Abstract

We ordinarily make statements of the form "They are the same thing," if there has been reason to distinguish what we now judge identical. But such statements seem not to make sense. "They" indicates that there are more than one thing, whereas "same" indicates that there is only one thing. How can many be one? Hume's obscure Principle of Identity passage in the Treatise addresses this problem . Call it the Number Problem for Identity. Clarifying Hume's account reveals that, despite its richness and suggestiveness, it is ultimately unsatisfactory. In discussing identity, he tries to solve an apparent number problem at the level of objects of ideas, only to leave himself with a real one at the level of ideas. He tries to solve the number problem for identity by appeal to one continuous whole and its many continuous parts, only to leave himself with a different number problem for parts/whole. He tries to make discontinuity necessary for numerical difference, but neither are. ;I develop his account into a satisfactory one, influenced by British Idealist discussions of Identity in Difference and by Hume's discussion of Distinctions of Reason. In response to I propose the concept of "Difference in Respect" based on the dictum that it is possible for an object in different respects to have and lack a property. In the problematic identity attribution I began with, "they" refers to the object in certain various respects, while "same" indicates that it is one object. This is a solution to the number problem for identity. In response to I solve the number problem for parts/whole by suggesting that parts and whole exist and count in different contexts. Which context is the better one to count in varies with purpose and interest. In response to I suggest that Hume's supposedly necessary conditions for numerical distinctness become criteria for judging between contexts, e.g. for many purposes it is better to count a single continuous whole rather than its many continuous parts. I end by applying my newly developed theory to other problems. I show how the same thing can have or lack a property depending on contexts determined by interest or purpose