Researchers Explore the Other-Race Effect Using AI and EEG
Researchers have combined artificial intelligence with EEG data to gain insights into the Other-Race Effect (ORE), a phenomenon where individuals recognize faces of their own race more accurately than those of other races. The studies indicated that participants tend to perceive faces from different races with less detail, often viewing them as more average, younger, and more expressive.
This diminished recognition capability may contribute to implicit biases and holds significant implications for many areas, such as enhancing facial recognition technology, eyewitness accuracy, and addressing social bias while advancing mental health diagnostics.
Key Findings
- Less Detail in Processing: Brain activity indicates that faces of different races are processed in a more general manner, affecting recognition accuracy.
- Misperceptions: Participants reconstructed other-race faces as more average-looking, younger, and more expressive.
- Possible Applications: These insights may help mitigate social bias, refine facial recognition tools, and inform mental health diagnosis practices.
The University of Toronto’s U of T Scarborough researchers have aimed to understand why recognizing faces from different races can be challenging.
In their studies, the researchers focused on the ORE, which suggests people are generally better at recognizing faces similar to their own. They utilized AI alongside EEG data to explore how we interpret other-race faces. Their findings reveal that these perceptual distortions may be more embedded in our cognitive processes than we previously realized.
“What we’ve discovered is quite remarkable—individuals are significantly adept at noticing facial details in people of their own race,” stated Adrian Nestor, an associate professor involved in the research.
This research included two studies. The first one, recently published in the journal Behavior Research Methods, employed generative AI to analyze participants’ responses to images of faces. Two groups—one East Asian, the other White—viewed various faces and rated their similarity. By leveraging a generative adversarial network (GAN), the researchers were able to visualize the mental images of the participants.
The results indicated that participants reconstructed same-race faces more accurately than those of different races. Notably, faces from other races tended to be perceived as younger.
Brain Activity Insights
The second study, published in the journal Frontiers, delved deeper into the brain activity related to ORE. It focused on the first 600 milliseconds of participants viewing images, reconstructing their visual processing of faces through digital means.
As complicated as it sounds, this approach is indeed innovative. Nestor’s lab had previously capitalized on EEG for visual perception studies, and their methods have since improved. The findings revealed that the brain processes faces of the same race distinctly compared to those of other races. Photos of different races elicited less neural differentiation, suggesting that these faces are categorized more generally.
Interestingly, other-race faces appeared not just more average but also younger and more expressive to participants, even when they weren’t. This raises potential explanations for the struggles individuals often face in recognizing different-race faces. The brain simply doesn’t process them with the same level of detail.
Real-World Applications
The research, backed by a grant from the Natural Sciences and Engineering Research Council of Canada (NSERC), could lead to wide-ranging consequences. Nestor suggests that understanding how biases form in the brain could aid in improving facial recognition software and enhance the reliability of eyewitness accounts. It might even help with diagnosing mental health disorders.
“Understanding distortions in emotional perception is crucial,” said Nestor. “For instance, identifying how someone interprets emotions can enhance mental health diagnostics and treatment strategies.”
Shoura further emphasized the importance of examining perceptual biases to improve various social interactions, from job interviews to reducing racial bias in initial face-to-face meetings. “If we can grasp how the brain interprets faces, we can devise strategies to lessen the impact of bias that surfaces when meeting someone from another race.”
