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.

Institutional Facts and Brute Values Author(s): A. C. Genova Source: Ethics, Vol. 81, No. 1 (Oct., 1970), pp. 36-54 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/2380015 Accessed: 31/08/2010 20:00 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. 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 http://www.jstor.org/action/showpublisher?publishercode=ucpress. 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 a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to Ethics. http://www.jstor.org

Institutional Facts and Brute Values A. C Genova Wichita State University Considering the substantial philosophical controversy which followed John Searle's well-known publication on how to derive an "ought" from an "'is,"1 it would only be with some hesitation that one would venture to resuscitate the issues and again challenge the argument. The history of the controversy shows that, for the most part, the philosophical camp eventually divided into well-entrenched supporters and dissenters. But after the dust had settled and the confrontation had begun to look like a philosophical deadlock, Searle's recent reformulation of the problem2 has again made it fair game. Searle's latest version accomplishes at least two things: (1) it clarifies the original argument by placing it in the context of a general theory of speech acts and thereby eliminates many objections which were really irrelevancies or due to misunderstanding; and (2) it invites and justifies new challenges to the argument by those who are prepared to accept Searle's hypothesis that speaking a language is an institution of rule-governed intentional behavior grounded on constitutive rules, that a theory of language is part of a theory of action, and that the speech act is the fundamental unit of analysis in any adequate philosophical analysis of language. In short, now that the contextual framework of the argument is clearly delineated, the controversy moves to a new level. In this paper, in the context of the theory of speech acts, I want to (1) examine Searle's claim that the "naturalistic-fallacy fallacy" is the source of a mistaken belief that evaluative statements cannot be logically deduced from descriptive statements alone; (2) criticize Searle's interpretation of his "is" -4 "ought" argument by showing that, just as Searle distinguishes between "brute facts" and "institutional facts." he must also 1. J. R. Searle, "How to Derive an 'Ought' from an 'Is,' " Philosophical Review (January 1964). I will refer to this as Searle's "is" -4 "ought" argument in the remainder of this paper. 2. J. R. Searle, Speech Acts (Cambridge: Cambridge University Press, 1969), chap. 8. All references cited in parentheses in my discussion refer to this work. 36

37 Institutional Facts and Brute Values admit a distinction between what I will call "brute values" and "institutional values"; and (3) respond to three likely misunderstandings of my argument by clarifying the significance of my position with respect to the fact-value distinction, the analytic-synthetic distinction, and the descriptive-evaluative distinction. In sum, I will argue that, although Searle does indeed derive an "ought" from an "is," he does not thereby do what he thinks he does, namely, show that those whom he calls the "classical linguistic moral philosophers" could not handle institutional facts and that they were wrong in maintaining the logical impossibility of validly deducing evaluative statements from descriptive ones. What Searle does do (and can do because of his theory of speech acts and his notion of an institutional fact) is to present a list of English sentences as premises which express various kinds of facts or tautologies, and validly deduce a conclusion which contains the English modal auxiliary "ought" but not the English copula "is." So my objections to Searle will not primarily concern the validity of his particular deduction as such but will challenge what he construes to be its significance, for I will contend that Searle's account does not refute the counter-thesis or expose the naturalistic fallacy as itself a fallacy-at least not in any sense in which those who endorsed these views could seriously have maintained them. To put the point somewhat crudely for the moment, Searle's deduction amounts to deducing an institutional fact from institutional facts or, alternatively, deducing an institutional value from premises containing institutional values (or even, if you like, deducing an institutional value from institutional facts); but what he does not do (and cannot do) is validly to deduce a brute value from any set of premises expressing facts alone (regardless of whether they be brute or institutional facts). Finally, I shall argue that Searle's own analysis presupposes the distinction between brute values and institutional values, and that it is precisely the deduction of brute values from facts that the classical moralists claimed could not be validly performed. I Those whom Searle characterizes as the "classical linguistic moral philosophers" fare badly on his analysis. Searle is explicitly referring to the successors of Moore (covering the period from the end of the Second World War to the early sixties), although it seems that his general attack would also apply to Moore himself and to many precursors of Moore throughout the history of philosophy. At any rate, we are told that the fatal weakness of the classical position lies in its failure to base particular linguistic analyses on a coherent, general approach to language (pp. 131 ff.). According to Searle, this flaw in the so-called classical position is exemplified in the various fallacies to which the position is vulnerablefallacies like the "speech act fallacy" (the confusion of the primary use of a word in the performance of a speech act with the meaning of the word), the "assertion fallacy" (the confusion of the conditions of a nondefective

