The robot basal ganglia: Action selection by an embedded model of the basal ganglia

Abstract

Action selection is the task of resolving conflicts between multiple sensorimotor systems seeking access to the final common motor path. Recently,1,2 we proposed that the basal ganglia may act to provide a biological solution to the problem of selection. To test this notion we have implemented a high level computational model of intrinsic basal ganglia circuitry and its interactions with simulated thalamocortical connections.3,4 The computational model was then exposed to the rigors of `real world’ action selection by embedding it within the control architecture of a small mobile robot.5 In a mock foraging task, the robot was required to select appropriate actions under changing sensory and motivational conditions, thereby generating sequences of integrated behavior. Our results demonstrate: (i) the computational model of basal ganglia switches effectively between competing channels depending on the dynamics of relative input ‘salience’; (ii) its performance is enhanced by inclusion of anatomically inspired thalamocortical circuitry; (iii) in the robot, the model demonstrates appropriate and clean switching between different actions and is able to generate coherent sequences of behavior.