MULTIMODAL GRADIENTS UNIFY LOCAL AND GLOBAL CORTICAL ORGANIZATION

Specialization of brain areas and subregions, as well as their integration into large-scale networks are key principles in neuroscience. Consolidating both local and global cortical organization, however, remains challenging. Our study developed a new approach to map global cortex-wise similarities of microstructure, structural connectivity, and functional interactions, and integrate these patterns with maps of cortical arealization. Our analysis combined repeated high-field in-vivo 7 tesla (7T) Magnetic Resonance Imaging (MRI) data collected in 10 healthy adults with a recently introduced probabilistic post-mortem atlas of cortical cytoarchitecture. We obtained multimodal eigenvectors describing cortex-wide gradients at the level of microstructural covariance, structural connectivity, and intrinsic functional interactions, and then assessed inter- and intra-area differences in cortex-wide embedding based on these multimodal eigenvectors. Inter-area similarities followed a canonical sensory-fugal gradient, with primary sensorimotor cortex being the most distinctive from all other areas, while paralimbic regions were least distinctive. This pattern largely corresponded to functional connectivity variations across different tasks collected in the same participants, suggesting that the degree of global cortical integration mirrors the functional diversity of brain areas across contexts. When studying heterogeneity within areas, we did not observe a similar relationship, despite overall higher heterogeneity in association cortices relative to paralimbic and idiotypic cortices. The results were replicated in a different dataset. Our findings highlight a close coupling between cortical arealization and global cortical motifs in shaping specialized versus integrative human brain function.Significance Our work situates cytoarchitecture-derived cortical areas within multimodal gradients of cortical microstructure, connectivity, and function derived from high-definition multimodal neuroimaging. We demonstrated that primary sensory and motor areas show most distinctive gradient profiles while paralimbic areas were least distinctive, overall recapitulating a sensory-fugal axis. This axis was shown to relate to the diversity of cortical areas across different functional contexts, and findings could be replicated across an independent dataset. Overall, our work shows how frameworks of cortical arealization and macroscale gradients converge in shaping functional specialization versus integration in the human brain.### Competing Interest StatementThe authors have declared no competing interest.