Abstract
During non-REM sleep, memory consolidation is driven by a dialogue between cortex and hippocampus. The reactivation of specific neural activity sequences – replay – is believed to represent a neuronal substrate of consolidation. In the hippocampus, replay occurs during sharp-wave ripples (SWRs), short bouts of excitatory activity in area CA3 which induce high frequency oscillations in the inhibitory population of area CA1. Despite growing evidence for the functional importance of replay, its neural mechanisms remain poorly understood. Here, we develop a novel theoretical model of hippocampal spike sequence replay during SWRs. In our model, noise-induced activation of CA3 pyramidal cells triggered an excitatory cascade, which induced local ripple events in CA1. Ripples occurred stochastically in the model, with Schaffer Collaterals driving coordination, so that localized sharp waves in CA3 produced consistently localized CA1 ripples. In agreement with experimental data, the majority of pyramidal cells in the model showed low reactivation probability across SWRs, defined by the overall network connectivity. We found, however, that a small portion of pyramidal cells which had high reactivation probability across multiple SWRs owed their reactivation properties to the fine variations within network connectivity, and hence the detailed spiking dynamic within SWRs. In particular, the excitatory inputs along synaptic pathway(s) to cells and cell pairs controlled emergent single cell and cell pair reactivation. Furthermore, we found that inhibitory synaptic inputs and intrinsic cell excitability only had an influential role on the activation of CA3 pyramidal cells, but not CA1 pyramidal cells, during SWRs. Our study predicts that hippocampal replay results from a network-wide coordination of activation probability across SWRs for cells and cell pairs, which is further refined by specific synaptic strengths. This suggests a possible competition among cell assemblies for activation during SWRs, where synaptic strengths mediate the chance of dominance of a given memory over others during spontaneous SWRs.
Author Summary During sleep, rhythmic activities in different brain regions are coordinated across multiple timescales and brain regions. The coordination of these events is important for consolidation of recently acquired memories. Sharp-wave ripples (SWRs) are one of such major sleep rhythms, seen in the hippocampal region, during which cells previously active during an awake task reactivate, in preserved order, during sleep (‘replay’). Replay is thought to contribute to consolidation by enabling re-elaboration of events of the day during sleep. However, the manner in which specific spiking patterns are selected for replay remains unknown. In this study, we apply computational models to reveal mechanisms behind the generation of SWRs and to explain the factors controlling which cell sequences reactivate during SWRs. We find that different hippocampal regions have different factors that promote replay. Our study predicts that when learning changes the strength of synaptic connections during wake, it would enhance the probability of reactivation of experience-specific groups of neurons during sleep.
Footnotes
Conflict of Interest: The authors declare no competing financial interests.