It is commonplace to make judgements about what a person ought or ought not believe. When a judgement of this sort is based solely on what we take the person’s evidence to be, the evaluation is epistemic. The practice of evaluating beliefs epistemically seems to commit us to the existence of a distinct class of obligations grounded in purely epistemic—as opposed to moral, legal or prudential—considerations. If we allow that there are moral, legal or prudential obligations, then why not epistemic ones? Is there any reason to doubt that epistemic duties exist?Footnote 1
A number of reasons have been offered. William Alston’s voluntarism argument claims that our lack of voluntary control over what we believe makes epistemic evaluations inappropriate.Footnote 2 Mark Nelson’s argument from justificational fecundity claims that our evidence is too rich for epistemic factors alone to pick out specific propositions that we ought to believe.Footnote 3 Keith DeRose’s argument from problematic cases claims that the epistemic ought is ambiguous.Footnote 4 And Anthony Booth’s argument from overriders claims that our doxastic attitudes are constrained generically, rather than epistemically, since we cannot be justly blamed for violating pro tanto—and by implication, merely ostensible—obligations.Footnote 5
Here I want to focus on Chase Wrenn’s argument from conceptual economy. What sets it apart is its powerful appeal to considerations of parsimony in theory choice. Though there may be duties to adopt particular doxastic attitudes, according to Wrenn, such duties are never epistemic. Allegedly epistemic duties are really a species of moral duty. Since every situation that imposes a supposedly epistemic duty morally requires the very same thing of us, it is conceptually economical, and therefore preferable, to eliminate the class of epistemic duties.
My aim is to show that Wrenn’s argument from conceptual economy fails for at least three reasons: (1) it is too powerful and removes classes of obligation that are only very implausibly denied; (2) it involves a misapplication of Occam’s razor; (3) it ignores the positive evidence for a class of distinctively epistemic duties. I present these objections over Sections 2 to 4. Section 1 is devoted to explaining Wrenn’s argument.
1. The Argument from Conceptual Economy
Wrenn assumes that we can be morally, legally or prudentially obligated to do certain things. Perhaps he would agree that:
(1) I am morally obligated to treat people fairly.
(2) I am legally obligated to register my car.
(3) I am prudentially obligated to meet my immediate neighbours.
Similarly, Wrenn suggests that there may be moral, legal or prudential obligations to believe certain things. He offers these examples for illustration:
(4) I may be morally obligated to believe in the equal moral value of all people [since I can’t act rightly without holding this belief].
(5) I may be legally obligated to have an accurate belief about the expiration date of the milk that I sell [since I can’t avoid selling expired milk without having an accurate belief about the date].
(6) I may be prudentially obligated to believe that my doctors are competent [since I can’t act in my own best interest without believing that their advice is sound].Footnote 6
If, in addition, a person can be epistemically obligated to believe, disbelieve or suspend judgement about something, then “there are purely epistemic sources of duty, over and above such other sources as morality, prudence, and the law.”Footnote 7 Epistemic duties are normally thought to be grounded in the available evidence.Footnote 8 If my total available evidence supports the proposition that the window is open, then I epistemically ought to believe that the window is open.
Interestingly, Wrenn claims that epistemic duties could in theory conflict with other sorts of obligation. He asks us to imagine the following scenario. The milk I sell expires in a week. The law requires me to have accurate beliefs about the expiration dates on the products that I carry.Footnote 9 Unfortunately, some pranksters secretly switched the labels on the bottles and I now have strong evidence for the proposition that my milk expires in 10 days. So I epistemically ought to believe it. And this conflicts with my legal duty to believe that the milk expires in a week.Footnote 10 The possibility of such conflict is consistent with the existence of epistemic duties. But it’s a possibility that is never realized. When it appears that an epistemic duty runs counter to a different obligation, what we really have, in Wrenn’s view, is a disguised moral duty competing with some other non-epistemic requirement. Epistemic requirements simply do not exist.
The argument Wrenn rolls out against them has undeniable appeal. First, he considers the prospects for a proposed category of parental duties. Parental duties would be duties that one has qua parent. Their source would be the biological fact of parenthood. If A is the parent of B then, on this view, there are things that A is obligated to do for B simply because A is the parent of B. Perhaps A has a parental duty to get proper medical care for B when B is ill or injured.Footnote 11 The source of obligation here would not be moral, legal or prudential. It would be parental.
One could quite reasonably deny that parental considerations alone give rise to any duties. This is not to say that parents have no duties, but to claim that the obligations one has with respect to one’s children issue from another source. Perhaps the source is morality. If so, it is the moral duty of parents to provide for their children and to promote their children’s interests. This is a plausible take on putative parental duties. A person who occupies the role of parent is bound by certain requirements, but they are ultimately grounded in moral rather than purely parental considerations.
It might be worth noting incidentally, however, that the example makes things a little too easy. The required actions or policies Wrenn considers—promoting another person’s interests, providing medical care for a person who can’t provide for herself—have a decidedly moral feel about them, so it’s no surprise that their source is morality. Consider an analogous case. If you have a cat, you have an obligation to feed her and not to hurt her, but no one would infer from this that there are distinctively felinal duties arising from felinal considerations alone. These are quite evidently moral duties. (And the alleged epistemic duties that Wrenn considers next are at least less palpably moral than are parental or felinal duties of this description.) Furthermore, one might argue (though I will not) that there are at least some parental duties that are not straightforwardly moral: perhaps a parent has an obligation—just by being a parent—to attend some of her child’s school plays, sports games or whatever. But it’s not obvious that these are moral duties. This contrasts with the (more clearly moral) obligation to get proper medical attention for one’s child.