38 Ethics speech act of assertion with the analysis or meaning of what is asserted), and the "naturalistic-fallacy fallacy" (the confusion between the differences in illocutionary forces of expressions with the truth-relations of the propositions expressed in the corresponding speech acts) (ch. 3). Since part of the task of this paper will be to argue that the so-called naturalisticfallacy fallacy is not a fallacy, we will restrict our attention to Searle's treatment of this issue. What is the origin of the "naturalistic-fallacy fallacy"? Searle argues that one of its props is the speech act fallacy. Thus, if one supposes that the meaning of a supposedly evaluative expression like "valid" or "Extra Fancy Grade" ties it to a particular range of speech acts (such as grading, evaluating, commending, etc.), then, since "entailment is a matter of meaning," it will seem impossible that other expressions standing for purely descriptive or logical characteristics which do not involve these evaluative illocutionary acts could ever be used to define "valid" or "Extra Fancy Grade," or that statements containing only expressions of the latter kind could entail a statement that an argument is valid or an apple is Extra Fancy Grade. In other words, because Searle's classical linguistic moral philosophers held a mistaken speech act analysis of certain words like "valid" or "Extra Fancy Grade," they assumed that certain logical relations involving these words could not obtain (p. 140). More generally, all the fallacies are due to the misapplication of the slogan, "Meaning is use." Linguistic philosophers, recognizing the different uses (different illocutionary forces) of indicative sentences, assumed that because the use of descriptive sentences was different from that of evaluative sentences, it followed that the meanings excluded the possibility of descriptive statements entailing evaluative ones (p. 148). How then is it possible that a statement having the illocutionary force of description can entail a statement having the illocutionary force of evaluation without violating the supposed principle that the conclusion of a valid deduction cannot "contain" more than is in the premises or is warranted by appropriate rules of inference? For Searle, the answer lies in the distinction between meaning and use, that is, respectively, between the truth conditions holding between propositions, on the one hand, and the point or function of the speech acts which express these propositions, on the other. As Searle says, "A statement P made in the utterance of a sentence S could entail a statement Q made in the utterance of a sentence T. even though the utterance of S characteristically had one illocutionary force and the utterance of T had another illocutionary force" (p. 155). Thus, according to Searle, entailment concerns the meaning of propositions and the truth relations between them; while illocutionary force concerns the purpose or function of speech acts. For example, if the meaning of "Extra Fancy Grade" is technically or institutionally defined as "having properties A, B, and C," then the statement that X has properties A, B, and C entails that X is Extra Fancy Grade, even though the former statement

39 Institutional Facts and Brute Values is typically used to describe and the latter to evaluate-urmson notwithstanding.3 Of course, at this point one could raise the standard objection that this "entailment" only works if we presuppose as an additional premise the synthetic statement, "All things having properties A, B, and C are those things which are Extra Fancy Grade," and therefore the evaluative expression "Extra Fancy Grade" is already presupposed, and the statement that X has properties A, B, and C is not sufficient by itself to entail that X is Extra Fancy Grade. Philosophers like Hare and Urmson would maintain that evaluative terms cannot be defined by means of the descriptive characteristics (A, B, C) which serve as the standards relative to a certain class of things by which we ascribe value to a member of the class. That is, a statement like "All things having properties A, B, and C are things which are Extra Fancy Grade" is a synthetic premise which connects the characteristics A, B, and C to being an Extra Fancy Grade X. But Searle maintains that such a statement is analytic-a definitional tautology-because its origin and context are typically such that it is introduced as a stipulative definition by some kind of institutional agency like the Ministry of Agriculture. So if the "additional premise" is indeed an analytic truth, then -the argument's validity is unaffected by its deletion; and this may then lead to the question whether such analytic truths are to be interpreted as some kind of required suppressed "premise" which nevertheless need not be stated for the preservation of validity, or whether they should be viewed as some kind of semantical rules of inference for valid arguments. But as we shall see, my concern will not turn on this latter issue. My interest is not to challenge the validity of this particular kind of argument in its context but to emphasize the fact that if the "premise" at issue is an analytic statement or definition, then it is a constitutive rule which grounds an institutional fact and that this consideration affects the significance of the deduction, not its validity. As a matter of fact, insofar as anyone might ground his interpretation of the naturalistic fallacy on what amounts to a confusion between meaning and use, and thereby has already fallen into the speech act fallacy, I would agree with Searle that the "fallacy" is itself a fallacy. Now, it seems to me that the bulk of Searle's argument rests on two basic claims. The first is that entailment is a relationship between propositions, not between illocutionary forces (nor even, strictly speaking, between statements or assertions, because these are illocutionary acts, not propositions). This means that when we are concerned with determining the validity of an argument, we are to examine the truth relations between the propositions involved because entailment is a matter of meaning (not use) and it is propositions which have meaning. The second important 3. Ibid., pp. 132-36. Searle's quarrel with Urmson centers on J. O. Urmson's "Some Questions concerning Validity," reprinted in Essays in Conceptual Analysis, ed. A. G. N. Flew (London, 1956) and "On Grading," Logic and Language, ed. A. G. N. Flew, 2d ser. (New York, 1953). Searle considers both of these papers as examples of the "naturalistic-fallacy fallacy."

40 Ethics claim concerns Searle's familiar distinction between a brute fact and an institutional fact. For it is by means of Searle's appeal to constitutive linguistic rules such as "To be Extra Fancy Grade is to have properties A, B, and C"-what he sometimes calls "meaning rules"-that he is able to move from the level of brute fact to institutional fact. And once he reaches this level, the rest of the way (the deriving of an "ought" from an "is") is clear sailing, dependent upon definitional connections and tautologies. The existence of institutional facts points to activities involving rulegoverned behavior in accordance with constitutive rules. Speech acts are institutional facts because speaking a language is constitutive, rule-governed, intentional behavior; and the making of a promise is an institutional fact which implies that the promiser is participating in the institution of promise making, just as exchanging a bishop for a knight implies that one is participating in the game of chess. Institutional facts are irreducible to brute facts because what will be missing are precisely those concepts formulated by constitutive rules which tell us what the fact is-the constitutive rules corresponding to the true statements one can make about the practices involved in the institutional activity as that activity. Thus, it is only given the institution (system of constitutive rules) of baseball that one can speak of home runs; only given the institution of chess that one can checkmate his opponent; only given the institution of promise making that one can promise. So institutions are systems of constitutive rules which constitute the bases for activities, the existence of which is logically dependent upon the underlying rules. Finally, constitutive rules usually take the form of "X counts as Y in context C" and are analytic statements, definitions, or tautologies. Thus, as Searle aptly says, to ask "How can making a promise create an obligation?" is like asking "How can scoring a touchdown create six points?" (p. 35). Both questions are answered by citing the relevant rules. Now it seems generally clear that institutions like games or promise keeping have a certain "upshot." I mean that the constitutive rules having the form "X counts for Y in context C" are such that the variable Y is typically satisfied by things like home runs, checkmates, promises, or obligations, that is, by institutional values. For example, in a perfectly straightforward sense, certain moves in chess are good, others bad; and surely the ultimate move resulting in checkmate is an object of desire, commendable or whatever. The same with baseball. Home runs are better than singles, and the cheers of the fans are sufficient evidence that they value the fact that their team has won. I say that these are institutional values because they, just like institutional facts, would never exist if constitutive rules did not define, among other things, just what is to count as a value and what is not-so they are institutionalized values. The same holds for institutional speech acts like promising. A constitutive rule tells us that to utter the words "I promise" under certain conditions just means that the utterer is obligated and a fortiori ought to keep his promise.

