In this paper, psychology is considered as a biological science within the context of the information sciences. Connectionist arguments about the computational architecture of the brain based on biological and/or computational plausibility are rejected. Following Hebb (1958), it is argued that analyses based on biological and/or computational plausibility may serve to tune a psychological model, but that behavioural accuracy must be the main arbiter of any psychological model. © 1990 Springer-Verlag.