Humean motivation and Humean rationality

MARK VAN ROOJEN HUMEAN MOTIVATION AND HUMEAN RATIONALITY* (Receivedin revisedform 15 March 1994) The Humean theory of motivation, encapsulated in the slogan "Reason alone cannot motivate action," has been both attracted and powerful. It is attractive, in part, because desire is required for intentional action. Further, it can easily explain weakness of will and accidie as stemming from the fact that beliefs alone cannot motivate action. Anti-Humean views of motivation, according to which beliefs alone are motivational, may even seem to make weakness of the will and accidie impossible. The power of the Humean thesis can be seen in the way accepting it can quickly lead to ethical relativism. On the plausible assumption that reasons for action must be able to motivate, the thesis can lead quickly to the conclusion that the reasons a person has depend on the desires that person has. Instrumental rationality will be the only kind of rationality. Thus, people with sufficiently different desires will have reason to do different things, even in the same circumstances. And, if internalism about ethics is right, and ethical evaluations provide reasons for actions, some version of ethical relativism seems inescapable. Now, as Thomas Nagel has pointed out, the desire necessary for intentional action may itself have been motivated by other considerations. 1 Thus, these strong conclusions are warranted only if these other considerations cannot themselves motivate without the help of some preexisting desire. Consider Michael Smith's formulation of the Humean thesis: R at t constitutes a motivating reason of agent A to • iff there is some ~ such that R at t constitutes a desire of A to k9 and a belief that were he to ~ he would g/.~ For this formulation to capture the issue, the desire mentioned must be neither entailed by the presence of, nor partially constitutive of, any Philosophical Studies 79: 37-57, 1995. Q 1995 Kluwer Academic Publishers. Printed in the Netherlands. 38 MARK VAN ROOJEN belief. Otherwise, that belief will itself, together with the appropriate means/ends belief, constitute a motivating reason. 3 So understood, the Humean thesis can play its role in the argument sketched above. But, it now itself needs defence, particularly if the anti-humean can leave room for weakness of will. For the claim now goes beyond the truism that motivation involves desire. Michael Smith's recent defence of the theory shows promise, in that it captures the most common reasons for accepting a Humean view. 4 But, as I will argue, it falls short of vindicating the view. Smith's argument fails, because it ignores the role of rationality conditions on the ascription of motivating reason explanations. Because of these conditions, we must have a theory of rationality before we choose a theory of motivation. Thus, we cannot use Humean restrictions on motivation to argue for a particular conception of rationality. I will not directly criticize using a Humean conception of rationality to defend a Humean theory of motivation. For my argument implies that such criticism must come more directly, as argument over the substantive content of rationality. AN ARGUMENT FOR THE HUMEAN THEORY OF MOTIVATION Smith argues that motivating mason explanations are teleological explanations and that teleological explanations essentially involve something having a goal. 5 He then argues that a proper filling out of a dispositional account of psychology will show that having a goal essentially involves mental states that can only be desires, but not beliefs. His argument involves the following steps. (1) Motivating reason explanations are teleological explanations. (2) Teleological explanations explain the actions of a system/ agent by showing those actions to be part of the pursuit of some goal. (3) Specific non-complex psychological states, including beliefs and desires, are differentiated by their directions of fit or functional roles. HUMEAN MOTIVATION AND HUMEAN RATIONALITY 39 (4) States that are such as to fit the world are beliefs. (5) States that are apt to lead to a change the world to fit with them are desires. (6) One way of changing the world to fit with a mental state is to motivate some further state which itself is such as to change the world to fit with it. (7) No particular non-complex psychological states both fit the world as in (4) and lead to a change in the world as in (5). (8) To have a goal is to be in a state apt to lead to a change in the world, therefore, it must be desiring. (9) Therefore, motivating reason explanations must cite a desire as well as a belief. "Direction of fit" here is supposed to be filled out by giving a dispositional account of belief and desire, of the sort given by various functionalist accounts of the mental. On such views, combinations of mental states ground dispositions to act in varying ways. Different kinds of states are differentiated, at least in part, by the ways in which they contribute to these dispositions. For example, the state of desiring water, together with the state of believing that water is in the glass before one, typically disposes one to drink what is in the glass. On a dispositional view, the disposition to combine with the relevant beliefs in this way partly constitutes the desire for water. Similarly, the disposition to combine with desire in this way to produce action partly constitutes beliefs, such as the belief that this is water. For my argument, I will accept the view that some such dispositional account is correct.6 Two other things are worth noting about Smith's argument. First, step (6) is necessary to cover the kind of case that Nagel highlights, the case where the desire directly involved in intentional action is itself motivated by some previous state of mind. Step (6) tells us that a state which will be such as to motivate the desire involved in intention or any other desire, will be characterizable as one with which the world must fit. 40 MARK VAN ROOJEN Second, step (7) plays a crucial role, both in the Humean's argument, and in using the Humean theory of motivation to argue for an instrument theory of rationality. For, if a state could have all of the characteristics of belief, and still be such as to change the world without some additional desire, the anti-Humean could point to it as a motive that could operate without any additional desire. The same could be said if the belief alone, without the existence of some pre-existing desire, were enough to bring about a desire which leads to a change in the world. Step (7) rules out such "besires," states with both directions of fit, which are at once fully cognitive and yet have the ability to change the world of desires. 