Dopamine Dynamics in Reward-based Learning

Dopamine signals are thought to be important for reward-based learning and appear to play important roles in regulating synaptic plasticity in the nucleus accumbens and the striatum. However, the precise neuronal circuit mechanisms underlying the learning of even the simplest goal-directed sensorimotor transformations remain to be precisely defined. Here, I first characterized the animal behavior traits, then I measured dopamine signals with fiber photometry using dLight expressed in the nucleus accumbens of head-restrained mice across reward-based sensorimotor task learning. Thirsty mice were first trained in a free-licking task, during which the mouse learned to lick a spout, for which they were sometimes rewarded with water delivery. In free-licking sessions, reward triggered a positive dopamine signal, while unrewarded licks evoked a negative dopamine signal. The amplitude of the reward-triggered dopamine response decreased across most individual sessions, likely reflecting the gradual reduction in thirst across each session with accumulated reward.Subsequently, the same mice were trained over days in a whisker-detection task, in which mice learned to lick the reward spout in response to a single brief whisker deflection. Reward delivery appeared to evoke a consistent dLight response across learning days. Whisker detection task learning was accompanied by an increase in a fast sensory-evoked dopamine signal, consistent with a large body of literature indicating dopaminergic reward prediction error signals. Muscimol inactivation experiments showed that nucleus accumbens is involved in the execution of the whisker detection task. Post-hoc analysis showed that, the time spent in the open arm of elevated plus maze correlated with the future Day 1 learning performance in the whisker detection task.Our results illustrated how the dopamine dynamics evolve in reward-based learning, and further suggest that inter-individual differences in behavioral traits may be a predictor of future learning performance.