are synthetic a priori judgements possible?" In both cases, i~thas usually been t'aken for granted in fife one case by Kant that synthetic a priori judgements were possible, and in the other case in contemporary,'d-". philosophical literature that contingent statements of identity are ppss. ible. I do not intend to deal with the Kantian question except to mention:ssj~".

Metaphysics, as I understand it, is the attempt to construct theories which give correct accounts in general terms of pervasive structural features of reality. Though not precise and not intended as an explicit definition, this characterization is comprehensive enough to include both descriptive and revisionary varieties of metaphysical theory. The enterprise of descriptive metaphysics, Strawson tells us, consists in describing “the actual structure of our thought about the world.” Presumably a philosopher would favor this approach to metaphysics if he or (...) she believed either that the actual structure of our thought about the world provides, at least for the most part, an accurate portrait of the world’s structure or that the actual structure of our thought at any time provides privileged access, or perhaps our only access, at that time to the structure of the world. Revisionary metaphysics, by contrast, is concerned to produce a structure of thought about the world better than the one we actually have. Obviously a philosopher would adopt this approach if he or she believed that the actual structure of our thought about the world provides, in at least some important respects, an inaccurate portrait of the world’s structure and stands in need of revision in the interests of correctness. And clearly a philosopher might hold that the actual structure of our thought is quite accurate in some respects and need only be described to yield correct theory and yet is thoroughly inaccurate in others and requires substantial revision if correct theory is to be the outcome. Philosophers may, as I suppose most of them do, practice by turns descriptive and revisionary kinds of metaphysics. No global commitment to one approach or the other is required; the choice between approaches can be allowed to rest on the exigencies of local problems. (shrink)