My takeaway from the discussion of the Buridan’s Ass Paradox is that the issue turns on the relationship between reasons and contrastive reasons. Here’s one version of such a principle:
Reason R is a reason to do A (or believe p) only if, for any B (or q) that competes with A (or p), R is a reason to do A rather than B (or believe p rather than believe q).
Armed with such a principle, one can argue that the ass has no reason to choose either bale, and also has no reason to adopt any strategy for breaking the tie between the two bales. For example, the ass can’t flip a coin to break the tie. Not just because asses don’t have opposable thumbs, but because the ass has no reason to identify any result of the flip with a preference for either bale.
Note that such a principle is part of a common skeptical strategy in epistemology: you have to be able to rule out alternatives to a favored opinion in order for that opinion to be justified. Of course, standard non-skeptical epistemologies resist. Mooreans, such as Peter Klein, will want to be able to come to rationally believe, and know, ordinary claims of common sense and then deduce the falsity of skeptical hypotheses on the basis of these ordinary claims. Relevant alternatives theorists deny that you have to be able to rule out all alternatives, opting instead for having to rule out relevant alternatives. (Neither strategy will help the ass, however, since s/he has no preference for either bale, such as occurs when comparing the ordinary claims of common sense to skeptical hypotheses, and it would be a bizarre account of relevance for one of the two bales to fail to be a relevant alternative to the other.) Similar rejections of the principle above can be found in defeasibility theories, safety theories, and sensitivity theories: they all reject the strong tie between epistemic status and contrastive epistemic status.
If we think about what explanation can be given for the failure of the principle in the arena of action, it it hard to see how to extend that explanation to the arena of belief. Suppose, just for simplicity, that we are preference-satisfaction consequentialists about rational action. The literature on the question of whether there are basic actions shows that there is a heirarchical structure to this domain, akin to the relationship between means and ends. In the paradox, there is a decisive reason to eat rather than not, but one cannot eat without performing some more specific action that constitutes a way of eating. So, here’s an idea in the theory of rational action: we can tolerate violations of the connection between reasons and contrastive reasons so long as such a connection exists at a higher level of generality. Arbitrary actions can be rational in virtue of being means to a goal which is rational in virtue of a convergence of reasons and contrastive reasons.
The same considerations would extend into the doxastic realm when we are talking about the practical rationality, or even the all-things-considered rationality of holding beliefs: to navigate the world and have any decent sort of life, beliefs are needed. Pyrrhonian suspension of belief does not work well (because the idea of acquiescing to the appearances has no suitable non-doxastic interpretation). But what about epistemic rationality itself, the rationality involving the idealized perspective in which we assume only purely cognitive interests? Sometimes I think we should just be hardheaded skeptics here, and say that from this purely epistemic position, nothing passes scrutiny because there is no available explanation why reasons don’t need to be contrastive reasons (and explain away contrary inclinations by appeal to the differences between epistemic rationality, practical rationality, and all-things-considered rationality). In my more accommodating moments, though, I find Ralph’s (Wedgwood, that is) contextualist ideas quite attractive (though perhaps replacing the contextualism with something more subjective). Maybe we should say that it’s a matter of how much risk of error is tolerable in the theory of epistemic rationality in order to make possible… Well, what? Well, what I’d say is: the achievement of the surpassing epistemic good of understanding. If we set the bar very high, we reduce or eliminate the possibility of achieving understanding, and that’s bad, especially from a purely cognitive point of view. So, maybe the great good of understanding functions with respect to specific beliefs the way more generic actions function with respect to specific realizations of them: in both cases, we have an explanation why reasons don’t always need to be contrastive reasons. In some quite interesting sense, both beliefs and actions can be arbitrary and yet rational (arbitrary in the sense of not conforming to the requirement that reasons are always and everywhere contrastive reasons).
Even granting this result, there’s still the hard question of determining in which range of cases reasons have to be contrastive reasons, for surely there is such a range of cases. But since reasons don’t always need to be contrastive reasons, arbitary actions can be rational and so, perhaps, can arbitrary beliefs. Such a conclusion doesn’t by itself block the paradox, but it blocks one way of developing it.