The Guise of the Good

  1. 1.  New York University

The agent portrayed in much philosophy of action is, let’s face it, a square. He does nothing intentionally unless he regards it or its consequences as desirable. The reason is that he acts intentionally only when he acts out of a desire for some anticipated outcome; and in desiring that outcome, he must regard it as having some value.[2] All of his intentional actions are therefore directed at outcomes regarded sub specie boni: under the guise of the good.[3]

This agent is conceived as being capable of intentional action—and hence as being an agent—only by virtue of being a pursuer of value. I want to question whether this conception of agency can be correct. Surely, so general a capacity as agency cannot entail so narrow a cast of mind. Our moral psychology has characterized, not the generic agent, but a particular species of agent, and a particularly bland species of agent, at that. It has characterized the earnest agent while ignoring those agents who are disaffected, refractory, silly, satanic, or punk. I hope for a moral psychology that has room for the whole motley crew.

I shall begin by examining why some philosophers have thought that the attitudes motivating intentional actions involve judgments of value. I shall then argue that their conception of these attitudes is incorrect. Finally, I shall argue that practical reason should not be conceived as a faculty for pursuing value.

One source of the view that intentional actions are aimed at the good has been a desire on the part of moral psychologists to reconcile two seemingly incompatible stories about how human action originates. These might be called the story of motivation and the story of rational guidance.[4]

The story of motivation says that an action is caused by a desire for some outcome and a belief that the action will promote it. The agent wants to know the time, for example, and believes that looking at his watch will result in his knowing the time; and he consequently looks at his watch.[5] The desire and belief cited in this story are conceived as propositional attitudes. That is, each is thought to consist in the agent’s grasping and being somehow disposed toward a proposition. He is, so to speak, desirous toward the proposition ‘I know the time’ and credent toward the proposition ‘Looking at my watch will result in my knowing the time’. These attitudes combine to cause a new attitude—a desire toward the proposition ‘I look at my watch’. And since looking at his watch is something that the agent can just do if he wants, this desire causes him to act.

The story of rational guidance tells how an agent acts for a reason. According to this story, a reason for acting is a proposition whose truth would reflect well on, count in favor of, recommend, or in some other sense justify an action.[6] A reason for performing an action exists so long as a proposition justifying the action is true.[7] But an agent cannot act for this reason unless he has mental access to it—unless he believes the proposition or at least grasps it in some related fashion. And even if he has appropriately grasped the reason, and is therefore in a position to act for it, he doesn’t ultimately act for the reason unless his grasp of it results in his being influenced or guided by its justifying force. An agent acts for a reason, then, when the action-justifying character of a proposition prompts his action via his grasp of that proposition.[8]

The apparent discrepancy between these stories lies in the relation posited between the agent’s action and the propositional objects of his attitudes. In the story of motivation, the objects of the agent’s attitudes are propositions that do not in themselves justify his action. They do, of course, help to determine which action he is moved to perform. The agent would not be moved to look at his watch if he didn’t want, in particular, to know the time and if he didn’t believe, in particular, that looking at his watch would result in his knowing the time. Yet the content of these attitudes doesn’t in any way reflect well on, count in favor of, or otherwise justify looking at his watch. Indeed, the propositions ‘I know the time’ and ‘Looking at my watch will result in my knowing the time’ do not reflect more favorably on watch-consulting behavior than on watch-ignoring behavior. When the agent’s attitudes toward these propositions move him to consult his watch, he is not responding to any action-justifying property of the propositions themselves; he is simply manifesting the valence of his attitude toward the former proposition. Because the agent has a desire toward ‘I know the time’, he is moved to look at his watch; whereas if he had an aversion toward the same proposition, he would be moved to ignore his watch instead. The content of his attitudes is in itself neutral between these alternatives.

The action performed in the story of rational guidance is also determined in part by the nature of the agent’s attitudes and not merely by their content. When the protagonist of this story acts for a reason, he acts partly because of grasping the reason in an attitude something like belief. But more than the nature of the agent’s attitude toward his reason must work in favor of his action. When an agent acts for a reason, he acts not only because his attitude toward the reason is more like belief than disbelief but also because the proposition involved militates in favor of his action rather than against it. The agent’s attitudes are thus conceived as having propositional objects that intrinsically favor a particular action, and their favoring the action is conceived as crucial to their behavioral influence.

This aspect of rational guidance is what makes rationally guided behavior, as we conceive it, comparable to the conclusion of an inference. The premises of an inference are propositions whose truth guarantees or makes probable the truth of the conclusion; and in this sense they favor believing the conclusion solely by virtue of their content, antecedently to any attitude in which one might fix them. In order for a particular set of premises to become one’s reasons for drawing a conclusion, one must somehow be influenced, in grasping them, by their antecedently favorable relation to the conclusion. Similarly, in order for a particular reason to become one’s reason for performing an action, one must be influenced by its bearing favorably on that action. To be motivated by a desire, by contrast, is to be guided by attitudes toward propositions that do not in themselves favor anything.


This fundamental difference between motivation and rational guidance wouldn’t necessarily render the two stories incompatible if neither purported to be the complete explanation of an action. One and the same action could be due to a confluence of motivation and rational guidance. Yet philosophers have tended to interpret each story as purporting to be the whole story; and they have therefore assumed that the apparent discrepancy between these stories has to be removed. They have sought to remove the discrepancy by slightly retelling one story or the other.

Noncognitivists, for example, retell the story of rational guidance in such a way that it collapses into the story of motivation. According to their version of the story, the agent’s reason for acting is the proposition ‘Looking at my watch will result in my knowing the time’, a proposition that recommends looking at his watch, not by virtue of its content, but rather because his belief in it inclines him to look at his watch, given his desire to know the time. Noncognitivists thus deny that the propositions constituting an agent’s reasons for acting are intrinsically favorable to his action. Propositions recommend acting, they think, only in relation to desires that lend motivational force to the agent’s belief in those propositions.[9]

Some noncognitivists may resent the suggestion that they are hereby retelling or revising anything—that there ever was any other story of rational guidance than theirs. But the noncognitivist story diverges from the commonsense story of rational guidance in one important respect: it reverses the order of explanation between justificatory and motivational force. In the commonsense story, the agent is moved toward an action because his reasons justify it; whereas in the noncognitivist story, his reasons justify the action in virtue of moving him toward it. The noncognitivist thus treats motivation as a constituent rather than an effect of justification.[10]

I shall not consider here whether this departure from the commonsense story of rational guidance is defensible. What suffices for my purposes is that some philosophers have preferred to avoid it, by adopting a different strategy for reconciling the two stories of human action. Rather than characterize rational guidance noncognitively, so that it collapses into motivation, they characterize motivation cognitively, so that it amounts to something like rational guidance; and they thereby introduce the evaluative conception of agency that interests me.


Proponents of this alternative strategy portray motivation itself as an inference, governed in part by action-justifying content to be found in the motivating attitudes. To this end, they incorporate the valence of desire into its content, by describing desire, not as a favorable attitude toward the representation of some outcome, but rather as an attitude toward a favorable representation of the outcome. The agent who wants to know the time is said, not to be favorably disposed toward ‘I know the time’, but rather to accept a proposition such as ‘My knowing the time would be good’. The content of this attitude and the content of the agent’s belief, ‘Consulting my watch would result in my knowing the time’, are sufficient to justify the conclusion ‘Consulting my watch would be good’. And the agent’s accepting this favorable representation of consulting his watch is now conceived as constituting a desire to consult his watch. Hence his transition from a desire to know the time to a desire to consult his watch appears to be dictated, in the fashion of an inference, by a privileged logical relation between the contents of the attitudes involved.

