Proceedings of the Annual Meeting of the Cognitive Science Society

Visual object categorization unfolds on the timescale of milliseconds as sensory representations are transformed into decisions through recurrent processing, yet it remains unclear whether this processing merely sharpens category structure shared across individuals or gives rise to individualized category representations. Using MEG, we examined object categorization along continua spanning basic-level category pairs in which category membership was supported either by graded, weakly diagnostic features (Experimental Continua) or by discrete, highly diagnostic features (Control Continua). Time-resolved multivariate decoding showed that early neural activity reflected shared perceptual structure for all continua, whereas later, response-locked activity encoded individual-specific category boundaries. Temporal generalization analyses further demonstrated that recurrent dynamics stabilized these individual category representations. Finally, a distance-to-bound analysis revealed that neural category representations predicted reaction times for Experimental Continua. These findings show that recurrent dynamics do not merely refine universal category axes but reorganize neural representations to reflect each observer's learned category structure.