Practical Interests and Reasons for Belief

P R A C T IC A L I N T E R E S T S A N D R E A S O N S F O R B E L IE F Baron Reed [email protected] In this paper, I examine the relationship between the practical and the epistemic. I reject two broad ways of thinking about that relationship— pragmatic encroachment and an epistemology centered on the truth norm— before offering a new approach, which explains epistemic normativity as arising from our practical commitment to a social practice that has arisen from our need to share information with one another. The social practice view captures the importance of knowledge and epistemic reasons to action, while preventing our practical interests from playing a disruptive role in how we arrive at our beliefs. ABSTRACT KEYWORDS reasons, rationality, knowledge, normativity, pragmatism, truth In a fair judicial system, judges and juries are impartial: they render their verdicts entirely on the basis of the facts in the cases they consider, without having anything personally at stake in the outcome. In the same way, it is often thought that inquirers in general ought to form their beliefs in an impartial way. Even when they do have something personally at stake, they ought to set those practical interests aside as they form an objective view of the reality they are confronted with. As David Hume said, “A wise man proportions his belief to the evidence.”1 Belief should not be fitted to what one wants to be true—to what would be profitable or useful to believe—but rather to what is true. If this is right, it means that belief cannot be made rational by practical reasons. Although having a particular belief can advance or impede your practical affairs, you cannot take those practical matters into account when deliberating about whether the proposition in question is true. This line of thought is developed in a variety of ways, but it is especially prominent in the debate over the norm or aim of belief.2 The main alternatives take belief to be directed toward truth or 1 Enquiry Concerning Human Understanding, section X. See also Clifford (1877). 2 See Williams (1973) for an influential early statement of the claim that belief aims at truth. The current literature on norms of belief is 2 knowledge or justification. Some philosophers will argue that belief answers to several different aims or normative standards; in addition to one or more of the above list, belief might also aim at understanding or wisdom. But, on any of these views, it looks like the epistemic and the practical are entirely different domains. Although this is a well-entrenched tradition in epistemology, there is an alternative picture, grounded in a kind of pragmatism, that takes the epistemic and the practical to be fundamentally linked. On this view, as C.I. Lewis says, “The primary and pervasive significance of knowledge lies in its guidance of action: knowing is for the sake of doing.”3 It is only in exploring the connection between knowledge and action that we can provide an explanation of the value or importance of knowledge. And some philosophers have further argued that this connection shows the pragmatic encroaches on the epistemic. 4 Suppose that knowledge gives you everything, from an epistemic point of view, you need for action.5 That is to say, if you have knowledge, you can rationally act on it. Put another way, if it is irrational for you to act on a particular belief, you do not count as knowing the proposition in question. Understood in that way, practical rationality is necessary for knowledge. And so, on this view, knowledge is deeply entangled with our practical interests. There is something to be said for each of these broad conceptions of epistemology, and yet they are in conflict with one another. As I shall argue in this paper, pragmatic encroachment should be rejected—it has implications that are simply unacceptable. But pragmatists are right to worry that the impartiality picture of epistemology is lacking something important: it does not explain why truth, knowledge, or evidence matter enormous, but see especially Chan (2013) and Gibbons (2013). 3 See, e.g., Williamson (2000). Lewis (1946), p. 3. See Craig (1990) and Williamson (2000) for more recent expressions of this sort of view. 4 See especially Fantl and McGrath (2002, 2009), Hawthorne (2004), Stanley (2005), and Hawthorne and Stanley (2008). The thesis of pragmatic encroachment is sometimes also called “interest-relative” epistemology or “anti-intellectualism.” 5 There may, of course, be non-epistemic impediments to action. It may be immoral, imprudent, or impolite to act on what one knows. Someone who acts in those circumstances may deserve criticism of various kinds, but epistemic criticism will not be appropriate. 3 to us. Even if we grant that truth is the norm or aim of belief, we may still wonder why that norm is compelling to us. An answer to this question can be given, I argue, without permitting the practical to encroach on the epistemic. The solution is to recognize that we have practical reason to engage in a broad epistemic practice. What we have reason to believe is determined by that practice, and the epistemic practice matters to us because it serves our deepest, most important practical needs.6 1. PRAGMATIC ENCROACHMENT There are three major considerations used to motivate pragmatic encroachment. First, its defenders argue that pragmatic encroachment provides the best explanation of our evaluative reactions to pairs of cases like the following7: TRAIN CASE 1: Matt is in Boston, waiting for the train that goes to Providence. He is out for a day of leisure and doesn’t mind how long it takes to get to Providence. He recalls having read the train schedule a few days ago and believes that the next train makes a stop in Foxboro on the way to Providence. TRAIN CASE 2: Jeremy is in Boston, waiting for the train that ultimately goes to Providence. He is running late for a job interview in Foxboro, and it is extremely important that he get there as quickly as possible. He recalls having read the train schedule a few days ago and believes that the next train makes 6 This paper brings together and develops two lines of argument begun in other work. For the rejection of pragmatic encroachment, and its replacement with stable invariantism—a view that takes knowledge to be independent of practical stakes but flexible enough to capture gradations in the strength of one’s epistemic position, see Reed (2010, 2013a, and 2013b). For the view that we have practical reasons to take epistemic reasons seriously, see Reed (2015). 7 See Fantl and McGrath (2002). See also the bank cases in Stanley (2005); the bank cases are adapted from DeRose (1992), who uses them in defense of contextualism. See also Cohen (1999) for a similar case used on behalf of contextualism. 