The Expressivist Circle: Invoking Norms in the Explanation of Normative Judgment

Philosophy and Phenomenological Research Vol. LXV, No. 1, July 2002 The Expressivist Circle: Invoking Norms in the Explanation of Normative Judgment JAMES DREIER Brown University "States of mind are natural states. They are extremely hard to define."1 1. Naturalizing Normative Judgment To naturalize normative judgment is to give some account of it, in naturalistic and non-normative terms. Simon Blackburn's Ruling Passions embraces naturalism, about ethics especially. To be a naturalist is to see human beings as frail complexes of perishable tissue, and so part of the natural order. It is thus to refuse unexplained appeals to mind or spirit, and unexplained appeals to knowledge of a Platonic order of Forms or Norms; it is above all to refuse any appeal to a supernatural order... So the problem is one of finding room for ethics, or of placing ethics within the disenchanted, non-ethical order which we inhabit, and of which we are a part. [48-49] Some kinds of naturalism are reductivist, but expressivism is not. All versions of expressivism agree that it is impossible to give any descriptive predicate that is analytically equivalent (roughly, synonymous) with a moral or other normative predicate. Normative assertions express psychological states that are not mainly representational. Normative predicates supply the expressive role. I need not go into details in this essay.2 Here is Blackburn: Valuing something, it [my theory] says, is not to be understood as describing it in certain terms, any more than hoping for or desiring something are describing it in particular terms. Rather, the state of mind of one who values something is distinctive, but nevertheless it is itself a natu- ral, and naturally describable, state. [49] 2 Ruling Passions (Oxford: 1998) p. 53. References to this book are hereafter by page number alone, cited in square brackets in the main body of the text. See especially pp. 48-50. 136 JAMES DREIER

So reductive analysis in the sense of providing a descriptive analytic equivalent for a normative predicate is out of the question. Nevertheless, expressivism is naturalistic. While I have never quite been able to believe expressivism, I am much more sympathetic than many to its kernel. It does seem to me that the core explanation of normative judgment must be broadly expressivist. The alternative, I think, is that there is no explanation at all. And while we can't rule out the possibility that the very idea of normative judgment cannot be any further explained,3 I regard that possibility as a last resort. So it comes as an uncomfortable surprise to me that there is a potential snag in the general form of expressivist explanations of normative judgment. If the snag cannot be untangled, it renders the explanation circular, or plunges it into an infinite regress. Blackburn observes the snag from the deck of Ruling Passions but does not seem to think it is very serious. I aim to alert fellow travelers (and critics) to the danger, and then to suggest how the snag might be untangled. II. The Problem The problem I am after has been mentioned, or at least approximated, by a few philosophers, but to my knowledge it has not yet appeared in any published article.4 In any case, for expository purposes I will present my own formulation. It is sometimes said that belief and desire are normative concepts. I will put it this way: ascriptions of beliefs and desires to agents are themselves normative judgments. I am neither endorsing this claim nor denying it, but it is undoubtedly popular. I will be quoting some of Blackburn's discussion of the claim. In a nutshell, the idea is that when we say that someone desires to drink, we imply (analytically) that insofar as he believes that the only way he can drink is by pouring water into the cup before him, he ought to pour the water. As it's sometimes put, the very idea of a desire (or a belief) with a certain content is analytically bound up with what it 'makes sense' for the desirer to do.5 If this is right, then the problem I'm worried about appears right away. The expressivist analysis of normative judgment invokes ascriptions of Jean Hampton raised the possibility that normativity is inexplicable in The Authority of Reason (Cambridge: 1998). As far as I can tell, she believed normativity is inexplicable. See especially pp. 100ff. Frank Jackson raises a related problem in "Noncognitivism, Normativity, Belief', Ratio 1999; 12(4), 420-35. In his unpublished "Five not-much-discussed problems for noncognitivism" Michael A. Smith counts it among the five. Nick Zangwill skirts very close to it in an unpublished paper, "An argument for normative realism". This notion of what 'makes sense' is regarded by Allan Gibbard, and now it seems by Blackburn too, as the most fundamental normative notion. See Gibbard, Wise Choices, Apt Feelings (Harvard: 1990) p. 6. BOOK SYMPOSIUM 137

