Abstract

In a famous argument, Hilary Putnam says that cats might turn out to be robots. Suppose that they might. The years since Naming and Necessity have seen followers of Kripke say that, necessarily, cats are animals. This may not contradict Putnam, as long as the occurrence of 'might' in Cats might be robots is epistemic. But we know that cats are not robots. How then can it be epistemically possible that cats are robots? ;In other words, how it could be true that cats might turn out to be robots, cats necessarily are animals, and we know that cats are animals and not robots. ;In Chapter 1, I discuss Putnam's robot cats argument. It is not really possible to complete this before we have a view about the epistemic sense of 'might'. But the discussion helps make clear what we should want from such a view. ;It is natural to think that, when we say Cats might be robots, we mean something like this: We could be in the epistemic situation that we are and epistemic counterparts of cats were robots. Epistemic counterparts of cats, roughly speaking, are things that look, and so on, just as cats actually do. After may early discussion of Putnam, I set aside some recent views about epistemic uses of 'might', and take up this idea. ;The analysis contains two relations: that of being in the same epistemic situation, and that of being an epistemic counterpart. These need formulation. I concentrate on the first, in chapters 2 and 3, give a formulation, and defend the analysis. ;Putnam's premise belongs to a family of epistemic claims where what might be true is, in fact, impossible. Logic and mathematics breed other members of this family. In Chapter 4, I ask whether the above view about epistemic 'might' could be extended to cases like these. I argue that it cannot