4 the stop in Foxboro. If our conception of knowledge were indifferent to practical matters, we should have the same reaction to both cases; either Matt and Jeremy both know the train stops in Foxboro, or neither of them does. But this is not the usual reaction people have to these cases. Typically, people judge that, while Matt knows the train will stop in Foxboro, Jeremy does not. These differing judgments are perhaps grounded in the thought that Jeremy would be irrational to simply rely upon his memory, without further confirming that the train does in fact stop in Foxboro. By contrast, there is no need for Matt to look for further confirmation in relying on his belief (e.g., in responding to a fellow passenger’s idle question about whether the train makes any stops on the way to Providence). Second, attributions of knowledge and ignorance are often used in defending and criticizing actions. Here is an example of knowledge being used in defense of an action: A: Why are you re-arranging the guests at the dinner party? B: I know that Thomas and Stuart will start arguing about politics and ruin it for everyone else. Here is an example of ignorance being used in criticism of an action: C: You shouldn’t turn off the lights yet—you don’t know that everyone is out of the theater! Of course, ignorance can also be used in defense of an action, and knowledge can be cited in criticism. Third, pragmatic encroachment makes possible a “ledger-keeping” conception of how knowledge is used in reasoning.8 According to this view, reasons are simply weighed against one another without any need to pay attention to how probable it is that those reasons are true. The idea here is that reasons are safe in the sense that, when you have a 8 See Fantl and McGrath (2009), pp. 77-82. The term ‘ledger-keeping’ is borrowed from Lewis Carroll. 5 reason, you can rely on it to justify you in doing anything.9 Knowledge is a safe reason, if pragmatic encroachment is correct, because a belief no longer counts as knowledge when it is practically irrational to rely upon it. The ledger-keeping conception of reasoning fits well with our practice of defending actions with knowledge claims. In answering a challenge by saying that I know some relevant bit of information to be true, it looks as though I am trying to convey that it is safe to rely upon it. What I do not do is weigh what I know to be true against the possibility that I might be wrong. Notice how the following conversation sounds odd: A: Why are you rearranging the guests at the dinner party? B: Thomas and Stuart will fight about politics, like they always do. A: They might not—there’s always a chance they’ll find something more congenial to discuss. B: You’re right. I know they will talk about politics. On the other hand, there’s a chance they won’t. As Fantl and McGrath argue, we do not mix consideration of reasons and their underlying probabilities in the same stretch of deliberation. 2. OBJECTIONS TO PRAGMATIC ENCROACHMENT A number of objections to pragmatic encroachment have been raised. The most straightforward of these are counterexamples in which a subject faces very high practical stakes but nonetheless plausibly retains her knowledge. Jessica Brown, for instance, points out that a surgeon who is preparing to amputate a limb knows at the outset which limb is to be removed, but she will still double check the charts just before cutting into the patient.10 If fallibilism is correct—that is to say, if it is possible to have knowledge that falls short of certainty—then knowing something leaves open a small chance of being wrong. In some high stakes situations, that small risk of error becomes practically important, even as the subject continues to know the proposition at issue. 9 Fantl and McGrath (2009), p. 77. 10 Brown (2008). See also PUNISHMENT/REWARD CASE 1 in Reed (2010). 6 Defenders of pragmatic encroachment can respond by denying this intuitive verdict; they can claim, for example, that Brown’s surgeon does not actually know which limb is to be amputated once she has entered the operating room. That seems like an implausible reply, as it is surely more proper to describe the surgeon as reconfirming, rather than relearning, which limb is to be amputated. In any case, the strategy of denying knowledge to everyone in a high stakes situation is even less plausible when the subject simultaneously faces, with respect to the same proposition, both a high stakes and a low stakes decision: PUNISHMENT/REWARD CASE 2: Margaret is taking part in a study of the effects of stress on memory. She is playing two games at the same time, in which she is asked questions about ancient history. In the first game, a correct answer gets her a jellybean, while an incorrect answer will bring a very painful electric shock. In the second game, a correct answer gets her $1000, while an incorrect answer will mean getting a gentle slap on the wrist. In both games, there is no penalty for not answering at all. Margaret is then asked when Julius Caesar was born. She remembers that the answer is 100 BCE, but she is not entirely confident in this memory.11 Let us suppose that Margaret’s memory is good enough that she should give her answer in the second game but not in the first. If we look only at the second game, there is no reason to deny that she knows when Julius Caesar was born; attributing knowledge to her explains why she behaves as she does. If we look only at the first game, on the other hand, then pragmatic encroachment tells us that Margaret does not know when Caesar was born. But, of course, Margaret is playing both games simultaneously. The defender of pragmatic encroachment has to reconcile these competing assessments of Margaret’s belief. When she is taking part in the study, does she know when Caesar was born? To say yes is to abandon pragmatic encroachment. To say no is to break the connection between knowledge and practical rationality; her behavior in the second game will have to be explained by something other than her 11 Reed (2010). 7 knowing when Caesar was born. That something else—justification, perhaps—might then be a better candidate for explaining her behavior in all cases.12 A third alternative is to relativize Margaret’s knowledge to the particular practical situation she is in. Relativizing in this way will allow us to say, without contradiction, that she simultaneously both knows and doesn’t know when Caesar was born. But it comes at a high cost; it means that knowledge is not available, ready for use whenever needed, as a reason that may be safely used. Instead, knowledge is something that arrives on the scene only when it is safe to do so. If this is the case, knowledge cannot play the central role in guiding action that defenders of pragmatic encroachment have thought it did. The same problem can be seen in another objection to pragmatic encroachment. The idea is that allowing the pragmatic to encroach on the epistemic puts THE CART BEFORE THE HORSE. 13 Suppose that Alicia has received an investment tip from her stockbroker: a company called QED is going to double in value in the next six months. Her broker has been a reliable source of this sort of information in the past. If nothing were at stake, Alicia’s belief that QED will increase in value would be sufficiently justified to count as knowledge. But her broker has recommended that she make a substantial investment in the company. Alicia asks her friend, Bea, for advice: A: If QED will double in value within six months, I should invest in it. Should I do it? B: That depends—do you know that it will double in value? A: That depends—should I invest in it? If pragmatic encroachment is correct, then knowing something is 12 Could the defender of pragmatic encroachment respond that the best explanation of Margaret’s action is that she knows that it is likely Caesar was born in 100 BCE? Given that this is another instance of knowledge, we should expect that it is also sensitive to practical stakes and disappears when they are sufficiently raised. (An easy way to see this is to imagine a variant of the case in which Margaret is asked, not when Caesar was born, but when it is likely he was born.) 13 Reed (2013a), pp. 103-4. For a related problem, see also the Dutch book objection at p. 103. 