41 Institutional Facts and Brute Values Hence, the constitutive rules which ground the institution of promise making, from which particular, datable institutional events of promising are derivative, are rules which generate institutional values in such a way that the expression "John ought to pay Smith five dollars" must necessarily express an evaluative force. But now I ask: If this is so, can we then make a distinction between institutional values and brute values just as we admitted the distinction between institutional facts and brute facts? I shall argue that we can and must make this distinction, that Searle himself indirectly recognizes this distinction, and that it is this distinction which undercuts Searle's claim that he has refuted the general thesis that evaluative statements cannot be validly deduced from descriptive ones. II In his "is" -> "ought" deduction (pp. 177-82), Searle interprets the thesis to be refuted as the thesis that "descriptive statements cannot entail evaluative statements." A counter-example is provided by taking statements which a proponent of the thesis would regard as purely descriptive and showing how they can logically entail a statement which is admittedly an evaluative one-in this case, an "ought" statement. Moreover, he emphasizes that this is a problem in the philosophy of language concerning modal auxiliaries and illocutionary forces (not political or moral issues) (p. 176). Thus, we might rephrase the issue by asking, "Can a sentence containing the English modal auxiliary 'ought' be logically derived from sentences each of which in effect contains only the English copula 'is'>" Searle's counter-example succeeds by appealing to a constitutive rule of linguistic usage which justifies an assertion of an institutional fact which in turn generates an institutional value. In other words, the argument proceeds by invoking an institution in such a way that a conclusion concerning what one ought to do is definitionally entailed. On the level of what we might call "institutionalized reality," Searle's analysis amounts to collapsing the dichotomy between "ought" and "is" or, as it were, superimposing both facts and values in a system of constitutive rules. Searle's deduction simply reflects the fact that once we understand the use of the word "promise," we then have the right-a kind of institutionalized inference ticket-to go on and say things about obligation and what one ought to do. The institutional fact of promise making is an instance of just that institution which is grounded on constitutive rules which define, create, and generate a set of institutional values. Searle thinks that his counter-example clearly shows that the traditional thesis that value statements cannot be validly deduced from factual statements is false, that the classical view cannot accomodate institutional facts, and that his opponents "cannot accomodate both the fact that promises obligate and the fact that it is a matter of fact that someone has made a promise" (p. 194). Also, for Searle, it follows that sentences like "One ought to keep one's promises," "One ought not to steal," "One ought not to tell lies," and

42 Ethics "One ought to pay one's debts" are members of a class of tautologies (pp. 185-86). I find this peculiar because among other reasons, if "One ought to keep one's promises" is a tautology, then when it is taken as a conclusion in relation to any set of premises whatever, we would have a valid argument because it would be impossible for the premises to be true and the conclusion false. Therefore, to produce a valid argument in which "ought" occurs in the conclusion while only "is" occurs in each premise would be a trivial accomplishment, to say the least. Now I take it to be agreed on all sides that the latter argument, offered as a counter-example to the nonderivability thesis, would be considered ludicrous because the proponents of the thesis (even if they granted that such statements were tautologies, which is highly doubtful) would require that the premises be relevant to the conclusion and that, in a much stronger sense than is suggested in our trivial counter-example, the premises be sufficient conditions which logically entail the conclusion as logically following from the assumed premises. Of course, the conclusion to Searle's deduction, namely, "John ought to pay Smith five dollars," is not itself a tautology, and, moreover, it is not a general "ought" conclusion but only a singular one. Nevertheless, what if a reasonable case can be made for the claim that, although Searle's argument is unlike our pseudo-counter-example because it is not trivial or absurd, yet it is analogous to our pseudo-counter-example because, when one logically derives an institutional value through definitional connections and tautologies from an institutional fact, one in effect has produced an irrelevant deduction of an "ought" from an "is" which does not logically deduce what has to be deduced if one is to have a genuine counter-example to the correct interpretation of the nonderivability thesis -although it does have the property of validity and truly shows us how institutional facts work? If this were the case, then Searle's argument against the naturalistic-fallacy thesis would be itself a fallacy. Perhaps we can christen it the "institutional fallacy"-the confusion of the conditions for logically deriving institutional values from institutional facts with the conditions for logically deriving what I call brute values from any kinds of facts at all. So far as I can understand Searle's analysis of the "is" -> "ought" issue, all he really accomplishes is to convince the sympathetic reader that language is a complex institutional fact, and, as with other institutional systems, there are necessary constitutive rules which define what is to count as what, for example, what is to constitute a home run in baseball, a checkmate in chess, a political victory in an election, and, indeed, an obligation or institutional value in the linguistic activity of promise making. Thus, while admitting that no institutional fact can be reduced to or derived from brute facts alone, Searle revises the naturalistic-fallacy claim: "No set of descriptive statements can entail an evaluative conclusion without the addition of at least one evaluative premise" by substituting: "No set of brute fact statements can entail an institutional fact statement without

