Alhassanain(p) Network for Heritage and Islamic Thought

The Rationality of BeliefAnd Some Other Propositional Attitudes

Appeared in Philosophical Studies 110: 163-196, 2002.

Thomas Kelly

Harvard Society of Fellows

University of Notre Dame

www.princeton.edu

www.alhassanain.org/english

Notice:

This workispuplished on behalf of www.alhassanain.org/english.

The typing errors are not corrected.

Table of Contents

ABSTRACT 4

1. Introduction 5

2. The Rationality of Belief 7

3. Belief and Other Propositional Attitudes 17

4. The Consequentalist Mistake 20

References 24

Notes 27

ABSTRACT

In this paper, I explore the question of whether the expected consequences of holding a belief can affect the rationality of doing so. Special attention is given to various ways in which one might attempt to exert some measure of control over what one believes and the normative status of the beliefs that result from the successful execution of such projects. I argue that the lessons which emerge from thinking about the case of belief have important implications for the way we should think about the rationality of a number of other propositional attitudes, such as regret, desire, and fear. Finally, I suggest that a lack of clarity with respect to the relevant issues has given rise to a number of rather serious philosophical mistakes.

1. Introduction

It is a philosophical commonplace that beliefs can be evaluated according to different standards. The evaluative concepts applicable to beliefs that have dominated the attention of philosophers--those of being certain or uncertain, warranted or unwarranted, and the like--are typically employed in the making of epistemic evaluations. But we can--and not infrequently do--evaluate beliefs practically as well asepistemically : that is, we can evaluate a given belief with respect to the consequences, or the expected consequences, of its being held, or with respect to how its being held would affect the achievement of some desired aim.

Consider contemporary debates over relativism.[ 1] One set of issues concerns the probative force of considerations that are adduced for and against various relativist theses. Is relativism self-refuting? Is there evidence for relativism, evidence afforded (as is sometimes claimed) by disciplines such as cultural anthropology? We might construe these debates as debates about whether someone who was acquainted with the relevant considerations would beepistemically rational in believing the contested theses.

A second set of issues--not always distinguished from the first set with sufficient care--concerns the consequences of belief in relativism. Thus, critics of relativism often condemn belief in relativist doctrines because of the bad consequences (e.g., the erosion of moral standards) which, they allege, would inevitably ensue if such belief became widespread. On the other hand, defenders of relativism often extol belief in relativist doctrines because of the good consequences (e.g., the fostering of greater tolerance among diverse groups) which, they allege, would result from belief in such doctrines. Here, both parties to the dispute are engaged in the practical evaluation of belief in relativism--they simply disagree over the question of whether the consequences of such belief would be good or bad on the whole.[ 2]

The conviction that consequences of a certain sort would result from a given belief's being held by others might lead one to attempt to propagate that belief in society, or, alternatively, to hinder its propagation. No doubt, various interesting ethical issues arise with respect to such possibilities. My concern though, is with the fact that one can evaluate one's own beliefs practically as well asepistemically , and in particular, with the question of whether such evaluations can make a difference to what it is rational for one to believe. It is uncontroversial, I take it, that one can evaluate one's own beliefs practically as well asepistemically : an athlete, for example, might realize that holding optimistic beliefs about her abilities would result in better performances than holding pessimistic beliefs about her abilities. More controversial is the suggestion that a realization of this sort might make a difference to what it is rational for the athlete to believe about her abilities. Can the expected consequences of holding a belief make a difference to whether it is rational to hold that belief? The core of the present paper consists of an exploration of this issue. In the remainder of the paper, I attempt to apply the lessons which emerge from that exploration. First, I argue that the lessons which emerge from thinking about the case of belief have important implications for the way we should think about the rationality of a number of other propositional attitudes, attitudes such as regret, desire, and fear. I then suggest that a lack of clarity with respect to the relevant issues has given rise to a number of rather serious philosophical mistakes.