propositional attitudes. In some early versions the attitudes were proper desires, or preferences; in some others the attitudes are 'valuings' or 'norm acceptance'. Whatever they are, the considerations that make it plausible that belief ascriptions and desire ascriptions are normative apply alike to the other propositional attitude ascriptions. And this means that the kernel of expressivist analysis invokes normative concepts. This circularity is a big problem. The expressivist analysis now appears circular. Insofar as all the various normative notions we use freely in ordinary language are supposed to be explained by the same basic type of expressivist analysis, an apparently vicious circularity (or possibly an infinite regress) will arise if each analysis must invoke some normative concept of its own.6 III. Really Any Problem? Blackburn seems to have noticed this problem, in two different parts of Ruling Passions. In the Common Questions appendix, he presents the following question and answer: Q. 20. If all descriptions of human psychology, including descriptions of what people believe and desire, are implicitly normative, how can normativity be essentially a matter of attitude? Ans. All such descriptions are plausibly regarded as normative because they seem to imply views about what it 'makes sense' for a person to do, if they are in the states attributed to them. And affirming that something did or did not make sense is entering a normative judgment. If we accept this line of reasoning, all I then add is that the verdict that a person's behaviour does or does not make sense itself expresses an attitude. And where is the harm in that? [319-20] Let's try to say explicitly where the harm is. As I said, if 'descriptions' of human psychology are really always normative, and these descriptions appear in the expressivist analysis of normative terms, then the analysis seems to be trapped in a circle or a regress. This point may not be so obvious as I have made it out to be. It is pretty clear 6 There is another problem. The observation that propositional attitude ascriptions are normative looks to be incompatible with the quotation from Ruling Passions that I've used as an epigram, and also with the remark, cited above: "Rather, the state of mind of one who values something is distinctive, but nevertheless it is itself a natural, and naturally describable, state." [49] States of mind, it might be said, are not natural states at all, any more than moral virtues are. (Of course, in one sense an expressivist is happy to agree that virtues are natural states-there is such a thing as courage, after all, and what could it be but a natural state? But attributing a virtue is not merely saying that a person is in some particular natural state; nor, it appears, is attributing a belief or desire.) And they are not "extremely hard to define" if by "define" is meant, to give a purely descriptive equivalent. They are impossible to define. But this is not the problem I want to focus on. Were it not for the problem of circularity discussed in the text, an expressivist could simply discard the idea that states of mind are natural states. 138 JAMES DREIER

that it some contexts, explaining one normative concept by means of another is perfectly acceptable. For example, a consequentialist theory that seeks to explain the idea of moral rightness by reference to the idea of moral goodness is not subject to any objection that this is entirely the wrong kind of explanation. Within normative theory there are interesting and controversial reductions of just this sort. But expressivism is not a view within normative theory at all. It is supposed to be explaining normativity, not just explaining one kind of normative judgment by reference to another. Consequentialism is not a competitor of expressivism.7 When Blackburn asked where the harm is, maybe he was suggesting that it might simply turn out to be true that moral judgment is the expression of some certain complicated intentional state, and that attributing such a state would indeed turn out to be expressing some further complicated state. If it is true, and expressivists are committed to it, well what could be the harm? It could be true, yes. But it could not be an explanation of normative judgment. For conveying it presupposes normative judgment. Using some normative terms to explain the meaning (or role) of some others is not, as I noted, always a pointless enterprise; it is one interesting project within normative theory. But if the aim is to explain normative judgment in general, then using some normative terms in the explanation of others is merely pushing the problem to one side. When we then try to explain what this judgment of 'making sense' is all about, we cite further attitudes (or the same ones) that the judgment expresses, and we are heading down an infinite regress (or around a circle). IV. How Blackburn Understands the Normativity of Belief and Desire Though Blackburn's answer to the question in the Appendix shrugs it off, in the text of Ruling Passions he takes the challenge more seriously. The strategy that he adopts is to 'de-fang' the norms associated with intentional states. Once again, there is a second potential harm. If 'descriptions' of human psychology are really always normative, then they are not really properly called 'descriptions' at all, according to expressivist theory. There may be no harm, in ordinary language or even some kinds of philosophizing, in calling them 'descriptions', but they are not best understood that way if they are normative. They are rather expressions of attitude. The contrast between genuine describing and expressing is a crucial one in expressivism. Valuing something, it says, is not to be understood as describing it in certain terms, any more than hoping for or desiring something are describing it in particular terms. Rather, the state of mind of one who values something is distinctive, but nevertheless it is itself a natural, and naturally describable, state. [49] But now it appears that saying what state someone expresses by a normative judgment is not describing, or not exclusively describing. Attributing an intentional state is at least partly expressing a state of one's own. The state of mind of one who values is no more a natural state than moral wrongness is a natural property. BOOK SYMPOSIUM 139

