What sorts of things are reasons?

Rain. It is raining. Furthermore, I have formed (through a good epistemic process) the belief that it is raining. I am about to go outside, and I take my umbrella, because I don’t want to get wet. What is my reason for taking the umbrella? Possibilities: (i) My belief that it is raining. (ii) The fact that it is raining. Here is an argument motivating (ii), which I've recently heard independently in lectures by Jaegwon Kim, Jamie Dreier, and John Broome: Suppose I am deliberating on the question of whether to take my umbrella. It would certainly not be rational of me to merely sit and introspect my beliefs in order to discover whether I believed it was raining – my belief at the time of my deliberation begins is irrelevant. If I want to know whether to take my umbrella, I ought to try to determine whether it is in fact raining, not whether I believe it to be raining. So I should look out the window, not into my head. But I think it's hasty to conclude that reasons are facts. Consider: New Age CD. Suppose that I am preparing to go outside and deliberating on whether to take my umbrella. I hear a heavy patter on the roof and occasional booming thunder-like sounds and form the belief that it is raining. I take my umbrella outside and surprised to find the skies sunny and clear – the sounds I heard were from a New Age CD being played loudly upstairs. It seems somewhat plausible to say that my reason for taking my umbrella outside was my belief that it was raining. It does not seem at all plausible to suppose that the reason was the fact that it was raining, for there is no such fact. And surely my action was not irrational – I certainly carried my umbrella outside for some reason. The defender of reasons-as-facts might reply that there is indeed a fact which provided my reason for action -- the fact that my neighbor was playing a CD which sounded like heavy rain. In one sense, this is satisfying -- in many contexts, it would provide a good answer to the question, "why did you bring your umbrella outside?". But in another, it is not. It seems odd to claim that my action was both rational and for a reason which I grossly misidentified. I didn’t believe that my neighbor was playing a CD. I believed that it was raining. This points to a criterion for reason-status. In agency-ideal circumstances (by which I mean circumstances where good, self-aware, non-akratic agency occurs) like Rain and New Age CD (which is agency-ideal but not epistemically-ideal), reasons must be internally accessible. This seems at least prima facia to be a case against the idea that facts in the world can be reasons for action. How can an external fact be cognitively accessible? It is not at all clear that I have access to the fact that it is raining in question in Rain, and in New Age CD, I certainly do not have access to either the fact that it is raining (which does not exist) or the fact that my neighbor is playing a rain-like CD (which I am completely unaware of). So both beliefs and actions seem to have serious problems. Could reasons be something else? We want our reasons to be internally accessible, yet also constitute good sensitivity to our environment. Here's a crazy idea: appearances. Appearances meet all the criteria I’ve discussed here for reasons. They are internally accessible. And if I am deliberating whether to take my umbrella with me when I go outside, it does not at all sound ridiculous for me to consider whether it appears to me that it’s raining (after all, this is the best that we can do when we wish to consider whether it’s raining). Furthermore, appearances as reasons easily account for deviant sources of belief: in New Age CD, I do take my umbrella for a reason – it sounded to me like it was raining. I know there's a ton of literature on this stuff, and what I'm saying is pretty simple. That suggests to me that it's either been discussed before or is obviously wrong. Probably the latter. It's not obvious yet to me, though, what's wrong with this view. Suggestions?