Abstract
Recent developments in cognitive psychology, particularly the theory of semantic memory and the theory of processes involved in problem solving, seem likely to provide a basis for strong developments in the theory of learning. These will be based on rigorous representations of cognitive states that are desired as the outcomes of instruction. It seems likely that a distinction between algorithmic knowledge and more general propositional knowledge will be needed in such a theory, and that knowledge of simple procedures will be distinguished from ability to interpret information in a problem, and also from skill in generating subgoals in order to organize problem‐solving activity.