Intentionality and Method. John R. Searle

Transcription

1 Intentionality and Method John R. Searle The Journal of Philosophy, Vol. 78, No. 11, Seventy-Eighth Annual Meeting of the American Philosophical Association Eastern Division. (Nov., 1981), pp Stable URL: The Journal of Philosophy is currently published by Journal of Philosophy, Inc.. Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at 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. JSTOR is an independent not-for-profit organization dedicated to and preserving a digital archive of scholarly journals. For more information regarding JSTOR, please contact support@jstor.org. Fri May 18 09:04:

2 720 THE JOURNAL OF PHILOSOPHY INTENTIONALITY AND METHOD' THIS paper is an attempt to draw some of the methodological conclusions implicit in a series of studies I have done of ~ntentionalit~.' Some claims are borrowed from these other studies and some of the arguments occur here in a very compressed form. Furthermore, the method proposed in this paper I, in fact, use in a longer work on this subject.' I believe that the philosophy of language is a branch of the philosophy of mind. According to this view the fundamental linguistic notions, such as reference, meaning, statement, etc., can be analyzed in terms of even more fundamental psychological notions such as belief, intention, and desire. When I wrote Speech ~cts,' I simply took these notions for granted, but any philosopher is bound to feel uncomfortable using such a rich family of unanalyzed concepts as primitives. If meaning is to be analyzed in terms of intentions, then one wants to know what an intention is; if speech acts are a class of actions, then what is an action?; if speaking a language is engaging in rule-governed behavior, then what is rule-governed behavior? More generally, what is the character of the Intentionality, directedness, or aboutness which some psychological states, such as beliefs, desires, and intentions have, but which some other psychological states, such as undirected anxiety or a sudden sense of elation, do not have? Actually, though I did not realize it until early in 1976, the answer to this last question had been staring me in the face all along: there is a theory of Intentionality already implicit in the theory of speech acts. This is hardly surprising since, in general, any speech act with a propositional content is an expression of the Intentional state specified in its sincerity condition: every statement is an expression of a belief, every order is an expression of a desire, every promise is an expression of an intention, and so on for most " To be presented in an APA symposium on Intentionality, December 28, J. N. Mohanty will be co-symposiast, and Robert Aquila will comment; see this JOURNAL, this issue, and 718/9, respectively. '"What Is an Intentional State?", Mind, LXXXVIII,349 (January 1979): 74-92; "Intentionality and the Use of Language", in A. Margalit, ed., Meaning and Use (Boston: Reidel, 1979), pp ; "Literal Meaning", Erkenntnis, XIII,2 (July 1978): ; "The Background of Meaning", in Searle, F. Kiefer, M. Bierwisch, eds., Speech Act Theory and Pragmatics (Boston: Reidel, 1980), pp ; "The Intentionality of Intention and Action," Inquiry, XXII,3 (Summer 1979): Zlntentionality: An Essay in the Philosophy of Mind (New York: Cambridge, forthcoming). New York: Cambridge, W81/7811/0720$01.40 O 1981 The Journal of Philosophy, Inc.

3 INTENTIONALITY 72 1 speech acts with a propositional content. This condition holds whether or not the agent has the psychological state that he expresses; indeed whether or not he has the state determines whether or not his speech act is sincere. If one attempts to follow out the implications of this connection one arrives.at the following points of structural isomorphism between speech acts and intentional states: the distinction between illocutionary force and propositional content has an exact analogue in the structure of intentional states, in the distinction between the type of state and its representative content; the distinction in direction of fit between words and the world has an exact analogue in the different directions of fit between the mental state and the world; the notion of the conditions of satisfaction of a speech act-truth conditions, obedience conditions, and fulfillment conditionswhich applies to any speech act with a direction of fit applies equally to any intentional state with a direction of fit. Indeed, for a very large class of cases, any speech act that has a propositional content, direction of fit, and conditions of satisfaction will be an expression of a psychological state with the same propositional content, the same direction of fit, and the same conditions of satisfaction. The picture that emerges is this: just as any speech act with a direction of fit is a representation of its conditions of satisfaction, so any intentional state with a direction of fit is a representation of its conditions of satisfaction. The notion of representation has a fairly sordid history in philosophy, but if we can strip it of its past for a moment and remember our theory of speech acts we can say the following. The key to understanding intentionality is representation, in a sense of representation we can make clear from our theory of speech acts; and the key to understanding representation in this sense is conditions of satisfaction. Any intentional state with a direction of fit is a representation of its conditions of satisfaction. And intentional states that do not have a direction of fit, such as feeling glad that p, or feeling ashamed that p, though they are not reducible to states with a direction of fit, such as beliefs and desires, nonetheless contain beliefs and desires and these do have directions of fit. Such a picture of intentionality is a natural extension of the theory of speech acts; and, to repeat, not really a very surprising extension. Since, in general, any speech act with a propositional content is an expression of a psychological state, we can begin a study of the structure of the psychological state by studying the structure of the speech act. This picture of intentionality is not reductive. The no-

