Resources
Authors & Affiliations
Jordan Elum,Larry Zweifel
Abstract
The mesolimbic dopamine system is believed to regulate both reinforcement learning and motivation to obtain rewards. A central question is how the mesolimbic dopamine system influences these distinct functions. A classical neural circuit model proposes a two-component system in which dopamine release in the nucleus accumbens core encodes prediction error signals to regulate reinforcement learning while dopamine release in the nucleus accumbens shell signals incentive salience and promotes motivated responses. Here we record bulk calcium dynamics in projection-specific ventral tegmental area (VTA) dopamine populations during instrumental conditioning, as mice undergo acquisition, extinction, and reinstatement of a food-reinforced operant task. We show that both accumbens core- and shell-projecting dopamine populations are activated by actions, cues, and rewards. However, we find differential activity dynamics in action, cue, and reward encoding between projection-specific dopamine populations. Accumbens shell-projecting VTA dopamine neurons preferentially encode animals’ action initiation (lever press) and display a sustained increase in activity during the cue and reward outcome periods. In contrast, these signals in the accumbens core-projecting population return to baseline between discrete events. Further, during unexpected reward omission, the accumbens core-projecting population displays temporally discrete decreases in calcium signals consistent with a prediction error-encoding model that are not observed in the shell-projecting dopamine cells. By contrast, during unexpected reward delivery, the accumbens shell-projecting population preferentially encodes high saliency reward outcomes. By optogenetically manipulating both the accumbens shell-projecting and core-projecting VTA dopamine populations we tested and confirmed the prediction that these populations differentially regulate motivation during cued reinstatement. These findings suggest that accumbens core-projecting dopamine neurons provide prediction error signals to facilitate reinforcement learning while accumbens shell-projecting dopamine neurons provide incentive salience signals to promote motivated behavioral responses. These results also provide evidence for a two-component neural circuit model of mesolimbic dopamine’s dual functions in reinforcement learning and motivation.