Local modulation of dopaminergic action signals during goal-directed learning

A key difference in dopamine signalling between the ventral and dorsal regions of the striatum lies in the hemispherically lateralized movement specificity of dopamine release in the dorsal striatum. This hemispherically lateralized signal cannot directly be reconciled with traditional theories that propose that dopamine transients accord with reward prediction error (RPE) signals, which reflect a scalar reward value that back-propagates to reward-predictive stimuli across learning. The predominant view is that dopamine release in the dorsal striatum signals, simultaneously and separately, a bilateral RPE and a lateralized movement signal. However, within this framework, it is unclear how movement and learning signals are disambiguated postsynaptically. We have recently proposed an alternative view, that lateralized action-related signals in the dorsomedial striatum reflect the specific action-outcome associations that underlie goal-directed action1, arguing that dopamine learning and movement signals are integrated in the dorsal striatum. Here, I will present new data suggesting that dopamine-mediated reward signals in the dorsomedial striatum emerge bilaterally, and are modulated by hemispherically separate action signals across learning, resulting in asymmetric, contralaterally dominant action-related signals, the magnitude of which accords with the rules of reward prediction error. I will also present new data that suggests that this local modulation occurs through a feedback loop involving striatal direct pathway neurons. Together, these results support a model of global RPE’s driving action-specific RPE’s within local circuits, medicated by canonical striato-nigro-striatal feedback loops.