The changing relations of representation and cognitive structure during the development of a cognitive map

In recent years, workers in cognitive science have come to recognize that cognitive structures should not be equated with computational ones. This realization has often been experienced as confusing. It is argued that cognitive science is moving closer to positions defended by genetic epistemologists. The course of development of a cognitive structure consists of two phases. During the first (Piaget's "phenotypic" adaptation), knowledge about successful ways of interacting with the environment is encoded in a relatively unspecialized format. Selected structural aspects of the resulting data base may later form the basis for a specifically appropriate format. It is during the first phase that knowledge about the environment is manifested in a behavioural structure which does not correspond to any computational entity. In the second phase, the internalized structure becomes computationally defined, which improves the system's efficiency. These ideas are illustrated with the Traveller, a detailed computational model of cognitive map development.