The leading contemporary proponent of this latter strategy is Donald Davidson. Consider the following passage, in which Davidson is discussing an agent who is moved to add sage to his stew by a desire to improve the taste:[11]

[L]et us suppose [the agent] wants to improve the taste of the stew. But what is the corresponding premise? If we were to look for the proposition toward which his desire is directed, the proposition he wants true, it would be something like: He does something that improves the taste of the stew (more briefly: He improves the taste of the stew). This cannot be his premise, however, for nothing interesting follows from the two premises: Adding sage to the stew will improve its taste, and the agent improves the taste of the stew. The trouble is that the attitude of approval which the agent has toward the second proposition has been left out. It cannot be put back in by making the premise ‘The agent wants to improve the taste of the stew’: we do not want a description of his desire, but an expression of it in a form in which he might use it to arrive at an action. The natural expression of his desire is, it seems to me, evaluative in form; for example, ‘It is desirable to improve the taste of the stew,’ or, ‘I ought to improve the taste of the stew’. We may suppose different pro attitudes are expressed with other evaluative words in place of ‘desirable’.

In this passage Davidson subjects the story of motivation to the retelling that I have just described. He demands that the outcome of motivation—the act of adding sage—be justified by some propositional content of the agent’s attitudes, as if it were a conclusion following from premises. And he obtains the required content by incorporating the valence of the agent’s attitude toward ‘I improve the taste’ into a new proposition: ‘Improving the taste is desirable’. The story of motivation is thus transformed into the story of an inference, in which the agent is under genuinely rational guidance.


Here, then, is one way in which rational agency comes to be conceived as a capacity for pursuing value. Desires are conceived as value judgments, with intrinsic justificatory force, so that the desire motivating an agent can be identified with the reason guiding him. The result is that all actions performed for reasons are conceived as arising from favorable value judgments, and hence as being aimed at the good.

This reconciliation of motivation and rational guidance comes under pressure from two different directions. If the cognitivist seriously means to characterize desire as an attitude toward an evaluative proposition, then he implies that the capacity to desire requires the possession of evaluative concepts. Yet a young child can want things long before it has acquired the concept of their being worth wanting, or desirable. Surely, the concept of desirability—of something’s being a correct or fitting object of desire—is a concept that children need to be taught. And how would one teach this concept to a child if not by disciplining its antecedently existing desires?[12]

This problem may explain Davidson’s apparent efforts to avoid saying that evaluations serve as the contents or propositional objects of desire. Davidson often favors alternative formulations, as in the passage quoted above, where he says that the relevant evaluation is “the natural expression of [the agent’s] desire” rather than its propositional object.

But this qualification leaves the cognitivist open to a different objection, since it seems to undermine his attempt to reconcile the stories of motivation and rational guidance. According to the latter story, acting for a reason entails being influenced by the force of a mentally grasped justification of one’s action. According to Davidson’s qualified formulation, however, a proposition that’s essential to the justification of the action—namely, the proposition that the action’s expected consequences are desirable—is merely a proposition that would naturally be used to express the agent’s desire. And the agent can be moved by his desire without either being able to express it or grasping the proposition with which it would naturally be expressed. He can therefore satisfy Davidson’s story of motivation without having mentally accessed anything that justifies his action. Hence the resulting story of motivation no longer corresponds to the story of rational guidance.[13]

These two objections seem to leave no room for the cognitivist strategy. But they do not rule out a sophisticated version of cognitivism—a version that is suggested, in any case, by an important feature of propositional attitudes in general. The feature in question is the so-called direction of fit that distinguishes conative attitudes such as desire from cognitive attitudes such as belief.[14] As we shall see, reflection on this feature naturally leads to a version of cognitivism that escapes the foregoing objections.


The term ‘direction of fit’ refers to the two different ways in which attitudes can relate propositions to the world. In cognitive attitudes, a proposition is grasped as patterned after the world; whereas in conative attitudes, a proposition is grasped as a pattern for the world to follow. The propositional object of desire is regarded not as fact—not, that is, as factum, having been brought about—but rather as faciendum, to be brought about; it’s regarded not as true but as to be made true.

There is a temptation to think that regarding something as to be brought about or made true is tantamount to holding a value judgment about it. Perhaps, then, when philosophers say that to want something is to regard it as good or desirable, they are thinking of the attitude’s direction of fit—of the distinctive way in which a proposition is regarded when it’s the object of desire rather than belief.[15]

I shall argue presently that the use of an evaluative term like ‘good’ to express desire’s direction of fit is a potential source of confusion. For the moment, however, I shall adopt that usage, in order to examine precisely what it might mean and where it might lead.

As for the meaning of this usage, note that even if desiring something entails making a value judgment about it, by regarding it as good, this attitude qualifies as a value judgment in only a rather unusual sense of the phrase, a sense corresponding to that in which a belief might be called a “truth judgment”. The desire that p is here conceived as a value judgment in the sense that it involves regarding p as to be brought about and hence, supposedly, as good, just as the belief that p involves regarding p as true. But to say that belief in p involves regarding p as true is not to say that it consists in a judgment whose object is the proposition ‘p is true’. That way lies a vicious regress of propositional attitudes. Similarly, to say that the desire involves regarding p as good is not to say that it consists in a judgment with an evaluative proposition as its object.

Expressions like ‘regarding . . . as true’ and ‘regarding . . . as good’ are intended to describe belief and desire in a way that elucidates the difference in their directions of fit. Because we conceive of belief and desire as alike in being attitudes toward propositions, and as differing in their treatment of the fit between propositions and the world, we unavoidably describe them with a common attitudinal verb (‘regarding’) and different predicate adjectives (‘as true’, ‘as good’). But this construction—attitudinal verb plus differentiating predicate—must not be interpreted as invoking a further attitude directed toward a proposition containing that predicate. The desire that p is not to be analyzed as an attitude toward the proposition that p is good; it must be analyzed as an attitude toward p as good.[16]


Even so, the resulting conception of desire does seem to allow for a reconciliation between the stories of motivation and rational guidance. For although the desire that p doesn’t entail grasping a proposition that justifies action conducive to p, the desire itself may appear to constitute an attitude that justifies such action, if it consists in regarding p as good.

Once we recognize that a propositional attitude must be characterized, not only by the proposition that embodies its content, but also by a predicate expressing how that proposition is regarded—that is, whether it’s regarded as factum or faciendum, as true or to be made true—we are less inclined to insist that the justificatory force influencing reason-guided behavior be lodged in the propositional objects of the agent’s attitudes. When the valence of a desire that p is represented by the expression ‘regarding p as good’, valence takes on the form of a content-like phenomenon and begins to seem like a potential bearer of justificatory force. For even if no action is justified by the fact that p, some action might well be justified by the faciendum that p—by p’s being something to be brought about.

Thus, if an attitude combines the propositional object p with a direction of fit expressible by the predicate ‘good’, then it would seem to harbor justificatory force—not in its propositional object alone but rather in the combination of its propositional object and its direction of fit. And the agent can be imagined as having mental access, not only to the propositions that he grasps in various attitudes, but also to the attitudes’ direction of fit, as expressed by their constitutive predicates.

