The Synaptic Theory of Memory: A Historical Survey and Reconciliation of Recent Opposition

Abstract

Trettenbrein (2016) has argued that the concept of the synapse as the locus of memory is outdated and has made six critiques of this concept. In this article, we examine these six critiques and suggest that the current theories of the neurobiology of memory and the empirical data indicate that synaptic activation is the first step in a chain of cellular and biochemical events that lead to memories formed in cell assemblies and neural networks that rely on synaptic modification for their formation. These neural networks and their modified synaptic connections can account for the cognitive basis of learning and memory and for memory deterioration in neurological disorders. We first discuss Hebb’s (1949) theory that synaptic change and the formation of cell assemblies and phase sequences can link neurophysiology to cognitive processes. We then examine each of Trettenbrein’s (2016) critiques of the synaptic theory in light of Hebb’s theories and recent empirical data. We examine the biochemical basis of memory formation and the necessity of synaptic modification to form the neural networks underlying learning and memory. We then examine the use of Hebb’s theories of synaptic change and cell assemblies for integrating neurophysiological and cognitive conceptions of learning and memory. We conclude with an examination of the applications of the Hebb synapse and cell assembly theories to the study of the neuroscience of learning and memory, the development of computational models of memory and the construction of “intelligent” robots. We conclude that the synaptic theory of memory has not met its demise, but is essential to our understanding of the neural basis of memory, which has two components: synaptic plasticity and intrinsic plasticity.

Is plasticity of synapses the mechanism of long-term memory storage?

“…Finally, it could well be that synaptic-specific plasticity and cell-wide intrinsic plasticity mechanisms both play critical roles in memory storage and maintenance. 127 The extent to which one predominates may depend on the nature of the memory and the circuit that is responsible for it. Sensorimotor reflex learning, as well as non-associative forms of learning, including habituation, sensitization, and, possibly, classical conditioning, may work perfectly, perhaps optimally, through cell-intrinsic mechanisms.…”

Systematic Analysis of Pigeons’ Discrimination of Pixelated Stimuli: A Hierarchical Pattern Recognition System Is Not Identifiable

“…Rather than employing a non-neural error backpropagation scheme one would want to employ a reinforcement-based evaluation loop widespreadly providing feedback to the connecting synaptomimes on how good or bad the outcome of an action was, thus emulating the action of the ‘self-stimulation’ (= reinforcing) system 88 . This may create a connectivity architecture, that is, a connectome that could resemble the functioning and performance of the real brain of pigeons faced with the present pixel pattern recognition task 89 . However, these studies show that the neural network approach demands considerable special computational-methodological knowledge and skills so that we must necessarily leave them to others.…”

Donald O. Hebb and the Organization of Behavior: 17 years in the writing

“…Much has been written about the Hebb synapse [32,105,109,137,144,166,168,174,176]. The idea of the 'Hebb synapse' was not new to Hebb but had been thought of by many others (see [137]).…”