Moral emotion processing in euthymic bipolar disorder: insights from network analysis

Abstract Background Bipolar disorder (BD) is associated with facial emotion recognition (FER) deficits. Among these, impairments in recognizing moral emotions may be especially relevant to social dysfunction, but their relationship with brain structure remains poorly defined. Methods In this cross-sectional case–control study, 48 euthymic adults with BD (BD-I, n = 20; BD-II, n = 28) and 45 sex- and age-matched healthy controls completed a computerized FER task and 3T T1-weighted MRI scan. FER efficiency (accuracy/response time) for facial moral (anger and disgust) emotion expressions and gray-matter volume (GMV) from 9 bilateral a priori fronto–limbic/salience regions were compared between groups. Group-specific psychometric networks (regularized partial-correlation, i.e. Gaussian graphical models; LASSO/EBIC) integrating FER and GMV nodes were estimated and compared using the Network Comparison Test. Results Relative to BD-II and controls, BD-I showed lower FER efficiency, with poorer decoding of moral but not neutral facial expressions. Network structure differed between BD-II and controls, whereas global network strength was comparable across all pairwise group comparisons. Exploratory centrality analyses suggested a more frontally weighted configuration in BD-I, with the superior frontal gyrus among the more central nodes. Conclusion In BD-I, persistently reduced ability to decode morally salient facial cues co-occurred with subtype-related differences in fronto–limbic structural covariance.