43 Institutional Facts and Brute Values the addition of at least one constitutive rule" (p. 185). But once the institutional-fact level is established, there is no problem at all in deriving an "ought" because the constitutive rules of English usage guarantee that the remaining connections are definitional or tautological. Now I ask: What does it take to allow the introduction of these crucial constitutive rules and the analytic truths (the reportings of which are purely factual because they merely report linguistic facts about the meaning of words)? I say that in an obvious sense it takes at least a willingness on the part of the auditor to speak English, and for the great majority of the speakers of the language, it usually implies an endorsement of the whole institution of promise making as an appropriate, worthwhile institution. Granting all this, then the conclusion with the word "ought" necessarily follows; and since the relevant constitutive rule in effect defines the word "promise" partly in terms of the word "ought," it makes little difference whether we say that the conclusion consists of a derivative institutional fact or institutional value because, as Searle noted, a word like "promise" is both evaluative and denotes a descriptive fact. So if one agrees (what else can one do?) to use the word "promise" in accordance with established usage, then one is thereby committed to the logical consequences, which include the evaluative notion of obligation, just as when one conforms to the usage of "triangle," one is thereby- cunkitted to the fact that triangles have three sides (p. 194). Finally, the mere fact that the commitment in the first case entails the evaluative notion of obligation (and therefore allows one to construct the counter-example to the classical thesis), while in the second case it does not, is for Searle totally irrelevant to the question of any kind of higher moral decision or evaluation. The two cases are exactly parallel in all relevant respects. The fact that you may not like promise making does not relieve you of your commitment to the meaning of the word any more than the fact that you despise geometry can change the truth that triangles have their three sides. Now what is it that makes some of us hesitate to let it go at that? Is it because, as Searle sometimes suggests, we are so dogmatically committed to the "metaphysical" dichotomy between fact and value, description and evaluation, "ought" and "is," that we make our very criterion for identifying an evaluative statement just the requirement that it be a statement that cannot validly be deduced from descriptive ones-thus making our whole thesis trivial? I think not. Rather, I think that we feel ambivalent when confronted with Searle's counter-examples because, on the one hand, we recognize the advantages of the speech act theory of language and are inclined to admit the validity of Searle's deduction when we understand that institutional facts grounded on constitutive rules are just those special kinds of facts which allow us to specify values that are logically dependent upon the existence of the underlying constitutive rules; but, on the other hand, we also recognize that there is a sense of "ought" (and certain other value terms) which, although it may well be determined by

44 Ethics constitutive rules of English usage as with any other words, does not presuppose the institutionalized facts (like promise making, debt paying, truth telling, or playing baseball, for that matter) to which the word is applied. This is a brute sense of "ought" which simply indicates whether we value, commend, recommend, endorse (or whatever) the evaluated institutional facts without presupposing that they are themselves good, bad, or indifferent. This does not mean that there are literally two meanings of "value" any more than that there are two meanings of "fact." There are, indeed, different uses of "value" and "ought," just as there are different uses of "fact" and "is," but like Searle (in his admonitions against the speech act fallacy and the "meaning is use" cliche), I contend that the meaning of these terms is not sufficiently determined by their use. The dictionary will tell us that "is" is the third person, singular, present indicative of "to be," which means to exist, have reality, take place or occur, etc.; and that "ought" is a modal auxiliary indicating the state of being bound in duty or being obligated in respect to some action or state of affairs, etc. As for "fact" and "value," let us simply say that a fact is an existing state of affairs-what is the case; and a value is something like a commitment to what ought to be the case. Then we distinguish brute facts from facts which are created by the process of institutionalization, and similarly, we distinguish brute values from institutionalized values; and the two distinctions are connected by the fact that institutionalized facts typically create institutionalized values. This latter point is just another way of saying that the concept of institutional fact, when unpacked, already logically contains the concept of institutional value. Searle is admitting this when he asserts, "What I am here arguing is that in the case of certain institutional facts, the evaluations involving obligations, commitments, and responsibilities are no longer left completely open because the statement of the institutional facts involves these notions" (p. 190). It should be obvious to the reader at this point that my criticism of Searle is neither to question the validity of his counter-example argument nor to make the inane objection that his argument is valid only because there are evaluative premises or terms lurking in the logical woodpile. Rather, my interest is to show that Searle's own analysis indirectly implies the fundamental distinction between brute and institutional values, that the naturalistic-fallacy thesis concerns the impossibility of logically deducing brute values from any kind of facts (or, derivatively, of deducing institutional values from brute facts), that Searle has not shown that such a deduction is possible (indeed, he grants the derivative claim), and that the so-called classical or traditional view can accomodate the existence of institutional facts-facts which have been recognized by philosophers from Plato to the present. Searle, although he apparently does not realize its significance, recognizes the distinction between what I call brute values and institutional values when he says that committing oneself to accept the institution of promising can mean either (a) undertaking to use the word "promise"

