Maps, routes, and the hippocampus: a neural network approach.

Abstract

This study describes hippocampal participation in maze navigation in terms of a real-time, biologically plausible neural network. The system is composed of (1) a goal-seeking mechanism, (2) a cognitive map system, and (3) a route system. The goal-seeking mechanism displays exploratory behavior until either the goal is found or a sufficiently strong prediction of the goal is generated. The cognitive map is a topological map that stores associations between places and views of accessible places, and between places and reward. The route system establishes associations between cues and reward. Both systems compete with each other to establish associations with the reward, with the cognitive system generally overshadowing the route system. In agreement with previous models, it is assumed that the hippocampus modulates the storage of cognitive maps in cortical areas and mediates the competition between cognitive maps and route systems. After hippocampal lesions, animals navigate through mazes making use of the route system. Computer simulations show that the network effectively describes latent learning, detour behavior, and place learning in normal and hippocampal- and cortical-lesioned animals.