A Neural Model of Schemas and Memory Consolidation

Abstract

The ability to behave differently according to the situation is essential for survival in a dynamic environment. This requires past experiences to be encoded and retrieved alongside the contextual schemas in which they occurred. The complementary learning systems theory suggests that these schemas are acquired through gradual learning via the neocortex and rapid learning via the hippocampus. However, it has also been shown that new information matching a preexisting schema can bypass the gradual learning

Coordinated hippocampal-thalamic-cortical communication crucial for engram dynamics underneath systems consolidation

“…In the long history of the field, many computational models and theories of systems consolidation have been proposed 4,5,9,11,[36][37][38][39][40] . While early computational studies relied on networks with highly abstract, simplified neuron models 9,11,36,37 , recent computational models have become increasingly more complex to incorporate a wider range of experimental findings: a three-stage Bayesian Confidence Propagation Neural Network was used to bridge the gap between working and long-term memory 38 , a spiking network was developed to explore the role of anatomical properties of the cortexhippocampus loop in systems consolidation 39 , and a rate-coded multi-layer network with a form of Hebbian learning was employed to investigate the effect of preexisting knowledge on memory consolidation 40 . Nevertheless, previous models have not addressed recent findings regarding engram cells and their role in systems consolidation.…”

“…This is at least in part due to existing theoretical and computational models lagging behind the groundbreaking advancements in engram cell research enabled by new technologies developed in the past decade 31,32,34,35 . In particular, previous computational studies have employed abstract neuronal models that are intended to capture high-level properties of systems consolidation (e.g., recent memory recall relies on HPC) but are unable to reproduce engram cell-level data produced by recent experiments 9,11,25,[36][37][38][39][40] .…”

Coordinated hippocampal-thalamic-cortical communication crucial for engram dynamics underneath systems consolidation

“…In the long history of the field, many computational models of systems consolidation have been proposed [9,11,[36][37][38][39][40]. While early computational studies relied on networks with highly abstract, simplified neuron models [9,11,36,37], recent computational models have become increasingly more complex to incorporate a wider range of experimental findings: a three-stage Bayesian Confidence Propagation Neural Network was used to bridge the gap between working and long-term memory [38], a spiking network was developed to explore the role of anatomical properties of the cortex-hippocampus loop in systems consolidation [39], and a rate-coded multi-layer network with a form of Hebbian learning was employed to investigate the effect of preexisting knowledge on memory consolidation [40]. Nevertheless, previous models have not addressed recent findings regarding engram cells and their role in systems consolidation.…”

“…This is at least in part due to existing theoretical and computational models lagging behind the groundbreaking advancements in engram cell research enabled by new technologies developed in the past decade [31,32,34,35]. In particular, previous computational studies have employed abstract neuronal models that are intended to capture high-level properties of systems consolidation (e.g., recent memory recall relies on hippocampus) but are unable to reproduce engram cell-level data produced by recent experiments [9,11,25,[36][37][38][39][40].…”