Biological plausible goal-directed navigation model based on direct reinforcement learning algorithm

Abstract

For the problem of the goal-directed navigation in continuous state and action space, a spiking neural network model from hippocampal place cells to the putative action cells in prefrontal cortex is proposed based on the firing characteristics of place cells and the information cycles in the hippocampus. The continuous state and action space are respectively characterized by a population of place cells and action cells, and the direct reinforcement learning algorithm combined with the spike response model has been used to goal-directed autonomous navigation. The simulation results in Morris watermaze task show that the algorithm used in the model can solve the problem of the goal-directed navigation in continuous state and action space, and obtain a better performance when compared to the classic methods based on temporal difference at the same given problem. When the number of the actions cell had been changed, the convergence of model remains the same. The model can still achieve the goal location, when the scale of the watermaze and the goal location had been changed.