Visualizing the Other-Race Effect with GAN-based Image Reconstruction
Bibliographic record
Abstract
The other-race effect (ORE) describes the advantage of recognizing faces of one’s own race better than other-race faces. While this effect has been extensively documented, its representational basis remains elusive. This study aims to bridge this gap by employing style-based generative adversarial networks (i.e., styleGAN2), a deep learning technique for generating photorealistic images (Karras et al., 2020), in conjunction with facial image reconstruction to investigate the characteristics and mechanisms underlying the ORE. Specifically, we explored how the ORE manifests in styleGAN2, by analyzing the similarity in face representations between GANs and adult participants. This involved assessing the pairwise visual similarity of GAN-generated face images by East Asian and Caucasian participants (N = 106). We then compared the structure of the human face space with that of the GAN latent face space and of other neural network face models (i.e., VGG16 and InsightFace). Our findings suggest that GANs offer insights into face recognition that are not captured by existing models. Furthermore, by leveraging the representational similarity between GANs and human participants, we were able to reconstruct perceptual face representations associated with viewing East Asian and Caucasian face stimuli. Last, we identified latent vector features associated with the ORE and we visualized systematic differences associated with the perception of other-race faces. In conclusion, this research provides a novel perspective on the ORE by integrating generative deep learning techniques in the behavioral study of face perception. The ability of GANs to complement other models of face space structure and perceptual bias underscores their potential as a tool in the study of face perception. Our findings not only contribute to the theoretical understanding of the ORE but also demonstrate the utility of GANs and image reconstruction in behavioral research.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".