Global Supervenience and Dependence 1

Two versions of global supervenience have recently been distinguished from each other. I introduce a third version, which is more likely what people had in mind all along. However, I argue that one of the three versions is equivalent to strong supervenience in every sense that matters, and that neither of the other two versions counts as a genuine determination relation. I conclude that global supervenience has little metaphysically distinctive value. Since its introduction, people have been alternately pleased and disappointed with global supervenience. On the one hand, it seems to have quite a lot of promise for capturing the idea that one set of properties depends upon another while neither making that dependency implausibly localized, nor implying that the very same things have the subvening and supervening properties. On the other hand, however, it has never been entirely clear whether it is strong enough to capture interesting dependencies. Clarifying these issues requires getting significantly clearer about just what global supervenience amounts to. I begin by distinguishing three different versions of global supervenience, and discussing the entailment relations among them, as well as between them and strong supervenience. I argue that one of the three forms is not interestingly different from strong supervenience, and then argue that neither of the other two capture ways in which one set of properties really determines another. They are not strong enough to be useful or important. Despite its apparent promise, then, global supervenience turns out to be little more than a chimera. 1. Versions and Versions of Global Supervenience Let us begin with the good old-fashioned definition of global supervenience. It is quite straightforward; it simply says that worlds that are alike in a certain way must also be alike in another way: 1 I am extremely grateful to Ted Sider, who provided unbelievably detailed comments on several drafts of this paper. I would also like to thank John Hawthorne and Mark Moyer for helpful discussion. Long after I submitted this for review, I learned that Oron Shagrir was working on similar issues; those interested should also see his forthcoming paper.