45 Institutional Facts and Brute Values in accordance with the literal meaning as it is determined by the internal constitutive rules of the institution, or (b) endorsing or valuing the institution itself as a good or acceptable institution from an external vantage point (p. 195). He grants that his whole argument assumes interpretation a. As for b, he admits the extravagant possibility that a "nihilistic anarchist" may well reject the whole institution of promise making, and, moreover, that even though "One ought to keep one's promises" is a tautology, it is "perfectly consistent" with his account for the nihilist to say (without contradiction I assume), "One ought never to keep a promise" (p. 188). I say "extravagant" possibility because it is much more typical of us to endorse the institution of promising as a good, worthwhile institution. But why? Well, there are many possible answers. One can justify his endorsement on utilitarian grounds, on pragmatic grounds, on hedonistic grounds, or perhaps even by appealing to some system of transcendental eternal values. The point is that for purposes of the present discussion it does not matter because what does matter is that all such endorsements are "secondorder" evaluations of institutions, subsumable under interpretation b. That is, they cannot themselves be grounded on the systems of constitutive rules and institutions which they are judging simply because these latter constitute the data which are being evaluatively judged. In short, if one judges that the institution of promising ought to exist, then this use of "ought" (although no doubt determined by English usage) is not logically determined by the use of "ought" which logically derives from the particular institution of promising-the use we find in "One ought to keep one's promises" where the latter is viewed as a tautology. I say that interpretation b corresponds to what the classical moralists meant by values as opposed to facts-brute values which differed from facts not by virtue of their "subjectivity" but by virtue of the different criteria which were required to support value statements as opposed to factual ones. Factual statements are supported by empirical evidence; logical or mathematical statements are supported by appeal to postulates, definitions, rules of inference, and the like; but brute value statements are usually supported by ethical theories-theories like those we find in Plato's Republic, Aristotle's Ethics, Hume's Treatise, Kant's second Critique, Mill's Utilitarianism, Moore's Principia Ethica, Hare's Language of Morals, Ayer's Language, Truth and Logic, and so on. And these theories are products of creative intelligence, not of merely linguistic rules. I have said that my position does not require two meanings of "value" or "ought" anymore than it requires two meanings of "fact" or "is." However, the last paragraph may well cause the reader to wonder whether I can consistently maintain this in the light of the distinction there drawn between two uses of "ought." Now, to be sure, there are two kinds of values and two kinds of facts, namely, brute and institutional. Look at it this way. When we are engaged in the activity of discovering brute facts, there presumably are constitutive rules for the use of "is" which give us

46 Ethics the right to use "is" to indicate what is the case (a brute fact). These linguistic rules do not themselves determine what it is that exists or is a fact. Similarly, when we are engaged in the activity of evaluation or making value judgments, that is, determining what is to constitute our brute values, there presumably are constitutive rules for the use of "ought" which give us the right to use "ought" to indicate what ought to be the case (a brute value). These linguistic rules do not themselves determine 'what it is that ought to be or is a value. Now let us move to the level of institutions like promise making and chess playing. When we are engaged in the institutional activity of promise making or chess playing, there are constitutive rules which do determine what is the case (institutional fact) and what ought to be the case (institutional value). They accomplish this not because they are different rules for the use of "is" and "ought" but because they are "meaning rules" for the use of such expressions as "to promise" and "to play chess." The constitutive rules which determine the meaning of "promise," for example, set forth conditions the satisfaction of which determines both that a particular promise has been made (institutional fact) and that it ought to be kept (institutional value). The constitutive rules which generate an institutional fact logically tie the notion of "promise" to "obligation" and to "ought." It seems to me that the meaning of "is" and "ought" remains unchanged in this process. Both "brute" and "institutional" describe fact, and this refers to what is the case in the same sense of "is"; both "brute" and "institutional" describe value, and this refers to what ought to be the case in the same sense of "ought." The difference lies not in the different meanings of "is" and "ought" but in the different uses of "is" and "ought" relative to two different levels of activity-the former use being logically independent of what actually is or ought to be the case (for surely particular brute facts or brute values are determined more by something like empirical discovery or rational insight, not linguistic rules), while the latter use is itself derivatively determined by the constitutive rules of the institution of promise making (for they just tell us, in this context, what is to count as a particular institutional fact or value). A similar analysis can be made for the institution of chess playing. Consider Searle's nihilist for a moment. He, in a stance which is "external" to the institution of promise making, must decide whether it is good or bad. But regardless of his decision, he, of course, must use "promise" as others do. He must grant that if a promise is made, the promiser undertakes an obligation and ought to keep his promise. He must grant all this because of the logical consequences of the meaning of the English word "promise." Now if one likes paradoxes, we can agree that the nihilist must say both that "One ought to keep one's promises" is a tautology and that "One ought never to keep a promise" is true. The paradox is resolved when we realize that the former statement is justified by interpretation a above, while the latter statement is justified by b. In other words, he can say, "One ought never to keep a promise" because

