Monday, January 23, 2012

What should evidentialists say about exculpation and inculpation?

NB: I'm currently writing up a paper on evidentialism and the epistemic norms that govern practical reasoning. Details need to be filled in, but this is a rough outline. One of its ultimate aims is to show that the widely held assumption in epistemology that what's rational must be justified is mistaken.

Everyone agrees that what you ought to believe and what you can justifiably believe depends upon your perspective. Nobody defends the view that the fact that your beliefs fit the facts is itself sufficient to justify belief. Facts that are obscure to you cannot justify your beliefs, only the evidence you have available can. There is some disagreement as to whether facts that are obscure to you can prevent your beliefs from being justified and whether your epistemic obligation can be to refrain from believing things that you reasonable believe because, say, your beliefs do not constitute knowledge. Evidentialists believe that justification is determined by relations between your beliefs and your evidence and does not depend upon any further facts that you might be non-culpably ignorant of. On their view, the fact that your experiences are not veridical might explain why your perceptual beliefs do not constitute knowledge, but it has nothing to do with whether your perceptual beliefs are justified.

Those who accept this view either have to say that you satisfy epistemic norms governing assertion, belief, and practical reason so long as you fit your beliefs to your evidence or they have to say that the justification of belief has nothing to do with whether you conform to these norms. I don’t think the second option is all that attractive. If Hawthorne and Stanley are right that you ought not treat p as a reason for action simply because you don’t know p and you can justifiably believe p without knowing p, situations can arise in which you justifiably believe p but ought to exclude the belief that p from deliberation about what to do even when you know that φ-ing would be the thing to do if p. If I know that the thing to do is take an umbrella when it rains and justifiably believe that it is raining, it seems strange to think that I cannot justifiably reason from the premise that it is raining in trying to decide whether to take the umbrella since I can justifiably reason from the premise that it is raining in reasoning about whether I ought to take an umbrella. (Reasoning about whether to take an umbrella is practical reasoning, but reasoning about whether it is true that I ought to take an umbrella is theoretical reasoning. Given the plausible closure principle for justification that one has justification to believe q when one knows that q follows from p and one has justification to believe p, we get the odd result that one may reason from the premise that p in reasoning about whether it is true that one ought to φ but may not reason from that premise when reasoning about whether to φ if the justification of belief does not depend upon whether you conform to the epistemic norms governing practical reasoning.)

The evidentialists have to say that just as the justification of belief depends entirely upon whether your beliefs fit the evidence you have on hand, conforming to the epistemic norms governing assertion, belief, and practical reason (if there are any such norms) depends entirely upon whether your beliefs fit the evidence you have on hand because you cannot justifiably believe what you do and then fail to conform to these epistemic norms. Thus, the evidentialist should say that Williamson is wrong when he says that you should not assert what you don’t know and Hawthorne and Stanley are wrong when they say that you should not treat what you don’t know as a reason for action. Justification justifies and whether your beliefs are justified does not depend upon whether they constitute knowledge.

Because it seems that we often do say things like, ‘You shouldn’t have let your dog off the leash because you didn’t know that he wouldn’t attack the other dogs’ and ‘You shouldn’t have used that needle because you didn’t know if it was clean’ and because it seems that the evidentialist has to deny that these sorts of claims are strictly speaking true, I think it is fair to ask why we should think that justification depends exclusively upon relations between beliefs and the evidence and not in any way upon further facts about the relations between your beliefs and the world. The standard argument for the evidentialist view goes something like this. If you consider two subjects, one subject in the good case and an internally similar subject in the bad, we know that one subject knows things that the other doesn’t (that’s what good and bad cases are) but it seems that these subjects are equally reasonable or equally rational in what they believe precisely because they believe what they believe for the same reasons. Moreover, believing for those reasons must be rational because one of these subjects has knowledge. So, whether your beliefs are rational depends upon the evidence you and your doubles share in common, not further facts that determine whether your reasonably held beliefs constitute knowledge. Because these authors think that any rationally held belief is a justified belief, intuitions about cases of error are supposed to support the internalist evidentialist view on which your belief about p is justified iff all of your non-factive mental state duplicates have sufficient justification to believe p.