2. The Rationality of Belief

Beliefs resemble actions in at least two theoretically interesting respects. First, token beliefs, like token actions, can be evaluated with respect to their rationality. Secondly, as we have already emphasized, token beliefs, no less than token actions, can be evaluated with respect to their expected consequences. It is curious then, that while the expected consequences of performing an action are ordinarily taken to be highly relevant to the rationality of that action, the expected consequences of holding a belief are ordinarily taken to be irrelevant to the rationality of that belief. That the expected consequences of performing an action are highly relevant to the rationality of that action--this will be conceded even by those who contend that the rationality of an action is not entirely determined by its expected consequences (other factors are also relevant), as well as by those who contend that, even if the rationality of an action is entirely determined by its expected consequences, one need not have performed the action with the best expected consequences in order to have acted rationally.[3] Indeed, perhaps the paradigm of an irrational course of action is a course of action which frustrates the agent's aims in foreseeable and predictable ways. In contrast, the rationality of a belief seems to depend not on its expected consequences but rather on its epistemic status. Our paradigm of an irrational belief is not that of a belief which predictably leads to the frustration of the believer's goals, but rather that of a belief which is held in the face of strong disconfirming evidence. An athlete who has an overwhelming amount of evidence that she is unlikely to do well, and bases her belief that she is unlikely to do well on that evidence, would seem to qualify as a rational believer--even if her rational belief frustrates, in foreseeable and predictable ways, her goal of doing well. (Indeed, in such circumstances her rationality would seem to be part of her problem.) What accounts for this asymmetry?

Of course, some philosophers have maintained (or at least, have been interpreted as maintaining) that the expected consequences of holding a belief are relevant to the question of whether it is rational to hold that belief. Perhaps the best known application of the idea is Pascal's Wager. In a passage which Ian Hacking has described as "the first contribution to what we now call decision theory",[ 4] Pascal recommended belief in God on the grounds that the expected utility of being a religious believer is higher than the expected utility of being a non-believer. Compare the justification that Kant offers for belief in the "postulates of practical reason". According to Kant, the proposition that human beings have free will (for example) is a proposition for which we lack, and indeed must inevitably lack, the kind of epistemic grounds that metaphysicians have often sought to provide. Nevertheless, belief in this proposition is rational inasmuch as its being believed is a necessary condition for full obedience to the moral law.[5] More recently, contemporary philosophers such as RobertNozick and Richard Foley have held that the expected consequences of holding a belief are relevant to its rationality.[6] Such philosophers face the task of delineating the respective roles that epistemic and practical considerations play in determining the status of a given belief as rational or irrational. On the other hand, those who reject the idea that practical considerations can make a difference to what it is rational to believe face the task of explaining why this should be so. Given that both actions and beliefs can be evaluated practically, why should practical considerations matter in the one case but not in the other?

One attempt to account for the putative asymmetry appeals to the apparent psychological inefficacy of practical considerations with respect to beliefs. The mere realization that my believing some proposition would issue in good consequences does not result in my believing that proposition. On the other hand, the realization that I have strong evidence that some proposition is true typically does result in my believing that proposition. With respect to beliefs, practical considerations seem to be psychologically impotent in a way that epistemic considerations are not. And it is tempting to conclude from this that practical considerations are irrelevant to a belief's rationality.

Compare the situation with respect to height. One can, of course, makejudgements about the expected consequences of being a certain height. For example, I am confident that I am considerably better off, on the whole, being as tall as I actually am as opposed to being two feet shorter. Still, no one would think that it is more rational for me to be some heights rather than others. Moreover, it's plausible to suppose that the reason why the expected consequences of my being a certain height make no difference to whether it is rational for me to be that height derives from my utter lack of control over my height. (Perhaps if I could control my height, then it would be more rational for me to be some heights rather than others.) Perhaps then holding a given belief is like being a certain height: although one can evaluate the expected consequences of holding that belief, such evaluations make no difference to whether it is rational to hold that belief.

