Standardized human tests show that most vision models are face-blind
- 1Georgia Institute of Technology
Abstract
Face recognition ability varies enormously across humans. Individuals with face-blindness (prosopagnosia) struggle to recognize even close family members while super-recognizers can identify strangers with exceptional accuracy. Where do ANNs fall within the human face recognition spectrum? Here, we administered the standardized tests used to characterize human face recognition ability to a diverse set of models, allowing us to contextualize a model's performance within the distribution of human behavior. We found that the majority 55% of models to be classified as face-blind and that even the best face-trained models do not cross the human threshold to be considered a super-recognizer. Interrogating the internal representations of these models showed that models that performed well on standardized face recognition tasks were more identity-selective and viewpoint-invariant. We also found that low-performing models did contain some identity information in independent representational subspaces. Removing viewpoint-dependent subspace improved face recognition abilities in 49 of the 53 models tested. Targeted unit ablations further identified opposing contributions, with viewpoint-dependent units disrupting identity coding and viewpoint-tolerant units supporting it. Together, our results show that most AI models are face-blind with worse face recognition ability than humans, and demonstrate how differences across AI models can be used generate testable hypotheses about the computational basis of human face recognition in humans which can then be probed in future studies.
Related
- Region-Aware Degradation-Conditioned Vision-Language Learning for Blind Image-Quality Assessment of Power Field-Operation Imagery
- Memory-based and surround-based pattern completion show distinct laminar profiles in human V1
- Standardized human granulocyte isolation using Mono-Poly™ resolving medium: protocol development, validation, and comparison with Lymphoprep™
- Gene Embedding Hub: a standardized resource for human gene representations
- Embeddings from standardized sorghum leaf images capture variation in disease response that human scoring misses
BibTeX
@misc{dudipala2026standardized,
title = {{Standardized human tests show that most vision models are face-blind}},
author = {K. R. Dudipala and F. Martinez-Addiego and D. Rahnev and B. C. Duchaine and N. A. Ratan Murty},
howpublished = {bioRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.18.750666},
url = {https://www.biorxiv.org/content/10.64898/2026.09.18.750666v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Standardized human tests show that most vision models are face-blind
AU - K. R. Dudipala
AU - F. Martinez-Addiego
AU - D. Rahnev
AU - B. C. Duchaine
AU - N. A. Ratan Murty
PY - 2026
DA - 2026/09/18
PB - bioRxiv
DO - 10.64898/2026.09.18.750666
UR - https://www.biorxiv.org/content/10.64898/2026.09.18.750666v1
AB - Face recognition ability varies enormously across humans. Individuals with face-blindness (prosopagnosia) struggle to recognize even close family members while super-recognizers can identify strangers with exceptional accuracy. Where do ANNs fall within the human face recognition spectrum? Here, we administered the standardized tests used to characterize human face recognition ability to a diverse set of models, allowing us to contextualize a model's performance within the distribution of human behavior. We found that the majority 55% of models to be classified as face-blind and that even the best face-trained models do not cross the human threshold to be considered a super-recognizer. Interrogating the internal representations of these models showed that models that performed well on standardized face recognition tasks were more identity-selective and viewpoint-invariant. We also found that low-performing models did contain some identity information in independent representational subspaces. Removing viewpoint-dependent subspace improved face recognition abilities in 49 of the 53 models tested. Targeted unit ablations further identified opposing contributions, with viewpoint-dependent units disrupting identity coding and viewpoint-tolerant units supporting it. Together, our results show that most AI models are face-blind with worse face recognition ability than humans, and demonstrate how differences across AI models can be used generate testable hypotheses about the computational basis of human face recognition in humans which can then be probed in future studies.
ER -