We may therefore be inclined to revise the story of rational guidance, by replacing its references to the agent’s grasp of action-justifying propositions with references to his action-justifying attitudes. We might say, for example, that an agent is mentally in touch with a justification for looking at his watch not only if he believes that knowing the time would be good but also if he regards it as good that he know the time—an attitude that supposedly constitutes a desire. We might also say that being guided by a desire’s direction of fit entails being guided by its evaluative aspect, which lends the desire its mentally accessible justificatory force. We might then conclude that being motivated by a desire can amount to acting for a reason.[17]


This version of cognitivism, like the previous version, implies that every action that’s motivated by a desire—or, equivalently, performed for a reason—is guided by some favorable value judgment, and hence that intentional action is always aimed at the good. The judgment involved is no longer conceived as an attitude toward an evaluative proposition; but it is still conceived as having the recommending force of an evaluation, so that it can serve as the agent’s reason for acting.

I believe that this version of cognitivism is an improvement on its predecessor, but that it is not ultimately more successful. For even if desiring something entails regarding it as good in some sense, regarding something as good in that sense does not in fact amount to making a value judgment about it. An agent’s motivating desire consequently lacks the justificatory force that the cognitivist attributes to it for the sake of identifying it with the agent’s reason for acting. The cognitivist strategy for reconciling motivation and rational guidance, and the resulting conception of intentional action as aimed at the good, thus turn out to rest on a mistake.

Before I attempt to demonstrate this mistake, however, I had better define more clearly what sort of mistake it is. When I deny that desire has the justificatory force with which it is credited by the cognitivist, I am not necessarily denying that having a desire provides one with reason for acting. It often does. Yet the cognitivist doesn’t merely claim that desire provides reason for acting; he claims that being moved by a desire amounts to acting for a reason. And this claim implies, as we have seen, that to fall under a desire’s motivational influence is to fall under the rational influence of a mentally grasped justification.

Yet one can agree that having a desire—and a fortiori being moved by a desire—sometimes entail the existence of a reason for acting, while denying that they entail being in the appropriate mental rapport with that reason. Suppose, for example, that the reason generated by a desire is the fact that one has the desire, a fact to which one can be utterly oblivious even as the desire moves one to act. In that case, one can have a desire and be moved by that desire without having grasped the reason that it generates; and so one can have and be moved by a desire without being in a position to be guided by the associated reason for acting.[18]

How the fact that one has a desire might justify action, and how an agent might be influenced by this justification, are questions that lie beyond the scope of this chapter.[19] I mention this view of the matter only for the sake of distinguishing the sense in which I concede that a motivating desire justifies action from the sense in which I deny it. I concede that desire can justify action objectively, by making true a proposition that could guide one’s actions if one gained appropriate access to it; but I deny that desire justifies action subjectively, by constituting an evaluative attitude whose justificatory force is already available to guide one’s actions. I thus deny that desire amounts to an evaluation, and that motivation consequently amounts to rational guidance, in the sense proposed by the cognitivist.[20]


The cognitivist thinks that desire provides a mentally accessible justification for acting because it harbors justificatory force in its direction of fit, as expressed by its constitutive predicate, which he takes to be a term of evaluation. But why would one think that the constitutive predicate of desire was ‘good’ or ‘desirable’ or some other evaluative term?

One reason, I suspect, is a tendency to psychologize various extrinsic descriptions of mental states. For example, to desire something is to be disposed toward it in a way that would be appropriate if the thing were good; and a person who desires something can therefore be said to regard or treat it as if it were good or, more concisely, as good. But this description of the person’s attitude, which cites a purely extrinsic fact about it, should not be mistaken as expressing a psychological aspect of the attitude itself. That someone’s attitude would be appropriate if its object were good is not something to which he has mental access simply by virtue of having the attitude; and so it’s not something whose justificatory force is necessarily available to guide him. Treating something as good in this sense is no more a value judgment than treating someone like dirt is a soil judgment.

Similarly, a person who desires something can be said to find it attractive, but this description does not necessarily mean that he makes an attractiveness judgment about it. It may mean simply that he is attracted to the thing and thereby has an experience that’s indicative of, or evidence for, its attractiveness. Although the combination of ‘to find’ with a predicate adjective has the superficial grammar of an attitudinal verb, it doesn’t necessarily express the content or valence of an attitude. Someone can find his dinner indigestible, for instance, without having any attitude toward it whatever: he may simply have a cramp. To say that he finds his dinner indigestible in this case is not to describe his cramp as an attitude; it’s to describe the cramp in terms of what it indicates. Of course, the phrase ‘find attractive’ does describe an attitude when it’s applied to desire; but it may still describe the attitude in extrinsic terms, as evidence of its object’s attractiveness, rather than in terms of the attitude’s content or direction of fit. Hence the subject’s finding something attractive may not entail that he has mental access either to a proposition about attractiveness or to an attitude that takes ‘attractive’ as its constitutive predicate.

These subtleties in our descriptions of propositional attitudes are compounded by the fact that predicates like ‘attractive’ and ‘desirable’ can have both normative and nonnormative senses. Calling something desirable can have the normative meaning that the thing is correct or fitting to desire; but it can also have the purely psychological meaning that the thing tends to be desired, that it’s easy or natural to desire. We must therefore be doubly careful with the observation that someone who desires something can be said to find it desirable. Desiring something may sometimes entail having an experience indicative of the thing’s being easily or naturally desired; and in such a case, it entails finding the thing desirable, in some sense. But to have an experience that’s evidence of something’s tendency to be desired is not necessarily to think of the thing as readily desired, much less to think of it as correct to desire or worth desiring. Hence we mustn’t assume that someone who finds something desirable thereby makes a desirability judgment that would subjectively justify action.


Although I think that some such confusion is responsible for the use of ‘good’ or ‘desirable’ as the constitutive predicate of desire, I also think that this usage can be harmless, if properly understood. I am not opposed to describing desire as the attitude of regarding something as good, so long as this description is taken merely to express the attitude’s direction of fit. Unfortunately, the description is also taken to imply that desire has the justificatory force of a value judgment; and in this respect, the description is misleading.

The resulting confusion can best be explained by analogy to a related misunderstanding about the justificatory force of belief. We are inclined to think that belief qualifies as a judgment of a proposition’s truth and carries the justificatory force of a truth judgment, simply because it entails regarding the proposition as true. Yet to say that belief entails regarding a proposition as true doesn’t exhaust the relation between belief and the truth. There are many cognitive attitudes other than belief, attitudes that have the same direction of fit and consequently take the same constitutive predicate. Hypothesizing that p, assuming that p, fantasizing that p, and the like are all attitudes in which p is regarded, not as a representation of what is to be brought about, but rather as a representation of what is. The propositional object of these attitudes is thus regarded as true. Yet fantasizing that p doesn’t amount to a judgment on the truth of p, and it lacks the justificatory force that would attach to such a judgment. Hence the reason why belief qualifies as a truth judgment cannot be simply that its constitutive predicate is ‘true’.

One might be inclined to say that fantasizing and hypothesizing don’t involve regarding anything as really true. But to say this is simply to acknowledge that there is more than one way of regarding a proposition as true. The sense in which hypotheses aren’t regarded as really true is, not that they aren’t regarded as true at all, but rather that they are only hypothetically so regarded. Fantasies aren’t regarded as really true because it is only imaginatively, or in imagination, that they are regarded as true—not because they aren’t so regarded at all. To regard something as “really” true must therefore be a particular way of regarding it as true—and, in particular, some way other than imaginatively or hypothetically.

