Large-scale brain mediation network based on resting-state functional MRI

Exploring the brain's complex interaction mechanisms is vital for understanding its functions. Traditional network models typically simplify brain region interactions into pairwise connections. However, such interactions may be significantly modulated by a third brain region. Given the widespread connections among numerous regions at the whole-brain scale, constructing an accurate model of brain triple-region interactions is challenging. Here, we develop an independent component-driven mediation brain network model that captures triple-region mediation relationships while controlling extraneous influences. Our results reveal an inverted U-shaped relationship between mediated strength and degree strength, indicating distinct mediation patterns in dense and sparse connected regions. Furthermore, the primary sensory and attention modules exhibit functional hierarchical differentiation: areas responsible for primary information processing belong to the super mediation set, while regions involved in higher-order cognitive functions belong to the super mediated set. These results emphasize the distinct influences of different mediation patterns on the cognitive capabilities of the brain.