On this page
Study design
Researchers recorded mouse prefrontal and parietal activity during a delayed match-to-sample task with a delayed report. They compared neural population subspaces with recurrent-network models constrained by the recorded data.
What the results suggest
The mice reused stimulus-processing subspaces for new inputs and memory-maintenance subspaces for different remembered information. Distinct neuron clusters supported these computations, and silencing corresponding units in data-constrained models selectively disrupted operations.
Limits
The evidence combines one mouse task with computational models; the silencing result was in the models. It does not establish that the same modular organization generalizes across tasks, species or human cognition, nor that these operations can be enhanced. The publisher page exposed the abstract only; detailed methods were not reviewed.
This sourced news summary and translation were prepared with AI assistance. Source links, the news date and research limitations are disclosed; this is not a claim of independent medical expert review. News production method.
Sources
Research news is not medical advice or a recommendation to use a substance. Read findings in the context of the study design, participants and limitations. Editorial standards.
← All news