Abstract

Three theories of cognitive representation are described, with emphasis on their implications for issues related to motor skill acquisition. Production system models make a distinction between declarative and procedural knowledge, and skilled performace is assumed to be based on procedural knowledge that is not ordinarily verbalizable or available to consciousness. Neural network models rely on error detection and correction, in a manner reminiscent of closed-loop theory, to develop a distributed representation of knowledge that captures relationships between task components. Instance theories of skill acquisition are founded on the assumption that expert performance derives from automatic retrieval of memory for individual training episodes. The instance memory approach contrasts with the schema theory of skill learning. In general, the cognitive systems described here constitute forms of representation that typically are not open to modification by intentional processes such as mental practice. They are constructed and influenced instead by direct experience with a task. Verf.-Referat