Tonically active neurons in the monkey dorsal striatum signal outcome feedback during trial-and-error search behavior

An animal’s choice behavior is shaped by the outcome feedback from selected actions in a trial-and-error approach. Tonically active neurons (TANs), presumed cholinergic interneurons in the striatum, are thought to be involved in the learning and performance of reward-directed behaviors, but it remains unclear how TANs are involved in shaping reward-directed choice behaviors based on the outcome feedback. To this end, we recorded activity of TANs from the dorsal striatum of two macaque monkeys ( Macaca fuscata ; 1 male, 1 female) while they performed a multi-step choice task to obtain multiple rewards. In this task, the monkeys first searched for a rewarding target from among three alternatives in a trial-and-error manner and then earned additional rewards by repeatedly choosing the rewarded target. We found that a considerable proportion of TANs selectively responded to either the reward or the no-reward outcome feedback during the trial-and-error search, but these feedback responses were not observed during repeat trials. Moreover, the feedback responses of TANs were similarly observed in any search trials, without distinctions regarding the predicted probability of rewards and the location of chosen targets. Unambiguously, TANs detected reward and no-reward feedback specifically when the monkeys performed trial-and-error searches, in which the monkeys were learning the value of the targets and adjusting their subsequent choice behavior based on the reward and no-reward feedback. These results suggest that striatal cholinergic interneurons signal outcome feedback specifically during search behavior, in circumstances where the choice outcomes cannot be predicted with certainty by the animals. Highlights TANs signal reward and no-reward outcome feedback when monkeys made search behaviors TANs respond regardless of predicted reward probability or chosen target location TANs may signal feedback outcomes when rewards cannot be predicted with certainty