Fig. 2

Methodology overview. a White matter tractography applied to diffusion MRI data for K = 200 adults participating in the HCP was used to map undirected (i.e., symmetric) weighted adjacency matrices representing the structural connectivity between N = 256,350,512 cortical regions. Navigation efficiency, diffusion efficiency and search information were computed between every pair of regions to generate asymmetric communication matrices. b Resting-state fMRI data for the same HCP participants was used to compute principal component (PC) time series summarizing the functional activity of M = 7,17,22 cortical subsystems. For each individual, effective connectivity between cortical subsystems was computed using spectral DCM. c Schematic of the communication asymmetry test. First, for a pair of nodes i and j, the difference in communication efficiency between the i → j and j ← i directions was computed. Performing this for K individuals yielded the distribution Δ(i, j, k = 1…K). Communication asymmetry was assessed by performing a one-sample t-test to determine whether the mean of this distribution is significantly different to 0, with A(i, j) defined as the resulting matrix of t-statistics. d The asymmetry test was applied to compute M × M matrices of communication and effective connectivity send-receive asymmetries between modules. We sought to test for correlations across the corresponding elements of these two matrices