Layer-specific functional gradients uncover intrinsic-network organization and feedback processing

Abstract

Functional cortical gradients capture the brains large-scale organization along sensory-to-association (G1) and visual-to-somatomotor (G2) axes, yet the laminar circuitry that supports these axes remains largely unknown. Leveraging whole-brain submillimeter 7T resting-state fMRI, we estimated functional connectivity within deep, middle, and superficial cortical depths, integrated these into a multilayer network, and derived depth-resolved functional gradients across the cortex. We then computed an inter-regional dissimilarity index quantifying how distinct each region is from the rest of the cortex, and discovered a systematic dissociation across depth: the superficial-layer index closely followed G1, whereas the deep-layer index aligned with G2. The deep-layer index peaked in receive-dominant regions, consistent with deep layers as principal targets of feedback projections, while the superficial-layer index was maximal in cytoarchitecturally less differentiated transmodal cortex, consistent with its dense recurrent circuitry. Together, these findings demonstrate that functional gradients relate to laminar connectivity, and establish a mesoscale framework for decomposing whole-brain cortical connectivity.