Figure 3. Task-induced changes in DMN and TPN functional connectivity.

Increasing working memory load is accompanied by decreasing functional connectivity within the DMN (negative ${\mathrm{w}}_1$ values) and increasing functional connectivity within the TPN (positive ${\mathrm{w}}_1$ values). (A) Mean task-induced functional connectivity change of each network edge (node-node connection) in the whole sample. Each edge (cell of the matrix) represents the regression coefficient (${\mathrm{w}}_1$ value) of connectivity change as a function of working memory load, averaged over the whole group. Arrows and black lines indicate the boundary separating nodes allocated to the empirical DMN vs TPN. (B) Mean task-induced functional connectivity change within the DMN, TPN and DMN-TPN across the whole group. Left: Group mean (± S.E.M) functional connectivity strength (Fisher z-transformed r-value) within the DMN, TPN and DMN-TPN as a function of working memory load. Right: Group mean (± S.E.M) task-induced functional connectivity change $\left({\mathrm{w}}_1\right)$ within the DMN, TPN and DMN-TPN. ${\mathrm{w}}_1$ was significantly different to zero in both the DMN (t₅₀ = -2.33, P = 0.02, one-sample t-test) and TPN (t₅₀ = 2.70, P = 0.01, one-sample t-test), but not the DMN-TPN (t₅₀ = 0.65, P = 0.52, one-sample t-test). Repeated measures ANOVA indicated that ${\mathrm{w}}_1$ was not equal within the DMN, TPN and DMN-TPN edges (F_2,100 = 11.16 P < 0.001), and post-hoc paired t-tests confirmed that the ${\mathrm{w}}_1$ of the DMN was significantly lower than both the TPN and DMN-TPN (t₅₀ = -4.56, P < 0.001 and t₅₀ = -3.01, P = 0.004, respectively), but that there was no significant difference between the ${\mathrm{w}}_1$ of the TPN and DMN-TPN (t₅₀ = 1.62, P = 0.11).