigned the task of defending the proposition that the British monarchy should be abolished and the other that it should be retained. A debating team will try to relate as a body to the proposition it is defending in the same way that an individual relates to a proposition he believes. But there is an element of discretion indeed of arbitrariness in the decision concerning which view will be the position of a debating team. Some passages in On Social Facts seem to fit this discretionary interpretation of believing as a body. Thus Gilbert says, one may feel that it is above all important that some view or other be jointly accepted by group members (p. 299). This strongly suggests that the decision concerning what view to adopt falls within the scope of discretion. More important, a number of her examples of what the joint commitment characteristic of collective belief is a commitment to do focus on downstream parts of the process that are under the control of the will, in contrast to the decision to adopt a particular position as the view of the group. Thus the obligations associated with the joint commitment underlying collective belief are obligations not to deny a view that has been accepted as the view of the group (p. 292), and to assert it when speaking in one s capacity as a member of the group (p. 303). The particular example that Gilbert mostly focuses on in On Social Facts, however, is continuing a conversation, and this brings us to the second way of understanding collective belief belief as a body. She says that joint acceptance in a conversational context can happen when the other parties decide to go along with what someone has said (p. 295). This suggests that we are still dealing with the discretionary sense of acceptance. It suggests that the contributions that the members make to the group s acceptance of a view are actions by them that are under the control of their wills. But in her subse-

352 352 Christopher McMahon quent discussion of the case of conversation, Gilbert slips into language that raises questions about whether discretionary acceptance is really what she has in mind. Gilbert speaks of the participants in a conversation as continuously negotiating propositions that are to be jointly accepted (p. 295). This seems to imply a process of bargaining under the control of the will. But consider the following passage: [P]eople can put their own personal views on hold in negotiating a jointly accepted view. They may wish to hear what is said in foro publico before venturing any opinion about what should be jointly accepted. And they may find a view which is put forward acceptable enough, without having any idea who is in the majority and who is not, or even whether there is a majority view at all. (Clearly there need not be any prior majority view on the matter.) As a result, they can easily come jointly to accept a view with others while at the same time the majority had always held a contrary view up to the point of acceptance (p. 301). The process of negotiation described here seems to be one that involves considering what the relevant evidence supports. And the result that Gilbert envisages appears to be something that could be characterized as a joint understanding of what the relevant evidence supports Let us explore this point further. Gilbert says, Some individual might personally believe that p but wish to have this view ratified by some others before really going public about it. In a conversation among a few friends, he might bring it up and find that no one objected (p. 298). Are we to say that if he brings it up and finds that no one objects, it constitutes the view of the group? The following passage suggests that the answer is yes. Suppose, first, that it is a general assumption in conversation that views which are not challenged are understood to have been jointly accepted, and that it is incumbent on one who jointly accepts a view with certain others to presuppose that view in the future course of the dialogue, at least (p. 296). Gilbert speaks here of the challenging of a view, and of an assumption within the group that a view that is not challenged is understood to have been accepted. These ideas are most readily understood in terms of the notion of epistemic in contrast to discretionary acceptance. To challenge a view usually means bringing forward evidence that it is false. The fact that no one challenges what someone says creates a presumption within the group that no one knows of evidence against it. The natural inference will be that all accept it as true. This means that we are in the domain of epistemic acceptance. We have a kind of joint acceptance where individual contributions

353 Two Modes of Collective Belief 353 take the form, not of actions or choices under the control of the will, but of acts of epistemic acceptance of each member s being led to accept something on the basis of the evidence On this second interpretation of collective belief, the idea of a group s relating to a proposition in the way that an individual who believes it relates to it acquires a new layer. The contributions that the members make to the group s believing something as a body do not merely take the form of downstream acts of argumentative defense. The members contribute to the group s acceptance of the proposition by adopting it themselves in the way appropriate to individual belief. The way of adopting a view that is appropriate to individual belief is to adopt it if and only if the evidence warrants it. So a collective belief of this sort is formed when all the members agree that the relevant evidence supports a particular conclusion. The significance of this for the question of whether collective belief is summative will be discussed further below. Some additional remarks by Gilbert can also be given a natural interpretation if we suppose that joint epistemic acceptance plays a role in the negotiation of collective beliefs characteristic of conversation. She notes that someone who dissents from a view that has been jointly accepted in a conversation will be perceived as doing something shocking (p. 290). But we can account for this without supposing that the person in question is violating an obligation of joint commitment. If joint acceptance is acceptance in the epistemic sense, then all the parties personally believe what is jointly accepted, and this is common knowledge. If one of them subsequently denies something that has been jointly accepted in this way, he will seem to be contradicting himself. Indeed, this may even elicit rebuke. But the basis of rebuke in a case of this sort will be a failure of proper epistemic functioning, not the violation of an obligation. If a member no longer believes something that was jointly epistemically accepted in a conversational group of which he is a member, he is not doing something wrong. Belief is not under control of the will. Indeed, it is arguable that in this case, he has an obligation to inform the others that the joint acceptance that previously existed no longer does. Similar points can be made about Gilbert s observation that the members of a conversational group negotiate the view they will accept. As was mentioned earlier, this could be understood as a process of bargaining under the control of the will, which would go with a discretionary interpretation of the decision concerning what view to adopt. But negotiation can also be understood as mutual reconsideration of what the evidence supports, prompted by the fact that there is disagreement about this within the group. A view is

354 354 Christopher McMahon jointly accepted when, after mutual reconsideration, it is found by all to be supported by the evidence. Before concluding this section, we should note a further point about the distinction I have drawn. Gilbert says that a group s joint acceptance of a view can put pressure on personal belief, and that personal belief can be a direct reflection of collective belief (p. 313). These points presuppose a discretionary view of collective belief. On the epistemic interpretation of joint acceptance, wherever there is collective belief there is already personal belief. The pressuring that Gilbert describes here may seem to undercut the distinction that I have drawn between two forms of joint acceptance, but it does not. Gilbert is not saying that if a proposition has been accepted as the view of the group in the discretionary sense, this constitutes evidence that supports its epistemic acceptance. The pressures in question are purely causal. Gilbert is no doubt right to note this causal pathway. But once we have the distinction between discretionary and epistemic acceptance firmly in hand, the sociological fact that this causal pathway exists constitutes a reason to ascertain in what mode a given view has been jointly accepted, and to put one s epistemic guard up in the presence of views that have been jointly accepted in the discretionary sense. II. Rational Cooperation I shall now propose a way of modeling these two different forms of believing something as a body. The model is based on the idea that collective phenomena we phenomena are to be understood as forms of rational cooperation. 9 In this section I set out the general features of the model. In the following sections, I apply it to the case of collective belief. Rational cooperation among individuals involves the production of a combination of actions (or more generally, acts, including acts of the mind) that constitutes, or brings about, something that each participant can regard as desirable. Let us take this to mean that each would regard the realization of the combination as preferable to its nonrealization, where its nonrealization is understood as the combination of actions that would exist if (1), each 9 The account of rational cooperation that follows draws on the discussion in Chapters 1 and 2 of my Collective Rationality and Collective Reasoning. I defend the idea the we phenomena are to be understood in terms of rational cooperation in Shared Agency and Rational Cooperation (forthcoming in Noûs).

355 Two Modes of Collective Belief 355 member supposed that none of the others would do their parts, and (2), each then performed whatever action would best satisfy her concerns under that assumption. I shall call this the noncooperative combination and the situation that it creates the noncooperative situation. We must, however, distinguish two ways that reason can be regarded as entering into cooperation. The first concerns the rationality of doing one s part in a cooperative undertaking. Let us say that a jointly feasible combination of actions is one that each member of the posited group of agents would regard as satisfying the above condition as being such that its realization would be preferable to the noncooperative combination. The term jointly feasible implies that each member of the group has a reason to contribute to such a scheme. How is this reason to be understood? In the case of a coordination problem, where the defection of any party prevents all from obtaining the benefits of cooperation, individual rationality (the rationality of acting so as to increase individual preference-satisfaction) suffices to make contribution rational. When, however, some can defect without losing the benefits of what the others produce, as in situations possessing the form of the prisoner s dilemma, individual rationality will dictate defection. Thus a theory of rational cooperation must somehow provide a reason for the parties to contribute in cases of this sort. Moral reasons, such as the reason of fair play, or obligations grounded in an agreement, are usually invoked here, but I have argued elsewhere that they cannot reliably secure the requisite contributions when agents have moral reasons for defecting. 10 The simplest solution, which I will adopt, is to suppose that the parties have an effective disposition to assign the same payoff to defection that they assign to the noncooperative situation. This has the result of turning a prisoner s-dilemma situation into a kind of a coordination problem. The key feature of a coordination problem, for present purposes, is that unilateral defection is not beneficial from any party s point of view. I shall say that individuals who have the sort of disposition described are cooperatively disposed. This strategy has the advantage of enabling us to give a single account of the rationality of contributing to a collective endeavor in both coordination problems and prisoner s dilemma situations. In both, a cooperatively rational agent will make her contribution to the combination at issue if the benefit (in light of the reasons and values she acknowledges) that will result from all the contributions that will in fact be made, hers included, exceeds what she would find in the noncooperative situation See Collective Rationality and Collective Reasoning, chap 1.

356 356 Christopher McMahon This has an important consequence. It means that a cooperatively rational agent will not judge herself to be justified in contributing to a cooperative endeavor unless she has some reason to believe that enough others will do their parts. In large-group contexts, this reason is often provided by the fact that the choice of a scheme to implement has been made by a choice mechanism that is commonly known to possess de facto authority in the group (or the ability to coerce compliance). But in the small-group contexts on which Gilbert focuses in discussing collective belief, the requisite assurances can be provided informally, by observation of the behavior of others, or common knowledge that everyone is cooperatively disposed. The second way that reason plays a role in cooperation concerns the choice of the combination of actions let us call it a cooperative scheme to be realized. For a group of cooperatively disposed agents, there will usually be several different combinations that would count as jointly feasible cooperative schemes, in the sense specified above. Before cooperation can take place, one of these must be designated as the combination on which cooperation is to proceed, and each member of the group must be able to regard this choice as rationally made. How are these conditions to be satisfied? On the view of rational cooperation that I favor, the mechanism of choice is, at the most basic level, agreement among the parties, but we must be clear about how this is to be understood. The notion of agreement can be interpreted either normatively or descriptively. In the normative sense, agreement involves an agreement of a particular kind: a joint undertaking which gives each party an obligation to do his part in some designated combination. In the descriptive sense, agreement is simply de facto coincidence of preference. Each, consulting his own values, finds the result acceptable. Given that the parties are cooperatively disposed, the agreement that secures the selection of one of the jointly feasible schemes need only be of the descriptive sort: a de facto coincidence of preference. It suffices for the rationality of the choice of a given cooperative scheme that each, viewing this scheme from the standpoint of his values, can endorse it as preferable to any of the other jointly feasible schemes. This understanding of the rationality of the choice of a scheme may seem to be severely limiting. In the normal case, the parties will rank the candidate schemes differently, and in particular, will disagree about which scheme should be ranked first. But the view I propose can accommodate such cases 11 In Collective Rationality and Collective Reasoning, chap. 1, I call the principle that codifies cooperative dispositions, so understood, the Principle of Collective Rationality.

357 Two Modes of Collective Belief 357 if we take the focus of agreement (of the coinciding preferences) to be a particular procedure for choosing among schemes. That is, each ranks a given procedure as preferable to all others. When this situation obtains, the parties will be able to regard the choice of the scheme on which they act as guided by reason. III. Collective Belief and Discretionary Acceptance Now let us consider how this account of rational cooperation applies to the phenomenon of collective belief. In Section I, we saw that Gilbert s notion of the joint acceptance of a view as the view of the group appears to elide an important distinction between discretionary acceptance (understood as bringing with it a commitment to do what is required if the group is to be related to a proposition in the same way as an individual who believes it) and epistemic acceptance in the standard sense of coming to believe something because the evidence supports it. How does this distinction look if we suppose that the realm of the collective is the realm of rational cooperation. To take the discretionary form of collective belief first, on the view of rational cooperation that I have proposed, cooperation effects the implementation of a cooperative scheme, understood as a combination of actions. These can be communicative actions. We do not need to be very specific about how cooperative performance of communicative actions the creation of a coordinated combination of communicative actions might benefit a group. Any purpose that might be served by a group s being able to defend a particular position can play the role of a cooperatively realizable benefit. Often, the purpose will have something to do with the group s maintenance of a certain posture in its dealings with outsiders. An example is the oil industry s taking the position and in that sense, believing as a body that drilling in the Arctic National Wildlife Refuge will not damage the environment. But a group might also adopt a position in the discretionary sense simply to have an established answer to a particular question that can serve as a basis for further cooperative enterprises in which it might engage. This appears to be the situation that Gilbert has in mind when she suggests that it may be important that some view or other be jointly accepted by a group. Related to this, the discretionary adoption of a particular position as the position of the group might facilitate coordination among the individuals within the group. Action is typically a product of desire and belief. Sometimes coordination requires agreement about which actions will be performed. But there may

358 358 Christopher McMahon also be a place for agreement on the view of the world that will guide each individual s choices, with the resulting actions determined by how that view combines with the particular desires of each agent. This may enable a group to maintain its distinctive identity as a group while dispensing with authoritative direction of the actions of its members. IV. Collective Belief and Epistemic Acceptance Approaching the domain of the collective in terms of rational cooperation enables us to provide a clear account of the difference between the form of collective belief just described, based on discretionary acceptance, and collective belief based on epistemic acceptance. The key is to introduce a particular sort of cooperative endeavor, collective reasoning. 12 In collective reasoning, the parties cooperate to amass a pool of reasons or criticized arguments on which they all can draw to form better-justified judgments. Often it is natural to understand these judgments individualistically. That is, the parties need not agree on what the amassed reasons support in order to realize the cooperative benefit: better-justified judgment. Although they continue to disagree, each may be able to form a better-justified judgment than the one she would have formed in the absence of collective action to amass the pool. As an example we might take an academic seminar or discussion group. The participants in a seminar or discussion group do not usually form the same judgments about the matters discussed, yet they have all received a benefit that none of them could have created alone. The judgment with which collective reasoning concludes can also, however, be an explicitly collective act. That is, the judgment concerning what the amassed reasons support can be made collectively. A collective judgment is generated when each member of the group suspends judgment about the matter at issue declines to draw a final conclusion from the assembled evidence until something all can regard as justified has been identified The discussion that follows draws on chapters 5, 6, and 7 of my Collective Rationality and Collective Reasoning. 13 The question of the extent to which contribution to a collective judgment is under the control of the will of each contributor is a delicate one. What happens when a collective judgment is needed is that each member of the group suspends judgment until a proposition that all can accept in the epistemic sense has been identified. One can suspend the influence of the evidence of which one is aware by deliberately seeking further evidence, and the decision to do this is under the control of the will. But finding something that

359 Two Modes of Collective Belief 359 I have argued elsewhere that the members of a group derive no additional epistemic benefit by making collective judgments of this sort. 14 The cooperating parties do not more fully achieve the goal they have as individuals forming better-justified judgments by incorporating their judgments into a collective judgment of the sort described. But if realizing some further cooperative benefit requires the group to agree on what a certain body of evidence supports, there can be good reason for reaching conclusions in this cooperative way for making collective judgments. For there to be agreement on what a certain body of evidence supports, all must epistemically accept the same conclusion. And holding out for a collective judgment about what the evidence supports ensures that this condition is satisfied, and is commonly known to be satisfied. A collective judgment of the sort described consists of a number of individual judgments made in a coordinated way. Any disagreement about what the relevant reasons support postpones the group s drawing of a conclusion. There is thus no place for the idea of a member s going along with the view of the group without having the requisite personal beliefs. We can contrast this with the discretionary form of joint acceptance. In that case, a member need not personally hold the belief that the group is committed to defending (although it will be disruptive for her to say what she believes without qualifying this as her personal view). When joint acceptance is discretionary, the reasons for behaving downstream in the way appropriate to belief are independent of what the members as individuals judge the relevant reasons to support. Some of Gilbert s observations about the phenomenon of conversation, quoted earlier, suggest that she is thinking about what I have called collective reasoning, and that what she identifies as collective beliefs formed in the course of a conversation propositions that are jointly accepted by the group in the course of a conversation are collective judgments in the sense I have described. She speaks of the parties as putting their personal views on hold until they have heard what others have to say. She also says that views that are not challenged are understood to be jointly accepted by the group. This suggests a procedure of suspending judgment as long as anyone expresses dissent. all can accept in the epistemic sense is finding something that all feel forced by the evidence to accept, so the ultimate contributions that the members make to a collective judgment are not under the control of the will. 14 Collective Rationality and Collective Reasoning, chap 5, section II

360 360 Christopher McMahon To be sure, Gilbert adds some twists. I have spoken of collective judgment as if it brings the process of collective reasoning to a conclusion. Gilbert s examples show that a group might want to make a collective judgment to settle an issue, creating something that could then serve as background for subsequent discussion. But if we see an act of joint acceptance as the sort of collective judgment I have described if we see it as involving joint epistemic acceptance a group s settling an issue in this way will have implications that depart from Gilbert s official theory of plural subject phenomena. According to this theory, any subsequent deviation from what has been jointly accepted is open to rebuke. This is not true of the collective judgments I have described. Someone might conclude, as discussion unfolds, that it is necessary to abandon a position already agreed to. But she could not legitimately be rebuked for this, in the way that someone who defects from a collective project to defend a position as the position of the group can be rebuked. Belief is not under the control of the will. To pursue this point further, a collective judgment is formed by a process that involves the suspension of individual judgment until something that all can accept has been identified. If someone subsequently changes her mind, the old collective judgment goes out of existence and a situation is created in which a new one must be formed. This process can be described, following Gilbert, as a process of negotiation. But negotiation of this sort is a matter of the exchange of arguments until something everyone can epistemically accept has been found. It is not as it could be in a case of discretionary acceptance a bargaining process designed to find something to which all can commit their wills. Similarly, we can accommodate Gilbert s observation that the other members of the group would be shocked if one of them denied something that had been previously jointly accepted. One might go along with a view expressed in a conversational context in the sense that one cannot see any reason to object to it. And this could be understood as tacit epistemic acceptance, and a tacit contribution to a collective judgment. As was noted earlier, if one subsequently expresses the view that what was jointly accepted is wrong, one will appear to be inconsistent, which is something that normally elicits an objection. But the objection here has nothing to do with violating an obligation. Similarly, if one is not to appear epistemically incompetent, when one withdraws from a collective judgment, one will have to explain one s change of mind to the group. But this will not be the same thing as asserting that one s personal point of view is different from the view of the group. There could be room for such an action if the group s decision to

361 Two Modes of Collective Belief 361 accept the view in question involved discretionary (will-based) joint acceptance. But when joint epistemic acceptance takes the form of a collective judgment about what a set of reasons supports, the jointly accepted view is the view of every group member. So one member s withdrawal means that the proposition that was previously jointly accepted no longer is. Finally, we should note the significance of the form of collective belief (believing as a body) that involves joint epistemic acceptance for the summative-nonsummative distinction. Gilbert, it will be recalled, rejects summative accounts of collective belief. In her view, belief by the individual members of a group is neither necessary nor sufficient for collective belief. But when joint acceptance is understood as involving collective judgment, in the sense described, individual acceptance is necessary for joint acceptance. So in one respect, joint acceptance in the epistemic sense brings with it a summative view of collective belief. But individual acceptance by all the members of a group is not sufficient for joint epistemic acceptance. Joint epistemic acceptance requires an explicitly collective act on the part of the members of the group, a suspension of individual acceptance by the parties until a proposition that all can accept as supported by the evidence has been identified, at which point the coordinated pattern of individual judgments in which collective judgment consists is brought into existence. The summative view that Gilbert criticizes does not have this feature. It is prepared to speak of a collective belief wherever most members of a group individually accept a proposition. So if we understand collective belief, in the sense that involves epistemic acceptance, as requiring a collective judgment, we will in an important respect be agreeing with Gilbert. We will be saying that there can be no collective belief without a particular kind of joint performance. V. Conclusion For Gilbert, we have collective belief when the members of a group jointly commit themselves to believing something as a body. She glosses this as their committing themselves to make it the case that the group of which they are members relates to a proposition in the same way that an individual who believes it relates to it. I have argued that we must distinguish two different aspects of relating to a proposition as an individual who believes it relates to it. Downstream from the decision to adopt a view as the view of the group, relating to a proposition in the way an individual who believes it relates to it

362 362 Christopher McMahon means making arguments defending the view. A group can mimic this way of relating to a proposition as an individual who believes it relates to it even when the decision to adopt the view as the view of the group has the character of discretionary acceptance. But an individual who believes a proposition will typically have been led to believe it by the evidence. If a group is to mimic this aspect of an individual s relation to a proposition she believes, the decision to adopt the view must take the form of a collective judgment. A collective judgment is something that a group brings into existence as a body, but it entails that all the members of the group believe the view adopted. In this case, we have a joint form of epistemic acceptance. When a group s joint acceptance of a proposition has the character of discretionary acceptance, the personal views of the members of the group can diverge from what the group believes as a body; there can be obligations of joint commitment to maintain the view of the group; and failure to maintain the view of the group is open to rebuke. But there is no place for these features of Gilbert s account of collective belief when a group s joint acceptance of a proposition has the character of epistemic acceptance. To put it another way, a group s displaying any feature as a body requires that its members make certain contributions. In the case of the downstream consequences of the joint acceptance of a proposition, these contributions are actions under the control of the will. Where these actions take the form of the making of arguments defending the proposition in question, there is some basis for ascribing a belief to the group even when the joint acceptance is discretionary. The group is acting as a body in the way an individual who has formed a belief does. It is defending a proposition as true. But when we turn to the process of belief formation, if the group is to conduct itself as an individual believer does, it must judge, as a group, that the evidence warrants the acceptance of the proposition in question. In this case, the contributions of the members will be individual judgments by each that the evidence warrants the acceptance of that proposition. Judgment in this sense entails belief. So when we consider the process of belief formation, we can speak of joint acceptance acceptance as a body only if all the members, as individuals, believe the proposition in question to be supported by the evidence.

363 What Really Divides Gilbert and the Rejctionists? 363 What Really Divides Gilbert and the Rejectionists? K. Brad Wray Abstract Rejectionists argue that collective belief ascriptions are best understood as instances of collective acceptance rather than belief. Margaret Gilbert objects to rejectionist accounts of collective belief statements. She argues that rejectionists rely on a questionable methodology when they inquire into the nature of collective belief ascriptions, and make an erroneous inference when they are led to believe that collectives do not really have beliefs. Consequently, Gilbert claims that collective belief statements are best understood as instances of belief. I critically examine Gilbert s criticisms of rejectionism. I argue that rejectionism is still a viable account of collective belief ascriptions. I also argue that Gilbert s most powerful criticism provides important insight into what really stands between her and the rejectionists. Gilbert and the rejectionists do not yet agree about what background assumptions can be made in developing an account of collective belief ascriptions. I. Introduction Margaret Gilbert (1989; 1994; 2000; 2002) has long defended the view that our common expressions attributing beliefs to collectives, like countries and organizations, ought to be construed non-reductively. She calls her account a plural subject account, for the bearer of such beliefs are plural, rather than individual, subjects. She contrasts her view with summative accounts. Summative accounts of collective belief statements are reductive, for advocates of such accounts construe a statement attributing a belief to a group as a claim to the effect that all, or some significant portion, of the individual members of the group hold the belief in question (see Quinton, 1975). In distinguishing her view from reductive views, Gilbert does not address the issue of whether or not her plural subject account is an account of something appropriately characterized as a form of belief. Many sociologists and philosophers of social science acknowledge the existence of plural subjects, groups of people acting on intentions that are properly understood to be the intentions of the group, rather than the intentions of the individual members of the group. But, not everyone who ac-

364 364 K. Brad Wray knowledges that groups of people are sometimes agents accepts that groups are capable of believing. Some theorists claim that, though some groups have the features required for agency, no group has the capacity for believing. These critics claim that when we attribute beliefs to groups, what the group does is accept the view in question, not believe it. Acceptance, the critics claim, is something quite different from belief, and only individual agents truly believe. 1 Indeed, this is the view I defend in Collective Belief and Acceptance (Wray 2001a). Significantly, unlike those who defend summative accounts of collective belief, those who insist that groups accept but do not believe willingly acknowledge that groups are capable of having intentions not reducible to the intentions of the constitutive members. Gilbert (2002) is not content with these half-hearted allies who only come part of the way with her. Indeed, she has singled three of us out: Raimo Tuomela (1992; 1995; 2000), Antonie Meijers (1999), and myself (Wray 2001a). She has labelled us rejectionists. We rejectionists claim that when people attribute beliefs to groups they are in error, for in reality groups do not believe. Gilbert is not alone in raising concerns against the rejectionists strategy (see Tollefsen 2002, and Preyer 2003). I have two aims in this paper. First, I want to defend a rejectionist account of collective belief ascriptions by showing that such an account is more plausible than Gilbert suggests. Second, I want to clarify what really divides Gilbert and the rejectionists. I will begin by reviewing the various criticisms Gilbert levels against rejectionism. I will then reply to the criticisms, and, in the process, refine and defend the rejectionists account of collective belief ascriptions. I argue that rejectionism is still a viable account of collective belief ascriptions, contrary to what Gilbert claims. Importantly, Gilbert s criticisms help us see what a viable rejectionist account must look like. Her most powerful criticism, I argue, can aid us in seeing what ultimately stands between her and the rejectionists. I argue that Gilbert and the rejectionists are unlikely to reach an agreement about what is the best account of collective belief ascriptions until they resolve the methodological issue of determining what background assumptions we should make in developing an account of collective belief ascriptions. 1 The most thorough analyses of the distinction between belief and acceptance are to be found in Cohen (1992), Van Fraassen (1980), and Stalnaker (1984).

365 What Really Divides Gilbert and the Rejectionists? 365 II. Gilbert s Criticisms In a recent paper titled Belief and Acceptance as Features of Groups, Gilbert (2002) raises three criticisms against rejectionists. The first pertains to the methodology commonly employed by rejectionists. Gilbert accuses rejectionists of employing a questionable, and question begging methodology in their efforts to show that groups do not believe. Gilbert s second criticism is that rejectionists make erroneous inferences about the nature of groups based on the characteristics of the individuals who constitute the groups. Gilbert s third criticism pertains to an alleged difference between the role and impact of evidence on collective and individual belief. Let us now review the criticisms in detail. First, Gilbert suggests that the rejectionists beg the question by the methods they employ in arguing for the view that groups accept but do not believe. In particular, she objects to the fact that rejectionists make their case by taking for granted that the proper understanding of the nature of belief is to be gleaned from examining the features unique to individuals beliefs (2002, 47). That is, when rejectionists argue that collectives are incapable of belief, they do so by constructing a list of features common to cases of individual belief, and then scrutinize cases in which beliefs are ascribed to collectives to see if the required features are present. Rejectionists are led to conclude that collectives do not really believe because, on examination of cases where a collective is alleged to have a belief, a key feature present in cases of individual belief is found to be lacking. This, Gilbert claims, betrays the rejectionists deep, and questionable, commitment to individualism. Their methodology is thus prejudiced against the collective case. Gilbert proposes an alternative method. She suggests that our understanding of belief might well be derived from an examination of cases of both individual and collective belief ascriptions. Gilbert believes that had the rejectionists proceeded with this method they would not be led to deny beliefs to plural subjects. Gilbert appeals to Plato as a model of someone who endorses her proposed method (Gilbert 2002, 48). In Republic, Socrates aim is to understand the nature of justice. In an effort to understand the nature of justice in the individual person, Socrates suggests that they examine justice as it is realized in something larger, in the city-state (Plato 1974, 368e-369a). Socrates reasons as follows. Given that the state is so much larger than the individual, it should be much easier to grasp what justice is there than in something as small as the individual person. What Gilbert likes about Plato s method is that it gives equal weight to the individual and the state in formu-

366 366 K. Brad Wray lating a general account of justice. Gilbert suggests that if rejectionists began their theorizing in a manner similar to that employed by Plato, they would have reached a very different conclusion. They would first develop an account of belief that is amenable to ascriptions to collectives and individuals alike. Given the suspicions Gilbert has about the rejectionists methodology, she is inclined to reject any conclusions reached employing it. Second, Gilbert believes that rejectionists are often guilty of committing the fallacy of composition when they inadvertently shift levels between individuals and collectives in their arguments for rejectionism (2002, 59, 61). In particular, she claims that when rejectionists argue that ascriptions of collective belief involve acceptance, they are inadvertently referring to the attitude of the individuals who constitute the collective, not the attitude of the collective. In analyzing the cognitive attitudes of the collective, rejectionists often start from conceptions of both acceptance and belief as the terms are applied to individuals. They then proceed to show that the attitude of the collective exemplifies the traits common to instances of acceptance rather than instances of belief. Gilbert s concern is that rejectionists inadvertently slide between an analysis of the traits of the group, and an analysis of the traits of the individual members. Finding that the individuals in the group exhibit features of acceptance rather than belief in cases where a belief is ascribed to a group, rejectionists then conclude that the group does not believe. Gilbert grants that this mistake is easy to make. But, she insists that if one hopes to show that collectives only accept views, then one must show that the features of the collective s attitude are characteristic of acceptance, and not belief. Gilbert believes that rejectionists have failed to do this. Consequently, she has yet to be shown that collectives cannot believe. A brief discussion of two of Gilbert s examples will help make this criticism more salient. First, as Gilbert notes, Meijers argues that we as a group may believe... a proposition that no one takes to be true individually (65) (58). Meijers takes this as evidence that the group does not believe the proposition in question. He insists that the cognitive attitude of the group is more aptly described as an instance of acceptance, rather than believe. In reply to this argument, Gilbert constructs an example where a committee comes to believe something that no individual member of the committee believes. But, as she notes, the fact that no single member of the group individually holds the belief is irrelevant to the stance of the collective they constitute, namely the committee itself (59). As Gilbert explains, it is a mistake to look at... the personal situation of the members of the committee when what is at issue

367 What Really Divides Gilbert and the Rejcetionists? 367 is the stance of the committee as a whole (59). According to Gilbert, this is a common pattern of reasoning employed by rejectionists. Let us now consider a second example. As Gilbert explains, both Meijers and Wray maintain that when an agent accepts some view they accept it voluntarily, whereas when an agent believes something it is believed involuntary. According to Gilbert, it is because an act of will underlies the creation of an alleged collective belief that Meijers and Wray are led to claim that collective beliefs are really instances of acceptance. But, Gilbert claims that both Meijers and Wray erroneously argue from the fact that the wills of the individual group members are involved in the production of a collective belief to the conclusion that collective beliefs... must be willed by the collective in question (61). She believes that such an inference is unjustified. Showing that the individuals involved in a collective must act voluntarily in order to become part of a collective does not demonstrate that the collective only accepts the views that are ascribed to it. Voluntarism at the level of individuals is consistent with involuntarism at the level of the group. Third, Gilbert claims that rejectionists are mistaken to think that plural subjects, unlike individuals, are not responsive to evidence (2002, 53-54). That is, rejectionists claim that when it is alleged that a plural subject believes something the belief is adopted for prudential reasons, and plural subjects are always moved by such reasons, rather than evidence for or against the truth of the belief. In contrast, rejectionists claim that individuals acquire and change beliefs on the basis of evidence for or against the beliefs. Gilbert believes that rejectionists are mistaken about this on two accounts. First, she claims that as a matter of fact individuals sometimes acquire beliefs for prudential reasons. Gilbert constructs an example of a man who wonders if it is in his interests to believe that his wife is unfaithful. He may decide that it is not, since the belief is making him miserable (56). Hence, the agent can choose to believe what serves his interests. Second, she claims that collectives are often responsive to evidence. Gilbert gives an example of an alliance of nations that, on a consideration of the evidence, decides to believe that bombing a neighbouring country is not a suitable means to achieve their goal. As she notes, the plural subject reaches this belief on the basis of a consideration of epistemic rather than practical reasons, contrary to what seems to be implied by rejectionism (55). Consequently, Gilbert believes that the rejectionists efforts to draw a distinction between the types of beliefs that individuals have and the types of beliefs that collectives have on the basis of their responsiveness to evidence fails. Plural subjects and individuals are no different in this respect. Conse-

368 368 K. Brad Wray quently, there seems to be no principled reason to regard the alleged beliefs of collectives as any less beliefs than the beliefs of individuals. Given these three criticisms, Gilbert claims that rejectionists fail to prove that plural subjects do not believe. Indeed, as far as she is concerned, these criticisms raise serious doubts about the plausibility of rejectionist accounts of collective belief ascriptions. III. Collective Beliefs and the Role of Evidence In the remainder of this paper, I want to address Gilbert s criticisms and assess the extent to which they require rejectionists to give up their attack on collective belief. I will address the three criticisms in reverse order. Gilbert s third criticism, I argue, is misguided. Her second criticism, I argue, is legitimate but applies to Meijers rejectionist account of collective belief acsriptions, not my own rejectionist account. And, Gilbert s first criticism, I argue, raises an important issue that neither she nor the rejectionists have adequately addressed. Gilbert s third criticism against rejectionism is that rejectionists are mistaken when they claim that responsiveness to evidence distinguishes belief from acceptance. Consequently, Gilbert argues, responsiveness to evidence cannot function as a criterion for distinguishing belief from acceptance, and classifying all alleged collective beliefs as instances of acceptance. In an effort to prove this, Gilbert argues that (i) there are instances of belief that are acquired for pragmatic reasons, rather than adopted in light of the evidence, and (ii) there are cases where a view is accepted, not for pragmatic reasons, but rather in light of evidence of its truth. This criticism, I believe, is based on a misreading of the rejectionists position. Gilbert ascribes stronger claims to the rejectionists than we either make or need commit ourselves to. As Gilbert implies, both Meijers and I recognize that sometimes individuals come to believe something even when it is contrary to the available evidence. This is David Hume s point in his discussion of the testimony of miracles in An Enquiry Concerning Human Understanding. As Hume (1977) explains, it is a miracle that so many people persist in believing something that is so contrary to experience (90). Clearly, in such cases, there is something besides evidence that is influencing what a person believes. Hence, Gilbert is correct to insist that sometimes our beliefs are caused by non-evidential considerations.

