Kant on the Number of Worlds

It has long been disputed whether Kant's transcendental idealism requires two worlds – one of appearances and one of things in themselves – or only one. The one-world view must be wrong if it claims that individual spatio-temporal things can be identified with particular things in themselves, and if it fails to take seriously the doctrine of double affection; versions that insist on one world, without making claims about the identity of individual things, cannot say in what way the world as we know it and the world of things in themselves can be ‘the same’. The two-world view must be wrong if it denies Kant's empirical realism, or offers a phenomenalist interpretation of it. On moral grounds Kant ‘identifies’ each human person with a particular thing in itself, but the relationship here cannot be strict identity; instead its closeness may warrant regarding the two distinct entities as part of a composite whole. Perhaps up to the first edition of the Critique, Kant thought that empirical knowledge required a particular kind of close correspondence between appearances and things in themselves, one that would make it appropriate to speak of composite wholes here also. By the time of the second edition, he saw that there could be no good grounds for thinking that. In this respect something a bit like the one-world theory might make more sense for the first edition than for the second; but in both cases there is room to speak of two worlds as well. Talk of the number of worlds is metaphorical, and both metaphors have their dangers.