47 Institutional Facts and Brute Values the institutionalized obligation alluded to in the former statement is overridden by what for him is a noninstitutionalized higher one, namely, the one based on his brute value judgement that the whole institution is evil because he thinks it interferes with his self-fulfillment, just as Searle allows, even in his own deduction, that it is legitimate to reject the conclusion when a superseding institutional obligation intervenes. Thus, if Jones promises Mary that he will murder Smith, then even though he is obligated to murder Smith (and by logic alone, as it were, he ought to), nevertheless we all (except perhaps for our egocentric nihilist) would say that Jones ought not to murder Smith because his obligation to fulfill his promise is overridden by his obligation to respect human life. Hence the latter case is no more paradoxical than the former. In any event, as I suggested earlier, there are at least three kinds of overriding obligation situations: (1) intrainstitutional cases, where, for example, one promise overrides another because the former one is more important; (2) interinstitutional cases, where one institutional obligation overrides an obligation of another institution; and (3) extrainstitutional cases, where a brute value obligation overrides particular institutional obligations. Clearly, our egocentric nihilist falls in category (3) because he would justify his practice of not keeping promises not by invoking some other institutional obligation (how could he?) but by direct appeal to his rejection of the institution of promise making. In his eyes this rejection would allow him to renege on any "promise" he might have pretended to make. This is reminiscent of Adeimantus's slight qualification of his brother Glaucon's suggestion to Socrates that justice is merely an advantageous convention originating by social contract, namely, that though this is true, the greatest advantages will come by appearing to be just while really being unjust -thereby appeasing both men and gods! And Machiavelli tells us that the prince need not actually be virtuous, but just appear to be so. Perhaps I can further clarify my argument by stating my case in still another way. Let us distinguish between what I will call de dicto obligation or "ought" and de re obligation or "ought." The sentence "One ought to keep one's promises" would involve a de re "ought"; the sentence "It ought to be the case that one keep one's promises" would involve a de dicto "ought." Now, just as de re necessity exhibited in "John is necessarily rational" is different from de dicto necessity exhibited in "It is necessary that John is rational" (because the first statement is a contingent statement depending on the existence of John and applies to "this possible world," while the second claims to be a necessary statement applying to "all possible worlds"), analogously, "One ought to keep one's promises" (as it occurs in Searle's usage) is contingent upon the actual institution of promising and its underlying constitutive rules, while "It ought to be the case that one keep one's promises" is not contingent upon the institution of promising or adequately understood by reference to any rules concerning the meaning of "promise." Rather, it must be grasped

48 Ethics as an assertion of brute values and a fundamental evaluation of the institution of promising. If asked why the former statement is true, we could answer by citing definitions and rules of usage; but if asked why the latter statement is true, the layman might say, "Well, I'm just committed to that principle," while the philosopher would probably start appealing to higher principles in the domain of ethical theory. Now, I immediately grant that these two different functions of the modal auxiliary "ought" are rarely distinguished grammatically. They are not distinguished when Searle treats "One ought to keep one's promises" as a tautology and yet does not consider the nihilist to be contradicting himself when he says, "One ought never to keep promises." What the nihilist really means is, "It ought to be the case that one never keeps promises." If he were strictly saying, "One ought never to keep promises," with the internal orientation, that is, in accordance with interpretation a above, then it would seem that his statement would indeed be self-contradictory. At any rate, the de re obligation as expressed in tautologous sentences like "One ought to keep one's promises" corresponds to an a priori analytic proposition; but the de dicto obligation expressed in sentences like "It ought to be the case that one never keeps promises" corresponds to a nonanalytic proposition because in no sense is its denial self-contradictory. To be sure, certain propositions as asserted in sentences like "It ought to be the case that one keep one's promises" or "It ought to be the case that one never should lie" may well be necessarily true (a priori), but, if so, then they are synthetic, not analytic. In other words, in these terms, statements asserting brute values may be a posteriori synthetic (Mill) or a priori synthetic (Kant); statements asserting constitutive rules which define institutional values are a priori analytic. On the basis of this discussion, I cannot agree that the version of the naturalistic-fallacy thesis which Searle refutes is the version that the great majority of philosophers have maintained or implied, nor can I see why these philosophers cannot accommodate Searle's notion of an institutional fact. Regardless of the interpretation of historical texts, it seems to me that the least we have shown is that there is a philosophical claim concerning the naturalistic fallacy in respect to which Searle's counterexample fails, namely, the claim that statements asserting brute values cannot be logically derived from statements asserting facts. I would be confident that a thorough analysis of the views of those philosophers who have maintained the naturalistic-fallacy thesis in its various forms would generally support my interpretation of the thesis. But this is not the place for that formidable task. III At this point in my discussion, I want to respond to three possible misunderstandings. In the first place, nothing I have said should be taken to imply that I consider it possible for an individual "to step out of the

49 Institutional Facts and Brute Values moral game," as it were. My example of the so-called nihilist anarchist was borrowed from Searle, and, moreover, everything I have said about any such nihilist, insofar as it pertained to the issue at hand (namely, the distinction between brute values and institutional values), applies equally to the antinihilist or anyone else. If nihilism implies the rejection of all values of all kinds, then nihilism is self-contradictory because the nihilist must admit that for him promise breaking and lying, etc., are values. The truth of the matter is that the nihilist, like everyone else, must have values; it is just that they do not conform to the values of practically everyone else. So the nihilist, like everyone, must "play the moral game." Unfortunately, there are different ways of playing it. Second, nothing I have said should give the reader the idea that I have surreptitiously slipped into the discussion an argument for absolute values or eternal verities. No such beings are lurking in the wings of the argument. My argument for the brute value-institutional value distinction applies regardless of the particular kind of ethics anyone may endorse. The distinction does not turn on any kind of dichotomy between subjective versus objective values, or relative versus absolute values; it depends primarily on the differences among intrainstitutional, interinstitutional, and extrainstitutional evaluation-or better, the difference between evaluation within the context of an institution and evaluation of the institution itself. The first two misunderstandings are relatively trivial, but the last is fundamental and would really require a lengthy argument to resolve. It concerns the whole notion of extrainstitutional evaluation. Consider the following objection to my argument: You have talked as if it were possible for an individual somehow to achieve a noninstitutionalized perspective or standpoint. This would be analogous to being suspended in empty space with nothing to stand on. But all human activities are instances of institutional facts. The activity of evaluating the institution of promise making is itself an activity governed by constitutive rules, and therefore, if (as you say) institutional facts only generate institutional values, then all values are institutional. Thus, it is indeed true that brute values cannot be deduced from facts-but only because there are no brute values to be deduced! Confronted with this argument, I feel somewhat as Socrates did after hearing the arguments of Simmias and Cebes in the Phaedo. Let me say first that I in no way meant to imply that an individual could engage in the activity of evaluating while extracting himself from all forms of institutionalized activity. On the contrary, all of us are probably involved in many institutions simultaneously; and in the activity of evaluating the institution of promise making, I already granted that our nihilist would be following constitutive linguistic rules which govern the use of "ought" as well as "promise" (that was why he had to use the word "promise" in the same way the rest of us do). The crucial question is whether, from the