But let’s suppose that Wrenn is right anyway: all situations that impose a parental duty to do x also impose a moral duty to do x. On this assumption, there is simply no need to introduce a distinct class of parental duties over and above the class of moral duties. Moral duties will do all the work required by parental duties. So we have a choice, according to Wrenn: we can recognize a class of parental duties, though doing so “adds nothing to our understanding of the world,” since parental duties fall within the scope of the moral. Or we can simplify our conceptual scheme and deny their existence. This is Wrenn’s recommendation: “Theoretical economy favours the latter attitude.”Footnote 12
Wrenn contends that his argument against the existence of “an autonomous realm of parental duty”Footnote 13 displays a pattern of reasoning that can be put to use elsewhere. It suggests a general strategy for arguing against the existence of duties of kind K. Judging from his remarks in the text, it seems that Wrenn has something like the following four-step process in mind:
Strategy for arguing against the existence of sui generis K-duties:
1. Show that all facts alleged to K-require x also morally require x.
2. Assert that a simpler conceptual scheme is always to be preferred.
3. Infer that a conceptual scheme recognizing the existence of moral duties but denying the existence of K-duties is preferable.
4. Deny the existence of K-duties.
Wrenn concedes that the method fails to provide a deductively valid argument against the existence of K-duties: “I do not claim that ‘it serves no purpose to posit K-duties, therefore there are no K-duties’ is a deductively valid form of inference.”Footnote 14 But he points to the success of this form of induction in the natural sciences. We deny the existence of vital forces, for example, because positing them does nothing to augment our biochemical understanding of life. So Wrenn proposes a corresponding application of Occam’s razor within the normative realm:
Once we have seen that the idea of sui generis K-duties adds nothing to our understanding of what people ought to do, their status comes to be like that of vital forces. We do not need to posit them, and so we might as well just deny that there are any.Footnote 15
The inference feels natural, even healthy: if positing entities of a certain kind does nothing to promote understanding, then we should exclude such entities from our ontology.
To complete the argument, Wrenn will need to demonstrate that epistemic duties can be collapsed into moral duties in the same way that parental duties can. We just need some reason to believe that alleged epistemic duties prescribe nothing beyond what is already required on purely moral grounds. To that end, Wrenn proposes a weak moral principle articulating an obligation to adopt the doxastic attitudes favoured by one’s evidence:
(MD): If a person S considers a proposition P, it is S’s pro tanto moral obligation to have the attitude toward P that S’s evidence favours.
The principle is weak to the extent that competing obligations can override it. But if (MD) is correct, there is no point in postulating an autonomous realm of epistemic duty, since any cognitive things we are required to do with respect to our beliefs will always be at least pro tanto morally necessary.Footnote 16 To recognize an additional, epistemic source of obligation would be redundant.
(MD) draws support from a higher principle, (MP), emphasizing the “moral importance of people’s ends and their efforts to realize them.”Footnote 17 It’s probably safe to say that any reasonable moral theory will endorse (MP) or something very much like it:
(MP): Whenever a person has a range of mutually incompatible options, she has a pro tanto moral obligation to take whatever option is best from the standpoint of promoting people’s ends (or one of those options, if there is a tie).Footnote 18
The basic idea is that, “other things being equal, one ought to do whatever is most likely to help people the most.”Footnote 19 The obligation to do what we can to help people is not absolute, however, since weightier moral considerations can trump it. For instance, though I might be able to help people the most by performing some action x that is wrong in itself, the intrinsic wrongness of x may be a more serious moral consideration than the help I can provide by doing it and so my pro tanto obligation to x is overridden. Alternatively, a more pressing moral obligation that precludes my doing x may take priority over my pro tanto obligation to x. So the principle is compatible with theories that allow for a range of competing moral concerns and the possibility that a person may sometimes be relieved of the duty to promote other people’s ends (or even her own).
Wrenn’s core thesis is that (MD) is a corollary of (MP). In other words, the apparently epistemic obligation to have fitting doxastic attitudes is just a special case of the moral obligation to further people’s ends. This explanation has the advantage of conceptual economy, since it tucks one source of normativity into another.
Is there any reason to think that helping people is best achieved by following the evidence? Wrenn must explain why, when considering any proposition, the best option from the point of view of promoting ends is to believe in accordance with the available evidence. The key is to consider “the ways in which our beliefs guide and constrain our actions in pursuit of our ends.”Footnote 20 There is a clear link between belief and goal-directed action, which Wrenn spells out as follows:
We choose and execute our actions not on the basis of how the world is, but how we believe it to be. Nevertheless, it is how the world is that determines whether our plans succeed or fail. Thus it is important for our beliefs to give us an accurate picture of how things stand in the world.Footnote 21
I think we can distinguish three claims here:
(1) Since we act on the basis of our beliefs (rather than on the basis of world-states), promoting people’s ends will require that we have beliefs.
(2) The world, not our beliefs about the world, determines whether our plans succeed.
(3) Since we act on beliefs, and since beliefs need to be hooked up with the world accurately in order for our actions to succeed, acting to promote people’s ends requires getting accurate beliefs.
There is also a hidden evidentialist assumption:
(4) The best way to get accurate beliefs is to adopt the doxastic attitudes favoured by our evidence.Footnote 22
Together these claims provide a satisfactory explanation of the connection between (MP) and (MD). Since we are morally required to promote people’s ends—this is the thrust of (MP)—and since promoting people’s ends requires having accurate beliefs, we are morally required to have accurate beliefs. And since our best bet for getting accurate beliefs is to follow our evidence, it is morally imperative that we do so—just as (MD) requires. To illustrate with a very simple example, I won’t be able to help a needy person S, if I don’t understand the respects in which S is needy or if I have inaccurate beliefs about which actions would alleviate the hardship. Aligning my beliefs about S’s circumstances with my evidence is therefore most likely to increase my chances of acting effectively with respect to S’s needs.