My point here is not the purely grammatical point that imagining or assuming something entails imagining or assuming it to be true; after all, wanting something entails wanting it to be true, as well. In this construction, ‘true’ attaches, trivially, to all propositional attitudes, simply by virtue of their being attitudes toward the bearers of truth values. My point is rather that what distinguishes belief from desire distinguishes assumption, hypothesis, and imagination from desire, too—namely, that they treat their propositional objects as reflecting antecedently fixed conditions rather than as dictating conditions to be achieved, as facta rather than facienda. To be sure, these attitudes don’t treat their propositional objects as reflecting the actual facts. But they still treat those propositions as factual reports rather than practical dictates—as being already true of some completed, though unreal, states of affairs rather than as to be made true by the completion of such states. These attitudes therefore share the distinctively cognitive direction of fit.

The definition that philosophers have traditionally offered for direction of fit has somewhat obscured the difference between cognitive and conative attitudes, as well as the differences among the attitudes within either category.[21] Direction of fit has traditionally been defined in terms of the locus of responsibility for correspondence between an attitude and the world. Whether an attitude has one direction of fit or the other is said to depend on whether the attitude is responsible for conforming itself to the world or makes the world responsible for conforming itself to the attitude. The difference in direction of fit between an expectation and an intention is thus supposed to entail that when an expectation isn’t fulfilled, the fault lies with the expectation, whereas when an intention isn’t fulfilled, the fault lies with the world.[22]

This definition doesn’t accurately characterize the difference between cognition and conation. A lack of correspondence between the world and an assumption, for example, doesn’t constitute a failure for which one party or the other must be to blame. If the assumption is made solely for the sake of argument, then it neither takes responsibility for fitting the world nor makes the world responsible for fitting it. Fit between such an assumption and the world is of no importance and is therefore neither party’s responsibility. Yet an assumption is still like an expectation, and unlike an intention, in that what’s assumed is regarded as true rather than as to be made true, as modelled after its intentional object rather than a model for it. Thus, an assumption still possesses the cognitive direction of fit, even though neither it nor the world is responsible for conforming itself to the other.


Where the traditional definition goes wrong, then, is in presupposing that whenever a proposition is regarded as true or to be made true, its truth thereby comes to constitute a success—and its falsity, a failure—for which either the attitude or the world must bear responsibility. Not every attitude of regarding something as true, or to be made true, has the thing’s being or coming true as a criterion of success. I shall now argue that different criteria of success—or, as I shall now put it, different constitutive aims—help to account for the differences among attitudes with the same direction of fit.

Let us say that to regard a proposition as true, in the sense that applies to all cognitive attitudes, is to accept the proposition. We can then distinguish believing that p from assuming or fantasizing that p, for example, as follows. Assuming or fantasizing that p consists in accepting it irrespective of whether it is really true; whereas believing that p requires accepting it as if in response to its being true. Thus, belief bears a double relation to the truth. Believing a proposition entails not only regarding the proposition as true but, in addition, so regarding it in a manner designed to reflect whether it really is true.[23] The latter relation to the truth is part of what distinguishes believing from the other cognitive attitudes, in which a proposition is regarded as true without concern for whether it really is.

The clearest way to analyze such differences between belief and the other cognitive attitudes is in terms of the subject’s dispositions to regulate his acceptance of a proposition. When someone assumes a proposition, he or his cognitive faculties are disposed to regulate his acceptance of it in ways designed to promote the ends of argument or inquiry: he comes to accept the proposition when doing so seems conducive to scoring a point or making a discovery, and he is disposed to continue accepting it only insofar as doing so seems to serve such polemical or heuristic purposes. When someone fantasizes, his acceptance of propositions is regulated in ways designed to whet his appetites, stimulate his mind, or provide a substitute for the fulfillment of his wishes: he accepts whatever propositions promise to provide the appropriate excitement or vicarious satisfaction.

When someone believes a proposition, however, his acceptance of it is regulated in ways designed to promote acceptance of the truth: he comes to accept the proposition, for example, when evidence indicates it to be true, and he’s disposed to continue accepting it until evidence indicates otherwise. Part of what makes someone’s attitude toward a proposition an instance of belief rather than assumption or fantasy, then, is that it is regulated in accordance with epistemic principles rather than polemics, heuristics, or hedonics. An attitude’s identity as a belief depends on its being regulated in a way designed to make it track the truth.[24]

Although the dispositional explication of this difference between belief and the other cognitive attitudes is perhaps the clearest, its import can be expressed, as I expressed it above, in attitudinal terms. That is, regulating one’s acceptance of the proposition by the exigencies of argument can be described as accepting the proposition without regard to its truth but rather with polemical or heuristic intent. And regulating one’s acceptance of a proposition by evidence of its truth can be described as accepting the proposition with an eye to its truth, or with truth as one’s aim. Belief and assumption are then described as two-tier attitudes, combining the first-order attitude of acceptance with different second-order attitudes—namely, the different aims or intentions with which a proposition can be accepted.[25]

We thus arrive at the familiar dictum that belief aims at the truth.[26] Properly understood, this dictum means that belief combines the attitude of regarding something as true with the aim of regarding as true what really is true—of getting the truth right. Hence belief not only has truth as its constitutive predicate but also has correctness in matters of truth—or, as one might put it, the “real” truth—as its constitutive aim.[27]

The difference between the constitutive aim of belief and those of other cognitive attitudes can also be expressed, even less clearly, by being incorporated into the constitutive predicates of these attitudes. Once we conceive of belief as a two-tier attitude, we tend to think of the second tier as having a constitutive predicate of its own, or as modifying the constitutive predicate of the entire attitude. Accepting a proposition is the attitude of regarding it as true; but accepting a proposition with the aim of accepting what’s really true must entail regarding the proposition’s acceptance as a means to that end. If one is aiming to get the truth right when one comes to regard p as true, then one must in effect regard p not only as true but also as something to be regarded as true for the sake of getting the truth right. The double relation between belief and truth can thus be expressed, somewhat obscurely, in the thought that believing a proposition entails regarding it as something that one is right to regard as true. Surely, that’s what we mean when we say that believing p entails regarding p as “really” true; we mean that it entails regarding p not only as true but also as correct to regard in this way. Belief can thus be conceived as having a constitutive predicate that expresses its own correctness.[28]


The constitutive aim of belief is necessary to belief’s being a judgment on the truth of a proposition, since nothing would count as a judgment on p’s truth if it didn’t aim at getting right whether p is true. The constitutive aim of belief is also essential to the attitude’s justificatory force as a premise of inferences.

Consider, for example, why my believing that p and that p q gives me subjective justification for believing that q. The answer cannot be simply that believing these propositions entails regarding them as true, and that the truth of the premises guarantees that of the conclusion. After all, fantasizing that p also entails regarding p as true, but fantasies justify nothing whatever, not even other fantasies.

The reason why one belief has subjective justificatory force for other beliefs is that the attitude of belief involves not only regarding a proposition as true but also doing so with the aim of getting the truth right. Because my belief that p is an attempt at tracking the truth, it makes a prima facie claim to be on the right track; and with this claim, the belief offers itself as a guide for other attitudes, provided that they, too, aim to track the truth. One belief guides the others because the latter aim at getting the truth right and the former represents my best efforts thus far toward the same end.


