The contention that psychological states are multiply realizable in different substrates has been used to support the contention that neuroscience is not likely to be very useful in guiding an understanding how cognition works. But in the context of scientific research, how seriously should we really take this threat of multiple realizability? By examining how brain areas are identified in neuroscience (where the approach is comparative and employs functional criteria), we show that the skepticism about neuroscience's role in understanding cognition is misguided and that the apparent success of multiple realizability is based on methodological error. With respect to the former point, we develop and analyze a representative case in which the organization of the brain provides crucial information about the organization of the cognitive system. Then we show that the initial plausibility of claims to multiple realizability rest on (a) mismatching a broad-grained criterion (to show sameness of psychological states) with a finegrained criterion (to differentiate brain states), and (b) a failure to attend to the purposes for which taxonomies of brain and psychological states are developed.