Dynamic belief representation and updating through learned attractor-like dynamics in the frontal cortex
- 1Harvard University
Abstract
To act adaptively, animals must infer hidden states of the world from incomplete sensory information and update beliefs as new observations accrue. While dopamine signals are well explained by reinforcement learning models that incorporate belief states, how the brain implements belief-state inference remains unknown. Prior modeling work showed that recurrent neural networks trained to predict value (Value-RNN) develop task-specific, attractor-like dynamics that mirror evolution of beliefs. Here we performed high-density electrophysiological recordings from orbitofrontal cortex (OFC) and other brain areas of mice performing two variants of a Pavlovian task that differ in reward probability, which produce different within-trial belief dynamics. We find that OFC population activity exhibits task-specific attractor-like dynamics that mirror the within-trial dynamics in the Value-RNN. These dynamics are absent in motor and olfactory regions, are not explained by behavioral differences, and emerge progressively with learning. Our findings indicate that the brain approximates belief-state inference through learned, task-specific attractor-like dynamics.
Related
- Cortical reinstatement of causally related events sparks narrative insights by updating neural representation patterns
- A single attractor governs the population dynamics of marine fishes worldwide
- Multiple-context extinction strengthens fear suppression through hippocampal dopamine D2R-dependent memory updating.
- Active locomotion predictively rescues head direction attractor dynamics in head-fixed mice
- Attractor dynamics underlie sequence working memory in data-constrained recurrent neural networks
BibTeX
@misc{pinto2026dynamic,
title = {{Dynamic belief representation and updating through learned attractor-like dynamics in the frontal cortex}},
author = {S. A. Romero Pinto and J. A. Hennig and M. Burrell and D. Okada and C. Benquet and D. Regester and Y. Isogai and S. W. Linderman and S. J. Gershman and N. Uchida},
howpublished = {bioRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.17.752495},
url = {https://www.biorxiv.org/content/10.64898/2026.09.17.752495v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Dynamic belief representation and updating through learned attractor-like dynamics in the frontal cortex
AU - S. A. Romero Pinto
AU - J. A. Hennig
AU - M. Burrell
AU - D. Okada
AU - C. Benquet
AU - D. Regester
AU - Y. Isogai
AU - S. W. Linderman
AU - S. J. Gershman
AU - N. Uchida
PY - 2026
DA - 2026/09/17
PB - bioRxiv
DO - 10.64898/2026.09.17.752495
UR - https://www.biorxiv.org/content/10.64898/2026.09.17.752495v1
AB - To act adaptively, animals must infer hidden states of the world from incomplete sensory information and update beliefs as new observations accrue. While dopamine signals are well explained by reinforcement learning models that incorporate belief states, how the brain implements belief-state inference remains unknown. Prior modeling work showed that recurrent neural networks trained to predict value (Value-RNN) develop task-specific, attractor-like dynamics that mirror evolution of beliefs. Here we performed high-density electrophysiological recordings from orbitofrontal cortex (OFC) and other brain areas of mice performing two variants of a Pavlovian task that differ in reward probability, which produce different within-trial belief dynamics. We find that OFC population activity exhibits task-specific attractor-like dynamics that mirror the within-trial dynamics in the Value-RNN. These dynamics are absent in motor and olfactory regions, are not explained by behavioral differences, and emerge progressively with learning. Our findings indicate that the brain approximates belief-state inference through learned, task-specific attractor-like dynamics.
ER -