4 722 THE JOURNAL OF PHILOSOPHY tions we use to explain intentional states, e.g., conditions of satisfaction and direction of fit, are themselves intentionalistic notions, and in that sense we are always moving within the circle of intentionality. Our aim is not to reduce intentionalistic notions to something non-intentionalistic, but to provide a general account of intentional states in terms of a small number of intentionalistic primitives. If the account is sufficiently general and if the amount of data it can accommodate is rich enough, then the analysis may be illuminating, though "circular." The account of the semantic notions, on the other hand, is intended to be reductive: we should be able to analyze such notions as meaning and reference in terms of intentional notions. If the intentionality of language is just a special case of intentionality, then it should be analyzable in terms of and reducible to the more fundamental forms of intentionality. We use our theory of speech acts as a heuristic to enable us to get a handle on intentionality. Once we have it, we can go back and do analysis of the semantic notions using the intentionalistic notions. My account of intentionality has certain implications for philosophical method which I want now to develop. The standard method of philosophical analysis is to analyze concepts, especially by investigating the truth conditions of sentences containing an expression of the concept to be analyzed. Thus, for example, one standard way to analyze the concept of promising is to ask under what conditions it would be true to say of a speaker who has made an utterance that in making the utterance he had made a promise to someone; and to do that is to state truth conditions for sentences of the form In utterance U speaker S made a promise that p to hearer H. But if we are right in thinking that the truth conditions of statements using such sentences are just a special case of the conditions of satisfaction of speech acts and if we are right in thinking that the linguistic notion of conditions of satisfaction is just an extension of the more primitive notion of conditions of satisfaction of intentional states, then a natural extension of the methods of philosophical analysis is to inquire into the conditions of satisfaction of intentional states. For example, instead of just investigating the logical structure of sentences containing the word 'believe', we investigate the logical structure of beliefs; instead of just investigating the truth conditions of descriptions using the words 'intentional' and 'intentionally', we actually investigate the conditions of satisfaction of intentions. The question, for example, under what conditions is the sentence "John intends to go to the movies" satisfied (=true), is not at all the same question as under what conditions is John's intention to go to the movies satisfied (=fulfilled).

5 INTENTIONALITY 723 In, theory one might say that the method of linguistic analysis must encompass the method of Intentional analysis, since anything represented by the states must also be represented by the statements that represent the states, assuming transitivity of representation. But, in practice, the direction of investigation varies depending on which method one selects. The two methods are not in conflict; indeed each is a natural extension of the other. Nonetheless, there are certain advantages to recognizing that the two methods are separate and employing both in philosophical analyses. In what follows I want to show some of the advantages of adding the method of analyzing the conditions of satisfaction of intentional states to the armory of the philosopher with respect to three problems: opacity, causation, and the preconditions of intentionality and meaning. OPACITY One of the mistakes that is endemic to linguistic analysis is the confusion between features of the reports of a phenomenon and features of the phenomenon being reported, and one common instance of this is the constant confusion between intensionality and intentionality. One often hears it said, for example, that beliefs and desires are intensional, or even that they are intensional entities along with propositions. But a moment's reflection reveals that this is simply a confusion. If our criteria for intensionality are failure to satisfy the standard tests for extensionality, such as existential inference and Leibniz's law, then there is nothing inherently intensional about beliefs and desires, or for that matter propositions. Statements that report beliefs and desires are often intensional on both of the above criteria, but from the fact that statements that report beliefs and desires are intensional it simply does not follow that the beliefs and desires reported are intensional. The way to avoid this mistake is to keep clearly in mind the distinction between the conditions of satisfaction of beliefs and desires and the conditions of satisfaction of reports of beliefs and desires. Thus, all hands agree that the statement Reagan lives in the White House. is extensional. But what about Ralph's belief that Reagan lives in the White House: is it extensional or intensional? Notice that the statement Ralph believes that Reagan lives in the White House. is an intensional statement, but if you look at the belief reported by that statement and not at the statement that reports it, you see immediately that the belief is extensional on both criteria. Ralph's belief is true if there is a unique x such that x = Reagan and a unique y such that y = the White House, and anything identical with x