50 Ethics standpoint of the evaluator of the institution of promise making, he is involved in the endorsing of values at a level which is logically independent of the endorsement of the institutional values which the institution of promise making internally requires. If he is (as I have argued), then the values he endorses are not logically deducible from statements describing the institution of promise making, statements asserting the constitutive rule for the usage of "promise," or definitional tautologies logically generated from within the institution. Now, is the activity of evaluating the institution of promise making itself an instance of an institutional fact at a higher level? Well clearly, if an institutional fact is one which cannot be sufficiently explained by reference to brute facts (physical-psychological explanations), then all rational activity, as activity which is rulegoverned behavior directed to some end, is a form of institutional fact. But the above objection is wrongheaded in its assumption that just because all activity is institutional, therefore all values are institutional. Even if this were so, the values endorsed at the level of evaluating the institution of promise making would still maintain their logical independence of the values within the institution-and this would be one reason why we are justified in calling them "brute values" in relation to the institutional values internal to the judged institution. There is no reason to make a metaphysical distinction between brute and institutional values or brute and institutional facts. Anscombe recognized the distinction between brute and institutional facts, but, unlike Searle, she plausibly argues that what is a brute fact in one context is an institutional fact in another, and vice versa.4 The same relativity can apply to the distinction between brute and institutional values. This would be my "weak" defence against the above argument. The "strong" defense does not turn on the issue of whether the "brute" versus "institutional" distinction is relative or nonrelative. The crucial point is that the fact that all activities are institutional (and therefore may generate certain institutional values at all levels) does not entail that all values are institutional, that is, that there are no brute values. To argue that the activities of value inquiry, evaluation, or value judging are themselves institutional and therefore cannot involve anything but institutional values is like arguing that since the activity of scientific inquiry is institutional, it cannot involve anything except institutional facts-but it is precisely by means of scientific inquiry that we discover brute facts. There is nothing in our analysis which requires us to say that institutional activities (like rational inquiry) cannot result in the determination or discovery of brute facts or brute values. Thus, what we called "extrainstitutional" was not the activity of inquiry into facts or values but the brute values themselves. It is the same with science. The brute facts are extrainstitutional, but the activity of inquiry in accordance with the canons of 4. G. E. M. Anscombe, "On Brute Facts," Analysis, vol. 18, no. 3 (1958).

51 Institutional Facts and Brute Values scientific method which leads to the discovery of brute facts is an institutional activity. So the activity of rational inquiry conforms to constitutive rules which (like other constitutive rules) create institutional fact. The fact they create is just the fact that human beings engage in the activity of rational inquiry, but this fact entails nothing as to what particular value judgments they ought to make or why they should make any at all, what particular things are valuable or what particular things exist. That is, although the institution of inquiry may commit the inquirer to any inherent institutional facts and values logically implied in the notion of inquiry itself, it clearly requires no a priori commitment to the particular values it may endorse or the facts it may discover. Searle's derivation cannot possibly fail because it amounts simply to setting forth the logical consequences of a systematic analysis of the speech act of promising-an institutional fact grounded on constitutive rules for the use of the word "promise." Nonetheless, Searle's explicit recognition of the distinction between internal and external endorsements of systems of constitutive rules does imply the distinction between institutional values and brute values. He argues that we must recognize the existence of institutional facts because there are many existing states of affairs which cannot be explained by brute facts alone. Likewise, we must admit of brute values because there are value commitments which cannot be explained by institutional values alone, much less by institutional facts or brute facts. And just as it is true that a description of institutional facts may well include a description of brute facts (but not conversely), so also a description of brute values may well include a description of institutional values. Our whole discussion of "brute" versus "institutional" has revolved around three notorious distinctions: (1) the fact-value distinction, (2) the analytic-synthetic distinction, and (3) the descriptive-evaluative distinction. It is a general consequence of our argument that these distinctions, if they are to be philosophically useful and interesting, depend on a clear understanding of what is meant by the brute and institutional levels of experience. Roughly, we can sketch our results as follows: (1) On the noninstitutional level, the distinction between fact and value is dichotomous and clear-cut; brute facts are existing states of affairs which may or may not be valued, and brute values concern desirable states of affairs which may or may not exist. At the institutional level this thesis no longer holds because institutions, as systems of constitutive rules, provide frameworks for rule-governed intentional activity directed to preestablished ends-frameworks based on constitutive rules which can analytically connect institutional facts with institutional values. The rules formalize the relation between means and ends, between activities which necessarily conform to the rules and the goals or "upshots" which are necessarily determined by the rules. Thus, "promise" implies "obligation" implies "ought to keep one's promises," and "play chess" implies "commitment to

