Learning-induced sequence reactivation during sharp-wave ripples: a computational study

Abstract

During sleep, memories formed during the day are consolidated in a dialogue between cortex and hippocampus. The reactivation of specific neural activity patterns - replay - during sleep has been observed in both structures and is hypothesized to represent a neuronal substrate of consolidation. In the hippocampus, replay happens during sharp wave - ripples (SWR), short bouts of excitatory activity in area CA3 which induce high frequency oscillations in area CA1. In particular, recordings of hippocampal cells which spike at a specific location ( place cells) show that recently learned trajectories are reactivated during SWR in the following sleep SWR. Despite the importance of sleep replay, its underlying neural mechanisms are still poorly understood.\n\nWe developed a model of SWR activity, to study the effects of learning-induced synaptic changes on spontaneous sequence reactivation during SWR. The model implemented a paradigm including three epochs: Pre-sleep, learning and Post-sleep activity. We first tested the effects of learning on the hippocampal network activity through changes in a minimal number of synapses connecting selected pyramidal cells. We then introduced an explicit trajectory-learning task to the model, to obtain behavior-induced synaptic changes. The model revealed that the recently learned trajectory reactivates during sleep more often than other trajectories in the training field. The study predicts that the gain of reactivation rate during sleep following vs sleep preceding learning for a trained sequence of pyramidal cells depends on Pre-sleep activation of the same sequence, and on the amount of trajectory repetitions included in the training phase.