369 What Really Divides Gilbert and the Rejectionists? 369 But, one can grant that people s beliefs are sometimes impervious to evidence and still maintain that people are not at liberty to choose what they believe, or that they are not capable of believing something merely for prudential rather than evidential reasons. To illustrate this, consider Gilbert s example of the man considering whether or not to believe that his wife is unfaithful. Presented with evidence, either robust or scanty, he is not at liberty to choose what to believe. Granted, as Hume suggests, even if the man were faced with insurmountable evidence against a particular belief, he may still persist in believing it. Initially, this may strike one as paradoxical. But the appearance of paradox vanishes when we realize that evidential considerations are not the only factors that play a causal role in belief-formation. This, though, does not prove that we have the capacity to believe what we choose. The non-evidential considerations that cause us to believe something may be as much out of our control as the impact that evidential considerations have on us. Hence, I do not think that Gilbert has made a convincing case for her claim that people can choose what they believe on the basis of prudential considerations. Insofar as the man in Gilbert s example chooses what to believe, he is best described as accepting rather than believing the view he chooses to adopt. Also, as Gilbert suggests, both Meijers and I admit that sometimes the evidence pertaining to the truth or falsity of a view will determine what view a plural subject or individual accepts. But this should not surprise us. After all, if the purpose of accepting a view is to advance some goal, something both Meijers and I claim, then knowing that a view is true will often have some impact on whether or not it is a view worth accepting. If accepting a particular view is to serve some purpose, then such considerations are apt to matter. Admitting this, though, does not entail that collective beliefs are more like cases of belief than acceptance. The important difference between acceptance and belief is that agents accept views in light of their goals, whereas beliefs are not acquired in light of goals. In fact, I believe that plural subjects are constituted by their goals. Without a goal or set of goals, there is no plural subject. A committee, an army, or a nation cannot exist without some goal. Lacking a goal, the various people who may come to form such a group are merely an aggregate of individual agents. Individual agents are unlike plural subjects in this respect. An individual agent can cease to have goals and yet still exist. Consequently, individual agents have capacities that plural subjects do not. Like plural subjects, individual agents have the capacity to accept things in light of their goals, but they also have the capacity to believe things. When an individual

370 370 K. Brad Wray comes to believe something, they are compelled to believe it whether or not it serves their goals. But, when either an individual or a plural subject decides to accept something, they do so in light of their goals. The importance of distinguishing between belief and acceptance lies in the fact that belief and acceptance can diverge. That is, an agent can accept things they do not believe. Indeed, one may often accept what one believes, and in turn believe what one accepts. But the fact that one has this capacity to accept what one does not believe suggests that some factor must influence acceptance that does not influence belief. I am suggesting that the distinguishing factor is the agent s goals. Given the differences between belief and acceptance, both Meijers and myself claim that collective belief ascriptions are more like instances of acceptance than belief. Consequently, both of us conclude that collectives do not believe. IV. Individual and Group Properties Let us now consider Gilbert s second criticism of rejectionism. Gilbert claims that rejectionists inadvertently ascribe qualities that are properly ascribed to the individual members of groups to the groups themselves. This, she argues, leads the rejectionists to mistakenly claim that groups do not have beliefs because the constitutive members of the groups have properties that are characteristic of agents when they accept views rather than believe them. As Gilbert notes, it is often difficult to keep straight which properties belong to which agents when one is discussing plural subjects, agents who are themselves composed of a number of individual agents. Further, as Gilbert implies, at least one rejectionist has mistakenly attributed properties of the individual members of a group to the group itself. Though I agree with Gilbert that Meijers makes this mistake, Gilbert s criticism does not apply to my own rejectionist account of collective belief ascriptions. Consider Meijers account. He notes that one of the features that distinguishes acceptance from belief is the fact that the former but not the latter is voluntary. That is, we can decide what we accept but not what we believe. Meijers (1999) then claims that the sorts of phenomena that Gilbert regards as instances of collective beliefs are voluntary because they depend on the will of the members of the group (70). Now Meijers may be correct in claiming that collective beliefs depend on the will of the members of the group,

371 What Really Divides Gilbert and the Rejectionists? 371 but one can grant this and still maintain that plural subjects believe rather than accept. After all, in appealing to the voluntary nature of acceptance to show that plural subjects accept rather than believe, one would need to show that the view adopted by the group is voluntarily chosen by the group, qua group. Meijers has yet to prove this. The group could act involuntarily, and thus believe the view attributed to it, even if the constitutive individuals accept the view of the group as the view of the group. Voluntarism at the level of the individual does not entail voluntarism at the level of the group. In this respect, Gilbert s criticism is well grounded. Meijers is guilty of committing the fallacy of composition. In my own analysis of collective acceptance, I also claim that acceptance is typically voluntary whereas belief is typically involuntary (Wray 2001a). But, I do not build my case against the claim that plural subjects do not believe on this fact. Unfortunately, my critics have failed to see this. Both Gilbert (2002) and Tollefsen (2002) suggest that I ground the claim that collective belief ascriptions really refer to instances of collective acceptance on the following two claims: (I) that belief is involuntary, and (II) that instances of collective acceptance are voluntary. The defining characteristic of acceptance that I emphasize in my argument for rejectionism is the fact that what an agent accepts is typically tailored to specific ends (see Wray 2001a, 325). This is true in both the individual and the collective cases. For example, an agent accepts a view for the sake of an argument, or in order to provide a shared background from which to resolve a problem with another agent. Belief, I argue, is generally not like this. That is, the beliefs that we hold we hold with no goal in mind. I argue that a collective belief ascription refers to a situation in which a plural subject adopts a view in light of their goals. Consequently, I argue that the cognitive attitude of the plural subject is most accurately described as acceptance, not belief. Hence, my analysis avoids the type of erroneous slip that Gilbert draws to our attention. That is, unlike Meijers, I do not commit the fallacy of composition. Consequently, I believe that a rejectionist account of collective belief statements can be developed without falling prey to Gilbert s second criticism. Indeed, my own account is just such an account. V. Methodology and Social Inquiry In this section, I want to examine Gilbert s criticism of the rejectionists methodology. I believe that this criticism is her most important criticism in

372 372 K. Brad Wray advancing our understanding of collective belief ascriptions. Her criticism, though, is not as damaging to rejectionism as she claims. Nonetheless, it does draw attention to what both parties, Gilbert and the rejectionists, must do in order to move forward in the debate about the proper understanding of collective belief ascriptions. At the outset I would like to say that Gilbert is absolutely right to emphasize the importance of reflecting on methodology whenever we embark upon an inquiry. As Helen Longino (1990) and other feminist philosophers of science have noted, in scrutinizing scientists methodological assumptions we frequently unearth dubious background assumptions that may favour one of the numerous competing hypotheses. For example, in physical anthropology Sally Slocum (1975) questioned the assumption that the males of our species are causally responsible for the evolution of our species when she proposed the woman-the-gatherer hypothesis to challenge the then-dominant manthe-hunter hypothesis. The man-the-hunter hypothesis suggests that the males of our species developed tool use and language use as means to facilitate big game hunting. The women-the-gatherer hypothesis suggests that the females of our species developed language use and tool use to facilitate the effective gathering of food. By developing this alternative hypothesis, Slocum demonstrates how prevailing sexist assumptions have inadvertently shaped science. Both of the competing hypotheses were equally able to account for the available data. A key factor that explains why the man-the-hunter hypothesis was developed first was its fit with the background assumptions that physical anthropologists were uncritically bringing to their research. Living and researching in a sexist society with a gendered division of labor, many anthropologists uncritically assumed that our early predecessors lived in a similar society. Their background assumptions thus prevented them from looking at the world objectively. Although all inquiry is necessarily mediated by background assumptions, not all inquiries are equally affected by biases (see Longino 1990; Wray 1999; Wray 2001b). Background assumptions become problematic when they bias inquirers toward accepting one hypothesis rather than a competitor hypothesis. Consequently, we should aim to ensure that our background assumptions are neutral with respect to the competing hypotheses under study. Hence, in analyzing the debate between Gilbert and the rejectionists, we will want to determine whether or not either side is guilty of working from background assumptions that beg the question in favor of one hypothesis. Thus, we want to ensure that our background assumptions are neutral with respect

373 What Really Divides Gilbert and the Rejectionists? 373 to the hypotheses that (I) groups are not capable of belief, and (II) groups are capable of belief. Gilbert argues that rejectionists rely on background assumptions that are not neutral with respect to the competing hypotheses, but I think that she is mistaken about this. Indeed, Gilbert is correct in claiming that Meijers, Tuomela, and myself all start our analysis of collective belief ascriptions with an examination of the characteristics of individual belief. But this in itself does not make us guilty of the charge of begging the question. When we rejectionists argue that alleged collective beliefs are not instances of belief, we do not merely stipulate a definition of belief, and then discover that it supports our conclusion. Rather, we appeal to what others have said about belief and acceptance (see Wray 2001a, ). Proceeding as we do in our inquiry into collective belief ascriptions, it is at least possible that we could have discovered that collectives do have beliefs. Consequently, contrary to what Gilbert claims, I think it is fair to say that rejectionists do not beg the question. Hence, Gilbert needs to do more than she has done so far in order to show that our methods are question begging. Let us now reconsider Gilbert s own proposed method. Ironically, I believe that it is her background assumptions that are not neutral with respect to the competing hypotheses. She starts from the observation that we ascribe beliefs to both groups and individuals, and then asks what are the features common to these cases such that we are led to make such ascriptions. Given her methodology, I would be surprised if one could possibly reach the conclusion that groups do not have beliefs. Her proposed method thus lacks the neutrality we seek. I believe that the source of the disagreement between Gilbert and the rejectionists is a particular unquestioned, though questionable, background assumption that she makes but her rejectionist critics do not make. Gilbert and rejectionists are in agreement that groups of people are, like individuals, capable of agency. But, Gilbert seems to assume that all agents are alike in more respects than the rejectionists assume. Consequently, she seems to uncritically assume that, like individual agents, plural subjects will have the capacity for belief. My concern is that she does not take seriously the differences between types of agents, collective and individual. Aristotle s (1998) discussion of the differences between individual persons, households and city-states will help to illustrate the differences between collective and individual agents. I assume that Gilbert would regard all of these as agents, and the latter two as plural subjects. Aristotle argues that the different types of entities differ in their qualities. Specifically, he argues that a

374 374 K. Brad Wray household is more self-sufficient than a single person, and a city-state than a household (1261b, 11-13). In contrast, a person is more unified than a household, and a household more unified than a city-state. Because of these differences in qualities the various entities will have different capabilities. Consequently, there are some things that a city-state can do that an individual person cannot do. Similarly, there are some things that an individual person can do that a city-state cannot do. Clearly, there are profound differences between individual and plural subjects. Indeed, one of the attractions to being part of a plural subject is the capabilities it affords us that we would not have acting alone, as an individual agent (see Wray 2001a, 322). Hence, I believe that Gilbert has yet to demonstrate that plural subjects have the capabilities that individuals have. In particular, she has yet to demonstrate that plural subjects are capable of belief. I suspect that Gilbert may still object to the method employed by rejectionists. She may claim that it unduly privileges individual belief ascriptions with the result that collective belief ascriptions do not get a fair hearing. Unfortunately, my analysis will have to leave the debate about the proper understanding of collective belief ascriptions unresolved. On the one hand, Gilbert s preferred method has been shown to be biased in favor of her own hypothesis. On the other hand, though I have argued that Gilbert s criticism of the rejectionists method is ungrounded, we have yet to reach agreement about what background assumptions should inform an inquiry into the nature of collective belief ascriptions. The question of which account is superior, Gilbert s own or some version of rejectionism, is not apt to be answered until the methodological issue is resolved. VI. Concluding Remarks Before closing it is worth briefly considering Gilbert s plea to common usage in defence of her thesis that plural subjects really do believe. Gilbert notes that in light of her responses to the rejectionists criticisms, it seems that alleged collective beliefs behave much like beliefs in the individual case, and certainly more like such beliefs than rejectionists have supposed (67). Consequently, she concludes that it is hard to see why the learned should not speak with the vulgar, and go ahead and call them beliefs (67). I have two remarks to make in response to Gilbert s suggestion that we should speak with the vulgar. First, I am not yet convinced that alleged collective beliefs do behave much like beliefs in the individual case. Indeed, my

375 What Really Divides Gilbert and the Rejectionists? 375 responses to Gilbert s criticisms discussed above are intended to show that this has yet to be proved. Second, given that belief is a theoretical term, appeals to common usage seem out of place. By theoretical I mean to suggest that it refers to a type of entity that is unobservable. Indeed, it is this observation that led Paul Churchland (1981) to endorse eliminative materialism. He claims that theories that employ the concept belief are unsatisfactory. We rejectionists are not as sceptical as Churchland about the explanatory power of the concept belief. It is a powerful concept for explaining individuals actions and behavior. But we are also careful not to overextend its use. And this is precisely what we believe Gilbert has done. Given that science is so often at odds with common sense, especially when scientists refer to unobservable phenomena, it seems that we should not be surprised to discover that a scientific understanding of collective belief ascriptions reveals that these phenomena are not as laypeople commonly think. Indeed, this is the core of the rejectionists claim. We are not satisfied with treating such phenomena as beliefs. When it comes to understanding the way plural subjects behave, we believe that there is much to be gained by drawing the distinction between belief and acceptance and recognizing that ascriptions of collective belief are really best understood as cases of acceptance. Though Gilbert has not yet proved that the rejectionists interpretation of collective belief ascriptions is untenable, she has drawn our attention to a type of error that we are at risk of making, and, more importantly, she has focussed all parties attention on the important methodological issue that seems to stand in our way of reaching a consensus. The two parties need to determine what the background assumptions are that should inform an inquiry into collective belief ascriptions. I would like to thank Margaret Gilbert, Lori Nash, Gerhard Preyer, Raimo Tuomela, and Lee Baker for valuable feedback on earlier drafts. References Aristotle Politics, translated by C. D. C. Reeve. Indianapolis: Hackett Publishing Company. Cohen, L. J An Essay on Belief and Acceptance. Oxford: Clarendon Press. Churchland, P Eliminative Materialism and the Propositional Attitudes, Journal of Philosophy, LXXVIII: 2,

376 376 K. Brad Wray Gilbert, M On Social Facts. Princeton: Princeton University Press Remarks on Collective Belief, in F. Schmitt, ed., Socializing Epistemology: The Social Dimensions of Knowledge. Lanham: Rowman and Littlefield Publishers, Inc., Sociality and Responsibility: New Essays on Plural Subject Theory. Lanham: Rowman and Littlefield Belief and Acceptance as Features of Groups, Protosociology, Volume 16, Hume, D Enquiry Concerning Human Understanding. Indianapolis: Hackett Publishing Company. Longino, H Science as Social Knowledge. Princeton: Princeton University Press. Meijers, A Believing and Accepting as a Group, in A Meijers, ed., Belief, Cognition and the Will. Tilburg: Tilburg University Press, Plato Republic. Tranlated by G.M.A. Grube. Indianapolis: Hackett Publishing Company. Preyer, G What is Wrong With Rejectionists?, in Interpretation, Sprache und das Soziale: Philosophische Artikel (Interpretaion, Language and the Social: Philosophical Articles). Frankfurt am Main. Quinton, A Social Objects, Proceedings of the Aristotelian Society, 76: Slocum, S Women the Gatherer: Male Bias in Anthropology, in R. Reiter, (ed.), Toward an Anthropology of Women. New York: Monthly Review Press. Stalnaker, R Inquiry. Cambridge: MIT Press. Tollefsen, D Challenging Epistemic Individualism, Protosociology, Volume 16, Tuomela, R Group Beliefs, Synthese, 91: The Importance of Us. Stanford: Stanford University Press Belief Versus Acceptance, Philosophical Explorations, 2: Van Fraassen, B The Scientific Image. Oxford: Clarendon Press. Wray, K. B A Defence of Longino s Social Epistemology, Philosophy of Science, Supplement to Vol. 66: 3, S538-S a. Collective Belief and Acceptance, Synthese, 129: b. Science, Biases, and the Threat of Global Pessimism, Philosophy of Science, Supplement to Vol. 68: 3, S467-S478.

377 Why Accept Collective Beliefs? Reply to Gilbert 377 Why Accept Collective Beliefs? Reply to Gilbert Anthonie Meijers Abstract Margaret Gilbert has recently argued in ProtoSociology against what she called my rejectionist s view according to which (i) we have to make a distinction between the intentional states of believing and accepting and (ii) genuine group beliefs, i.e. group beliefs that cannot be reduced to the beliefs of the individual members of a group, should be understood in terms of the acceptance of a view rather than of beliefs proper. In this reply I discuss Gilbert s objections. 1. What Rejectionists Believe In my paper in ProtoSociology (Meijers 2002) I defended a pluralist view according to which there is a whole spectrum of belief-like attitudes that we can attribute to groups. The extremes are formed by what I called the weak opinion poll conception of collective beliefs, on the one hand, and the strong agreement-based conception on the other. The opinion poll conception is reductionistic in that it conceives collective beliefs in terms of individual beliefs plus a certain distribution of these beliefs among the members of the group. This conception applies, for example, to the Dutch people believing that the current change of climate is due to the greenhouse effect. On the other side of the spectrum is the agreement-based conception of collective beliefs, according to which groups do have genuine and irreducible belieflike states of a certain type. These states are attributed to groups conceived as structured wholes (as opposed to aggregates). An example would be the Dutch cabinet believing that the current change of climate is due to the greenhouse effect. The agreement-based conception is the more interesting one, since it raises the question how this belief-like state is to be understood. Should it be conceptualised in terms of belief proper or in terms of something else? I have argued that these states are to be understood in terms of the acceptance of a view rather than belief proper. Several authors have argued for a distinction between belief and acceptance, including Michael Bratman (1993), Pascal Engel (1998) and Raimo

378 378 Anthonie Meijers Tuomela (2000). They do not make this distinction in exactly the same way, but for our discussion the following characterization will suffice. Beliefs aim at truth, are shaped by evidence, and are subject to the ideal of integration. Acceptances, on the other hand, aim at utility or success, are shaped by evidence and practical concerns, and are subject to coherent policies. The crucial element in the context of our present discussion is whether or not nonepistemic considerations enter into the grounds for having the belief-like state. To use our earlier example: there may not be conclusive evidence for the Dutch cabinet to really believe that the current change of climate is due to the greenhouse effect, but it may accept this proposition for prudential reasons in its practical reasoning; for example, when the cabinet reasons about its energy conservation policy it may choose to play on the safe side when our climate is concerned and accept the proposition. Practical concerns prevail here in the absence of conclusive epistemic evidence. Note that this is not to say that epistemic evidence does not play a role when accepting a view. The arguments for the view that the agreement-based conception of collective beliefs should be understood in terms of the acceptance of a view rather than of belief proper are, among other things, based on the following two observations. Firstly, beliefs of this type involve non-epistemic commitments among the members of the group. If the Dutch cabinet believes that the current change of climate is due to the greenhouse effect, its ministers will have established this view in a certain procedure and are subsequently committed to this view. They may and probably will correct each other if they do not live up to the view. This may occur even in the face of changing epistemic evidence. Secondly, the content of collective beliefs may be different from the beliefs held by the individual members of the group. This happens when the ministers of the cabinet compromised on the view they take together, where no one holds this view individually. These two characteristics are incomprehensible on the conception of belief proper, which is guided by truth and evidence. In the example above among the reasons for the cabinet to believe the proposition that the current change of climate is due to the greenhouse effect, that will be non-epistemic ones (such as prudence, pragmatism, practical considerations, group dynamics, etc.). This makes this collective belief an acceptance rather than a belief proper. Now suppose that a group would decide only to allow epistemic reasons to enter in their deliberations about their collective beliefs. Truth and evidence would be included, practical considerations, prudence and the rest would be excluded. There would thus be no room for negotiation, interindividual

379 Why Accept Collective Beliefs? Reply to Gilbert 379 commitments, or a radical asymmetry between the group beliefs and the individual beliefs. Such a group belief would be analysable in terms of individual beliefs plus a certain distribution of these beliefs among the members of the group, possibly combined with common knowledge about these beliefs. If only epistemic reasons for believing are taken into account, it is impossible that there is a difference in content between what I as an individual believe and what I believe as a member of the group, for epistemic access is not role bound. Interindividual commitments would also not be involved in such a collective belief, for they brings into the grounds for believing non-epistemic considerations, which the example explicitly intends to exclude. A collective belief of this type would thus be analysable in terms of the weak opinion poll conception mentioned before. 2. Gilbert s Criticism Margaret Gilbert defends a so-called plural subject account, according to which the members of a population, P, collectively believe that p if and only if they are jointly committed to believe that p as a body (Gilbert, 42). In her account the commitments among the members of a group play a central role. These members are not required to believe what the group believes, or to act as-if they believe so. They are to act as mouthpiece of the group taken as a body (Gilbert, 45). Gilbert s objections to the rejectionist position are the following. The Question of Method The proper way to proceed in a philosophical analysis is, in Gilberts view, to take belief attributions of all kinds seriously and to develop a general notion of belief on that basis. The analysis should not be merely stipulative, but responsive to the everyday use of the relevant terms (Gilbert 48). Given a general account, we can then investigate the similarities and differences between the various attributions of beliefs. Such a well-founded account of belief in general will take collective belief statements seriously (Gilbert, 47). It is not prejudiced towards the individual case. This is not, however, the way philosophers normally proceed. They beg the question in that they are methodologically biased. They are reluctant to take collective belief attributions seriously because of their prior view that indi-

380 380 Anthonie Meijers vidual beliefs are the paradigm case. Rejectionists in particular start with a set of features that they take to apply to the individual case and conclude subsequently that collective beliefs cannot be beliefs since they lack some of these properties. The right approach would be, according to Gilbert, to conclude that these features might not be essential features of beliefs as such (Gilbert, 49). In my view, the issue here should certainly not be about the use of terms, which is rather uninteresting, but about the concepts involved. Gilbert seems to think that it is crucially important whether we proceed in our analysis in a top-down or a bottom-up way. She favours a top-down approach where we start with a more general notion of belief and then analyse beliefs of a more specific kind. To me both ways are equally valid and should have the same result provided the analysis is carried through completely. The relevant differences and similarities between the concepts of belief that apply to individual agents and collective agents (excluding for the moment animals and artefacts) will come out no matter how we proceed. If there is a bias involved it is Gilbert s bias for an essentialist view on the relation between the various concepts of belief, where there is a common core of essential features that apply to all notions of belief. In her methodology she advises us first to look for the more general, necessary features of belief and proceed from there when analysing individual and collective belief attributions. Her paradigm case of Plato s analysis of justice, which covers both the individual agent and society, clearly shows her essentialist orientation (Gilbert 48). This approach excludes from the outset a family resemblance view on the relation between the various concepts of belief. However, I don t think the issue is of great importance. Once the analysis has been carried out in depth the right way to look at the relations between these concepts will emerge. In the background of Gilbert s methodological argument with the rejectionists seems to be the fear that collective cognitive states are not taken seriously enough if we start from an individualistic concept of belief. We then tend to take statements about collective beliefs only metaphorically, as façons de parler (Gilbert 47), and as something that cannot be taken at face value. If this observation is right Gilbert either makes a straw man of the rejectionist s position, or has another opponent. Rejectionists do not question whether we can attribute cognitive intentional states to groups. Their concern is the nature of the state attributed. Her main target here seems to be those who defend methodological or metaphysical individualism. They claim that in the explanation of human behaviour or in our ontology we can only allow for intentional states that are features of individuals, not of collectives.

381 Why Accept Collective Beliefs? Reply to Gilbert 381 The Notion of Acceptance Before I turn to Gilbert s other objections, I want to address a misunderstanding about to the notion of acceptance. Gilbert takes as a canonical example of acceptance the case of assuming something for the sake of the argument or for present purposes (Gilbert 38, 66). This is certainly not the conception of acceptance I have been defending and which I have outlined above. Acceptance in Gilbert s notion is very close to entertaining a thought. Acceptance in my sense is much more committal. It is a notion that belongs to the domain of practical reasoning and action. What we are prepared to accept as premise in our practical reasoning has real-life consequences. It guides our actions in an important ways. For example, whether or not the cabinet accepts the proposition that the current change of climate is due to the greenhouse effect will make a big difference in its policies, regulations, and investments. Gilbert s weaker notion of acceptance does not have such worldly consequences. This may also explain why Gilbert opposes the rejectionist s view that collective cognitive states are to be understood in terms of the acceptance of a view rather than of belief proper. Given her own weak conception of acceptance, the collective acceptance of a view does not have much explanatory power indeed when it comes to the explanation of the actions of groups. Collective beliefs, on the other hand, are supposed to have this power. Belief and Context Gilbert s subsequently discusses the role of the context for understanding the difference between belief and acceptance. She takes issue with my view that collective beliefs are context bound in ways that individual beliefs are not and that they should therefore be understood as acceptances rather than beliefs. She gives examples of collective beliefs that are not context bound and examples of individual beliefs that are context bound, thus apparently undermining the difference claimed by the rejectionist to exist between the two. Gilbert is certainly right in emphasizing that there is a trivial sense in which beliefs are context bound, whether individual or collective. They depend on the evidence available in a given situation and nobody will deny this. If the evidence changes over time we will change our beliefs consequently. She is also right in that collective beliefs are not tight to a specific context in a sense she explains by giving the following example. It is one where a person accepts

382 382 Anthonie Meijers a proposition for the purpose of some argument (Gilbert 52). This case of what Gilbert calls intentional binding to a specific context shows such a tight connection between context and acceptance that it does not fit the notion of acceptance I have outlined before and thus does not apply in my view to collective beliefs in the agreement based sense I am analysing here. But more interestingly, there are at least two senses in which collective beliefs are context bound and individual beliefs are not and which Gilbert ignores in her discussion. Firstly, in order for a belief to be a collective belief, there has to be a shared understanding of the content of such a belief among the members of the group. This cannot be taken for granted in many cases. For example, a proposition such as the current change of climate is due to the greenhouse effect has many terms that need clarification. The meaning of the terms is often negotiated among the members of a group, which can be witnessed at international conferences. The resulting collective belief is then context bound in the sense that the content of the belief is fixed relative to a group. Relative to another group the belief can and often will have a different content. In the individual case there is no such social process negotiating the content of the beliefs one holds. Secondly, collective beliefs are context bound in the sense that they are role bound. If a selection committee believes that a particular candidate is the best candidate for the job, it may be possible that none of its members believes this individually. They could have a different candidate as their first choice. However, in their role as members of the committee they believe the selected candidate to be the best for the job and they will act as mouthpiece of the group, to use Gilbert s phrase. To give another example, in one s role as minister of the Dutch cabinet one may believe that that the current change of climate is due to the greenhouse effect, while not having views on this individually. These examples show that collective beliefs commit the members of a group to certain views, which they are supposed to hold in their role as member of the group. Individual beliefs do not have this role dependence. An individual believes that p regardless of the role taken. Belief and Evidence I now turn to Gilbert s criticism of another disanalogy that rejectionists claim to exist between collective beliefs of the strong agreement-based type and individual beliefs. Many theories of belief agree on the idea that beliefs have a mind-to-world direction of fit, that they are to be assessed in terms of their

383 Why Accept Collective Beliefs? Reply to Gilbert 383 truth and falsehood and that they are shaped by the available evidence. Collective beliefs, rejectionists argue, are different in these respects in that they involve non-epistemic elements as well. To give an example, let us assume that there is no conclusive evidence for the view that the current change in climate is due to the greenhouse effect. The Dutch cabinet may nevertheless believe this proposition in the light of the consequences that may follow if this belief turns out to be true, and will use this proposition in its deliberations about future actions. Many cases where the precautionary principle is used are like that. Gilbert s strategy here is to argue that (i) collective beliefs may be given up for epistemic reasons and (ii) individual beliefs may be given up for nonepistemic reasons. She concludes that the rejectionists fail to show that there is a major disanalogy here. With respect to her first point, it is simply a misunderstanding of the rejectionist s position to suppose that according to this view epistemic evidence does not play a role in the formation of collective beliefs, a point I have made several times in my earlier paper and in this paper. The crucial point, however, is that collective beliefs of the strong agreement-based type are never changed or given up solely for epistemic reasons. The commitments that play a role in the constitution of a collective belief also play a role in giving up or changing a collective belief-like state. When a group is jointly committed to believe p as a body, to use Gilbert s phrase, the members of the group need to follow a group s procedure or have such a procedure in place in order to change this belief. Such a procedure is similar to the one needed to arrive at the original s group s belief and is never purely epistemic. Her second point that individual beliefs may also be given up by a person for non-epistemic reasons is highly debatable. Gilbert gives the example of a man called Joe who believes that his wife is unfaithful to him. This belief makes him miserable and he decides to the contrary that his wife is faithful. As a consequence he may start believing the opposite without recourse to epistemic evidence (Gilbert 56). I find this example very troublesome, since it is open to so many different interpretations that I wonder whether it shows something at all. For example, Joe could also have suppressed his earlier belief in this case, or have misled himself, or have pretended to believe the opposite. The difference between believing and accepting is, among other things, introduced to cope with these cases. To me it seems more to the point to say that Joe accepts the proposition that his wife is faithful and act on this, given the miserable consequences that would follow if he would act on his belief that his wife is unfaithful, a belief which he probably tries to forget

384 384 Anthonie Meijers about. But as I said before, the case is so problematic that it is very hard to draw meaningful conclusions here. Belief and Truth Probably the important characteristic of the notion of belief is its relation to truth. Beliefs are assessed in terms of their truth and falsehood. In my earlier paper I argued that collective beliefs involve commitments among the members of the group. They may thus correct each other when they do not live up to their group beliefs. These interindividual commitments are incomprehensible if truth and evidence were the only things that matter for collectively believing. Furthermore, it is possible to believe a proposition as a group which none of the members of the group believe individually (the selection committee example). This phenomenon is also incomprehensible on the truth view of collective beliefs. This indicates that non-epistemic elements are involved in collectively believing a proposition. Given these nonepistemic elements, I argued, it is better to understand collective beliefs as acceptances rather than beliefs proper. Gilbert s discussion of these issues centres around two claims: (i) that collective beliefs hold more or less the same relation to truth as individual beliefs do and (ii) that it is a mistake to look at the beliefs of the members of a group when analysing the group beliefs (Gilbert 59). Before I proceed it is useful to clarify a number of things. What confuses the discussion of these issues is that the phrase we believe that p has at least four readings. The we may concern an aggregate or a group (structured whole) and the generic expression believe may concern an acceptance or a belief proper. The claim I am making is that groups do not have beliefs proper. The overall picture looks like this: mental state individual aggregate group belief proper yes yes no acceptance yes yes yes A further clarification concerns against the difference between believing and accepting. Beliefs and acceptances can both be taken as assent to a particular content. However, it is important to see that their role is different, as are the reasons for assenting to the content. Beliefs are building blocks of a correct picture of our world and are assessed in terms of truth and falsehood. Acceptances, on the other hand, are propositions we use when we deliberate our

385 Why Accept Collective Beliefs? Reply to Gilbert 385 actions and they are assessed in terms of their usefulness for achieving our goals. Given this clarification, it follows that the same proposition can both be believed to be true and be accepted as a premise in our practical reasoning about what to do. However, not all the beliefs we hold to be true are accepted as premises in our practical deliberations. The complexity of our belief system simply prevents us from doing so. We have to select the beliefs that we hold to be true and which we expect to be useful in a given situation. We accept them as premises in our reasoning about future actions. Furthermore, we may accept useful propositions in our reasoning that we know to be false. For example, modelling in science is simplifying the world to such an extent that only the relevant aspects are covered and even they are idealised. Models lie is common sense in science. In our truth seeking activities we thus accept propositions we know to be false in the hope of arriving at true beliefs. Lastly, we may accept in our deliberations propositions for non-epistemic, reasons, for example prrudentialone s. For example, when we make conservative assumptions in a situation where great risks are involved. The picture that emerges enables us to better understand why I think that collective beliefs should be conceptualised as acceptances rather than beliefs and why they have a different relation to truth than individual beliefs. The joint commitment (or agreement) that is constitutive of a group s believ, introduces reasons for assenting to this view that are non-epistemic. This make a collective belief an acceptance of a view rather than a belief proper. The view that is accepted guides the group s subsequent actions. This is not to deny of course that there are propositions that a group may be said to believe to be true, but these should be understood in terms of the weak opinion-poll conception mentioned earlier and not in terms of the strong agreementbased conception discussed here. The issue of truth is probably the most serious disagreement between Margaret Gilbert and me. She questions whether the motivation for being committed to a belief have anything to do with the nature of that belief (Gilbert, 57), thus blurring in my view the distinction between belief and acceptance. Gilbert introduces an example intended to show that a group belief that p can aim at truth, while none of its members believes p to be true (Gilbert, 58). The example is about a committee believing that a café would improve the campus, while none of the individual members believe this. The grounds for believing are non-epistemic in this case: the chancellor of the university is wedded to this idea and nobody in the committee dares to challenge him on this. In Gilbert s view it would still be appropriate to say that

386 386 Anthonie Meijers such a group belief aims at truth. The evidence she provides for this claim is the language expressions that the committee still might use (we believe that p is true, that p is the case, etc.). I find such a claim very problematic. First of all, everyday language cannot be taken for granted here, but requires an in depth analysis. In a sense, the discussion between Gilbert and me is an attempt to provide such an analysis. Secondly, I do believe that there has to be a relation between the beliefs one holds individually and the beliefs one is prepared to hold as a group, assuming that beliefs aim at truth and are shaped by the available evidence. One cannot be the mouthpiece of a committee believing that p is true, while personally believing that p is not true. In such a situation there is a clear conflict that has to be resolved, usually by distancing oneself from the committee s view. 1 The analysis of collective beliefs should in my view also take into account their relation to the beliefs of the group members. Belief and the Will The final issue between Margaret Gilbert and me concerns the question whether involuntarism is true with respect to collective beliefs. Can we believe at will as a group? I argued in my earlier ProtoSociology paper that given rejectionism and the distinction between belief and acceptance, group beliefs are voluntary. The members of a group commit themselves voluntarily to the view they accept as the group s view. It requires an act of the will of the members of the group. Gilbert takes the issue of involuntarism in a different way. The analogy between the collective and the individual case implies for her that the relevant question here is whether or not a collective will needs to be involved in the production of a collective belief. Gilbert claims that collective beliefs are not necessarily willed by the relevant collective and are in that sense certainly similar to individual beliefs. The resulting discussion shows a different ontological orientation between her and me as well as the usefulness of the distinction between belief and acceptance (in my sense). I find Gilbert s ontological view that one needs clearly to distinguish between the individual wills of the members of a collective and the will of the collective itself problematic for reasons that I have 1 The issue of the relation between the group s views and the views of the members of the group is also related to the context dependency of collective beliefs, which I discussed above.

