The effect of effort on reward prediction error signals in midbrain dopamine neurons

Abstract

Midbrain dopamine neurons encode reward prediction error (RPE) signals and are involved in reinforcement learning, which in turn influences various decision-making behaviors. Recently, it was revealed that dopamine neurons incorporate non-reward information in the computation of RPE signals. Here, we review the influence of non-reward information, particularly the cost of effort expenditure (the ‘effort cost’), on dopamine neuron computation of RPE signal. We discuss evidence that (a) predicted costs function to reduce overall reward prediction, and (b) incurred costs enhance dopamine neuron response to reward. This evidence and consistent behavioral results indicate that consequent to costs, such as effort expenditure, RPEs are enhanced and animals perceive greater relative reward. The integration of non-reward information in dopamine neurons may therefore partially explain irrational decision making in animals, including humans.