8 sufficient for it being practically rational to act on it. Put in a logically equivalent way, in order to know something, it is necessary that one be practically rational in acting on it. Suppose Alicia were to act on her broker’s advice; in that case, the stakes would be very high, and she would lack the knowledge in question. On the other hand, if Alicia were not to act on her broker’s advice, the stakes relative to the proposition would remain low, and she would know the advice to be correct. What she knows will depend on what she should do, and what she should do will depend on what she has at stake, but what she has at stake will depend on what she does. Until her action is settled, there is no fact of the matter as to what she should do and what she knows. This is a problem because, if knowledge is to be useful, it should be available for use in deliberation—but deliberation can’t begin until the action has been performed. Knowledge enters the picture only after it is no longer needed, and so it would be practically useless if pragmatic encroachment were correct. According to a third objection, if pragmatic encroachment were correct, knowledge would be worse than practically useless—it would undermine one of our basic moral practices. Consider the following case: THE READY EXCUSE: Graham has just arrived at a conference. He remembers, from an email received two weeks earlier, that the conference schedule was recently changed, and the first session begins at noon—ten minutes from now. But he’s tired from traveling, and he would love to visit the sauna. Graham remembers that his department chair is also at the conference; he further remembers that she threatened to cut his funding if she found out that he was skipping conference sessions again. Clearly, a lot rides on his decision to go to the sauna. Because the stakes are so high, Graham’s memorial evidence is not good enough for him to be able to know that the schedule was changed. Graham contentedly lets himself into the sauna, happy that he now has an excellent excuse for missing the first session: he just didn’t know the schedule had been changed. We often excuse one another for actions (or omissions, or consequences of actions or omissions) that would otherwise be blameworthy when the 9 agent doesn’t know some relevant fact. For example, the superintendent of a building might reset the circuit breaker, not realizing that a maintenance worker deliberately tripped it in order to work on an electrical problem in one of the apartments. Resetting the circuit breaker has the unfortunate consequence of electrocuting the worker. This is a bad outcome, but it is not one for which the superintendent is responsible.14 If pragmatic encroachment is correct, Graham doesn’t know the conference schedule has changed; his belief does not meet the elevated epistemic standards that are in place in virtue of the high practical stakes he faces. When the chair of his department begins to angrily criticize him for skipping the first session of the conference, Graham can truthfully say that he didn’t know the schedule had changed. In light of his excuse, her criticism is no longer proper. This is problematic for at least two reasons. First, our ability to fully hold one another responsible in cases where the stakes are high—which are the cases in which responsibility is most important—is compromised. In Graham’s case, he is excused from missing the opening of the conference because his job is at stake; obviously, the effect will be even more pronounced in cases that involve the potential for pain, suffering, or death.15 Second, pragmatic encroachment would 14 To be clear, the superintendent’s ignorance must be warranted. That is, she cannot have been in a position to know that, if the circuit breaker is tripped, it should not be reset without confirmation from all the maintenance workers in the building. So, we can suppose that the building follows a practice of putting a note on the circuit breaker box to this effect whenever someone is doing electrical repairs and that the worker in this case forgot to do so. 15 In other cases, the agent’s responsibility may be diminished but not eliminated. Suppose the highest degree of blame will attach to an agent who performs action A while knowing that p. When the stakes are very high, pragmatic encroachment will mean that the agent does not know that p, even while possessing a significant degree of justification for the belief that p. This is compatible with there being a lesser degree of blame for an agent who performs A without knowing that not-p. For example, a waiter who knowingly serves peanuts to someone with a severe peanut allergy would be doing something utterly reprehensible. 10 allow agents to manufacture their own excuses. If not knowing a relevant piece of information would excuse some particular wrongdoing, an unscrupulous agent could ensure that he does not have that knowledge, simply by raising the stakes at issue. If that strategy became widespread, our practice of giving and accepting excuses would be undermined. More generally, our ability to criticize and defend actions by making attributions of knowledge or ignorance would be impaired. In this way, a universal acceptance of pragmatic encroachment would undercut one of the considerations used to motivate the view. Finally, there is a still deeper problem with pragmatic encroachment—one that brings it squarely into conflict with the ideal embodied in the impartiality conception of epistemology: PRACTICAL PERSUASION: Holly’s boss has complained that someone is stealing her lunches from the company kitchen. Holly is a fundamentally honest person: if she knows who is taking the lunches, she will tell her boss. She accumulates good evidence that Nigel is the culprit—he never seems to bring food to the office but always has something to eat, food wrappers matching the boss’s favorite meals are often found near his desk, etc. At the same time, Holly knows that Nigel’s recommendation is both necessary and sufficient for a major job promotion that she very much wants. She understands that the Pragmatic encroachment will make it turn out that waiters rarely do this knowingly, even in cases where they are fairly well justified in thinking that they are serving peanuts to customers with a peanut allergy. Nevertheless, these waiters would still be guilty of serving food to customers without knowing that it doesn’t contain peanuts. In other words, they would be guilty of negligence but not of deliberate indifference. In THE READY EXCUSE, this is the most for which Graham can be blamed: he was negligent in not being sure about when the conference began, but it looks like he cannot be criticized for deliberately skipping the opening session. (Some philosophers may be tempted to criticize Graham because he was justified in believing that the schedule had changed. But to lodge this sort of complaint is to break the explanatory/justificatory connection between knowledge and action meant to motivate pragmatic encroachment.) 