7 The claim in step (7) is argued for in two ways, each aimed at showing the impossibility of besires. The first starts with a particular interpretation of step (3), which claims that states are differentiated by their directions of fit. On this reading, states are differentiated by the direction of fit of their propositional content to or from the world. The belief that I now drink a soft drink, and the desire that I now drink a soft drink, on this way of thinking, have the same propositional content, namely that I now drink a soft drink. What distinguishes them is the way in which that content fits or comes to fit with the world. Given this account of direct of fit, the idea of a state with both directions of fit is incoherent. Smith puts the point: ... [A]s we have understood the concept of direction of fit, the direction of fit of a state with content p is determined, inter alia, by its counterfactual dependence on a perception that notp. A state with both directions of fit would therefore have to be such that both, in the presence of such a perception it tends to go out of existence, and, in the presence of such a perception, it tends to endure, disposing the subject to bring it about thatp. 8 Construing direction of fit as describing the relations between contents of mental states and the world, and postulating a state with both directions of fit will lead to contradictory requirements. The state must both persist and disappear in the face of evidence that its propositional content is not true. Thus, no state could have two directions of fit so construed. But, this is problematic as an argument for the Humean view. For, as we shall see, on this reading of direction of fit, it is not completely clear that the goal involved in teleologically explaining behavior, i.e. in motivating reason explanations, has to be the goal found in the propo- HUMEAN MOTIVATION AND HUMEAN RATIONALITY 41 sitional content of the states mentioned in the explanation. Thus, step (8) is not yet clearly warranted when direction of fit is construed as a relation between c o n t e n t s and the world. Consider, for example, the belief that drinking coffee soon is desirable. This might be responsive to evidence of various kinds. If evidence called into question the desirability of coffee drinking we would be disposed to modify the belief accordingly. Perhaps the coffee is carcinogenic, or proceeds from the sale of most coffee go toward propping up corrupt governments. At the same time, that belief might be such as to dispose us to pursue the imbibing of coffee should appropriate circumstances arise. Such states might be such as to have both directions of fit. For such states, however, their being such as to change the world would not involve changing the world so as to make their propositional content true. That direction of fit would rather involve changing the world so as to realize the state of affairs recommended by the content, in our example the drinking of coffee. And, we might think that this propensity to bring about changes in the world might be such as to ground motivating reason explanations. So the first argument fails. At this point we find the second line of argument against besires: ... [M]oral practice seems better explained by a theory that thus weakens the connection between evaluative beliefs and motivation - a tendency is not the necessary connection postulated by the n o n - H u m e a n . . . Stocker reminds us that in certain fits of depression, or self-deception, or in conditions of physical tiredness, we sometimes believe that a certain course of action is good and yet seem totally indifferent to it; not motivated at all to do what we believe good. Such examples are an embarrassment to the non-Humean who thinks that the evidence provided by moral practice supports the view that there are desires = beliefs. 9 Part of my task in this paper is to remove the sting from this objection. For, I will argue, a proper understanding of the dispositional or functional roles of beliefs and desires is compatible both with the motivational efficacy of cognitive states and with weakness of will and accidie. I agree with Smith,that motivating reason explanations are teleological explanations, and their teleological character restricts the ways in which a dispositional account of belief and desire must fill in their respective roles. But, my account will class all intentional explana- 42 MARK VAN ROOJEN tions as teleological. As a result, the teleological character of motivating reason explanations cannot be used to show that they involve desire; it will show only that they are a species of intentional explanation, which is essentially a kind of rationalizing explanation. The argument that motivating reason explanations are rationalizing explanations will have important consequences for the issue between Humeans and non-Humeans. That argument does not imply that the 6ognitive status of belief is incompatible with the motivational efficacy typical of desires. But it does imply that our accepting that such states exist should depend on whether or not we think that being motivated by such beliefs is rational for a person. In other words, the theory of motivation depends on our view of rationality and cannot be used as part of an independent argument for any particular view of rationality. Thus, while I use Smith's argument as a foil, the upshot of my argument is a very general point. It will have application against any account of motivation, Humean or otherwise, that does not independently motivate an account of rationality. WHAT ARE TELEOLOGICAL EXPLANATIONS? To assess whether motivating reason explanations are teleological, we need an account of teleological explanations. Since teleological explanations can be loosely characterized as explanations which explain by displaying a phenomenon as conducive to some "goal", a proper account needs to say what exactly is a "goal". My account will imply that the goals referred to in teleological explanation need not be the objects of intentions, and hence, teleological