387 Why Accept Collective Beliefs? Reply to Gilbert 387 outlined elsewhere. 2 Furthermore, Gilbert gives an example of a case intended to show that individual wills may be involved in the production of a collective belief, while a collective will is not involved. Sam is under pressure from his boss to propose a budget cut in the department. He expresses a readiness to be jointly committed with others to believe as a body that cost must be cut (Gilbert, 61). In order for the department to believe that cost must be cut, the others have to do the same. No collective will is involved in this process, according to Gilbert, thus showing that the analogy between individual and collective beliefs holds. However, Gilbert s example of a collective belief is in my view a perfect case of the collective acceptance of a view, though not in her sense of for the sake of the argument. It is the acceptance of the proposition that budget cuts are necessary by the members of the group. This proposition will subsequently be used in the department s practical deliberations about future actions. So apart from the ontological issues related to a collective will there is the issue here of how to understand the phenomenon described here in the first place. Taken together, they make it probably very hard to settle the controversy of involuntarism. 3. Conclusion Margaret Gilbert s criticism of my view that collective beliefs should be understood in terms of the acceptance of a view rather than of belief proper has raised many interesting issues, ranging from methodological ones to issues in (social) epistemology. I am grateful for her thought provoking remarks, analyses, and counterexamples, which challenged me to clarify my ideas at several points; for example, with respect to the context dependency of collective beliefs and the distinction between belief and acceptance. However, I do not believe that my position, which she calls rejectionist, is undermined by her arguments. There are also additional arguments in favour of rejectionism developed by Raimo Tuomela (2000) and K. Brad Wray (2001) that I have not discussed here. So contrary to Gilbert I believe that ordinary language misleads us in suggesting that the notion of belief in collective belief and individual belief is the same notion. Therefore, with respect to collective beliefs the learned should not speak with the vulgar and go ahead and call them beliefs, to use Gilbert s own phrase (Gilbert 67). It is important to keep in mind is that I have been analysing collective 2 See also Meijers (2003), Can Collective Intentionality be Individualized?

388 388 Anthonie Meijers beliefs of the strong, agreement-based type. For this type of belief rejectionism holds, i.e., these beliefs should be understood in terms of acceptances. For the weaker opinion poll type rejectionism does not hold, since this type of collective belief can be understood in terms of individual beliefs plus a certain distribution of these beliefs among the members of the group. It is also important to note that rejectionism does not imply epistemic individualism, a claim made by Tollefsen (2002). But the question in what sense collective agents can be said to know something is a complicated one which requires, among other things, an analysis of the idea of distributed cognition, a topic that is clearly beyond the scope of this paper References Bratman, M. (1993), Practical Reasoning and Acceptance in a Context, Mind 102, Engel, P. (1998), Believing, Accepting, Holding True, Philosophical Explorations 1, Gilbert, M. (2002), Belief and Acceptance as Features of Groups, ProtoSociology 16, Meijers, A.W.M. (2002), Collective Agents and Cognitive Attitudes, ProtoSociology 16, Meijers, A.W.M. (2003), Can Collective Intentionality be Individualized?, American Journal of Economics and Sociology 62, Tollefsen, D. (2002), Challenging Epistemic Individualism, ProtoSociology 16, Tuomela, R. (2000), Belief versus Acceptance, Philosophical Explorations 3, Wray, K. Brad (2001), Collective Beliefs and Acceptance, Synthese 129, I would like to thank Melissa van Amerongen, Giacomo Romano and Marc de Vries for stimulating discussions of these issues in a research group on collective beliefs at Eindhoven University and Melissa van Amerongen en Marcel Scheele for carefully reading this reply.

389 Rejecting Rejectionism 389 Rejecting Rejectionism Deborah Perron Tollefsen Abstract There is a small, but growing, number of philosophers who acknowledge the existence of plural subjects collective agents that act in the world and are the appropriate subject of intentional state ascriptions. Among those who believe in collective agency, there are some who wish to limit the types of intentional state ascriptions that can be made to collectives. According to rejectionists, although groups can accept propositions, they cannot believe them. In this paper I argue that, given the centrality of belief and the similarities between individual belief and collective attitudes, we ought to reject rejectionism. If one believes in collective agency, one must also believe in collective belief. Introduction Until recently, the notion of collective intentionality was likely to be discussed only in conjunction with the work of Emile Durkheim and early 20 th century debates over the proper methodology of the social sciences. We now find, however, philosophers of language, mind, and action theory turning their attention to the issue. With the publication of John Searle s The Construction of Social Reality (1995) the topic of collective intentionality has been acknowledged as central to discussions in social ontology, legal theory, and moral theory. 1 Indeed, with the help of Searle, the topic has become somewhat fashionable. Among those who acknowledge that collectives can be the subject of intentional state ascription, there is a debate raging over which type of intentional states are appropriately attributable to collectives. There are some, like Margaret Gilbert and myself, 2 who argue that it is appropriate to attribute to groups a wide range of intentional states including beliefs. Others, like K. Brad Wray (2002), Raimo Tuomela (2000), and Anthonie Meijers (1999), 1 See Lagerspetz, et al. (2001) for a discussion of the role collective intentionality plays in a variety of contexts. 2 For Gilbert s account of collective intentionality see (1989), (1994), and (2000). For Tollefsen s account of collective intentionality see (2002a,b,c).

390 390 Deborah Perron Tollefsen have argued that, although groups may accept a proposition, they cannot believe a proposition. The nature of belief, according to these philosophers, is such that groups cannot be believers. The latter camp has been labeled by Gilbert as the rejectionists because they reject the possibility of group belief. For ease, I refer to the former camp as the believers. In What Really Divides Gilbert and the Rejectionists (2003, this volume), Wray responds to Gilbert s recent attack on rejectionism (2002). In this paper, I continue the dialogue and attempt to respond on behalf of the believers. Wray s defense of rejectionism is commendable and it contributes a great deal to the debate in terms of clarifying key issues. Despite this, I think there are serious problems with rejectionism. In section I, I develop more fully a response to rejectionism I provided in Challenging Epistemic Individualism (2002b). Following Wray, I think our methodological starting point must be with the conception of belief (and intentionality in general) that has arisen by reflection on the individual case. But I will argue that reflection on the nature of intentionality at the individual level puts the rejectionist in a rather difficult position. Given the centrality of belief, the rejectionist must either reject rejectionism or reject non-reductive accounts of collective intentionality. If groups don t have beliefs, they don t have any other intentional states either. This is not, of course, particularly troubling for the individualist who will willingly reject non-reductive accounts of collective intentionality. But it should be troubling to Wray, and other rejectionists, who want to acknowledge collective agency. In section II, I build upon my comments about the centrality of belief and focus on what D.S. Clarke has called the entailment thesis: the thesis that if one accepts that p, one believes that p. This thesis has been rejected by defenders of the distinction between belief and acceptance. According to J. L. Cohen (1989), R. Stalnaker (1984), M. Bratman (1999), and many rejectionists, one can accept what one does not believe. This is thought to distinguish belief from acceptance, as one cannot believe what one does not also accept. D.S. Clarke makes a compelling argument that the entailment thesis holds in most cases of acceptance. This has significant consequences for the debate between the rejectionists and the believers. If acceptance entails belief then collective acceptance entails collective belief. Indeed, one might just as easily see collective acceptance as a species of collective belief. In section III, I focus on what Wray now identifies as the crux of his argument for rejectionism. According to his most recent article, the main emphasis of his argument for rejectionism is the fact that belief is not goal-directed, whereas acceptance is. Since, groups typically have a goal in mind

391 Rejecting Rejectionism 391 when they make decisions, acceptance is the appropriate intentional state to attribute to them. I argue in response to this that there are many cases of individual belief formation that are directed and informed by goals. The fact that groups typically have a goal in mind when they form their attitudes, is not, therefore, enough to rule out the existence of group beliefs. In section IV, I turn briefly to the issue of eliminativism. Wray has argued that there are good reasons for eliminating collective belief ascriptions. Although I agree that there are some cases in which it would be more appropriate to attribute acceptance than belief to a collective, I will suggest that there are powerful explanatory reasons for including collective belief ascriptions in our folk sociological repertoire. I. The Centrality of Belief In Belief and Acceptance as Features of Groups (2002) Margaret Gilbert criticizes the rejectionist methodology. She argues that the rejectionists begin their analysis of collective intentionality by focusing on characteristics of individual belief. According to Gilbert, this methodology begs the question. It favors individual belief over collective belief and leads to an unjustified rejection of collective belief. Although Gilbert and I are of the same mind on many issues, I disagree with her here. 3 Whether an account of belief begs the question or not with respect to collective intentionality depends on the details of the account, rather than its being developed by reflection on the individual. Admittedly, discussions of collective intentionality are still governed by individualistic assumptions, but I think our only option is to focus, at least initially, on the theories of intentionality that have been developed by contemporary philosophy of mind. It is at the level of individuals that we have the most sophisticated and well-defended accounts of intentionality. As my own research suggests (2002a,b,c), contemporary philosophy of mind is a rich and important resource for thinking about collective intentionality. Indeed, reflection on the nature of belief at the individual level raises a serious problem for the rejectionist. There is considerable agreement among philosophers of mind that belief is central. This is what Donald Davidson 3 To be fair, Gilbert is not suggesting that we completely give up reflecting on the nature of intentionality at the individual level. She does go on to note that taking central criteria for belief from current theories of individual intentionality and applying them to collectives is not as problematic as various rejectionists make it out to be. Thus, it is not clear that we disagree all that much.

392 392 Deborah Perron Tollefsen has called the principle of the centrality of belief (1984, 156). Having a thought or any other mental state requires that there be a background of beliefs. The principle does not say that in order to desire that p one must have a belief that p. Rather, in its strongest version the principle claims that desires, intentions, fears, hopes, and other attitudes get their content and identity from the beliefs in which they are embedded. I could not have a fear of flying in airplanes, for instance, unless I had a variety of other beliefs, presumably, some about airplanes. I could not desire a hot fudge Sunday unless I had a variety of other beliefs, presumably, some about hot fudge. What particular beliefs one must have in order to have a particular intentional state is difficult to determine. But the principle of the centrality of belief does not require that we settle this. It simply points to the holistic nature of mental content. Desires, intentions, fears and so on, get their content against the backdrop of other beliefs. As Davidson puts it, Not only does each belief require a world of further beliefs to give it content and identity, but every other propositional attitude depends for its particularity on a similar world of beliefs. In order to believe the cat went up the oak tree I must have many true beliefs about cats and trees, this cat and this tree, the place, appearance and habits of cats and trees and so on; but the same holds if I wonder whether the cat went up the oak tree, fear that it did, hope that it did, wish that it had, or intend to make it do so. Belief-indeed, true belief-plays a central role among the propositional attitudes. (Rational Animals, p. 475) If belief plays the central role that Davidson claims it does, it poses a problem for the rejectionist. The rejectionist claims that groups can accept propositions but they cannot believe propositions. But if all propositional attitudes depend for their particularity on beliefs of some kind or other, then without group beliefs there are no group attitudes at all, including acceptance. Wray and other rejectionists might respond to this argument by rejecting the centrality of belief principle. There have been previous attempts to do so. Jerry Fodor, for instance, has argued that if the principle is true, it is true only contingently, because it is at least possible that desires eventuate in behavior without having to interact with beliefs (1992,119). But the principle need only be contingently true to make my point. Wray himself continually stresses that he is dealing with the typical case of belief. Typically, though not necessarily, one could not accept that p without having a variety of other beliefs, presumably some related to p. If groups typically can t have beliefs about or related to p, then they cannot accept that p.

393 Rejecting Rejectionism 393 Further, the principle can be weakened in order to avoid objections to meaning holism. Davidson s point in the passage above is that propositional attitudes get their content from the web of beliefs in which they are embedded. Although meaning holism is a well-defended position, we need not be committed to this in order to give belief a central role. On a weaker reading, the principle of the centrality of belief simply entails that a creature cannot have propositional attitudes unless it has some beliefs. On this reading the principle seems trivially true. A creature that does not have any beliefs at all or that lacked the capacity to have beliefs would surely be unable to have any other attitudes. But this is just what the rejectionist is claiming about groups. Because of the nature of belief and the ways in which group decisions are made, groups aren t capable of having beliefs. But if they lack the capacity for belief, how can they have any attitudes at all? 4 The rejectionist, then, is faced with a dilemma. Either he can reject rejectionism, or he can accept the consequences that follow from the principle of the centrality of belief and admit that groups are not legitimate subjects of attitude ascription. Although the latter option is open to someone like Raimo Tuomela 5, Wray, at least, seems to want to acknowledge that there are plural subjects with the capacity for intentional agency (2003, 139) and that this agency is to be understood non-reductively (2003, 140). There is, however, no intentional agency without belief. The rejectionist, then, if he wants to acknowledge group agency, must reject rejectionism. II. The Entailment Thesis In this section I want to focus on a thesis that suggests that the relationship between acceptance and belief is even stronger than the principle of the centrality of belief (in either its strong or weak form) suggests. This thesis states 4 There is another way that Wray and other rejectionists might respond to my argument. They might argue that group acceptance is a function of individual beliefs and that the web of individual beliefs serves as the necessary backdrop for group acceptance. This doesn t seem to help those that want to acknowledge that groups are, themselves, plural agents. Indeed, it opens the door to a reductive account of group attitudes. If group acceptance is merely a function of individual attitudes then why not simply reduce group acceptance to individual attitudes? 5 As I understand it, for Tuomela, a group can think (e.g. have an attitude) and act only in a somewhat metaphorical sense: when a group thinks or acts its relevant members acting as group members must think or act in suitable ways, and these ways, in the general case, involve joint thinking or joint acting. See Tuomela (1992), (2000).

394 394 Deborah Perron Tollefsen that the relationship between belief and acceptance is one of entailment. Following the work of J. L. Cohen (1989), R. Stalnaker (1984), M. Bratman (1999), rejectionists have adopted that view that one of the things that distinguishes belief from acceptance is the fact that one can accept what one does not believe. Part of the argument for rejectionism, then, rests on denying what D.S. Clarke (1994) has labeled the entailment thesis; the thesis that if one accepts that p, one believes that p. Clarke provides a compelling defense of the entailment thesis. If he is correct, then there is additional reason for thinking that an agent, either individual or collective, cannot be an acceptor without also being a believer. Cohen, Stalnaker, and Bratman provide several examples to motivate rejection of the entailment thesis. Consider what one does when one measures a window for curtains. 6 I measure my kitchen windows as 3 feet wide and 6 3/4 feet tall. When I go to buy a curtain I might accept as true the proposition that the window is 7 feet tall. Although I accept the proposition My window is 7 feet tall as true relative to the purposes as hand, I believe independently of this context the measurement I previously made. Therefore, what is accepted is not believed. Bratman (1999, p. 22) provides another example that attempts to undermine the entailment thesis. Suppose that a person receives several estimates on construction for a house. In order to be financially prudent, the person accepts the highest estimate, say $300,000. She uses this estimate as a basis for deciding when to begin construction and for further decisions she makes about her financial situation. But she may not in fact believe that it will cost that much to complete the construction. Prudence suggests, however, that one ought to accept a projected cost that is higher than what one really believes. Thus, she accepts the proposition that construction will cost 300,000 but she does not believe it. Therefore, according to Bratman, acceptance does not presuppose belief. These examples do not, however, support rejection of the entailment thesis. As Clarke points out (1994, 147) these examples misdescribe the content of the proposition being accepted as true. Once the content of the proposition is accurately described, the entailment thesis still stands. What I accept in buying the curtain is an estimate of what I believe is the window s length. The curtain does not have to be exactly the same size as the window. Window treatments can often be an inch or so larger or smaller than the window. If so, then the proposition being accepted is that the window s length is 8 feet plus 6 See Clarke (1994, 147) for a similar example involving a tablecloth.

395 Rejecting Rejectionism 395 or minus a few inches. If this is so, there is no inconsistency between this and what a person believes on the basis of prior measurements. We can redescribe the content in Bratman s example as well. Surely the person is not accepting the simple proposition that the construction will cost 300,000. The evidence does not support this hypothesis. Clarke (1994, 148) describes the content more accurately as the proposition that there is a low probability of.1 that actual costs will exceed the estimate of $300,000. Acceptance of this proposition is not inconsistent with a belief that construction will not be $300,000. Thus, there is no reason to reject the view that acceptance presupposes belief. J.L.Cohen provides a legal example to motivate rejection of the entailment thesis. Consider a jury that deliberates and decides on a verdict of not guilty. Members of the jury, nevertheless, may believe that the defendant is guilty. Thus, they accept that he is not guilty and this acceptance does not entail that they believe he is not guilty. But this example involves a mistake similar to those found in the Bratman and Stalnaker examples. If we redescribe the situation there is no conflict between what the juror believes and accepts. The juror accepts that under the laws of the state the defendant cannot be convicted of the crime. Thus, the juror believes that under the laws of the state the defendant cannot be convicted of the crime. She may, however, also believe that the defendant is not innocent (in the moral sense). Once we better specify the content of the beliefs we see that there is no inconsistency between what is accepted and what is believed and hence no need to reject the entailment thesis. Cohen also provides arguments based on scientific practice to support his rejection of the entailment thesis. Scientists often accept a proposition p based on empirical evidence supporting p and on the basis of its relation to other accepted theories. The accepted proposition is then used in making predictions and explanations and justifies new lines of research. But the acts of acceptance, according to Cohen, can be completely independent of the scientist s beliefs. In fact, Cohen argues that it is often better if the scientist does not believe the proposition he is trying to verify empirically. This prevents the scientist from being biased and promotes objectivity. Does scientific practice provide a refutation of the entailment thesis? Again, Clarke makes a convincing case that it does not. Scientists will often withhold final endorsement of a theory or proposition until they have adequate empirical research to back it up. At some point in time, however, answers are given to the questions the theory is meant to answer and decisions are then made to terminate further research. Physicists have accepted

396 396 Deborah Perron Tollefsen as true Hubble s theory of the expanding universe and the Watson-Crick model DNA, not simply as adequate to the existing data, but in the sense that the questions to which these were proposed as answers are no longer open in the sense they once were (1994, 149). Clarke makes the distinction between two sorts of acceptance: the provisional acceptance of a hypothesis and terminal acceptance. Provisional acceptance of a hypothesis h is a decision to consider h for further testing, to include h in the set of those hypotheses that experimental data must choose among (Clarke, 1994, 149). At the provisional stage it may be that acceptance does not entail belief. Although, given that belief can vary in degree, one could easily explain this situation by claiming that scientists believe h to a minimal degree until further evidence is provided. After all, they must believe h is true to some degree to pursue it as part of a research program or to incorporate it into their theory. Regardless of how we describe the provisional stage of scientific inquiry, the entailment thesis, according to Clarke, has always applied to terminal acceptance rather than provisional acceptance. Terminal acceptance does, indeed, imply belief. If we consider those theories which the scientific community has terminally accepted we will find that, in general, scientists believe the hypothesis under question. Those scientists that do not believe h are those that are challenging h and thus have not granted h terminal acceptance. At the terminal stage members of the scientific community have accepted the hypothesis based on a great deal of evidence. Of course, science is still fallible so that they may not believe the hypothesis with certainty, but there is some degree of belief implied by their terminal acceptance. It is the belief (to some minimal degree) in the hypothesis that allows scientists to move on. Clarke provides convincing evidence that the entailment thesis still stands. Acceptance, in most cases, presupposes belief. If one accepts that p, one also believes that p to a certain degree. This suggests that the distinction between belief and acceptance is not as clear as the rejectionists make it out to be. Indeed, one way to understand the relationship between acceptance and belief is to see acceptance as a species of belief. To accept that p is just to believe p is true 7 to some minimal degree in a particular context. If we under- 7 It has been suggested to me by one of the editors of this issue that there is a distinction to be made (one that I do not make) between acceptance of a proposition as true and acceptance of a proposition as useful. I must admit that I do not see that there is such a distinction and it is not one that arises in the literature on belief and acceptance. To

397 Rejecting Rejectionism 397 stand the relationship between belief and acceptance in this manner, one cannot accept a proposition without also believing it to some degree. Thus, collective acceptance is just a form of collective belief. 8 III. Goal Directedness In this section I want to turn to the refined version of the rejectionist s argument found in Wray (2003, this issue). It will be helpful, however, to recap Wray s original argument for rejectionism. In Collective Belief and Acceptance (2002), Wray identifies four differences between acceptance and belief. 1. You can accept things that you do not believe but you cannot believe what you do not accept. (Rejection of the entailment thesis) 2. Acceptance often results from a consideration of one s goals, and thus results from adopting a policy to pursue a particular goal. (2002, p. 7). 3. Belief is a disposition to feel that something is true. 4. Belief is involuntary, whereas acceptance is voluntary. Wray then proceeds to show that the examples that Gilbert gives of group belief (1989), (1994), are actually instances of acceptance. Because group attitudes are formed against the background of goals, because they are formed voluntarily, and because their formation does not entail that the members believe the content of the attitude, group views are more aptly described as instances of acceptance. accept a proposition is to accord it a role in one s reasoning. But in order for a proposition to play any role (either by providing rational grounds for inferences or by providing psychological motivation) it must be accorded some probability of being true. A proposition that we know for certain is false won t be very useful to us. Indeed, I think there is good empirical evidence that people simply cannot accept propositions and act on them when they believe them to be absolutely false. We can entertain propositions that we know to be false. But to accept a proposition is to license the proposition to give it a place and some force in our cognitive economy. This is simply not something we do unless we think the proposition has some possibility of being true. To accept a proposition as useful is just to accept it as true to some, perhaps minimal, degree. 8 We should note here that Clarke not only defends the entailment thesis but also provides positive arguments in support of it. I do not consider those arguments here.

398 398 Deborah Perron Tollefsen In Challenging Epistemic Individualism (2002b) I responded to this argument by showing that the issue of voluntarism is not as clear cut as Wray suggests. First, although we, as individual human beings, cannot will to believe, it has not been established that belief itself is necessarily involuntary. Involuntarism about belief is, at best, a contingent thesis about human abilities, not a necessary truth about belief. It may be that individuals cannot will to believe, but groups can. Further, although we cannot will to believe there is a way in which we have control over our beliefs. We have control, to some extent, over the conditions under which we can acquire beliefs. I will return to this form of voluntarism about belief below. Second, I questioned the account of belief that Wray adopts in 3. The link between intentionality and phenomenology, although gaining popularity, 9 is by no means an established view of belief in contemporary philosophy of mind and language. Granted, we must all begin somewhere in our theorizing, but this starting point isn t philosophically prudent. It is here that we find a good dose of begging the question. The view that belief states have a phenomenological content or character, begs the question not just with respect to collective belief but with respect to artificial intelligence, alien cognition, and quite possibly some forms of animal cognition. If belief is defined in terms of phenomenology then creatures that lack phenomenology (or that lack a specific sort of phenomenology) will be ruled out as believers by definition. In What Really Divides Gilbert and the Rejectionists? (2003, this issue) Wray has refined his argument. His focus, he says, is not on the issue of voluntarism or the phenomenological account of belief, but the goal-dependency of acceptance. Wray writes, In my own analysis of collective acceptance, I also claim that acceptance is typically voluntary whereas belief is typically involuntary (Wray 2002). But, I do not build my case against the claim that plural subjects do not believe on this fact. Unfortunately, my critics have failed to see this. Both Gilbert (2002) and Tollefsen (2002) seem to suggest that I ground the claim that collective belief ascriptions really refer to instances of collective acceptance on the following two claims: (I) that belief is voluntary, and (II) that instances of collective acceptance are voluntary. The defining characteristic of acceptance that I emphasize in my argument for rejectionism is the fact that what we, individually or collectively, accept is typically tailored to specific ends. We accept a view for the sake of an 9 See for instance Horgan and Tienson (2002), and Siewert (1998)

399 Rejecting Rejectionism 399 argument, or in order to provide a shared background from which to reason. Belief, I argue, is generally not like this (2003, p.11). Wray is correct. I did fail to see that the case was not built on the issue of voluntarism, but this is because the goal-directness of acceptance (and the lack of goal-directedness of belief) is essentially tied to the issue of voluntarism. If one could form beliefs voluntarily or at will, then one could form them with a goal in mind. It is precisely because belief seems to be out of our control that we can t form beliefs in response to particular goals we have. It is, therefore, extremely difficult to pry the issue of voluntarism apart from the issue of goal-directedness. But I want to try, as Wray does, to set the issue aside. I think we can do so by acknowledging a minimal level of voluntarism with respect to belief. As I noted above, although we cannot will to believe, we can do various things to put ourselves in the right position, so to speak, to acquire beliefs. My 4 yr old son has recently become interested in insects. He spends a great deal of time in the backyard searching for creepy crawlers. In an attempt to foster and promote learning, I have searched out various websites and books on insects that we have explored together. By putting ourselves in the appropriate context, we have acquired a great number of beliefs (hopefully, true beliefs) about various insects. To this extent, we had control over our beliefs. We didn t will to believe, but we willed to acquire beliefs of a specific sort. Assuming that this level of voluntarism about belief is plausible, let us return to the issue of goal-directedness. Wray argues that typically belief is not tailored to specific ends. What a person or group accepts is accepted in light of a goal or set of goals. Beliefs are not like this. They are not tailored to our purposes. It will be useful here to turn to Wray s original article to figure out just what he means by attitudes being tailored to specific ends. Here is what he says about Gilbert s case of the parents who are jointly committed to raising their children. The parents are jointly committed to raising their child effectively. Thus, as a unit, the parents develop views with this goal in mind. It is their collective goal that determines what they claim to collectively believe. Sometimes they will even adopt views that conflict with their personal preferences and beliefs (2000, 7). The basic idea seems to be that the overall goal of raising children effectively informs and directs the attitudes the parents choose to adopt as a parental

400 400 Deborah Perron Tollefsen unit. The attitudes are dependent on the goal in some way. In some cases, the attitude that is adopted is a means to realizing a goal. Belief, according to Wray, does not typically function in this manner. 10 Is it really the case that beliefs are not goal dependent? If we grant the minimal voluntarism described above, I don t believe it is. Indeed, I think that typically there are lots of goals that inform and direct us in our acquisition of beliefs. My son s insect education is a particularly good case. My son and I had the goal of learning about creepy crawlers and this goal and various subgoals determined the ways in which we acquired beliefs. We acquired beliefs about their lives, for instance, rather than how to exterminate them. We learned about their eating habits rather than their reproductive habits. We acquired beliefs about certain insects rather than others. As the mother of young children, one my goals is to foster self-esteem in my children. This goal informed the way I went about acquiring information (beliefs) about child development and child-rearing techniques. I chose to read Dr. Sears over Dr. Spock, as there is some evidence that attachment parenting promotes self-esteem. As a consequence, I have come to acquire a radically different set of beliefs than those I would have acquired by reading Dr. Spock. My goal of attaining a Ph.D. determined in various ways the beliefs I acquired as a graduate student. In order to fulfill that goal I had to write a dissertation which itself became a subgoal. In order to pursue this subgoal I placed myself in a position to acquire all sorts of beliefs about collective intentionality. What I believe today about collective intentionality is, in a sense, believed in light of my goal of pursuing a Ph.D. in philosophy. I read certain books rather than others. I went to certain conferences rather than others. If I did not have the goal of receiving a PhD and writing a dissertation on collective intentionality, I would not have the beliefs I now have about collective intentionality. Indeed, at the risk of sounding Aristotelian, I would argue that the desire to understand seems to be a general epistemic goal that directs and informs the belief acquisition of all human beings. We may not have this goal in mind at all times when we are out in the world looking into microscopes, 10 I take the issue of the goal-directness of acceptance to be different from the Bratman/ Stalnaker view that acceptance is relative to a context of practical deliberation. I suspect, however, that these two claims are related. See Clarke (1994, 154) for an example that challenges the Bratman/Stalnaker claim. It appears that there are cases where belief varies with respect to context also and so this cannot be the basis of a distinction between belief and acceptance.