11 boss will fire Nigel if she learns that he has been stealing her lunches, and she also knows that she is very unlikely to get the promotion without his help. Worried that she has begun to suspect him, Nigel tells Holly that he is looking forward to recommending her for the promotion in the coming year. Given the details of the scenario, it is clear that for Holly there are high stakes riding on the proposition that Nigel is the lunch thief. If she comes to know that it’s true, then her honesty will compel her to turn him in to the boss, and she will lose her chance at the promotion. When Nigel offers his assistance with the promotion, it might look as though he is doing something morally wrong—offering her a bribe in exchange for her silence. But, if pragmatic encroachment is correct, that need not be the only way to characterize his action. We can see his offer as an attempt to persuade Holly away from a particular point of view. On the traditional, impartial conception of epistemology, rational persuasion is possible only through the provision of evidence, which can then convince the subject either to disbelieve the proposition or to withhold judgment on it. On a pragmatist conception, however, an interlocutor can accomplish this either by providing evidence or by presenting an altered set of practical stakes. Nigel, perhaps recognizing that he has little hope of giving Holly counterevidence, has opted for the latter course of action. Practical persuasion, as it may be called, has the further advantage of being easily augmented, for those who have the means to do so. Hence, Nigel can strengthen his case by holding out an even greater promotion, with a higher salary and a better parking space, to Holly. In doing so, Nigel has prevented Holly from knowing that he has stolen the boss’s lunches, not by coercing her or corrupting her, but by making it harder for her to meet the requisite epistemic standard for knowledge. When Holly comes to withhold judgment on Nigel’s guilt, this is a rational, epistemic response—if pragmatic encroachment is correct. Given the intimate connection between knowledge and action, this means that Nigel will also have made it harder for her to turn him in to the boss. It will strike many people—correctly, I think—that there is something fundamentally perverse about this sort of practical persuasion. Using one’s wealth and power to bend another’s epistemic point of view to suit oneself is not the sort of thing that should be 12 condoned by epistemologists. The problem for pragmatists is that they don’t have a principled way of ruling it out, once the pragmatic is taken to encroach upon the epistemic. This is excellent reason, then, to think that the connection between the practical interests and reasons to believe cannot be as direct as the pragmatists would have it. 3. THE “AIM” OF BELIEF Where advocates of pragmatic encroachment have argued that practical rationality is essential to knowledge, defenders of the traditional, impartiality conception of epistemology have tried to enforce a strict line between the epistemic and the practical. Views of this sort try to flesh out the idea that, as Bernard Williams noted, beliefs “aim at the truth.”16 One of the most common ways of explaining this seemingly obvious claim is to say that it is true in virtue of the nature of belief. Belief is constituted by the fact that it aims at the truth. And perhaps the most plausible way to understand the metaphor of “aiming” at the truth is to think of belief as being governed by a normative principle like the following: “a belief is correct if and only if the proposition believed is true.”17 Because one’s practical interests typically have no bearing on the truth of a belief (setting aside cases in which one’s belief is about those interests), one’s practical interests are irrelevant to whether the belief is correct. So, practical interests cannot serve as reasons for belief. Nevertheless, it is widely recognized that there can be cases in which it is practically beneficial to have a particular belief. In the classic example from William James, someone who is climbing in the mountains and has no alternative but to attempt to leap over a broad chasm may be in such a position that she will safely make the jump if she believes she can, but she will fail if she doubts she can jump that far. James says, “In this case (and it is one of an immense class) the part of wisdom clearly is to believe what one desires; for the belief is one of the indispensable preliminary conditions of the realization of its object.”18 What one should believe from an epistemic point of view, in this sort of 16 Williams (1973), p. 148. See also, e.g., Wedgwood (2002), Shah (2003), and many of the essays in Chan (2013). 17 Wedgwood (2013), p. 124. 18 James (1897), p. 97. 13 case, is not independent of, but rather follows from, what one should believe from a practical point of view. The contrast between these two points of view can be heightened if we suppose that the climber has ended up in a position where she will certainly fail to make the jump if she doubts she can make it, but she will at least raise the probability of successfully making it to the opposite side if she believes she can make it. In this version of the scenario, believing she can jump the chasm does not make it more likely than not that she will in fact make it across. She still is likely to fall. But having that belief at least gives her the best possible chance of survival. In this sort of case, there appears to be an outright conflict between what is recommended to her by the epistemic point of view (don’t believe you can jump across) and the practical point of view (do believe you can jump across). Do cases of this sort show that the recommendations, or reasons, provided by the practical point of view can be brought to bear on our beliefs in the same way that epistemic reasons do? If so, won’t this lead to situations, like PRACTICAL PERSUASION, that seem to violate our deepest sense of what the epistemic ideal should be? To rule out this possibility, some philosophers have drawn a distinction between different kinds of reasons. One sort of reason is “object-given” or “constitutive.” In the case of belief, it is an evidential consideration, which weighs in favor of the truth of the belief’s content. A reason for belief in this sense is a reason to believe something. The other sort of reason is “state-given” or “extrinsic.” In the case of belief, it is a reason for believing something, even if it is not a consideration that counts in favor of the truth of the content of the belief.19 Returning to the case of James’s climber, then, we can now say that she has a reason for believing she will successfully jump the chasm, even though she does not have a reason to believe she will successfully make the jump. The former is a practical reason; the latter is an epistemic reason. Drawing this distinction does not by itself allow us to determine whether the climber should, or should not, believe that she can jump across the chasm. In 19 See Parfit (2001) for the distinction between object-given and state- given reasons. See Hieronymi (2005) for the distinction between constitutive and extrinsic reasons. See Adler and Hicks (2013) for the distinction between reasons to believe and reasons for believing. 14 other words, it does not show us how practical reasons may be weighed against epistemic reasons. But it does allow us to segregate the two kinds of reasons, so that changes in one’s practical circumstances cannot affect one’s epistemic situation, as in PRACTICAL PERSUASION. Although the distinction between reasons to believe and reasons for believing promises to keep the practical and the epistemic neatly partitioned, a broader perspective reveals its limitations. To see this, notice that merely calling attention to a norm does not at all explain why we care about adhering to it. When Herman Melville lived with the islanders of Nuku Hiva in the mid-19th century, he found that they had an elaborate system of norms of taboo—various people, places, and things would be off-limits to some people but not to others, and perhaps at some times but not at others. These norms were seemingly arbitrary to him; for example, he learned—from the horrified shrieks of the women making it—that it was taboo for a man to touch a fabric called tappa. 