6 724 THE JOURNAL OF PHILOSOPHY lives in anything identical with y. The belief is as fully transparent as the original statement that Reagan lives in the White House, because it has exactly the same truth conditions. Of course the belief that Ralph believes that Reagan lives in the White House is an opaque belief for the same reason that the statement that Ralph believes that Reagan lives in the White House is an opaque statement: it fails both tests for extensionality. Parallel remarks apply to propositions. If we keep in mind the distinction between the truth conditions of a proposition and the truth conditions of propositions about a proposition we can easily see that the proposition that Reagan lives in the White House is fully transparent even though the proposition that the proposition that Reagan lives in the White House is transparent, is itself opaque.4 But what about the claim that beliefs, desires, and propositions are intensional entities or that they are intensions? Notice that if we define intensionality in terms of tests for extensionality (and I have never seen any other coherent attempt to define it), then intensionality isn't defined over entityhood at all. The tests for extensionality divide representations into two categories, those which meet the tests and those which don't; but they do not, in addition, create an ontological category of intensional entities. I think that the view that beliefs and propositions are intensional entities is based on a muddled pun as follows. In the old doctrine that "intension determines extension," 'intension' is presumably used to mean meaning, semantic content, or intentional content generally. If so, the claim that propositions, beliefs, etc., are intensional entities is just the claim that they are identical with intentional contents or at least that they intrinsically have intentional contents. But this sense of 'intension' is quite different from the sense in which intensional representations are distinguished from the extensional ones. Thus, on this latter sense, there is nothing inherently intensional about intensional entities. Some intensional.entities are extensional, e.g., the proposition that Reagan lives in the White House; some are intensional, e.g., the proposition that Ralph believes that Reagan lives in the White House. Another more subtle confusion that arises out of failing to distinguish between the truth conditions of.representations about representations, e.g., statements about beliefs, and the truth conditions of ground-floor representations, e.g., ground-floor beliefs, is the belief that there is a fundamental distinction between de re and de To see this, note that the second proposition can be true even if Reagan never existed, hence exportation is not valid.

7 INTENTIONALITY 725 dicto beliefs. There are indeed two ways of reporting beliefs with two different sets of truth conditions. Thus, if Ralph sees a man with a bottle in his hand and forms the belief that the man is drunk, we can report that belief either in the form Ralph believes that the man with the bottle is drunk. About the man with the bottle, Ralph believes he is drunk. The first of these is a de dicto report, for it reports only the intentional content of the belief; the second is a de re report because it commits the reporter to the existence of an object that the belief is about. The occurrence of 'the man with the bottle' is intensional in the first, extensional in the second. But, though the truth conditions of the reports are different, it does not follow that the beliefs reported are different. Notice that the distinction we the reporters can make from our third-person point of view is not a distinction that Ralph can make from his first-person point of view. In our representation of his representation we can decide how much of the original representation we shall just report and how much we can commit ourselves to; but that is not a distinction that Ralph can make because he is committed to the whole content of the representation, since it is all his belief. To see this, imagine the absurdity of the following conversation. Ralph: I believe that the man with the bottle is drunk. Ortcutt: But about the man with the bottle, do you believe he is drunk? Ralph: No, I never said that. That would be a de re belief and I have merely a de dicto belief. I never said (a) About the man with the bottle I believe he is drunk. But only (b) I believe that the man with the bottle is drunk. The reason that this conversation is ridiculous is that (a) and (b) report exactly the same belief, even though the truth conditions of the two reports are different. If it turns out that there never was a man with a bottle, that the whole thing was a hallucination, then (b)is true (even though the belief reported isn't true), but (a) is not true, since there isn't any man with the bottle about whom the belief is reported. Same belief, different reports of the belief. The belief is identified by its conditions of satisfaction, and these need not be the same as the conditions of satisfaction of the statement reporting the belief even if the believer is doing the reporting. But, it might be objected, in the case as imagined the belief is a perceptual belief, of the sort that would be expressed by a demon-

