Department of Psychology, University of Arizona, USA
Wednesday 4 June 2008
Seminar Room B10 (Basement)
Alexandra House, 17 Queen Square, London, WC1N 3AR
Dopaminergic function in the basal ganglia is thought to support reinforcement learning to support behaviors that maximize rewards. The majority of models of this system focus on the reward prediction error "signal" conveyed by dopamine neurons. I present a detailed neural network model of the basal ganglia that makes predictions about how these prediction error signals drive learning and action selection via distinct postsynaptic mechanisms. Model predictions are tested with a series of novel experiments using patient, pharmacological, and genetic data. Abstract reinforcement learning models are used to derive best fitting model parameters corresponding to individual subject learning, and which are predictably modulated by dopaminergic genotypes in a manner that is consistent with the more complex but physiologically constrained model.