Biology session

Reward prediction error (RPE) is the difference between actual and expected reward. According to reinforcement learning theory, value is updated based on RPE. The release of dopamine (DA) in the ventral striatum (VS) was shown to represent RPE during classical condition. However, DA release in the prefrontal cortex (PFC), an area that is crucial to value-based decision making, was much less studied. Basal forebrain cholinergic neurons have already been shown to play an important role in associative learning, suggesting that they may be involved in the processing of stimuli that predict future outcomes. It has been previously demonstrated that the release of norepinephrine (NE) follows the threat prediction error, but its relationship with RPE has not been investigated yet. We aimed to investigate the role of DA, ACh and NE in associative learning and the correlation of their release with the RPE.

To address this, we trained mice (n=15) on a sound detection Pavlovian conditioning task with a 50% reward schedule that allowed examining clean representations of positive (rewarded trials) and negative (reward omission trials) RPE, while we measured DA, ACh and NE release by fiber photometry, simultaneously measuring DA release in both PFC and VS, and ACh/NE release in both PFC and basolateral amygdala (BLA). Behavioral updating of value representations based on the outcome of the previous trial was indexed by the licking activity of mice in the anticipation of reward.

As expected, anticipatory licking during the stimulus decreased after omitted rewards but increased after rewarded trials. DA release followed a similar pattern not only in the VS but also in the PFC. Moreover, we found significant positive correlations between DA release and anticipatory lick rate difference both in VS and PFC. In the case of ACh, we did not experience anything similar in any of the brain areas. No clear correlation was observed between the change in ACh release in any of the brain areas and the change in the anticipatory lick rate. NE release in the BLA followed a similar pattern to DA, but in the PFC, it decreased after rewarded trials and increased after omitted rewards. Furthermore, we observed a negative correlation in the PFC and a mostly positive correlation in the BLA between NE release and anticipatory lick rate difference.

Thus, these results indicate that the dopaminergic system broadcasts similar RPE signals to both striatal and frontal cortical targets. However, the cholinergic system does not follow the changes of the RPE. Contrary to expectations, the noradrenergic system showed opposite activity to the RPE towards the PFC.