Chapter 34 - Phasic Dopamine Signaling in Action Selection and Reinforcement Learning

Abstract

The present chapter begins by reviewing some of the influential ideas on basal ganglia involvement in action selection and reinforcement learning. Within this context, we assess the specific contribution of the phasic response of dopamine (DA) neurons to the fundamental processes of reinforcement learning. Since the turn of the century, the dominant hypothesis has been that sensory-evoked DA responses signal reward prediction errors—events that are better or worse than expected. The reinforcing effects of these signals are to ensure that future reward acquisition is maximized. One of the few arguments against this view was that phasic DA signals are unlikely to signal a generally useful reward prediction error because their short latency ( before the identity of the event has been established? However, recent evidence that cortical visual processing can also elicit phasic DA responses means the argument that short latency of phasic DA signaling is too short to signal reward prediction errors is no long applicable. This important new perspective is incorporated into our previously articulated hypothesis that phasic DA signaling would be useful for the determination of agency and the discovery of novel actions.