Schedule of Events | Symposia

Semantic Space Organization of Fifteen Emotional States Decoded from Task fMRI Data

Poster Session D - Monday, March 31, 2025, 8:00 – 10:00 am EDT, Back Bay Ballroom/Republic Ballroom

Yaohui Ding¹ (yaohui.ding@duke.edu), Nathan M. Muncy², Leonard Faul³, John L. Graner¹, Joel S. White¹, John M. Pearson¹, Kevin S. LaBar¹; ¹Duke University, ²University of Nebraska-Lincoln, ³Boston College

We investigated the organization of fifteen decoded emotional states (amusement, anger, anxiety, awe, calmness, craving, disgust, excitement, fear, horror, joy, neutral, romance, sadness, and surprise) in the semantic space. We collected fMRI data from 115 participants (67 female, mean age =30.21 (8.47)) while they underwent emotion induction from movie clips and text scenarios, presented in blocks of five stimuli for each emotion and induction modality. Mass univariate general linear models were used to identify block-level activation (emotion stimuli > washout) in all grey matter voxels, which were subsequently used for classification using partial least squares discriminant analysis (PLS-DA). The PLS-DA classifiers achieved significant above-chance performance for all fifteen emotions from both movies (accuracy = 37.2 %, p < 0.05, CI = [35.5 %, 38.9 %]; AUC = 0.82, p < 0.05, CI = [0.81, 0.87]) and scenarios (accuracy =16.0 %, p < 0.05, CI = [14.8 %, 17.3 %]; AUC = 0.65, p < 0.05, CI = [0.63, 0.67]). The number of classification errors negatively correlated (r = -0.29, p = 0.0026) with the pairwise Euclidean distances among all emotions in a categorical 15-dimensional space, but they were not significantly correlated (r = - 0.18, p = 0.062) with the distances in a 2-dimensional arousal-valence space. Clustering analyses revealed both distinct and grouped structures, e.g., a fear, horror, and anxiety cluster, and a joy and amusement cluster. Taken together, our results contribute to an understanding of the dimensionality and distribution of these fifteen emotions in the semantic space.

Topic Area: EMOTION & SOCIAL: Other