One potential problem with this idea is that many of my beliefs don’t seem at all relevant to anyone’s needs or interests, not even my own. I believe that my purple tie is next to a pink one in the drawer upstairs. But how is that going to help anyone? It doesn’t seem to matter, morally-speaking, whether I believe this because I have good evidence for it or on the basis of a wild guess. The belief just doesn’t connect with anyone’s problems. Wrenn’s response to this worry is convincing, though it will cause some difficulty for his view later.
Call the proposition that my purple tie is next to a pink one ‘P’. I haven’t looked in the drawer in a couple of days and I have zero recollection of what order the ties were left in. So my evidence does not favour P. Suppose I made a guess, however, and now believe that P. Since P is irrelevant, this shouldn’t matter.Footnote 23 But as Wrenn points out, it’s certainly possible that a seemingly irrelevant proposition is relevant to someone’s problems and if not, it might become relevant someday.Footnote 24 The fact that I can’t imagine how P might affect someone’s interests doesn’t mean that it doesn’t really, or won’t sometime. Secondly, believing the truth is presumably one of my interests, and believing P without evidence impedes my ability to promote that interest.Footnote 25 Thirdly, and most significantly, failing to heed the evidence in connection with irrelevant propositions that one considers either reinforces bad habits of mind or else paves the way for their development: “To settle for an unfavoured attitude towards an irrelevant proposition is either to manifest or to cultivate a habit that will lead to unfavoured attitudes on relevant propositions.”Footnote 26 Since this potentially interferes with the promotion of people’s ends, it is morally impermissible.Footnote 27
This completes my exposition of the argument from conceptual economy. I will now raise several objections.
2. Collapsing One Source of Obligation into Another
My first objection is that the argument from conceptual economy does too much. Though epistemic requirements are a sort of conceptual extravagance on Wrenn’s theory, moral, legal and prudential requirements are not. They constitute distinct and irreducible normative domains. But the reductive strategy Wrenn deploys against epistemic duties can be turned with equal plausibility against these other types of requirement as well. If the argument works, we will be left with no independent source of obligation, no category of normativity that doesn’t simply collapse into another. And so it will never be true to say that we ought to do something on moral, legal or prudential grounds. This is unlikely enough to shake our confidence in the strategy itself.
My case rests on a number of arguments copying the pattern of reasoning proposed by Wrenn. We’ll have to rethink the strategy, if even one of them succeeds. And success here is relative: I am attempting a sort of reductio on Wrenn’s argument, so the reasoning need only be as plausible as his in order to sink it.
Let’s look at legal requirements first. It’s pretty clear that we have legal obligations. But this is a claim that we won’t be able to make if the argument from conceptual economy goes through. Recall that, aside from considerations of parsimony, Wrenn’s only basis for collapsing epistemic obligations into moral obligations is that every situation imposing an epistemic duty to do x also imposes a moral duty to do x. Presumably, then, if every situation imposing a legal duty to do x also imposes a moral duty to do x, considerations of parsimony will justify the same inference. And indeed many moral philosophers hold us morally accountable to the law in just this way.Footnote 28 But legal duties are arguably distinct from the corresponding moral duties. On certain interpretations of legal validity, anyone living under a particular legal system is ipso facto bound by its prescriptions. One could quite conceivably accept this notion of automatic legal liability while premising in addition a pro tanto moral requirement to abide by the law. I have a legal duty to pay my taxes, for instance, because I reside in a country where this is the law. I will also have a moral duty to pay my taxes, however, if I have a moral duty to do everything the law prescribes.
But why think that we have such a moral obligation in the first place? Here is one possible explanation. Law and the idea of equality go hand-in-hand, so perhaps considerations of fairness could supply the foundation for a pro tanto moral obligation to respect and obey the law.Footnote 29 The laws codify rules for arbitrating objectively between competing interests and it is therefore natural to think of them as providing guidance—albeit imperfect guidance—for promoting fairness in our dealings with other people. Taking this intuition as a starting point, we might think to dispose of legal obligations with the help of Wrenn’s reductive strategy.
The argument will go like this. Respecting persons is a moral requirement, if anything is, and since respecting persons involves treating them fairly, we morally ought to treat people fairly. Any decent moral theory will therefore endorse some such principle as the following:
(MP*): Whenever a person has a range of mutually incompatible options, she has a pro tanto moral obligation to take whatever option is best from the standpoint of promoting fairness (or one of those options, if there is a tie).
Assuming that obeying the law is normally best from the standpoint of promoting fairness, (MP*) provides some support for (MD*):
(MD*): For any person S and any possible action A, it is S’s pro tanto moral obligation to obey the law with respect to A.
Basically, (MD*) says that if A is legally obligatory for S, then S is pro tanto morally required to do A, and if A is legally impermissible for S, then S is pro tanto morally required not to do A.