8 726 THE JOURNAL OF PHILOSOPHY strative, and what warrants the claim that it is irreducibly de re is the fact that such beliefs place the believer in direct causal and perceptual relations to the object the belief is about. In such a case Ralph would normally report his belief with a sentence of the form That man is drunk. and the occurrence of the perceptual demonstrative indicates that the belief relates Ralph directly to the man, and in that sense the belief is irreducibly de re. Once again, I want to maintain that if we shift our attention from the sentence to the intentional content of the mental states and events, we can see that this belief, like any other, is individuated entirely by intentional content and is in that sense de dicto, and that the appearance of an irreducibly de re belief is a grammatical illusion. In this case part of the content of the belief is a perceptual experience. The identification of 'that man' is based on the fact that Ralph is seeing the man, that he has a visual experience of the man. But that visual experience, like any other intentiocal content with a direction of fit, has conditions of satisfaction, and those conditions of satisfaction are part of the conditions of satisfaction of the belief. The relevant portion of the conditions of satisfaction of the visual experience can be represented as follows: Vis exp (there is a man x there and the fact that x is there is causing this Vis exp) where the intentional content is represented by the part in parentheses, and that intentional content simply plugs into the rest of the belief. The relevant portion of the whole belief, then, has an intentional content that can be represented as follows: (There is a man x there and the fact that x is there is causing this Vis exp and x is drunk) In this case, as in any case, to identify a belief one must identify its intentional content. But to identify its intentional content one must do more than just look at the sentence used to express the belief; one must look at the intentional content in the man's head. Before turning to causation, let us step back a moment and ask how the puzzles about opacity and transparency arise in the first place. Statements and beliefs are representations; their conditions of satisfaction are matters of how things are in the world. But statements and beliefs about statements and beliefs are representations of representations, and their conditions of satisfaction depend on how things are in the representations being represented. This

9 INTENTIONALITY 727 has the consequence that the ground-floor representations to be satisfied require the existence of objects satisfying their referential content (hence existential inference is valid), and it doesn't matter to the satisfaction of the representations how the objects are referred to (hence substitutability preserves truth). But the second-order representations, the representations of representations, require rather that the representational content of the first-order representation, and only that content, be preserved in its representation. Thus for the representation of the representation it doesn't matter to its truth whether or not the objects represented by the ground-floor representation really exist (hence existential inference is not valid), but it does matter how the objects are presented, under what "mode of presentation" they are represented (hence substitutability does not preserve tr~th).~ And this is why (pace Frege) the same words with the same meaning can occur in both the ground-floor and secondorder representations and yet the logical properties of the representations are not the same. In this sense one might say that the secondorder representation does not so much represent the content of the first-order, as rather it presents it, since ideally the same representational content as occurred in the first-order state also occurs in its report. The reader is advised to test his intuitions against mine by going step by step through the claims made in this paragraph with examples of representations and representations of representations. For example, compare the conditions of satisfaction of Jones's belief (not the sentence) that the evening star twinkles with Smith's belief that Jones believes that the evening star twinkles. INTENTIONAL CAUSATION Causal analyses of intentional notions have recently become quite common, but there has been very little investigation of the implications of the fact that causation enters into the conditions of satisfaction of many intentional states. Thus it is frequently and correctly remarked that statements of the form x saw y. entail statements of the form y caused a visual experience in x. But that fact is a reflection of a causal element in the conditions of satisfaction of the visual experience itself, and those conditions of satisfaction need exploration. The point is not just that the concept his parallelism between statements and beliefs is disguised by the fact that criteria of extensionality are often stated in syntactical terms, in terms of bound variables and substitutability of expressions, but the underlying ideas are still semantic.

