Structural magnetic resonance imaging (MRI) was performed ex vivo in resilient (Res-POL) and susceptible (Sus-POL) MIA-exposed offspring, which were stratified based on their behavioral performance (see Fig. 3). Behaviorally characterized reference control (Ref-CON) offspring were used as a comparison. a Total brain volume (mm3) across groups. b MRI fly-through of absolute volume differences across groups. No changes survived stringent correction for multiple comparisons (family wise error rate set at p < 0.05). The data shown represent voxel clusters with significant (p < 0.05) group differences revealed in exploratory analyses that were uncorrected for multiple comparisons. The clusters of voxels with the largest effect in the retrosplenial cortex (RSC) are highlighted by arrows. c Heat map showing effect sizes (in units of SD) for regional volume differences in absolute volume (mm3) across each group. Effect sizes are shown for brain areas with significant group differences revealed by uncorrected analyses. d Structural covariance between the seed (RSC) and thalamic reticular nucleus (in both hemispheres). For each group, the linear mixed-effects model fit (solid line) is shown, along with a linear model separately fitted for each subgroup/hemisphere (dotted line and shaded area). A similar pattern of positive covariation in Ref-CON offspring, and loss of covariation in Sus-POL offspring, is also seen in other brain structures (see Supplementary Fig. S7). e Fixed effects for the linear mixed-effects model predicting the volumes of all target structures in the structural covariance analysis. n(Ref-CON) = 16 (8m, 8f), n(Res-POL) = 14 (8m, 6f), and n(Sus-POL) = 14 (8m, 6f).