401 Rejecting Rejectionism 401 reading books, looking up words in the dictionary, listening to lectures, but it is there implicitly tailoring our attitudes in various ways. If we didn t have such general epistemic goals, it would be difficult to understand why anyone would bother getting up off the couch. If belief acquisition is typically informed and governed by overarching goals and desires then there is no need to reinterpret Gilbert s parent case as a case of collective acceptance. The parental unit has the goal of raising children effectively. The parents may choose, as a unit, to pursue an attachment parenting philosophy. In order to fulfill this goal they put themselves in the position to acquire all sorts of beliefs about attachment parenting. They read and discuss together Dr. Sears work, books on breastfeeding, co-sleeping, and so on, and they come to believe both individually and collectively (through discussion, joint deliberation, and perhaps, joint commitment) lots of different things. And they engage in the joint activity of raising their children in light of these collective attitudes. They may, of course, also come to accept (and believe) things as a unit that they do not personally believe. They may collectively accept some things and collectively believe others. But this possibility does not rule out collective belief. Given a certain understanding of voluntarism about belief (one that Wray could, if he wanted to, accept since he claims that his argument does not rest on the issue of voluntarism), I see no reason to believe (or accept) that there is a distinction between acceptance and belief based on goal dependency and thus, no reason to deny that groups can believe on the basis of this distinction. IV. Eliminativism I have argued that, given the various ways in which belief is central, an agent cannot be an acceptor without also being a believer. Thus, if groups accept, they are believers. I have also argued that individual belief acquisition is often informed and directed by goals. Thus, the fact that groups acquire their attitudes with goals in mind does not support the view that they cannot be believers. From all that I have said in this article and in Challenging Epistemic Individualism (2002b) it would seem that I am completely denying any distinction between belief and acceptance. This is not so. There are differences between acceptance and belief. Acceptance appears to be a dateable psychological act whereas belief is typically a persistent state or disposition. Accepting that something is true, does not always commit oneepis-

402 402 Deborah Perron Tollefsen temically in the ways that believing something to be true does. And acceptance does, in many cases, seem to be something I have more control over than belief (although, as I have shown, we often have control over the acquisition of our beliefs). I am even in agreement with Wray that there will be cases (as there are at the individual level) where it is more appropriate to ascribe acceptance rather than belief to a collective. I suspect that these will be cases in which the group commitment to the proposition is not very strong and is based simply on practical or pragmatic considerations. But there are, I think, strong reasons for resisting Wray s insistence that we do away with collective belief ascriptions altogether. Wray is essentially arguing for a form of eliminativism with respect to collective intentionality. Paul Churchland (1981) famously defends eliminitivism with respect to individual intentional states. According to Churchland, our common sense view of the mind constitutes a sort of folk psychology. But mature neuroscience will most likely show that this folk theory of the mind is false and so we will no longer be justified in believing that the entities postulated by the theory (things like beliefs, desires, and so on) exist. Just as the failure of the phlogiston theory of combustion removed any justification for believing in the existence of phlogiston, so the failure of folk psychology removes the justification for believing in the existence of folk psychological entities. Wray offers us a form of limited eliminativism with respect to collective intentionality. He seems to want to acknowledge that groups, themselves, can accept propositions, but belief and other psychological states ought to be eliminated from our repertoire when it comes to ascribing attitudes to groups. I think the major reason for rejecting Churchland s eliminativism with respect to individual intentional states applies equally well to Wray s form of eliminativism. Folk psychology at the individual level has not been shown to be false by neuroscience. On the contrary, its explanatory power strongly suggests that it is ineliminable. Likewise, the explanatory power of, what I shall call, folk sociology, suggests that it is ineliminable. This folk theory includes a whole range of collective psychological states-including beliefs. As I have argued elsewhere (2002c), prediction and explanation of collective behavior in terms of collective beliefs, desires, and intentions, is ubiquitous and successful. This is so, not only in colloquial contexts but in social scientific research. Rational choice theory, for instance, a theory steeped in belief/ desire psychology, is routinely applied to collectives like nations in order to explain and predict their behavior. 11 Further, if Wray is correct, attribution of 11 See, for instance, Elster (1983).

403 Rejecting Rejectionism 403 collective beliefs are literally false because there is no such thing as collective belief. This leaves us with the odd result that social scientists, economists, and many of us everyday folk 12 are saying false things all the time. Although, we do say false things sometimes, any theory that has us saying false things all the time should be looked upon as suspect. 13 My appeal to common usage here will, no doubt, be dismissed by Wray, as he dismissed Gilbert s appeal to it in What Really Divides Gilbert and the Rejectionists (this volume, 2003). Perhaps it is this which really divides us. If Wray wants to dismiss common usage and practice as relevant to the issue of collective intentionality then it seems we must agree to disagree. For me, collective intentionality ascriptions have their home in our everyday social lives. These ascriptions function much the same way they do at the individual level. They help us make sense of ourselves and the world around us and are governed by normative considerations. Understanding them, then, will necessarily involve reflection on common sense practice. Wray suggests that a scientific study of collective intentionality will show us that we have been mistaken about the nature of collective attitudes, just as Churchland thinks that a scientific study of the individual brain will reveal that there are no psychological entities like belief. I would suggest that a scientific study of either collective intentionality or individual intentionality is doomed to failure. Intentional agency is present at a level of description that is foreign to the natural sciences and it is governed by normative considerations that are not amenable to scientific laws. A scientific study of intentional agency, therefore, is likely to lose sight of the phenomenon altogether Another odd result of eliminating collective belief ascriptions is that it will no longer be appropriate to talk about the beliefs of a religious institution. The Nicene Creed is a statement of what the Catholic Church (and many individual members) believes. It would be odd to say that the Church simply accepts those statements in the service of some pragmatic goal. Mere acceptance of a proposition doesn t seem to provide the requisite motivation for action. The Catholic Church has, sometimes violently, defended its creed. It is difficult to explain the institutions behavior by appeal only to collective acceptance. 13 Rejectionism, in effect, gives us an error theory. J.L. Mackie (1990) famously defended an error theory about affimative moral judgments: they purport to refer to objective moral properties, but (he claimed) there are good metaphysical and epistemology reasons to doubt there are such (instantiated) properties. Hence, common sense moral judgments are (taken literally) false. See Crispin Wright (1992) for a refutation of error theories, in general. My claim here is that an error theory about common sense attributions to collectives is as untenable as Mackie s error theory is in ethics. I do not have space here to substantiate this claim. 14 What divides this believer from the rejectionists, then, are deep seated views about the nature of intentionality.

404 404 Deborah Perron Tollefsen ***** I have attempted in this paper to shift the explanatory burden back to the rejectionists. Wray has argued that it has yet to be proven (2003, 372) that collective beliefs behave much like beliefs in the individual case. I have tried to show in this paper that the similarities between collective attitudes and individual beliefs are quite strong. Further, given the primacy of belief, if Wray and other rejectionists want to acknowledge collective agency, they must acknowledge collective belief. Thus, they must reject rejectionism. References Bratman, M Practical Reasoning and Acceptance in a Context, in Faces of Intention. Cambridge, UK: Cambridge University Press, pp Churchland, P.M Eliminative Materialism and the Propositional Attitudes Journal of Philosophy, 78, pp Cohen, J.L Belief and Acceptance, Mind XCVIII(391), pp Clarke, D.S Does Acceptance Entail Belief? American Philosophical Quarterly, 31 (2), pp Davidson, Donald Rational Animals, in Actions and Events, ed. E. Lepore and B. McLaughlin. Oxford: Basil Blackwell, pp (1984). Thought and Talk in Inquiries into Truth and Interpretation. Clarendon Press: Oxford. pp Elster, J Explaining technical change. Cambridge: Cambridge University Press. Fodor, J Holism. Oxford: Basil Blackwell. Gilbert, M On Social Facts. Princeton, N.J.: Princeton University Press Remarks on Collective Belief, in F. Schmitt, ed. Socializing Epistemology:The Social Dimension of Knowledge. Lanham: Rowman and Littlefield, pp Sociality and Responsibility: New Essays on Plural Subject Theory. Lanham: Rowman and Littlefield Belief and Acceptance as Features of Groups, Protosociology, vol. 16, pp Horgan, T. and Tienson, J The intentionality of phenomenology and the phenomenology of intentionality. Philosophy of Mind: Classical and Contemporary Readings. Oxford: Oxford University Press, pp Mackie, J.L. (1990) Ethics: Inventing Right and Wrong. USA: Penguin Press. Lagerspetz, E. IkaHeimo, H., Kotkavirta, J. (2001) On the Nature of Social and Institutional Reality. Jyvaskyla, Finland: SoPhi. Searle, J The Construction of Social Reality. New York, N.Y.: Free Press. Siewert, C The Significance of Consciousness. Princeton, N.J: Princeton Uni-

405 Rejecting Rejectionism 405 versity Press. Stalnaker, R. (1984). Inquiry. Cambridge, MA: MIT Press. Tollefsen, D. 2002a. Collective Intentionality and the Social Sciences. Philosophy of the Social Sciences 32 (1): b. Challenging Epistemic Individualism. Protosociology, vol. 16, pp c. Organizations as True Believers. Journal of Social Philosophy, vol. 23, pp Tuomela, R Group Beliefs, Synthese, 91: pp Belief versus Acceptance, Philosophical Explorations, 2: pp Wray, K.B Collective Belief and Acceptance, Synthese, 129: pp Wray, K.B Collective Belief and Acceptance, Synthese, 129: pp Wright, C Truth and Objectivity. Cambridge, MA: Harvard University Press.

406 406 Deborah Perron Tollefsen

407 Goldman s Knowledge in a Social World: 407 On Social Epistemology and Terrorism in Time of Globalization

408 408 Patrick Rysiew

409 Goldman s Knowledge in a Social World: 409 Goldman s Knowledge in a Social World: Correspondence Truth and the Place of Justification in a Veritistic Social Epistemology Patrick Rysiew Abstract Knowledge in a Social World (KSW) is Alvin Goldman s sustained treatment of social epistemology. As in his previous, individualistic epistemology, Goldman s lodestar is the idea that it is the truth-aptness of certain processes/methods which marks them out for our epistemic approval. Here, I focus on issues concerning the framework of KSW: Goldman s claim that a correspondence theory of truth is favoured/required by his veritistic social epistemology (VSE); and the issue of whether a VSE of the sort Goldman elaborates and defends shouldn t be (not replaced but) supplemented by more procedural or justification-centred considerations. 1. Introduction While he has made substantial contributions in other areas of philosophy, it is of course for contributions to epistemology that Alvin Goldman is best known. Until recently, Goldman has tended to follow epistemological tradition in focusing on individuals on what it is for an individual to possess some piece of propositional knowledge, on the nature of epistemic justification, and so on. With Knowledge in a Social World (KSW), however, Goldman has now given us a sustained treatment of social epistemology (SE). Now, as perhaps befits the topic, social epistemology applies to quite a diverse, and sometimes competing, range of issues and projects (see, e.g., Goldman 2001b and 2002b). Goldman s own lodestar here, however, as in his non-social epistemology, is the idea that it is the truth-aptness of certain processes/methods which marks them out for our epistemic approval. In KSW we are given (in Part I) an elaboration and defence of this veritistic framework, followed by specific applications thereof, with regard to both (Part III) rather specialized domains of socio-epistemic activity (e.g., science and the law) and (Part II) more generic forms of collaborative and/or co-

410 410 Patrick Rysiew operative information-sharing practices (e.g., testimony and argumentation). In a short piece such as this, it s not possible to do justice either to KSW or to the many questions and issues on which Goldman touches. So I shall confine myself to discussing Part I of the book, and two of the (to my mind) more salient questions which it prompts. To that end, I ll begin with a very brief sketch of the general framework that Goldman develops and defends in KSW, focusing on those aspects of it that with be at centre stage here. 2. A Very Brief Statement of the Framework As Goldman treats it, Social Epistemology (SE) is social in at least two ways: it focuses on social routes to knowledge my coming to get knowledge through another s testimony, say; and it addresses not just my (perhaps social) routes to knowledge, but how we together acquire and share knowledge, how information and various aspects of the epistemic burden is distributed among us, and so on (4). In addressing these topics, Goldman s concern is evaluative and not just descriptive: he wants to assess just which social practices are epistemically praiseworthy hence, which we should aspire to implement, insofar as they are feasible and our goals are strictly epistemic. Nor does Goldman think that such practices are to be evaluated in terms of their intrinsic features; the right approach, rather, is a consequentialist one. Specifically for Goldman, the right/good social practices are those which lead to knowledge on the part of its users, where knowledge is understood in the weak sense of true belief (5; 23ff.). Even supposing some form of consequentialism is the right one, why s that the proper evaluative standard? Because that is the over-arching epistemic goal: we want true belief, both because we are intrinsically curious creatures, and because where our beliefs make a practical difference (some don t), ceteris paribus it s true belief, not false, that helps us to satisfy our goals (e.g., 3, 69; pace, e.g., Stich 1990, Chapter 5). Nor will just any notion of what truth is do for Goldman. Specifically, he spends substantial time addressing both (in Chapter 1) veriphobes and (Chapter 2) some specific approaches to truth which he rejects instrumentalist, relativist, epistemic, and deflationary theories. Each of these approaches is problematic, Goldman argues, and none does justice to the the basic correspondence idea that what makes sentences or propositions true are real-world truth makers (68). It s this the root idea, Goldman says the

411 Goldman s Knowledge in a Social World: 411 idea that truth involves a relation to reality (59) and that idea alone, that s required for the veritistic project of KSW (68). Only a notion of truth that is broadly in the correspondence tradition, according to Goldman, does justice to our intuitive understanding of truth (42), makes sense of our having interests, both intrinsic and extrinsic, in acquiring knowledge (true belief) and avoiding error (69), and legitimizes a veritistic approach to SE. 3. Veritistic Social Epistemology and the Basic Correspondence Idea Now, there are no doubt those who, in spite of Goldman s arguments in Chapter 1, remain sceptical of the very notion of truth; and those too who, in spite of his arguments in the early sections of Chapter 2, would remain unconvinced that truth is not best understood in, say, pragmatic, epistemic, or verificationist terms. For myself, though, I agree that a proper SE should go beyond mere doxology and sociological observations; and I think an SE divorced from any interest in (at least) true belief is arguably not a social epistemology at all. Further, I agree with Goldman that pragmatic, epistemic, and verificationist theories of truth are deeply unsatisfactory: The problem is not just that such theories face various technical objections; rather, it s that they seem simply to get off on wrong track, confusing from the outset truth with sui generis things: on the face of it, utility and truth for example needn t coincide (there are many useless truths, and many useful falsehoods). So too for truth and justified/warranted belief: many of the things we justifiedly believe are false; and there are many truths which, because of our present ignorance, we would not be justified in believing. Moreover, as Goldman points out, many of these views seem actually to require an antecedent handle on the notion of truth: Thus, though justified belief and truth seem to come apart, it might be suggested that this isn t so of truth and what would be justifiedly believed in the ideal epistemic situation. Arguably, though, the notion of truth is needed to define (ideal) justification; if truth-conducivity isn t built into the understanding of (ideal) justification, however, we re back to the problem of divergence for why suppose that what we d be justified in believing, even in the limit, would be true? To Goldman s mind, what the failure and specific way(s) of failing of such approaches to truth suggest is the need for some sort of correspondence theory. Once again, we have both an intrinsic and an instrumental interest in acquiring true beliefs. What does that amount to? Well, the natural sugges-

412 412 Patrick Rysiew tion is something like this: to say that we re intrinsically curious creatures is to say that we have a natural desire to understand the world; and to say that our satisfying our goals depends on our navigating successfully through the world, is to say that we have a practical interest in believing things which track the way the world is. So our interest in truth is an interest in the way things actually are, the way reality really is. And it s because they fail, ultimately, to conceive of truth in this way that epistemic and instrumentalist theories (e.g.) prove unsatisfactory. But this is just to say that the failure of these views points to the need for a general relation like correspondence to reality (59) as the proper basis of a theory of truth. Goldman s own version of the correspondence approach, his descriptive success (DS) theory, runs thus: (DS) An item X is true if and only if x describes reality as being a certain way and reality is that way (59). Goldman is quite up front in allowing that as it stands (DS) is best regarded as a sketch of a theory within the correspondence tradition (41, 59; emphasis added): a full theory on this topic will have to answer such questions as what determines the contents of such (putative) descriptions of reality, just what the worldly truth-makers are, and what it means for reality to be the way it is described (cf. 61-3). Answering such questions as these, however, takes us beyond the scope of Goldman s Chapter 2; for the intent of that chapter is merely to motivate and defend the basic correspondence idea that what makes sentences or propositions true are real-world truth makers. The tenability of this basic idea, Goldman says, is all that is required for the veritistic epistemology [developed in KSW] (68). And it s only a correspondence theory of truth that accommodates this basic correspondence idea. But is that so? Again, I think that Goldman s criticisms of epistemic, instrumentalist, etc., theories of truth are on the mark, and that they don t respect the basic correspondence idea. (At any rate, I won t be arguing to the contrary here.) What s less clear to me is whether Goldman s own preferred approach to truth is, in this regard, really superior to deflationary views. Of course, there are different theories that get grouped under the general deflationist heading. As Goldman puts it, though, what they have in common what makes them all deflationist is that they all involve denying that sentences using the word true involve the predication of a property (or a substantive property) to propositions or anything else (51). Thus, on Quine s disquotationalism, attributing truth to the sentence, Snow is

413 Goldman s Knowledge in a Social World: 413 white, is no different from attributing whiteness to snow; the adjective true is dispensable when attributed to sentences that are explicitly before us (1987: 214). Where true really earns its stripes is when we want to talk about sentences not available to us ( The final sentence Churchill uttered was true ), when it s too cumbersome to actually token them ( The longest sentence ever uttered is true ), or when we want to generalize over sentences ( Every sentence of the form, P or not-p, is true ). Even so, for Quine the distinctiveness of true is not that it gives us, so to speak, something new to say a way of picking out some substantive property over and above, say, snow s being white, etc.; rather, it gives us a new (sometimes better) way of saying what we already could (in principle, anyway), by replacing talk of the world with logically equivalent talk about words (Williams 2001: 141). In the same way, on Horwich s (1990) minimalist theory (MT), the correct theory of the meaning of true consists in all instances of the equivalence schema, (T) <P> is true iff P And a person s understanding of the truth predicate consists in his being disposed to accept any instance of (T). Horwich thinks that this minimal theory suffices to explain what needs explaining. For instance, it explains why, from the sentences, (1) Everything Cassie says is true, and (2) Cassie says that snow is white, it follows, given (MT), that (3) Snow is white. So, he thinks, there is no need to search for any deeper theory of truth (which is just as well, Horwich thinks, since such a search would fail to turn up anything). Now, I m not going to argue that either of these deflationary views is correct; while I m not sure that the situation is quite as dire as Goldman makes it out to be (54-9), both Quine s and Horwich s views face real difficulties. The present point is that it s not clear that Goldman s (DS) does a better job than either of these theories of respecting the basic correspondence idea that what makes sentences or propositions true are real-world truth makers (68).

414 414 Patrick Rysiew In the final pages of his discussion of truth (66-8) Goldman suggests the following way of making his (DS) compatible with a deflationary view of truth (supposing that any technical problems facing the latter could be overcome): As Goldman says, there are a number of distinct projects that might cited under the heading, a theory of truth. Among these are: attempts to give the meaning of a specific piece of language, true ; and attempts to say something about the nature of truth itself (41, 66; cf. Horwich 1990: 37-8 and Kirkham 1992: Chapter 1). Now, Horwich s or Quine s view might turn out to be the correct semantic theory the correct theory of the predicate, true. The deflationary aspect of these views consists in their authors declining the invitation to give a theory of the nature of truth itself; and on that matter it may turn out that (DS) is the right way to go. Hence: [E]ven if some form of deflationism can surmount its obstacles and be rendered fully attractive, this would not force us to relinquish the basic correspondence idea that what makes sentences or propositions true are real-world truth makers. (68) Surely, though, the deflationist will respond that the only reason Goldman s (DS) doesn t face such troubles itself is because it is too much a sketch of a theory of truth. Goldman writes that the crucial feature of the correspondence theory that distinguishes it from its competitors is its claim that the truth concept involves truth-makers: worldly entities of some sort that make propositions (or other truth-bearers) true (61). But Horwich, for example, regards the thought that (crudely put) nothing is true but that the world makes it so to be just about the only [sort of] uncontroversial fact to be found in discussions of truth (1990: 126); and he thinks that this fact is just one of a family of closely related ideas which grow out of the innocuous idea that whenever a sentence or proposition is true, it is true because something in the world is a certain way something typically external to the sentence or proposition (ibid.: ). ((DS), on this view, is merely a harmless generalization from instances of (T).) Quine too thinks that the correspondence theorist is right that truth should hinge on reality, and it does. No sentence is true but reality makes it so (1971: 10). According to Goldman, deflationists saying such things is music to the ears of a correspondence theorist (67); but should it be? As I read it, the disagreement between Goldman, on the one hand, and Horwich and Quine, on the other, is not over the basic correspondence idea that what makes sentences or propositions true are real-world truth makers (68). The disagreement, rather, concerns the prospects of turning this basic correspon-

415 Goldman s Knowledge in a Social World: 415 dence idea into something worth calling a theory of truth. For (DS) to represent an explanatory advance over the (uncontroversial) basic correspondence idea, the deflationist is liable to say, Goldman needs to provide (i.a.) theories of descriptive content and descriptive success (of a content s fitting reality ); and he must do so without invoking the notion of truth (cf. Horwich 1990: 113). But don t we need a correspondence theory of truth in order to do justice to our interest in the truth both for its own sake, and because it helps us achieve our goals? Again, that s not so clear. What is uncontroversial is that, for both intrinsic and extrinsic reasons, we want it to be the case that, to a first approximation, we believe that p just in case p (where p, typically, will refer to some worldly state of affairs): we want to believe that a particular food in poisonous if it is (but, epistemically anyway, not otherwise); we want to know whether it was disease, a meteor, or some other thing that led to the extinction of the dinosaurs; etc. Uncontroversial too, when taken on its own, is the summarizing of all of this with the claim that we have both an intrinsic and an extrinsic interest in the truth, where it s presumed that this interest is an interest in having our beliefs be appropriately connected with reality. The controversy, once again, is whether this claim is one the vindication of which rests upon discerning the nature of truth (along correspondence lines), or whether we should say instead that that which it summarizes needs no vindicating in the first place. It bears emphasizing that none of this undercuts the larger project of VSE, or Goldman s specific version thereof. That, in fact, is the point: insofar as the tenability of [the basic correspondence idea] is all that is required for the veritistic epistemology [developed in KSW] (68), that that idea doesn t favor a correspondence theory of truth over deflationism (supposing it doesn t) hardly threatens VSE. On the contrary, inasmuch as we ve yet to be presented with a complete and satisfactory correspondence theory, that it doesn t require such a theory actually brightens the prospects for Goldman s VSE. 4. Knowledge, Justification, and the Pursuit of Truth If the aspiration to truth strikes some as immodest, a notable feature of Goldman s KSW is that, in another respect, it s strikingly modest: contrary to how knowledge is usually treated within the North American analytic tradition, Goldman treats of knowledge (only) in the weak sense of true belief. For a couple of reasons, this is notable, and possibly worrisome.

416 416 Patrick Rysiew First, of course, there is the fact that most epistemologists, including Goldman in his earlier work, take it that, while knowledge (in the sense of concern to them) requires true belief, it s not constituted thereby: something more justification, warrant, or what have you is required. So one might worry that Goldman s KSW marks a departure, not just from traditional, individualistic epistemology (include Goldman s own), but from a concern with genuine knowledge. But there is a second reason why Goldman s focusing on mere true belief as the chief epistemic end might be regarded as problematic. Some of those who ve been most critical of KSW have complained that Goldman doesn t really do justice to the real spirit of social epistemology at all. Here is how Martin Kusch puts the point: Goldman agrees that traditional epistemology is excessively individualistic. But he thinks this shortcoming can be remedied simply by adding a social wing to the old edifice. (2001, 188) Why think that adding a new wing won t do? Why think that the edifice itself needs renovation? Well, for some anyway, it is because social factors figure in the very truth conditions of statements of the form, S is justified in believing that p. Thus, e.g., Stew Cohen (1987) has argued that knowledge itself has a social component. On Cohen s view, knowledge entails having good reasons; but how good one s reasons must be in order for one to possess knowledge depends upon social (intersubjectively determined, and variable) standards. Similarly, Helen Longino (1994) argues that a scientific belief is justified (and so a candidate for knowledge) to the extent that it results from the application of objective methods, where objective methods are explicitly characterized in terms of certain features of the relevant social practices whether they there are genuinely public, allow for critical interaction, exhibit equality of intellectual authority among diverse perspectives, and so on (1994, 153). On either of these views, knowledge itself in general, and not just when it comes to SE is deeply social. To be clear: this is not for ontogenetic reasons because, as a matter of fact, the conditions which enable us to come to know include social conditions but because knowledge requires justified/warranted belief, and the truth conditions for a statement of the form, S is justified in believing that p, make ineliminable reference to social factors or conditions. One might worry, then, that in focussing on knowledge in the weak sense of true belief, Goldman has effectively side-stepped what has seemed to some to be a if not the route to an epistemology that s deeply social, much more deeply social than Goldman s KSW makes it out to be.

417 Goldman s Knowledge in a Social World: 417 Having raised some worries about Goldman s taking knowledge in the weak sense to be his target, let me say that I think they re actually misplaced. Not because there aren t these problems with focussing merely on true belief, but because, and contrary to what Goldman himself suggests, there is a sense in which it is incorrect to think that his concern is really just with weak knowledge. Goldman wants, recall, to evaluate social practices in terms of whether they produce true belief(s) versus false belief, or no belief in their users. However, that a practice just happens on a given occasion to issue in someone s, even most people s, having a true belief won t suffice that would be a fluke; our (intrinsic and extrinsic) interest in the truth is best served by practices that tend to produce (/sustain) true belief. When Goldman says that the main concern for veritistic [social] epistemology is: Which [social] practices have a comparatively favourable impact on knowledge as contrasted with error and ignorance? (5), he is speaking of the propensities of various practices to produce knowledge (versus error or ignorance) (91). But this just means that there is a reliability requirement implicit within Goldman s VSE: for reliability just is the tendency of a practice (/process/etc.) to produce a high ratio of true beliefs to false. And Goldman, famously, has argued that the reliability condition is what separates (mere) true belief from (genuine) knowledge (e.g., Goldman 1976), and that it constitutes the proper justification condition on knowing (e.g., Goldman 1979, 1988). Of course, both of these claims that reliabilism is the right approach to knowledge, and that it s the right approach to justification are controversial. But they re not controversial for Goldman! So it s not at all clear that, by Goldman s own (reliabilist) lights, the knowledge with which he s concerned in KSW is knowledge in the weak sense of, simply, true belief. Granted, in the case of social processes, the relevant practices/processes won t always supervene simply on properties of the subject; in picturesque terms, they won t always be inside the subject s head. But that s only to be expected, given that they are social processes; their reliability (or not) is unaffected by their location. Does the presence of this justification condition make knowledge itself interestingly social? Does it show that the edifice of traditional epistemology needs serious reworking? Not that I can see. The feature of, say, Cohen s view that makes knowledge itself interestingly social, is that the truth conditions of S knows that p make explicit mention of social facts: S doesn t know that p (/S s belief that p isn t justified) if socially determined standards aren t met. Whereas, in Goldman s case, the requirements for knowledge (/justified be-

418 418 Patrick Rysiew lief) with respect to beliefs arising from social processes is the same as they always are S doesn t know that p (/S s belief that p isn t justified) unless that belief is produced (/sustained) by reliable processes. Granted, none of the arguments of KSW require that reliabilism be correct. What making explicit the implicit reliabilist condition explicit does do, however, is to show that KSW may be much more continuous, both with Goldman s own previous work and with the epistemological tradition, than Goldman himself suggests. Even so, Goldman does wish to avoid taking his target in KSW to be anything other than weak knowledge. He gives two reasons for this. First, because he wishes to speak to SE, and wants to avoid getting drawn into the tangled epistemological topic of what constitutes knowledge in the strong sense. More importantly, Goldman says, his view is that people s dominant epistemic goal is to obtain true belief, pure and simple (24). Of course, the usual route to true belief might be the obtaining of evidence, say; but the value of such evidence is wholly instrumental: we want it only because (we think) it will lead us to true belief (ibid.). However, one can be wholly sympathetic to Goldman on both of these points, and still worry that there is a problem lurking here. Above, we saw that classical epistemology is concerned with knowledge, in the sense of (at least) justified/warranted belief. Nor is the inclusion of the justification condition gratuitous: for even if all we really care about, in the end, is correct information (true beliefs) about the world, and even if that s all that our getting by in the world requires, justification is supposed to be what helps get us there. As Goldman says, a even the guiding question of classical epistemology is, How can an individual engage in cognitive activity so as to arrive at true belief and avoid false belief? (2001b: Section 2). However, as critics of epistemic externalism are constantly pointing out, injunctions such as, use reliable methods (e.g.), are no help at all if what you re wondering about is how to maximize truth and minimize falsehood among your beliefs. Hence at least one common rationale for the inclusion of the justification condition: it constitutes a more tractable, proximal goal; it s often hard to know when we ve got the truth, but less hard to know when we ve got justified belief; and if all goes well, in pursuing justified belief we re getting closer to knowledge as well. Thus motivated, the need for justification is the need for (what s usually called) an internalistic construal thereof. Whereas, insofar as Goldman has an implicit justification condition at work in KSW, it s reliabilist (hence, externalist). Not that reliabilism is therefore the wrong account

419 Goldman s Knowledge in a Social World: 419 of knowledge, or even of justification. Goldman can say that the demand to know whether we ve got knowledge, or to know how to get more, is metaepistemological, and shouldn t affect our understanding of knowledge (/justification) itself (cf. 25, note 16). Fair enough; but whether or not we see an internalistic sort of justification as required for knowledge per se, or as something that is independently desirable, one might insist that it is desirable nonetheless. Returning now to KSW, one common/natural worry about Goldman s VSE, even among veriphiles, concerns its feasibility that is, our ability actually to employ veritistic norms. For suppose we agree that it s (reliably produced) true belief we want from our socio-epistemic practices; how exactly, and how often, are we actually going to be in a position to determine how well we re doing on this score? This worry, note, is not an especially sceptical or pessimistic one (cf. 80-1): what s at issue is not whether our faculties, in general, issue in true beliefs surely they do ( there goes a squirrel ); and surely, when they don t, we re often in a position to see that things have gone wrong ( I was wrong it was a bird ). Often, though, especially as concerns, not the methods and results of simple perception but, say, those of scientific inquiry, it s hard to know how well we re doing, veritistically speaking; here, we encounter veritistic failure as often as success; and the discovery of such failure tends to be backward-looking we tend to see that we were wrong in either our beliefs or our methods only once we ve moved on to different ones. Goldman addresses this sort of concern in Chapter 3, giving examples in which subjects are able to select the veritistically best practice, even when they start out using different means of choosing among the available alternatives (81-2). As Goldman notes, however, it s a feature of these examples that the subjects know what the prior probability that a candidate practice (/expert) will get it right is; and this, Goldman admits, is a condition that may be met only infrequently (82, note 8). Indeed, he writes that the problem of determining the performance of various practices, both actual and possible, across a wide range of prospective circumstances which is precisely what measuring veritistic value is all about makes [t]he implementation of veritistic epistemology extremely difficult (91; emphasis added). Still, Goldman says, [m]y measures of V-value are intended to provide conceptual clarity, to specify what is sought in an intellectually good practice, even if it is difficult to determine which practices in fact score high on these measures. Conceptual clarity about desiderata is often a good thing, no matter

420 420 Patrick Rysiew what hurdles one confronts in determining when those desiderata are satisfied. (91) Fair enough. But we re still left with the question of how actually to proceed the next time we re deciding what views or methods to adopt (cf. Maffie 2000: 250). This concern the concern about implementing VSE suggests, to my mind, that a veritistic SE stands in need of supplementation by a procedural one. Consider, say, Longino s suggestion that scientific communities and their activities should include a shared standards that critics can invoke, a general responsiveness to such criticism, and a rough equality of intellectual authority among qualified practitioners (78). Now, as Goldman observes, in discussing this (and other) more procedural conceptions of SE, it s not clear that the features Longino s criteria pick out are intrinsically good, as opposed to being good because (we think) they promote good veritistic outcomes. But no matter: we don t have to choose between procedural and consequentialist (specifically, veritistic) means of evaluating practices; and there are going to be cases where, though we hope the meeting of the procedural criteria will have good veritistic results, we really can t be sure, and will have to evaluate candidate social practices solely in terms of these more tractable proximal goods. Putting it another way, there are likely cases in which we ll have no choice but to adopt a justification-centred approach with the proviso, of course, that given the source of the present concern it won t help to have the justification in question be externalist in character. This leaves the question of how far any of this is strictly incompatible with what Goldman says in KSW. In Chapter 3, he presents proceduralism as though it were necessarily a competitor to veritism. Yet, if Longino s proceduralist view ends up being impure, perhaps so too does Goldman s veritistic approach. For instance, some of the features of science that Goldman takes to be veritistically significant - e.g., that there be in place systems of credit and reward, and for the critical assessment of theories and findings (see 250, Chapter 8) are quite like Longino s criteria for good scientific exchange. (Though they, in turn, might be well-complemented by consensus considerations another of the alternative conceptions of SE Goldman discusses in Chapter 3; for our shared verdict on the matter of whether those criteria have been fulfilled might be all we have to go on in the end.) And one might say that the rules for good argumentation that Goldman puts forward (Chapter 5) are excellent examples of good (nonexternalistic) socio-epistemic norms, and perhaps a model of a practice that can be applied to the very question of whether a given social practice is

421 Goldman s Knowledge in a Social World: 421 veritistically good (when we lack any direct knowledge thereof). (Here, such argumentation would constitute justification in the sense of the right kind of public, inter-personal justifying of beliefs.) Of course, for Goldman, both Longino s criteria and good rules of argumentation derive their epistemic value from their tending (we think) towards true belief. The issue, though, is whether, in a given instance, that feature of a candidate practice will be available to us. If not, then in discussing a feasible VSE, perhaps justification and more procedural considerations should have a place at, or near, the centre of KSW. Then again, insofar as Goldman s book contains both an implicit externalistic justification requirement, and at least the resources for an account of the proper role of internalistic justification in VSE, perhaps they already do. Acknowledgements An ancestor of this paper was presented at the Annual Congress of the Canadian Philosophical Association (Halifax, 2003), as part of a symposium on KSW organized by Yves Bouchard and David Matheson. The author also acknowledges the support of the Social Sciences and Humanities Research Council of Canada, and the Peter Wall Institute for Advanced Studies. References Alston, William (1986) Internalism and Externalism in Epistemology. Philosophical Topics, Vol. 14, No. 1. Reprinted in Epistemic Justification (Cornell University Press, 1989). (1996) A Realist Conception of Truth; Cornell University Press. Cohen, Stewart (1987) Knowledge, Context, and Social Standards. Synthese 73: Goldman, Alvin (1976) Discrimination and Perceptual Knowledge. The Journal of Philosophy 73: Reprinted in Goldman 1992: (1979) What is Justified Belief?, in Justification and Knowledge, edited by George Pappas. Reprinted in Goldman 1992: (1986) Epistemology and Cognition; Harvard University Press. (1988) Strong and Weak Justification. Philosophical Perspectives, Volume 2. Reprinted in Goldman 1992: (1989) Psychology and Philosophical Analysis. Proceedings of the Aristotelian Society 89: Reprinted in Goldman 1992:

422 422 Patrick Rysiew (1992) Liaisons: Philosophy Meets the Cognitive and Social Sciences; MIT Press. (1994) Naturalistic Epistemology and Reliabilism. Midwest Studies in Philosophy, XIX: (1999) Knowledge in a Social World; Oxford University Press. (2001a) Social Routes to Belief and Knowledge. The Monist 84: Reprinted in Goldman 2002a. (2001b) Social Epistemology, in Stanford Encyclopedia of Philosophy, on-line at: (2001c) The Unity of the Epistemic Virtues, in Virtue Epistemology, edited by Fairweather and Zagzebski; Oxford University Press. Reprinted in Goldman 2002a: (2002a) Pathways to Knowledge: Public and Private; Oxford University Press. (2002b) What is Social Epistemology? A Smorgasbord of Projects, in Goldman 2002a: (2002c) Précis of Knowledge in a Social World, Philosophy and Phenomenological Research, Vol. LXIVC, No. 1: Horwich, Paul (1990) Truth; Basil Blackwell. Kirkham, Richard L. (1992) Theories of Truth: A Critical Introduction; MIT Press. Kusch, Martin (2001) A General Theory of Societal Knowledge? Aspirations and Shortcomings of Alvin Goldman s Social Epistemology, Studies in the History and Philosophy of Science, Vol. 32, No. 1: Longino, Helen (1994) The Fate of Knowledge in Social Theories of Science, in Schmitt, ed. (2002) The Fate of Knowledge; Princeton Paperbacks. Maffie, James (2000) Alternative Epistemologies and the Value of Truth, Social Epistemology, Vol. 14, No. 4: Nelson, Lynne Hankinson (1993) Epistemological Communities, in Feminist Epistemologies, edited by Linda Alcoff and Elizabeth Potter; Routledge. Quine, W.V. (1970) Philosophy of Logic; Prentice-Hall. (1987) Quiddities; Harvard University Press. Schmitt, Frederick F. (1994) Socializing Epistemology; Rowman & Littlefield. Shapin, Steven (1994) A Social History of Truth. Chicago: University of Chicago Press. Stich, Stephen P. (1990) The Fragmentation of Reason; MIT Press. Talbott, William (2002) The Case for a More Truly Social Epistemology, Philosophy and Phenomenological Research, Vol. LXIVC, No. 1: Williams, Michael (2001) Problems of Knowledge: A Critical Introduction to Epistemology; Oxford University Press.