20 Although he recognized that these norms of taboo existed, Melville didn’t seem to feel particularly bound by them. To the extent that he followed them, it was out of a desire not to offend his hosts and not because the norms themselves motivated him. (If no one were around, for example, he surely wouldn’t have hesitated to touch the tappa if he were curious about it.) Let’s say that these norms lacked force for Melville but not for the islanders. Now, the question naturally arises: which norms, if any, have force for us? And, more importantly, why do they have it? There is a familiar menu of theories from which to choose. According to realism, it’s a matter of objective fact that some norms simply do have force for us. Expressivists, on the other hand, will say that there is nothing objective here at all; our normative judgments merely express our approval or disapproval of the relevant actions, objects, or mental states. Other philosophers will take those norms that have force for us to have it in virtue of our rational nature. And some, following Hume, will trace normative force ultimately to desire. These views might be more or less plausible, depending on the sorts of norms we are trying to explain. The taboo norms, for example, are surely neither objective nor grounded in our rational nature. In the case of epistemic norms, most philosophers have been attracted to some sort 20 Typee, ch. 29. 15 of Humean account. There are two compelling considerations in its favor. First, it would obviously explain why we do care about following epistemic norms; doing so would presumably allow us to satisfy our fundamental desires. And, second, it would presumably explain why epistemic norms are truth-directed. No matter what our desires are, the thought goes, having beliefs that are true will be quite useful in allowing us to satisfy them. So, even if we do not have a pure desire for the truth itself, having true beliefs will be instrumentally useful in satisfying those desires we do have. But a problem arises for anyone who tries to combine a Humean account with the distinction between reasons to believe and reasons for believing. Suppose that you have a reason to believe something. Why does this move you? That is, why does this epistemic reason have force for you? For the Humean, the answer is that its force stems from the fact that treating it as a reason ultimately serves your purposes. You have a reason to believe—i.e., an epistemic reason—because you have a reason for having that belief—viz., a practical reason. In the simplest case, your goal is to believe the truth. Reason r makes it likely that p, so it is a reason to believe that p. You treat r as a reason—as something that moves you—because it helps you accomplish your goal: having a true belief. And so r is a reason for having a belief. In the more complex case, your goal is to accomplish some end e other than simply believing the truth. If there is evidence that makes it likely that p is true, and having the true belief that p will help you accomplish e, then that evidence gives you a reason r to believe that p. But r moves you because having the true belief that p helps you attain e. So, again, r gives you reason for having the belief that p. In both cases, if your goal changed— if you no longer wanted to have a true belief or you no longer wanted to accomplish e—reason r would no longer move you. It might still be the case that r is a reason to believe that p, but, when r no longer gives you reason for having the belief that p, ignoring r will be no more troublesome to you than breaking the taboo was to Melville. The idea that belief “aims” at the truth—that it is governed by a truth norm—was supposed to explain the distinction between the epistemic and the practical. The distinction between reasons to believe and reasons for believing was supposed to bolster this account, by showing how the epistemic and the practical provide very different sorts of reasons. And the Humean account was supposed to explain how the 16 truth norm has force for us and why epistemic reasons move us. But, as we have seen, this turns out to be a self-defeating set of views; if all of them were correct, it would mean that epistemic reasons matter to us only when they are also practical reasons. And this suggests that the fundamentally perverse sort of manipulation we saw in PRACTICAL PERSUASION has not been excluded after all. A change in your practical circumstances—your goals, stakes, or interests—can bring with it a change in whether you feel bound by the norm of truth. One might object that, whether or not you feel the force of the truth norm, it still is a norm. But that answer doesn’t carry us very far. There are countless norms, like the norms of taboo, the norms of dress in Victorian England, or the food norms of the Pythagorean school, you might be violating at this very moment. A norm without force for you is hardly a norm at all.21 There are also problems for the Humean view and for the truth norm view taken separately. Regarding the former, Thomas Kelly argues that a Humean (or instrumental) account of epistemic reasons fails to capture their apparently categorical nature.22 When you are confronted with evidence for a belief you don’t want to have, you can’t simply ignore it. Regarding the latter, it’s unclear how well the claim that belief by its very nature is governed by a truth norm fits with a naturalistic account of the mind. If our fundamental capacities have been shaped by natural selection, there’s little reason to think that belief responds only to the truth. 23 For example, a monkey that believes there is danger nearby whenever it hears an unexpected noise is likelier to survive than one that stays to confirm the presence of a predator. These aspects of our biological heritage continue to shape human psychology, even in circumstances when they are unhelpful—as may happen, for instance, when one’s judgments about risk are distorted by various belief heuristics. 21 This is why realism about norms—i.e., taking the existence of norms to be a matter of objective fact—is ultimately an unsatisfying (or incomplete) theory. There are many objective facts that we simply disregard in our practical lives. What prevents norms from being among them? 22 Kelly (2003). 23 See Plantinga (1993), ch. 12; Street (2009); and Reed (2015). 17 Taken altogether, these arguments seem to show that there is little hope of grounding an appropriate separation between the epistemic and the practical in the nature of belief or the normativity that applies to it. But I think this pessimistic conclusion can be forestalled. The basic Humean strategy can be made successful if we apply it, not to beliefs, but rather to practices. 