Sunday, October 13, 2013

The Best Case for Uncertaintism

Liz Harman's excellent Madison Metaethics talk on 'The Irrelevance of Moral Uncertainty' [pdf]* got me thinking again about whether false normative beliefs can be justified.  As I previously noted, we might think that the fundamental normative truths, being non-contingent and presumably a priori, are always what it'd be ideally rational to believe.  Then again, insofar as they can also be highly non-obvious, we might think there's an important (less "strict") sense in which we rationally ought to apportion our credences broadly over a range of competing normative views.

One reason you might think that false moral beliefs can be justified is that we can have "procedural" (e.g. testimonial) evidence for any claim whatsoever.  For example, if everyone a mafioso trusts and respects since childhood has drummed into them that it's right to kill innocent people to promote the family's financial interests, perhaps you'll allow that there's some sense in which they end up being "justified" or "reasonable" in believing this.

Nonetheless (as Liz argued), this doesn't seem to justify their immoral actions in any significant sense -- the mafioso is still blameworthy when they kill innocent people, since they are in fact showing insufficient regard for others' interests, no matter how laudable they believe themselves to be.  So this kind of case -- involving what I've called "procedural" kinds of evidence -- seems problematic for Uncertaintism: the view that what we ought to do depends on our moral credences (as well as our ordinary credences about non-moral matters of fact).  As this case shows, false moral views supported on merely "procedural" grounds seem not to influence what we ought to do.

However, I think it would be too quick to conclude on these grounds that uncertaintism is false.  For that would be to neglect that, "procedural" evidence aside, one might also think we can have "substantive" reasons to believe certain false normative views -- reasons to believe them simply in virtue of the intrinsic credibility of their content.  Not beliefs like the mafioso's, presumably (they seem intrinsically morally repugnant).  But the kinds of moral views that philosophers reasonably dispute.  To borrow an example suggested by Jamie Dreier in the Q&A: Suppose you can either prevent one death or one million headaches (supposing that the latter is just slightly greater in terms of aggregate utility).  Suppose utilitarians are actually right in claiming that you should prevent the headaches in this sort of case (though it's a close call), and that the competing anti-aggregative view (i) is intrinsically credible, at least to some non-trivial extent; and (ii) holds that it would be much worse, morally speaking, to prevent the headaches.  In such a case, the "uncertaintist" claim -- that you should "play it safe" by preventing the one death -- seems more plausible.

Now, I'm not sure whether uncertaintism is right even about this case. Perhaps no false a priori claims can be "intrinsically credible" in the required sense -- perhaps any candidates we might put forward are merely psychologically "credible"-seeming to us, and not really intrinsically credible (or warranting credence simply in virtue of their content) in the way that true a priori claims presumably are.  To think otherwise would certainly raise puzzling questions about how (some but not other) false contents could be intrinsically credible in this way.

But it seems to me that the best hopes for uncertaintism rest on this question.  If some false moral views are intrinsically credible in this way, then it's plausible that we really ought to take these "moral risks" into account.  But if false moral views are never intrinsically credible -- so that any possible "justification" for them takes the merely procedural form found in the Mafioso case -- then Liz's case against uncertaintism seems to me very persuasive.

* [Update: PDF shared with permission]


  1. [Andrew Sepielli writes:]

    Does the move I make from subjective-probability-relative to minimal-probability-relative norms help here? If not, why not?

    Relatedly, an example: it strikes me as very plausible that the probability (minimal, or epistemic on the expressivist construals of epistemic probability) of consequentialism is higher conditional on Jonathan Bennett's theory of acts and omissions being true than on that theory's being false. That is why, IF you accept Bennett's theory, it is a perfectly reasonable thing to do to increase your confidence in consequentialism -- which is, of course, what he wants you to do. That's why he wrote up his argument. What's wrong with that? My suspicion is that any answer is going to be grounded in considerations that in fact support the view, defended by Clayton Littlejohn (and maybe others -- I don't know this literature), that there are no false justified beliefs of any sort. But again, I'm open to hearing the other side.

    1. Hi Andrew, could you say a little more about how you see the shift to "minimal probability" talk helping here? Unless our MP-judgments are justified then I don't see them as having much normative interest (they won't excuse objective wrongdoing, etc.)

    2. Though the example of philosophical argument is a nice one for bringing out the intuitive force of the idea that we really should (for some normatively significant sense of "should") have non-extreme credences about all sorts of a priori matters.

    3. I claim that the norms that we most fundamentally use to guide our behavior under uncertainty are not credence-relative norms, but are, rather, minimal-probability-relative norms. MP-relative norms are relative to MP's, not to our judgments about MP's. And claims about MP's express, rather than report, our credences, and their semantic features are to be explained by the credences they express. It is perfectly coherent to say "The minimal probability that Ayn Rand's moral theory is right is very low, even though Leonard has a very high credence in it", for JUST THE REASON that it's coherent to say "P, but Leonard believes not-P." This is all to say MP's are mind-independent in the relevant senses, and norms about what to do given the MP's of normative propositions are utterly insensitive to whatever credences various people happen to have. Similarly, I think anyone who has credences and expresses them either to herself or others has made MP-claims, just as anyone who has a belief in P and expresses it to herself or others has made "P"-claims.

      Now, does MP-relative normativity have anything to do with blame? I don't know, and to be honest, I don't so much care. But it has everything to do with the guidance of action, and since it's not belief-relative or credence-relative in some *allegedly* unseemly way, why not think that there are genuine facts about what you MP-relative ought to do.

      (This is a quick summary of stuff that's been developed at greater length in my 2013 Nous paper, my 2012 OSNE paper, and a forthcoming article on Subjective and Objective Reasons in the Oxford Handbook on Reasons and Normativity.)

      But back to my example -- what's supposed to be the problem again?


Visitors: check my comments policy first.
Non-Blogger users: If the comment form isn't working for you, email me your comment and I can post it on your behalf. (If your comment is too long, first try breaking it into two parts.)

Note: only a member of this blog may post a comment.