Abstract
Own-race faces are recognized more effectively than other-race faces. This phenomenon is referred to as other-race effect (ORE). Existing behavioral evidence suggests that one of the possible causes of ORE is that own-race faces are processed more holistically than other-race faces. However, little is known about whether such differences in processing also produce distinctive neural responses in the cortical face processing network. To bridge this gap, the present study used fMRI methodology and the composite face paradigm to examine the response patterns of the traditional face-preferential cortical areas (i.e., the bilateral fusiform face areas [FFA] and the bilateral occipital face areas [OFA]) elicited by own-race faces and other-race faces. We found that the right FFA exhibited a neural composite face effect only for own-race faces but not for other-race faces, even with the absence of the race-related difference in behavior composite face effect. These findings suggest that the right FFA plays a greater role in holistic processing of individual own-race faces than other-race faces. They also suggest that the neural composite effect observed in the right FFA is not the exact neural counterpart of the behavioral face composite effect. The findings of the present study revealed that, along the pathway of the bottom-up face processing, own-race faces and other-race faces presented the holistic processing difference as early as when they were processed in the right FFA.
Introduction: The Other-Race Effect
Psychologists have long documented the Other-Race Effect (ORE): the phenomenon where people are significantly better at recognizing and remembering faces of their own race compared to those of other races.
A leading theory suggests this happens because of holistic processing. We tend to perceive own-race faces as an integrated "whole," whereas we process other-race faces more "featurally" (focusing on individual parts like the nose or eyes). While behavioral evidence for this theory is strong, the neural "where" and "how" remained fuzzy. This paper seeks to bridge that gap by pinpointing which parts of the brain's face-processing network—specifically the Fusiform Face Area (FFA) and the Occipital Face Area (OFA)—drive these differences.
Methods: Probing the Composite Face Effect
The researchers recruited 40 Han Chinese adults and placed them in a 3.0 Tesla fMRI scanner. They used a classic psychological tool called the composite face paradigm to measure holistic processing.
In this task, participants viewed two faces in quick succession. Each face was split into a top and bottom half. Participants had to decide if the top halves were identical while ignoring the bottom halves. The study manipulated two key variables:
- Alignment: In "Aligned" trials, the halves were joined perfectly, making it hard for the brain to ignore the bottom half if it's processing the face holistically. In "Misaligned" trials, the bottom half was shifted to the side, making it easier to isolate the top half.
- Identity: The study looked for "release from adaptation." If the brain processes a face holistically, changing the bottom half (even if the top stays the same) should make the brain treat it as a "new" person, triggering a spike in fMRI activity.
The stimulus set consisted of gray-scale Chinese and Caucasian face images, originally validated by Ge et al. (2009), balanced for luminance and contrast with external features removed.
Key Insights: The Right FFA as a Neural Hub
The most significant finding was that the right FFA showed a neural composite effect only for own-race faces. It treated a change in the bottom half of a Chinese face as a change in identity, but it didn't do the same for Caucasian faces. For own-race faces, "bottom-different" aligned trials elicited significantly higher percent signal change (PSC) than "all-same" aligned trials ($p < 0.017$). This effect vanished for other-race faces.
Because the FFA is an early-to-mid stage in the face-processing stream, this suggests that the ORE isn't just a high-level social bias; it’s baked into the bottom-up visual processing of the face. Interestingly, the neural effect in the right FFA was present even when the participants' behavioral performance inside the scanner didn't show a massive difference between races. This suggests the right FFA is a sensitive marker for holistic processing that can precede overt behavior.
Discussion and Future Directions
While the right FFA emerged as a specialist, the left FFA and the bilateral OFA did not show the same race-sensitive holistic signature. This reinforces the idea that the right FFA is uniquely specialized for the "whole-face" integration required for expert recognition.
A limitation of this study is its homogeneous sample of Han Chinese participants. To truly generalize these findings, a "cross-over" study is needed where Caucasian participants are tested on the same stimuli to see if their right FFA shows the inverse pattern. Future research may also utilize multi-variate pattern analysis (MVPA) to examine how information is distributed across the entire face-processing network.
Summary updated