Then the question is how to take references to 'making sense' in these clauses. They seem to be statements about what you would expect from an agent. It would surprise you, and lead you to scramble for another interpretation of X, if X does not behave in ways that make sense. Similarly we might say that the circuit is wired up if it is in a state such that, when the key is turned, there ought to be a spark. But the 'ought' here speaks not of duties and values, but just about what you would expect. If there is no spark, then something has gone wrong, but this too means simply loss of expected or intended function. [56] This passage makes it appear that when we say that someone who believes that if Socrates is a man then he is mortal ought to believe (say) that if Socrates is not mortal, he is not a man, the 'ought' is merely the 'ought' of expectation. It is not really normative at all. No meaning would be lost by replacing all occurrences of "ought to" with "is likely to". When I conclude "there ought to be a spark" from "the key was turned in the ignition", there is no implication at all that the proposition that the key was turned in the ignition is normative. Saying that if you fear dogs, then when you come upon one you ought to flee, is just saying that this is the kind of behaviour you would expect of people who fear dogs. There is no implication that if you do not flee, you are somehow irrational. You are only to some extent doing something unexpected, to be explained with further investigation, and even indicative that your fear of dogs is not all that real. [56] I have two doubts about this strategy. First, it simply seems wrong to say that our ordinary idea of what someone ought to think or do, given what else he thinks and wants, is not normative. When someone believes that all men are mortal, and that Socrates is a man, then he ought to believe that Socrates is mortal, and if he doesn't then he is flouting rationality. While there certainly does seem to be a non-normative notion of what 'makes sense', the ought of what someone ought to believe seems to be paradigmatically normative. Second, while in the example of the key in the ignition we are inclined to say that there ought to be a spark, we do not think that it is analytic that there ought to be a spark when the key is turned. On the other hand, we are inclined to say (though I haven't given much of an argument for this claim) that it is analytic, constitutive of the very idea of belief and desire, that when someone has certain beliefs and certain other desires he ought to behave a certain way. This difference is important, I think. It suggests that (insofar as these oughts are normative) the belief and desire attributions are normative themselves, while the proposition that someone turned the key in the ignition is not. V. Toward a Solution Maybe there are two different ways in which something could 'make sense' to do. Or rather, as an expressivist would prefer to put it, there are two different 140 JAMES DREIER

