Stable memory with unstable synapses
Lee Susman, Naama Brenner, and Omri Barak
Received Date: 1st August 2018
What is the physiological basis of long term memory? The current dogma in theoretical neuroscience attributes changes in synaptic efficacy to memory acquisition. This view implies that in the absence of learning, synaptic efficacies should be constant; in other words, stable memories correspond to stable connectivity patterns. However, an increasing body of experimental evidence points to significant, activity-independent dynamics in synaptic strengths. These fluctuations seem to be occurring continuously, without specific reference to a learning process, but with similar magnitude. Motivated by these observations, we explore the possibility of memory storage within a specific component of network connectivity, while individual connections fluctuate in time. We find a family of neural network models in which memory is acquired by means of biologically-inspired learning rules; the properties of these rules, in turn, determine the storing component in which memory can persist indefinitely. Memory representations manifest as time-varying attractors in neural state-space and support associative retrieval of learned information. Our results suggest a link between the properties of learning rules and those of network-level memory representations, which generates experimentally-testable predictions.
Read in full at arXiv.
This is an abstract of a preprint hosted on an independent third party site. It has not been peer reviewed but is currently under consideration at Nature Communications.