There is a standard response to concerns about the psychological impotence of practical considerations, a response which dates back to Pascal. Immediately after presenting his Wager Argument, Pascal has his interlocutor respond as follows:

Yes, but my hands are tied...I am not free...I am so made that I cannot believe. What do you want me to do then?[ 7]

After acknowledging the force of this concern, Pascal offers his interlocutor some advice. Roughly put, the advice is this: one should act in ways that are conducive to the acquisition of religious belief. Pascal's idea is that what one believes is, to a considerable extent, determined by what one does. And inasmuch as one can control what one does, one can exert a certain amount of indirect control over what one believes. Since having been first articulated by Pascal, this Standard Response to concerns about our lack of control over our beliefs has been embraced by many.[ 8]

The possibility of influencing one's beliefs indirectly is often noted but seldom explored at any length. For this reason, I want to examine some of the more interesting ways in which one might engage in such a project. The cases in which I am interested have the following form: (i ) An individual knows that her believing some proposition (which she does not now believe) would have good consequences, so she deliberately undertakes a course of action which results in her acquiring the desired belief, or (ii) An individual knows that her believing some proposition would have bad consequences, so she deliberately undertakes a course of action which results in her not holding the undesirable belief. Not only are projects of this sort possible, they are, I think, not infrequently pursued in the course of everyday life. Moreover, contrary to what is sometimes claimed, successfully completing such a project need not require any self-deception.

In some cases, one can acquire the belief that p is the case by changing the world so that p is the case. This point is noted by RichardFeldman[ 9] who provides the following example: if I'm offered a large sum of money if I acquire the belief that the lights are on in my office, then I can win the money simply by turning my office lights on. Given the highly contrived nature of Feldman's example, it's natural to think that the possibility of manipulating one's beliefs in this way is uncommon, or to think that it is philosophically uninteresting. (Feldman himself notes the point only in passing, and quickly sets it aside as irrelevant to the main issue with which he is concerned.) Neither of these natural thoughts is correct.

Manipulating one's beliefs in this way is not uncommon--in fact, a fair bit of human behavior is best explained by attributing projects of the relevant sort to the agents in question. Consider, for example, the countless number of people who exercise regularly and carefully monitor their diets in order to avoid being overweight. No doubt, there are many reasons why people act in this way, but one fairly common motivation seems to be this: if one is overweight, then that makes it considerably more difficult to avoid believing that one is overweight. And for many, the psychological costs of believing that one is overweight are (or would be) considerable. Given this, one might--and, I suspect, many do--avoid being overweight at least in part in order to avoid having the belief that one is overweight.

Consider, more generally, self-respect--according to John Rawls, the most important of the primary goods.[ 10] Plausibly, self-respect requires that one hold certain beliefs about oneself--or at least, it requires that one not hold certain beliefs about oneself. Someone who believes that his or her activities are utterly worthless does not possess self-respect. One might then act in worthwhile ways in order to maintain (or acquire) the belief that one's activities are worthwhile, and (thereby) achieve self-respect.

The possibility of deliberately acquiring desired beliefs about oneself by acting in ways designed to result in the holding of those beliefs raises interesting issues for moral psychology. Consider, for example, the proper role of conscience in moral motivation. In explaining why he refrains from performing an immoral action, an individual might say: "I couldn't live with myself if I did that". One might avoid performing immoral actions in part because of the psychological costs of believing that one has performed such actions (or perhaps: the psychological costs of believing that one is the sort of person who performs actions of the relevant kind). To what extent do such "impure" motivations detract from the moral praiseworthiness of an agent? Kant held that only conduct performed out of respect for the moral law possesses moral worth; to the extent that a person refrains from immoral behavior in order to avoid the sanctions of a guilty conscience, his conduct lacks moral worth. There is, however, this much to be said on behalf of such an individual: he is at least the sort of person who would be bothered by the belief that he has acted immorally. (Contrast a person who would be undisturbed by this belief, or worse, one who would perversely derive pleasure from it.)