4. PRACTICES Social life is often characterized by a need for coordination. For example, cars would have to be driven much more slowly if there were not a set of rules governing where and how fast they may go, how drivers take or defer precedence, and so on. These rules, obviously, can vary from one place to another. Americans drive on the right side of the street, the British on the left. Neither is doing it “wrong,” so long as there is clarity about which rules are in effect.24 Let us say that a practice is an organized way of life that arises in response to a need for coordination. The sort of practice that drivers are engaged in is shaped to a large extent by a set of explicit rules—the traffic laws. There can also be unwritten expectations; e.g., in China, smaller vehicles tend to give up the right of way to larger ones. Practices can be more or less widely shared, and there are some tricky questions about their boundaries. Do American and British drivers engage in the same practice, or are they too different? Is there an Illinois driving practice, distinct from an Iowa driving practice? Practices can also change qualitatively over time while retaining their numerical identity. The maximum speed limit in New Mexico has changed over the years from 65 to 75 miles per hour, though presumably the practice itself has remained numerically the same. On the other hand, the practice as it currently exists is surely different in number from the practice that prevailed when the roads were traveled only by horse-drawn vehicles. Where the need for coordination in driving is fairly specific and explicit, there are also many other kinds of practices that have grown 24 Of course, one can be doing it wrong relative to the rules where one is driving. An American who insists on the rightness of driving on the right side of the road, even in London, is wrong twice over—wrong to think that’s how to drive in London and wrong to think there is an absolute rightness to his own way of driving. 18 into wide-ranging, largely implicit ways of life. The taboo system Melville encountered in the Marquesas, the chivalry of medieval Europe, the standards of politeness in the antebellum American South, and the moral codes we abide by today are all practices of this sort. Where some of these practices seem utterly compelling—like morality, one hopes— others have entirely lost their force. (Chivalry is indeed dead.) These kinds of practices can be distinguished from one another in at least two ways. First, practices may address different coordination problems. Where driving practices coordinate the movements of different vehicles, the taboo practice regulates the interaction of the genders and provides a connection between the sacred and the mundane. Second, practices may address similar coordination problems but in different ways. The customs and traditions that are part of the practice of politeness help smooth interactions with others by, among other things, ensuring that various social ranks and roles are properly respected. Moral codes, on the other hand, also help smooth social interactions, but they tend to do so by steering us toward respecting the moral standing of others. To be sure, these different practices can overlap in ways that make it hard to tell them apart on a particular occasion—being impolite is often an immoral thing to do, given that it can cause suffering to the recipient of the impoliteness—but the differences between them are clear enough in general, given that they tend to work in different ways. Practices cannot be reduced to a set of norms. Although they include norms, practices also have characteristic properties, objects, institutions, symbols, identities, virtues, values, and expectations. All of these phenomena are directed, mediately or immediately, toward the need for coordination, which gives significance to all of them. This is how the norms, values, etc., gain their normative force: where the need for social coordination is compelling, the individual agent has a very strong practical reason to take part in the practice that has grown up to solve it. To be sure, some needs subside over time; the taboo and chivalry practices do not have the same grip on people today as they once did. But some needs are of fundamental and, apparently, unending importance to us. They will persist as long as we continue as social beings. There are several important implications of this account. First, 19 norms do not have force in a piecemeal way. Because they are part of a much broader practice, the context in which they bind us cannot be ignored. The same normative principle may feature in different practices and will, for that reason, have a different meaning in each. For example, the prohibition on killing animals may appear in a religion that takes animals to be sacred, and it may also appear in a moral system that finds animal pain to be morally bad. In the contexts of these two practices, the principle counts as different norms; in each, it is expressive of different concerns, and it will have a different significance in interacting with a different set of values, institutions, and so forth. Call this feature of the view normative holism. Second, if we take a practice to have its identity in virtue of the need for coordination it addresses, there can be conflict between different ways that practice takes shape. The moral code of a small, rural, deeply religious community will look quite different from the moral code that can be found in a large, urban, secular community. These differences can arise for a variety of reasons, including differences in the local environment. For example, marital infidelity will tend to be far more disruptive and damaging in a small community than in a big city, and so there will be a harsher prohibition on adultery in the former than the latter. Communities whose members live at a subsistence level may tolerate infanticide, where this is strictly forbidden in wealthier communities. Conflicts between alternative ways a practice addresses its target need can be difficult to resolve, as it can be easy to confuse the non-essential parts of a practice with the elements of it that are irreplaceable. But reflection on how the different aspects of the practice serve that basic need for coordination is perhaps the best way to peacefully resolve these differences, especially when this reflection is accompanied by an awareness of other ways the practice might be developed. Understanding that infanticide in a subsistence community is the result of a difficult calculation about how many non-productive family members can be supported without endangering the entire family allows us to see both the genuinely moral nature of that part of the community’s moral code and its irrelevance to people outside of that community. And, of course, we can discard parts of our own practice, in cases where they are holdovers from an earlier set of considerations we no longer find compelling, as has happened for many, e.g., who have recently come to see same-sex marriage as morally acceptable. Call this 20 sort of elaboration of a practice normative plasticity. Third, what the agent has practical reason to do is to take part in the practice as a whole. This does not necessarily mean, however, that the guidance the practice provides on a particular occasion will also be practically rational for the agent.25 For example, an elderly man may feel that politeness dictates that he give up his seat on the train for a young woman who has just entered the car. Doing so will mean that he has to stand for an hour—a burden that will be difficult for him to meet. From the practical point of view, he would be better off sitting; that is to say, a narrow focus on his self-interest would not make rational his polite gesture. 