The Once and Future Hebb Synapse

Abstract The Hebb synapse has become better known than Donald Hebb himself. In this respect he has joined an exclusive club along with the Ising model in condensed matter physics and Parkinson's disease in medicine. This is not to say that Hebb has not made other important contributions, as Peter Milner and Bryan Kolb document, but the Hebb synapse has eclipsed these other achievements. The goal of this essay is to examine how this happened. The Hebb synapse remains a vital organizing concept for both experimental studies and theoretical analysis, as Geoffrey Hinton emphasizes. I am sometimes asked to identify important advances made by theory and computational modeling in neuroscience. Most would agree that the achievement of Hodgkin and Huxley in modeling the action potential was of seminal importance. Not only did they provide a mechanistic explanation of the action potential that has withstood the test of time, they also outlined a research strategy for explaining even more complex internal properties of neurons that has served us well over the last 50 years (Destexhe & Sejnowski, 2001). It is more difficult to find success stories at the systems level, but Donald Hebb would qualify in my mind. Not only did he make a prediction about the conditions for synaptic plasticity that was subsequently confirmed, he also outlined a framework for building links between neurophysiology and psychology that has become a major research program. I wrote a review of Hebb's 1949 book on the occasion of the 50th anniversary (Sejnowski, 1999) and this essay gives me the opportunity to put recent discoveries in synaptic plasticity into the context of Hebb's research program. The central problem that Hebb posed in The Organization of Behavior: A Neuropsychological Theory was the origin of autonomous activity in the cerebral cortex: ... we know practically nothing about what goes on between the arrival of the excitation at a sensory projection area and its later departure from the motor area of the cortex. (p. xvi) Hebb conjectured that cortical circuits admit selfsustaining activity that reverberated in "cell assemblies," inspired by anatomical evidence for recurrent connections between neighbouring cells in the cerebral cortex and reverberatory activity lasting for up to half a second. Hebb further suggested that activity in one cortical circuit could, through converging projections, activate other areas of cortex and lead to a sequence of activations or "phase sequence." Hebb needed a way to sustain persistent reverberatory activity or "trace" in cortical circuits. He proposed that patterns of connections between neurons could sustain reverberatory activity if their strengths could be adjusted by an activity-dependent mechanism for synaptic plasticity that he called a "Neurophysiological Postulate": When an axon of cell A is near enough to excite cell B and repeatedly or persistently takes part in firing it, some growth process or metabolic change takes place in one or both cells such that A's efficiency, as one of the cells firing B, is increased. (p. 62) These words have been interpreted to mean that the conditions for synaptic plasticity should depend on coincidence detection; that is, strengthening of the synapse should occur when the release of neurotransmitter molecules from a presynaptic terminal coincides with the depolarization of the postsynaptic cell. Instead of being used to develop models of sustained activity in recurrent networks, the first theoretical applications of Hebb's postulate were to models of distributed associative memory in feedforward network models (Hinton & Anderson, 1981). Hebb's postulate admits an alternative, deeper interpretation, which has unfolded over the last five years in a surprising and satisfying way. HEBB SYNAPSES IN THE BRAIN Biological evidence for the Hebb rule had to wait for neurobiology to discover conditions that elicited longterm changes in synaptic strength and could be reliably studied at the cellular level. …