IBL Finds 14,000 Neurons Favor Diverse Codes Over Categories Across 43 Cortical Regions
Updated
Updated · Nature.com · Jul 15
IBL Finds 14,000 Neurons Favor Diverse Codes Over Categories Across 43 Cortical Regions
3 articles · Updated · Nature.com · Jul 15
Summary
Using 14,000-plus neurons from 43 mouse cortical regions, the International Brain Laboratory found most individual areas encode tasks through highly diverse responses rather than distinct neuron categories.
That diversity supported high-dimensional representations and strong linear separability, letting cortical populations distinguish many experimental conditions even when obvious functional clusters were absent.
Categorical coding appeared mainly in primary sensory regions such as visual and auditory cortex, while pooling neurons across larger modules—or the whole cortex—revealed clearer category-like structure aligned with anatomy.
Across the sensory-to-cognitive hierarchy, higher regions encoded more independent condition combinations, showed less clustering, and exhibited greater representational dimensionality.
The study argues cortical computation is organized by scale: anatomical specialization shapes whole-cortex patterns, but within regions circuits prioritize neural diversity to maximize flexible readout.