423 Is Terrorism Globalizng? 423 Is Terrorism Globalizng? Albert J. Bergesen Is today s global terrorism a relatively recent phenomena, as some argue globalization in general is, or have been earlier waves of international terrorism, that is, earlier globalizations of terrorism. To try and answer this question some general similarities between the social characteristics of contemporary international terrorism since 1968 are compared with the assassinations and bomb throwing terrorism associated with anarchist unrest at the end of the 19 th century, 1870s This is followed by some general comparative questions one would want to answer to b e able to determine if, in fact, terrorism is globalizing and whether it has gone through a similar globalization period earlier. Finally, a technique is proposed that allows researchers to measure the degree of globalization of terrorist events over time in these two time periods. Before turning to the comparison of international terrorism since 1968 with the 1870s-1914 wave of largely anarchist terrorism, the complexity of the meaning of terrorism requires a discussion of how it is to be defined. Realizing that one person s terrorist is another s freedom fighter, and that states as well as individuals and groups engage in terrorism (Chomsky, 2001), I seek as clear a definition of terrorism as possible. In this regard, much of the early work in terrorism research centered upon definitions (see the discussions in Jenkins, 2001; Cooper, 2001; Gibbs, 1989; Hoffman, 1999; Ruby 2002; Schmid and Jongman, 1988; Senechal de la Roche, 1996, 2001), but as Jenkins (2001) notes there seems to be a consensus emerging on a definition of terrorism. For example, Enders and Sandler (2002) argue terrorism involves a focus upon underlying political, social or religious motives, as its violence is separable from that of just a crime or personal vengeance or an act of someone mentally deranged, although over the centuries all these labels have been applied to terrorists. The act itself is also tied to intimidation or attempting to influence an audience of some sort which is often not that of the terrorist victims/targets themselves, such that terrorism is most often directed toward non-combatants or civilians, and is random so that everyone feels at risk. For them, Terrorism is the premeditated use or threat of use of extranormal violence or brutality by subnational groups to obtain a political, religious, or ideological objective through intimidation of a huge audience, usually not directly involved with the policy making that the terrorists seek to influence (Enders and Sandler, 2002: ). The US Department of

424 424 Albert J. Bergesen State defines terrorism as politically motivated violence perpetrated against noncombatant targets by subnational groups or clandestine agents, usually intended to influence an audience (quoted in Ruby, 2002: 10). This is similar to Chomsky s (2001:19) definition: Terrorism is the use of coercive means aimed at populations in an effort to achieve political, religious, or other aims. I will adopt Stern s (1999: 30) definition, which is similar to those above. For her, terrorism is, an act or threat of violence against non-combatants, with the objective of intimidating or otherwise influencing an audience or audiences. This conception leaves open questions of motivation and ideology, which is important in research that involves different historical periods and needs to be flexible enough to include, say, the ideology of anarchists and social revolutionaries at the end of the 19 th and fundamental Islamic beliefs at the end of the 20 th century. I also realize the importance of state terrorism (Oliverio, 1998; Chomsky, 2001) but this particular analysis will only focus upon subnational groups and individuals. Changes Since 1968 Following incidents such as the bombing of the World Trade Center (WTC) in 1993, US embassies in Africa in 1998, and the attacks on the Pentagon and WTC in 2001, the conventional wisdom of researchers and commentators on terrorism was that the world had entered a new phase since the 1990s that departed dramatically from what went before. It was variously called the new terrorism (Jenkins, 2001; Lesser et. al. 1999), or spoken of as involving new types of post-cold war terrorists (Hudson, 1999: 5), or a new breed of terrorist (Stern, 1999: 8), or new generation of terrorists (Hoffman, 1999), or terror in the mind of God (Jurgensmeyer, 2000), or a clash of fundamentalisms (Ali, 2002), or simply a new wave of terrorism (Rapoport, 1999, 2001). In general, the argument has been that terrorism is changing in the following ways: (1) Organizational shift toward an international/network form. Newer terrorist organizations have moved away from the earlier model of professionally trained terrorists operating within a hierarchical organization with a central command chain, toward a more loosely coupled form of organization with a less clear organizational structure. Similarly, whereas from the 1960s through the 1980s groups were more clearly nationally bound (German, Japanese, Italian, Spanish, Irish, Palestinian, etc.) more recent organizations

425 Is Terrorism Globalizng? 425 like Al Queda have members from multiple nationalities and organizational sites outside the leadership s country of origin. (2) Organizational identities more difficult to identify. Terrorist organizations also seem to identify themselves or claim responsibility for specific acts less often, such as the bombing of the US Embassies in Africa or the events of September 11 th, which while purportedly organized by Bin Laden and Al Queda, were never clearly claimed by that organization. This is in contrast with earlier terrorist organizations which were, it is argued, much more clear in taking responsibility for their actions and defining who they were, often with elaborate radical political ideologies. (3) Terrorist Demands have become more vague and hazy. Earlier terrorist demands were often quite specific: Black September demanded the release of comrades when they attacked Israeli athletes at the 1972 Munich Olympics; the IRA want the British out of Northern Ireland, the PLO wants the Israelis out of the West Bank, and Basque independence is a clear desire of ETA. Similarly the Marxist/radical/left logic behind earlier attacks on businessmen, diplomats, and so forth, were also well known. In contrast there were no demands surrounding the 1998 US Embassy attacks, or the events of September 11 th. (4) Ideological shift from political to religious motives. What has been called the new religious terrorism, or holy terrorism, reflects the increasing prevalence of religion in the ideology of terrorist organizations, with the most notable being Islamic fundamentalism, or Political Islam, but also Christian fundamentalism (anti-abortion terrorism), Messianic Zionism (behind Yitzhak Rabin s assassination) or the religious sect, the Aum Shinrikyo, a Japanese terrorist group that released poisonous gas in a Tokyo subway in There also seems to be an increase in groups with more vague, millennial, and religious ideologies than earlier radical groups such as the German Red Army Faction, the Italian Red Brigades, or the Japanese Red Army. (5) A more global dispersion of targets/victims. Where international terrorism of the 1960s and 1970s was largely centered in Europe and the Middle East, by the 1990s it had moved to Africa (1998 attack on US Embassies), Argentina (1992 truck bombing of Israeli Embassy and 1994 bombing of a Jewish Community Center) and the US mainland proper (1993 WTC bombing and 2001 WTC/Pentagon suicide airplane attacks). (6) Violence becomes more indiscriminate. Along with a geographical dispersion of targets, there seems to be a move away from specific targets, as for instance when hundreds of civilian Kenyan and Tanzanian embassy em-

426 426 Albert J. Bergesen ployees and passers by were killed to achieve the objective of bombing the US embassy. The 1993 and 2001 attacks of the WTC were also examples of more indiscriminate targets as opposed to earlier skyjacking of a national airline s plane in order to attain specific demands, or the kidnaping a particular politician (such as Aldo Morro by the Italian Red Brigades). If one reflects for a moment upon these changes, many of them suggest the process of globalization (Sassen, 1998; Tomlinson, 1999; Sklair, 1995; Tilly, 2002c; Robertson, 1992; Hoffman, 2002; Barber, 2001; Castells, 1993; Boli and Thomas, 1999). That is, arguments about terrorist organizational forms becoming more network like and multinational in locale and membership, remind of ideas about new global network forms of business organizations. Arguments about ideologies and demands becoming less nationally specific, reminds of theories of how globalization involves cultural deterritorialization (Meyer, et. al., 1997; Tomlinson, 1999). Finally, arguments about a growing dispersion and indiscriminateness of terrorist violence also expresses a disregard for national boundaries and, as such, a growing global, as opposed to national, character of terrorism. But is this just a product of recent international conditions, as some have argued about trade and cultural globalizations, or, as others have argued about globalization in general, does history show that there are, in fact, other periods of globalizations (Phillips, 2002; Ferguson, 2001a; Chase-Dunn, et. al., 2000; Arrighi, 1994; Arrighi and Silver, 1999; Davis, 2001). And if that is so, might there be other periods of globalizations of terrorism? From an historical perspective terrorism seems endemic to organized social life, as it has appeared, and reappeared, throughout history. Waves of terrorism have been documented in the first century CE with the Zealots-Sicarii, a Jewish group involved in assassinations and poisonings of Roman s occupying Palestine, and with the Assassins, who operated in 11th-13th century Persia and Syria, assassinating political and religious leaders (Lacqueur, 1997, 1999; Stern, 1999: 15). The modern meaning of the term terrorism is associated with the Reign of Terror during the French Revolution of the 18 th century, and Rapoport (1999, 2001) speaks of modern terrorist waves since the 1870s, with the first being the one associated with anarchists and social revolutionaries in the late 19 th century.

427 Is Terrorism Globalizng? 427 Anarchist Terrorism, 1870s-1914 This well documented historical period of terrorism ran from roughly the 1870s up to the First World War. Associated with the idea of propaganda by deed, Russian, Italian, Spanish, French, American, Serbian, Macedonian terrorists were involved in a period of assassination and bomb throwing from the Russian and Ottoman Empires to the east though the Austrian Empire and Western Europe to the United States on the west. There were well known individual terrorists (Emile Henry, Ravachol) and terrorist groups. In Serbia, the Black Hand; in Russian Narodnaya Volya, or People s Will; among Slovenes, Croats, and Serbs, the Young Bosnians and the Narodna Obrana, or the People s Defense (Hoffman, 1999). Terrorists from one country also killed people from another. In 1892 the Italian Angiolillo assassinated the Spanish prime minister; in 1894 the Italian Santo Jeronimo Caseiro assassinated French president Carnot; in1898 the Italian Luigi Luccheni assassinated Empress Elizabeth of Austria, and in 1914 the Bosnia Serb Princip assassinated the Austrian Archduke Francis Ferdinand (Joll, 1964). While the contemporary period is known as one of international terrorism there are clear grounds for considering the anarchist period as one which also had international or global aspects in that terrorism appeared in different parts of the world and involved crossing national boundaries for many attacks. This possible similarity with today s terrorism has not gone unnoticed, as there are scholars who suggest that recent terrorist events have had precursors, whether as individual terrorists (the Sudanese Mahdi in the 1880s and Bin Laden today), terrorist organizations (The Black Hand and Al Queda), or reactions of public officials (T. Roosevelt after a terrorist attack declaring in 1901, the anarchist is the enemy of all mankind and G.W. Bush after the 9/11 attack, blaming the evil-doers ). More specifically, Ferguson (2001b) and Gray (1996) make comparisons between figures like Bin Laden and late 19 th century Russian terrorists, with Ferguson (2001b:119) going on to caution against not taking seriously the similarity of the pre-1900 terrorist wave with today s wave. He notes similarities in the political religion of their ideologies, the diasporic or transnational nature of both sets of terrorists who often resided and planned attacks abroad, and the similarity of global political economic conditions at the end of the 19 th and 20 th centuries. Schweitzer (2002) argues that the major powers in 1900 would have considered giving terrorism top billing on their agenda as had President Clinton at a Group of Seven meeting after the bombing of a US compound in Saudi Arabia, and that while 19 th century

428 428 Albert J. Bergesen anarchist assassinations were often done by individuals they were part of a larger movement of their time which the more general public was fearful, much like contemporary reactions to terrorism. Chalmers Johnson (1990) has noted a similarity in today s terrorist reactions to American empire to earlier terrorist attacks against the Tsarist, Ottoman, and Habsburg Empires at the end of the 19 th century. If Al Queda is a reaction to American empire, as Johnson argues, then one could see earlier terrorist resistance in the form of pre-1914 terrorist groups attacking the empires of their day (the Serbian Black Hand vs. the Austrian Empire; Inner Macedonian Revolutionary Organization vs. the Ottoman Empire, and the terrorists of Narodnaya Volya vs. the Tsarist Russian Empire). In the case of fundamentalist Islamic terrorism, a comparison with the Sudanese revolt of the Mahdi in the 1880s against the British Empire and Bin Laden against the US, has been made by Ferguson (2001b: 124), who also notes that 19 th century Sudan would probably have been considered a rogue state in its time. Paul Kennedy (2001: 56) notes a similarity between the hatred of London as the financial center of world capitalism at the end of the 19 th century and the hatred by fanatical Muslims today of the dominance of Wall Street and the Pentagon. Similarly Davis (2001) links late 19 th century globalization of trade and finance to outbreaks of terrorist activity in the developing world, and Brooks and Wohlforth (2002: 30) note a similarity between the growing multipolarity of the great powers at the end of the last century (like today) and the presence of anarchist assassinations. Comparative Research Questions This literature, then, suggests the following questions: How similar is today s terrorism with that of the 1870s-1914 period? And, are the changes observed in terrorist organization, ideology, recruitment, demands, tactics, and violence, previously characterized as the new terrorism, really instances of a globalization of terrorism? And, if so, is it also possible there was a similar globalization of terrorism during the earlier period? More specifically, we can ask three sets of questions which addresses different aspects of terrorism at different levels of analysis. The first deals with questions about the organization, ideology, demands, violence and so forth, of the individual terrorist event (the assassination, kidnapping, bombing, and so forth). The second set of questions centers upon the terrorist wave (1870s-1914 and ) as

429 Is Terrorism Globalizng? 429 the unit of analysis, looking for developmental patterns, trends, or signs of a life cycle in the sequences of event characteristics identified in the first set of questions. Finally, a third set of questions deals with whether the terrorist waves are historically unique to each period, or whether they cycle that is show repeated tendencies along the same set of dimensions or spiral, that is advance in a systematic way along certain, or all, of the structural dimensions of the cycle that were identified with the second set of questions. In a real sense this is also about the identification and understanding of social processes at different levels of analysis and how they might be interrelated or nest within each other, as the terrorist event nests within the terrorist wave, which in turn nests within the geographically larger, and historically longer, global spiral of century to century terrorism. This suggests a number of questions for each level of analysis, from the individual terrorist event to properties of the terrorism wave to properties of global cycles and spirals of terrorism. We can consider each in turn. 1. Properties of the Individual Terrorist Event The first set of questions addresses a number of properties of the individual terrorist incident that concern the basic elements of terrorist social organization, their group ideology and identity, the substance of demands, the range of tactics employed, and types of violence directed against different social and national types of targets/victims. Organizational Characteristics: What are the basic forms of terrorist organizations in these two periods? Are they more hierarchical organizations staffed by professional terrorists, or more amateur and loosely organized groupings, or more diffuse network forms of organization that link individuals and groups around the globe? Are these differences found in both periods, or only the present one? Or, contrary to what the literature suggests, will empirical evidence point to no differences at all? Ideological Content: What are the different values, rationales, and legitimations utilized by terrorists and their organizations in both the 19 th and 20 th centuries. How are they similar and how are they different with regard to social, religious, or political themes. Are they more specific or more open ended and vague? And, is there an association between different forms of terrorist organizational and ideological structures? Do politically radical

430 430 Albert J. Bergesen groups have a different organizational form than religious groups, or groups who seem to have no identity at all? Membership and Recruitment: Who is recruited, joins, or starts terrorist organizations? This has been an area of major research concern (see the exhaustive review of studies in Hudson, 1999 and Schmid and Jongman, 1988), but the issue is usually treated in isolation looking only at individual personality types or social backgrounds. We will ask about the interrelationship of different forms of terrorist organization, ideology, and membership. Are they more national or international, for instance? One would imagine that recruitment to the more politically radical groups would be different from more religiously millennial groups, but that is an empirical question. Terrorist Demands: What is the substance of terrorist demands, or do they make them at all, and, again, to utilize our other questions, are there certain kinds of terrorist organizations that make certain kinds of demands? Are some of them more concrete and specific like freeing comrades from prison or are they more generalized and non-specific. And, do these differences vary by type of organization and ideological outlook? There are also questions about demands and kinds of violence and the success or failure of a mission. That is, are more specific demands easier to meet, resulting in less fatalities, or do they require political acts by authorities that they are not willing to make, and therefore more often end in a violent resolution? Similarly, how are demands formulated, and does the structure of the organization affect the content and form of demands? Social Characteristics of Targets/Victims. Who is actually targeted for a terrorist attack? Are there some sectors of the population that are more preferred, such as politicians, business executives, bank presidents, or members of the military? And, do these social characteristics of targets vary with the organizational form, ideology, and type of demands of the terrorist group? There is also a question of region and whether targets are domestic or foreigner. So, for instance, are targets in the Middle East different from those in Europe or Asia? And if it is foreigners, are they from different social backgrounds when they are, say, traveling abroad versus being stationed in a foreign country? And if they are stationed in the country of the terrorist organization, is the social type of target different than when the terrorist travels to another country to target a foreigner there?

431 Is Terrorism Globalizng? 431 Tactics and Violence. We already have a sense of the kinds of violence involved in terrorism, from assassinations to bombings, kidnapings, and hijacks. The questions we can raise here concern which types of violence are used more often by which types of terrorist organizations, or which types of ideologies, or are different kinds of tactics dependent upon the social location of the target. Are bankers, say, kidnaped or assassinated more often than politicians? Or do groups with certain kinds of ideologies engage in certain kinds of tactics? 2. Properties of the Terrorist Wave Moving to the next level of analysis, we note that terrorist incidents bunch and cluster to form periods or waves, or outbreaks of terrorism, as in the two periods selected for the proposed research. The focus here is not upon the individual incident and its properties, but upon the overall pattern and structure of the period as a whole. For this analysis, a second set of questions are addressed to the period, looking for emergent structural and cultural properties that characterize this higher level of analysis. The comparisons shift from the structure, pattern, and organization of each event, to the structure and life cycle of each wave. And when these waves are compared, will something like a generalized life cycle of terrorist activity emerge? To guide our search for these patterns we ask the following focal questions. Geographical concentration/dispersion: During an historical outbreak of terrorism do terrorist incidents tend to cluster in certain areas, or regions of the world, either in general, or at various times in the terrorist cycle? Does terrorism appear more in advanced industrial, developing, democratic, or autocratic countries? There are numerous hypotheses in the literature which suggest various combinations of the above, making this an important question to ask. Similarly, does terrorism at certain points in the larger cycle cluster within certain types of countries, or zones of the world (developed, underdeveloped, core or periphery)? Is there an international concentration of terrorist organizations in any particular sector of the global system? In the present situation one thinks of Middle Eastern groups, but earlier in the same wave the concentration was more in Europe (the German Red Army Faction, and so forth). This raises the question of geographical shifts in centers of terrorist organization from one sector of the global system to another as the overall wave develops and unfolds. Similarly, there is the question of the dispersion

432 432 Albert J. Bergesen of terrorist events. Overtime, do they spread in any particular direction, from developed to underdeveloped states, or visa versa, and does this spread show any patterning over the life cycle of the terrorist wave? Changing social and national location of targets/victims: As a wave of terrorism progresses does the social location of the targets/victims systematically vary? Are more elites targeted in the early stages and then does the violence become more socially indiscriminate as the period progresses? Are social locations of targets (class, political, status, power, positions) ordered over time in both of these waves of terrorism? Is there something like a life cycle of targets? And, while the present period isn t as yet complete, it may be possible, if there are similarities in the progression of targets through social space, to make predictions as to the next set of targets in the future. That is an issue with obvious policy implications. A life cycle of terrorism: Taking all the information so far we can ask questions about the overall life cycle of terrorism. For instance, do terrorist waves start with one type of violence and then turn to others? Karstedt-Henke (1980) theorizes a moderate/radical split in protest organizations that results in terrorist groups, while Della Porta and Tarrow (1986) and Tarrow (1989a, b) see terrorist violence as a means of differentiation within an overcrowded social movement field. Both find a tendency toward radicalization in the later stages of a protest wave. This work can be build upon, in that as they speak of a protest wave turning more violent in its later stages, we can look for structural and developmental properties of a terrorist wave, to see if it too turns more indiscriminate and violent in its later stages. 3. Properties of Global Cycles and Spirals A third set of questions could be asked about the global condition of terrorism when both waves are considered at once in sequential order. Are there global geographical spirals of terrorism? Do terrorist events spiral outward from centers of the global system? In the first wave terrorist organizations were centered in Europe and the Ottoman/Russian/Austrian Empires, but in the second it seems to have moved eastward toward the Middle East and Central Asia in terms of points of origination, while the central target seems to have moved westward to the US. This, of course, is based on casual

433 Is Terrorism Globalizng? 433 observation, but it can be tested as a hypothesis by plotting the international location of terrorism from wave to wave to see if such a pattern emerges. The same spiral idea can be used to examine changing degrees of violence or dispersion of events across global space. Is there a spiral in the social characteristics of victims? We can also ask if there is a spiral in the width of the social structure from which victims are selected. Has their social location widened out from the upper strata of political elites in the 19 th century, who were the subjects of assassinations and bomb throwing, to broader circles of the population today? Is there an occupational, or status, or political power pattern discernable in the overall direction of target shifts between waves, or is it more cyclic, in that the same social locations appear in both centuries? And, finally, is there no pattern at all? Are things random or so scattered that it is impossible to discern any cycling let alone spiraling? An Indicator of Globalization Finally, let me propose a coding scheme to measure the degree of globalization of terrorist incidents, that builds upon Chase-Dunn et. al. s (2000) technique to measure trade globalization. The Home/Away Technique. This technique is quite simple, actually, and can be applied to terrorist events recorded in both time periods. What it does is to code each event in terms of whether the terrorist (individual or organization) and the target/victim are either at home (within the country or region where they reside) or whether they are away (abroad; not in their home country/region). Each terrorist event recorded in both data sets will be coded as to whether the terrorist perpetrator is home or away, and whether the victim/ target is home or away. This creates a simple 2 by 2 table with 4 cells. When the terrorist is home and the victim is home (TH/VH) this is the least international or globalized form of terrorist incident. Anwar Sadat being assassinated by Islamic militants in Egypt would be an example. Where the terrorist is at home and the victim is away (TH/VA), we would have a more international incident. This would include events such as the bombing of the US Marine barracks in Lebanon. Where both terrorist and victim are away (TA/ VA) one would have an indicator of an even more globalized terrorist incident (the irrelevance of national boundaries). An example here would be the

434 434 Albert J. Bergesen bombing of US Embassies in Africa by Al Queda. Finally, there is the case where the terrorist is away and the victim home (TA/VH), as in September 11 th. With this simple code we get a progression from attacks within the home country on home targets, to being at home but attacking foreign targets, to being away and attacking foreign targets away, to finally being away and going to the home country of the target for the attack. This procedure has already been pre-tested on a sample of 135 events in the period and 24 events in the 1870s-1914 period, and preliminary results suggest that there is a tendency for home/home events to appear earlier, on average, than away/ away and away/home events, lending tentative support to the hypothesis that there is globalization going on within both terrorist waves. Terrorist Event Analysis: What is suggested here is a research program where terrorism is conceptualized and studied as its own form of collective violence, with its own structures and dynamics. At present empirical research on terrorist events occurs primarily within the social movements and collective violence literature, where terrorism is usually considered an instance of a more general type of collective violence. Gurr (1990) for example, codes sporadic terrorism, political banditry and unsuccessful coups in one category and successful coups and campaigns of terrorism in another. Terrorism incidents are not coded separate. Likewise Tilly (2002a, b) conceptually mixes terrorist incidents in with other forms of violent claim making seeing them as forms of coordinated destruction along with other forms such as lethal contests and campaigns of annihilation, while White (1993: 576) codes terrorist incidents in Northern Ireland as political violence, and Koopmans (1993: 640), in what seems clearly to be terrorist events, speaks of acts of severe and unusually conspiratorial violence directed against property (arson, bombings, sabotage) or people (political murders, kidnaping) as heavy violence. Terrorism is a form of political violence, coordinated destruction, and heavy violence, but so are other violent events, and for some research questions mixing them up is a reasonable procedure. But terrorism, like the strike, the riot, or the protest, is its own form of social violence and a fuller understanding of it will occur when it is separated out from understandably similar, but clearly different, kinds of collective violence.

435 Is Terrorism Globalizng? 435 References Ali, Tariq The Clash of Fundamentalisms: Crusades, Jihads and Modernity. London: Verso. Arrighi, Giovanni The Long Twentieth Century: Money, Power and the Origins of our Time. New York: Verso. Arrighi, Giovanni and Beverly J. Silver Chaos and Governance in the Modern World-System. Minneapolis: University of Minnesota Press. Barber, Benjamin R Jihad vs. McWorld: Terrorism s Challenge to Democracy. New York: Ballentine Books. Boli, John and George M. Thomas (eds.) Constructing World Culture: International Nongovernmental Organizations Since Stanford, CA: Stanford University Press. Brooks, Stephen and William Wohlforth American Primacy in Perspective. Foreign Affairs 81: Castells, Manuel The Rise of the Network Society. Cambridge: Blackwell. Chase-Dunn, Christopher; Kawano, Yukio and Benjamin D. Brewer Trade Globalization since 1795: Waves of Integration in the World-System American Sociological Review 65, 1, Feb, Chomsky, Noam U.S. A Leading Terrorist State. Monthly Review. 53: Cooper, H. H. A Terrorism: The Problem of Definition Revisited American Behavioral Scientist 44, 6, Davis, Mike Late Victorian Holocausts. London: Verso. Della Porta, Donatella and Sidney Tarrow Unwanted Children: Political Violence and the Cycle of Protest in Italy: European Journal of Political Research 14: Patterns of Transnational Terrorism, : Alternative Time-Series Estimates. International Studies Quarterly. 46: Ferguson, Niall. 2001a. The Cash Nexus: Money and Power in the Modern World, New York: Basic Books. Gibbs, Jack P Conceptualization of Terrorism American Sociological Review, 54, 3, Gurr, Ted Robert Terrorism in Democracies: Its Social and Political Bases. Pp in W. Reich (ed.) Origins of Terrorism. Princeton: Princeton University Press. Hoffman, Bruce Inside Terrorism. New York: Columbia University Press. Hudson, Rex R Who Becomes a Terrorist and Why: The 1999 Government Report on Profiling Terrorists. Guilford, Conn. The Lyons Press. Jenkins, Brian Michael Terrorism and Beyond: A 21 st Century Perspective. Studies in Conflict and Terrorism. 24: Johnson, Chalmers Blowback: The Costs and Consequences of American Em-

436 436 Albert J. Bergesen pire. New York: Henry Holt and Company. Joll, James The Anarchists. London: Methuen. Juergensmeyer, Mark Terror in the Mind of God: the Global Rise of Religious Violence. Berkeley: University of California Press. Karstedt-Henke, Sabine Theorien zur Erklarung terroristischer Bewegungen Pp in Politik der inneren Sicherheit E. Blankenberg (ed.). Frankfurt,Germany: Suhrkamp. Kennedy, Paul Maintaining American Power. Pp in Strobe Talbott and Nayan Chanda (eds.) The Age of Terror. New York: Basic Books. Koopmans, Ruud The Dynamics of Protest Waves: West Germany, 1965 to American Sociological Review 58: Lacqueur, Walter Terrorism. London; Weidenfeld and Nicolson The New Terrorism: Fanaticism and the Arms of Mass Destruction. New York: Oxford University Press. Meyer, John W., John Boli, George M. Thomas, and Francisco Ramirez World Society and the Nation-State. American Journal of Sociology 103: Oliverio, Annamarie The State of Terror. Stony Brook: State University of New York Press. Phillips, kevin Wealth and Democracy. New York: Broadway Books. Rapoport, David C Terrorism Pp in Lester R. Kurtz and Jennifer E. Turpin (eds.) Encyclopedia of Violence, Peace, and Conflict, Volume 3. San Diego: Academic Press The Fourth Wave: September 11 in the History of Terrorism. Current History (December): Robertson, Roland Globalization: Social Theory and Global Culture. London: Sage. Ruby, Charles L The Definition of Terrorism Analyses of Social Issues and Public Policy. Pp Sassen, Saskia Globalization and its Discontents. New York: The New Press. Scheitzer, Glenn E A Faceless Enemy: The Origins of Modern Terrorism. New York: Perseus Publishing. Schmid, Alex P. and Albert J. Jongman Political Terrorism: A New Guide to Actors, Authors, Concepts, Data Bases, Theories and Literature. Revised, expanded and updated edition. New Brunswick, NJ: Transaction Books. Senechal de la Roche, Roberta Collective Violence as Social Control. Sociological Forum 11: Why is Collective Violence Collective? Sociological Theory 19: Sklair, Leslie Sociology of the Global System. Baltimore, MD: Johns Hopkins University Press. Stern, Jessica The Ultimate Terrorists. Cambridge: Harvard University Press. Tarrow, Sidney. 1989a. Struggle, Politics, and Reform: Collective Action, Social

437 Is Terrorism Globalizng? 437 Movements, and Cycles of Protest. (Occasional Paper No. 21). Western Societies Program, Center for International Studies, Cornell University, Ithaca, New York b. Democracy and Disorder: Protest and Politics in Italy Oxford: Clarendon Press. Tilly, Charles. 2002a. Violence, Terror, and Politics as Usual Unpublished manuscript, Columbia University. Tilly, Charles. 2002b. Collective Violence. Unpublished book manuscript, 284 pages. Columbia University. Tilly, Charles. 2002c. Past, Present, and Future Globalizations. Unpublished manuscript, Columbia University. (Draft chapter for Gita Steiner-Khamsi, ed., Lessons from Elsewhere: The Politics of Educational Borrowing and Lending, Teachers College Press). Tomlinson, John Globalization of Culture. Chicago: University of Chicago Press. White, Robert W On Measuring Political Violence: Northern Ireland, 1969 to American Sociological Review 58:

438 438 Contributors CONTRIBUTORS Ulrich Baltzer, Irmgardstr. 28, D München, Germany. Albert J. Bergesen, Department of Sociology, University of Arizona, Tucson, Arizona 85721, USA. Cristiano Castelfranchi, University of Siena & ISTC- Institute of Cognitive Sciences and Technologies, Viale Marx 15, Roma, Italy. John Davis, Faculty of Economics and Econometrics, University of Amsterdam, Roeterstraat 11, 1018 WB Amsterdam, The Netherlands, and Department of Economics, Marquette University, PO Box 1881, Milwaukee, WI , USA. Kay Mathiesen, Department of Philosophy,Massachusetts College of Liberal Arts North Adams, MA 01247, USA. Christopher McMahon, Department of Philosophy, University of California, Santa Barbara CA 94010, USA. Anthonie Meijers, Philosophy and Social Sciences, Den Dolech 2, P.O. Box 513, 5600 MB Eindhoven, The Netherlands. Seumas Miller, The Australia National University, Center for Applied Philosophy and Public Ethics, GPO, Box A 260 ANU Canberra, Australia. Steven Miller, Loyola University. Department LFCD, School of Education, Mallinckroth Campus, 1041 N. Ridge Road, Wilmetta IL , USA. Patrick Rysiew, Department of Philosophy, Department of Philosophy, University of British Columbia, Vancouver BC V6T-1Z1 Canada. Antti Saaristo, Centre for the Philosophy of natural and Social Science, London School of Economics and Political Science, Houghton Street, London WC2A 2AE, Great Britain. Amie Thomasson, Department of Philosophy, University of Miami, P.O. Box Coral Gables, FL , USA.