26 Call this gap between practical rationality and the guidance delivered by a practice practical indirectness. Fourth, when there is a conflict of this sort, between the guidance of a practice and what is practically rational, the guidance of the practice is not undermined, though it may be overridden. Even if the elderly man decides that the practical cost of giving up his seat is too great, he still recognizes that politeness calls for him to do it. His choice is, in that sense, impolite. The same happens when two different practices offer conflicting guidance on a particular occasion. For example, an eighteenth-century man who is challenged to a duel might feel torn between what the norms of honor require and what morality dictates he do. Whether or not he takes part in the duel, he will fail to abide by one of his practices. He can be moral, in which case his honor will have been tarnished, or he can defend his honor by taking part in an immoral contest. Either way, he will feel the normative force of the practice he is violating. Call this phenomenon normative persistence. 25 Let us say that, in a set of circumstances, a practice guides an agent to a particular outcome—an action, an emotion, a belief, etc. (This is an idealization, of course; a practice may not resolve to a single outcome. For present purposes, though, this complication can be ignored.) 26 There is a complication here: one aspect of a practice that should not be overlooked is the fact that it may provide an identity that is of the utmost importance to the practitioner. In this case, being polite may be more important to the elderly man than the physical discomfort of standing for an hour. If so, then his practical interests are in fact linked to the guidance of the practice in question. My point here is simply that this does not always happen. 21 5. THE EPISTEMIC PRACTICE With this general account of practices, we can now return to epistemic matters. Recall that there is an evolutionary objection to the view that belief by its very nature is governed by a truth norm. If our capacity for belief has been shaped by natural selection, then our beliefs will be optimal when they promote survival and reproductive success. Beliefs that are true will often—but not always—promote those outcomes. This leaves defenders of the truth norm with the burden of explaining how it is that belief comes to acquire that nature. The solution to the problem is not to focus narrowly on the nature of belief but rather on the way it functions in a social setting. Given the tremendous benefits that come with sharing information, we have a need to coordinate our beliefs. When there is a conflict between my belief and yours, we need a way of resolving our difference. Although each of our beliefs may have been produced by idiosyncratic subpersonal processes, attuned as much to speed as to accuracy, these are not suitable for the sort of inter-personal dialogue needed to reach a shared view of reality. In order to sort out our differences, we will each need to be able to articulate what we believe and why it would be good to believe it. The good-making feature of our beliefs must be accessible to both of us to play this role. The shared standard we use is the fit our beliefs have with the world around us. In the simplest case, you can simply point to that part of the environment that makes your belief true. In more complicated—and cognitively more powerful—cases, you present something that indicates your belief is likely to be true. Much more could be said about the various ways in which an enormously complex epistemic practice has grown up in response to this need for coordination. 27 What is relevant here is that Williams’s platitude is not quite right; it is not belief that aims at truth but we who do so—together—in following our epistemic practice. To the extent that culture is independent of biology, this account allows us to see how we can share the capacity for belief with other animals, equally shaped by natural selection, but who do not exhibit the same concern for truth that humans do. At the same time, though, there is nothing in the account that would require a non-natural source for epistemic 27 See Reed (2015). 22 normativity. Cultural practices are, after all, as much a part of the natural world as straightforwardly inherited traits are. It is worth noting a welcome consequence of the shift away from a stand-alone truth norm to an epistemic practice animated by a concern for truth. Although cognitive success in many cases will involve having a true belief, there are some false beliefs that can play a vital role in sophisticated cognition. A simplifying falsehood can make some kinds of inferences easier to perform, by abstracting away inessential details. A series of hypotheses that fall short of the truth but get increasingly close to it represents an important sort of cognitive progress. And a theory chosen over an empirically equivalent rival because it is simplest is also a cognitive achievement, despite it being unclear whether it satisfies the truth norm. Each of these positive outcomes is possible because our epistemic practice is complex, holistic, and plastic. It can be elaborated in many different ways, where each of these elements is in some way in service of discovering the truth, but the connection with truth need not be immediate.28 The epistemic practice view can also answer the two other objections considered in section 3 above. According to the first, the distinction between reasons to believe and reasons for believing collapses, once we try to explain why reasons to believe have normative force. On the most plausible explanation, the Humean approach, reasons to believe matter to us because they are related to our fundamental desires—either for the truth itself or for some other end, for which having a true belief is instrumentally valuable. But this would mean that we care about reasons to believe when we also have reasons for having the belief in question; that is, we would feel bound by epistemic reasons only when we also have the relevant practical reasons. According to the second objection, from Tom Kelly, the Humean approach runs into trouble when the agent loses her desire for the truth (or for whatever end made it instrumentally rational to have a true 28 There is a further complication: we can possess the truth to a greater or lesser degree. Elsewhere, I argue, in effect, that the epistemic practice recognizes a variety of kinds of knowledge. Depending on the practical circumstances in which one finds oneself, some kinds of knowledge may be inadequate (and thus irrelevant) for one’s purposes. See Reed (2013). 