Brian Weatherson has an argument that intuition isn't unreliable. My own view of the argument is that it is one of those cases of 'so close yet so far'. Weatherson notes that a lot of falsehoods are counterintuitive, so there is some reason to attribute reliability to intuition; but one might still refuse to consider it reliable for one of two reasons.
(1) We may think it's impossible to measure how reliable or unreliable it is (because there would be infinitely many successes and failures alike, or because there is no clear way to decide what counts as a single intuition).
(2) We may think that the question, "Is intuition reliable?", is ill-formed because 'intuition' doesn't mark a single type of thing, but several different types of things.
He gives arguments in each case that we could still at least deny that appeal to intuition is epistemologically suspect.
I think (2) is where the money is at; as I've noted before, in the eighteenth century James Beattie identified eight distinct types of things, each with its own quirks, to do the same general sort of work our contemporaries are trying to do with the one word, 'intuition'. And what is more, Beattie and his contemporaries had more sophisticated accounts of each of the eight types than we have of this supposedly single type of thing, 'intuition'. Thus I think any defense of intuition has to recognize that it is not a univocal term, whatever different kinds of intuitions may have in common to deserve the single label. So "Is intuition reliable?" is not a good question to ask, because it can only be answered with a question: "Which kinds of intuition do you mean?" Weatherson recognizes this, but the way he tries to handle it involves dividing the classes of intuition in an odd way: he divides them into philosophical, epistemological, moral, etc. This is a very bizarre way to divide the class of things called 'intuition' because it still leaves the question unanswerable. Consider philatelic intuitions. It makes no sense to ask as a general question, "Are my philatelic intuitions reliable?" If someone were to ask such a question, it would still require the clarificatory question, "What kinds of intuitions do you mean?" If we are tracking reliability we don't want to divide intuitions up according to the fields for which they are useful; we want to divide them according to the kinds of things they are. And to add 'moral' to 'intuition' no more clarifies what the intuition is supposed than adding 'philatelic' does; it simply clarifies the subject matter we're thinking about, not the way we're thinking about it. If we were to divide intuitions into classes based on kinds of intuition rather than subject matter (e.g., by adapting Beattie's classification), we could then give a thorough and excellent account of the reliability of this or that kind of intuition. Then we could run the sort of argument Weatherson wants to run.
