Particle.news

Download on the App Store

Researchers Use AI and Brain Data to Uncover Bias in Cross-Racial Face Recognition

A University of Toronto study reveals that other-race faces are processed with less neural detail, appearing more average, younger, and less distinct.

Image

Overview

  • The Other-Race Effect (ORE) was studied using EEG and generative AI, showing people recognize same-race faces more accurately than other-race faces.
  • Neural data revealed that other-race faces elicit less distinct brain responses within the first 600 milliseconds of perception, leading to less detailed mental processing.
  • Participants reconstructed other-race faces as younger, more average, and more expressive than they actually were, highlighting cognitive distortions in face perception.
  • These findings could inform advancements in facial recognition technology, improve eyewitness testimony accuracy, and aid in mental health diagnostics.
  • Researchers are now exploring practical interventions to mitigate racial biases in perception and their broader societal impacts.