Fig. 4: Network matrices comparing changes in resting-state correlations among older adults (65 years and over) with and without college degrees.

Matrices depict significant changes in resting-state correlations (paired t-tests between baseline and the last time point) at the system level for older adults (65 years and over) in each education group. We highlight certain larger circuits that appear to exhibit education-related differences in patterns of change over time. This includes changes (typically increases) in between-system correlations involving regions of the default system, the frontal–parietal control system and the memory-retrieval system (white dotted borders) and decreases in within-system correlations of cingulo–opercular control and visual systems (white arrows), all of which are evident in ‘below college’ older adults but less so in ‘college+’ older adults. Some of these changes also contribute to changes in brain system segregation that were more evident in this ‘below college’ older adult group. Permutation was used to identify which system-level effects (blocks) were significant within each education group (see Methods for permutation details). Within each group, system-level changes (calculated by paired t-tests) that were not significant (cutoff of P < 0.05) (based on 10,000 permutations) are in gray, and the effects that survived false discovery-rate correction (at P < 0.05) are enclosed by solid gray borders.