439 Titel 439 Deborah Perron Tollefsen, Department of Philosophy, University of Memphis, 327 Clement Hall, Memphis, TN Maj Tuomela, Department of Social Psychology, P.O. Box 9, University of Helsinki, Finland. Raimo Tuomela, Department of Social and Moral Philosophy, P.O. Box 9, University of Helsinki, Finland. Stephen Turner, Department of Philosophy, University of South Florida, University of South Florida, Tampa, FL K. Brad Wray, Department of Philosophy, State University of New York, Oswego, 128 Piez Hall, Oswego, NY USA. Petri Ylikoski, Department of Social and Moral Philosophy, P.O. Box 9, University of Helsinki, Finland.

440 440 Pulished Volumes IMPRINT ProtoSociology: An International Journal of Interdisciplinary Research ISSN Editor: Gerhard Preyer Johann Wolfgang Goethe-Universität Frankfurt am Main, Dep. of Social Sciences Editorial Staff: Georg Peter Editorial of the Vol , 2003: Raimo Tuomela, Gerhard Preyer, and Georg Peter Layout and digital publication: Georg Peter Editorial Office: ProtoSociology, Stephan-Heise-Str. 56, Frankfurt am Main, Germany, Phone: (049) , Bank: Dresdner Bank AG, Frankfurt am Main, account: , BLZ: SWIFT-BIC: DRES DE FF IBAN DE Die Zeitschrift soll 1/2jährlich erscheinen. Die Anzahl der jährlich erscheinenden Hefte und Sonderhefte bleibt jedoch vorbehalten. Copyright: Die in dieser Zeitschrift veröffentlichten Beiträge sind urheberrechtlich geschützt. Alle Rechte sind vorbehalten. Übersetzungen, Nachdruck, Vervielfältigung auf fotomechanischem oder ähnlichem Weg oder im Magnettonverfahren, Wiedergabe durch Vortrag, Funk- und Fernsehsendungen sowie Speicherung in Datenverarbeitungsanlagen, auch auszugsweise, sind nur mit Genehmigung des Herausgebers möglich. Für eingereichte Beiträge wird keine Haftung übernommen. Weitere Publikationsrechte von Artikeln bleiben vorbehalten. Zweitpublikationen seitens des Autors werden eingeräumt. Bei einer Zweitpublikation ist das Heft (Nummer, Titel, Erscheinungsjahr) der PROTOSOCIOLOGY zu zitieren. Für unaufgefordert eingesandte Manuskripte wird keine Haftung übernommen. Gerichtsstand ist Frankfurt am Main. Copyright: All rights reserved. This publication may not be reproduced, stored or transmitted in any form or by any means without the prior permission in writing of the publisher. Additional publications of the articles are reserved. The authors retain the personal right to re-use their own articles. Authorization to photocopy items for internal or personal use, or the internal or personal use of specific clients is garanted by PROTOSOCIOLOGY, provided that the base fee is paid directly to VG Wort, Goethestr. 49, München RFA.The publisher accepts no responsibility for submitted manuscripts. ProtoSociology im Internet:

441 Titel 441 ON PROTOSOCIOLOGY Protosociology occupies an important position in the European intellectual scene, bridging philosophy, economics, sociology and related disciplines. Its volumes on rationality bring together concerns in all these topics, and present an important challenge to the cognitive sciences. Donald Davidson, Berkeley (USA) Protosociology publishes original papers of great interest that deal with fundamental issues in the human and social science. No academic library is complete without it. Nicholas Rescher, Pittsburgh (USA) Protosociology has been remarkably successful in publishing interesting work from different tradition and different disciplines and, as the title signals, in giving that work a new, eye-catching slant. Philipp Pettit, Canberra, Australia Protosociology is a truly premier interdisciplinary journal that publishes articles and reviews on timely topics written by and for a wide range of international scholars. The recent volumes on rationality are remarkable for their breadth and depth. Protosociology would be a great addition to any library. Roger Gibson, St. Louis (USA

442 442 Pulished Volumes PROTOSOCIOLOGY An International Journal of Interdisciplinary Research Vol. 17, 2002 Indirect and Reported Speech CONTENTS The Meaning of What is Said and the Ascription of Attitudes The Semantic Significance of What is Said Emma Borg Representing What Others Say Cara Spencer The Things People Say Jonathan Sutton Reported Speech and the Epistemology of Testimony Sanford C. Goldberg Reports and Imagination Eros Corazza Propositional Content and Cognitive Structure On Representing Content David Hunter Conceptual Realism and Interpretation Max A. Freund Content Partialism and Davidson s Dilemma Corey Washington Vagueness, Indirect Speech Reports, and the World Steven Gross On Contemporary Philosophy and Social Sciences Epistemological Remarks Concerning the Concepts Theory and Theoretical Concepts Hans Lenk Social Science and (Null) Hypothesis Testing:Some Ontological Issues Steven Miller and Marcel Fredericks

443 Published Volumes 443 PROTOSOCIOLOGY An International Journal of Interdisciplinary Research Vol. 16, 2002 Understanding the Social: New Perspectives from Epistemology CONTENTS FROM MENTALIZING FOLK TO SOCIAL EPISTEMOLOGY Alvin I. Goldman The Mentalizing Folk Margaret Gilbert Believe and Acceptance as Features of Groups Antonie Meijers Collective Agents and Cognitive Attitudes Deborah Perron Tollefsen Challenging Epistemic Individualism EXTERNALISM, EVENTS AND INSTITUTIONAL FACTS Gerhard Preyer From an Externalistic Point of View: Understanding the Social Steven Miller Are Context and Event Equivalent? Possibilities for an Ontological Symbiosis Frank Hindriks Institutional Facts and the Naturalistic Fallacy: Confronting Searle (1964) with Searle (1995) GROUP RESPONSIBILITY AND THE CONSENSUS ACCOUNT OF JUSTIFIED BELIEF Pekka Mäkelä and Raimo Tuomela Group Action and Group Responsibility Lamber Royakkers Collective Commitments: a Theoretical Understanding of Human Cooperation Fred Schmitt Justification and Consensus: The Peircean Approach ON CONTEMPORARY PHILOSOPHY Roger F. Gibson How I Came to Know Quine: A Reminiscence Hans Lenk and Matthias Maring Responsibility and Globalization

444 444 Pulished Volumes PROTOSOCIOLOGY An International Journal of Interdisciplinary Research Vol. 15, 2001 On a Sociology of Borderlines: Social Process in Time of Globalization Gerhard Preyer, Mathias Bös (eds.) CONTENTS Introduction: Borderlines in Time of Globalization (Gerhard Preyer, Mathias Boes) I RECONCEPTIONALIZATIONS OF THE GLOBAL: BORDERLINES IN WORLD SOCIETY Shmuel Noah Eisenstadt The Continual Reconstruction of Multiple Modern Civilizations and Collective Identities Christopher Chase-Dunn: Globalization: A World-Systems Perspective Thomas D. Hall: World-Systems, Frontiers, and Ethnogenesis: Incorporation and Resistance to State Expansion Richard E. Lee: After History? The Last Frontier of Historical Capitalism II DEFINING BORDERLINES IN WORLD SOCIETY: THE EMERGENCE OF NEW MEMBERSHIPS Gerhard Preyer: Globalization and the Evolution of Membership Barrie Axford: Enacting Globalization: Transnational Networks and the Deterritorialization of Social Relationships in the Global System Mathias Bös Immigration and the Open Society: The Normative Patterns of Membership in the Nation State Uta Gerhardt, Birgitta Hohenester: A Transformation of National Identity? Refugees and German Society After World War II Marja Keränen: Citizenship, Universal- Speak, and Local-Speak III THE GLOBAL AND THE LOCAL: THE COLLAPSE AND RECONSTRUCTION OF BORDERLINES Christie Davies, Eugene Trivizas The Collapse of the National Morality and National Moral Boundaries of Small Peripheral Countries: not Globalisation but the Imposition of Liberty Walter L. Bühl: Former GDR between Transformation and Social Evolution F. Peter Wagner: Beyond East and West : On the European and Global Dimensions of the Fall of Communism Ramón Grosfoguel Cultural Racism and Borders of Exclusion in the Capitalist World-Economy: Colonial Caribbean Migrants in Core Zones Francisco Entrena: Socio-Economic Restructurings of the Local Settings in the Era of Globalization ON A SOCIOLOGY OF VIOLENCE Konrad Thomas Ein anderes Verständnis von Gewalt: Der gesellschaftsanalytische Beitrag des Literaturwissenschaftlers René Girard CRITICAL REVIEW: ProtoSciology, Vol. 12, 1998-Special Edition: After the Received View. Developments in the Theory of Science George N. Schlesinger

445 Published Volumes 445 PROTOSOCIOLOGY An International Journal of Interdisciplinary Research Vol. 14, 2000 Folk Psychology, Mental Concepts and the Ascription of Attitudes: On Contemporary Philosophy of Mind CONTENTS Alvin I. Goldman Folk Psychology and Mental Concepts Philip Pettit How the Folk Understand Folk Psychology Jane Heal Understanding Other Minds from the Inside Christopher S. Hill From Assertion to Belief: The Role of Linguistic Data in the Practice of Belief-Ascription David Rosenthal Content, Interpretation, and Consciousness Jay L. Garfield Thought as Language: A Metaphor Too Far Robert M. Gordon Sellars s Ryleans Revisited Louise Röska-Hardy Self-Ascription and Simulation Theory Brian P. McLaughlin Why Intentional Systems Theory Cannot Reconcile Physicalism With Realism about Belief and Desire Gerhard Preyer Primary Reasons: From Radical Interpretation to a Pure Anomalism of the Mental Rebecca Kukla How to Get an Interpretivist Committed David Pitt Nativism and the Theory of Content Raffaella De Rosa On Fodor s Claim that Classical Empiricists and Rationalists Agree on the Innateness of Ideas Consuelo Preti Belief and Desire Under The Elms Erwin Rogler On David Lewis Philosophy of Mind Barbara Von Eckardt, Jeffrey S. Poland In Defense of the Standard View ON CONTEMPORARY PHILOSOPHY OF LANGUAGE, ONTOLOGY AND MORAL THEORY Thomas Baldwin Nearly Logic James E. Tomberlin Logical Form, Actualism, and Ontology Arend Kulenkampff, Frank Siebelt What a Noncognitivist might tell a Moral Realist

446 446 Pulished Volumes PROTOSOCIOLOGY An International Journal of Interdisciplinary Research Vol Reasoning and Argumentation D. Mans, G. Preyer (eds.) CONTENTS Ralph H. Johnson: Reasoning, Argumentation and The Network Problem Leo Groarke: The Fox and the Hedgehog: On Logic, Argument, and Argumentation Theory J. Anthony Blair: Presumptive Reasoning/Argument: An Overlooked Class Robert C. Pinto: Argument Schemes and the Evaluation of Presumptive Reasoning: some Reflections on Blair s Account Douglas Walton: The New Dialectic: A Method of Evaluating an Argument Used for Some Purpose in a Given Case Manfred Kienpointner: Comments on Douglas Walton s Paper Christopher W. Tindale: The Authority of Testimony John Woods: Peirce s Abductive Enthusiasms Henry W. Johnstone, Jr.: Any, Every, and the Philosophical Argumentum ad Hominem ON CONTEMPORARY PHILOSOPHY Hans Lenk Interdisziplinarität und Interpretation Ellery Eells Causal Decision Theory

447 Published Volumes 447 PROTOSOCIOLOGY An International Journal of Interdisciplinary Research VOL. 12, 1998 Special Edition AFTER THE RECEIVED VIEW Developments in the Theory of Science Gerhard Preyer, Georg Peter, Alexander Ulfig (Eds.) CONTENTS in memoriam Wolfgang Stegmüller Introduction: Developments in the Theory of Science (Gerhard Preyer, Georg Peter, Alexander Ulfig) LOGICAL OPERATIONALISM SIGNIFICANCE AND MEANING Wilhelm K. Essler: Truth and Knowledge. Some Considerations concerning the Task of Philosophy of Science Gerhard Preyer: The Received View, Incommensurability and Comparison of Theories Beliefs as the Basis of Theorizing Robert Schwartz: Reflections on Projection Jeffrey E. Foss: The Logical and Sociological Structure of Science STRUCTURALISM MEANINGFUL MEASUREMENT THE CONCEPTION OF PHYSICAL LAW C. Ulises Moulines: Structuralism vs. Operationalism Nicholas Rescher: Meaningless Numbers R. I. G. Hughes: Laws of Nature, Laws of Physics, and the Representational Account of Theories James R. Brown: Einstein s Principle Theory INDUCTIVE INFERENCES INTERPRETATION OF PROBABILITY GAME THEORY Kevin T. Kelly, Cory Juhl Transcendental Deductions and Universal Architectures for Inductive Inferences Howard H. Harriott: R.A. Fisher and the Interpretation of Probability Brian Skyrms: Evolution of an Anomaly PROPERTIES UNDERDETERMINATION SCIENTIFIC REALISM George N. Schlesinger: Degrees of Characterizations Carl A. Matheson: Observational Adequacy as distinct from the Truth about Observables Thomas R. Grimes: Scientific Realism and the Problem of Underdetermination Paul C. L. Tang: On Paul Churchland s Treatment of the Argument from Introspection and Scientific Realism RATIONALITY METAPHORS VALUES IN SCIENCE David Resnik: Scientific Rationality and Epistemic Goals Aldo Montesano: Rationality in Economics: A General Framework Joseph Agassi: Science Real and Ideal: Popper and the Dogmatic Scientist Michael Bradie: Models and Metaphors in Science David Gruender: Values and the Philosophy of Science

448 448 Pulished Volumes PROTOSOCIOLOGY An International Journal of Interdisciplinary Research Vol. 11, 1998 Cognitive Semantics II Externalism in Debate CONTENTS RADICAL INTERPRETATION, ONTOLOGY AND THE STRUCTURE OF LANGUAGE Gerhard Preyer, Michael Roth: On Donald Davidson s Philosophy: An Outline Richard Manning: All Facts Great and Small Barbara Fultner: Of Parts and Wholes: The Molecularist Critique of Semantic Holism Louis Goble: Re-Evaluating Supervaluations David Simpson: Interpretation and Skill: On Passing Theory Wulf Kellerwessel: Katz on Semantics and Pragmatics EXTERNALISM AND THE INDIVIDUATION OF CONTENT Ron Wilburn: Knowledge, Content, and the Wellstrings of Objectivity Anthony Brueckner: Content Externalism and A Priori Knowledge Consuelo Preti: The Irrelevance of Supervenience Michael Liston: Externalist Determinants of Reference Arnold Silverberg: Semantic Externalism. A Response to Chomsky Gerhard Preyer: Interpretation and Rationality: Steps from Radical Interpretation to the Externalism of Triangulation

449 Published Volumes 449 PROTOSOCIOLOGY An International Journal of Interdisciplinary Research Vol. 10, 1997 Cognitive Semantics I Conceptions of Meaning CONTENTS MEANING, TRUTH AND THE USE OF LANGUAGE Herman Cappelen, Ernie Lepore: Semantic Theory and Indirect Speech Kirk Ludwig: The Truth about Moods Louise Röska-Hardy: Language, Use and Action Jeffrey King: The Source(s) of Necessity Filip Buekens: The Genesis of Meaning (a Myth) RATIONAL DIALETISM OF ANTINOMIES Mark Sainsbury: Can Rational Dialetheism Be Refuted By Considerations about Negation and Denial? CONTEMPORARY DEBATES IN THE PHILOSOPHY OF MIND Klaus Sachs-Hombach: Philosophy of Mind: Die Simulationstheorie REFERENCE, INDEXICALS AND SPEAKER MEANING Robert Hanna: Extending Direct Reference Peter Ludlow: Semantics, Tense, and Time: a Note on Tenseless Truth-Conditions for Token-Reflexive Tensed Sentences Gerhard Preyer: Verstehen, Referenz, Wahrheit. Zur Philosophie Hilary Putnams Reinaldo Elugardo: Descriptions, Indexicals and Speaker Meaning IN RETROSPECT Joseph Agassi: Wittgenstein The End of a Myth Kent A. Peacock, Richard Feist: The Einstein-DeSitter Controversy

450 450 Pulished Books CONCEPTS OF MEANING Framing an Integrated Theory of Linguistic Behavior ed. by Gerhard Preyer, Georg Peter, Maria Ulkan Introduction Gerhard Preyer, Georg Peter, Maria Ulkan I Speaker Meaning, Communication, and Intentions 1 Communicative and Illocutionary Acts Maria Ulkan 2 Language Acts and Action Louise Röska-Hardy 3 Reflections on the Intentionality of Linguistic Behavior Jan Nuyts 4 Descriptions, Indexicals, and Speaker Meaning Reinaldo Elugardo 5 Informatives and/or Directives? (A New Start in Speech Act Classification) Georg Meggle, Maria Ulkan 6 Constructive Speech-Act Theory Dirk Hartmann II Truth, Semantic Content, and Externalism 7 The Truth about Moods Kirk Ludwig 8 Semantic Theory and Indirect Speech Herman Cappelen and Ernie Lepore 9 All Facts Great and Small Richard N. Manning 10 Knowledge, Content, and the Wellsprings of Objectivity Ron Wilburn 11 Interpretation and Skill: On Passing Theory David Simpson Contributors Index Kluwer Academic Publisher, Philosophical Studies 96, 2003Dordrecht, Netherland

451 Published Books 451 LOGICAL FORM AND LANGUAGE Gerhard Preyer, Georg Peter (eds.) To Donald Davison This volume brings together some of the world s most distinguished linguistically-minded philosophers and philosophically-minded linguists in an outstanding collection of new papers on what logical form is supposed to be and why it matters for the study of language. - Zoltan Szabo, Cornell University CONTENTS Gerhard Preyer, Georg Peter: Introduction PART I: THE NATURE OF LOGICALFORM Stephen Neale Logically Unformed Ernie Lepore, Kirk Ludwig What is Logical Form Paul M. Pietroski Function and Concatenation Jeffrey King Two Sorts of Claim about Logical Form Peter Ludlow LF and Natural Logic. The Syntax of Directional Entailing Environments Robert May/R.F. Fiengo Identity Statements PART II: INTENSIONALITY, EVENTS AND SEMANTIC CONTENT James Higginbotham Why is Sequence of Tens Obligatory? Richard Larson The Grammar of Intensionality Barry Schein Events and the Semantic Content of Thematic Relations Norbert Hornstein A Grammatical Argument for A Davidsonian Semantics Jason Stanley Nominal Restriction PART III: LOGICAL FORM, BELIEF ASCRIPTION AND PROPER NAMES Bernard Linsky Russell s Logical Form, LF and Truth Conditions Lenny Clapp, Robert Stainton Obviously Propositions are Nothing. Russell and the Logical Form of Belief Reports Robert Matthews Logical Form and the Relational Conception of Belief Marga Reimer Ordinary Proper Names Reinaldo Elugardo The Predicate View of Proper Names Oxford University Press, Oxford

452 452 Pulished Books Analytische Ästhetik Eine Untersuchung zu Nelson Goodman und zur literarischen Parodie Georg Peter I. ANALYTISCHE ÄSTHETIK: DIE ZEICHENTHEORIE NELSON GOODMANS 1. Eine andere Fragestellung / 2. Zu Nelson Goodman in allgemeiner Absicht / 3. Zeichenbedeutung: Sprachliche und nichtsprachliche Denotation / 4. Von der Probe zum passenden Etikett: Exemplifikation und Ausdruck / 5. Die Nebenbeschäftigung der Symbole: Ein bedeutungstheoretischer Exkurs zu Metapher und Wahrheit / 6. Ästhetische Bedeutung und kognitive Funktion: Das Kunstwerk als Probe / 7. Wann oder doch Was ist Kunst? II. SCHEMAINTERPRETATION UND DIE IDENTIFIZIERBARKEIT ÄSTHETISCHER ZEICHEN 1. Schemainterpretationismus und Symboltheorie: Vom Symbol zum Schema / 2. Der konstruktive Schemainterpretationismus von Hans Lenk / 3. Hermeneutik und Schemainterpretation / 4. Die soziale Verankerung der Interpretation: Ästhetische Kompetenz und Kultursoziologie / 5. Thesenhaftes zu einer schemaorientierten Symboltheorie: Wann ist (es) Parodie III. ROBERT NEUMANNS PARODIEN: DIE PROBE AUF DIE EXEMPLIFIKATION 1. Literaturwissenschaft und Parodie: eine vorbelastete Beziehung / 2. Die Parodie als eine Form des Komischen / 3. Das Problem der Wertung: Die kritische und die (bloß) komische Parodie / 4. Nachahmung, Bezugnahme und Robert Neumanns Theorie der Parodie / 5. Von der Funktion der Zeichen zur Funktion der Parodie: Goethe und Brecht parodiert / 6. Komplexität der Parodie und die komplexen Parodien Robert Neumanns / 7. Zeigen und Sagen: Funktion und Qualität der Parodie anhand einer Benn-Parodie / 8. Die Grenzen der Parodie und der Parodierbarkeit / 9. Parodien und andere Formen der Nachahmung IV. ZUR METHODIK DES ANSATZES UND SEINER WEITERFÜHRUNG ca. 330 Seiten, Erschienen in der Reihe: PHILOSOPHISCHE ANALYSE / PHILOSOPHICAL ANALYSIS, Herausgegeben von / Edited by Herbert Hochberg Rafael Hüntelmann et al., Deutsche Bibliothek der Wissenschaften, Dr. Hänsel-Hohenhausen AG, Frankfurt a.m., München, New York

453 Published Books 453 Borderlines in a Globalized World New Perspectives in a Sociology of the World System Gerhard Preyer, Mathias Bös (eds.) To Walter L. Bühl Introduction Gerhard Preyer, Mathias Bös Borderlines in Time of Globalization: New Theoretical Perspectives I RECONCEPTIONALIZATIONS OF THE GLOBAL: BORDERLINES IN WORLD SOCIETY Shmuel N. Eisenstadt: The Continual Reconstruction of Multiple Modern Civilizations and Collective Identities Christopher Chase Dunn: Globalization: A World-Systems Perspective Thomas D. Hall: World-Systems, Frontiers, and Ethnogenesis: Incorporation and Resistance to State Expansion. Richard Lee: After History? The Last Frontier of Historical Capitalism II DEFINING BORDERLINES IN WORLD SOCIETY: THE EMERGENCE OF NEW MEMBERSHIPS Gerhard Preyer: Globalization and the Evolution of Membership Barrie Axford: Enacting Globalization - Transnational Networks and the Deterritorialisation of Social Relationship Mathias Bös: Immigration and the Open Society: The Normative Patterns of Membership in the Nation State Uta Gerhardt, Birgitta Hohenester: A Transformation of National Identity? Refugees and German Society after World War II III. THE GLOBAL AND THE LOCAL : THE COLLAPSE AND RECONSTRUCTION OF BORDERLINES Christie Davies, Evgenios Trivizas: The Collapse of the National Morality and National Moral Boundaries of Small Peripherial Countries not Globalization but the Imposition of Liberty F. Peter Wagner: Beyond East and West : On the European and Global Dimensions of the Fall of Communism Francisco Entrena:Socioeconomic Restructurings of the Local Settings in the Era of Globalization Kluwer Academic Publishers Social Indicators Research Series Vol. 9 Dordrecht, Netherlands 2002

454 454 Pulished Books The Contextualization of Rationality Gerhard Preyer, Georg Peter (eds.) Introduction: Problems, Concepts and Theories of Rationality RADICAL INTERPRETATION, NATURALIZED EPISTEMOLOGY AND NORMATIVITY John Heil The Propositional Attitudes David K. Henderson Epistemic Rationality, Epistemic Motivation, and Interpretive Charity Gerhard Preyer Interpretation and Rationality: Steps from Radical Interpretation to the Externalism of Triangulation Roger F. Gibson Stich on Intentionality and Rationality Paul K. Moser and David Yandell Against Naturalizing Rationality Harvey Siegel Naturalism, Instrumental Rationality, and the Normativity of Epistemology INTENTIONS AND THE SOCIAL ASPECT OF RATIONALITY Alfred R. Mele Rational Intentions and the Toxin Puzzle Peter Gärdenfors The Social Stance Timo Airaksinen and Katri Kaalikoski Instrumental Rationality Raimo Tuomela Rational Cooperation and Collective Goals Peter French Rationality and Ethics Julian Nida-Rümelin The Plurality of Good Reasons and the Theory of Practical Rationality CONCEPTS OF EXPLANATION, JUSTIFICATION AND REALITY Philip Pettit Three Aspects of Rational Explanation Keith Lehrer Rationality and Trustworthiness Alexander Ulfig Validity, Justification and Rationality Nicholas Rescher Reason and Reality 250 Seiten mentis Verlag, Paderborn 2000

455 Published Books 455 PROTOSOZIOLOGIE IM KONTEXT Lebenswelt und System in Philosophie und Soziologie Gerhard Preyer, Georg Peter, Alexander Ulfig (Hrsg.) in memoriam Paul Lorenzen Einleitung: Lebenswelt und System in Philosophie und Soziologie, Gerhard Preyer, Georg Peter, Alexander Ulfig ZUM BEGRIFF DER LEBENSWELT Ernst W. Orth: Lebenswelt als unvermeidliche Illusion? Husserls Lebensweltbegriff und seine kulturpolitischen Weiterungen, Walter Biemel: Gedanken zur Genesis der Lebenswelt, Alexander Ulfig: Lebenswelt und Reflexion. Anhang: Lebenswelt als Fundament der Wissenschaft, Gerhard Preyer: Hintergrundwissen: Kritik eines Begriffs, Hubert A. Knoblauch: Soziologie als strenge Wissenschaft? Phänomenologie, kommunikative Lebenswelt und soziologische Methodologie LEBENSWELT BEGRÜNDUNG WISSENSCHAFT Jürgen Mittelstraß: Das lebensweltliche Apriori, Peter Janich: Die Rationalität der Naturwissenschaften, Jürgen Mittelstraß: Rationalität und Reproduzierbarkeit, Elisabeth Ströker:Lebenswelt durch Wissenschaft, Paul Janssen: Lebenswelt, Wissen und Wissenschaft, Richard T. Murphy: E. Husserl s Phenomenology of Reason LEBENSWELT/LEBENSFORM SPRACHE Pierre Kerszberg: Lifeworld and Language, John F.M. Hunter: The Motley Forms of Life in the Later Wittgenstein, Peter A. French: Why did Wittgenstein read Tagore to the Vienna Circle? Georg Peter: Die Nebenbeschäftigung der Symbole. Zu Wahrheit und Funktion der Metapher SYSTEM SOZIALSYSTEM GESELLSCHAFT Niklas Luhmann: Die Lebenswelt nach Rücksprache mit Phänomenologen, Niklas Luhmann: Observing Re-entries, Gerhard Preyer: System-, Medien- und Evolutionstheorie. Zu Niklas Luhmanns Ansatz, Richard Münch: Autopoesis per Definition, Hans Zitko: Codierungen der Kunst: Zur Kunstsoziologie Niklas Luhmanns, James Bohman: The Completeness of Macro-Sociological Explanations: System and Lifeworld, Göran Ahrne: Outline of an Organisational Theory of Society, Anhang: Karl Otto Hondrich: Zu Göran Ahrnes Ansatz Digitale Neuauflage bei: 392 Seiten Verlag Königshausen & Neumann, Würzburg 1996 Humanities Online, Frankfurt am Main2001

456 456 Pulished Books INTENTION BEDEUTUNG KOMMUNIKATION Kognitive und handlungstheoretische Grundlagen der Sprachtheorie Gerhard Preyer, Maria Ulkan, Alexander Ulfig (Hrsg.) Einleitung: Zu kognitiven und handlungstheoretischen Grundlagen der Sprachtheorie, Gerhard Preyer, Maria Ulkan, Alexander Ulfig I INTENTIONEN UND KOMMUNIKATIVE HANDLUNGEN Maria Ulkan: Kommunikative und illokutionäre Akte; Georg Meggle/Maria Ulkan: Grices Doppelfehler. Ein Nachtrag zum Griceschen Grundmodell; Jan Nuyts: Intentionalität und Sprachfunktionen II INTERPRETATION UND BEDEUTUNG Gerhard Preyer: Kognitive Semantik, Anhang Sprechaktsemantik: J.L. Austin, J.R. Searle, H.P. Grice, P.F. Strawson; Louise Röska-Hardy: Sprechen, Sprache, Handeln; Frank Siebelt: Zweierlei Holismus. Überlegungen zur Interpretationstheorie Donald Davidsons; Peter Rothermel: Semantische Implikaturen; Volkmar Taube: Referenz und Interpretation. Zur Theorie nichtsprachlicher Symbolisierung; Georg Peter: Zu Richtigkeit und Interpretation der Metapher: Kognitive Funktion und rekonstruktive Schemainterpretation III KLASSIFIKATION VON SPRECHAKTEN Maria Ulkan: Informations- und Aufforderungshandlungen; Dirk Hartmann: Konstruktive Sprechakttheorie; Volkmar Taube: Bildliche Sprechakte IV KOMMUNIKATIVES HANDELN UND INTERSUBJEKTIVE GÜLTIGKEIT Jürgen Habermas: Sprechakttheoretische Erläuterungen zum Begriff der kommunikativen Rationalität; Karl-Otto Apel: Illokutionäre Bedeutung und normative Gültigkeit. Die transzendentalpragmatische Begründung der uneingeschränkten kommunikativen Verständigung; Peter-Paul König: Kommunikatives und strategisches Handeln. Kritische Bemerkungen zu zwei zentralen Begriffen der Theorie des kommunikativen Handelns von Jürgen Habermas; Alexander Ulfig: Präsuppositionen und Hintergrundwissen. Eine Kritik am formalpragmatischen Präsuppositionsbegriff V DIALOGSTRUKTUR UND ARGUMENTATION Wilhelm Franke: Konzepte linguistischer Dialogforschung; Franz Hundsnurscher: Streitspezifische Sprechakte: Vorwerfen, Insistieren, Beschimpfen; Dieter Mans: Argumentation im Kontext, Exkurs: Zu Christoph Lumers Praktische Argumentationstheorie 408 Seiten Westdeutscher Verlag, Opladen 1997 Digitale Version: Humanities Online 2001

