Face ensembles reshape the neural other-race effect

Technical Summary: Face Ensembles Reshape the Neural Other-Race Effect

Problem Statement
The other-race effect (ORE) is a well-established phenomenon in face perception, characterized by the superior recognition of same-race (SR) faces compared to other-race (OR) faces. While robust behavioral and neural evidence exists for this effect when viewing individual faces, the neural mechanisms underlying the ORE during the perception of face groups (ensembles) remain unclear. This study addresses the gap in understanding how the neural expression of the ORE changes when faces are viewed individually versus as part of a six-face ensemble.

Methodology
The study employed a within-subjects design involving twenty-two East Asian adults. Participants were exposed to both East Asian (SR) and White (OR) faces under two distinct viewing conditions:

1. Individual Format: Single faces presented one at a time.
2. Ensemble Format: Six-face arrays presented simultaneously.

Neural activity was recorded using electroencephalography (EEG). The researchers utilized a multi-faceted analytical approach to characterize the data:

• Decoding Analysis: To assess neural discriminability between SR and OR faces.
• Generative Reconstruction: To determine the recoverable content from neural signals, allowing for the visualization of what the brain "sees."
• Geometric and Dynamic Analysis: To map the structure of face space and the temporal dynamics of information processing.
• Behavioral Testing: To confirm the presence of the behavioral SR advantage.
• Independent Judgments: To evaluate shifts in perceived age, valence, and arousal in the reconstructed images.

Key Contributions and Results
The study provides a nuanced characterization of the neural ORE, revealing that its expression is highly dependent on the viewing context and the specific representational measure used.

• Individual Face Processing: Consistent with established literature, single SR faces demonstrated superior neural discriminability, occupied a more dispersed position in face space, and were reconstructed with higher accuracy compared to OR faces. This confirms a robust neural SR advantage at the individual level.
• Ensemble Processing: The neural dynamics shifted significantly when faces were viewed in groups. While ensembles successfully preserved information regarding racial composition, the cumulative decoding difference between SR and OR faces was reduced. Notably, the ensemble condition revealed an early OR advantage, contrasting with the SR advantage seen in individual processing.
• Reconstruction and Content: Generative reconstructions supported the identification of both individual faces and ensemble summaries. While an overall SR advantage persisted in reconstruction accuracy, there was no reliable interaction between race and format.
• Perceptual Shifts: Independent judgments of the reconstructed images revealed that race and format influenced perceived age, valence, and arousal, indicating that the neural representation of face groups involves complex shifts in affective and demographic attributes.

Significance
The paper claims that these findings fundamentally reshape the understanding of the neural ORE. Rather than being a static deficit in processing other-race faces, the neural ORE is shown to be a dynamic phenomenon whose expression is contingent upon the viewing context (individual vs. ensemble) and the specific metric of neural representation (decoding vs. reconstruction). The study demonstrates that while the SR advantage is a dominant feature of individual face processing, the neural system adapts when processing face ensembles, altering the temporal and geometric properties of racial discrimination.