The Hebb synapse is a powerful idea that unifies psychology, physiology, and computation. Perhaps too powerful. To help you think outside the Hebbian box, I will explain the idea of the hedonistic synapse, which is based on the fact that synaptic transmission is unreliable: when depolarized by an action potential, a presynaptic terminal may release neurotransmitter, or it may fail to release. The hedonistic synapse responds to a reward signal by increasing its probability of release or failure, depending on which action immediately preceded reward. I will argue that the hedonistic synapse is analogous to operant conditioning in psychology, can solve the optimization problems important in computational accounts of learning, and is also physiologically plausible. The time is ripe to look for it and other forms of synaptic plasticity related to reinforcement learning.

Sebastian Seung is Associate Professor of Computational Neuroscience at MIT, and Assistant Investigator of the Howard Hughes Medical Institute. He studied theoretical physics with David Nelson at Harvard University and completed postdoctoral training with Haim Sompolinsky at the Hebrew University of Jerusalem. Before joining the MIT faculty, he was a member of the Theoretical Physics Department at Bell Laboratories. He is a Sloan Research Fellow, a Packard Fellow, and a McKnight Scholar.