We must be careful not to overstate the point. Just as following the evidence does not always promote people’s ends most effectively—some evidence is misleading and sometimes false beliefs lead to better outcomes than do true beliefs—so following the laws does not always promote fairness: some laws are unjust and application of a just law can sometimes produce unfair consequences. But on the whole the assumption should be that upholding the law is better from the point of view of fairness than just making up one’s own rules.Footnote 30 To be sure, promoting people’s ends sometimes requires acting without or even contrary to the evidence and maintaining fairness sometimes requires bending or breaking a law. But on balance these are the best options we have.Footnote 31
Though attractive, the view that we have both legal and moral obligations with respect to the law is not available to advocates of conceptual economy. If every putative legal obligation is also a moral obligation, considerations of parsimony demand that we simplify our conceptual scheme by eliminating legal duties. But does the fact that there is always a moral presumption in favour of obeying the law mean that we don’t also have the relevant legal obligations? That’s a clear non sequitur. We are legally required to pay our taxes just because it’s the law. And this legal obligation has an independent moral backing: fairness. Both types of obligation may coexist, even if the truth of (MD*) implies that every legal obligation to x is accompanied by a moral obligation to x. This parallels a situation where (MD) is true—and thus, where every epistemic obligation to x is accompanied by a moral obligation to x—without epistemic obligations disappearing into moral obligations.
A tempting counterargument must be dealt with straightaway. Perhaps someone will want to claim that we have a strictly general moral obligation to obey the law but that it doesn’t translate into any particular moral obligations to comply with any particular laws. The particular obligations are legal and cannot, therefore, be collapsed into the moral. But this response won’t do. First, it’s not even clear what it means. Suppose we transcribe all of the laws into one document: Call it list L. It makes no sense to say that we have a general K-duty to adhere to L but no particular K-duties to perform any of the particular actions listed in L. What, exactly, are we being told to do if not to perform the particular actions from L? Let’s say that L contains actions A, B, C…n. One would think that my K-duty to act according to L implies that I am K-required to A, K-required to B, and so on, all the way to n. If not, it’s unclear what K-accountability to L really amounts to. Second, even if there is something to this distinction—which I very much doubt—it must save epistemic duties from collapse as well. If we interpret Wrenn’s (MD) as implying only a general moral duty to follow the evidence, we’ll have to deny that we have particular moral duties to believe any particular propositions. The particular duties could all be epistemic. But this is exactly what Wrenn denies. They are moral duties, according to him. So either we lay the objection aside as confused or else we admit that there may be epistemic duties.
A second counterargument was suggested to me by a referee for this journal. Wrenn will want to identify some important disanalogy between legal and epistemic duties. He might argue that legal duties are not as closely linked with moral duties as he’s claiming that epistemic duties are.Footnote 32 Finding a relevant disanalogy would prevent the extension of Wrenn’s reductive strategy to legal requirements.
I agree entirely with the referee that Wrenn should do this. He must identify some relevant difference between epistemic and legal requirements that will allow the reduction in the one case and block it in the other. This is what I am challenging him to do. The onus is on Wrenn to show that my parody of his argument doesn’t work. I am indicting the strategy by using it to generate a number of implausible conclusions. Wrenn must explain why the strategy works for him but doesn’t work for me. As things stand, we have an argument for the moral necessity of obeying the law that is no worse off than Wrenn’s argument for the moral necessity of heeding the evidence. So by his own reckoning legal requirements should reduce to the moral.
But the fact is that legal requirements don’t simply vanish on account of being morally required as well. Their concurrence with the principles of morality poses no conceptual problems whatever. Wrenn agrees, of course, that the inference from ‘Every K-duty to x is also a G-duty to x’ to ‘There are no K-duties’ is deductively invalid. He’s claiming only that it’s reasonable. That it is reasonable, however, is less certain now that we’ve used it to draw the unlikely conclusion that we have no legal obligations. This should raise suspicions. There’s something fishy about this form of inference despite its elegance and obvious appeal. I’ll see if I can put a finger on the problem in Section 3. Another worry must be dealt with first: exaggeration. Perhaps I’ve made too much of an isolated case? A few more arguments will show that I haven’t.
The arguments can be set out quickly, since the pattern is now familiar. We’ve just seen legal duties fold into moral duties by way of conceptual economy. But it won’t stop there. They can also be converted into purely prudential requirements. Though there is undoubtedly such a thing as a genuinely legal obligation, few would deny that it’s also prudent to conform one’s actions to the law. Consider these sober remarks from David Hume:
If the reason be asked of that obedience, which we are bound to pay to government, I readily answer, “Because society could not otherwise subsist.”Footnote 33
The general obligation which binds us to government is the interest and necessities of society; and this obligation is very strong.Footnote 34
In effect, Hume is suggesting that the general obligation to obey the law is a matter of prudence, since obeying the law promotes the interests of society and the interests of society are clearly our own. In other words, we ought to submit to the law as a necessary condition of maintaining a stable social order, for the resultant benefits of security outweigh the personal costs of obedience. It’s a question of pure utility, for Hume. If this analysis is correct, then legal obligations will be subsumed by prudential obligations on the grounds that a simpler conceptual scheme—one recognizing only prudential requirements—is always to be preferred.
For those who doubt that the interests of society coincide with our own, the point can be made more directly. Breaking the law comes with a penalty. It’s prudent to avoid fines and to stay out of jail, so it’s generally prudent to obey the law. But what if no one is looking? Sitting at a traffic light in the middle of the night on an empty street, one often contemplates the possibility of running the red. There’s no risk of detection (or of injury), so perhaps the legal obligation to wait it out fails to be a prudential obligation. Possibly, cases like this will block the collapse of legal obligations into prudential obligations, since conceptual economy comes into play only if every legal obligation is also a prudential obligation.
But we’re forgetting about my purple tie. Lacking the relevant evidence I just guessed where it was in the drawer. Since this potentially detracts from my ability to promote people’s ends, Wrenn would hold me morally accountable for the unfounded belief. And even if it’s not relevant to anyone’s interests, I fall into a bad habit by guessing in default of evidence. The same thing must hold here, if we’re going to be consistent. Though I may be quite alone at the traffic light, I really shouldn’t jump it, since this encourages a practice that will get me into trouble on other occasions. It’s just bad policy to break the law. So every situation imposing a legal duty to do x also imposes an overrideable prudential duty to do x. And that spells the end for legal duties.
