Memory Consolidation from Seconds to Weeks Through Autonomous Reinstatement Dynamics in a Three-Stage Neural Network Model

Abstract

Long-term memories for facts and events are not created at an instant. Rather, memories stabilize gradually over time and involve various brain regions. The shifting dependence of acquired declarative memories on different brain regions – called systems consolidation – can be tracked in time by lesion experiments and has led to the development of the Complementary Learning Systems framework, which focuses on hippocampal-cortical interaction. Observations of temporally graded retrograde amnesia following hippocampal lesions, point to a gradual transfer from hippocampus to cortical long-term memory. Spontaneous reactivations of hippocampal memories, as observed in place cell reactivations during slow-wave-sleep, are supposed to drive cortical reinstatements and facilitate this process. We propose a functional neural network implementation of these ideas and furthermore suggest an extended three-stage framework that also includes the prefrontal cortex and bridges the temporal chasm between working memory percepts on the scale of seconds and consolidated long-term memory on the scale of weeks or months. We show that our three-stage model can autonomously produce the necessary stochastic reactivation dynamics for successful episodic memory consolidation. The resulting learning system is shown to exhibit classical memory effects seen in experimental studies, such as retrograde and anterograde amnesia after simulated hippocampal lesioning.