Psychological researchers should deal with the concrete stimulus-response principles of learning on which behavior is based, and study behaviors that are representative of real life behaviors. The present research strategy has come from two faulty ideas: first, a concern with underlying, inferred mental processes, rather than with actual tasks or behaviors; and second, a belief that behavior and problem solving can be functionally divided into perception and performance. These ideas lead (1) to the use of tasks that are not representative of the universe of behaviors of everyday living (e.g. the anagram problem, the water jar problem), (2) to a lumping together of behaviors that are different in nature (e.g. classical conditioning of a cat to "no!" and instrumental conditioning of a cat to his name), and (3) to the separation of behaviors that are similar in nature (e.g. learning to read letters and learning to write letters). Only by a shift of emphasis to concrete functional analysis can the artificial perception-performance dichotomy be discarded and a great deal about behavior and acquisition of skills be discovered. (MH)