Empathy and the Structural Representation of Facial Affect: Evidence from a Genetic-Algorithm Face Synthesis Task

Empathy has been linked to facial emotion recognition, but whether empathy is associated with the structural representation of facial affect (how observers position different affects relative to one another in face-shape space) remains largely unexplored. 53 adults completed a genetic-algorithm face task that generated prototypes for 13 affects using the Basel Face Model, and completed the 60-item Empathy Quotient (EQ). The genetic-algorithm task produced structurally distinct prototypes for all 13 affects (all paired tests p < .001 Bonferroni-corrected; Cohen’s dz 1.74–3.13), confirming that participants generated reliable, affect-specific face representations. A 13 × 13 between-affect distance matrix was then compared between higher-EQ (n = 30) and lower-EQ (n = 23) groups. 1 pair survived full correction across all 156 off-diagonal cells: Amusement × Contempt (Cohen’s d = −1.31), with higher-empathy participants representing these two affects as structurally closer to each other. Both amusement and contempt are social-evaluative affects that share overlapping facial action components, and this convergence may reflect heightened sensitivity to shared expressive structure among higher-empathy observers. In exploratory analyses, permutation testing and continuous-EQ correlations pointed to a broader pattern centered on social-evaluative affects (Amusement, Awe, Contempt, Fear, Happiness, Pride, Sadness, Interest). Individual differences in empathy appear most prominently associated with how social-evaluative affects are structurally positioned in face-shape space, suggesting that empathy modulates not just emotion recognition accuracy but the representational geometry of facial affect itself.