Traces of Semantization, from Episodic to Semantic Memory in a Spiking Cortical Network Model

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• Episodic memory involves recalling personal experiences tied to specific times and places.
• Semantization, or the loss of contextual details in episodic memory, is a common phenomenon.
• The neural mechanisms driving semantization remain largely unknown despite behavioral insights.

Purpose of the Study:

• To propose and test a novel hypothesis linking Bayesian-Hebbian synaptic plasticity to episodic memory semantization.
• To investigate the neural basis of context loss in episodic memory using a computational model.
• To explore how modulating synaptic plasticity might enhance memory retention and resist semantization.

Main Methods:

• Implemented a cortical spiking neural network model utilizing the Bayesian Confidence Propagation Neural Network (BCPNN) learning rule.
• Simulated episodic memory encoding across multiple contexts to observe item-context decoupling.
• Compared the BCPNN rule with spike-timing dependent plasticity (STDP) in the same memory task.

Main Results:

• The BCPNN model successfully replicated the semantization phenomenon, demonstrating item-context decoupling with multiple exposures.
• Unlike BCPNN, the STDP rule did not account for the decontextualization process in the model.
• Selective plasticity modulation during salient events was shown to enhance memory retention and resistance to semantization.

Conclusions:

• Bayesian-Hebbian synaptic plasticity offers a plausible neural mechanism for episodic memory semantization.
• The BCPNN model provides a mechanistic explanation for context loss, bridging synaptic function and behavioral observations.
• Targeted plasticity modulation presents a potential strategy to strengthen episodic memory traces and counteract forgetting.