Those who have noted that belief aims at the truth are often inclined to think (mistakenly, I shall argue) that desire correspondingly aims at the good.[29] What are the consequences of the assumption that desire and belief are analogous in this respect?

Well, desire is like belief in being only one of many attitudes with its characteristic direction of fit. Wishing, hoping, and the like are also attitudes in which a proposition is regarded as a pattern for the world to follow, as something to be brought about or made true. Let us say that these attitudes are different ways of approving a proposition, just as the cognitive attitudes are different ways of accepting one. Suppose, then, that the difference between desire and other modes of approval was analogous to that between belief and other modes of acceptance.[30] In that case, wishing that p, for example, might entail regarding p as to be brought about, but so regarding it irrespective of whether it really was to be brought about; whereas desiring that p would entail regarding p as to be brought about, and doing so with an eye to whether it really was.[31] The analogy between desire and belief would thus lead to the conclusion that, just as belief aims at tracking the actual facts, so desire aims at tracking the actual facienda.

Such an attitude would indeed have the justificatory potential claimed for it in Davidson’s theory of rational guidance. As an attempt to track the actual facienda, it would present itself as a guide for one’s actions, representing one’s best efforts to identify what really was to be brought about. What’s more, the feature lending justificatory force to this attitude would also give the attitude a constitutive predicate expressing its correctness. For if desiring something entailed regarding it as to be brought about, with the aim of so regarding what really was to be brought about, then it would entail regarding the thing as correct to regard in that way—as correct to approve.

We would thus arrive in the vicinity of Davidson’s claim that to desire something is to regard it as desirable.[32] This claim seems to say that to desire something is to have an attitude toward it as worthy of that very attitude. We can now see that such a claim would make sense if it rested on the assumption that desire, like belief, had correctness as its constitutive aim.

Yet when the use of an evaluative term as the constitutive predicate of desire was first introduced, it was understood merely as a way of expressing the attitude’s direction of fit—as a colorful alternative to the predicates ‘to be brought about’ or ‘to be made true’. Now a different interpretation has emerged. The claim that to desire something is to regard it as desirable is now being interpreted to imply, not just that the propositional object of desire is regarded as something to be made true or brought about, but also that it is so regarded with the aim of getting things right.

This implication can easily infiltrate our understanding of every evaluative term used to express a desire. Whereas the word ‘good’ in ‘regarding p as good’ was initially taken as a synonym for ‘to be brought about’, a phrase that merely expressed the attitude’s direction of fit, it can easily be understood as meaning “really to be brought about” or “correct to approve”—phrases expressing the aim of tracking the actual facienda. Regarding p as good can thus be thought to entail a potentially action-guiding judgment as to what is worth bringing about or making true.


In order to assess the validity of this new interpretation, we must ask whether desire really has a constitutive aim analogous to that of belief. The answer, I’m afraid, is no. When we consider how desire differs in aim from other modes of conation, we find that the difference is not analogous to that between belief and other modes of cognition. The difference in aim between desire and other conative attitudes appears to be that desire aims, not at the good, but rather at the attainable.[33]

One cannot desire something if it seems impossible or if it seems already to have come about; one can desire that p only if p seems attainable, in the sense of being a possible future outcome. Yet the obstacle to desiring what seems unattainable, or already attained, is not that such things cannot be objects of approval. One can wish that p even if the truth of p seems quite impossible, and one can hope that p when p already seems to be true. Thus, one can approve the unattainable or the attained, and what prevents one from desiring them must be something else.

The obstacle, I would suggest, is that desire has the attainable as its constitutive aim. That is, unless approval is regulated in a way designed to track what’s attainable, it doesn’t qualify as desire; and approval isn’t being regulated in a way designed to track the attainable if it’s directed at what already seems actual or impossible.[34]

Although this explanation for the limits on desire is only an hypothesis, I think that it is clearly more plausible than the alternative hypothesis about desire’s constitutive aim. The grounds that I have just outlined for thinking that desire aims at the attainable are not matched by any comparable grounds for thinking that it aims, more narrowly, at what really is to be brought about. Nothing other than allegiance to the cognitivist program would tempt us to think that desire must be regulated in a way designed to track the facienda.

The upshot is that nothing about desire entitles us to credit it with the justificatory force of a value judgment. A judgment on something’s value would be an attempt to get things right, and it would consequently have the standing to guide one’s actions. But even if desiring something consists in regarding the thing as good, in a sense synonymous with “to be brought about”, it isn’t an attempt at getting right whether the thing really is to be brought about, and so it doesn’t amount to a judgment on the thing’s goodness. Desiring something consists in regarding it as to be made true only in the sense that imagining something consists in regarding it as true. Hence desire has the same subjective justificatory force as fantasy—that is, none at all.


That desire doesn’t aim at correctness explains why desire can be perverse.[35] As Michael Stocker has pointed out, one can often desire things conceived as worthless, or even bad, and desire them precisely under those descriptions.[36] A tendency to desire things under negative descriptions is an essential element of various emotions and moods such as silliness, self-destructiveness, or despair. A mood of playfulness is, in part, a disposition to form desires for things conceived as having no particular value; a self-destructive mood is, in part, a disposition to form desires for things conceived as harms; and so on. None of these desires could retain its characteristic idleness or perversity if it involved an attempt at getting things right.

After all, what makes a desire perverse is that its propositional object implies that it is inappropriate. That is, the perverse subject desires that something undesirable occur, and its being undesirable is part of the description under which he desires it. He thus holds the attitude of desire toward something under the description that it is unworthy of that attitude. This discrepancy between his attitude toward an object and his conception of the object’s deserts would be impossible in an attitude whose nature was to aim at getting things right and whose constitutive predicate consequently implied the attitude’s correctness. Correctness in approval simply cannot be one’s aim when one approves of something under the description that it is unworthy of approval.

Consider, by way of analogy, the prospects for perversity in belief. Because belief entails not only regarding a proposition as true but, in addition, doing so with the aim of getting the truth right, belief cannot be transparently perverse. That is, one cannot believe a proposition that presents itself as false—say, the proposition ‘I am five inches taller than I really am’.[37] The reason why one cannot believe such a proposition is not that one is incapable of accepting it. The reason is rather that the only way of accepting such a self-evident falsehood would be to accept it irrespective of its truth; and accepting a proposition irrespective of its truth wouldn’t amount to believing it. If one accepts this proposition irrespective of its truth, one will then be assuming it, as one might do for the sake of reasoning counterfactually (“Let’s assume that I’m five inches taller than I really am . . .”).

In short, a proposition that presents itself as false cannot be the object of an attitude that aims at getting the truth right. Similarly, a prospect that presented itself as bad could not be the object of an attitude that aimed at correctness in regarding things as to be brought about. Thus, if aiming at correctness were constitutive of desire, as it is of belief, perverse desire would be inconceivable.


The assumption that desire aims at the good forces the cognitivist to misdescribe examples of perverse desire. Consider, for a particularly vivid example, the figure of Satan in Paradise Lost, who responds to his defeat with the cry, “Evil be thou my Good.”[38] Satan is here resolving to desire and pursue evil, and hence—as he himself puts it—to regard evil as good. But he cannot reasonably be interpreted as adopting new estimates of what’s valuable—that is, as resolving to cease judging evil to be evil and to start judging it to be good. If Satan ever loses sight of the evil in what he now desires, if he ever comes to think of what he desires as really good, he will no longer be at all satanic; he’ll be just another well-intentioned fool. The ruler of Hell doesn’t desire what he wrongly thinks is worthy of approval; he desires what he rightly thinks isn’t. He thereby illustrates my point that regarding something as good, in the sense requisite to desiring it, does not amount to making a favorable judgment on its value.