explanation need not be intentional. In so far as Humeans identify the two, their argument will need revision. According to Smith, teleological explanations are "explanations that explain by making what they explain intelligible in terms of the pursuit of a goal. ''1° Taken literally, characterizing all teleological explanations as aiming at a goal makes all teleological explanations into intentional explanations. This is too strict in that many plausible cases of teleological explanation do not involve the intentional pursuit of a goal. For example, functional explanations in biology purport to be teleological, HUMEAN MOTIVATION AND HUMEAN RATIONALITY 43 although we now believe that no one designed creatures or their organs. The function (or a function) of the heart is pumping blood, but that is not it's goal. We need a characterization which allows for teleological explanation where no intention is involved. Still, Smith's characterization of teleologically explicable phenomena as goal directed has a point. Typically, teleological explanations explain phenomena by referring to some effect of the phenomena to be explained. But without some restriction on which effects are relevant, too many explanations are teleological. Any system will have many effects, yet only some of these can figure in teleological explanations as the end which organizes the explanation. To take an example from biology, we can explain the construction of the heart in terms of its conduciveness to pumping blood. But that construction also enables the heart to make a thumping noise. We need a restriction that rules out some of these causal effects as accidental. Smith's restriction of the effects to goals serves that purpose. Smith's formulation also serves the function of distinguishing teleological explanations from explanations that display some event as an instance of a law we have inductive evidence to accept. Regularizing laws obtain, because their instances occur. To keep from counting regularizing explanations as teleological, we need a restriction on the kinds of effects teleological explanations can refer to. Here again, Smith's restrictions to effects that are goals serves the purpose. But, it rules out too much. t 1 We can, however, get the advantages of the account without the disadvantages by taking talk of pursuing goals metaphorically, and filling out the metaphor. The general idea is that the adaptiveness of function of certain phenomena explains their persistence, not their other accidental characteristics. Teleological explanations seem to be in the offing wherever a counterfactual of the form "Event E would not have happened (existed, come into being, been as likely, continued to occur, etc.), if E had not been necessary for the realization of goal G is true, and where our expectation that G will come about is based on something other than merely inductive evidence that G is realized. 12 Without the last clause, any account of the workings of nature that postulates regularities would ground a teleological explanation. A ball's falling would 44 MARK VAN ROOJEN not have happened if it had not been necessary for the law of gravity's holding. But that is not a teleological explanation. We can now distinguish the goal mentioned in the explanation from accidental effects of selection. It will not be true of the accidental effects of E, that E would not have occurred if E had not been necessary for their coming about. To go back to our example of the heart, it is not true that the heart would not have the character that it has if that character was not necessary for making a pounding noise. But it is true that the heart would not have the character that it has if that character was not necessary for pumping blood. For, if the heart's character did not enable it to pump blood, natural selection would not operate to favor creatures with such organs, and organs having the character of hearts would not continue to exist. We can, then, restrict goals to relevant rather than accidental effects, while distinguishing teleological explanations from regularizing explanations, without requiring the goals to be contents of intentions. This reinforces a conclusion reached earlier, which undermined the first interpretation of direction of fit. 13 Construing direction of fit as a relation between contents of propositional attitudes and the world, does not immediately entitle one to equate the contents of such attitudes With the goals mentioned in teleological explanations, even if we admit that such explanations are teleological. ARE MOTIVATING REASON EXPLANATIONS TELEOLOGICAL? For all that, we should ask whether motivating reason explanations are teleological, and if so, why they are. For their teleological character does have implications for the debate between Humeans and non-Humeans. Intentional explanations, including motivating reason explanations, will turn out to be teleological because of the role rationality constraints play in the attribution of attitudes. And their status as rationalizing explanations itself directly has implications for Smith's argument against the possibility of besires. As Davidson and Lewis emphasize in their accounts of radical interpretation, we can attribute propositional attitudes only if we rely on HUMEAN MOTIVATION AND HUMEAN RATIONALITY 45 rationality constraints. Smith should also accept this. As he emphasizes, the epistemology of belief and desire attribution is such that it suggests a strong connection between being in a mental state and being disposed to act in a particular way, given certain other beliefs and desires. Yet, since these dispositions issue in action only given that the person has certain other beliefs or desires, the actions leave the intentional states underdetermined. Rationality constraints fill the gap. If the correct account of a person's mental states is the one that makes her out to be most rational consistent with her behavior, the possible interpretations of her actions converge. David Lewis, for example, recommends that we employ two rationality constraints: the principle of charity, and the "rationalization principle."14 Charity requires that a person be represented as believing what she ought to believe and desiring what she ought to desire (in light of her experience); the rationalization principle requires that the beliefs and desires ascribed provide good reasons