VPCE: A visual clustering model of place cells with multiscales, multifields, and spatial elongation
Year: 2026
Type: preprint
Abstract: Until recently, most studies of hippocampal place cells have represented spatial distributions using single place fields at a single scale, while representing the fields as circular. More recent recordings in rats and bats, particularly in large environments, have shown multi-field and multi-scale spatial distributions, with some cells showing elongated fields. Based on most of these studies, traditional computational models of place cells represent place-cell spatial distributions as circular single-field, single-scale maps for individual place cells. While there are some models that do consider multiple scales, most of them are based on circular single fields, with the field locations and the number of scales specified in advance. In this work, we introduce VPCE (Visual Place Cell Encoding), an adaptive visual clustering model of place-cell spatial distributions, where multiple field scales result from the clustering of observations rather than being set in advance, and where an individual place cell may produce spatially elongated fields that are active in multiple separate regions in the environment, at varying scales. Such a multiscale and multifield population of place cells is derived exclusively from visual and depth observations, without coordinates, odometry, path integration, or reward. Agglomerative clustering of the observations is used to yield a clustering hierarchy at varying scales. We evaluated the model in multiple simulated environments and feature-space configurations against properties reported in experimental studies in both rats and bats, where smaller fields greatly outnumbered larger ones, with many place cells being active in separate regions, at multiple scales, and producing multiple fields spatially elongated near walls.
Source: Research Square
Authors: Chance J. Hamilton, Alfredo Weitzenfeld
Institution: University of South Florida
Language: English
Topic: Memory and Neural Mechanisms
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7234780023
