Reusable Modular Architecture Enables Flexible Cognitive Operations

Abstract

Flexible behaviors emerge from the coordination and reuse of multiple distinct cognitive processes. Theoretical studies have posited that artificial neural networks achieve this flexibility through modular architectures, where small functional modules, each specialized for specific computations, are reused across tasks. However, empirical evidence for reusable modular networks in animal brains has been lacking. Neural recordings in mice performing a delayed match-to-sample with delayed report (DMS-dr) task identified reusable subspaces for stimulus processing and memory maintenance. Clustering analysis revealed functionally distinct clusters in medial prefrontal cortex (mPFC) and posterior parietal cortex (PPC), each contributing to a different subspace of neural activity. Recurrent neural networks trained on these dynamics demonstrated that silencing specific clusters disrupted specific computations, highlighting the modular and reusable organization of these networks. By bridging theoretical predictions with empirical evidence, our findings suggest that the brain utilizes a reusable modular network architecture to achieve flexible cognitive behavior.

Note:
Funding Information: This work was supported by NIH grants R01MH133066 and R01NS130361, MURI Grant W911NF2110328, and The Picower Institute Innovation Fund (M.S.); NIH grants R01MH126022 and R01MH129492 (T.J.B.); Japan Society for the Promotion of Science (JSPS) Overseas Research Fellowships, and The Uehara Memorial Foundation Postdoctoral Fellowship (Y.O.)

Declaration of Interests: The authors declare no competing interests.

Ethics Approval Statement: All procedures conducted in this study were approved by the Massachusetts Institute of Technology’s Animal Care and Use Committee and conformed to the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health.

Keywords: Cognitive flexibility, Mouse, Prefrontal cortex, Parietal cortex, Reccurent neural network, Working memory, Stimulus processing, optogenetics