Until we have settled the question of whether moral beliefs necessarily have motivational force, for example, we are in no position to say whether it is a point in favor of a given account of our moral practice that it endows them with such a force; and until we have decided whether mathematical beliefs can be known a priori, we will be unable to say whether it is a point in favor of an account of our mathematical practice that it allows them to have such a status. A realist or antirealist conclusion therefore represents the terminus of philosophical inquiry into a given area rather than its starting point; and so it is hardly surprising that such slight progress has been made within realist metaphysics, even by comparison with other branches of philosophy.
Kit Fine, ‘The Question of Realism’, Philosophers’ Imprint, vol. 1, no. 1 (June, 2001), p. 25
[T]he much-vaunted analogy with natural kinds is of little help, and actually stands in the way of seeing what the mechanism might be. For our beliefs concerning natural kinds are not in general independent of perceptual experience. If we were to learn that most of our perceptual experience was non-veridical, then little would be left of our knowledge of natural kinds. The brain-in-the-vat is at a severe epistemic disadvantage in coming to any form of scientific knowledge; and if there really were an analogy between our understanding of scientific and of ethical terms, then one would expect him to be at an equal disadvantage in the effort to acquire moral wisdom. It is for this reason that the continuity in moral and scientific inquiry so much stressed by writers such as Boyd and Railton appears entirely misplaced. A much better analogy is with our understanding of mathematical terms, for which the idea of a hookup with the real world is far less plausible.
Kit Fine, ‘The Varieties of Necessity’, in Modality and Tense: Philosophical Papers, Oxford, 2005, p. 258