In previous work I’ve argued that this view is mistaken because it denies the possibility of consequential moral luck. Suppose that you do use a dirty needle in giving an injection and your patient gets an infection as a result. Your belief that the needle was clean might have been justified and you might have known that if the needle was clean, you should use it to give an injection. I claimed that while you might be excused for causing your patient the infection, it was nevertheless wrong to inject using the dirty needle. (Among other things, you have reparative duties when you cause these sorts of injuries whether or not you were culpable in your beliefs and such duties suggest that you’d be responding to some previous wrong.) The evidentialists respond by saying that the objection assumes an objective account of obligation and insist that because the belief that the needle was clean was reasonable, the use of the needle was regrettable but not wrong. Just as rational beliefs are justified beliefs, rational actions performed by morally conscientious agents are justified.

What’s interesting (to me) is that my opponents seem to agree that if it would be wrong to φ, you should not judge that you ought to φ. That is, they agree that there is some principled link between the normative standing of a belief and the normative standing of the actions that they rationalize, but they insist that because rationality and justification come to the same thing features of the situation that the subject is non-culpably ignorant of cannot be the features of the situation that would make it wrong for the agent to act. They say that because what’s rational is justified (R=J), you cannot use cases of consequential moral luck to cause trouble for their view. They acknowledge that given their commitment to R=J, they have to accept that there’s a principled link between the justification of belief and action, but they think we should be internalists about the justification of both.

This debate is an in-house debate between unificationists, people who think that the demands of theoretical and practical reason are unified in such a way that your obligation will never be to act against your own justified judgments about your obligations. We all accept (a version of) Ewing’s Principle (or claims that obviously entail it), which states that it is proper to φ if one properly believes that one ought to φ. While I would argue from an objectivist account of practical obligation against the evidentialist account of justified belief, my opponents argue that there are epistemic constraints on practical obligation and that my approach gets things backward.

It looks as if you have to accept their internalist account of the unificationist view given R=J. Surely it is not rational to act against your own rational judgment about what you ought to do. In this paper, I want to explain why we should reject the internalist unificationist account. Contrary to what these authors suggest, the rational is not the mark of the permissible or the justified. I’ll set aside cases involving consequential moral luck. Even if all moral evaluation is of the agent’s will and is never concerned with the consequences of an agent’s actions and even if an agent’s obligation is never to successfully bring about some state of affairs but only to try, we should reject the internalist unificationist view.

The unificationist either thinks that there are practical constraints on epistemic normativity or epistemic constraints on practical normativity.
The objectivist believes that you ought to φ iff φ-ing is the best option you have (i.e., the option that maximizes deontic value). On Zimmerman’s formulation, the prospectivist believes that you ought to φ iff φ-ing is the prospectively best option you have (i.e., the option that maximizes expectable value). Expected value is a function of the probability of an outcome and that outcome’s value. Expectable value is a function of the probability of an outcome and the probable value associated with that outcome. The probable value of an outcome is determined by the agent’s evaluative evidence, not facts about value that might be obscure to the agent.

Some prospectivists (e.g., Jackson) prefer to think of an agent’s obligation as the obligation to maximize expected value, but I think that this cannot be the view that the internalist unificationists accept since it ranks options in terms of values that the agent might know nothing about or values that the agent might rationally believe not to be values.

On the objectivist account, ignorance and mistaken belief exculpate (when they do) by blocking the inference from observations about someone’s behavior to judgments about the quality of the agent’s will. While the objectivist is not committed to the view, the objectivist can say that factual ignorance and mistaken factual belief exculpate in a way that moral ignorance and mistaken moral belief do not.

On the prospectivist account, ignorance and mistaken belief prevent facts that the objectivist takes to be wrong-making features from constituting reasons not to act and so obviate the need to excuse many of the behaviors that the objectivist classifies as wrongdoing. Mistakes and ignorance (whether moral or factual) subvert obligation rather than excuse the failure to meet an obligation. (I think this is somewhat misleading, but you might say that while the objectivist view will say that factual mistakes excuse and moral mistakes typically don’t, the prospectivist takes all reasonable mistakes and non-culpable ignorance to be on par when it comes to determining what an agent’s responsibilities are. You might consider these epistemic defects and limitations to be justifying conditions rather than excusing conditions.)

