Reusable modular architecture enables flexible cognitive operations in the mouse brain and artificial recurrent networks.

Mice and artificial networks both reuse distinct neuronal modules for stimulus processing and memory, demonstrating a flexible, modular organization in complex cognitive tasks.

• Why it matters: Understanding whether the brain employs reusable modules is crucial for unraveling the neural basis of flexible cognition and for designing better artificial systems.
• What they did: The study analyzed neuronal activity in mice performing a delayed match-to-sample task, identifying specialized subspaces supported by distinct neuron clusters, and tested artificial recurrent networks constrained by neural data.
• The result: Disrupting specific neuron clusters impaired related computations, confirming a modular, reusable architecture that enables flexible cognitive operations in both biological and artificial systems.