Abstract
We examined whether Asian individuals would show differential sensitivity to configural vs. featural changes to own-and other-race faces and whether such sensitivity would depend on whether the changes occurred in the upper vs. lower regions of the faces. We systematically varied the size of key facial features (eyes and mouth) of own-race Asian faces and other-race Caucasian faces, and the configuration (spacing) between the eyes and between the nose and mouth of the two types of faces. Results revealed that the other-race effect (ORE) is more pronounced when featural and configural spacing changes are in the upper region than in the lower region of the face. These findings reveal that information from the upper vs. lower region of the face contributes differentially to the ORE in face processing, and that processing of face race is influenced more by information location (i.e., upper vs. lower) than by information type (i.e., configural vs. featural).
Introduction
The "Other-Race Effect" (ORE) is a well-documented psychological phenomenon where individuals are significantly better at recognizing and discriminating faces from their own ethnic group compared to those of other races. While this effect is robust, the underlying mechanisms remain a subject of intense debate. A prominent theory suggests that we process own-race faces "holistically"—integrating features into a single gestalt—while processing other-race faces "featurally," or part-by-part.
However, this study identifies a critical oversight in the "configural vs. featural" debate: the importance of facial location. Given that the eyes (upper face) are highly diagnostic for identity, the researchers hypothesized that the ORE might stem specifically from a failure to process the upper region of other-race faces effectively, regardless of whether the information is configural or featural.
Methodology
The researchers conducted two experiments using a "same/different" discrimination task with 19 Chinese college students. The study utilized the Face Dimensions Test, which allows for precise, systematic manipulation of facial attributes.
Stimuli and Datasets
The stimulus set was derived from an original pool of 36 Asian and 36 Caucasian male face photos. From these, 8 Asian and 8 Caucasian faces were selected based on high "typicality" ratings to ensure results were not skewed by unusually distinctive features. All faces were converted to grayscale and standardized to approximately 250x320 pixels.
Experimental Design
- Experiment 1 (Configural/Spacing): Participants viewed pairs of Asian and Caucasian faces. Researchers manipulated the spacing between the eyes (upper region) or the distance between the nose and mouth (lower region) at three difficulty levels (Easy, Medium, Hard).
- Experiment 2 (Featural/Size): Researchers manipulated the size of the eyes (upper region) or the size of the mouth (lower region) by 10% or 20%.
In both experiments, participants decided if two side-by-side images were identical or different within a 3,000ms time limit. Discrimination accuracy was measured using d' (d-prime), a sensitivity index from Signal Detection Theory that accounts for guessing.
Key Results and Insights
The study revealed that the ORE is not a generalized deficit in processing other-race faces but is instead highly localized to specific facial regions.
The Dominance of the Upper Face
In Experiment 1, a massive ORE was found for the eye region ($p < 0.001$), with participants demonstrating far greater sensitivity to spacing changes in Asian faces. Conversely, for the mouth region, there was no significant difference in performance between own-race and other-race faces ($p > 0.82$).
Experiment 2 showed that while an ORE was present for both regions when manipulating feature size, it remained much more pronounced and stable in the upper face. In the lower face, the ORE only appeared at specific difficulty levels, suggesting the mouth is a less reliable source for racial discrimination.
Location Over Type
Traditionally, researchers have debated whether the ORE is driven by "spacing" (configural) or "parts" (featural) information. This study suggests that the location of the information—upper vs. lower—is more critical than the type of information. The ORE was robust in the upper face for both spacing and size changes, while it was inconsistent or absent in the lower face. This suggests that our expertise in discriminating own-race faces is primarily an expertise for the eye region.
Conclusion and Future Directions
The findings confirm that the ORE is "region-selective," localized primarily in the upper half of the face. For the Chinese participants in this study, the mouth region provided much less "race-sensitive" information, and changes there were harder to detect overall.
Limitations and Future Work
- Participant Diversity: This study focused on Chinese participants. Future research should include Caucasian participants to determine if they show a reciprocal reliance on the upper region of Caucasian faces.
- Cultural Scanning Patterns: Differences in how Westerners and East Asians scan faces (eyes vs. nose/center) may influence which regions drive the ORE.
- The Role of the Nose: While used as a reference point, the nose was not manipulated as an isolated feature. Future studies could isolate the nose to explore its unique role in face processing.
Summary updated