457 Published Books 457 STRUKTURELLE EVOLUTION UND DAS WELTSYSTEM Theorien, Sozialstruktur und evolutionäre Entwicklungen Gerhard Preyer (Hrsg.) Einleitung: Gerhard Preyer: Strukturelle Evolution und das Weltsystem: Theorien, Sozialstruktur und evolutionäre Entwicklungen ZENTRUM UND PERIPHERIE INSTITUTIONELLE ENTWICKLUNG ASKRIPTIVE SOLIDARITÄT Shmuel Noah Eisenstadt: Social Division of Labor, Construction of Centers and Institutional Dynamics: A Reassessment of the Structural Evolutionary Perspective, Bernhard Giesen, Kay Junge: Strukturelle Evolution, Gerhard Preyer: Mitgliedschaftsbedingungen. Zur soziologischen Kerntheorie einer Protosoziologie, Anhang: Die modernen Gesellschaften verstehen. Richard Münchs Entwicklungstheorie moderner Gesellschaften, Erwin Rogler, Gerhard Preyer: Relationslogische Darstellung der sozialen Gesetze, Gerhard Preyer: Die modernen Gesellschaften verstehen. Zu Richard Münchs Entwicklungstheorie moderner Gesellschaften, Dieter Claessens: Bemerkungen zur Entstehung der modernen Ökonomie: Das Organistionsproblem, Richard Pieper: Strukturelle Emotionen, elementare Strukturbildung und strukturelle Evolution DIE EVOLUTION POLITISCHER ORDNUNGEN Rainer C. Baum: Parsons on Evolution of Democracy, Mathias Bös: Zur Evolution nationalstaatlich verfaßter Gesellschaften, Konrad Thomas: Das Ethnische und das Staatliche, Volker Bornschier: Die westeuropäische Integration als Gesellschaftsmodell im Zentrumswettbewerb ZUR SOZIOLOGIE DES WELTSYSTEMS Immanuel Wallerstein: Evolution of the Modern World-System, Christopher Chase- Dunn, Thomas D. Hall: The Historical Evolution of World-Systems: Iterations and Transformations, Albert Bergesen: Postmodernism: A World System Explanation, Richard Münch: Modernity and Irrationality: Paradoxes of Moral Modernization, Walter L. Bühl: Transformation oder strukturelle Evolution? Zum Problem der Steuerbarkeit von sozialen Systemen STATE OF THE ART: Michael Schmid: Soziologische Evolutionstheorien Suhrkamp Verlag, Frankfurt am Main 1998, Suhrkamp Taschenbuch Wissenschaft

458 458 Pulished Books LEBENSWELT SYSTEM GESELLSCHAFT Konstruktionsprobleme der Theorie des kommunikativen Handelns von Jürgen Habermas Gerhard Preyer I DIE ENTWICKLUNGSLOGIK VON WELTBILDERN 1. Die Rationalisierung von Weltbildern 2. Strategien der Analyse von Weltbildern 3. Folgeprobleme und Kritik II GESELLSCHAFT ALS LEBENSWELT UND SYSTEM 1. Die Strukturen der Lebenswelt 2. Allgemeine Bezugsprobleme der soziologischen Evolutionstheorie 3. Die Verständigungsformen 1. Weltbilder und soziale Integration 2. Zur Durkheim-Interpretation 4. Kommunikationsmedien und generalisierte Kommunikationsweisen 5. Folgeprobleme und Kritik III DIE FORMAL-PRAGMATISCHE BEDEUTUNGSTHEORIE 1. Die sprechakttheoretische Grundlegung 2. Interpersonal geltende Bedeutungskonventionen 3. Regelbewußtsein und Handlungskompetenz 4. Die Expansion des semantischen Gehalts 5. Verständigung und die Herstellung interpersonaler Beziehungen IV DER ERWERB DES MORALISCHEN BEWUßTSEINS 1. Die sozial-kognitive Grundausstattung 2. Diskurs und moralisches Bewußtsein 3. Folgeprobleme V KONSTRUKTIONSPROBLEME UND KRITIK 1. Rekonstruktionshypothesen 2. Zu den Konstruktionsproblemen ANHANG: Max Webers Religionssoziologie als eine Typologie des Rationalismus 290 Seiten digital bookpublication: Verlag Humanities Online, Frankfurt am Main

459 Published Books 459 Die globale Herausforderung Wie Deutschland an die Weltspitze zurückkehren kann Gerhard Preyer Der Begriff Globalisierung hat bereits Schlagwortcharakter bekommen und ist in aller Munde. Globale Märkte haben die Grundsituation wirtschaftlichen und politischen Handelns grundsätzlich verändert. Das heute entstehende globale Weltsystem und der Medienverbund auf dem es beruht, fordert uns alle heraus. Was bedeutet Globalisierung? Worin bestehen die globalen Herausforderungen, die zu bewältigen sind? Darauf wird eine Antwort gegeben. NEUE MEDIEN, WISSENSCHAFT UND TECHNIK IM ZEITALTER DER GLOBALISIERUNG Was heißt Globalisierung? Die neuen Medien: Eine Kopernikanische Wende Der Überlebensimperativ: Technologieentwicklung Die Zukunft der deutsche Universitäten VERÄNDERTE KONSTELLATIONEN Globale Wirtschaft und globale Ordnung Die Globalisierung der Finanzmärkte Auf dem Weg zur virtuellen Organisation Das Ende des klassischen Arbeitsmarktes Zu einer neuen Wirtschaftspolitik EUROPA IM ZEITALTER DER GLOBALISIERUNG Zur Ausgangssituation im Zentrumswettbewerb Frankreichs Zentralismus: Grenzen der Marktwirtschaft Das Überlebenssystem Italien Großbritannien zwischen Tradition und Modernisierung Deutsche Stärken im Umbruch Neu gemischte Karten Die Wiedergewinnung des Standorts Deutschland DIE EVOLUTION DES MITGLIEDSCHAFTSCODES Zur Soziologie der Grenzziehungen Gesellschaft, Organisation und Interaktion Gesellschaftsinterne Globalisierung und Entwicklungstrends Die globalisierte Gesellschaft 290 Seiten Frankfurter Allgemeine Zeitung/Gabler Edition, Frankfurt am Main 1998 Neuauflage Reprint Verlag Science-Digital, Haiger

460 460 Pulished Books INTEGRIERTES MANAGEMENT Was kommt nach der Lean-Production? Gerhard Preyer/Jakob Schissler Es gibt heute im Weltmaßstab kein Unternehmen, das seine Umstrukturierung nicht an den Bestandteilen des japanischen Modells der schlanken Produktion und des schlanken Managements, der Lean-Modelle, orientiert. Die Lean-Modelle sind zu einem integrierten Management fortzuentwickeln: dem über Vernetzung integrierten segmentierten Unternehmen, mit durchlässigen, ständig wechselnden Grenzen zwischen Unternehmen, Lieferanten und Kunden. Integriertes Management ist das Unternehmensmodell, das eine erhöhte Qualifizierung der Mitarbeiter und die Fähigkeit, in sich selbst steuernden Teams mitzuarbeiten, erfordert. Das bedeutet aber auch die Entwicklung eines neuen Führungsstils. Mit integriertem Management kann jedes Unternehmen sofort anfangen. Dazu wird in diesem Buch ein Weg gezeigt. Es wird eine wichtige Funktion als Impuls für weitere Entwicklungen haben. Inhalt DIE NEUEN HERAUSFORDERUNGEN 1. Vom japanischen und amerikanischen Management lernen? 2. Information, Wissen und Entscheidungen KREATIVITÄT ORGANISATION FÜHRUNG 1. Managementfähigkeiten 2. Innovative Organisation 3. Management und Führungsstile STRATEGIE MARKT KULTUR 1. Unternehmensstrategie 2. Controlling 3. Marketing 4. Zur Unternehmenskultur STICHWORT Was kommt nach der Lean Production? Frankfurter Allgemeine Zeitung, Verlagsbereich Wirtschaftsbücher, Blickbuch- Wirtschaft, Frankfurt am Main 1996, 164 Seiten. Neuauflage Reprint Verlag Science-Digital, Haiger

461 Published Books 461 ÜBER DAS KÄMPFEN Zum Phänomen des Kampfes in Sport und Gesellschaft Axel Binhack Kämpfen ist stets als ein Beziehungsphänomen, gleichsam als Fortsetzung der Kommunikation mit anderen Mitteln zu verstehen. Anhand von sechs Strukturmerkmalen wird eine Prototypik des Kampfes entwickelt. Ihre zusammenhängende Dynamik ist gesellschafts- und kulturunabhängig in vielfältigen Konfliktformen zu erkennen. In dem vorgelegten interdisziplinären Anschnitt wird die Struktur der Kampfbeziehung von vergleichbaren Sozialphänomenen abgegrenzt und am Beispiel des Kampfsports in Form des Zweikampfes analysiert. 1. Merkmale einer Formalstruktur des Kampfes Der duale Beziehungsaspekt: Antagonismus und Ambivalenz Der duale Trägeraspekt: Entscheidungsorientiertheit und tendenzielle energetische Totalität Der duale Inhaltsaspekt: Zweckgerichtetheit und riskante Offenheit 2. Die Strukturmerkmale des Kampfes im Vergleich mit Arbeit, Spiel und den Beziehungsphänomenen Tausch, Flirt und Macht Der antagonistische Unterschied Freiheit und Zweckgerichtetheit Zeitenthobene innere Unendlichkeit und zeitbeschleunigende Entscheidungsorientiertheit Die formalstrukturelle Folgenhaftigkeit des Kampfes 3. Zum Kampfsport Kampfsport als Radikalisierung des sportlichen Kampfes Der Kampfsport als kulturelles Phänomen Der Kampfdiskurs als Symbol männlichen Agierens Ästhetische Perspektiven 4. Die Kampfsportart Karate-Do als Beispiel Historische Entwicklungslinien des Karate-Do Der Schwertkampf und die Leere Hand : Zum karatespezifischen Aspekt der Waffenlosigkeit Die Kultivierung des Kampfphänomens im Karate-Do Der Kampf als Lehrmeister im Sinne des Zen Campus Verlag, Campus Forschung Bd. 768, Frankfurt am Main 1998

462 462 Pulished Books System der Rechte, demokratischer Rechtsstaat und Diskurstheorie des Rechts nach Jürgen Habermas hrsg. von Werner Krawietz / Gerhard Preyer Editorial: Zivilgesellschaftliche Assoziationen und spätmoderner Rechtssstaat (Werner Krawietz) I JURISTISCHE ENTSCHEIDUNG, BEGRÜNDUNG UND RECHTFERTIGUNG DES RECHTS IN DER PERSPEKTIVE DISKURSIVER REFLEXIONSTHEORIE Enrique P. Huba: Standortbestimmung in der zeitgenössischen Rechtstheorie - Rawls, Dworkin, Habermas, Alexy und andere Mitglieder der modernen Heiligen (Rede-) Familie, Thomas McCarthy: Legitimacy and Diversity: Dialectical Reflexions on Analytical Distinctions, Gerhard Preyer: Rechtsgeltung Argumentation Entscheidung II ORDNUNG, RECHT UND RATIONALITÄT IM JURISTISCHEN DISKURS Karl-Heinz Ladeur: Rechtliche Ordnungsbildung unter Ungewißheitsbedingungen und intersubjektive Rationalität, Ulfrid Neumann: Zur Interpretation des forensischen Diskurses in der Rechtsphilosophie von Jürgen Habermas, Ota Weinberger: Diskursive Demokratie ohne Diskurstheorie III KONSENSUSTHEORIE VERSUS SUBJEKTPHILOSOPHIE? VORAUSSETZUNGEN UND FOLGEN DES JURISTISCHEN DISKURSES IM DEMOKRATISCHEN RECHTSTAAT Michael Pawlik: Die Verdrängung des Subjekts und ihre Folgen. Begründungsdefizite in Habermas System der Rechte, Uwe Steinhoff: Probleme der Legitimation des demokratischen Rechtsstaats, William Rehg: The Place of Consensus in Democratic Legitimation: A Recommendation Rechtstheorie Zeitschrift für Logik, Methodenlehre, Normentheorie und Soziologie des Rechts , Habermas-Sonderheft (erschienen 1998) Duncker und Humblot, Berlin

463 Published Books 463 Donald Davidsons Philosophie Von der radikalen Interpretation zum radikalen Kontextualismus Gerhard Preyer Für Bruce Aune EINLEITUNG I RADIKALE INTERPRETATION, LOGISCHE FORM UND EREIGNISSE 1. Donald Davidsons Philosophie: Ein Überblick 2. Wahrheit, Bedeutung und radikale Interpretation 2.1 Radikale Interpretation als ein radikaler Externalismus (1) Der Zirkel zwischen Überzeugung und Bedeutung: Die Asymmetrie von RI (2) Die Strukur von Inhaltssätzen: sagen, daß... (3) Der Grundsatz der Nachsicht (4) Die Autonomie der Bedeutung (5) Radikaler Externalismus: Der innertheoretische Schritt (6) Die Ontologie von RI 2.2 Von der Idiolekttheorie zum dritten Dogma des Empirismus (1) Ausgangs- und Übergangstheorien der Interpretation (2) Die Demontage eines Mythos (3) Die epistemischen Beschränkungen von Verstehen 3. Die logische Form von Handlungssätzen und die singuläre Kausalaussage 3.1 Logische Form und adverbiale Modifikation 3.2 Kausale Beziehungen 4. Wahrheit und Überzeugung 4.1 Zum Haupteinwand 4.2 Zur empirischen Offensichtlichkeit von Einstellungszuschreibung 4.3 Rationalität als normativer Begriff II KÖRPERBEWEGUNGEN UND HANDLUNGEN 1. Das logische Verknüpfungsargument 2. Primäre Gründe und die Identitätsthese: Die synkategormatische Fassung von Absichten 3. Basisakte 4. Flucht vor den Körperbewegungen 4.1 Thalbergs Handlungstheorie und der Akkordeon-Effekt 4.2 A.I. Goldmans Kritik an der Identitätsthese 4.3 H.L.A. Hart: Zuschreibungen 4.4 Körperbewegungen als Bestandteil von Handlungen

464 464 Pulished Books III EINE RADIKALE THEORIE DES HANDELNS 1. Der Begriff der Einstellungsrationalität 1.1 Die Homogenität der Interpretation 1.2 Bewertende Einstellungen 2. Handlungsbeschreibungen und Handlungsverursachung 3. Eine Handlungserklärung 3.1 Überzeugungen, Absichten, Situationen 3.2 Praktische Gedanken 3.3 Handlungsgründe 3.4 Zu Hempels Ansatz 4. Praktische Schlüsse 4.1 Zur Gültigkeit praktischen Schließens 4.2 Entscheidungen und die Ausführung von Absichten 4.3 Überzeugungen und willentliche Handlungen 5. Radikaler Kontextualismus Literatur 290 Seiten Verlag Humanties Online, Frankfurt am Main,

465 Published Books 465 LANGUAGE, MIND AND EPISTEMOLOGY On Donald Davidson s Philosophy Gerhard Preyer, Frank Siebelt, Alexander Ulfig (eds.) Introduction: On Donald Davidson s Philosophy (Gerhard Preyer, Frank Siebelt, Alexander Ulfig) PHILOSOPHY OF LANGUAGE Jerry Fodor, Ernie Lepore (New Brunswick, USA): Meaning, Holism, and the Problem of Extensionality; Olav Gjelsvik (Oslo, Norway): Davidson s Use of Truth in Accounting for Meaning; Wilhelm K. Essler (Frankfurt/Main, Germany): Was ist Wahrheit? Arend Kulenkampff (Frankfurt/Main, Germany): Eigennamen und Kennzeichnungen EPISTEMOLOGY Roger F. Gibson (St. Louis, USA): Quine and Davidson: Two Naturalized Epistemologists; Eva Picardi (Bolongna, Italy): Davidson and Quine on Observation Sentences; Ralf Naumann (Düsseldorf, Germany): Events and Externalism; Dorit Bar-On (Chapel Hill, USA): Conceptual Relativism and Translation; David K. Henderson (Memphis, USA): Conceptual Schemes after Davidson; Frank Siebelt (Frankfurt/Main, Germany): Singular Causal Sentences and two Relational Views PHILOSOPHY OF MIND AND THEORY OF ACTION Louise M. Antony (Raleigh, USA): The Inadequacy of Anomalous Monism as a Realist Theory of Mind, Louise Röska-Hardy (Darmstadt, Germany): Internalism, Externalism and Davidsons Conception of the Mental, Marcia Cavell (Berkeley, USA): Dividing the Self, Ralf Stoecker (Bielefeld, Germany): Willensschwäche - Wie ist das nur möglich? Klaus Puhl (Graz, Austria): Davidson on intentional Content and Self-Knowledge, Johannes Brandl (Salzburg, Austria): Sharing Beliefs and the Myth of the Subjective, Kirk Ludwig (Gainesville, USA): First Person Knowledge and Authority, Gerhard Preyer (Frankfurt/Main, Germany): Rationalität: Absichten, Primärgründe und praktisches Denken. Donald Davidson: Dialectic and Dialogue 445 pages Kluwer Academic Publishers, Synthese Library, Dordrecht

466 466 Pulished Books Reality and Humean Supervenience Essays on the Philosophy of David Lewis Gerhard Preyer, Frank Siebelt (eds.) This is a very fine collection of essays about David Lewis s work. Centered around such central Lewisian doctrines as Humean supervenience, modal realism, the counterfactual analysis of causation, and physicalism, the collection is a fitting tribute to his outstanding work and to its influence. Jeremy Butterfield, University of Cambridge This collection gathers papers on David Lewis s metaphysics. Since the editors have assembled a superb group of authors, the result is at the cutting edge of the best work in contemporary metaphysics. Together with Lewis s writings it would make for an outstanding seminar in metaphysics. Ernest Sosa, Brown University David Lewis s work occupies a central place in contemporary metaphysics, and in this volume some of our leading metaphysicians comment, interpret, and critically appraise Lewis s important and influential claims and arguments on issues like modality and possible worlds, Humean supervenience, endurance and perdurance, counterfactuals and time, causation, and the mind-body problem. This is a timely and highly useful collection indispensable to students of Lewis s work in metaphysics and philosophy of mind. Jaegwon Kim, Brown University Preface Gerhard Preyer, Frank Siebelt Reality and Humean Supervenience - Some Reflections on David Lewis Philosophy MODAL REALISM Phillip Bricker: Island Universes and the Analysis of Modality, John Bigelow: Time Travel Fictions, Peter Forrest: Counting the Cost of Modal Realism, Paul Teller: Lewis s Defence of Counterpart Theory, Harold W. Noonan: The Case for Perdurance PHYSICALISM, CAUSATION AND CONDITIONALS Daniel Bonevac: Naturalism for the Faint of Heart, D. M. Armstrong: Going throug the Open Door again: Counterfactual vs. Singularists Theories of Causation, Jonathan Bennett: On Forward and Backward Counterfactual Conditionals REDUCTION OF MIND Terence Horgan: Multiple Reference, Multiple Realization and the Reduction of Mind, Michael Tye: Knowing what it is like: The Ability Hypothesis and the Knowledge Argument Rowman and Littlefield Publishers, Lanham, USA, 2001

467 Published Books 467 Wilhelm K. Essler Unser die Welt Sprachphilosophische Grundlegungen der Erkenntnistheorie Ausgewählte Artikel herausgegeben von Gerhard Preyer Einleitung: Zur Struktur von Erfahrung 1. Fundamentals of a Semi-Kantian Metaphysics of Knowledge 2. Kant und kein Ende 3. Tarski on Language and Truth 4. Am Anfang war die Tat Semantische und epistemologische Anmerkungen zur Sprachstufung 5. Was ist Wahrheit? 6. Gorgias hat recht! 7. Was ist und zu welchem Ende treibt man Metaphysik 8. Das logische Aufbauen von Welten 9. Unsere Welt trotz alledem 10. Selbst das Selbst ist nicht selbst 11. Erkenne dich selbst! Der Versuch von Hinweisen auf die Weisheit des Sokrates und auf die des Buddha Schakyamuni 12. Offenes Philosophieren Digitale Publikation: Verlag Humanities Online, Frankfurt am Main,

468 468 Pulished Books ProtoSociology: Digital volumes available Vol. 11 is a free download. Visit our homepage and go to Free Downloads. Vol. 17 Sematnic Theory and Reported Speech Vol. 16 Understanding the Social: New Perspectives from Epistemology Vol. 15 On a Sociology of Borderlines: Social Process in Time of Globalization Vol. 14 Folk Psychology, Mental Concepts and the Ascription of Attitudes Vol. 13 Reasoning and Argumentation Vol. 12 After the Received View Developments in the Theory of Science Vol. 11 Cognitive Semantics II Externalism in Debate (free download!) Vol. 10 Cognitive Semantics I Conceptions of Meaning Vol. 8/9 Rationality II &III (double volume) Order and download directly from our hompepage: Payment by credit card or bank tranfer: 15.- Euro each Or use the orderform on the next page and send to to: PROTOSOCIOLOGY, editorial office: Stephan-Heise-Str. 56, D Frankfurt am Main, Tel. 069/ Via preyer@em.uni-frankfurt.de or peter@protosociology.de Editor: Gerhard Preyer, Johann Wolfgang Goethe-Universität Frankfurt am Main FB 3: Department of Social Sciences. Editorial staff: Georg Peter.

469 Published Books 469 Subscription / Digital Version For the easiest and fastest ordering please visit our shop at: www. protosociology.de For printout: I subscribe PROTOSOCIOLOGY pdf-file 12,- Euro each issue, payable by credit card or bank transfer. Date, signature... Your address: Name... Address Send the order to: PROTOSOCIOLOGY Editorial office: Stephan-Heise-Str. 56, D Frankfurt am Main, preyer@em.uni-frankfurt.de, Tel. 069/ Editor: Gerhard Preyer, Johann Wolfgang Goethe-Universität Frankfurt am Main, Department of Social Sciences. Editorial staff: Georg Peter

470 470 Pulished Books Science -digital Publishing House and Service Provider for Scientific ebooks The founding of the publishing house Science-Digital is a result of the changes in modern sciences and ten years of experience with ProtoSociology: ebooks Sciences become more and more specialized, international and dynamic. So there is a need for a media that is fast, reaches a worldwide readership, inexpensive, and that will fit (in)to modern communication structures. Science-Digital offers scientific publications as ebooks for immediate download: first and new editions, serials, reprints or scientific journals with the main focus on philosophy, sociology and literary studies. With an easy to use, save payment system and encrypted to save the author s rights. Gerhard Preyer Die globale Herausforderung. Wie Deutschland an die Weltspitze zurückkehren kann 164 pages. PDF-file, 15.- Euro Gerhard Preyer, Jakob Schissler Integriertes Management. Was kommt nach der Lean Production? 260 pages. PDF-file, 15.- Euro International authors and editors will find a state of the art surrounding and network especially for topical readers and reprints of important works. For further information visit: Or contact: info@science-digital.com Georg Peter Publishing House Science-Digital Postal Address: Frankfurter Str. 122 D Offenbach am Main (Germany)

471 Published Books 471 Susanne Boshammer Gruppen, Rechte und Gerechtigkeit Die moralische Begründung der Rechte von Minderheiten X, 248 Seiten. Gebunden. S 49,95 [D] / sfr 80, ISBN Sollten muslimische Lehrerinnen im Unterricht ein Kopftuch tragen dürfen? Ist es gerecht, wenn Frauen bei der Vergabe von Arbeitsplätzen bevorzugt werden? Sollten die Kinder der Amish von der Schulpflicht befreit sein? Rechte von Minderheiten beschäftigen zunehmend die aktuelle Rechtsprechung und Politik. In der philosophischen Debatte sind sie nicht zuletzt deswegen so umstritten, weil sie dem Gleichheitsgedanken zu widersprechen scheinen. Es zeigt sich jedoch, dass die Anerkennung von Minderheitenrechten unter bestimmten Umständen moralisch gefordert ist. de Gruyter Philosophie Marco Iorio Karl Marx Geschichte, Gesellschaft, Politik Eine Ein- und Weiterführung XII, 372 Seiten. Broschur. S 29,95 [D] / sfr 48, ISBN Mit diesem Buch liegt eine beachtenswerte Neuinterpretation zentraler Teile des Marxismus der Marx schen Geschichts- und Gesellschaftstheorie vor. In einem ersten Schritt lässt Iorio Widersprüche und Mehrdeutigkeiten im Denken von Marx sichtbar werden. Es gelingt dem Autor jedoch, diese Widersprüche im Rahmen einer einheitlichen Theorie der Gesellschaft und des gesellschaftlichen Wandels zu überwinden, die vor allem darum bemüht ist, eine tiefgreifende Spannung zwischen dem Historischen Materialismus und dem marxistischen Konzept des Klassenkampfes aufzulösen. Diese Auflösung zeigt, wie die überindividualistisch strukturalistischen Aspekte des Historischen Materialismus mit Marxens Überzeugung zu versöhnen sind, dass menschliche Individuen Gesellschaften konstituieren und selbst ihre Geschichte machen. Bitte bestellen Sie bei Ihrer Buchhandlung oder direkt beim Verlag. Preisänderungen vorbehalten. Newsletter: WALTER DE GRUYTER GMBH & CO. KG Genthiner Straße Berlin Telefon +49-(0) Fax +49-(0) wdg-info@degruyter.de

472 472 Pulished Books Frank Schulz-Nieswandt Herrschaft und Genossenschaft Zur Anthropologie elementarer Formen sozialer Politik und der Gesellung auf historischer Grundlage Schriften zum Genossenschaftswesen und zur OÈ ffentlichen Wirtschaft, Band Seiten, 2003 h i A 56,80 / sfr 96,± Risikolagen und vulnerable Gruppen gab es zu allen Zeiten und Orten. Soziale Fragen, die gestaltend bewaèltigt werden wollen, existierten in allen kulturellen RaÈumen. Die elementaren Formen, in denen auf diese Herausforderungen in den unterschiedlichen Kulturen reagiert wurden, sind morphologisch aèhnlich. Frank Schulz-Nieswandt nimmt diesen Tatbestand zum Anlaû, aus anthropologischem Blickwinkel zwei archetypische Urformen sozialer Politik zu betrachten: Herrschaft und Genossenschaft. Im Mittelpunkt erschlieût der Autor die strukturale Formenlehre sozialer Politik im Dualismus des gemeinorientalischen Typus der vertikalen WohltaÈtigkeit ¹von obenª Barmherzigkeit) und des griechisch-roèmischen Typus der genossenschaftsartigen Hilfe auf Gegenseitigkeit BuÈrger-EgalitaÈt, aber auch als maskulin-agonale res publica). Aus diesen archetypischen Urkategorien heraus differenzieren sich die binaèren elementaren Formen als Typen vertikaler und horizontaler ReziprozitaÈt. Die herausgestellten Archetypen leben heute noch fort, beispielsweise in der steuerfinanzierten staatlichen Transferpolitik beduèrftigkeitsgepruèfte Sozialhilfe) und der asymmetrischen freiwilligen WohltaÈtigkeit sowie in der sozialversicherungsartigen Risikovergemeinschaftung und in Kleingenossenschaften der sozialen Selbsthilfe. Allen diesen differenzierten Formen sozialen Helfens und sozialer RisikobewaÈltigung gemein ist ihr Ursprung in dem Gabemechanismus als conditio humana. Duncker & Humblot GmbH Berlin Postfach D Berlin Telefax 0 30) Internet:

473 Published Books 473 Transactions of the Charles S. Peirce Society A Quarterly Journal in American Philosophy Peter H. Hare Randall R. Dipert Editors Since its founding in 1965, this has been the premier journal specializing in the history of American philosophy. Although named after the founder of American pragmatism, all types of American thought are covered, from the colonial period until the recent past. All books published in the field are discussed in essay reviews. Membership/Subscription Rates Special* $20 Individual $35 Institution $60 Sustaining $100 * Special Rate: Students/Retired/Unemployed Postage Outside USA Surface Mail: $10 Air Mail: $25 (USA postage included in Rate above) Back volumes available upon request. Transactions of the C. S. Peirce Society SUNY-Buffalo, Philosophy Department 135 Park Hall Buffalo, New York fax: phhare@acsu.buffalo.edu

474 474 Pulished Books REVIEW FERNAND BRAUDEL CENTER A Journal of the Fernand Braudel Center for the Study of Economies, Historical Systems, and Civilizations Vol. XXVI in 2003 has a special section in No. 1 on Uneven Development and a special issue on Ecology of the Modern World-System in No. 2 Previous Special Issues and Sections still available include: XXV, 3, 2002 Utopian Thinking XXIV, 1, 2001 Braudel and the U.S.: Interlocuteurs valables? XXIII, 4, 2000 Development Revisited XXIII, 1, 2000 Commodity Chains in the World- Economy, XXII, 4, 1999 Caribbean Migrants to Core Zones XXII, 3, 1999 ReOrientalism? XXI, 3 & 4, 1998 The States, the Markets, and the Societies: Separate Logics or a Single Domain? XX, 3/4, Sum./Fall 1997 Nomothetic vs. Idiographic Disciplines: A False Dilemma? XVIII, 1, Winter 1995 Labor Unrest in the World-Economy, XV, 4, Fall 1992 Two Views of World History A brochure containing the Table of Contents of past issues is available on request. Institutions $90/yr. Individuals $28/yr. Non-U.S. addresses, postage $8/yr. Special rate for low gnp per capita countries $10/yr. Managing Editor, Review Fernand Braudel Center Binghamton University State University of New York PO Box 6000 Binghamton, NY

475 Published Books 475 WHEN TALK IS A SCIENCE Linguistics and Language Behavior Abstracts Comprehensive, cost-effective, timely coverage of current ideas in linguistics and language research Abstracts of articles, books, and conference papers from nearly 1,500 journals published in 35 countries; citations of relevant dissertations as well as books and other media. Available in print or electronically through the Internet Database Service from Cambridge Scientific Abstracts ( Contact sales@csa.com for trial Internet access or a sample issue. Linguistics Language Behavior Abstracts Published by CSA

476 476 Pulished Books sociological abstracts Comprehensive, cost-effective, timely Add dimension to your sociological research Abstracts of articles, books, and conference papers from nearly 2,500 journals published in 35 countries; citations of relevant dissertations as well as books and other media. Available in print or electronically through the Internet Database Service from Cambridge Scientific Abstracts ( Contact for trial Internet access or a sample issue. sociological abstracts Published by CSA

477 Titel 477 Jetzt: Die Studienausgabe in der Reihe UTB Manuel Castells Das Informationszeitalter Band 3: Jahrtausendwende UTB L Ca. 600 Seiten. Kart. Ca. 16,90 Euro ISBN UTB-ISBN Band 2: Die Macht der Identität UTB L Ca. 500 Seiten. Kart. Ca. 16,90 Euro ISBN UTB-ISBN Band 1: Der Aufstieg der Netzwerkgesellschaft UTB L Ca. 600 Seiten. Kart. Ca. 16,90 Euro ISBN UTB-ISBN Der fesselndste Versuch, der je gemacht wurde, die Konturen des globalen Informationszeitalters nachzuzeichnen. Anthony Giddens, New Statesman (London) w w w. l e s k e - b u d r i c h. d e

Comments on Seumas Miller s review of Social Ontology: Collective Intentionality and Group agents in the Notre Dame Philosophical Reviews (April 20, 2

Comments on Seumas Miller s review of Social Ontology: Collective Intentionality and Group agents in the Notre Dame Philosophical Reviews (April 20, 2 Comments on Seumas Miller s review of Social Ontology: Collective Intentionality and Group agents in the Notre Dame Philosophical Reviews (April 20, 2014) Miller s review contains many misunderstandings

More information

Mikhael Dua. Tacit Knowing. Michael Polanyi s Exposition of Scientific Knowledge. Herbert Utz Verlag Wissenschaft München

Mikhael Dua. Tacit Knowing. Michael Polanyi s Exposition of Scientific Knowledge. Herbert Utz Verlag Wissenschaft München Mikhael Dua Tacit Knowing Michael Polanyi s Exposition of Scientific Knowledge Herbert Utz Verlag Wissenschaft München Bibliografische Information Der Deutschen Bibliothek Die Deutsche Bibliothek verzeichnet

More information

Philosophical Perspectives, 16, Language and Mind, 2002 THE AIM OF BELIEF 1. Ralph Wedgwood Merton College, Oxford

Philosophical Perspectives, 16, Language and Mind, 2002 THE AIM OF BELIEF 1. Ralph Wedgwood Merton College, Oxford Philosophical Perspectives, 16, Language and Mind, 2002 THE AIM OF BELIEF 1 Ralph Wedgwood Merton College, Oxford 0. Introduction It is often claimed that beliefs aim at the truth. Indeed, this claim has

More information

Varieties of Apriority

Varieties of Apriority S E V E N T H E X C U R S U S Varieties of Apriority T he notions of a priori knowledge and justification play a central role in this work. There are many ways in which one can understand the a priori,

More information

Moral Argumentation from a Rhetorical Point of View

Moral Argumentation from a Rhetorical Point of View Chapter 98 Moral Argumentation from a Rhetorical Point of View Lars Leeten Universität Hildesheim Practical thinking is a tricky business. Its aim will never be fulfilled unless influence on practical

More information

Lecture 3. I argued in the previous lecture for a relationist solution to Frege's puzzle, one which

Lecture 3. I argued in the previous lecture for a relationist solution to Frege's puzzle, one which 1 Lecture 3 I argued in the previous lecture for a relationist solution to Frege's puzzle, one which posits a semantic difference between the pairs of names 'Cicero', 'Cicero' and 'Cicero', 'Tully' even

More information

THE MEANING OF OUGHT. Ralph Wedgwood. What does the word ought mean? Strictly speaking, this is an empirical question, about the

THE MEANING OF OUGHT. Ralph Wedgwood. What does the word ought mean? Strictly speaking, this is an empirical question, about the THE MEANING OF OUGHT Ralph Wedgwood What does the word ought mean? Strictly speaking, this is an empirical question, about the meaning of a word in English. Such empirical semantic questions should ideally

More information

Final Paper. May 13, 2015

Final Paper. May 13, 2015 24.221 Final Paper May 13, 2015 Determinism states the following: given the state of the universe at time t 0, denoted S 0, and the conjunction of the laws of nature, L, the state of the universe S at

More information

The ontology of human rights and obligations

The ontology of human rights and obligations The ontology of human rights and obligations Åsa Burman Department of Philosophy, Stockholm University asa.burman@philosophy.su.se If we are going to make sense of the notion of rights we have to answer

More information

Ayer on the criterion of verifiability

Ayer on the criterion of verifiability Ayer on the criterion of verifiability November 19, 2004 1 The critique of metaphysics............................. 1 2 Observation statements............................... 2 3 In principle verifiability...............................

