The problem of epistemic encroachment is that of keeping epistemic matters from encroaching into practical matters in ways they should not. To get a sense of the problem, it seems that as we ordinarily describe moral rules or principles they seem to be the sorts of things that someone could breach without realizing it. Moreover, they seem to be the sorts of things that someone could breach even when their evidence suggests that their actions conform to the relevant rules or principles. Umpires are not supposed to call a batter out until the third strike. If the umpire harbors the non-culpably mistaken belief that the batter has already had two strikes called and pronounces the batter out upon calling the next strike, the umpire’s call might be excusable but cannot be justified. We are supposed to keep our promises. If we promise to meet with a student and simply misremember the time of the meeting, the mistaken belief that our afternoon is free might be non-culpably held and factor into our decision to spend the day reading at home. The decision not to come in to the office is not justified. The most we can hope for is an excuse. The justification of the decision to stay home requires either that the reason to head into the office is cancelled (e.g., the student calls off the meeting) or overridden (e.g., an emergency requires you to break off the meeting). As ignorance neither cancels the reason to keep the promise nor ensures that there is some overriding reason for breaking it, the failure to keep the meeting is excusable at best. It does not take too much imagination to describe cases in which epistemic counterparts perform actions on the basis of the same motives and acting from the same intentions in which one, but only one, acts with justification. The other agent, we might imagine, is to be excused for breaking a rule or breaching some norm. That this is possible is our first assumption:
(A1) It is possible for two subjects with just the same evidence to act on just the same explanatory reasons where one, but only one, subject acts permissibly.
The problem emerges when we combine this first assumption with a second and then try to reconcile these assumptions with Evidentialism. The second assumption is about motivation. It is often said that there is some internal or necessary connection between judgments or beliefs about what we ought to do and the motivation to do what we judge we ought to. It is difficult to articulate the connection precisely, but a rough account should suffice for our purposes:
(A2) If a subject judges that she ought to Φ and neither lacks the capacities for assuming responsibility for her deeds nor suffers from some sort of incapacitation, she will thereby be motivated to Φ.
It seems that these two assumptions about the justification of action and the link between action and the beliefs that constitute part of the motivation for performing them are incompatible with Evidentialism. If we imagine a pair of epistemic counterparts who both judge that they ought to Φ, we might imagine that the facts are such that it is permissible for only one of them to Φ. (Deny this, and you have effectively denied (A1).) In judging that they ought to Φ, they will thereby Φ unless they suffer from incapacitation or lack the faculties necessary for assuming responsibility for their deeds. Assuming that they do not, however, one will act impermissibly whereas the other will act permissibly. The beliefs however in light of which they are moved to act (i.e., the belief that they ought to Φ) is a belief the evidentialist cannot say that these subjects ought not hold for nothing has been said thus far to suggest that either subject lacks sufficient justification for the belief that they ought to Φ. However, given our two assumptions and the principle ‘ought’ implies ‘can’, we have an argument against Evidentialism. It cannot be that an agent ought not Φ unless they can refrain from Φ-ing. According to Evidentialism, it permissible for an agent to believe she ought to Φ even if she ought not Φ. However, on the assumption that our agent can assume responsibility for her deeds and suffers no incapacitation, she cannot both judge that she ought to Φ and refrain from Φ-ing. So, either she is permitted to both believe and act on that belief or permitted to do neither. The reason that this is the problem of epistemic encroachment is that unless the evidentialist rejects what seems to be an eminently plausible assumption about motivation, the evidentialist must reject (A1) if the evidentialist is to remain true to Evidentialism.
Thinking about this problem should help us see what is wrong with the standard defenses of Evidentialism as well as the arguments offered in its support. Unless the evidentialist rejects (A1), the evidentialist would have to say that while the counterpart of the agent who acted permissibly who failed to do what she ought to have done was no less than fully reasonable, responsible, or respectable insofar as this person was led to act on the basis of the evidence at hand. However, they would have to say, that the agent was no less than fully reasonable, responsible, and is perfectly respectable they failed to act with justification. The evidentialist would have to go on to say that the marks in light of which we say that the subject was rational, reasonable, and responsible for having Φ’d are not the marks of justification. The evidentialist would then be hard pressed to explain why the mark of epistemic justification is that the subject was rational, reasonable, or responsible for having believed. There would be nothing to the concept of justification that would suggest that it followed from the subject’s being fully reasonable, responsible, and rationale that the subject’s responses to the reasons available were justified. So, the supervenience of epistemic justification on the evidence would not be something we could understand in terms of the concept of justification. It would have to be the ‘epistemicness’, as it were, of epistemic justification that was the key to understanding the evidentialist’s supervenience thesis. But, it is obscure why the epistemicness of epistemic justification helps us see why the evidentialist’s supervenience thesis should hold in spite of the absence of any connection between the generic concept of justification and evidence or concepts having to do with responsibility or rationality. What seems distinctive of epistemic justification is that it justifies something aimed at truth whereas practical justification justifies something aimed at something other than truth, but nevertheless external to the agent. If the reason (A1) is true is that action is, or ought to be, concerned with matters external to the subject, then the thesis that belief is, or ought to be, concerned with matters external to the subject (e.g., conforming to the norm that enjoins us to refrain from misrepresenting the world), the epistemicness of epistemic justification is not going to help shed light on the evidentialist’s supervenience thesis.