See how Satan’s horns are blunted when his desires are misinterpreted as full-blooded value judgments, in this passage from Elizabeth Anscombe’s Intention’?[39]

‘Evil be thou my good’ is often thought to be senseless in some way. Now all that concerns us here is that ‘What’s the good of it?’ is something that can be asked until a desirability characterisation has been reached and made intelligible. If then the answer to this question at some stage is ‘The good of it is that it’s bad’, this need not be unintelligible; one can go on to say ‘And what’s the good of its being bad?’ to which the answer might be condemnation of good as impotent, slavish, and inglorious. Then the good of making evil my good is my intact liberty in the unsubmissiveness of my will.

What sort of Satan is this? He is trying to get things right, and so he rejects the good only because he has found respects in which it is unworthy of approval. He rejects the good, that is, only because it is slavish and inglorious, and hence only because shunning the good is a means to liberty and glory. But then he isn’t really shunning the good, after all, since the goods of liberty and glory remain his ultimate goals. Anscombe’s Satan can desire evil only by judging it to be good, and so he remains, at heart, a lover of the good and the desirable—a rather sappy Satan.


Let me summarize my argument thus far by updating one of Anscombe’s own devices for analyzing the difference between cognition and conation. Anscombe suggests that the difference between cognition and conation is analogous to that between an inventory and a shopping list.[40] An inventory is modelled after one’s existing stock: it represents things as being on the shelves, having already been obtained. A shopping list is a model for one’s stock to follow: it represents things as not yet on one’s shelves but to be obtained.

What I have argued, in effect, is that Anscombe’s analogy neglects differences within the categories of possible inventories and possible shopping lists. When a list of items falls into our hands, we must of course ascertain whether it represents those items as having been obtained or as to be obtained. But we must also ascertain the aims with which the contents of the list have been regulated. A list of things on the shelves may have been compiled in a way designed to represent what’s actually on the shelves, but it may also have been compiled in a way designed to include whatever might be on the shelves, as a checklist against which the actual contents of the shelves can then be compared. A checklist is also an inventory; it’s just a hypothetical inventory. Similarly, a list of things to be obtained may have been compiled in a way designed to track what’s actually needed, but it may also have been compiled in way designed to lend excitement to one’s bet on the lottery. A wish list is also a shopping list; it’s just a fantasy shopping list.

Thus, to find a list headed “Things in stock” is not yet to have grounds for any conclusions about what’s available for dinner; and to find a list headed “Things to buy” is not yet to have grounds for making purchases. If the inventory is hypothetical—merely a checklist—then it doesn’t embody a judgment of what’s in stock and doesn’t justify any conclusions about what’s available; and if the shopping list is fantastical—merely a wish list—then it doesn’t embody a judgment of what to buy and doesn’t justify any purchases. Judgmental and justificatory force attaches only to those lists whose entries have been compiled and regulated with the aim of getting things right.


The cognitivist conception of desire as an action-justifying value judgment thus depends on a misinterpretation of the sense in which desiring something entails regarding it as good. Properly understood, the use of ‘good’ as desire’s constitutive predicate doesn’t support the cognitivist conception of desire as a value judgment of the sort that would subjectively justify action; and it therefore doesn’t support identifying an agent’s motivating desire with his guiding reason.

Of course, to show that desire lacks the subjective justificatory force of a value judgment is not to show that such force may be lacking from whatever else might constitute an agent’s reason for acting. Indeed, one might think that my arguments thus far prove only that desire cannot be what constitutes an agent’s reason for acting, precisely because his reason, whatever it may be, must somehow present his action as a good thing to do.[41] One might then continue to claim that an action performed for a reason must be aimed at the good, without assuming that the guiding evaluation is embodied in a desire.

Yet I think that one of my arguments against identifying reasons with desires suggests an argument against identifying reasons with value judgments in any form. In that argument, I explained the capacity of desires to be perverse in terms of their not being aimed at the good; here I shall contend that reasons for acting can be perverse as well. That is, an agent’s reason for doing something can be that it’s a bad thing to do; and so its justificatory force cannot depend on that of a favorable evaluation.

Suppose that I have suffered a profound disappointment that has cast me into a mood of bitterness and despair. In this mood, the very thought of ameliorating my condition, or the condition of the world, strikes me as a sick delusion. All attempts at constructive action seem absurd. No more earnest efforts for me, I say to myself, no more worthy endeavors: to hell with it all.[42]

Being in despair doesn’t prevent me from being moved to act, however. I am moved to stay at home, refuse all invitations, keep the shades drawn, and privately curse the day I was born. I may even be moved to smash some crockery—though not in order to feel better, mind you, since trying to feel better seems just as ludicrous a project as any other. (Someone who smashes crockery in order to feel better didn’t feel all that bad to begin with.) What’s more, I engage in these actions not only out of despair but also in light of and on the grounds of despair. That is, despair is part of my reason as well as part of my motive for acting.

But do I regard my actions, in light of my despair, as good or desirable or positive things to do? Far from it. I am determined never to do a good or desirable or positive thing again. If smashing things seemed like a good thing to do, I would pointedly avoid it; even if it seemed good only for someone in despair, I would still avoid it; indeed, it’s seeming good for someone in despair would count most strongly against it, since doing things that are good for someone in my condition is exactly what, in light of that condition, makes sense for me to avoid.[43] I’m smashing things because this seems like an utterly worthless act, worthless from every perspective but especially from mine. My reason for acting thus includes not only my mood but also an unconditionally negative evaluation of the action.

To be sure, there is a kind of perversity that would preclude my acting for reasons; but the perversity that would preclude acting for reasons is not a counter-evaluative state of mind but rather a counter-rational one. That is, I cannot act for reasons if I don’t care about doing what’s justified or (as I would prefer to put it) what makes sense. But I can still care about doing what makes sense even if I don’t care about the good. This possibility is demonstrated by my capacity to be guided by what makes sense in light of a counter-evaluative mood such as despair, since what makes sense in light of such a mood just is to do what’s bad rather than what’s good.

Yet pursuing the bad on the grounds of despair would be impossible if every reason for an action had to present that action as a good thing to do. The problem is not that the badness of an action could never make the action seem good. The problem is that if the badness of an action weighed with me by making the action seem good, then I would be once again engaged in pursuing the good—a pursuit incompatible with the very mood that helps to constitute my reason for acting. Just as Satan would have to shed his satanism in order to value evil sub specie boni, so I would have to shed my despair in order to pursue a self-destructive course under that positive guise. If I were swayed toward an action because its badness made it seem like a good thing to do, then I’d be in the business of finding silver linings, a business that’s closed to me so long as I am truly acting out of despair.


If my arguments are correct, then practical reasoning is nothing like what it has traditionally been conceived to be. Desires lack the evaluative force that is thought to make them reasons for acting; and the justificatory force of reasons is not evaluative, in any case. Since reasons do not recommend an action by presenting it as a good thing to do, actions performed for reasons need not be performed under the guise of the good.


