Presentation Encoding expected reward value for formulating goal-directed decision in the rostro-medial caudate and the ventral pallidum.

藤本, 淳  ,  堀, 由紀子  ,  永井, 裕司  ,  菊池, 瑛理佳  ,  W McCairn, Kevin  ,  平林, 敏行  ,  Takada, Masahiko  ,  須原, 哲也  ,  南本, 敬史

2017-11-13
Description
The limbic basal ganglia networks are conjectured to mediate neural activity underlying reward prediction and goal-directed behavior. Having recently demonstrated that the rostromedial caudate (rmCD) is essential for normal goal-directed decision based on reward size (Nagai et al., 2016), the ventral pallidum (VP)—an area that is reciprocally connected with the rmCD—has also been implicated in reward-based neural processing. How rmCD and VP interact, however, and formulate value-based decisions remains unclear. To address this, we recorded neuronal activity from the rmCD and the VP from rhesus monkeys (N=2), while they performed the ‘reward-size’ task (Minamimoto et al., 2009). In this task, the monkeys performed an instrumental action (lever release < 1 sec after a ‘go’ signal) following presentation of a cue associated with the reward size (1, 2, 4, or 8 drops of juice). Both rmCD (39/107) and VP (63/105) transiently encoded the expected reward size following a cue presentation. Neuronal latency for the value coding in the rmCD neurons was significantly longer than the VP neurons (rmCD: 235 ms, VP: 115 ms, P < 0.01), suggesting faster recruitment of the VP rather than the rmCD in the value expectation process. To examine the causal role of these two regions, we inactivated bilateral VP by local infusion of GABAA receptor agonist (muscimol) in one monkey, and compared the behavioral effects with those from rmCD inactivation (N=2, Nagai et al., 2016). Silencing in either structure impaired the normal relationship between error rate and reward size, while the VP inactivation resulted in higher error rates than the rmCD inactivation. These results imply that both the rmCD and the VP encode an expected reward signal, which is critical for formulating decisions in goal-directed behavior. In conclusion, the VP encodes reward and goal-directed behavior in a complex, reciprocating manner, with rmCD.
Society for Neuroscience 2017 Annual Meeting

Number of accesses :  

Other information