things we can do by saying that something 'makes sense'. One thing we can do is to express norms approving of that sort of thing. The other thing we can do also makes use of norms, but does not express them. We can say, for instance, that in Monopoly it makes sense to build houses as soon as you own a monopoly, or that in chess it makes sense to castle early, or that in a close baseball game it makes sense to sacrifice a baserunner to second. Sometimes we might be expressing our approval of such tactics. But not always. Even if I think that baseball is a stupid waste of time, I can still see that it makes sense to use various tactics relative to the goals and rules of baseball itself. If I make this relativization explicit, or even if I leave it tacit, I drain the expressive (or as Hare says, 'prescriptive') function of the other- wise normative terms. Usually when I say what makes sense in a game, the norms to which I relativize are the goals of the game. But I may also relativize to the rules of the game. I can say what one 'may' do in chess (capture diagonally with a pawn), what one 'must not' do (move into check), what one is 'required' to do (move or interpose or capture the attacking piece if possible when in check). When I say such things, it is not always plausible to understand me as expressing norms I accept.8 I am not telling anyone what to do, except in the sense of telling them what to do if they are to play chess. Compare Blackburn: Consider the example of game-playing. Here, too, there is a definite normative order: a game is defined by its rules. There is a limit to the extent to which people can fail to conform to the rules. [57] You could break a rule of chess and still be said to be playing-just playing very badly, as a six year old who jumps his bishop over a pawn. But if you break too many rules too often, then we just don't count you as playing chess. And then we would stop telling you what you are 'required' to do, we'd say, "Oh, well, do what you like, of course, but that isn't chess." To summarize the point of the analogy with games: Some rules are regulative, and the normative judgments powered by them express normative attitudes. Some rules are constitutive, and the normative judgments underwritten by them are analytic. Maybe the words 'belief and 'desire' and the verb 'to value' are like that. We cannot attribute beliefs and desires without 'buying into' the normative game that defines them, any more than we can speak seriously about promises while staying neutral about whether the man we say has promised has really obligated himself, or speak seriously about 'castling' without invoking the rules of chess. 8 Thanks to John Devlin for forcing me to see this point, which now seems perfectly obvious, but which I inexplicably resisted at the time. Thanks also to Russ Shafer-Landau for helpful comments. BOOK SYMPOSIUM 141

Suppose this is right. What does it mean for expressivist analysis of normative expressions? I think the upshot is that the analysis can be given, but expressivists have to be careful in giving it. I suggest a judicious use of the operator, so-called. If I am speaking with or about the adherents of a religion I do not share, I might say that chastity is a so-called virtue and pride a so-called sin. I mean that chastity is a virtue according to that religion, and pride a sin according to it. The operator keeps me clear of commitment myself. In a real conversation, I might often just say that chastity is a virtue and pride a sin, leaving the audience to supply the inverted commas. Note that by saying that pride is a so-called sin, I am not saying that it isn't a sin. In avowing my atheism, I might say something like, "It's all a mistake, there is no such thing as sin, so pride is no sin." But the assertion that pride is not a sin is most naturally understood as a "move within the game," and I don't want to be playing the game. And in giving their analysis of normative expressions, expressivists (nowadays) certainly do not want to say things like, "It's all a mistake, there is nothing at all that we ought to do." In presenting their analyses, they don't mean to be playing at all. So my suggestion is that expressivists should say something like this: Saying what people ought to do is expressing a so-called desire that such things be done. Of course, the attitude could be any of a number; I have chosen desire only as an example. This analysis is free from the circularity problem. The norms that are built in to the concept of desire are not expressed here, though they are made use of; the notion of a so-called desire is not a normative notion but a descriptive one, so no norms are invoked in giving the analysis. Suppose I am trying to explain to you the chess game on television. "White is in whatthey-call check," I say. If I said that white was in check, I could not resist the conclusion that white must move his king. But I do not want to draw that conclusion, because I think chess is a dangerous throwback to feudal values, so I want to say instead that white ought to insist on moving his pawns forward, thus engaging in radical rebellion against the oppressive FIDE lords. With approximately equal plausibility, I might decline to call your mental state a 'desire to avoid dogs', because I do not want to be committed to the conclusion that you ought to run away when you see a dog approaching. Instead my attitude toward people in your mental state is that I approve of them hugging any nearby dog. So I say instead that you have a so-called desire to avoid dogs. Of course, the bizarre attitudes aren't the only reasons for avoiding committing oneself to one norm or another; I might instead be doing metaphysics. Not that metaphysicians of normativity are frightened of 142 JAMES DREIER

commitment, or that they have bizarre attitudes, but the explanation they are giving would be ruined if it included a normative judgment. Needless to say, I hope, I am not claiming to have shown that any particular expressivist analysis is correct, nor that the general expressivist approach is correct. My goal is much more limited: to show how an expressivist analysis might explain normativity without invoking it. BOOK SYMPOSIUM 143