This is all very puzzling. At first it seemed that considerations of conceptual economy would force us to regard putative legal obligations as moral obligations in disguise, but now we have an equally strong argument for treating them as ordinary prudential obligations.Footnote 35
Things get even murkier when we consider that it’s also prudent to be moral. Any time you have a moral duty to undertake some action it’s also prima facie advisable to do so from an entirely self-serving point of view since your reputation may depend on it. No one likes people who are immoral and it’s useful to be liked. So you ought to be moral for your own sake.Footnote 36 Inversely, many people believe that we also have moral duties to ourselves. A case could certainly be made that we are morally required to promote our own interests to the extent that they do not interfere with the promotion of anyone else’s. But this is just to say that we have a defeasible moral duty to be prudent. So it’s both prudent to be moral and moral to be prudent. Which class of obligation, then, reduces to the other? The argument from conceptual economy provides no criterion for determining which direction the reduction should go and I can’t think of any either.
The outlook for Wrenn’s reductive strategy is thus very poor. He needs to preserve the autonomy of the moral realm if he’s going to claim that epistemic requirements are really just moral requirements in disguise. And for this claim to be at all credible, it must not preclude there being genuinely legal and prudential requirements as well, since we generally do recognize the existence of moral, legal and prudential sources of obligation. But as the arguments presented above have shown, all three categories of normativity break down when we start thinking in terms of conceptual economy. To prevent a total collapse of the normative, Wrenn would need to exempt moral, legal and prudential obligations from the demand for parsimony, but unless he offers some convincing reason to do so, such a move would be merely arbitrary. And even if some such reason were to be found, it still wouldn’t give Wrenn the conclusion that he wants, at least not unambiguously. At best, the argument from conceptual economy justifies subsuming epistemic requirements under one of the other types of obligation, not under moral obligations specifically. This is a distinct point that requires some explanation.
Return briefly to Wrenn’s argument for the existence of a moral duty to follow the evidence. When you boil it right down it comes to this:
1. We morally ought to promote people’s ends.
2. Promoting people’s ends requires having accurate beliefs.
3. The best way to get accurate beliefs is to follow the evidence.
4. Therefore, we morally ought to follow the evidence.
Coupled with an appeal to parsimony, the argument favours simplifying our conceptual scheme by replacing normative facts of one type (epistemic requirements) with normative facts of another type (moral requirements). This will reduce, if not the overall number of entities in our ontology (quantitative parsimony), the number of types of thing postulated (qualitative parsimony).Footnote 37 To persist in positing epistemic requirements when we do not need to would be extravagant.
Seeing the bare bones of the argument like this exposes a new weakness. The reasoning can be repeated, almost verbatim, to show that we have both a prudential and—surprisingly—a legal duty to follow the evidence:
1. We prudentially ought to promote our own ends.
2. Promoting our own ends requires having accurate beliefs.
3. The best way to get accurate beliefs is to follow the evidence.
4. Therefore, we prudentially ought to follow the evidence.
And:
1. We legally ought to obey the law.Footnote 38
2. Obeying the law requires having accurate beliefs.
3. The best way to get accurate beliefs is to follow the evidence.
4. Therefore, we legally ought to follow the evidence.Footnote 39
None of these claims are the least bit contentious and the arguments are nearly identical to Wrenn’s. The implications, however, are unacceptable. Consistent application of Occam’s razor requires that every apparent epistemic duty to follow the evidence be replaced with a corresponding prudential as well as a corresponding legal duty. This is problematic for a couple of reasons. It is false, of course, that we are legally required to follow the evidence in matters of belief. Plus, it just can’t be that every apparent epistemic duty is simultaneously nothing but a moral duty, nothing but a prudential duty and nothing but a legal duty. And as far as the reductive strategy goes there is simply no basis for privileging moral considerations over legal or prudential considerations as the ultimate ground of our obligations with respect to belief. We are left wondering what putative epistemic requirements really are and what force they can possibly have when the appeal to parsimony has eliminated every other candidate source of obligation.
I want to be very clear that I do not stand behind any of the arguments introduced in this section. My claim is just that Wrenn’s argument is no better. Since the conclusions I have drawn above are neither reasonable nor consistent, the reductive strategy that produced them will need to be re-examined. What we will find, I think, is that Wrenn is playing fast and loose with Occam’s razor. Showing this is the task of the next section.
3. Abusing Occam’s Razor
The idea that evidence can be morally significant is hardly controversial and Wrenn’s particular way of spinning it out reveals some possible connections between ethics and epistemology that are worth exploring. Even the derivation of (MD) from (MP) is fairly convincing in the details and (MP) itself is a plausible moral principle. Objections could be raised, I’m sure, but it’s not here that we should focus attention. To find the underlying defect in the reductive strategy we need to think in more general terms about what Wrenn is doing. At bottom he seems to be saying that if we can get away with denying the existence of epistemic requirements, then we should, since it’s always a bonus to streamline our conceptual scheme. The key claim, remember, was this: “We do not need to posit them, and so we might as well just deny that there are any.”Footnote 40 Wrenn characterizes this move as “an application of Occam’s razor,”Footnote 41 but I’m not convinced that it is, at least not a valid one. Without denying that conceptual extravagance is generally undesirable, we may ask whether an appeal to economy is warranted in the circumstances Wrenn describes and whether the purging of epistemic duties from our ontology is what economy really requires. My answer to both questions is: no.Footnote 42
Consider what Wrenn might mean by “we do not need to posit them.” Obviously we never really need to posit the existence of anything, with the possible exception of our own minds. Wrenn must be thinking about what it is reasonable not to posit. I suppose we would need to posit epistemic requirements—in the sense that it would be irrational not to—if either or both of the following claims were true:
(1) The evidence favours positing epistemic requirements.
