Two-world views

In his new book on Kant, Guyer makes an interesting claim about two-object views. He says that it is misleading to describe ontological interpretations of transcendental idealism as 'two-object' or 'two-world' views. "This makes it sound as if Kant is being supposed to have entirely made up a mysterious world behind the world of ordinary objects, to have needlessly duplicated the objects of our experience while at the same time stripping them of their most important properties. But, for Kant, as for virtually every philosopher in the seventeenth and eighteenth centuries, there already were two sorts of objects to hand, namely, ordinary objects and our mental representation of them, and all that Kant was doing, as he saw it, was relocating spatial and temporal properties from one kind of object that everybody recognized - non-representations - to the other kind of object that everybody recognized - representations. So of course he held a 'two-object' view: everyone (except Berkeley) did, though few would have agreed with Kant's reassignment of spatio-temporal properties from ordinary objects to representations. Any interpretation of Kant's transcendental idealism needs to take account of the fact that, like the vast majority of his contemporaries, he was in fact committed to a 'two-object' view independently of transcendental idealism." (Guyer: 2006, p. 69)

While I agree that Kant believed in the distinction between representations and objects, it is questionable to what extent this already implies a 'two-world' view and moreover, it is not clear whether Kant did transfer the primary properties from objects thus conceived to representations. As he says in the Prolegomena, he made these primary qualities into secondary qualities, but that does not mean that they are not in the object any longer. Rather, the object is seen as a construction out of the representations; it has these qualities, but what makes them secondary qualities is that they are due to the representations rather than due to the noumenal object. In a way we end up with three objects, namely (i) representations, (ii) their objects, and (iii) objects as they are in themselves. The distinction between primary and secondary properties is applicable within the phenomenal realm. We have to distinguish between representations and the phenomenal objects of these representations, i.e. the causal nexuses characteristic of Kant's scientific realism. This is usefully cashed out by van Cleve's distinction between objects1 and object2 - both of them are phenomenal. In addition to these two kinds of phenomenal objects we have noumenal objects. What is at issue is whether noumenal and phenomenal objects occupy different domains or worlds. The question is not whether representations and their objects occupy different domains since it is evident that they both belong to the phenomenal realm. That is, the distinction that Guyer appeals to is a distinction within the phenomenal realm and as such has no direct bearing on the dispute about how many worlds there are. What is special about Kant's system is that he proposes a bifurcation of reality into the noumenal and phenomenal realms. It is for that reason that the 'two-world' label is adequate and not for any reason regarding the relation between representations and phenomenal objects.