Biologically inspired alignment reveals distributed and directionally asymmetric correspondence between rat and human cortices

Abstract

Rodent models enable causal interrogation of brain circuits, but translating their systems-level organization to humans requires correspondence beyond one-to-one anatomical homology. Guided by the view that cortical areas are characterized by their relationships within distributed functional networks, we developed BrainMapping, a biologically inspired framework for aligning rat and human cortex using resting-state functional MRI. The framework combined a shared representation with optimal transport to derive probabilistic, bidirectional maps. These alignments revealed distributed, many-to-many correspondences that generalized to held-out subjects and preserved functional connectivity architecture, cortical gradients, and graph topology. Organizational preservation was directionally asymmetric. Human-to-rat mapping preserved organization more uniformly across rat cortical systems, whereas rat-to-human mapping showed stronger preservation in human association-related networks than in primary sensory and motor networks. The maps recovered literature-consistent default-mode and salience-system correspondences and generated testable non-canonical predictions. These findings provide a framework for expressing rat functional organization in human functional-network coordinates.