(2) Positing epistemic requirements significantly increases the explanatory power of our conceptual scheme (and the evidence is not otherwise against them).Footnote 43
So we can reasonably exclude epistemic requirements from our ontology—we can get away with not positing them—only if they are not favoured by the balance of evidence and positing them doesn’t significantly increase the explanatory power of our theory. It wouldn’t make much sense to deny the existence of x’s on the ground of economy if the evidence for x’s were very strong or, alternatively, if the evidence were in other respects neutral but positing x’s allowed us to explain many things that we couldn’t otherwise explain. For example, it would be irrational for me to deny the existence of bacteria. Given the evidence and given their power to explain diverse phenomena, I can’t reasonably get away with dropping bacteria from my ontology.Footnote 44
But we’ve taken an extremely well-supported hypothesis with the example of bacteria, and we needn’t. Far weaker evidence than this will still trump considerations of parsimony. When evidence for the existence of exoplanets orbiting sun-like stars first began to emerge, it would have been foolish for scientists to deny their existence on the grounds that a conceptual scheme without exoplanets is simpler, and thus more elegant, than one with them. One’s ontology must reflect one’s evidence, after all, not one’s aesthetics and it’s enough that the evidence favoured their existence to make the postulate of exoplanets rational. When the question is whether or not to recognize the existence of something, considerations of parsimony don’t even factor in unless the evidence—including the evidence from explanatory power—is more or less indeterminate. Parsimony is at best a tie-breaker. So if the total available evidence supports the existence of x’s, it is reasonable to postulate x’s with a degree of conviction proportionate to the evidence, regardless of conceptual economy. Put simply, evidence always takes precedence over elegance in theory choice.Footnote 45
What role precisely considerations of parsimony should play in theory choice is a vexed question in the philosophy of science, which I cannot begin to settle here. The one point that I have been pushing—albeit rather crudely—is that evidential considerations are primary.Footnote 46 It is therefore astonishing that Wrenn says nothing on the subject of evidence. He never asks if there is any independent evidence for or against the existence of epistemic requirements and he merely asserts, without argument, that if every alleged epistemic requirement to x coincides with a moral requirement to x, then everything explainable on the assumption of epistemic requirements is explained equally well without them. And so, for the sake of economy, “we might as well just deny that there are any.”Footnote 47 Before whisking away our intellectual obligations, however, we ought to at least look at the evidence—in fact, we ought to look closely. Wrenn simply bypasses the evidential question and gets right to housekeeping. But absent a careful consideration of the evidence for and against the existence of epistemic requirements, the appeal to economy has no punch.
In the final section, I run through some of the evidence that needs to be taken into account before any decision can be made regarding the existence of epistemic requirements. For now I’d like to just give Wrenn the point, and to give him moral requirements as well. I’ll assume for the sake of argument that:
(1) Moral requirements exist.
And I’ll suppose that the total available evidence fails to discriminate between:
(2) Epistemic requirements exist.
(3) Epistemic requirements do not exist.
In supposing that the evidence is entirely neutral with respect to (2) and (3), I am also allowing that no loss in explanatory power will result from the removal of epistemic requirements. In other words, I assume that we have two equally good ways of explaining all normative constraints on belief—one of them simple, the other complex. The simple explanation is committed to (1); the complex explanation is committed to (1) and (2).
In the circumstances imagined, it seems to me that an appeal to conceptual economy would be warranted and that the simpler explanation would be justifiably preferred. If the facts are precisely as I have framed them above, then our conceptual scheme should commit us to the existence of moral requirements without committing us to the existence of epistemic requirements. But even granting all of this, are we entitled to assert (3) on the grounds of economy? Are we entitled to deny the existence of epistemic requirements? I see no reason to think so.
To convince you of this I’m going to bring the discussion down to earth for a minute and ask you to consider the situation in my backyard. Suppose I tell you that I’ve often seen racoons behind my house at night and you have every reason to believe me. And suppose you’ve gathered from some pictures that my backyard contains a couple of hollow tree stumps. Your evidence, we’ll suppose, favours:
(4) There are hollow tree stumps in my backyard.
Racoons, you probably know, like hollow tree stumps. So the existence of hollow tree stumps in my yard adequately explains the presence of racoons. But suppose you also know that racoons like open garbage bins and that you have no idea how I store my garbage (the pictures you saw didn’t show the top part of the yard, we’ll say). Your evidence, therefore, does not discriminate between:
(5) There are open garbage bins behind my house.
(6) There are no open garbage bins behind my house.
It appears that you have at least two good explanations at hand for the racoon sightings—one simple, the other complex. The simple explanation is committed to (4); the complex explanation is committed to (4) and (5).
My sense is that in these epistemic circumstances you ought to prefer the simpler explanation. You should suppose that there are hollow tree stumps in my backyard and let that, pending further investigation, suffice as your explanation for the racoons. You should not postulate open garbage bins since your simple explanation is adequate and you have no evidence that there are any. You can get away with not postulating any bins. But it would be totally irrational for you to assert (6). “We do not need to posit open bins, and so we might as well just deny that there are any” is not a line of reasoning that you should take. You should suspend judgement on the question of open bins. Your evidence requires it.
