Learning from performance errors

A theory of how people detect and correct their own performance errors during skill practice is proposed. The basic principles of the theory are that errors are caused by overly general knowledge structures, that error detection requires domain-specific declarative knowledge, that errors are experienced as conflicts between what the learner believes ought to be true and what he or she perceives to be the case, and that errors are corrected by specializing faulty knowledge structures so that they become active only in situations in which they are appropriate. A computer simulation model that embodies the theory learns cognitive skills in ecologically valid domains. The theory generates novel and testable predictions about error correction. It is also consistent with learning phenomena that are seemingly unrelated to errors, including transfer of training and the learning curve.