10 728 THE JOURNAL OF PHILOSOPHY of causation enters into the description of the visual experience, but that the experience of causation enters into the visual experience itself. How? For a start, consider the example above of Ralph seeing the man with the bottle. What are the conditions of satisfaction of his visual experience? They are, first, that there must be a man in front of him and the man must have such and such features, e.g., holding a bottle in his hand; but, secondly, these conditions of satisfaction must cause the very visual experience whose conditions of satisfaction they are. Thus, the relevant portion of the conditions of satisfaction can be represented as follows: Vis exp (there is a man with F features over there and that there is a man with F features over there is causing this Vis exp) The visual experience is self-referential; its conditions of satisfaction make reference to the very visual experience whose conditions of satisfaction they are, and the form of the self-referentiality is causal. The causation is part of the content, though not the object, of the visual experience. When one sees anything, the visual experience occurs as caused by the thing one sees, even though, of course, one never sees the visual experience. The fact that causation is always part of the content and thus part of the conditions of satisfaction of a visual experience does not imply that the causal condition is in fact always satisfied. It is always possible, as traditional epistemologists are eager to point out, that Ralph could be having a hallucination, he could be having that visual experience even if it wasn't caused or wasn't caused "in the right way" by the presence and features of a man in front of him. But this is irrelevant to the claim that the causal condition is part of the content of the experience. Even if Ralph's experience is a hallucination, the content of the experience determines what must be the case if it is not to be an hallucination, and to say that is to say that the conditions of satisfaction, including the causal condition, are internal to the experience. Exactly parallel remarks apply to intentional actions. Suppose I perform a simple intentional action such as raising my arm. Keeping in mind the distinction between prior intentions (those intentions which occur prior to actions) and intentions in action (those intentions which are component parts of action), we can represent the conditions of satisfaction of the intention in action in a way that is closely parallel to the conditions of satisfaction of the visual experience. My intention in action will be satisfied if and only if my arm goes up and my intention in action causes it to go up.

11 INTENTIONALITY 729 Thus: Intention in action (my arm goes up and that my arm goes up is caused by this intention in action) Intentions in action, like perceptual experiences, are causally selfreferential. Their conditions of satisfaction make reference to the very intention of which they are the conditions of satisfaction. And the differences in the formal structure of the conditions of satisfaction of intentions in action and perceptual experience are as revealing as the similarities: perceptual experiences have the mind-toworld direction of fit and the world-to-mind direction of causation (roughly, that means that they are satisfied only if the world is as it perceptually seems to be and if its being that way causes its perceptually seeming that way), whereas intentions in action are exactly opposite in both direction of fit and direction of causation. They have world-to-mind direction of fit and mind-to-world direction of causation (that means that they are satisfied only if the world comes to be the way one tries to make it be and if its coming to be that way is caused by one's trying to make it that way). Notice further that it is a feature of intentional causation that either the cause is a representation of the effect or the effect is a representation of the cause, and in that sense there is a "logical" or "internal" relation between cause and effect. Since my aim here is not so much to carry out intentionalistic analyses as to argue for the methodological advantages of doing so, I want to point out some of the results of analyses of states that have causally self-referential conditions of satisfaction which are not readily obtainable from doing traditional-style analyses of the truth conditions of sentences containing the word 'cause'. I will mention just two. (1) It is often claimed by adherents of a Humean analysis of causation that one never directly experiences causal connections. One can experience events that are, in fact, causally related; but, in addition to the experience of regular occurrence of sequences of events, one has no experience of causation. Causation must h la Hume be inferred from regularity (the "constant conjunction of resembling instances"). On the intentionalistic analysis sketched above, not only is it false that we never experience causation, but in fact we experience it pretty much all of the time, viz., whenever we perceive or act. The Humeans were looking in the wrong place. They sought causation as the object of our experiences, whereas in its most primitive form it is part of the content of experiences of perceiving and acting. When I look at a tree I don't see causation, I