This very mundane example is meant to illustrate an important point that has been made by Elliott Sober. According to Sober, there are two versions of Occam’s razor—what he calls the ‘razor of denial’ and the ‘razor of silence.’ The razors differ on what it is rational to believe when one’s evidence fails to discriminate between the existence and nonexistence of some contested entity:
Razor of denial: If your evidence does not discriminate between “X exists” and “X does not exist,” you should deny the former and affirm the latter.
Razor of silence: If your evidence does not discriminate between “X exists” and “X does not exist,” you should suspend judgement about both.Footnote 48
Sober claims that the razor of silence “is the standard bearer for evidentialism.”Footnote 49 This statement is correct. If the propositions we are justified in believing are the ones our evidence supports, then, whenever our evidence is silent on the existence of x’s, we ought to be silent too. To deny the existence of something when the evidence doesn’t favour denial is as irrational as affirming the existence of something when the evidence doesn’t favour assertion. Going beyond the evidence in either direction is a mistake.
A minor point of disagreement needs to be mentioned. I maintain that the razor of denial is never of any use to an evidentialist, whereas Sober allows for one exception. He considers an oft-repeated slogan from the natural sciences: absence of evidence isn’t evidence of absence. I have no evidence that there is a storm on the surface of Jupiter right now.Footnote 50 But absence of evidence for a storm on Jupiter is decidedly not evidence for absence of a storm. I clearly have no grounds for denying that a storm is raging on the surface of Jupiter: “The razor of denial gives bad advice here.”Footnote 51
So far I agree entirely with Sober. Here’s where I think he slips up:
The razor of denial makes sense if the putative entity under consideration can be assumed to have the following property: If X exists, then we’d have evidence that X exists. Given this assumption, absence of evidence is evidence of absence.Footnote 52
There’s no question that absence of evidence is evidence of absence in cases where it’s true to say that if x existed, we would have evidence that x exists. My quibble is with Sober’s suggestion that the razor of denial “makes sense” in any such case. Where absence of evidence is evidence of absence, we simply don’t need any extra-evidential input from Occam’s razor. The evidence alone would favour denial. I infer that the razor of denial is a needless appendage and that evidentialists, like Wrenn, are wrong to invoke it. If the evidence does not discriminate between ‘Epistemic requirements exist’ and ‘Epistemic requirements do not exist,’ then we should refrain from making either assertion while selecting the simplest normative theory consistent with what we know.
It is important to remember that the preceding discussion has been carried out under the assumption that the evidence for epistemic duties does not incline either way. This is a point that Wrenn failed to establish—failed even to address. So he is not justified in reaching for Occam’s razor. Had he first surveyed the evidence and found it lacking, then silence, not denial, would have been called for. Either way, Wrenn has done nothing to support his thesis that there are no epistemic duties. The more fitting conclusion is that there may be epistemic duties.Footnote 53 Even supposing the evidence to be entirely neutral, this is the only assertion we are entitled to make.
4. Evidence of Epistemic Normativity
Wrenn wants us to deny the existence of epistemic duties but has given us no reason to do so. Defusing his argument was the goal above and is a necessary preliminary to establishing the opposite. But that is not something I attempt here. I am not going to argue that we do have epistemic duties. The positive conclusion would be difficult to prove even without limitations of space and the arguments I envision are incomplete.Footnote 54 Lowering my sights accordingly, I present some evidence that is prima facie suggestive of epistemic normativity and claim that its existence makes the negative stance somewhat harder to maintain. Were the evidence indifferent, denial would be rash; as signs of normativity are revealed, it becomes increasingly unreasonable.
So what is the evidence that we are subject to purely epistemic requirements? One way to appreciate the evidence we do have is to ask what it would be like for there to be evidence against them. Assuming the idea of an epistemic requirement is coherent, the only possible disconfirming evidence I can think of is of the negative kind that implies an exception to the absence of evidence slogan. Here is what I’m getting at. Suppose that epistemic requirements were the sort of thing that would tend to manifest themselves if they existed. Then, absent any indications of them, it would be reasonable to infer that they don’t—the absence of evidence would be evidence of absence.Footnote 55 But is that really how things are? Are there no signs of epistemic normativity? It seems to me that the evidence we should expect to find if epistemic requirements existed is exactly like the evidence we do find—of the same kind and quality as the evidence we have for the existence of purely moral requirements. In short, we would hold ourselves and others accountable to certain standards of reasoning just as we hold ourselves and others accountable to certain standards of conduct.
That we do hold each other accountable to certain standards of reasoning is evident from two notable features of our epistemic practice and from our inclination to regard the rules of logic as binding. In addition, our rational nature itself may be a source of epistemic normativity. That gives four types of evidence for epistemic requirements, though possibly there are more. I list them now and comment briefly on each:
1. Negative Epistemic Appraisal: It is an undeniable fact that we evaluate irrational belief negatively. When someone makes a hasty or unjustified inference we say things like, “You shouldn’t have inferred that” or “You’re jumping to unwarranted conclusions.” If someone’s beliefs don’t square with their evidence we correct them or simply alert them to the discrepancy and expect that it will motivate—if not actually bring about—a doxastic change. “Your belief is unfounded” constitutes a criticism. And, significantly, we blame people for habitual irrationality or for failing to reflect with sufficient care or frequency on the grounds of their beliefs. “He’s overly emotional and ignores the evidence” and “She rarely thinks about what she believes” give natural expression to a common concern—that someone is managing their belief system poorly. Such evaluations needn’t make any reference to the moral, legal or prudential ramifications of transient or chronic irrationality. In many cases they stem from epistemic considerations alone. Negative epistemic appraisal is a ubiquitous and deep-seated feature of our epistemic practice and can therefore be regarded as prima facie indicative of epistemic normativity. If we were bound by epistemic duties, then it stands to reason that we would take ourselves to be. The latter, at least, appears to be the case.
