Preference encoding in ventral pallidum mediates reward-seeking behavior
Ottenheimer, David Joshua
MetadataShow full item record
An essential function of the nervous system is to direct reward-seeking behavior in order to maximize the acquisition of preferred rewards. This process requires a method for evaluating all available outcomes on a common scale and using these valuations to organize the appropriate behavioral response. The ventral pallidum (VP) is a key node in a basal ganglia circuit hypothesized to convert limbic information, like reward values, into reward-seeking actions. Previous work has linked VP neural activity to the availability and palatability of rewards, and VP has been functionally implicated in the motivation to pursue rewards. Open questions include how VP encodes the values of multiple available rewards and whether its activity contributes to preference-driven behaviors. For this dissertation, we conducted a series of electrophysiological and optogenetic experiments to characterize the role of VP in navigating scenarios with multiple rewarding outcomes. First, we demonstrated that, following reward delivery, the activity of a majority VP neurons reflected the value of the delivered outcome relative to the locally available options; notably, this activity preceded and outnumbered reward-specific activity in nucleus accumbens, the most frequently studied input to VP. Further analysis of VP activity revealed that, consistent with a reward prediction error signal, a subset of neurons' reward-evoked activity incorporated the outcomes from the most recent previous trials. The prediction error hypothesis was further supported by optogenetic manipulations of VP activity during this epoch, which altered rats' engagement in the reward-seeking task according to changes in their estimate of the task's value. In a final set of experiments, we linked VP neural activity to the evolution of rats' choice behavior under changing physiological conditions and demonstrated a causal role for VP outcome signals in driving behavioral preference. Our results not only establish VP as a crucial site for encoding reward preferences; they also provide insight into fundamental principles of reward signaling in the nervous system, with particular consideration for the interface between prediction errors and preference, both static and dynamic.