Fig. 5.

Baseline and dynamic effective connectivity components of the default mode network during a session of an exemplar participant. The intrinsic effective connectivity for whole time series under stationarity assumption (A₀stationary) is displayed in (A). (B) reports the average model (A₀avg) of all windows in the session. The dynamic components of effective connectivity are displayed according to the temporal basis function model; DCT (C), PCA(D) and fPCA (E). All models have a design matrix with up to the 6th temporal basis (among total 11 bases), denoted with a superscript for each model. The first regressor is a vector of ones. Effect sizes that survived a criterion of 95% posterior confidence, or more, are shown in rectangles. The basis sets or design matrices used to model the dynamics across windows are displayed in (F) for DCT, (G) for PCA and (H) for fPCA. The adjacency matrix element A(i,j) can be interpreted as effective connectivity from the column to row. For better illustration purpose, the colors in the adjacency matrix was trimmed at two standard deviations deviated from the average of all elements in the adjacency matrix.