for her behavior. These constraints on the attribution of belief and desire to explain rational actions meets our criteria for teleological explanations. We can take rationality itself as the end, or "goal," postulated, and now display the state in question as necessary against the backdrop of the person's other beliefs, desires and actions, if the person is to remain rationalJ 5 And our expectation that the person will be rational is independent of inductive evidence that such action does occur. For our expectation rests on a theory of rationality which is itself(at least somewhat) independent of inductive support. Our actual practice of explaining actions in terms of motivating reasons, does seem to work in just this way. We ask why someone drank from the glass in front of her. We are told that she wanted to quench her thirst, believed that the glass in front of her held water, and that water would quench thirst. We find this a perfectly satisfactory explanation. On the proposed account we find it satisfactory because we believe that, all other considerations being equal, it is rational to drink what is in the glass, and somewhat irrational not to. Drinking the contents of the glass and the desire to do so which is part of intending to act are rationally necessary. 16 46 MARK VAN ROOJEN HOW DOES THE RATIONALIZING CHARACTER OF MOTIVATING REASON EXPLANATIONS CONSTRAIN THE THEORY OF MOTIVATION? The teleological explanation of behavior constrains the content of motivating reason explanations. Smith claims that this constraint forces us to be Humeans about motivation. I claim that it merely constrains us to adopt a motivational theory which allows states which normatively require other states, to motivate those states as well. 17 Thus, is we think having certain desires is rationally required if we have certain beliefs, we should think that such beliefs can motivate such desires. This leads to an interesting possibility: Whether we are Humeans about motivation might depend on whether we are Humeans about rationality. 18 But here I get ahead of myself. That motivating reason explanations are teleological, only because they are rationalizing explanations, has implications for the possibility of besires. Smith's Humean argument needed to rule out the possibility of besires in order to vindicate the claim that no cognitive states could by themselves motivate action. To do so he claimed that cases of weakness of will and accidie show that there is a gap between belief and motivation. Thus, providing such motivation cannot be part of the functional role of belief. But the phenomena cited do not warrant any such strong conclusion, even on a dispositional account of mental states. On a dispositional account, the functional roles which at least partly constitute various mental states, are defined by commitments of folk-psychology. 19 For, if the rationality constraints are to help with the epistemology of belief and desire, they must be used in the everyday interpretation of people's attitudes. Now, on any plausible view of human psychology, people are only for the most part rational. Common-sense psychology does not claim that the norms of rationality are observed without exception. If it did, common-sense psychology would not even have a chance as an explanatory theory of human action. For that claim is obviously false. Common sense claims only that conjunctions of certain states make other states or actions more likely in rational creatures. Furthermore, the generalizations of common sense psychology cover people in general. Thus a HUMEAN MOTIVATION AND HUMEAN RATIONALITY 47 state that is generally apt for a certain functional role, might not play that role in some individual. For example, believing that tomorrow is Tuesday, and believing that Wednesday follows Tuesday, will make it more likely that the person will come to believe that the following day is Wednesday. But it does not entail that the person will. Since people may be irrational, the constraints on attitude attribution may, in some cases, work against one another. No one description of the subject may meet all the constraints fully. Still that is as we would have expected, and a system of attitudes which comes near to meeting the constraints will be close enough to vindicate our expectations. As Lewis observes: I t . . . seems hard to deny t h a t . . , indeterminacy can arise because no solution fits all the constraints perfectly, and many different ways to strike a balance give many different compromise solutions. 2° In such cases we will be struck trying to strike the best balance that we can. Though we will attribute irrationality, we will try to attribute as little as possible. But the possibility remains that in some cases of irrationality we won't know exactly what to say. These points are important because they allow that a person can be in a certain functionally specified state and yet not be disposed in all of the ways that the theory says the state is apt to dispose one. This allows the anti-Humean about motivation to deny the charge that the presence of certain cognitive states entails desires, while maintaining that these states have more than one direction of fit. In other words, for the Humean's objection from weakness of will, accidie, and like cases to work, the tendencies that constitute the functional role of mental states have to be exceptionless laws. But such a requirement is too strong, for Humeans and anti-Humeans alike. The tendencies relied upon are not exceptionless laws of mental functioning, but generalizations meant to cover the normal case. What the dispositionalist construes as a state's direction of fit is its aptness for playing a certain role, even if the ability may not be manifested while both in that state, and in generally suitable conditions for the manifestation of its aptness for the role. 21 Thus, what makes a state a desire for an end is (in part) its aptness for causing a desire for the means, but this does not entail that a person desire the 48 MARK VAN ROOJEN means in every case. For if it did, we would have an a priori proof that a certain kind of irrationality could not occur. We all, Humeans and non-Humeans alike, have empirical reasons for rejecting that view. It is worth stressing that accepting the objection's presuppositions would commit the Humean to denying that a person could ever be in the state of desiring