A model of autonomous interactions between hippocampus and neocortex driving sleep-dependent memory consolidation

Abstract

How do we build up our knowledge of the world over time? Many theories of memory formation and consolidation have posited that the hippocampus stores new information, then “teaches” this information to the neocortex over time, especially during sleep. But it is unclear, mechanistically, how this actually works—How are these systems able to interact during periods with virtually no environmental input to accomplish useful learning and shifts in representation? We provide a framework for thinking about this question, with neural network model simulations serving as demonstrations. The model is composed of hippocampus and neocortical areas, which replay memories and interact with one another completely autonomously during simulated sleep. Oscillations are leveraged to support error-driven learning that leads to useful changes in memory representation and behavior. The model has a non–rapid eye movement (NREM) sleep stage, where dynamics between the hippocampus and neocortex are tightly coupled, with the hippocampus helping neocortex to reinstate high-fidelity versions of new attractors, and a REM sleep stage, where neocortex is able to more freely explore existing attractors. We find that alternating between NREM and REM sleep stages, which alternately focuses the model’s replay on recent and remote information, facilitates graceful continual learning. We thus provide an account of how the hippocampus and neocortex can interact without any external input during sleep to drive useful new cortical learning and to protect old knowledge as new information is integrated.

Aversive experience drives offline ensemble reactivation to link memories across days

“…memory consolidation, recent studies have suggested that offline periods following learning may be important for memory integration processes as well [22][23][24][25] .…”

A complementary learning systems model of how sleep moderates retrieval practice effects

“…Other studies suggest that sleep selectively facilitates weakly learned information (e.g., Schapiro, McDevitt, Rogers, Mednick, & Norman, 2018). These findings suggest that SWS does not simply stabilize memories, and instead, it might play a more complex role, reorganizing representations of past experience (Lewis & Durrant, 2011;Singh, Norman, & Schapiro, 2022).…”

“…It is noteworthy that other mechanisms may also potentially account for the integration of related information during sleep. For example, (Singh et al, 2022) simulated the alternation between SWS, during which the hippocampus trained the neocortex with newly acquired memory, and REM sleep, during which the neocortex primarily replayed the consolidated remote memory without influence from the hippocampus. Therefore, it is possible that the replay of semantic connections during REM sleep directly overwrites the impairment of non-targets caused by retrieval practice and leads to facilitation.…”

Memory reactivations during sleep: a neural basis of dream experiences?

“…Memory consolidation is a process that unfolds over timehowever, due to methodological restrictions, our understanding of the time course of memory reactivations within a single night and especially across multiple nights is still lacking. Recently proposed frameworks suggest that within a single night, the combination of high-fidelity replay of recent memories in early NREM sleep with lower-fidelity replay, and integration with more remote memories, in later REM sleep optimizes memory consolidation over time (e.g., [153,154]). However, experimental data from rodent and human research to verify these models are still lacking.…”