It is widely assumed that the explanatory states of scientific psychology are type-individuated by their semantic or intentional properties. First, I argue that this assumption is implausible for theories like David Marr's  that seek to provide computational or syntactic explanations of psychological processes. Second, I examine the implications of this conclusion for the debate over psychological individualism. While most philosophers suppose that syntactic states supervene on the intrinsic physical states of information-processing systems, I contend they may not. Syntatic descriptions must be adequately constrained, and the most plausible such constraints appeal to a system's teleological function or design and hence to its history. As a result, physical twins may not realize the same syntactic states.