A major goal of physiological psychology is to determine the physical basis of mental representation. Animal models are essential to this project. Dretske's influential analysis of the concept of mental representation suggests that operant and classical conditioning involve mental representation. This analysis comports well with known physiological mechanisms of conditioning, but fails to capture necessary features of mental representation at the human level. We conclude that the applicability of animal models to the problem to human mental representation is more restricted than previously thought.