Many other examples in which the desirability of believing that p is the case provides a powerful motive for acting so that p is the case could be provided; I confine my personal favorite to an extended footnote.[ 11]

Insofar as I amepistemically rational, the acquisition of strong evidence that p is true will result in my acquiring the belief that p is true. In some cases then, I can acquire a desired belief by exploiting my epistemic rationality: I deliberately manipulate my epistemic position, acquiring evidence for the relevant proposition, knowing that belief will follow. Cases in which my means of acquiring evidence that p is true consists in my acting so that p is true are simply special cases of this more general possibility. For even if one cannot affect whether p is true, one can often deliberately acquire the belief that p by acquiring evidence for p, evidence that one would not have acquired if one was not motivated to believe that p. Again, the possibility of engaging in such behavior is one that is routinely exploited in the course of everyday life, and again, our doing so need not involve any self-deception. A nervous parent, unable to sleep because he or she does not believe that his or her child has arrived safely, will often acquire the desired belief by means of a simple phone call.

Consider the neo-Pascalian project of maximizing the expected utility of one's beliefs. One

might have thought, offhand, that this is not the sort of project that could be fruitfully pursued by an individual who is perfectlyepistemically rational, i.e., an individual who, at any given time, believes all and only those propositions which it isepistemically rational for her to believe at that time. But it should now be clear that this natural thought is mistaken. In particular, an individual who is perfectlyepistemically rational might take into account the expected utilities of holding certain beliefs in deciding which questions to investigate further. Such an individual could adopt a policy of investigating topics with respect to which the present state of the evidence is unfavorable (in the hope that further investigation will alter the state of the evidence in desirable ways), while choosing not to actively investigate topics with respect to which the present state of the evidence is favorable. We would expect that such a policy, if consistently followed, would skew the individual'sdoxastic corpus in the direction of beliefs with greater utility--while preserving the impeccable epistemic rationality of that corpus at every given moment. Perhaps surprisingly, allowing considerations of expected utility to play a role in determining one's beliefs is consistent with the highest degree of epistemic purity.[12] [13]

Because human beings are largely (although imperfectly)epistemically rational, the strategy of acquiring a desired belief by means of acquiring evidence for that belief will often be the most effective strategy available. Of course, one can only acquire evidence for a desired belief if in fact evidence for that belief is forthcoming. This is just to say that project of acquiring a desired belief by means of acquiring evidence for it will--like any other project--be frustrated if the world proves sufficiently recalcitrant. In cases in which evidence for the desired belief is not forthcoming, one might, of course, resort to alternative strategies. Thus, Pascal advised his interlocutor to thoroughly immerse himself in a life of religious ritual and to imitate, in every way, the behavior of those who do believe. By doing so, the non-believer might hope to become more "docile",[ 14] and hence, more susceptible to religious belief. In a similar vein, Alston writes of "selective exposure to evidence, selective attention to supporting considerations, seeking the company of believers and avoiding non-believers, self-suggestion, and more bizarre methods like hypnotism".[ 15] It is an empirical question how much control we might exercise over our beliefs by utilizing strategies such as these--and one whose answer might very well change, if more effective strategies are developed, or currently available strategies are refined.

Can the expected consequences of holding a given belief make a difference to the rationality of holding it? In answering this question, we should distinguish carefully, as Eugene Mills hasurged[ 16] , between the assessment of a token belief and the assessment of the actions which are intended to result in the acquisition of that belief. Just as it might be rational to cause oneself to actirrationally[ 17] , it might be rational to cause oneself to hold an irrational belief. That is, the mere fact that a token belief is the intended result of a course of action which it was rational to undertake does not guarantee that the belief is rational. But can the expected consequences of holding a belief make a difference to the rationality of the belief itself? Taken in one straightforward sense, this question should be answered in the affirmative: the expected consequences of holding a belief can make a difference to whether it is rational to hold that belief.