The internalist unificationist is committed to the prospectivist view on which an agent’s obligation depends (in part) upon the agent’s evaluative evidence. It is possible for an agent’s evaluative evidence to strongly support the view that only the suffering of creatures that can assume responsibility for their actions carries any moral weight, in which case this subject can reasonably believe that they have no obligation to refrain from performing acts that harm animals and have no obligation to perform actions that could help a suffering animal. While those who perform acts that show indifference to the suffering of sentient creatures should not be excused for their wrongful behavior, the prospectivist account says otherwise because it says that there is no wrong to excuse. Reasonably mistaken moral views and non-culpable moral ignorance neither excuses nor subverts obligation.

Response: This objection assumes something like the quality of will account of moral responsibility defended by Arpaly (among many others) according to which what makes an agent’s actions blameworthy is that it manifests the lack of proper care about what is morally significant where this is understood de re instead of de dicto (i.e., care about features that themselves have moral significance vs. care to respond to morally significant features of the situation under that description). But surely the internalist unificationist should reject this account. This objection is no more persuasive than the objection that focuses on cases of consequential moral luck.

I think it’s clear that the internalist unificationist has to reject the quality of will account of moral responsibility. Because of this, I don’t think the internalist unificationist can explain why Rosen and Zimmerman’s argument for skepticism concerning moral responsibility fails. They argue that you can only be culpable for wrongdoing when you act in the belief that your action is wrong. (Cases where you φ believing that φ-ing is wrong are cases of ‘clear-eyed akrasia’.)

1. Suppose Adam φ’d and φ-ing was wrong, but Adam was ignorant of this fact simply because Adam did not believe that φ-ing was wrong.
2. One is culpable for ignorant behavior only if one is culpable for the ignorance in or from which it was performed.
3. Thus, Adam is culpable for having φ’d only if he is culpable for the ignorance from which he φ’d.
4. However, one is culpable for something only if one was in control of that thing.
5. Thus, Adam is culpable for having φ’d only if he was in control of the ignorance.
6. One is never directly in control of whether one believes or does not believe something. Control over belief is always indirect.
7. If one is culpable for something over which one has merely indirect control, then one’s culpability for it is itself merely indirect.
8. One is indirectly culpable for something only if that thing was a consequence of something else for which one is directly culpable.
9. Thus, Adam is culpable for having φ’d only if there is something else, ψ, for which he is directly culpable and of which the ignorance in which he φ’d was a consequence.
10. But, whatever ψ-ing was, it cannot itself have been an instance of ignorant behavior because then the argument would apply all over again to ψ-ing. ψ-ing must have been an item of behavior that Adam believed at the time to be wrong.
11. Thus, Adam is culpable for having φ’d only if there was some act or omission, ψ-ing, for which Adam is directly culpable and of which his failure to believe that φ-ing was wrong was a consequence and ψ was such that Adam believed that it was wrong to ψ at the time he ψ’d.

Rosen argues that your ignorance is culpable only if you culpably fail to manage your procedural epistemic obligations and maintains (rightly, I think) that those who form mistaken moral views or fail to hold correct moral views are not invariably guilty of a failure to meet their procedural epistemic obligations. He then says:
When we engage the examples [of moral ignorance and mistake] in imagination, bearing it fully in mind that the agent is not responsible for his moral ignorance, then our capacity to blame is neutralized by this very thought. Moreover, this is not simply a psychological observation. When we find ourselves unwilling to blame the agent who acts from blameless ignorance, it is because we have come to think that it would be a mistake to blame him (2003: 71).
Perhaps his position seems intuitive if we remind ourselves that to deny it would be to blame someone for an action while at the same time conceding that the action was done because the agent had arrived at her beliefs by taking every reasonable step to ensure that the agent would act rightly. If by our lights the agent had taken every reasonable precaution against doing the wrong thing, how can we then turn around and blame them for doing the wrong thing?

The reason I think that (2) is false and Rosen’s defense of (2) fails is that (in keeping with the quality of will account) I think beliefs are culpable only when they are formed in such a way that it shows that the agent does not have proper concern for what has epistemic value, not that the agent does not have proper concern for moral values. Whether an action is culpable, however, is determined by whether the agent who performed it showed proper concern for what has moral value. I don’t see any reason to think that any agent whosoever is epistemically on the up and up will thereby show proper concern for the relevant moral values. (This is what cases of reasonable disagreement show.)

