Abstract

A key role ascribed to midbrain dopamine (DA) neurons rests with learning about rewarding events by reflecting reward prediction errors (RPE). Research has shown that during reward learning a positive prediction error (e.g. surprising reward), leads to phasic excitation, while a negative prediction error (e.g. omission of an expected reward) leads to phasic inhibition in DA neurons. It remains unclear, however, how DA regulates learning about aversive events. Using behavioral electrophysiology we recorded from DA neurons in the ventral tegmental area (VTA) during a Pavlovian task in which auditory cues were trained as predictors of either an appetitive sucrose reward or aversive footshock. Our analyses confirmed a role for VTA DA neurons in tracking reward prediction error (RPE), that is, elevation in firing rate (FR) to the reward predictor and depression in FR at time of reward omission in a correlated fashion. Further, our goal was to determine whether DA firing would represent reward and aversion in line with a valence-based prediction error signal. We found that cue related phasic DA activity to both reward and aversion predicting cues contained both information about stimulus identity, as well as valence. Additionally, outcome omission was represented as state of opposite valence. These results support the hypothesis that midbrain DA neurons support learning by signaling valence-based prediction errors.