Fig. 3. Multivariate prediction of social network position characteristics based on patterns of white matter microstructural integrity.

For each subject, average FA was extracted from each white matter tract in a given brain network, and the resulting set of FA values were used as predictors (as shown in different colors) in a ridge regression-based algorithm to predict individuals’ social network position characteristics (see Methods for further details). This procedure was performed for the mirroring, affective processing, mentalizing, and face perception brain networks.