Indeed, we have already seen that this is so. Upon realizing that my believing some proposition (which I presently neither believe nor have any evidence for) would be practically advantageous, I might set out to acquire evidence for that proposition. Suppose that my efforts are successful: I subsequently acquire a significant amount of evidence that the relevant proposition is true. Even those who contend that believing a proposition is rational if and only if the believer possesses a significant amount of evidence that the proposition is true will agree that, in these circumstances, my holding this practically advantageous belief would be rational. Moreover, they will also presumably concede that there is a clear sense in which the expected consequences of holding this belief has made a difference to whether it is rational for me to hold it: I would not now possess evidence for the belief if I had not realized that my holding it would be practically advantageous. That is, even if the rationality of a given belief is wholly determined by its epistemic status, the expected consequences of holding the belief might make a difference to whether it is rational to hold it, simply because its epistemic status might be historically dependent on the expected consequences of its being held. In this sense then, the expected consequences of holding a belief can make a difference to the rationality of holding it.

However, those philosophers who have defended the idea that the expected consequences of a belief can make a difference to its rationality have meant something considerably stronger than this. Such philosophers have held that practical considerations can make a difference to whether it is rational to hold a given belief even when practical considerations do not make a difference to the epistemic situation of the believer. For example, Foley claims that

All things being considered, it can be rational for an individual to believe what it is notepistemically rational for him to believe.[ 18]

And this is because it might be clear, in a given case, that holding anepistemically irrational belief would best promote one's goals.

Here, the suggestion seems to be that practical considerations can rationalize beliefs, in much the way that practical considerations can rationalize actions.[ 19] Suppose then that we interpret the question of "Can the expected consequences of holding a belief make a difference to whether

it is rational to hold that belief?" so that it is equivalent to the question: "Can practical considerations ever rationalize the holding of a belief?" So interpreted, the question should, I think, be answered in the negative: practical considerations do not rationalize beliefs. With respect to beliefs, rationality just is epistemic rationality.

Asa first step towards seeing why this is so, consider a distinction familiar from both epistemology and action theory: the distinction between

(1)some individual's F-ing while having a reason R to F, and

(2)some individual's F-ing for reason R

That the first state of affairs (1) obtains does not guarantee that the second state of affairs (2) obtains. For example, I might have strong epistemic reasons to believe that today will be a bad day, believe that today will be a bad day, but not hold the belief for those strong epistemic reasons. (Suppose that I hold the belief simply because of an unduly pessimistic temperament.) When an individual not only has a reason to F, but in fact F-s for that reason, we say that her F-ing is based on that reason. The basing relation is the relation which obtains between a reason R and a token belief (or a token action) when the belief is held for that reason (or the action is performed for that reason).

It is because of the gap between (1) and (2) that one can beepistemically irrational in holding beliefs that are overwhelmingly supported by one's evidence, and practically irrational in performing actions that one has overriding practical reasons to perform. Inasmuch as I believe that today will be a bad day solely because of an irrationally pessimistic attitude towards life, my belief isepistemically irrational. Even though I in fact possess compelling epistemic reasons for this belief, my belief is not rationalized by these reasons because it is not based on them.

How the basing relation should be analyzed is extremely controversial.[ 20] Some philosophers advocate causal analyses of the basing relation. According to such analyses, an individual's

F-ing is based on a reason R just in case the individual's recognition that R plays the appropriate causal role in his or her F-ing . Other philosophers reject causal accounts of the basing relation and attempt to develop alternatives. Fortunately, we need not enter into this dispute, for the point which is essential for our purposes is conceded by both sides. The point in question is the following. Even if

one's recognition that R plays an essential role in the causal history of one's F -ing

this is not sufficient for

one's F -ing to be based on R.

Even those philosophers who seek to analyze the basing relation in causal terms admit that not just any causal relation is sufficient; the causal relation must be of "the right sort".[ 21] Suppose, for example, that my recognition that I have strong evidence that today will be a bad day leads me to consult my horoscope--in the hope, perhaps, of being told otherwise. Suppose further that upon consulting my horoscope, I immediately forget about my original evidence, but believe that today will be a bad day simply because this is what my horoscope portends. In this case, my belief is not based on my original evidence--although my possession of that evidence plays an indispensable role in my coming to hold the relevant belief.