52 Ethics the rules of chess" implies "ought to play to win in such and such a way," etc. In this context, the fact-value dichotomy collapses because a "fact" is typically an activity conforming to certain rules necessarily directed to certain ends or values, and a "value" is typically what ought to be achieved by means of some actual activity. The fact of promising entails that one ought to keep his promise, and if one ought to keep it, one must in fact have made a promise. An existing activity, as a fact, is just the realization of some value; and a desirable goal, as a value, is just that which exists as a result of an instituted activity. (2) On the noninstitutional level, any proposition expressing what is the case or what ought to be the case-for example, that X exists, that X is Y, or that X ought to be Y-is always synthetic because the truth value of any claim concerning a brute state of affairs is not determined by logic alone but by evidence and reasons. As such, it is always debatable, logically deniable, and subject to modification or revision in the light of new evidence or better reasons. On the institutional level, the analytic-synthetic distinction becomes hazy and is dependent upon the context of utterances-even though conceptually (as Searle insists) we are quite clear concerning the logical criteria for the distinction. But in practice the institution of language can, in an appropriate context (regardless of whether this happens by actual enactment, natural evolution, habit, or whatever), provide a constitutive rule of linguistic usage such that what was previously treated as a synthetic statement now becomes analytic. For example, is "All bodies are heavy" synthetic or analytic? Originally, it may well have been treated as synthetic because it seems to report a brute fact-a fact which presumably could have been otherwise. But with long-established usage coupled with increased knowledge about bodies, we may well decide to declare that "body" just means "heavy" (among other things). In short, if we become sufficiently convinced that all bodies must have mass, then we may have no misgivings about defining "body" in these terms. What about "Extra Fancy Grade apples are those which are red, juicy and sweet"? Can one reasonably argue about this? Hare and Urmson maintained that evaluative terms cannot be defined by expressions which contain only nonevaluative terms. But were they talking about institutional or noninstitutional values? Clearly, the business about apples would be a stipulative definition pronounced by an institution, namely, the Ministry of Agriculture. My point is that brute facts and values cannot be defined into existence, but institutional facts and values can; and in the latter context, the analytic-synthetic distinction must take account of institutional reality and constitutive rules if it is to have any coherent application. (3) On the noninstitutional level, the descriptive-evaluative distinction refers to the illocutionary force associated with propositions expressing brute facts or brute values. If a proposition of this kind is expressed descriptively, it merely describes or gives an account of certain states of affairs-regardless of whether they concern facts or values; while if it is expressed with the force of evaluation, it implies

53 Institutional Facts and Brute Values endorsement or valuation (or their opposites) of those states of affairs. It is at this level, I submit, that the nonderivability thesis holds. At the institutional level the thesis fails because, as Searle's argument shows, once the level of institutional fact is postulated, an evaluative statement can be logically entailed through definitional tautologies; and this is possible because institutional facts can be defined in evaluative terms. Thus, "This apple is red, juicy and sweet" is both a descriptive statement (because it simply states the fact that the apple has certain characteristics) and an evaluative statement (because of the institutionalized definition which identifies the characteristics as value-giving characteristics). When the "classical moral philosophers" argued that evaluative terms (especially "good") could not be defined by means of nonevaluative terms, and that expressions like "X is a good Y" could not be validly deduced from expressions asserting only nonevaluative characteristics of X (unless a synthetic major premise was assumed which connected the characteristics with being a good Y), they were in a way wrong and in a way right. I think they recognized that brute values cannot be defined into existence, but, as Searle shows, they did not adequately recognize the place of institutional facts and values in our ordinary thinking and discourse. They failed to recognize that "All apples having properties A, B, C are Extra Fancy Grade" is a "tautology" because it is a constitutive rule defining the use of "Extra Fancy Grade" relative to the institution of apple grading as practiced by the Ministry of Agriculture. Yet, they were correct in insisting that, in another important sense, the attribution of value to properties A, B, C is a synthetic judgment. They failed to appreciate the fact that institutional values can be logically derived from institutional facts because (I think) they were more concerned with the notion that institutional values cannot be derived from brute facts, and that brute values cannot be derived from any facts at all. Philosophers like Hare and Urmson were wrong if they identified the meaning of an expression with its illocutionary force (speech act fallacy) or with the conditions for its successful assertion (assertion fallacy), and if they were also guilty of the naturalistic-fallacy fallacy (as Searle defines it), it was not because the nonderivability and nondefinist theses are false but because the interpretation of these theses as theses asserting the impossibility of logically deriving or defining institutional values by means of institutional facts is indefensible. Thus, to maintain in a general way (ignoring what kinds of facts or values are involved) that no evaluative expressions can be defined by, reduced to, or derived from expressions which lack evaluative terms, is highly ambiguous and vulnerable to any number of counter-examples. This way of stating the thesis fails to take account of the primary distinction between institutional and noninstitutional facts and values, along with all the secondary distinctions related to the problem. But the thesis becomes true (or at least not demonstrably

54 Ethics false) when we make explicit that the values we are talking about are brute values. If Jones made a promise, then, indeed, it is a fact that he made a promise. But the issue as to whether promises obligate is ambiguous. As institutional facts, they do by definition; but whether they ought to obligate concerns brute values. How does one determine what these brute values are? That is a little like asking how one determines what the brute facts are. Is the answer to the latter query, "By empirical observation"? Then perhaps the answer to the former question is, "By rational insight." Incidentally, in spite of my extended argument, I am not altogether convinced that statements expressing brute values cannot logically be derived from statements expressing only facts. It is just that if what I have said is correct, then I have yet to see a counter-example to the thesis.