Representations converge as brain maps diverge along the cortical hierarchy.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Cognitive Science

Background:

• Individual differences in brain maps (e.g., retinotopy, somatotopy) are presumed to reflect computational variations.
• Artificial neural networks (ANNs) demonstrate that similar performance can arise from diverse circuit layouts.

Purpose of the Study:

• To investigate whether spatial diversity in brain topography correlates with representational diversity across individuals.
• To explore the relationship between functional topographies and representational geometries in the human brain.

Main Methods:

• Utilized task and resting-state functional magnetic resonance imaging (fMRI) data from a large cohort (n=414).
• Compared regional functional topographies and representational geometries (within-individual activity pattern dissimilarities).
• Examined twin data (n=394) to assess the heritability of topographies and representations.

Main Results:

• Representational convergence was observed in higher-order cortex despite significant topographic diversity across individuals.
• Functional topography correlated with representational differences primarily in sensory-motor cortices and architecturally constrained regions.
• Brain topographies were found to be more heritable than representations.

Conclusions:

• Similar information is encoded by distinct, individual-specific activity patterns in higher-order brain regions.
• Architectural constraints, rather than solely learned representations, can lead to idiosyncratic neural layouts.
• The interplay between neural architecture and implementation flexibility shapes the relationship between brain topography and representation.