Poster D26, Monday, March 27, 8:00 – 10:00 am, Pacific Concourse
Deriving a neural representation of interpersonal guilt from multivariate brain patterns
Hongbo Yu1,6, Leonie Koban2,3, Luke Chang2,4, Ullrich Wagner5, Patrik Vuilleumier3, Xiaolin Zhou1, Tor Wager2; 1Peking University, 2University of Colorado Boulder, 3University of Geneva, 4Dartmouth College, 5University of Münster, 6University of Oxford
Interpersonal guilt is a negative feeling arises from the awareness of one’s own moral transgression. Neuroimaging studies have identified a number of neural substrates of guilt but have not yet derived a neural representation of guilt that is sensitive and specific for guilt processing and generalizable in predicting guilt states in new observations. We used machine learning to derive a neural representation of guilt on the basis of two existing neuroimaging datasets of interpersonal guilt. This pattern discriminated different states of interpersonal guilt in cross-validation (n = 24; Chinese population) and independent test (n = 19; Swiss population) samples. Moreover, it performed at chance level when applied to discriminate different levels of thermal pain or different types of recalled emotions (including guilt), indicating that it is specific to direct experience of interpersonal guilt and not to arousal or salience itself. Within the multivariate pattern, the voxels in the anterior middle cingulate cortex and anterior insula contribute most significantly. Overall, this work highlights an alternative approach for investigating the neural representation of complex social emotion and has implications for theory and measurement of social emotion.
Topic Area: EMOTION & SOCIAL: Emotion-cognition interactions