Reinforcement Learning: Back to the Brain

Artificial reinforcement learning (RL) initially drew inspiration from psychological theories about learning in animals, before developing into a rich field with its own concepts and results. Conversely, in recent years these RL concepts have come to play an important role in our current understanding of how specific brain circuits guide adaptive decision-making. In this talk I will first summarize several examples of the use of RL in contemporary neuroscience, before describing some ongoing, related projects in my laboratory. I aim to spark broad discussion of how RL in computer science and RL in neuroscience can productively cross-fertilize.
Josh Berke is Associate Professor of Psychology at UM. He studied Natural Sciences at the University of Cambridge, UK, before performing his PhD in Neurobiology at Harvard investigating the molecular biology of dopamine signaling. His postdoctoral training in behavioral neurophysiology was with Howard Eichenbaum at Boston University, comparing information processing in distinct hippocampal and striatal memory systems. Since 2004 his laboratory at UM has primarily focused on how neural circuitry at different spatial scales within the basal ganglia contribute to learning and behavioral control.