12 730 THE JOURNAL OF PHILOSOPHY just see a tree, and when I raise my arm I don't raise causation, I just raise my arm. But in both cases the experience of causation is part of the content of the experience; it is part of the conditions of satisfaction that are internal to the experience. (2) On some versions of the Humean view genuine causes and effects must always be "logically independent." Indeed, some years ago a Wittgensteinian version of this principle took the form of denying that reasons could be causes, i.e., denying that desires, intentions, etc., could cause actions, on the ground that the cause and effect were not logically independent in the required way. The intention to perform an action a, so the claim went, could not be a cause of doing a, since the specification of the alleged cause had to make mention of the alleged effect. This thesis was countered by pointing out that the extensionality of sentences of the form "x caused y" permits substitution of other descriptions of the cause and effect under which the logical independence of cause and effect is preserved. Thus, one could have both a Humean account of causation and a belief that reasons can be causes. According to the view of intentional causation I am here advocating, both sides in this dispute were muddled. It isn't in spite of the fact that there is a logical connection between cause and effect that intentions can cause actions; rather it is precisely because the relation is "logical" in this sense that the causation works the way it does; since the form of causation in question is intentional causation. In the causally self-referential forms of intentionality, the conditions of satisfaction of the intentional state require either that it cause (in the cases of prior intention and intention in action), or be caused by (in the cases of memory and perceptual experience), the rest of its conditions of satisfaction. The way, for example, that a prior intention works causally is that it is part of the conditions of satisfaction of the intention that it must cause the action represented by the rest of the conditions of satisfaction. It is, e.g., part of the intentional content of my prior intention to raise my arm that if that intention is to be satisfied it must cause the action of my raising my arm. NETWORK AND BACKGROUND The attempt to spell out the conditions of satisfaction of intentional states helps to dispel a certain naive conception of intentional states according to which they neatly and atomistically individuate and have clear conditions of satisfaction. It is less easy to combat this form of atomism as long as one confines oneself to sentences, for sentences do fairly neatly individuate and can be made to seem to have clear truth conditions.

13 INTENTIONALITY 731 Consider, for example, what must be the case in order that I could form the intention to go to a philosophy convention. As long as one confines one's investigation to sentences of the form X intends to do A. one is likely to concentrate on entailment relations, and one will come up with certain interesting results, e.g., 'X intends to do A' entails 'X believes it is possible to do A' and 'X wants in some sense to do A'. But the effort to spell out the conditions of satisfaction will reveal that I must have other intentional states in order to have had my quite specific intention. One finds that any such intention is embedded in a network of intentional states and that it can have the conditions of satisfaction that it has only in virtue of its position in the network. In order that I should have specifically that intention, with the content that it has, I must believe and desire a whole lot of other things which go far beyond the logical consequences of the corresponding statement. I have, for example, a whole series of beliefs about airlines, and geographical locations, and hotels, and professional associations. And these beliefs, in turn, are connected to a series of desires about tickets, reservations, and travel arrangements. The content of my original intention cannot stand on its own, but functions only in relation to this holistic network of other intentional states. In general, one can say that any complex intentional state has its conditions of satisfaction only in relation to other intentional states, but, if that is so, then such intentional states cannot be independently individuated. Each stands in internal relations to other states, and to say they have "internal relations" simply means that they could not have the contents they have and thus could not be the states they are, except in relation to the rest of the network. Furthermore-and this is an independent and more controversial point-the holistic network of representational states shades off into a background of mental capacities that are not and could not be representations. I do indeed have sets of beliefs and desires about conventions, traveling, and philosophical discussion, but these function only against a background of nonrepresentational capacities. My ability to cope with travel and meetings, like my ability to interpret utterances, does not consist entirely in beliefs, desires, and other representations. Though I know of no knockdown arguments to establish this point, there are various ways of coming to see it. If you actually try to follow out the threads of the network, on the assumption that the network is self-sufficient, you eventually reach a set of proposi-

