A mechanistic theory of planning in prefrontal cortex.
4 affiliations
- 1Sainsbury Wellcome Centre, University College London, London, United Kingdom.
- 2Oxford Centre for Integrative Neuroimaging, University of Oxford, Oxford, United Kingdom.
- 3Department of Experimental Psychology, University of Oxford, Oxford, United Kingdom.
- 4Gatsby Computational Neuroscience Unit, University College London, London, United Kingdom.
Abstract
Planning is critical for adaptive behaviour in a changing world, because it lets us anticipate the future and adjust our actions accordingly. While prefrontal cortex is crucial for this process, it remains unknown how planning is implemented in neural circuits. Prefrontal representations were recently discovered in simpler sequence memory tasks, where different populations of neurons represent different future time points. We demonstrate that combining such representations with the ubiquitous principle of neural attractor dynamics allows circuits to solve much richer problems, including planning. This is achieved by embedding the environment structure directly in synaptic connections to implement an attractor network that infers desirable futures. The resulting 'spacetime attractor' excels at planning in challenging tasks known to depend on the prefrontal cortex. Recurrent neural networks trained by gradient descent on such tasks learn a solution that precisely recapitulates the spacetime attractor - in representation, in dynamics, and in connectivity. Analyses of networks trained across different environment structures reveal a generalisation mechanism that rapidly reconfigures the world model used for planning, without the need for synaptic plasticity. The spacetime attractor is a testable mechanistic theory of planning. If true, it would provide a path towards detailed mechanistic understanding of how prefrontal cortex structures adaptive behaviour.
Related
- A Mechanistic Theory Unifying Value and Schema in the Prefrontal Cortex
- Hippocampus-evoked polysynaptic responses in the medial prefrontal cortex are attenuated in aged rats.
- Serotonin2A receptors in the medial prefrontal cortex do not drive DOI-induced head-twitch responses in rodents
- Reference-Gene Stability for RT-qPCR in the Medial Prefrontal Cortex and Hippocampal Subregions of the 5xFAD Mouse Brain
- Large-Scale Network Embedding Predicts Opposing Functions of Adjacent Regions Within Human Prefrontal Cortex
BibTeX
@article{jensen2026mechanistic,
title = {{A mechanistic theory of planning in prefrontal cortex.}},
author = {Kristopher T Jensen and Peter T Doohan and Mathias Sablé-Meyer and Sandra Reinert and Alon Boaz Baram and Maneesh Sahani and Thomas Akam and Timothy E Behrens},
journal = {ELife},
year = {2026},
month = oct,
doi = {10.7554/elife.109757},
url = {https://europepmc.org/article/MED/42904774},
note = {CC BY},
}
RIS
TY - JOUR
TI - A mechanistic theory of planning in prefrontal cortex.
AU - Kristopher T Jensen
AU - Peter T Doohan
AU - Mathias Sablé-Meyer
AU - Sandra Reinert
AU - Alon Boaz Baram
AU - Maneesh Sahani
AU - Thomas Akam
AU - Timothy E Behrens
PY - 2026
DA - 2026/10/21
JO - ELife
DO - 10.7554/elife.109757
UR - https://europepmc.org/article/MED/42904774
AB - Planning is critical for adaptive behaviour in a changing world, because it lets us anticipate the future and adjust our actions accordingly. While prefrontal cortex is crucial for this process, it remains unknown how planning is implemented in neural circuits. Prefrontal representations were recently discovered in simpler sequence memory tasks, where different populations of neurons represent different future time points. We demonstrate that combining such representations with the ubiquitous principle of neural attractor dynamics allows circuits to solve much richer problems, including planning. This is achieved by embedding the environment structure directly in synaptic connections to implement an attractor network that infers desirable futures. The resulting 'spacetime attractor' excels at planning in challenging tasks known to depend on the prefrontal cortex. Recurrent neural networks trained by gradient descent on such tasks learn a solution that precisely recapitulates the spacetime attractor - in representation, in dynamics, and in connectivity. Analyses of networks trained across different environment structures reveal a generalisation mechanism that rapidly reconfigures the world model used for planning, without the need for synaptic plasticity. The spacetime attractor is a testable mechanistic theory of planning. If true, it would provide a path towards detailed mechanistic understanding of how prefrontal cortex structures adaptive behaviour.
SN - 2050-084X
ER -