Multi-scale place fields provide a predictive basis for hierarchical control
- 1University of Cambridge
Abstract
Hippocampal place cells typically exhibit unimodal spatial selectivity in small enclosures, but express multiple, irregularly spaced fields in environments of naturalistic scale. However, existing predictive-map theories of the hippocampus do not scale well to large environments and do not account for these heterogeneous place field characteristics. Here, we introduce the factorised hierarchical successor representation (fHSR), a predictive map that supports hierarchical control using subgoal-directed subroutines. We show that the fHSR allows efficient generalisation across tasks, scaling favourably with environmental size, and it reproduces key statistical and topological characteristics of multi-scale CA1 place cell activity in rats and bats. Analyses of experimental data further confirmed a novel prediction of the fHSR: population firing rates exhibit spatial periodicity with a characteristic wavelength independent of environmental size. These results identify the integration of hierarchical control and predictive representations as a normative principle that unifies efficient navigation with the spatial organisation of hippocampal activity.
Related
- Multi-agent meta-reinforcement learning based on contrastive predictive coding—For rapid adaptive traffic control after emergencies
- Localizing predictive concepts encoded in neural representations via hierarchical testing
- Multi-Trait GWAS Analyses of Blood Pressure Traits Provide Limited Gains in Locus Discovery or Polygenic Risk Score Performance Compared to Single-Trait GWAS.
- Tuned inhibitory control of neuronal firing thresholds explains predictive sensorimotor behavior
- A hierarchical Bayesian model for dynamic phenotypes in multi-environment trials with genotype by environment interaction
BibTeX
@misc{yu2026multiscale,
title = {{Multi-scale place fields provide a predictive basis for hierarchical control}},
author = {C. Yu and S. R. Maimon and P. D. Rich and T. Eliav and L. Las and A. K. Lee and N. Ulanovsky and M. Lengyel},
howpublished = {bioRxiv},
year = {2026},
month = oct,
doi = {10.64898/2026.10.28.762461},
url = {https://www.biorxiv.org/content/10.64898/2026.10.28.762461v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Multi-scale place fields provide a predictive basis for hierarchical control
AU - C. Yu
AU - S. R. Maimon
AU - P. D. Rich
AU - T. Eliav
AU - L. Las
AU - A. K. Lee
AU - N. Ulanovsky
AU - M. Lengyel
PY - 2026
DA - 2026/10/28
PB - bioRxiv
DO - 10.64898/2026.10.28.762461
UR - https://www.biorxiv.org/content/10.64898/2026.10.28.762461v1
AB - Hippocampal place cells typically exhibit unimodal spatial selectivity in small enclosures, but express multiple, irregularly spaced fields in environments of naturalistic scale. However, existing predictive-map theories of the hippocampus do not scale well to large environments and do not account for these heterogeneous place field characteristics. Here, we introduce the factorised hierarchical successor representation (fHSR), a predictive map that supports hierarchical control using subgoal-directed subroutines. We show that the fHSR allows efficient generalisation across tasks, scaling favourably with environmental size, and it reproduces key statistical and topological characteristics of multi-scale CA1 place cell activity in rats and bats. Analyses of experimental data further confirmed a novel prediction of the fHSR: population firing rates exhibit spatial periodicity with a characteristic wavelength independent of environmental size. These results identify the integration of hierarchical control and predictive representations as a normative principle that unifies efficient navigation with the spatial organisation of hippocampal activity.
ER -