Fig. 8: Association between GWM-HFN connectivity and Behavior and Cognition measures.

A–C Scatterplots illustrating the significant correlations between PLS1 scores and cognitive measures. A Correlation between GWM-HFN PLS1 scores and the estimated IQ derived from Shipley-Hartford Age-Corrected T-Scores. B Correlation between GWM-HFN PLS1 scores and the Shipley vocabulary task scores. C Correlation between the benchmark GM-GM PLS1 scores and the Shipley vocabulary task scores. The color of the data points in the scatterplots indicates the density of overlapping points, with warmer colors representing higher density. D–F Brain visualizations of the stable, high-weight connectivity patterns driving the significant PLS associations shown in A–C, respectively. Edges displayed are those with absolute standardized PLS1 weights exceeding 3σ and confirmed to be stable via bootstrap analysis. G Density plots of 10,000 permutation tests for the correlation coefficient (r) between connectome data and Shipley vocabulary task scores using the BBS modeling method. The results demonstrate significant individual-level prediction for both the GWM-HFN (left panel) and GM-GM (right panel) frameworks. Circos plots visualizing the key connectivity patterns with the highest contributions ( | z | > 3σ) to the significant BBS predictions for the GWM-HFN (H) and GM-GM (I) vocabulary models. Positive weights are shown in red, and negative weights are in blue. n = 1564 biologically independent participants from the BGSP dataset.