The neurobiology of human reinforcement learning throughout the lifespan

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): The proposed project investigates the neural mechanisms underlying human reinforcement learning. Electrophysiological recordings from primates suggest that dopamine neurons encode a reward prediction error (RPE) signal required by reinforcement learning models. There is, however, scant evidence from humans supporting this conclusion. The proposed project uses a combination of clinical and neuroimaging approaches to study the neural correlates of age-related changes in reinforcement learning. Our specific aims are as follows: 1) To determine to what extent the choice behavior of young and elderly subjects matches the predictions of animal reinforcement learning models in a task with a dynamic reinforcement schedule and to identify any age-related changes in reinforcement learning that might stem from normal age-related decline in the number of dopamine neurons. 2) To determine if reinforcement learning in elderly subjects that suffer from Parkinson's disease, an atypically severe reduction in the number of dopamine neurons, is affected by the disease and modulated by dopaminergic medication. This finding would provide causal evidence that dopamine underlies human reinforcement learning. 3) To determine if activity in dopamine areas measured by functional MRI is consistent with those areas encoding a RPE signal and to see if behavioral and neural estimates of reinforcement learning variables throughout the lifespan match consistently at all ages. We hope to find correlative and causal evidence for dopamine encoding the RPE signal in humans and hope to account for differences in reinforcement learning across the lifespan by changes in dopamine levels.