The thing to notice is that the internalist unificationist is committed to (2). Given that they take failures to believe and act with justification to be culpable failures and believe that justified beliefs justify the actions they rationalize, they cannot say that the subject that justifiably believes that she should φ could have an obligation to do otherwise and could not be culpable for acting on her moral belief. In light of this, the only way for the internalist unificationist to resist the skeptical argument is to identify some other premise that they think is false and reject it.

I should hasten to add that I don’t see how the internalist unificationist could make sense of why we are culpable in cases of clear-eyed akrasia. Because they reject the quality of will account, it is not simply the wrong-making features of the action that the subject is cognizant of that makes the agent’s actions blameworthy. (Moreover, it’s not clear that the action will always have WMFs in the relevant cases because the subject’s belief about her own obligations might be mistaken.) They might say that the action is culpable because the agent shows insufficient concern for the rational authority of morality, but Rosen points out that someone can have mistaken views about the rational authority of morality without arriving at these views by culpably mismanaging their opinions. Moreover, cases of ‘inverse akrasia’ suggest that acting against your own conscience doesn’t make your actions culpable.

If somehow (never mind how) the internalist unificationists find a way of resisting the skeptical argument and showing that someone can be culpable for their actions, I think that they face serious problems with the way in which ignorance and mistake would seem to inculpate on their view. The following case seems possible. Adam and Steve have been in a loving monogamous relationship but their intuitions about the permissibility of allowing yourself to give in to sexual desire have started to change. They have the firm intuition that it is wrong to give in to sexual desire unless one has an overriding reason to do so and further believe that the only justifying reason would be to try to procreate. Since they know they cannot procreate, they both judge that it would be wrong for them to continue having sex. Not that that’s ever stopped anybody.

There’s nothing the prospectivist could say to rule out the possibility that subjects like this might have a body of evaluative evidence such that engaging in sex that could not lead to procreation will score rather low in terms of its expectable moral value. Indeed, there’s nothing that the prospectivist could say to rule out the possibility that Adam and Steve would know (assuming prospectivism) that their obligation is to refrain from having sex. Not that that’s ever stopped anybody.

Here’s one part of the objection. We (and I’m assuming that we don’t think that there’s any reason for people not to engage in non-procreative sex) don’t think that Adam and Steve’s actions display a willingness to act against any genuine moral reasons, but it seems we would have to blame them given a very plausible principle concerning blame, which is that we blame people who can be held responsible for their actions if they knowingly perform actions that are overall wrong when they are not under duress. Yet, I think it would be wrong to blame the protagonists of my story for giving in to sexual desire.

Here’s the other part of the objection. In blaming people for their actions, we often do so in part because we desire that they not act badly. As before, desiring that you do right and avoid doing wrong can be understood de dicto or de re. Insofar as we don’t think Adam and Steve’s actions have any WMFs, our desire for them not to act badly could not be the desire for them not to respond to the underlying features of the situation in the way that we do. Rather, our desire for them not to act badly would have to be understood as the desire for them to either refrain from acting on the de dicto desire to do wrong or to act as if they have the de dicto desire to do right. But this makes moral fetishists of us. (The moral fetishist is, as Smith puts it, “precious, overly concerned with the moral standing of their acts when they should instead be concerned with the features in virtue of which their acts have the moral standing they have”.)

So, not only does the internalist unificationist view suggest that if we should ever blame people we should blame the wrong people, it threatens to turn us into moral fetishists.

There is a fourth objection to consider, which is that the internalist unificationist view offers us an impossibly easy refutation of the error theory, but this is still under development and that’s best saved for another time.

If these objections have any force, they suggest that the make of the permissible and the justified cannot be the rational or the reasonable. My own view is that excuses exculpate only by showing that an agent with the right values did everything that could be expected of them and still failed to meet her obligations. If this view is right in the practical sphere, I think given Ewing’s Principle it has to be right for the theoretical sphere. The main lesson here is negative. It is that R=J is false and that the standard rationale for the evidentialist view is unsound. I would argue that the considerations offered here also show that the evidentialist view is itself mistaken since I do think that the justification of belief depends upon whether that belief could justify the actions the belief rationalizes and don’t think that the justification of your actions depends entirely upon the evidence you have on hand, but that’s better left for another occasion.