1. This chapter originally appeared in Noûs 26 (1992) 3–26, and is reprinted by permission of Blackwell Publishers. The material in this chapter was presented, in various forms, at the University of Michigan, Yale University, and the University of Dayton. I am grateful to these audiences for their comments. I am also indebted to Rüdiger Bittner, Paul Boghossian, Jennifer Church, Carl Ginet, Jonathan Lear, Richard Miller, Donald Regan, Connie Rosati, Geoffrey Sayre-McCord, Sydney Shoemaker, Michael Slote, Michael Smith, and Dennis Stampe for comments on earlier drafts or discussion of related issues. Thanks are also due to Lloyd Humberstone for allowing me to read the manuscript of a paper entitled “Direction of Fit”, which was subsequently published in Mind 101 (1992) 59–83.

2. G. E. M. Anscombe, Intention (Ithaca: Cornell University Press, 1963), 70 ff.; Donald Davidson “Intending”, reprinted in Essays on Actions and Events (Oxford: Clarendon Press, 1980), 97, n. 7; Dennis Stampe, “The Authority of Desire”,The Philosophical Review 96 (1987), 355; Alvin I. Goldman, A Theory of Human Action (Princeton: Princeton University Press, 1970), 94.

3. See Ronald B. De Sousa, “The Good and the True”, Mind 83 (1974) 534–51; see also Donald Davidson, “How is Weakness of the Will Possible?”, reprinted in Essays on Actions and Events, 22; David Pears, Motivated Irrationality (Oxford: Clarendon Press, 1984), 198.

4. Michael Smith refers to these two stories as the “intentional perspective” and the “deliberative perspective” (“Valuing: Desiring or Believing?”, in David Charles and Kathleen Lennon [eds.], Reduction, Explanation and Realism [Oxford: Oxford University Press, 1992], 323–60). I understand that Rüdiger Bittner is also working independently on the relation between these stories.

5. I have borrowed this example from Davidson, “How is Weakness of the Will Possible?”, 31 ff. I shall presently discuss Davidson’s own analysis of the case.

6. I realize, of course, that some philosophers would deny that genuine propositions justify action, independently of their actually or potentially exerting some motivational influence on the agent. I discuss this view briefly below, under the name of noncognitivism.

7. I shall avoid giving an example of an action-justifying proposition, so as to remain neutral on the precise nature of practical reason.

8. To say that acting for a reason entails being influenced by the intrinsic action-justifying character of a belief is not to say that it requires the operative belief to have intrinsic motivational force that’s independent of any desires. One may have a desire to perform actions that are justified, or to perform actions that are related to one’s circumstances in particular ways, which are in fact justifying. Any belief that gains motivational force from such a desire will also owe that motivational force, in part, to its own action-justifying character, and will therefore fulfill the story of rational guidance. I thus assume that beliefs can be intrinsically justifying even if they cannot be intrinsically motivating. Here I differ with, e.g., Thomas Nagel, The Possibility of Altruism (Princeton: Princeton University Press, 1970), and John McDowell, “Are Moral Requirements Hypothetical Imperatives?”, Proceedings of the Aristotelian Society (1978, supp. Vol.) 13–29.

9. This approach can be traced back at least as far as Hobbes’s definition of deliberation, at Leviathan, Part I, chapter vi.

10. The noncognitivist cannot remove the discrepancy by pointing out that he conceives of justificatory force as a disposition to motivate, which can indeed explain particular instances of motivation. In the commonsense story, the justificatory force of a reason explains (rather than consists in) the reason’s disposition to motivate as well as its motivating on particular occasions.

My remarks on the noncognitivist view echo much that is said by W. D. Falk in “Action-Guiding Reasons”, (1963), reprinted in Ought, Reasons, and Morality (Ithaca: Cornell University Press, 1986), 82–98. Yet Falk does not ultimately reject noncognitivism, since he, too, believes that reasons “are . . . choice-supporting in proportion to their choice-influencing potential” (p. 92).

11. Davidson, “Intending”, 86.

12. Of course, the young child may not be susceptible to rational guidance, either; but this point hardly counts in Davidson’s favor. When Davidson characterizes belief-desire motivation as equivalent to rational guidance, he leaves no room for agents who are moved by desires without being guided by reasons. The fact that children, who pursue desired ends, can nevertheless be too young for rational guidance is therefore a point against Davidson, on a par with my point that they can be too young for the concept of the desirable. (For an alternative argument against identifying desires with evaluative judgments, see David Lewis, “Desire as Belief”, Mind 97 [1988] 323–42.)

13. I think that Davidson sometimes betrays an uncomfortable awareness of this difficulty. He wants to avoid the implausibly strong claim that desires consist in value judgments. (See, e.g., the discussion of wanting to drink a can of paint in “Actions, Reasons, and Causes”, reprinted in Essays on Actions and Events, 4.) And yet he senses that he is committed to that claim by his strategy of equating motivation with rational guidance. See, e.g., “Intending”, 102 (desires “constitute” value judgments); “How is Weakness of the Will Possible?”, 31 (desiring something is “setting a positive value” on it); “Intending”, 97, n. 7 (desiring something entails “holding” it to have “some positive characteristic”).

14. The reader will find that I use the term ‘direction of fit’ in a somewhat different sense from others who have used the term—including, for example, John Searle, Intentionality: An Essay in the Philosophy of Mind (Cambridge: Cambridge University Press, 1983), 7 ff.; and Mark Platts, Ways of Meaning (London: Routledge and Kegan Paul, 1979), 257. See also G. E. M. Anscombe, Intention (Ithaca: Cornell University Press, 1963), 56. The definition offered by these authors will be explained and criticized in the text below.

15. Direction of fit is clearly what Dennis Stampe has in mind in his version of this claim: “[W]hile the belief and the desire that p have the same propositional content and represent the same state of affairs, there is a difference in the way it is represented in the two states of mind. In belief it is represented as obtaining, whereas in desire, it is represented as a state of affairs the obtaining of which would be good,” (“The Authority of Desire”, The Philosophical Review 96 [1987] 355.

16. What psychological realization, if any, attends the predicates used to characterize propositional attitudes? A behaviorist might insist that regarding a proposition as true is nothing more than being disposed to behave as would be appropriate if the proposition were true. Alternatively, one might imagine that regarding a proposition as true entails having a representation of it in a particular mental compartment, which might be called the “true” box, because of its role in the mental architecture. Another alternative would be to give a phenomenological rather than functional account of the relevant mental posture toward a proposition—whatever such an account might be. The most attractive alternative, for present purposes, is a stance of neutrality among these and other models of propositional attitudes. One needn’t adopt any particular model of propositional attitudes in order to adopt the descriptions whose realization is the point of contention among such models. One can thus insist that whatever believing that p consists in, it must be something that would appropriately be called regarding p as true; and whatever desiring that p consists in, it must be something that would appropriately be called regarding p as to be made true. For further discussion of this issue, see Chap. 10, below.

17. This version of cognitivism is articulated by Dennis Stampe: “Desires constitute reasons for us to act because their contents are represented as states of affairs the realization of which would be good” (ibid.). I do not know whether Davidson accepts or would accept this formulation. His accepting it would explain why he says both that desires are, and that they are expressed by, value judgments.

