Some things I've been thinking about while wrapping up my "History of Ethics" paper....
Rawls, in his "Theory of Justice," makes it clear that his primary enemy is utilitarianism. His grand project is to come up with a plausible more or less Kantishly-flavored alternative to it. Despite this, he praises utilitarianism for giving us a single consistent principle of justice, thereby ruling out the possibility of conflicting obligations. He admits that this would be best, but since he doesn't find their one principle defensible, he thinks the next best thing is to postulate a lexical ordering of obligations, such that when obligations stemming from Principle A conflict with alleged obligations rooted in Principle B, the A-obligations always win and so on down the line.
Now, to the extent that I'm a moral realist (which I try to be, or at least I do on Mondays, Wednesdays, alternate Tuesdays and maybe on Yom Kippur even when it falls on one of the other days), I take it for graned that it is never true that one is morally obliged to do P & ~P. (On Thursdays-through-Sundays, I'm even more sure of it!) So, programmatically, my sympathies are entirely with Rawls here.
What I find interesting, though, is that he makes no arguments, none whatsoever, to tell us *why* to think this should be the case. In what is, at least to me, one of the most interesting lines in the whole of the ToJ, Rawls attacks intuitionism by saying that if we don't have a knowable principle for deciding between conflicting prima facie obligations, "the means for rational discussion come to an end."
So....why? Let's put it this way. There are at least three options when dealing with a prima facie obligation to do P and a prima facie obligation to do ~P.
(1) Use some principle to decide between P & ~P, as Rawls and his utilitarian opponents both do.
(2) Admit, as Rawls castigates the inuitionist for doing, that there's no way to decide, that you just have to go with your gut on a case-by-case basis, but take it as a given that of course the conjunction of an action and its negation can't be obligatory.
(3) Say, as Graham Priest does in his chapter on Philosophy of Law in "In Contradiction," that there is no general reason to assume that contradictory prima facie obligations need always to be merely prima facie.
Now, perhaps Rawls is right that there's nothing much to discuss given (2)--it's hard not to think of Stephen Colbert's inspired White House Correspondents dinner riff on knowledge based on the gut rather than on the head--but Rawls takes it as a given that (2) being unpalatable, we must go with (1). Why?
Well, if we assume that the underlying logic of this rational discussion need be classical, it's certainly the case that in standard classical logic extended with deontic operators, you can conclude ~O(P) from O(~P), and hence [O(P) & ~O(P)] from O(P & ~P). This could be seen as problematic due to the alleged explosiveness of contradictions--that is to say, on the assumption that the underlying logic of Rawls' "rational discussion" is classical, and [O(P) & ~O(P)] is sometimes true, then any randomly chosen O(Q) would also be true. Hence, if in a lifeboat situation, you were morally obliged to save your sister and your mother from drowning, and it was impossible to do both, then it would follow that you were morally obliged to go around killing puppies. The strong moral intuition that we are not in fact so obliged might be considered, on an intuition-reliant reflective-equilibrium sort of model of moral reasoning, to be taken to be pretty good evidence that we never have inconsistent obligations.
So far, so good. By why should the means of rational discusison about morality have to be based on classical logic? It seems fairly clear that--since, as we've seen in earlier posts, the Duns Scotus proof follows from logical rules that only make sense if we assume that the Law of Non-Contradiction is universally true--if there are indeed inconsistent obligations, then the underlying logic of rational moral discussion had best be paraconsistent.
So, given that, why should we rule out (3)?
A moral philosopher could leave the refutation of dialetheism to the logician and assume that, since dialetheism is false, there are no moral dialetheias, but if inconsistent obligations are taken to be part of the motivating evidence for dialetheism, the burden is distributed the other way around. (Indeed, when Graham Priest came to Miami last spring, he told me that inconsistent obligations are the most compelling cases of true contradictions. While his primary focus in this work is on the philosophy of law, he's very clear in the chapter on this in "In Contradiction" that any normative system is likely to give rise to similar examples.) As such, the logician needs the moral philosopher to do his part for the refutation of dialetheism by giving us independent grounds for supposing moral obligations to be necessarily consistent.
The obvious move is to say that "ought implies can," and we can't engage in contradictory actions. You can't both save your mother and (by saving your sister on the other end of the lifeboat) not save your mother. Even Graham Priest, who has a detailed argument in "Doubt Truth to Be a Liar" that there are no contradictions in the "observable world," would grant this much. (As a side note, I think that his argument for the consistency of the observable world is extremely dubious, and that if the Law of Non-Contradiction is not universally and necessarily true--which of course I think it is--there are no particularly good reasons to believe the observable world to be consistent. BUT that would get us well off-track from the present discussion, so for the moment, I'm happy to grant Priest the point.) But why should we suppose that ought does indeed imply can?
After all, in contemporary work on moral responsibility, that principle is less secure than ever. Frankfurt in particular has given us some extremely compelling thought experiments that pump our intuitions in the opposite direction. In his cases, we do in fact morally judge people even when they could not to otherwise, and statistical evidence from experimental philosophers have shown that most people's pre-philosophical intuitions lie with Frankfurt.
So what's a moral realist anti-dialetheist who takes Frankfurt's examples seriously to do? There may be a way of reconstructing "ought implies can" that freely admits that "can" need not be an actual physical possibility for a given agent as required in libertarian (or even traditional compatibilist) conceptions of free will, but that is still carefully enough construed to rule out things that are never physically possible for any agent under any circumstances, but caution and independent grounding would be sorely required here to avoid making this completely ad hoc and question-begging.
Anyone have any ideas?