an end and not desiring the means. For the principle the Humean deployed against his opponent took the absence of an effect used to define the functional role of a given state as incompatible with a subjects' being in that state. If that principle is applied to the functional characteristics of desires that the Humean says c a n motivate, no person could irrationally neglect to take the best means to some end desired. On the strong principle that the absence of an effect shows the absence of the disposition, the Humean will have to deny that people are ever irrational about pursuing their desires. But few Humeans would accept a conclusion that strong and those who did would soon find themselves at odds with the empirical evidence. A more reasonable view will vindicate the commitment to this extent: All other things being equal, we are committed to ascribing the desire for the means where agents have a motivating reason as specified. But, all other things are not always equal, and we may be unwilling to give up other commitments which militate against ascribing that desire. If this is right, the anti-Humean can accept that we sometimes attribute an evaluative belief to a person without attributing a desire for the appropriate means as well. For that may just be because, despite our commitment to attributing that desire when all other things are equal, in this case all other things are not equal. The defender of the Humean theory of motivation can, of course, regroup and try to show that we have to such commitment. But the most likely way of accomplishing that will, I think, turn most crucially on what the account of rationality tells us to believe. Why is that? Well, recall that the method of radical interpretation has us attributing to people just those beliefs and desires we think it would be reasonable to have in those circumstances, and which would be most reasonable given their behavior. If we think that their rationality requires that people who think an option "good", "practically necessary" or whatever, should be motivated to bring about that option, and thus motivated to desire the HUMEAN MOTIVATION AND HUMEAN RATIONALITY 49 means, then we will also have a commitment to attributing those desires. As we have already seen, that commitment will be defeasible, but that will not differentiate it in kind from the commitment where desires are concerned. Perhaps, the Humean may suggest, the commitment is not as strong as in the case of ordinary desires together with means-end beliefs. Very likely for many evaluative beliefs that will clearly be so. But what extent we find it so for any particular kind of evaluative belief, will depend precisely on how unreasonable we think having the belief but not the motivation is. And even if the commitment with evaluative beliefs is not as strong, provided it is strong enough, the contrast with desires will be one of degree and not of kind. Defenders of the Humean view must explain why this difference in degree matters enough to disqualify evaluative beliefs from counting as sources of motivation when desires, with an only somewhat stronger connection, do count. THE INELIMINABILITY OF NORMATIVE RATIONALITY The argument so far has proceeded from an account of our practice of attitude attribution. Thus, its upshot is at least epistemological: our theory of normative rationality will play a heuristic role in constructing reason explanations. Thus, I claim, to refute an anti-Humean account of motivation the Humean must show that an instrument conception of rationality is superior to a more substantial conception that postulates normative connections between beliefs and desires. Two interpretations of this result are possible. It might follow from the kind of explanation that intentional explanations are, so that they need the normative component for some of their explanatory force. Or, this might be a contingent result of our limitations at predicting how people will behave. On this latter view, the problem is just that we are less than perfect predictors of human behavior. We need the theory of rationality to tell us what pattern of behavior to expect from people, but if we had some other way of getting at that pattern we could do without the theory of rationality. The fact that people are not perfectly rational just underscores the fact that our dependence on the theory of rationality is a condition to be 50 MARK VAN ROOJEN lamented. If we could only get hold of a theory that got the empirical story more nearly right, we could do an even better job of explanation. While such a position should strictly be seen as denying both the Humean and the anti-Humean views, the Humean might try to seize on it as showing that his view was more right than the anti-Humeans'. For if people turned out to violate instrumental norms less often than the more substantive anti-Humean norms, the Humean might claim that his theory was closer to the empirical truth. We might call an advocate of this position a quasi-Humean. Now the quasi-Humean has little chance of vindicating a truly Humean theory. For an this view, empirical'science, not a partially a priori interpretive theory, should determine whether the Humean or anti-Humean theory is more nearly correct. And the empirical evidence shows that people are sometimes not motivated to take the most efficient means to what they desire, and sometimes do not take any means at all. Thus, empirical research might yield some new category to supersede that of desire, the members of which do always motivate. The quasihumean will have little to say against this, since this new category will capture the empirical regularities even better than the current categories of belief and desire. Since the rationality constraints were just handy ways to get on to these regularities, and the new concepts have a better grip on the actual regularities, the old, rationality-linked, concepts of belief and desire can be abandoned. I think, however, that regarding rationality constraints as essential to motivational explanations is more tenable, though both attitudes are compatible with acknowledging that our motivational theory must presuppose an account of normative rationality. Reflection on the following examples may make clear why I think so. Suppose we come across Davidson's paint drinker and are trying to understand his behavior. 