18. Davidson seems to entertain a view like this when he says, “[I]f someone acts with an intention, he must have attitudes and beliefs from which, had he been aware of them and had the time, he could have reasoned that his action was desirable” (“Intending”, 85, emphasis added). Here Davidson seems to suggest that the agent needs to become aware of his attitudes in order to be in a position to draw a conclusion about the desirability of his action. I do not understand, however, how this suggestion squares with Davidson’s claim that an intentional action constitutes the agent’s conclusion that the action is desirable (“Intending”, 99). For how can his action constitute a conclusion that he might not have been in a position to draw?

19. But see the discussion of the issue in my Practical Reflection (Princeton: Princeton University Press, 1989).

20. That Davidson regards desire as justifying action subjectively is confirmed by passages such as this: “Thus there is a certain irreducible—though somewhat anaemic—sense in which every rationalization justifies: from the agent’s point of view there was, when he acted, something to be said for the action” (“Actions, Reasons, and Causes”, 9, italics mine.)

21. In this paragraph and the next I have benefitted from Lloyd Humberstone’s manuscript “Direction of Fit”.

22. See the references to Searle, Platts, and Anscombe in n. 14, above.

23. For an explication of this expression, see the following note. The expression gives, at most, a necessary condition of belief, not a sufficient condition. Indeed, even to call it a necessary condition may still be too strong. My suspicion is that attitudes qualify as beliefs or desires or intentions, etc., by virtue of approximating to a paradigm or ideal specimen. What are usually called the necessary conditions for belief or desire are in fact a definition of the paradigm case; and instances that fail to meet one condition may still qualify as beliefs or desires if they sufficiently resemble the relevant paradigm in other respects. (See my Practical Reflection, 136.)

24. What do I mean when I describe a cognitive mechanism as “designed to track the truth”? I don’t mean that it’s designed in such a way that it succeeds in tracking the truth, since beliefs can be false. Rather, I mean that its design is governed by the goal of tracking the truth, although it may not attain that goal invariably or completely. And I am of course assuming that the concept of design encompasses the work of the pseudo-designer known as natural selection.

In the text I have assumed that a mechanism designed to track the truth would operate in response to evidence, as it probably would if designed by natural selection. But other mechanisms might also qualify as being designed to track the truth—particularly if they were designed by the subject himself, in accordance with an alternative epistemology. For example, someone who believes that the truth about something is to be found by consulting scripture or seeking revelation may regulate his acceptance of the relevant propositions by those means. The resulting attitudes will qualify as beliefs, however, since the method of their regulation will be designed to track the truth, in the requisite sense. (I am indebted to Michael Slote for pointing out this possibility.)

25. Note that the aim or intention with which a proposition is accepted may belong to the subject’s cognitive faculties rather than to the subject himself, depending on who or what is regulating the subject’s acceptance of propositions. I shall henceforth ignore this distinction. In “Direction of Fit”, Lloyd Humberstone cites J. O. Urmson, “Memory and Imagination”, Mind 76 (1967) 83–91, as offering a similar account of the difference between imagining and remembering.

26. The notion that belief aims at the truth figures prominently in the literature on believing at will. See, e.g., Bernard Williams, “Deciding to Believe”, reprinted in Problems of the Self (Cambridge: Cambridge University Press, 1970).

27. This aspect of belief is, I suspect, what leads Davidson to say, “Someone cannot have a belief unless he understands the possibility of being mistaken” (“Thought and Talk”, reprinted in Essays on Actions and Events, 168). Because believing entails aiming to get the truth right, it would seem to entail understanding the possibility of a mistake. But the soundness of this argument depends on the assumption that the constitutive aim of belief must be an attitude of the believer rather than a purpose inherent in the design of his cognitive faculties. I have suggested that a person’s acceptance of a proposition can be aimed at getting the truth right—and hence qualify as a belief—so long as it is regulated by a mechanism designed to track the truth. (See n. 24, above.) In that case, the believer himself might lack the conceptual resources for framing the requisite aim in an attitudinal sense.

28. Or, more precisely, the correctness of the acceptance involved in the belief. The view that belief involves an intimation of its own correctness is similar to the view of judgment that Hannah Ginsborg attributes to Kant. Ginsborg argues that reflective judgment, for Kant, is “the capacity to take one’s mental state in perceiving a particular object to be universally valid with respect to that object.” Ginsborg, “Reflective Judgment and Taste”, Noûs 24 (1990) 63–78, 64.

29. See Anscombe, Intention, 76: “Truth is the object of judgment and good the object of wanting. . . .” See also DeSousa, “The Good and the True”, Mind 83 (1974) 538: “[T]ruth and good are the targets of belief and want.”

30. Note that the distinction developed here is different from Pears’s distinction between “weak” and “strong” value judgments, even though Pears regards his distinction as defining the line between those value judgments which are and those which are not necessarily implicated in preference and noncompulsive intentional action (Motivated Irrationality, 196 ff.). “Weak” value judgments, in Pears’s terminology, are still judgments; what distinguishes them from their “strong” counterparts is the set of interests on which they are based. My distinction, by contrast, divides value judgments from valuings or preferrings that aren’t judgments at all.

31. What would it mean to say that something really was to be brought about? I’m not sure. This phrase is generated when the formula describing the constitutive aim of belief is adapted for the purpose of attributing an analogous aim to desire. Since I reject the analogy, I needn’t take responsibility for the language it generates.

32. I say “in the vicinity” because to regard something as worthy of approval is not quite to regard it as worth desiring—approval and desire being slightly different attitudes.

33. For this point I am indebted to Geoffrey Sayre-McCord.

34. Note, as before, that an attitude’s constitutive aim should be treated as a necessary but not a sufficient condition. An instance of approval must have other features—including, perhaps, actual or counterfactual behavioral manifestations—in order to qualify as a desire. Compare n. 23, above, for further qualifications.

35. Although I believe that we can indeed have perverse desires (and shall assume so in the text), I am not strictly committed to this empirical claim. I am committed only to the claim that desires can in principle be perverse, in the sense that there is nothing about the concept of desire that would prevent a perverse attitude from satisfying it. That is, an attitude can qualify as a desire even if it is perverse. This conceptual claim would be compatible with there being some contingent feature of human psychology that prevented us from having perverse desires.

36. “Desiring the Bad: An Essay in Moral Psychology”, Journal of Philosophy 76 (1979) 738–53.

37. The word ‘really’ in this proposition rigidly designates our world, so that the proposition is evidently false in this world but true in some other possible world.

38. Book IV, line 110.

39. Intention, 75.

40. Ibid., 56 ff.

41. For the view that evaluation is embedded in the very concept of a reason, see, e.g., Kurt Baier, “Rationality, Reason, and the Good”, in D. Copp and D. Zimmerman (eds.), Morality, Reason and Truth, (Totowa, N.J.: Rowman & Littlefield, 1984), 202. “What I call ‘cognitive’ reasons are employed in the activity of examining what to believe, and ‘practical’ ones are employed in the activity of examining what to do. The specific aim of the first is beliefs that are true; the second is intentions or acts that are good. We could say that the aim in the case of the first is the true and in the case of the second the good.” See also E. J. Bond, Reason and Value (Cambridge: Cambridge University Press, 1983), 2–3. For a persuasive attack on this view, see Anderson, “Pluralism, Deliberation, and Rational Choice”.

42. Again, see Stocker, “Desiring the Bad”, 745.

43. Ibid., 748.

Originally published at




This article and its reviews are distributed under the terms of the Creative Commons Attribution 4.0 International License, which permits unrestricted use, distribution, and redistribution in any medium, provided that the original author and source are credited.