More information

Vol 2 Bk 7 Outline p 486 BOOK VII. Substance, Essence and Definition CONTENTS. Book VII

Vol 2 Bk 7 Outline p 486 BOOK VII. Substance, Essence and Definition CONTENTS. Book VII Vol 2 Bk 7 Outline p 486 BOOK VII Substance, Essence and Definition CONTENTS Book VII Lesson 1. The Primacy of Substance. Its Priority to Accidents Lesson 2. Substance as Form, as Matter, and as Body.

More information

To link to this article:

To link to this article: This article was downloaded by: [University of Chicago Library] On: 24 May 2013, At: 08:10 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office:

More information

Are There Reasons to Be Rational?

Are There Reasons to Be Rational? Are There Reasons to Be Rational? Olav Gjelsvik, University of Oslo The thesis. Among people writing about rationality, few people are more rational than Wlodek Rabinowicz. But are there reasons for being

More information

In Defense of Radical Empiricism. Joseph Benjamin Riegel. Chapel Hill 2006

In Defense of Radical Empiricism. Joseph Benjamin Riegel. Chapel Hill 2006 In Defense of Radical Empiricism Joseph Benjamin Riegel A thesis submitted to the faculty of the University of North Carolina at Chapel Hill in partial fulfillment of the requirements for the degree of

More information

1 What is conceptual analysis and what is the problem?

1 What is conceptual analysis and what is the problem? 1 What is conceptual analysis and what is the problem? 1.1 What is conceptual analysis? In this book, I am going to defend the viability of conceptual analysis as a philosophical method. It therefore seems

More information

Ayer and Quine on the a priori

Ayer and Quine on the a priori Ayer and Quine on the a priori November 23, 2004 1 The problem of a priori knowledge Ayer s book is a defense of a thoroughgoing empiricism, not only about what is required for a belief to be justified

More information

FUNDAMENTAL PRINCIPLES OF THE METAPHYSIC OF MORALS. by Immanuel Kant

FUNDAMENTAL PRINCIPLES OF THE METAPHYSIC OF MORALS. by Immanuel Kant FUNDAMENTAL PRINCIPLES OF THE METAPHYSIC OF MORALS SECOND SECTION by Immanuel Kant TRANSITION FROM POPULAR MORAL PHILOSOPHY TO THE METAPHYSIC OF MORALS... This principle, that humanity and generally every

More information

CONVENTIONALISM AND NORMATIVITY

CONVENTIONALISM AND NORMATIVITY 1 CONVENTIONALISM AND NORMATIVITY TORBEN SPAAK We have seen (in Section 3) that Hart objects to Austin s command theory of law, that it cannot account for the normativity of law, and that what is missing

More information

Oxford Scholarship Online Abstracts and Keywords

Oxford Scholarship Online Abstracts and Keywords Oxford Scholarship Online Abstracts and Keywords ISBN 9780198802693 Title The Value of Rationality Author(s) Ralph Wedgwood Book abstract Book keywords Rationality is a central concept for epistemology,

More information

Wright on response-dependence and self-knowledge

Wright on response-dependence and self-knowledge Wright on response-dependence and self-knowledge March 23, 2004 1 Response-dependent and response-independent concepts........... 1 1.1 The intuitive distinction......................... 1 1.2 Basic equations

More information

Verificationism. PHIL September 27, 2011

Verificationism. PHIL September 27, 2011 Verificationism PHIL 83104 September 27, 2011 1. The critique of metaphysics... 1 2. Observation statements... 2 3. In principle verifiability... 3 4. Strong verifiability... 3 4.1. Conclusive verifiability

More information

Introduction. I. Proof of the Minor Premise ( All reality is completely intelligible )

Introduction. I. Proof of the Minor Premise ( All reality is completely intelligible ) Philosophical Proof of God: Derived from Principles in Bernard Lonergan s Insight May 2014 Robert J. Spitzer, S.J., Ph.D. Magis Center of Reason and Faith Lonergan s proof may be stated as follows: Introduction

More information

Raimo Tuomela: Social Ontology: Collective Intentionality and Group Agents. New York, USA: Oxford University Press, 2013, 326 pp.

Raimo Tuomela: Social Ontology: Collective Intentionality and Group Agents. New York, USA: Oxford University Press, 2013, 326 pp. Journal of Social Ontology 2015; 1(1): 183 187 Book Review Open Access DOI 10.1515/jso-2014-0040 Raimo Tuomela: Social Ontology: Collective Intentionality and Group Agents. New York, USA: Oxford University

More information

Saving the Substratum: Interpreting Kant s First Analogy

Saving the Substratum: Interpreting Kant s First Analogy Res Cogitans Volume 5 Issue 1 Article 20 6-4-2014 Saving the Substratum: Interpreting Kant s First Analogy Kevin Harriman Lewis & Clark College Follow this and additional works at: http://commons.pacificu.edu/rescogitans

More information

THE CONCEPT OF OWNERSHIP by Lars Bergström

THE CONCEPT OF OWNERSHIP by Lars Bergström From: Who Owns Our Genes?, Proceedings of an international conference, October 1999, Tallin, Estonia, The Nordic Committee on Bioethics, 2000. THE CONCEPT OF OWNERSHIP by Lars Bergström I shall be mainly

More information

Natujwa Umbertina Mvungi. Challenges in the Implementation of the East African Community Common Market Protocol

Natujwa Umbertina Mvungi. Challenges in the Implementation of the East African Community Common Market Protocol Natujwa Umbertina Mvungi Challenges in the Implementation of the East African Community Common Market Protocol GUC - Verlag der Gesellschaft für Unternehmensrechnung und Controlling m.b.h. Chemnitz 2011

More information

The Inscrutability of Reference and the Scrutability of Truth

The Inscrutability of Reference and the Scrutability of Truth SECOND EXCURSUS The Inscrutability of Reference and the Scrutability of Truth I n his 1960 book Word and Object, W. V. Quine put forward the thesis of the Inscrutability of Reference. This thesis says

More information

Jeu-Jenq Yuann Professor of Philosophy Department of Philosophy, National Taiwan University,

Jeu-Jenq Yuann Professor of Philosophy Department of Philosophy, National Taiwan University, The Negative Role of Empirical Stimulus in Theory Change: W. V. Quine and P. Feyerabend Jeu-Jenq Yuann Professor of Philosophy Department of Philosophy, National Taiwan University, 1 To all Participants

More information

Ayer s linguistic theory of the a priori

Ayer s linguistic theory of the a priori Ayer s linguistic theory of the a priori phil 43904 Jeff Speaks December 4, 2007 1 The problem of a priori knowledge....................... 1 2 Necessity and the a priori............................ 2

More information

Direct Realism and the Brain-in-a-Vat Argument by Michael Huemer (2000)

Direct Realism and the Brain-in-a-Vat Argument by Michael Huemer (2000) Direct Realism and the Brain-in-a-Vat Argument by Michael Huemer (2000) One of the advantages traditionally claimed for direct realist theories of perception over indirect realist theories is that the

More information

Semantic Foundations for Deductive Methods

Semantic Foundations for Deductive Methods Semantic Foundations for Deductive Methods delineating the scope of deductive reason Roger Bishop Jones Abstract. The scope of deductive reason is considered. First a connection is discussed between the

More information

Philosophy 5340 Epistemology. Topic 6: Theories of Justification: Foundationalism versus Coherentism. Part 2: Susan Haack s Foundherentist Approach

Philosophy 5340 Epistemology. Topic 6: Theories of Justification: Foundationalism versus Coherentism. Part 2: Susan Haack s Foundherentist Approach Philosophy 5340 Epistemology Topic 6: Theories of Justification: Foundationalism versus Coherentism Part 2: Susan Haack s Foundherentist Approach Susan Haack, "A Foundherentist Theory of Empirical Justification"

More information

2.3. Failed proofs and counterexamples

2.3. Failed proofs and counterexamples 2.3. Failed proofs and counterexamples 2.3.0. Overview Derivations can also be used to tell when a claim of entailment does not follow from the principles for conjunction. 2.3.1. When enough is enough

More information

The Paradox of the stone and two concepts of omnipotence

The Paradox of the stone and two concepts of omnipotence Filo Sofija Nr 30 (2015/3), s. 239-246 ISSN 1642-3267 Jacek Wojtysiak John Paul II Catholic University of Lublin The Paradox of the stone and two concepts of omnipotence Introduction The history of science

More information

Reply to Robert Koons

Reply to Robert Koons 632 Notre Dame Journal of Formal Logic Volume 35, Number 4, Fall 1994 Reply to Robert Koons ANIL GUPTA and NUEL BELNAP We are grateful to Professor Robert Koons for his excellent, and generous, review

More information

Two Kinds of Ends in Themselves in Kant s Moral Theory

Two Kinds of Ends in Themselves in Kant s Moral Theory Western University Scholarship@Western 2015 Undergraduate Awards The Undergraduate Awards 2015 Two Kinds of Ends in Themselves in Kant s Moral Theory David Hakim Western University, davidhakim266@gmail.com

More information

Remarks on the philosophy of mathematics (1969) Paul Bernays

Remarks on the philosophy of mathematics (1969) Paul Bernays Bernays Project: Text No. 26 Remarks on the philosophy of mathematics (1969) Paul Bernays (Bemerkungen zur Philosophie der Mathematik) Translation by: Dirk Schlimm Comments: With corrections by Charles

More information

Philosophy 1100: Introduction to Ethics. Critical Thinking Lecture 1. Background Material for the Exercise on Validity

Philosophy 1100: Introduction to Ethics. Critical Thinking Lecture 1. Background Material for the Exercise on Validity Philosophy 1100: Introduction to Ethics Critical Thinking Lecture 1 Background Material for the Exercise on Validity Reasons, Arguments, and the Concept of Validity 1. The Concept of Validity Consider

More information

Choosing Rationally and Choosing Correctly *

Choosing Rationally and Choosing Correctly * Choosing Rationally and Choosing Correctly * Ralph Wedgwood 1 Two views of practical reason Suppose that you are faced with several different options (that is, several ways in which you might act in a

More information

WHY IS GOD GOOD? EUTYPHRO, TIMAEUS AND THE DIVINE COMMAND THEORY

WHY IS GOD GOOD? EUTYPHRO, TIMAEUS AND THE DIVINE COMMAND THEORY Miłosz Pawłowski WHY IS GOD GOOD? EUTYPHRO, TIMAEUS AND THE DIVINE COMMAND THEORY In Eutyphro Plato presents a dilemma 1. Is it that acts are good because God wants them to be performed 2? Or are they

More information

THE NATURE OF NORMATIVITY IN KANT S PHILOSOPHY OF LOGIC REBECCA V. MILLSOP S

THE NATURE OF NORMATIVITY IN KANT S PHILOSOPHY OF LOGIC REBECCA V. MILLSOP S THE NATURE OF NORMATIVITY IN KANT S PHILOSOPHY OF LOGIC REBECCA V. MILLSOP S I. INTRODUCTION Immanuel Kant claims that logic is constitutive of thought: without [the laws of logic] we would not think at

More information

THE SEMANTIC REALISM OF STROUD S RESPONSE TO AUSTIN S ARGUMENT AGAINST SCEPTICISM

THE SEMANTIC REALISM OF STROUD S RESPONSE TO AUSTIN S ARGUMENT AGAINST SCEPTICISM SKÉPSIS, ISSN 1981-4194, ANO VII, Nº 14, 2016, p. 33-39. THE SEMANTIC REALISM OF STROUD S RESPONSE TO AUSTIN S ARGUMENT AGAINST SCEPTICISM ALEXANDRE N. MACHADO Universidade Federal do Paraná (UFPR) Email:

More information

Informalizing Formal Logic

Informalizing Formal Logic Informalizing Formal Logic Antonis Kakas Department of Computer Science, University of Cyprus, Cyprus antonis@ucy.ac.cy Abstract. This paper discusses how the basic notions of formal logic can be expressed

More information

Scanlon on Double Effect

Scanlon on Double Effect Scanlon on Double Effect RALPH WEDGWOOD Merton College, University of Oxford In this new book Moral Dimensions, T. M. Scanlon (2008) explores the ethical significance of the intentions and motives with

More information

Reply to Kit Fine. Theodore Sider July 19, 2013

Reply to Kit Fine. Theodore Sider July 19, 2013 Reply to Kit Fine Theodore Sider July 19, 2013 Kit Fine s paper raises important and difficult issues about my approach to the metaphysics of fundamentality. In chapters 7 and 8 I examined certain subtle

More information

Skepticism and Internalism

Skepticism and Internalism Skepticism and Internalism John Greco Abstract: This paper explores a familiar skeptical problematic and considers some strategies for responding to it. Section 1 reconstructs and disambiguates the skeptical

More information

Hello again. Today we re gonna continue our discussions of Kant s ethics.

Hello again. Today we re gonna continue our discussions of Kant s ethics. PHI 110 Lecture 29 1 Hello again. Today we re gonna continue our discussions of Kant s ethics. Last time we talked about the good will and Kant defined the good will as the free rational will which acts

More information

Aboutness and Justification

Aboutness and Justification For a symposium on Imogen Dickie s book Fixing Reference to be published in Philosophy and Phenomenological Research. Aboutness and Justification Dilip Ninan dilip.ninan@tufts.edu September 2016 Al believes

More information

Kantian Humility and Ontological Categories Sam Cowling University of Massachusetts, Amherst

Kantian Humility and Ontological Categories Sam Cowling University of Massachusetts, Amherst Kantian Humility and Ontological Categories Sam Cowling University of Massachusetts, Amherst [Forthcoming in Analysis. Penultimate Draft. Cite published version.] Kantian Humility holds that agents like

More information

TWO VERSIONS OF HUME S LAW

TWO VERSIONS OF HUME S LAW DISCUSSION NOTE BY CAMPBELL BROWN JOURNAL OF ETHICS & SOCIAL PHILOSOPHY DISCUSSION NOTE MAY 2015 URL: WWW.JESP.ORG COPYRIGHT CAMPBELL BROWN 2015 Two Versions of Hume s Law MORAL CONCLUSIONS CANNOT VALIDLY

More information

Received: 30 August 2007 / Accepted: 16 November 2007 / Published online: 28 December 2007 # Springer Science + Business Media B.V.

Received: 30 August 2007 / Accepted: 16 November 2007 / Published online: 28 December 2007 # Springer Science + Business Media B.V. Acta anal. (2007) 22:267 279 DOI 10.1007/s12136-007-0012-y What Is Entitlement? Albert Casullo Received: 30 August 2007 / Accepted: 16 November 2007 / Published online: 28 December 2007 # Springer Science

More information

Deontological Perspectivism: A Reply to Lockie Hamid Vahid, Institute for Research in Fundamental Sciences, Tehran

Deontological Perspectivism: A Reply to Lockie Hamid Vahid, Institute for Research in Fundamental Sciences, Tehran Deontological Perspectivism: A Reply to Lockie Hamid Vahid, Institute for Research in Fundamental Sciences, Tehran Abstract In his (2015) paper, Robert Lockie seeks to add a contextualized, relativist

More information

1 Hans Jonas, The Imperative of Responsibility: In Search of an Ethics for the Technological Age (Chicago: University of Chicago Press, 1984), 1-10.

1 Hans Jonas, The Imperative of Responsibility: In Search of an Ethics for the Technological Age (Chicago: University of Chicago Press, 1984), 1-10. Introduction This book seeks to provide a metaethical analysis of the responsibility ethics of two of its prominent defenders: H. Richard Niebuhr and Emmanuel Levinas. In any ethical writings, some use

More information

24.01 Classics of Western Philosophy

24.01 Classics of Western Philosophy 1 Plan: Kant Lecture #2: How are pure mathematics and pure natural science possible? 1. Review: Problem of Metaphysics 2. Kantian Commitments 3. Pure Mathematics 4. Transcendental Idealism 5. Pure Natural

More information

CONTENTS A SYSTEM OF LOGIC

CONTENTS A SYSTEM OF LOGIC EDITOR'S INTRODUCTION NOTE ON THE TEXT. SELECTED BIBLIOGRAPHY XV xlix I /' ~, r ' o>

More information

Faith and Philosophy, April (2006), DE SE KNOWLEDGE AND THE POSSIBILITY OF AN OMNISCIENT BEING Stephan Torre

Faith and Philosophy, April (2006), DE SE KNOWLEDGE AND THE POSSIBILITY OF AN OMNISCIENT BEING Stephan Torre 1 Faith and Philosophy, April (2006), 191-200. Penultimate Draft DE SE KNOWLEDGE AND THE POSSIBILITY OF AN OMNISCIENT BEING Stephan Torre In this paper I examine an argument that has been made by Patrick

More information

1/12. The A Paralogisms

1/12. The A Paralogisms 1/12 The A Paralogisms The character of the Paralogisms is described early in the chapter. Kant describes them as being syllogisms which contain no empirical premises and states that in them we conclude

More information

Freedom as Morality. UWM Digital Commons. University of Wisconsin Milwaukee. Hao Liang University of Wisconsin-Milwaukee. Theses and Dissertations

Freedom as Morality. UWM Digital Commons. University of Wisconsin Milwaukee. Hao Liang University of Wisconsin-Milwaukee. Theses and Dissertations University of Wisconsin Milwaukee UWM Digital Commons Theses and Dissertations May 2014 Freedom as Morality Hao Liang University of Wisconsin-Milwaukee Follow this and additional works at: http://dc.uwm.edu/etd

More information

Can logical consequence be deflated?

Can logical consequence be deflated? Can logical consequence be deflated? Michael De University of Utrecht Department of Philosophy Utrecht, Netherlands mikejde@gmail.com in Insolubles and Consequences : essays in honour of Stephen Read,

More information

Comments on Truth at A World for Modal Propositions

Comments on Truth at A World for Modal Propositions Comments on Truth at A World for Modal Propositions Christopher Menzel Texas A&M University March 16, 2008 Since Arthur Prior first made us aware of the issue, a lot of philosophical thought has gone into

More information

part one MACROSTRUCTURE Cambridge University Press X - A Theory of Argument Mark Vorobej Excerpt More information

part one MACROSTRUCTURE Cambridge University Press X - A Theory of Argument Mark Vorobej Excerpt More information part one MACROSTRUCTURE 1 Arguments 1.1 Authors and Audiences An argument is a social activity, the goal of which is interpersonal rational persuasion. More precisely, we ll say that an argument occurs

More information

Constructing the World

Constructing the World Constructing the World Lecture 3: The Case for A Priori Scrutability David Chalmers Plan *1. Sentences vs Propositions 2. Apriority and A Priori Scrutability 3. Argument 1: Suspension of Judgment 4. Argument

More information

The Greatest Mistake: A Case for the Failure of Hegel s Idealism

The Greatest Mistake: A Case for the Failure of Hegel s Idealism The Greatest Mistake: A Case for the Failure of Hegel s Idealism What is a great mistake? Nietzsche once said that a great error is worth more than a multitude of trivial truths. A truly great mistake

More information

SUPPOSITIONAL REASONING AND PERCEPTUAL JUSTIFICATION

SUPPOSITIONAL REASONING AND PERCEPTUAL JUSTIFICATION SUPPOSITIONAL REASONING AND PERCEPTUAL JUSTIFICATION Stewart COHEN ABSTRACT: James Van Cleve raises some objections to my attempt to solve the bootstrapping problem for what I call basic justification

More information

ProtoSociology An International Journal of Interdisciplinary Research

ProtoSociology An International Journal of Interdisciplinary Research ProtoSociology An International Journal of Interdisciplinary Research Volume 27, 2011 Modernization in Times of Globalization II www.protosociology.de 2011 Gerhard Preyer Frankfurt am Main http://www.protosociology.de

More information

What is a counterexample?

What is a counterexample? Lorentz Center 4 March 2013 What is a counterexample? Jan-Willem Romeijn, University of Groningen Joint work with Eric Pacuit, University of Maryland Paul Pedersen, Max Plank Institute Berlin Co-authors

More information

Kolumne. ProtoSociology. An International Journal of Interdisciplinary Research. Concepts Contemporary and Historical Perspectives

Kolumne. ProtoSociology. An International Journal of Interdisciplinary Research. Concepts Contemporary and Historical Perspectives Kolumne 1 ProtoSociology An International Journal of Interdisciplinary Research Volume 30, 2013 Concepts Contemporary and Historical Perspectives www.protosociology.de ProtoSociology Volume 30/2013: Concepts

More information

World without Design: The Ontological Consequences of Natural- ism , by Michael C. Rea.

World without Design: The Ontological Consequences of Natural- ism , by Michael C. Rea. Book reviews World without Design: The Ontological Consequences of Naturalism, by Michael C. Rea. Oxford: Clarendon Press, 2004, viii + 245 pp., $24.95. This is a splendid book. Its ideas are bold and

More information

Rule-Following and Constitutive Rules: A Reconciliation

Rule-Following and Constitutive Rules: A Reconciliation Rule-Following and Constitutive Rules: A Reconciliation Cyril Hédoin University of Reims Champagne-Ardenne (France) Version 2.0: 19 th March 2017 Abstract: This article contrasts two broad approaches of

More information

Primitive Concepts. David J. Chalmers

Primitive Concepts. David J. Chalmers Primitive Concepts David J. Chalmers Conceptual Analysis: A Traditional View A traditional view: Most ordinary concepts (or expressions) can be defined in terms of other more basic concepts (or expressions)

More information

Practical Rationality and Ethics. Basic Terms and Positions

Practical Rationality and Ethics. Basic Terms and Positions Practical Rationality and Ethics Basic Terms and Positions Practical reasons and moral ought Reasons are given in answer to the sorts of questions ethics seeks to answer: What should I do? How should I

More information

Can A Priori Justified Belief Be Extended Through Deduction? It is often assumed that if one deduces some proposition p from some premises

Can A Priori Justified Belief Be Extended Through Deduction? It is often assumed that if one deduces some proposition p from some premises Can A Priori Justified Belief Be Extended Through Deduction? Introduction It is often assumed that if one deduces some proposition p from some premises which one knows a priori, in a series of individually

More information

Rationalism. A. He, like others at the time, was obsessed with questions of truth and doubt

Rationalism. A. He, like others at the time, was obsessed with questions of truth and doubt Rationalism I. Descartes (1596-1650) A. He, like others at the time, was obsessed with questions of truth and doubt 1. How could one be certain in the absence of religious guidance and trustworthy senses

More information

2 FREE CHOICE The heretical thesis of Hobbes is the orthodox position today. So much is this the case that most of the contemporary literature

2 FREE CHOICE The heretical thesis of Hobbes is the orthodox position today. So much is this the case that most of the contemporary literature Introduction The philosophical controversy about free will and determinism is perennial. Like many perennial controversies, this one involves a tangle of distinct but closely related issues. Thus, the

More information

Rationalist-Irrationalist Dialectic in Buddhism:

Rationalist-Irrationalist Dialectic in Buddhism: Rationalist-Irrationalist Dialectic in Buddhism: The Failure of Buddhist Epistemology By W. J. Whitman The problem of the one and the many is the core issue at the heart of all real philosophical and theological

More information

semantic-extensional interpretation that happens to satisfy all the axioms.

semantic-extensional interpretation that happens to satisfy all the axioms. No axiom, no deduction 1 Where there is no axiom-system, there is no deduction. I think this is a fair statement (for most of us) at least if we understand (i) "an axiom-system" in a certain logical-expressive/normative-pragmatical

More information

An alternative understanding of interpretations: Incompatibility Semantics

An alternative understanding of interpretations: Incompatibility Semantics An alternative understanding of interpretations: Incompatibility Semantics 1. In traditional (truth-theoretic) semantics, interpretations serve to specify when statements are true and when they are false.

More information

Review of David J. Chalmers Constructing the World (OUP 2012) David Chalmers burst onto the philosophical scene in the mid-1990s with his work on

Review of David J. Chalmers Constructing the World (OUP 2012) David Chalmers burst onto the philosophical scene in the mid-1990s with his work on Review of David J. Chalmers Constructing the World (OUP 2012) Thomas W. Polger, University of Cincinnati 1. Introduction David Chalmers burst onto the philosophical scene in the mid-1990s with his work

More information

OSSA Conference Archive OSSA 5

OSSA Conference Archive OSSA 5 University of Windsor Scholarship at UWindsor OSSA Conference Archive OSSA 5 May 14th, 9:00 AM - May 17th, 5:00 PM Commentary pm Krabbe Dale Jacquette Follow this and additional works at: http://scholar.uwindsor.ca/ossaarchive

More information

A CONTRACTUALIST READING OF KANT S PROOF OF THE FORMULA OF HUMANITY. Adam Cureton

A CONTRACTUALIST READING OF KANT S PROOF OF THE FORMULA OF HUMANITY. Adam Cureton A CONTRACTUALIST READING OF KANT S PROOF OF THE FORMULA OF HUMANITY Adam Cureton Abstract: Kant offers the following argument for the Formula of Humanity: Each rational agent necessarily conceives of her

More information

THE RELATION BETWEEN THE GENERAL MAXIM OF CAUSALITY AND THE PRINCIPLE OF UNIFORMITY IN HUME S THEORY OF KNOWLEDGE

THE RELATION BETWEEN THE GENERAL MAXIM OF CAUSALITY AND THE PRINCIPLE OF UNIFORMITY IN HUME S THEORY OF KNOWLEDGE CDD: 121 THE RELATION BETWEEN THE GENERAL MAXIM OF CAUSALITY AND THE PRINCIPLE OF UNIFORMITY IN HUME S THEORY OF KNOWLEDGE Departamento de Filosofia Instituto de Filosofia e Ciências Humanas IFCH Universidade

More information

the aim is to specify the structure of the world in the form of certain basic truths from which all truths can be derived. (xviii)

the aim is to specify the structure of the world in the form of certain basic truths from which all truths can be derived. (xviii) PHIL 5983: Naturalness and Fundamentality Seminar Prof. Funkhouser Spring 2017 Week 8: Chalmers, Constructing the World Notes (Introduction, Chapters 1-2) Introduction * We are introduced to the ideas

More information

METHODENSTREIT WHY CARL MENGER WAS, AND IS, RIGHT

METHODENSTREIT WHY CARL MENGER WAS, AND IS, RIGHT METHODENSTREIT WHY CARL MENGER WAS, AND IS, RIGHT BY THORSTEN POLLEIT* PRESENTED AT THE SPRING CONFERENCE RESEARCH ON MONEY IN THE ECONOMY (ROME) FRANKFURT, 20 MAY 2011 *FRANKFURT SCHOOL OF FINANCE & MANAGEMENT

More information

Kant s Fundamental Principles of the Metaphysic of Morals

Kant s Fundamental Principles of the Metaphysic of Morals Kant s Fundamental Principles of the Metaphysic of Morals G. J. Mattey Spring, 2017/ Philosophy 1 The Division of Philosophical Labor Kant generally endorses the ancient Greek division of philosophy into

More information

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. The Physical World Author(s): Barry Stroud Source: Proceedings of the Aristotelian Society, New Series, Vol. 87 (1986-1987), pp. 263-277 Published by: Blackwell Publishing on behalf of The Aristotelian

More information

Ulrich Haarmann Memorial Lecture

Ulrich Haarmann Memorial Lecture Ulrich Haarmann Memorial Lecture ed. Stephan Conermann Volume 6 Irmeli Perho Ibn Taghrībirdī s portrayal of the first Mamluk rulers EBVERLAG Ibn Taghrībirdī s portrayal of the first Mamluk rulers Ulrich

More information

WHY THERE REALLY ARE NO IRREDUCIBLY NORMATIVE PROPERTIES

WHY THERE REALLY ARE NO IRREDUCIBLY NORMATIVE PROPERTIES WHY THERE REALLY ARE NO IRREDUCIBLY NORMATIVE PROPERTIES Bart Streumer b.streumer@rug.nl In David Bakhurst, Brad Hooker and Margaret Little (eds.), Thinking About Reasons: Essays in Honour of Jonathan

More information

A Priori Bootstrapping

A Priori Bootstrapping A Priori Bootstrapping Ralph Wedgwood In this essay, I shall explore the problems that are raised by a certain traditional sceptical paradox. My conclusion, at the end of this essay, will be that the most

More information

Etchemendy, Tarski, and Logical Consequence 1 Jared Bates, University of Missouri Southwest Philosophy Review 15 (1999):

Etchemendy, Tarski, and Logical Consequence 1 Jared Bates, University of Missouri Southwest Philosophy Review 15 (1999): Etchemendy, Tarski, and Logical Consequence 1 Jared Bates, University of Missouri Southwest Philosophy Review 15 (1999): 47 54. Abstract: John Etchemendy (1990) has argued that Tarski's definition of logical

More information

Chapter 5: Freedom and Determinism

Chapter 5: Freedom and Determinism Chapter 5: Freedom and Determinism At each time t the world is perfectly determinate in all detail. - Let us grant this for the sake of argument. We might want to re-visit this perfectly reasonable assumption

More information

A Review of Neil Feit s Belief about the Self

A Review of Neil Feit s Belief about the Self A Review of Neil Feit s Belief about the Self Stephan Torre 1 Neil Feit. Belief about the Self. Oxford GB: Oxford University Press 2008. 216 pages. Belief about the Self is a clearly written, engaging

More information

Florida State University Libraries

Florida State University Libraries Florida State University Libraries Electronic Theses, Treatises and Dissertations The Graduate School 2011 A Framework for Understanding Naturalized Epistemology Amirah Albahri Follow this and additional

More information

An Inferentialist Conception of the A Priori. Ralph Wedgwood

An Inferentialist Conception of the A Priori. Ralph Wedgwood An Inferentialist Conception of the A Priori Ralph Wedgwood When philosophers explain the distinction between the a priori and the a posteriori, they usually characterize the a priori negatively, as involving

More information

Class 4 - The Myth of the Given

Class 4 - The Myth of the Given 2 3 Philosophy 2 3 : Intuitions and Philosophy Fall 2011 Hamilton College Russell Marcus Class 4 - The Myth of the Given I. Atomism and Analysis In our last class, on logical empiricism, we saw that Wittgenstein

More information

Philosophy Epistemology Topic 5 The Justification of Induction 1. Hume s Skeptical Challenge to Induction

Philosophy Epistemology Topic 5 The Justification of Induction 1. Hume s Skeptical Challenge to Induction Philosophy 5340 - Epistemology Topic 5 The Justification of Induction 1. Hume s Skeptical Challenge to Induction In the section entitled Sceptical Doubts Concerning the Operations of the Understanding

More information

Since Michael so neatly summarized his objections in the form of three questions, all I need to do now is to answer these questions.

Since Michael so neatly summarized his objections in the form of three questions, all I need to do now is to answer these questions. Replies to Michael Kremer Since Michael so neatly summarized his objections in the form of three questions, all I need to do now is to answer these questions. First, is existence really not essential by

More information

Has Nagel uncovered a form of idealism?

Has Nagel uncovered a form of idealism? Has Nagel uncovered a form of idealism? Author: Terence Rajivan Edward, University of Manchester. Abstract. In the sixth chapter of The View from Nowhere, Thomas Nagel attempts to identify a form of idealism.

More information

GS SCORE ETHICS - A - Z. Notes

GS SCORE ETHICS - A - Z.   Notes ETHICS - A - Z Absolutism Act-utilitarianism Agent-centred consideration Agent-neutral considerations : This is the view, with regard to a moral principle or claim, that it holds everywhere and is never

More information

Summary of Kant s Groundwork of the Metaphysics of Morals

Summary of Kant s Groundwork of the Metaphysics of Morals Summary of Kant s Groundwork of the Metaphysics of Morals Version 1.1 Richard Baron 2 October 2016 1 Contents 1 Introduction 3 1.1 Availability and licence............ 3 2 Definitions of key terms 4 3

More information

Interpretation: Keeping in Touch with Reality. Gilead Bar-Elli. 1. In a narrow sense a theory of meaning (for a language) is basically a Tarski-like

Interpretation: Keeping in Touch with Reality. Gilead Bar-Elli. 1. In a narrow sense a theory of meaning (for a language) is basically a Tarski-like Interpretation: Keeping in Touch with Reality Gilead Bar-Elli Davidson upheld the following central theses: 1. In a narrow sense a theory of meaning (for a language) is basically a Tarski-like theory of

More information