Figure 6. .

Node identities across data-driven modules and their relationship to task performance. (A) Schematic shows a theoretical Subject 1’s connectivity matrices for two conditions A and B. The unique values in the matrix (outlined in red) are vectorized and correlated with each other using Pearson’s r. This produces a single value metric, called similarity, to assess how close the global connectivity patterns are across the network. (B) The group-level connectivity matrices for the spatial encoding (top panel) and temporal encoding (bottom panel) are shown. (C) Each individual’s similarity value between the spatial encoding and temporal encoding networks (both unthresholded) were plotted against their retrieval performance, resulting in a significant negative relationship (robust regression line shown in gray). (D) The computation described in Panel C was performed across different network wiring costs, and the resulting uncorrected p value is plotted. At higher network thresholds, the similarity-performance relationship disappears.