Is race erased? decoding race from patterns of neural activity when skin color is not diagnostic of group boundaries

Kyle G. Ratner, Christian Kaul, Jay J. Van Bavel

Research output: Contribution to journalArticlepeer-review


Several theories suggest that people do not represent race when it does not signify group boundaries. However, race is often associated with visually salient differences in skin tone and facial features. In this study, we investigated whether race could be decoded from distributed patterns of neural activity in the fusiform gyri and early visual cortex when visual features that often covary with race were orthogonal to group membership. To this end, we used multivariate pattern analysis to examine an fMRI dataset that was collected while participants assigned to mixed-race groups categorized own-race and other-race faces as belonging to their newly assigned group. Whereas conventional univariate analyses provided no evidence of race-based responses in the fusiform gyri or early visual cortex, multivariate pattern analysis suggested that race was represented within these regions. Moreover, race was represented in the fusiform gyri to a greater extent than early visual cortex, suggesting that the fusiform gyri results do not merely reflect low-level perceptual information (e.g. color, contrast) from early visual cortex. These findings indicate that patterns of activation within specific regions of the visual cortex may represent race even when overall activation in these regions is not driven by racial information.

Original languageEnglish (US)
Article numbernss063
Pages (from-to)750-755
Number of pages6
JournalSocial cognitive and affective neuroscience
Issue number7
StatePublished - Oct 2013


  • Face network
  • Fmri
  • Fusiform gyrus
  • Multivariate pattern analysis
  • Race

ASJC Scopus subject areas

  • Experimental and Cognitive Psychology
  • Cognitive Neuroscience


Dive into the research topics of 'Is race erased? decoding race from patterns of neural activity when skin color is not diagnostic of group boundaries'. Together they form a unique fingerprint.

Cite this