Hippocampal CA2 modulates its geometry to solve the memory-generalization tradeoff for social memory
Neuroscience ePoster by Lorenzo Posani, Lara Boyle, Steven A. Siegelbaum, Steven A Siegelbaum and Stefano Fusi
Part of COSYNE 2023 posters, Columbia University; Center for Theoretical Neuroscience, Zuckerman Institute; Columbia University
Sunday
Abstract
Social memory, one's ability to recognize and remember experiences with other conspecifics, consists of several discrete psychological processes. These include the ability to rapidly detect whether an individual is novel or familiar (“I know that person, but from where?”) and the recollection of an individual’s specific identity and the associated set of past experiences (“ah, we met at COSYNE last year”). These processes have conflicting demands and requirements. Familiarity is an abstract concept that can readily generalize to any individual in any context. In contrast, social episodes consist of complex multi-dimensional experiences with a particular individual at specific places and times. How does the brain manage these conflicted memory requirements? Are there brain mechanisms that provide for the abstraction of familiarity and simultaneously enable detailed storage and recall of social experience? Here, we use calcium imaging from large populations of hippocampal CA2 pyramidal neurons to show that their activity is able to accommodate these competing demands by representing novel and familiar individuals in distinct representational geometries. We found that through the encoding of novel animals in low-dimensional representations, CA2 activity enables the identity of novel animals to be readily disentangled from their position. Such low-dimensional representations suffer from having a limited memory storage capacity. We found that CA2 solves this problem by transforming the representations of familiar individuals into a higher dimensional geometry, favoring memory storage at the cost of generalization. Crucially, the degree of this familiarity-induced increase of dimensionality correlated with behavioral performance in a social recognition test. Our results provide the first evidence that experience-dependent transformations in the geometry of neural representations can enable the hippocampal activity to meet the distinct demands of generalization and high-capacity storage needed to support social memory.
Topics
Details
- Session
- Poster Session III
Related ePosters
- Mar 19, 2022Lisbon, Portugal
Social and spatial codes in the hippocampus
Vienna, Austria
More in Poster Session III
All postersAbstract structure and generalization in sensorimotor networks configured with semantic-based instruction embeddings
Mar 12, 2023Reidar Riveland et al.Accounting for visual cortex variability with distributed neural activity states
Mar 12, 2023Anna Li et al.Alignment of ANN Language Models with Humans After a Developmentally Realistic Amount of Training
Mar 12, 2023Eghbal Hosseini et al.Arousal dynamics: diverse measurements of a universal manifold
Mar 12, 2023Ryan Raut et al.
More with Lorenzo Posani
All their workNeuralNet 2026: 15th Annual Meeting of the GDR NeuralNet
Conference · Nov 25–27, 2026Weakly structured selectivity and high-dimensional representations across the cortical hierarchy
ePoster · Mar 27, 2025COSYNE 2025Unstructured representations in a structured brain: a cross-region analysis of the neural code
ePoster · Sep 29, 2024Bernstein Conference 2024Hippocampal place field formation by sparse, local learning of visual features in virtual reality
ePoster · Jun 25, 2024FENS Forum 2024