Abstract:
A model for motivational learning of spatial behavior is presented. It postulates a spatial representation in the form of a directed graph with nodes containing drive-related information and edges containing sensorimotor features. This representation is called the World Graph (WG). The model specifies the general drive dynamics, the ways in which WG is updated, and the ways in which the organism moves about in the world. Two hypotheses on the interaction of choice of nodes and drive levels are discussed. Two hypotheses on the way in which weights are assigned to the different nodes in the WG are developed. The model is contrained by experiments by Brown (1942, 1948), Gough (1962), Hull (1933), Leeper (1935), and Strain (1953).