Fig. 2. Salience network flexibility construction and results.

a Illustrates time courses of blood-oxygen-level-dependent (BOLD) signals from one of the 264 brain regions defined by ref. ⁵⁷. Functional connectivity between the time courses of each pair of brain regions is calculated via wavelet coherence within overlapping time windows of 20 TRs in length. b Illustrates connectivity matrices indicating the functional connectivity among each pair of brain regions. Four of the 15 sliding windows are illustrated. c Illustrates the construction of node flexibility. Multilayer modularity maximization is used to assign each brain region (node; y-axis) to a community (assignment indicated by color) at each sliding window (x-axis). Panel d highlights the coordinates of regions in the salience network as defined by ref. ⁵⁷. Flexibility of an individual node captures the number of times the node changed community, normalized by the number of times the node could have changed communities. e Illustrates the values of salience network flexibility at which the association between previous moment’s sadness and repetitive negative thinking is significant. The dashed orange vertical line at 0.005 indicates the value of salience network flexibility at which the effect of previous moment’s sadness on negative thinking becomes non-significant. The upper bound of the region of significance for the salience network flexibility variable (0.04) is not shown as values below this bound are not observed in the sample. The range of values we observe in the sample is indicated by the horizontal green line at previous moment’s sadness = 0. The ribbon indicates the standard error of the mean. f Indicates the effect of previous moment’s sadness on negative thinking at low and high levels of salience network flexibility. Low and high values of between-person, sample-mean centered salience network flexibility reflect plus and minus 1 standard deviation about the mean (−0.03, 0.03). Values for the previous moment’s sadness on the x-axis reflect plus and minus 1 standard deviation about the mean. The slope of the simple regression of negative thinking on sadness at low levels of salience network flexibility is significant such that people with lower than average levels of salience network flexibility experienced significant increases in negative thinking following higher than usual levels of sadness at the previous measurement occasion. The slope of the simple regression at high levels of salience network flexibility is not significant. The ribbon indicates the standard error of the mean