22 We are offered the explanation that he has always had a yen to drink a can of paint. Does this explain the action? I agree with Humeans that it does, albeit minimally. For it displays the drinking of this can of paint here now, as a means to satisfying the general desire to drink a can of paint. But, I claim that its explanatory force rests on our accepting a norm something like this: If one desires to ~ , and one can • by ~b-ing them, all other things being equal, (desire HUMEAN MOTIVATION AND HUMEAN RATIONALITY 51 to) ~b. We can see that not just any desire can play the explanatory role that the yen to drink a can of paint does in this explanation. Citing a yen to stand on one's head would not explain drinking the can of paint. That's because the explanation in the first case depends on our thinking desiring the means for what one desires generally makes sense. But we don't think that desiring to drink a particular can of paint makes sense if one wants to stand on one's head. 23 A Humean who accepts my account of motivating reason explanations can accept all of this so far. And even those who disagree with me about the essential role of normative rationality, can accept much of this. They will agree that no just any desire can explain the action. And they can agree that there are rational norms, of the kind I say ground the explanatory force of the motivational explanation. 24 They will disagree only with my claim that these norms play the essential explanatory role I claim. We will need to do more to decide the issue. We can make the explanation of the paint drinker's action even better. We might further explain why he always had a yen to drink a can of paint. Perhaps his yen stems from seeing such an action as funny. If we can attribute this belief to the agent, we will have a better explanation than before. And, this is not just because we have gone up one more level in attributing motivating reasons, though we have done that. It is also because we can see humor as a goal a person might intelligibly pursue. We have made our explanation a better teleological explanation, since it now shows the agent as aimed at something we have even more reason to expect a person to aim at. Someone resistant to my hypothesis can still accept most of this, by saying that our expectation rests on noticing that many people do like to be funny. Thus, our expectation does not rest on any commitment to the rationality of wanting to be funny, but just on the general observation that lots of people are this way. Where I look to a normative principle to do some of the explanatory work, my opponent can try to make do with an empirical regularity. There is nothing inconsistent in this. But in some cases our expectations outstrip our empirical evidence for such generalizations. Sometimes our commitment to interpreting agents as rational goes beyond the empirical evidence and seems to rely on some a priori principle. In fact, our attribution of even Humean 52 MARK VAN ROOJEN motivating reasons sometimes seems to depend on just such a maneuver using the norms of practical reasoning regarding ends and means that Humeans accept. We can imagine a chain of means-ends reasoning that is too complicated for many people to successfully complete. And we can imagine an accompanying desire, which with that chain of meansend reasoning would result in action. If someone has that desire, and engages in the action recommended by that means-ends reasoning, it will be natural to view that desire and that reasoning as the motivation behind the action, particularly if the agent consciously engaged in that reasoning process. But, empirical evidence suggests that people will not generally be motivated in such circumstances. So an empirically justified expectation that they will cannot be the one grounding the seemingly teleological explanation of this motivation. Can we identify an expectation that we are entitled to? The only candidate I can think of is something like: I f an agent acts in line with the canons of rationality, then she will act in this manner. The bit of reasoning and desiring to be explained is conducive to acting in the required manner. Thus, the presence of the motivation is explained by its necessity in light of the canons of rationality. I admit to being a bit surprised by this conclusion. For I can find no analogous cases of normative principles of this kind grounding teleological explanations where agency is not involved. But I can't see any way around using this as the principle we appeal to. For it just seems fight to say that we can explain a person's action where she reasons correctly even if that reasoning was so hard we expect very few people to be able to do it. Normative expectations seem to fill in where we do not have a statistical expectation. And this observation leads me to suspect that motivating reason explanations are a special kind of teleological explanation in so far as they essentially involve normative considerationsY The suspicion is reinforced by reflection on another example. We might notice that some person exhibits a regularity of the following kind: Whenever the person desires to stand on his head, he desires to drink the nearest can of paint. Enough observational evidence could establish this regularity with some degree of certainty. Perhaps we even get a story about how the person was conditioned to have this occur. Yet in some sense we remain baffled by the person's desire, even when HUMEAN MOTIVATIONAND HUMEANRATIONALITY 53 we see it as an instance of a regularity. In fact, it is tempting to say that while we might have an explanation of the causal genesis of the desire to drink paint, we have not been given a motivating explanation at all. If this is right, we now have some grounds for resisting the interpretation of motivating reason explanations as merely regularizing explanations, whose explanatory force comes from displaying the action as necessary if some empirical regularity is to obtain. Rationalizing explanations have force even where they outstrip the empirical regularities. And motivating reason explanations are such rationalizing explanations. 