Neuronal allocation and sparse coding of episodic memories in the human hippocampus

TL;DR

The study shows that human hippocampus uses a sparse coding scheme for episodic memories, with neuron allocation influenced by firing activity during encoding.

Contribution

The paper provides human evidence that sparse, pattern-separated coding in the hippocampus is linked to encoding-related neuronal firing increases.

Findings

Remembered items with increased firing at encoding are associated with sparse hippocampal codes at retrieval.

This sparse coding effect is specific to the hippocampus and not observed in other brain regions.

Encoding-related firing activity may influence the recruitment of neurons into memory codes.

Abstract

Neurocomputational models hold that individual episodic memories are represented by a sparse, pattern-separated coding scheme in the hippocampus. Animal studies further suggest that the allocation of neurons to such codes is non-random and may be biased by their excitability at the time of encoding. Here, utilizing an independent dataset of single-unit recordings from epilepsy patients, we report that only remembered items that elicited a relative increase in firing at encoding were associated with a sparse, pattern-separated neural code at retrieval, and this effect was specific to the hippocampus. These findings provide evidence in humans that remembered episodic memories are represented by sparse codes in the hippocampus. Although excitability is not typically defined in terms of firing rates, the present results raise the possibility that excitability at encoding may influence the likelihood of recruitment into such codes.

The online version contains supplementary material available at 10.1038/s41598-025-21967-7.

Figures