23 belief). Without that desire, she has no reason to pay attention to whatever evidence she may be confronted with. But that can’t be correct; it is never right for us to ignore counter-evidence just because we don’t want to know the relevant truth. Kelly objects, then, that the Humean approach fails insofar as it makes epistemic reasons hypothetical rather than categorical in nature. The epistemic practice view is broadly Humean, in the sense that it explains the normative force of epistemic reasons as stemming ultimately from what we desire. Nevertheless, the view has two characteristic features that permit answers to these objections: practical indirectness and normative persistence. To see this, recall James’s climber, who had practical reason for believing that she could jump across the chasm, as this would make it likelier to happen, even though the evidence shows that she very probably will not successfully make that jump. If we limit our focus to how this single belief relates to her practical interests, it looks as though believing in accordance with her evidence would not be practically rational, as she doesn’t desire the truth for its own sake in this case, nor would a true belief be instrumentally useful for her. But, from a broader point of view, it is clear that she does still have good practical reason to take part in an epistemic practice. Doing so best serves her desire for truth in general, if she does indeed have that desire, and it is the most effective way to pursue the great majority of her other ends as well. Given her ongoing commitment to the epistemic practice, she will feel the normative force of the evidence indicating that she will not get across the chasm. Of course, what it is practically rational to do may still outweigh the guidance of the epistemic practice on any particular occasion—she may choose to believe that she’ll make it to the other side, even in the face of that evidence. The point is simply that there will be a conflict between what it is practical to do and what it is epistemically rational to do.29 29 Hence, there will be no possibility of the perverse manipulation to be found in PRACTICAL PERSUASION. The climber may ultimately believe that she will be able to jump across the chasm—and this may be the rational thing to do, all things considered—but it will not be the rational thing to do from an epistemic point of view. In having that belief, she will still be doing something epistemically wrong, and she will be susceptible to feeling the force of that criticism—though one hopes none of her 24 In this way, the evidence she has constitutes an epistemic reason to believe something, no matter how her practical interests bear on having that belief. The normative force of the epistemic reason does derive from her practical interests, but only indirectly. The epistemic reason is not fully categorical—if the agent had no practical reasons at all, she wouldn't feel the force of epistemic reasons, either. Nevertheless, the epistemic practice view does provide a satisfactory approximation of categorical reasons. Insofar as participation in an epistemic practice will be practically rational for any social being who has practical ends to pursue, epistemic reasons will have universal normative force. 30 And, given the close connection between pursuing practical ends of any kind and being able to draw on the resources of the epistemic practice—its characteristic properties, virtues, institutions, etc.—the force of the reasons it delivers will be deeply felt. Only the barest outline of what an epistemic practice is, how it works, and how it relates to our practical interests has been sketched here. More could be said about how our practical needs on a particular occasion interact with specific aspects of the epistemic practice.31 And, certainly, much more could be said about how a shared concern for truth animates the vast range of phenomena that epistemic practice comprises, from the intellectual virtues to the methodology of science. But those parts of the story will have to be told elsewhere. 32 climbing partners will be cold-bloodedly honest enough to mention it. 30 Notice that the epistemic practice has a better claim to universality than does our moral practice. Even psychopaths care about the truth, if only because it is instrumentally useful. 31 See Reed (2013) for an account of knowledge pluralism, epistemic possibility, and some aspects of how we talk about knowledge in different practical environments. 32 For helpful comments, I am grateful to the audience at a conference on epistemic normativity at the University of Southampton, who heard a very early version of this paper. I am also grateful to the editors of this volume, both for their generous patience and for their insightful feedback. I have benefitted tremendously from interacting with my students, both in conversation and in reading their written work on some of these issues—especially Amy Flowerree, Nathan Weston, and 25 REFERENCES Adler and Hicks. 2013. “Non-Evidential Reasons to Believe,” in The Aim of Belief, T. Chan (ed.). Oxford: Oxford University Press, 140-166. Brown, Jessica. 2008. “Subject-Sensitive Invariantism and the Knowledge Norm for Practical Reasoning,” Noûs 42: 167-189. Chan, Timothy (ed.). 2013. The Aim of Belief. Oxford: Oxford University Press. Clifford, W.K. 1877. “The Ethics of Belief,” Contemporary Review 29: 289-309. Cohen, Stewart. 1999. “Contextualism, Skepticism, and the Structure of Reasons,” Philosophical Perspectives 13: 57-89. Craig, Edward. 1990. Knowledge and the State of Nature. Oxford: Clarendon Press. DeRose, Keith. 1992. “Contextualism and Knowledge Attributions,” Philosophy and Phenomenological Research 52: 913-929. Fantl, Jeremy, and Matthew McGrath. 2009. Knowledge in an Uncertain World. Oxford: Oxford University Press. ——. 2002. “Evidence, Pragmatics, and Justification,” Philosophical Review 111: 67-94. Gibbons, John. 2013. The Norm of Belief. Oxford: Oxford University Press. Hawthorne, John. 2004. Knowledge and Lotteries. Oxford: Clarendon Press. Hawthorne, John, and Jason Stanley. 2008. “Knowledge and Action,” The Journal of Philosophy 105: 571-590. Kathryn Pogin. As always, my deepest debt is to Jennifer Lackey. 26 Hieronymi, Pamela. 2005. “The Wrong Kind of Reasons,” Journal of Philosophy 102: 437-457. Hume, David. 1975 [1748]. Enquiries Concerning Human Understanding and Concerning the Principles of Morals, 3rd ed. Oxford: Clarendon Press. James, William. 1897. The Will to Believe, and Other Essays in Popular Philosophy. New York: Longmans, Green, and Co. Kelly, Thomas. 2003. “Epistemic Rationality as Instrumental Rationality: A Critique,” Philosophy and Phenomenological Research 66: 612-640. Lewis, C.I. 1946. An Analysis of Knowledge and Valuation. La Salle, IL: Open Court. Melville, Herman. 1846. Typee: A Peep at Polynesian Life. New York: Wiley and Putnam. Parfit, Derek. 2001. “Rationality and Reasons,” in Exploring Practical Philosophy, D. Egonsson et al. (eds.). Aldershot: Ashgate, 17-39. Plantinga, Alvin. 1993. Warrant and Proper Function. Oxford: Oxford University Press. Reed, Baron. 2015. “Reasons for Reasons,” Episteme 12: 241-247. ——. 2013a. “Practical Matters Do Not Affect Whether You Know,” in Contemporary Debates in Epistemology, 2nd ed., Matthias Steup and John Turri (eds.). Oxford: Wiley Blackwell: 95-106. ——. 2013b. “Fallibilism, Epistemic Possibility, and Epistemic Agency,” Philosophical Issues 23: 40-69. ——. 2010. “A Defense of Stable Invariantism,” Noûs 44: 224-244. ——. 2001. “Epistemic Agency and the Intellectual Virtues,” The 27 Southern Journal of Philosophy 39: 507-526. Shah, Nishi. 2003. “How Truth Governs Belief,” Philosophical Review 112: 447-482. Stanley, Jason. 2005. Knowledge and Practical Interests. Oxford: Oxford University Press. Street, Sharon. 2009. “Evolution and the Normativity of Epistemic Reasons,” in Belief and Agency, D. Hunter (ed.). Calgary: University of Calgary Press, 213-248. Wedgwood, Ralph. 2013. “The Right Thing to Believe,” in The Aim of Belief, T. Chan (ed.). Oxford: Oxford University Press, 123-139. ——. 2002. “The Aim of Belief,” Philosophical Perspectives 16: 267-297. Williams, Bernard. 1973. Problems of the Self. Cambridge: Cambridge University Press. Williamson, Timothy. 2000. Knowledge and Its Limits. Oxford: Oxford University Press.