26 A decision on what kinds of states have motivational force will tum on the prior question of what kinds of states make certain kinds of motivation and action rational. Thus, the question of which kinds of beliefs and desires make which kinds of motivation and action reasonable, must receive an independence answer, without the aid of a theory of motivation. The result is that the Humean Theory of Motivation no longer provides us with an independent way in to deciding the truth of relativism. It may be true, but we can affirm that it is only if we accept a theory of rationality (the instrumental theory) which would already confirm a form of relativism on its own. There might of course be various ways of approaching the more fundamental issue of the acceptability of this instrument account. If, as I believe, no reductive account of rationality is in the offing, the best way to approach the fundamental issue is head on. NOTES * I owe many thanks to Gil Harman and Michael Smith for extensive discussion and comments on previous drafts of this paper. I also owe thanks for comments and discussion to Robert Audi, Sarah Buss, A1 Casullo, Josh Cohen, Jamie Dreier, Berys Gault, Margaret Gilbert, Jennifer Hatey, Harry Ide, Mark Johnston, Mark Kalderon, Cliff Landesman, Joe Mendola, Michaelis Michael, Dick Moran, Alexander Nehamas, Pauline O'Connor, Lee Overton, Peter Railton, Charles Sayward, Scott Sehon, and Lyle Zynda, as well as an anonymous referee for this journal. Drafts of this paper were presented to audiences at Princeton University in 1990, and to the University of Nebraska-Lincolnin 1991. I thank the people present on each of those occasions for many useful comments. i Nagel, ThePossibilityofAltruism(Princeton, PrincetonUniversityPress, 1970). 54 MARK VAN ROOJEN 2 Michael Smith, "The Humean Theory of Motivation," Mind (1987), p. 36. 3 This is in fact the reading Smith intends. For the issue between the Humean and the anti-Humean is supposed to be about whether or not cognitive states by themselves can motivate action. For more on this see, Pettit, "Humeans, Anti-Humeans, and Motivation," Mind (1987), p. 31, and Smith, "On Humeans, Anti-Humeans, and Motivation: A Reply to Pettit", Mind (1988), pp. 589-594. 4 I agree with R. Jay Wallace in his survey of the relevant literature that Smith's argument captures the most common concerns of those who accept a Humean view. See, Wallace, "How to Argue About Practical Reason," Mind (July 1990), pp. 355-385. 5 Smith, "The Humean Theory of Motivation," p. 55. 6 One might think that this way of pursuing the question presupposes the truth of functionalism, a dubious procedure when the task is to assess an argument critical of, among others, John McDowell, who denies the truth of functionalism. (See, for example, "Functionalism and Anomalous Monism," in LePore and McLaughlin, eds., Actions and Events: Perspectives on the Philosophy of Donald Davidson (Oxford, OUP, 1985), pp. 387-398.) Two comments are appropriate. First, since I end up vindicating McDowell, my argument is stronger for not starting out by assuming the falsity of functionalism. Second, McDowell seems most concerned to counter functionalism as a reductive thesis. Accepting some sort of dispositional account need not commit one to a reductive account of the mental. One can think that the causal relations are partially constitutive of what it is for an actual mental state to be the state that it is, without thinking that what it is to be a particular kind of mental state can be reduced to only such causal relations. 7 The term "besire" comes from James Altham's "The Legacy of Emotivism," in MacDonald and Wright, eds., Fact, Science and Morality: Essays on A. J. Ayer's Language, Truth and Logic (Oxford, Basil Blackwell, 1986). In "The Humean Theory of Motivation," Smith uses "quasi-belief" for a state with both directions of fit. In his reply to Pettit he takes over Pettit's term "desire=belief". 8 Smith, "The Humean Theory of Motivation," p. 56. 9 Smith, "On Humeans, Anti-Humeans, etc.," p. 594, footnote to Michael Stocker, "Desiring the Bad: An Essay in Moral Psychology," Journal of Philosophy (1979), omitted. Smith is not the only philosopher who thinks that such examples support Humean conclusions. S tocker himself presented his examples as reasons to include various affective states within any adequate theory of motivation. And Bernard Williams uses the possibility that a person might at once recognize some need while remaining uninterested in fulfilling that need, even after deliberation, to show that needs are not "internal" reasons, where internal reasons are reasons which may potentially motivate the agent. See Williams, "Intemal and Extemal Reasons", in Moral Luck (Cambridge, Cambridge University Press, 1981), p. 105. 10 "The Humean Theory of Motivation," p. 44. 11 Isn't taking Smith's talk of "goals" this literally unfair. In fact, don't I go on to use the term myself. Yes, that would be unfair if the literal sense was not what Smith HUMEAN MOTIVATIONAND HUMEANRATIONALITY 55 intended. The problem is that he says that "... having a goal just is desiring" ("The Humean Theory of Motivation," p. 37). Perhaps this was an infelicitous remark, and we agree on more than it seems. 12 The account here owes much to Philip Pettit, "Broad-minded Explanation and Psychology," in Pettit and McDowell, eds., Subject, Thought and Context (Oxford, Clarendon Press, 1986). Pettit, in turn traces the genealogy of his account through Charles Taylor, The Explanation of Behavior (London, Routledge & Kegan Paul, 1964), Larry Wright, TeleologicaIExplanation (London, University of California Press, 1976), and G. A. Cohen Karl Marx's Theory of History (Oxford, Clarendon Press, 1978). We may with to follow Pettit even further, and fill out the counterfactual dependence causally."... [A]n event E is teleological explained as a means to a goal G if it is caused to occur by the circumstances of an E-type event's being necessary for the goal." But for current purposes we can remain agnostic. 13 This account of teleological explanations is contentious. But even if that account is wrong, my main conclusion about motivation still holds. That does not depend on any particular account of teleological explanation. It requires only the reason for thinking that intentional explanation are teleological are also reasons for thinking that rationality must have a constitutive role in the attribution of beliefs and desires, and thus on the structure of motivating reason explanations. 14 Lewis, "Radical Interpretation," in PhilosophicaIPapers, Volume 1 (Oxford, OUP, 1983), pp. 108-121. 