Sunday, June 29, 2008

Reasons as evidence

I've posted some of this over at Ethics Etc., but as the thread is dead I thought I'd post a little on the matter here. Daniel Star and Stephen Kearns have written a few papers explaining and defending the view on which normative or justifying reasons are pieces of evidence about what we should believe or should do. It's an interesting view and the arguments they offer appeal to a number of considerations I'd appeal to in arguing that pieces of evidence do not constitute reasons in the relevant sense.

The first worry, which I posted over at Ethics Etc., went something like this. I believe it was Bernard Williams who observed that there can be a kind of conflict among practical reasons that there cannot be among theoretical reasons. If we identify both reasons for action and belief with evidence, I think you’ll have to deny Williams’ observation. So, the observation. You can judge that you ought to X knowing that there are real reasons not to that serve as reasons for regret. Moreover, you can _knowingly_ judge that you ought to X while knowing that there are, as it were, real reasons not to X that are defeated by reasons that favor X-ing. However, if you judge that you ought to believe, you cannot knowingly judge that you ought to believe while knowingly judging that there are real reasons not to believe. The evidence that speaks against believing would be regarded as misleading. I can’t see how this comes out true on a view on which all reasons are pieces of evidence.

The second worry concerns the status of the following thesis:
(OR) If S ought to X/S ought not X, there is a reason for S to X/there is reason for S not to X.

I don't see why we'd say the following is true:
(OE) If S ought to X, there is some evidence that S ought to X/there is evidence that S ought not X.

But, if Kearns and Star are right, I don't see how OE could not be a consequence of OR. Maybe there are good reasons to accept (OE), but I really cannot think of any offhand. Consider cases in which the subject's wrongful action is due to a non-culpably mistaken belief. To say that the beliefs are non-culpably mistaken is to say that they are the proper response to the evidence _available_. But, to say that the action was wrongful was to say that there were reasons not to do it. It seems that what makes the action wrongful is a set of facts that does not strongly supervene on the evidence at hand. To accommodate such cases, it seems that Kearns and Star would have to say that the sort of evidence they have in mind is not necessarily evidence possessed but evidence there is. I'm not sure how to make sense of evidence possessed or not that would not trivialize (OE) so that it just becomes the view that if you ought to do something there are some facts in the world in light of which this is true.


Tony Booth said...

Hi Clayton,

I've been thinking about this issue since last time you posted about it. Here are some concerns:

You seem to be saying that defeated or overidden reasons to believe would (should?) be regarded as misleading, whereas defeated reasons for action are not. But why aren't defeated reasons for action regarded as misleading? I guess that your answer (following Williams) is that such reasons still have normative force over us in terms of their residue, where defeated reasons for belief do not leave behind such residue.

Now, are we to construe this residue in purely psychological terms? If so, I don't think there is a good argument for thinking that we can;t treat defeated reasons for action as misleading via appeal to residue (such reasons could well still be misleading even if we have a contigent psychological attachment to them). So we might contrue 'residue' in normative terms...but then there is no reason to deny that there can be residue in the case of reasons for belief. Just because we don't *feel* the residue does not mean that the defeated reason does not still carry normative force...

Hope that makes some sense!!


Errol Lord said...

I know what they would say about the second worry because I had an email conversation with Daniel about something similar. The 'ought' they are talking about is from the perspective of the all-knowing beneficient by-stander. From that perspective each time OR is true OE is true. Of course, from some agent's perspective it might be that one ought to phi without having evidence for the proposition that they ought to phi. But Star and Kearns don't think that's a problem for their view.

I don't know what to say about the first point. It obviously depends on what you take to be a reason for belief. I like the Williamson view that p is a reason for q iff the probability of q conditioned on p is greater than the probability of q on its own. On that conception, I think that you could believe q while knowing that there is some consideration that lowers the probability of q. Just as long the reasons in favor of q sufficiently raise the probability. But you might want to reject that view of reasons.

My big problem with their project is that they treat 'ought' as a primitive. I think that this skirts a major source of data as to whether they're right--viz. the details about the semantics of the deontic 'ought'. As it turns out, I think that their 'if and only if' claim is true, but their 'if and only if (and because)' claim is false. In other words, propositions are evidence for 'ought' propositions because they're reasons, not propositions are reasons because they're evidence for 'ought' propositions. I'm writing a paper about such things right nowd, as it turns out.

Errol Lord said...

I take back what I said w.r.t your first point.