The evidentialist is free to reject (A1), but this comes with costs. Among them is that the evidentialist would have to say that epistemic encroachment is something we must live with and that the permissibility of an action does not depend on what ordinary subjects take it to. Staying home to read is perfectly fine regardless of the promises that you made because what matters are the promises you seem to have made. Forgotten promises have no bearing on permissibility. Imagined promises, I imagine, would. One of the oddities of this view is that it seems to clash with certain intuitively plausible claims about the kinds of value that ought to factor into a judgment about what to do. Whereas it seems that the values served by kept promises and the values harmed by broken ones are what the agent takes to factor into the judgment about where to be, on this view this sort of value does not bear on whether to keep the promise. Rather, the only value that bears on whether to keep a promise is the value that attaches to, say, an agent’s trying to keep a promise from the right sort of motive. As someone with pluralist sympathies, I would not wish to deny that there is something of value to see in someone’s failures to keep promises, say, when they truly desire to keep them. However, it seems that this sort of value calls for a different sort of response than the value that attaches to, say, a person’s actually keeping a promise. The value that is served by promise keeping and discharging the duty of fidelity is the kind of value that calls for a kind of response, which is the agent’s keeping of the promise. The value that we see in some failures to keep a promise (e.g., the failure that is due to a mistaken belief) is not one that bears on whether the agent ought to act in a certain way, but one that calls out for respect or admiration. It is not the value the deliberating agent has in mind when deciding what to do, but what others have in mind when trying to decide whether to praise the agent or punish her. In fact, it is not clear that it is a value that calls for promotion rather than, say, honoring or appreciation. I cannot see how those who reject (A1) while retaining (A2) in order to remain true to their evidentialist commitments can avoid making just this confusion about the types of value and the kinds of responses these values call for.
The reason that the problem of epistemic encroachment is a problem is that we ought not rewrite ethics in order to accommodate Evidentialism. It seems we would have revise our ethical views to accommodate Evidentialism given our assumption about the link between belief and motivation. If the reader agrees that this much is correct, we not only have an argument against Evidentialism, we can now see what is wrong with a recent argument that has been offered in support of it. In support of the claim that the beliefs an individual ought to have are those that are supported by the evidence that individual possesses, Feldman claims that, “following one’s evidence is the proper way to achieve something of epistemic value”. As he notes:
While true beliefs may have considerable instrumental value, a person who irrationally believes lots of truths is not doing well epistemically. In contrast, a person who forms a lot of rational but false beliefs is doing well epistemically. While knowledge has a kind of value, seeing it as the only thing of epistemic value fails to explain what is valuable about forming beliefs that fall short of knowledge. We avoid these problems associated with identifying epistemic value with true belief or with knowledge if instead we say that what have value are rational beliefs.
These claims about epistemic value have their plausibility, but evidentialism is a claim about justified belief and permissible belief. We need a way to translate talk of epistemic value into claims about justified belief, and Feldman makes two further claims. First, that we maximize epistemic value by forming those beliefs we are rational for forming (i.e., that we maximize epistemic value by following the evidence). Second, that the believer ought to form those beliefs that would maximize epistemic value.
In light of what we have seen, we ought to see now why this argument cannot be used to establish Evidentialism. There might be a kind of epistemic value that attaches to the false beliefs that lead us to act in ways that we ought not. After all, there is a kind of moral value that we can see in someone’s failing to do what she should provided that the failure is excusable. However, just as the kind of value we see in someone’s trying to do what she should but failing to do so is not the value bears on whether the subject should act in the ways that she does, Feldman offers no argument whatsoever for the claim that the kind of epistemic value that attaches to rational, false beliefs is the value that we ought to promote and thus ought to guide our belief formation. Moreover, we have an argument for denying that it is the sort of value that we ought to respond to by forming a belief about some matter. If we were to, say, form the false, but rational belief that we ought to Φ when the evidence indicates that we ought to Φ, insofar as we are rational, we will thereby be motivated to act in ways that we should not. By hypothesis, the kind of value that we can see our wrongful Φ-ing is not the sort of value that calls for action on our part. Thus, the kind of value that we can see in our mistaken belief that we ought to Φ is not a value that calls for forming the belief that we ought to Φ. To deny this is to deny (A2) and sever the connection between belief and motivation. So, I submit, this recent argument for Evidentialism rests on a mistake. The mistake is in thinking that the kind of value that attaches to our rationally held false beliefs is a value that calls for promotion by means of the mistaken belief.