Notice that in cases in which one succeeds in acquiring a desired belief indirectly, the acquired belief is not based on one's recognition that the belief would be practically advantageous to hold.If my desire to hold a belief results in its acquisition via the acquisition of evidence which supports that belief, my belief is not based on practical considerations: rather, it is based on the newly-acquired evidence. Of course, my recognition that the belief in question would be practically advantageous might very well have played an indispensable role in the causal history of my coming to hold it. But from this it does not follow that the belief is based on that recognition. (In contrast, the actions which lead to the acquisition of the relevant evidence are based on my recognition that the belief would be practically advantageous.) Similarly, if, having been convinced by Pascal, an agnostic undertakes a program of religious self-indoctrination, then her undertaking this program is based on the belief that the expected utility of being a religious believer is greater than the expected utility of not being a religious believer. But if the program ultimately succeeds, the newly-arrived at belief that God exists will not be based on this belief about the expected utility of religious belief.

Here, I think, is why practical considerations do not rationalize beliefs. Although practical considerations can make a difference to what one believes, they do not do so by constituting grounds on which beliefs are based. (Contrast the way in which practical considerations do constitute grounds on which actions are based, and epistemic considerations constitute grounds on which beliefs are based.) And rational beliefs, like rational actions, are rationalized by those considerations on which they are based.[ 22]

It is tempting to think that the reason why practical considerations do not rationalize beliefs is the following: insofar as practical considerations can influence one's beliefs, their influence is indirect--that is, their influence is mediated through intervening actions. Epistemic considerations, on the other hand, can influence beliefs directly. But the distinction between "direct" and "indirect" is not, I think, crucial here. In general, S's F-ing might be based on a reason R (and hence rationalized by R) even if S cannot F directly, but can only F by performing some intervening action which leads to her F-ing . Consider the relationship between basic andnonbasic actions. It is often the case that my F-ing is rationalized by my recognition that R, even though I cannot respond directly to my recognition that R by F-ing , but can only respond by performing some series of actions A1...An which culminate in my F-ing . For example, my recognition that my young child would be better off if she attends a good university might rationalize my ensuring that she attends a good university, although I can only do this by setting aside money each month, working with her on her academics, and performing various other actions over a period of years. Thus, the fact that practical considerations influence beliefs only indirectly (i.e., through intervening actions) does not suffice to show that they do not rationalize those beliefs. The important contrast is not between "direct" and "indirect"; rather, the important contrast is the contrast between the kinds of considerations on which a given response can be based and the kinds of considerations on which it cannot.

Given the centrality of the basing relation to the preceding claims, it would be useful to have an account of it which renders theoretically perspicacious the intuitive difference between (1) R's being a reason on which S's belief that p is based and (2) R's playing a role in the history of S's believing that p. Otherwise, how can we be sure that, e.g., the belief that God exists is not based on practical considerations, in those cases in which it is in fact arrived at inPascalian fashion? Ideally, we would like an analysis of the basing relation, in the form of necessary and sufficient conditions for S's belief that p to be based on R. Unfortunately, I have no analysis of the basing relation to offer.[ 23] We do not require an analysis, however, in order to note a crucial mark of the distinction between (1) and (2). Whether R is a reason on which S's F-ing is based, or whether R merely plays a role in the history of S's F-ing will be reflected in the conditions under which S would (or would not) continue to F. If my belief that Smith is rich is based on my belief that Smith has won the lottery, and I subsequently discover that Smith has not won the lottery,then I will (all else being equal) abandon the belief that Smith is rich. Contrast this to a case in which my belief that Smith has just won the lottery plays an indispensable role in my coming to believe that Smith is rich, but is not the grounds on which my belief is based:

Having been told (falsely) that Smith has just won the lottery, I immediately drive to his home in order to congratulate him and to ask for money. Upon my arrival, I see that Smith lives in a great mansion and realize that he has been extremely wealthy all along. When Smith informs me that the lottery story is false, I do not abandon my belief that he is rich.[ 24]

