Abstract:Memories are thought to be stored in neural ensembles known as engrams that are specifically reactivated during memory recall. Recent studies have found that memory engrams of two events that happened close in time tend to overlap in the hippocampus and the amygdala, and these overlaps have been shown to support memory linking. It has been hypothesised that engram overlaps arise from the mechanisms that regulate memory allocation itself, involving neural excitability, but the exact process remains unclear. Indeed, most theoretical studies focus on synaptic plasticity and little is known about the role of intrinsic plasticity, which could be mediated by neural excitability and serve as a complementary mechanism for forming memory engrams. Here, we developed a rate-based recurrent neural network that includes both synaptic plasticity and neural excitability. We obtained structural and functional overlap of memory engrams for contexts that are presented close in time, consistent with experimental and computational studies. We then investigated the role of excitability in memory allocation at the network level and unveiled competitive mechanisms driven by inhibition. This work suggests mechanisms underlying the role of intrinsic excitability in memory allocation and linking, and yields predictions regarding the formation and the overlap of memory engrams. Significance statement In the brain, memories are not formed in isolation from each other. For example, two memories of events that happened close in time tend to be linked, so that recalling one memory leads to recall of the second one. Although memories are thought to be formed by reinforcement of synapses among neurons, understanding memory linking requires us to consider intrinsic properties of the neurons themselves. In this study, we modelled a neural network aiming at explaining how memories are formed and linked in the brain. This model is able to reproduce experimental results and allows us to make predictions about how to link or dissociate memories.

Retrospective memory integration accompanies reconfiguration of neural cell assemblies

How the Brain Formulates Memory: A Spatio-Temporal Model Research Frontier.

Spiking Neuron Based Cognitive Memory Model

Intrinsic neural excitability biases allocation and overlap of memory engrams

Differentiation and Integration of Competing Memories: A Neural Network Model

A hippocampus CA3 spiking neural network model for storage and retrieval of sequential memory

Neurogenesis and multiple plasticity mechanisms enhance associative memory retrieval in a spiking network model of the hippocampus

A neural network model of differentiation and integration of competing memories

An Excitatory Neural Assembly Encodes Short-Term Memory in the Prefrontal Cortex.

Memory Retention in Pyramidal Neurons: a Unified Model of Energy-Based Homo and Heterosynaptic Plasticity with Homeostasis

Behavioral decision-making of mobile robots simulating the memory consolidation mechanism of human brain

A working memory model based on recurrent neural networks using reinforcement learning

Recurrent Network Models Of Sequence Generation And Memory

A model for integrating elementary neural functions into delayed-response behavior

A Neural Dynamic Model based on Activation Diffusion and a Micro-Explanation for Cognitive Operations

A dynamic attractor network model of memory formation, reinforcement and forgetting

Effective connectivity of the anterior hippocampus predicts recollection confidence during natural memory retrieval

A unifying account of replay as context-driven memory reactivation

Emotional learning retroactively promotes memory integration through rapid neural reactivation and reorganization

Segregation-to-Integration Transformation Model of Memory Evolution

The dynamic reconfiguration of the functional network during episodic memory task predicts the memory performance