Characterizing Individual Differences in Behavior and Neural Activation Through the Organization of Social Semantic Information

Abstract

This thesis investigates whether individual differences in the semantic organization of social information correspond to differences in neural activation. Building on recent critiques of the assumption of cognitive universality, this study embraces an individual-focused approach to conceptual representation. Participants completed a triadic judgment task using DALL·E-generated face stimuli, allowing for the construction of participant-specific conceptual spaces via ordinal embedding. These behavioral motifs were compared to fMRI data collected during a face-viewing task using both anatomical and functional analyses. Functional connectivity matrices were generated using the fusiform face area (FFA) as a seed region, and individual differences were explored through ROI analyses, Procrustes alignment, and representational similarity metrics. While behavioral data revealed two dominant motifs emphasizing race and gender, neural connectivity profiles varied widely, with limited overlap across subjects. A Mantel test revealed a strong negative correlation between behavioral and neural dissimilarity matrices (r = –0.975), indicating that shared semantic judgments can emerge from divergent neural architectures. These findings provide preliminary evidence that semantic motifs reflect not only behavioral but also neurobiological individuality, and support calls to move beyond group-averaged cognitive models in favor of more personalized frameworks.