ABSTRACT
A competitive neural network model for the process of recurrent choice is described and a possible explanation for the associated adaptive effects of reinforcement learning is offered. We review a set of relevant experiments that characterize the framework. A minimum set of functional units and a mechanism that links them are proposed. The way in which the model reacts while exposed to various experimental conditions is carefully examined and its specific behavioral patterns are emphasized. The model shows remarkably complex behavior that resembles the general features of the behavior of pigeons and rats under several different choice-procedures.
