FIGURE 3.

Relationship between amygdala-mPFC connectivity and negative valence bias as a function of PDS score. (a) A seed region in the right amygdala was defined using the contrast of surprised facial expressions versus baseline (p < .0005). The dorsal position of this cluster within the amygdala is consistent with previous work demonstrating that content conveying ambiguous valence recruits the amygdala/substantia innominata in particular (Kim, Somerville, McLean, et al., 2003; Whalen et al., 2009). (b) A PPI analysis based on surprise > neutral activity in the amygdala seed revealed a relationship between amygdala connectivity and valence bias that was moderated by PDS score in the mPFC (peak-t₃₆ = 4.50, p = .00007). (c) The estimated regression slopes between valence bias and the surprise > neutral amygdala-mPFC connectivity betas, at a lower (gray line; 1 standard deviation below the mean PDS score; t₃₆ = −4.89, p = .00002) and relatively higher (black line; 1 standard deviation above the mean PDS score; t₃₆ = 3.52, p = .001) PDS scores. More mature children (within this relatively immature sample) that have a mature (inverse) connectivity pattern were more likely to have a positive valence bias while those that have the less mature (positive) connectivity pattern were more likely to show a negative valence bias; this relationship reached significance for PDS scores between 2.0 and 2.8 (blue shaded area). Lower PDS scores predicted the opposite relationship between valence bias and amygdala-mPFC connectivity; this relationship reached significance for PDS scores between 1.0 and 1.4 (pink shaded area)