15 Some may think it funny that I postulate acting and believing rationally as the end which is presupposed in the teleological explanation of action. I suspect that part of the oddness comes from having already assimilated all teleological explanations to intentional explanations on a Humean model. That's part of why I bicker with Smith's characterization of teleological explanations in terms of "having a goal." 16 Onemight wonder about an explanation of an action of the form: "Shedid it because she wanted to." I think such explanations qualify as teleological, because the action is rationally necessary, given the desire for an action like it, all other things equal. Thus, the action itself fulfills an explanation raised by the theory of rationality. 17 I have presented the constraints as entailed by a conception of motivating reason explanation as a species of teleological explanation. But, Lewis's account of interpretation gives us independent reason to endorse such constraints. For we decided that such explanations are teleological only because given the methodology as described by Lewis, reason explanations meet certain rationality constraints. And these constraints are why our attribution of motivating reasons meet the requirements of teleological explanation. 18 Christine Korsgaard uses a somewhat different argument to support this point in "Skepticism About Practical Reason," Journal of Philosophy (1986), pp. 5-25. 19 Lewis, "Mad Pain and Martian Pain," in Philosophical Papers, Volume I (Oxford, OUP, 1983)p. 124. 20 Lewis, "Radical Interpretation," p. 118. 56 MARK VAN ROOJEN 21 HOW could that be? Well, one way might be if the state's aptness for producing a certain result were statistical. I should admit to a certain tension in my treamaent of how to characterize the relations between beliefs in what follows. I start out with Lewis's statistical/causal interpretation of the relations, but wind up with something that allows explanation even where there are only normative relations that go beyond the empirical regularities backing them up. In so far as the move away from the causal/statistical model is motivated by examples that are supposed to provide arguments, I think the shifts can be seen as refinements of an approach to the attribution of attitudes. And the shift does not abandon the causal/statistical view, entirely, for while I do allow some explanation that goes beyond those regularities, this still must be against the background of empirical regularities that are, for the most part, as the norms indicate. Otherwise we would abandon the attempt to interpret the system in question intentionally. 22 The example is taken from "Actions, Reasons, and Causes," in Essays on Action and Events (Oxford, Clarendon Press, 1983), pp. 3-19. 23 I am not claiming that together with the appropriate beliefs it could not form such an explanation. But in such cases the appropriate beliefs would be of a kind that would enable us to locate normative requirements in the background of the explanation. 24 In fact, Michael Smith has himself argued that the Humean should accept norms of the kind I am postulating here. See "Reason and Desire," Proceedings o f the Aristotelian Society (1987). 25 It is reasonable to worry about this kind of explanation, particularly given that it looks like coming to the correct conclusion may have been partly a matter of luck. I'm putting some weight here on my sense that we would find it explanatory to cite the correctness of the result in explaining the action. If this argument can not bear the weight by itself, the argument immediately following should carry some of the load. 26 The reason I think I need to pursue this line of argument is connected to my worries about the account of teleological explanation offered earlier. It may be the case with teleological explanations in biology that the selection mechanism on which they rely can be explained completely in terms of certain empirical causal regularities. In that case, while teleological explanation would be respectable enough, there would seem to be no principled objection to replacing the teleological explanation with an explanation relying only on the causal regularities underlying the selection process. Even if this is right for natural selection, my argument here is intended to help us resist such a move with respect to intentional explanation. REFERENCES Altham, J. (1986) "The Legacy of Emotivism," in MacDonald and Wright, eds., Fact, Science and Morality: Essays on A. J. Ayer's Language, Truth and Logic (Oxford: Basil Blackwell), pp. 275-88. HUMEANMOTIVATIONAND HUMEANRATIONALITY 57 Cohen, G. A. (1978) Karl Marx's Theory of History (Oxford, Clarendon Press). Davidson, D. (1983) "Actions, Reasons, and Causes,"in Essays on Action and Events (Oxford, Clarendon Press), pp. 3-19. Korsgaard, C. (1986) "Skepticism About Practical Reason," Journal of Philosophy 83, pp. 5-25. Lewis, D. K. (1983) "Radical Interpretation," in Philosophical Papers, Volume 1 (Oxford, OUP), pp. 108-121. Lewis, D. K. (1983) "Mad Pain and Martian Pain," in Philosophical Papers, Volume I (Oxford, OUP), pp. 122-32. McDowell, J. (1985) "Functionalism and Anomalous Monism," in LePore and McLaughlin, eds., Actions and Events: Perspectives on the Philosophy of DonaM Davidson (Oxford, OUP), pp. 387-398. Nagel, T. (1970) The Possibility of Altruism (Princeton, Princeton University Press). Pettit, E (1986) "Broad-minded Explanation and Psychology," in Pettit and McDowell, eds., Subject, Thought and Context (Oxford, Clarendon Press), pp. 17-58. Pettit, P. (1987) "Humeans, Anti-Humeans, and Motivation," Mind 96, pp. 530-33. Smith, M. (1987) "The Humean Theory of Motivation," Mind 96, pp. 589-95. Smith, M. (1987-8) "Reason and Desire," Proceedings of the AristoteIian Society 88, pp. 243-58. Smith, M. (1988) "On Humeans, Anti-Humeans, and Motivation: A Reply to Pettit," Mind 97, pp. 589-594. Stocker, M. (1979) "Desiring the Bad: An Essay in Moral Psychology," Journal of Philosophy 76, pp. 738-53. Taylor, C. (1964) The Explanation of Behavior (London, Routledge & Kegan Paul). Wallace, R. J. (1990) "How to Argue About Practical Reason,"Mind 99 (July), pp. 355385. Williams, B. (1981) "Internal and External Reasons," in Moral Luck (Cambridge, Cambridge University Press), pp. 101-13. Wright, L. (1976) Teleological Explanation (London, University of California Press). Department o f Philosophy Brown University B o x 1918 Providence, R I 02912 USA