The attempt to derive mental function from brain structure is highly constrained by study of the allometric changes among brain components with evolution. In particular, even if homologous structures in different species produce similar computations, they may be constituents of larger systems (e.g., cortical-subcortical loops) that exhibit different composite operations as a function of relative size and connectivity in different-sized brains. The resulting evolutionary constraints set useful and specific conditions on candidate hypotheses of brain circuit computation.