Sorry that so many of my posts lately have been rough, loose, and sometimes cryptic notes; I've been pressed for time when at the computer. Here's another. It would be nice to have some workable taxonomy of the the way reasons relate to what they support, one based on structural links between the reason and thing believed rather than on content. Here's a rough first attempt.
One way in which reasons for believing X might be related to X is directly: that is, the reason (defeasibly or indefeasibly) implies X (either on its own or with additional assumptions regarded as plausible). So let's call those implicative reasons.
Another way might be this. Some (relatively rare) feature or quality is taken to be desirable, and X has it. (For instance, you might think that a good explanation of moral practices should also explain ordinary etiquette; and if this is hard to find, and X allows you to explain both moral practices and ordinary etiquette, that might be taken as a reason for believing X.) Let's call those desiderative reasons, because they tell us that the thing believed conforms to certain independent desiderata.
I think we can add a third genre. A moral realist might think his moral realism to be a reason for believing some form of aesthetic realism, because they are alike, despite not thinking that moral realism implies (defeasibly or otherwise) aesthetic realism and despite not thinking that being like moral realism is itself a desideratum for an aesthetic theory. We can say that this sort of reason is an extrapolative reason, because it says that a plausible extrapolation from the original reason would, on some supposed principle of extrapolation, include the thing to be believed. I suspect that extrapolative reasons are incidental byproducts of particular combinations of implicative and desiderative reasons, but I haven't thought it through completely.
Am I missing any? And no doubt there are important structural subgenres (e.g., based on the type of implication for implicative reasons, on the type of of means-end reasoning for desiderative reasons, and on the underlying principle of extrapolation for extrapolative reasons). Are any of those of special note?