This paper has attempted to outline the basis for a new theory of learning. One important feature of the theory has to do with those respects in which it is not new--it has incorporated some ideas contained in presently competing theories into a formulation that offers the possibility of unifying these rivals. In addition, new suggestions concerning the analysis of performance, the interpretation of punishment and the role of information feedback in learning have been offered. Data from three experiments described in the paper have both supported predictions based on the new theory and have indicated difficulties in previous interpretations of learning. These three experiments also demonstrated that the new theory has the desirable feature of lending itself readily to experimentation on human subjects.