2. Excuse-Making in Epistemic Contexts: That our intellectual conduct is governed by epistemic norms is further evidenced by the common practice of excuse-making in epistemic contexts—the flipside of negative epistemic appraisal. Though we don’t normally excuse ourselves for adopting particular doxastic attitudes, we do make excuses for sloppiness and failure to deliberate: “I was rushed and didn’t think it through,” “I should have put my feelings aside and focused on the facts,” “I need to think more about why I believe some of the things I do.” If we didn’t think that reflecting carefully on the grounds of our beliefs were in some way obligatory, then it wouldn’t make sense for us to offer any such excuses. It is telling that we try to rationalize cognitive lapses and laziness even in circumstances devoid of moral significance. Anyone who wants to deny the existence of uniquely epistemic requirements will need to explain why this rationalizing tendency is so widespread. If there were no epistemic requirements, then excuse making in epistemic contexts would be out of place or even incoherent. It’s possible, of course, that we’re simply confused when we offer excuses for failing to heed or sort through our evidence, but I think it’s more likely that we regard ourselves as epistemically obligated to do certain things because we are epistemically obligated to do certain things. Though hardly decisive, the pervasiveness of this practice puts some additional weight behind the idea that we are bound by duties of a purely epistemic description.Footnote 56
3. The Authority of Logic: Certain patterns of reasoning have a special connection with truth—they preserve it, either perfectly (by way of deduction) or imperfectly (by way of induction, abduction or a looser sort of inference). Truth-preserving patterns of reasoning lead from truths to truths with a higher frequency than from truths to falsehoods. Such patterns are ‘logical’ in the broadest sense of the term. Principles of logic are formalized statements of these patterns. Their connection with truth makes them normative. They distinguish good from bad thinking in general, correct inferences from incorrect. Though logic normally prescinds from questions of evidence and belief, there is a sharp propensity to regard the rules of logic as authoritative in epistemic contexts and logical support translates into evidential support in matters of belief. If P entails Q, or makes it likely that Q, then, if S justifiably believes that P, S is justified in believing that Q on the basis of P (whether or not S in fact believes that Q).Footnote 57 The normativity of logic thus seems to imply the truth of certain epistemic ought statements. Suppose I justifiably believe both that 2 is prime and that 2 is even. This entails that not all primes are odd. So two of my justified beliefs strongly support the proposition, ‘Not all primes are odd.’ There is a sense of ‘ought,’ therefore—a non-moral, non-prudential and non-legal sense—in which I ought to believe it. The way these ‘oughts’ link up with duties is not straightforward, in my view, and difficult questions will need to be faced in working out a theory.Footnote 58 Even so, the perceived authority of logic in epistemic contexts is another reason to suppose that some epistemic ought statements are true and it seems obvious that their truth imposes certain constraints on what we believe independently of any moral, legal or prudential considerations.Footnote 59
4. Our Rational Nature: It is plausible to think that beings capable of reasoning correctly ought to reason correctly. Since this is a capacity we do have, perhaps there’s a further argument to be made for epistemic normativity. It’s not yet clear to me exactly how the argument should go, but it could involve reflections like the following: because we have the faculty of reason, we occupy the role of believers. This implies that certain ‘role oughts’ pertain to us. The idea of role oughts comes from Richard Feldman. Feldman claims that many true ought statements result directly from one’s occupying a role. Such statements articulate the right ways to perform the role—they define good performance, in other words, and set the standards for evaluation. For example, an historian ought to consider historical documents without prejudice and a private investigator ought to be thorough. Feldman holds that the role of believer is no different in this respect. There are things we ought to do just insofar as we are believers. There is a way of doing it right and that’s what we ought to do.Footnote 60 It strikes me that more is needed, however, for such oughts to qualify as duties. A carriage horse ought to stick to the road (and it ought not to gallop), but it doesn’t have any obligations. At least two additional conditions must be satisfied—beyond the fact of occupying a certain role—for a ‘good performance’ ought to amount to a duty, and both are grounded in reason. First, a subject must be able to understand that there is a right and a wrong way to perform her role. She needn’t be cognizant of what the standards actually are, but must be capable of recognizing that they exist. The horse lacks this capacity and so it has no duties. Second, failure of compliance must be possible as a result of choice. I ought to salivate when there’s food in my mouth, but it’s not a duty, since I can’t fail to salivate either by willing it directly or by willing other things that would prevent it. Believing is different. I occupy the role of believer and I understand that there is a right and a wrong way to believe: in general, I ought to believe in accordance with my evidence. And to some extent I can fail in this task by ignoring my evidence and neglecting to think. Many details are missing from this brief account and many questions are unanswered, but the fact that some of the core necessary conditions of obligation are met in the case of belief suggests that believing may be the sort of thing that comes with duties. To get clear on that we must ascertain what else is needed for turning a role ought into a duty. Whatever the additional factors may be, I think they will depend on the presence of reason.
None of this is meant to confirm that we have epistemic duties. All I am recommending is caution.Footnote 61 The view I find most attractive holds that epistemic considerations never impose duties to adopt particular doxastic attitudes. The ‘ought’ in ‘S ought to believe that P’ must be deflated so as not to imply that S is blameworthy for failing to believe that P. Epistemic blame is incurred—if at all—for more persistent intellectual shortcomings. Defending these claims is not part of the plan. The point was to prove that the argument from conceptual economy misfires. As the analysis has shown, there is no reason to doubt that we have epistemic duties and some reason to think that we do. Suspension of judgement, therefore, is more reasonable than denial. I don’t know if there are epistemic duties. But I think there might be.
