Abstract

The brain contains multiple yet distinct systems involved in reward prediction. To understand the nature of these processes, we recorded single-unit activity from the lateral prefrontal cortex (LPFC) and the striatum of monkeys performing a reward inference task using an asymmetric reward schedule. Prefrontal neurons inferred the reward value (amount) of a stimulus, even when the monkeys had not yet learned the stimulusreward association directly. Striatal neurons, however, predicted the reward for the stimulus only after directly experiencing the stimulus-reward contingency. Our results suggest dissociable functions in their reward prediction, i.e., that the LPFC utilizes stimulus categorization in a generative process of reward inference, whereas the striatum applies direct experiences of stimulus-reward associations in the guidance of behavior. However, while the LPFC and striatum have mutual connections through direct and indirect pathways, it is not clear that how the two structures are interactive to make a reward prediction. To investigate this issue, we simultaneously recorded local field potential (LFP) from the both areas during monkey’s performing the reward inference task and performed the Granger-causality analysis with the LFP data. Information flow from LPFC to striatum was larger than that from striatum to LPFC , particularly in correct trials, in trials with new stimulus and in small reward trials.