14 732 THE JOURNAL OF PHILOSOPHY tions that look fishy, not to say preposterous, if you try to treat them as beliefs. I do indeed believe without ever thinking about it that Washington, D.C., is east of San Francisco, but I don't in that way believe that philosophy conventions occur near the surface of the earth, or that tables and chairs are generally solid. Such things could, of course, be believed, but in the actual functioning of my intentionality, including my intentional behavior, such facts figure in the various stances that I adopt and the various capacities I exercise in coping with the world. I do not represent such facts in my beliefs; I don't need to. They show themselves, rather, in my having the intentional states that I do and in their having the conditions of satisfaction that they have. Another way to see the same point is to note that intentional contents are not self-interpreting. One has to know how to apply them, but the "know-how" in question cannot consist entirely in further intentional contents, without an infinite regress. Each further intentional content is as much in need of "interpretation" or an "application" as the original. Suppose, for example, I now intend to drink a glass of water. That very token intention will not be satisfied if I "drink" the water through my ear, if the "water" is in gaseous or frozen form, or if the "glass" is a ten-thousand-gallon glass. But none of these conditions are represented in the original intention; it contains no representation of any of these requirements. Indeed, in different circumstances we could imagine a token of the same intentional type, the intention to drink a glass of water, where it couldn't be satisfied unless the water came in through the ear or was frozen solid or was a ten-thousand-gallon swig. And we cannot evade the nonrepresentational character of these requirements by assuming that they are all in the network, for as such they would be subject to further analogous requirements. My aim in this paper has been to urge that the methods of analyzing the conditions of satisfaction of intentional states should be added to the traditional linguistic methods of philosophical analysis. Truth conditions of sentences are, after all, just one form of conditions of satisfaction of intentionality. It used to be widely believed, and perhaps in some quarters still is, that there is a fundamental difference between the methods of something called "phenomenology" and something called "analytic philosophy." But if by phenomenology we mean not the attempt to give introspective characterizations of the various forms of tiresonieness and hassle involved in hanging around the Lebenswelt, but rather the examination of the conditions of satisfaction of intentional states, then we could say that phenomenology is a branch of analytic philoso-

15 APA INVITED PAPERS 733 phy: it is that branch of conceptual analysis which is concerned with analyzing the truth conditions and other sorts of conditions of satisfaction of intentional states. Or, alternatively, we could say that analytic philosophy is a branch of phenomenology; it is that branch of the investigation of conditions of satisfaction which is concerned with words and sentences. University of California, Berkeley JOHN R. SEARLE AMERICAN PHILOSOPHICAL ASSOCIATION EASTERN DIVISION Abstracts of Invited Papers to Be Read at the Seventy-eighth Annual Meeting TWO PARADOXES OF ANALYSISf In this paper I distinguish two paradoxes of analysis, and propose solutions to each. The paradoxes can be illustrated as follows: (1) Assume that (a) 'An analysis of the concept "brother" is that to be a brother is to be a male sibling' expresses a true proposition. Then it seems that (b) 'An analysis of the concept "brother" is that to be a brother is to be a brother' must also express a true proposition, and in fact must express the same proposition as (a). Yet (b) expresses a proposition that clearly seems neither true nor the same proposition as that expressed by (a). (2) Assume that an analysis of the concept "knowledge" is that knowledge is justified true belief not essentially grounded in any falsehood. Then it seems that whoever knows that knowledge is knowledge thereby knows that knowledge is justified true belief not essentially grounded in any falsehood. Yet this consequence seems clearly false. (2) illustrates the sort of problem that many philosophers nowadays call "the paradox of analysis." But something along the lines of (1) can also be gleaned from classic writings on analysis. Both (1) and (2) rest on the assumption that analysis is purely a relation between concepts and that, moreover, in a true analysis the analysandum and the analysans are the same concept. *Abstract of an APA invited paper, to be presented December 29, 1981; Carl J. Posey will comment. I am grateful to Alvin Plantinga, Ernest Sosa, and James Van Cleve for discussions of issues bearing on the material in this paper.