Imagine an agnostic who, having become convinced that the expected utility of being a religious believer is higher than the expected utility of not being a religious believer, undertakes a project designed to induce religious belief. The agnostic thoroughly immerses herself in a life of religious ritual, seeks out the company of religious believers while scrupulously avoiding that of non-believers and (following Pascal's advice) imitates in every way the behavior of those who do believe.[25] In time, she genuinely becomes convinced that God exists. Suppose further that a tragic irony subsequently ensues: the expected utility of belief in God suddenly and dramatically changes. (A despot bent on persecuting religious believers unexpectedly seizes power.) Even if she recognizes that the expected utility of being a believer is now lower than the expected utility of being a non-believer, this recognition will typically not prompt the abandonment of the newly-acquired belief. (Although it might, of course, prompt an anti-Pascalian project ofdeconversion .) Here, the fact that the belief is not abandoned in response to the change in expected utility indicates that the belief is not based on considerations of utility.

Now let us alter the example slightly. In the altered version, the despot seizes power at a somewhat earlier time--the agnostic has begun the project of acquiring belief in God, but the project has not yet reached fruition. Upon recognizing that the expected utility of being a religious believer is now lower than that of being a non-believer, she simply discontinues the project. Here, the fact that she discontinues the project in response to the change in expected utility indicates that her participation in the project is itself based on considerations of utility. The considerations on which a given belief (or course of action) is based are revealed by the circumstances which would prompt one to abandon that belief (or course of action).

But isn't the claim that beliefs cannot be based on practical considerations undercut by familiar psychological phenomena such as wishful thinking? However, the acknowledged possibility of wishful thinking should be carefully distinguished from the (alleged) possibility of basing one's beliefs on practical considerations. To a first approximation: wishful thinking involves holding a belief because one thinks that things would be better if that belief were true. This, of course, is quite different from holding a belief because one thinks that things would be better if one held that belief. (Indeed, perhaps the most objectionable thing about engaging in wishful thinking is that doing soso often leads to bad consequences, in foreseeable and predictable ways.)

Even if one accepts the claim that practical considerations cannot rationalize beliefs because beliefs cannot be based on practical considerations, one might well wonder: Why can't beliefs be based on practical considerations? This is a fair question, and one for which I do not have a confident answer. Nevertheless, I offer the following hypothesis. The reason why actions but not beliefs can be based on practical considerations is simply this. Actions are not beliefs. One of the central features whichdistinguishes a given response as an action rather than a belief is that it is the kind of response which can be based on practical considerations. Conversely, one of the central features thatmakes a given state a belief--as opposed to an action, or some other kind of propositional attitude--is that it is the kind of response which can be based on epistemic considerations but not on practical considerations. That is, it is part of the nature of belief that beliefs are states which can be based on epistemic considerations but not on practical considerations. (Compare the way in which we would answer someone who wondered why actions could not be based on epistemic considerations.[ 26] )

In this section, I have argued that practical considerations do not rationalize beliefs. I do

not want, however, to exaggerate the probative force of the argument on offer. I take the dialectical situation to stand as follows. The apparent psychological inefficacy of practical considerations with respect to beliefs constitutes a prima facie case that practical considerations do not rationalize beliefs. The Standard Response seeks to undermine this prima faciecase, by showing that (in at least some instances) practical considerations are psychologically efficacious with respect to beliefs. I have attempted to undermine the Standard Response by showing that the kind of psychological efficacy which it attributes to practical considerations, though real and not uncommon, is essentially irrelevant to the question of whether such considerations can rationalize beliefs. Although this kind of argument is obviously insufficient to prove that practical considerations do not rationalize beliefs, it does, I believe, effectively shift the burden of the argument back to the side of those who would defend the opposite claim. In the remainder of the paper, I want to consider some of the consequences that would follow in the event that this burden proves too heavy to discharge.[ 27]