Reading Smiles: Proxy Bias in Foundation Models for Facial Emotion Recognition

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Tsangko, Iosif, Triantafyllopoulos, Andreas, Abdelmoula, Adem, Mallol-Ragolta, Adria, Schuller, Bjoern W.
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866912733548511232
author Tsangko, Iosif
Triantafyllopoulos, Andreas
Abdelmoula, Adem
Mallol-Ragolta, Adria
Schuller, Bjoern W.
author_facet Tsangko, Iosif
Triantafyllopoulos, Andreas
Abdelmoula, Adem
Mallol-Ragolta, Adria
Schuller, Bjoern W.
contents Foundation Models (FMs) are rapidly transforming Affective Computing (AC), with Vision Language Models (VLMs) now capable of recognising emotions in zero shot settings. This paper probes a critical but underexplored question: what visual cues do these models rely on to infer affect, and are these cues psychologically grounded or superficially learnt? We benchmark varying scale VLMs on a teeth annotated subset of AffectNet dataset and find consistent performance shifts depending on the presence of visible teeth. Through structured introspection of, the best-performing model, i.e., GPT-4o, we show that facial attributes like eyebrow position drive much of its affective reasoning, revealing a high degree of internal consistency in its valence-arousal predictions. These patterns highlight the emergent nature of FMs behaviour, but also reveal risks: shortcut learning, bias, and fairness issues especially in sensitive domains like mental health and education.
format Preprint
id arxiv_https___arxiv_org_abs_2506_19079
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Reading Smiles: Proxy Bias in Foundation Models for Facial Emotion Recognition
Tsangko, Iosif
Triantafyllopoulos, Andreas
Abdelmoula, Adem
Mallol-Ragolta, Adria
Schuller, Bjoern W.
Computer Vision and Pattern Recognition
Artificial Intelligence
Human-Computer Interaction
Foundation Models (FMs) are rapidly transforming Affective Computing (AC), with Vision Language Models (VLMs) now capable of recognising emotions in zero shot settings. This paper probes a critical but underexplored question: what visual cues do these models rely on to infer affect, and are these cues psychologically grounded or superficially learnt? We benchmark varying scale VLMs on a teeth annotated subset of AffectNet dataset and find consistent performance shifts depending on the presence of visible teeth. Through structured introspection of, the best-performing model, i.e., GPT-4o, we show that facial attributes like eyebrow position drive much of its affective reasoning, revealing a high degree of internal consistency in its valence-arousal predictions. These patterns highlight the emergent nature of FMs behaviour, but also reveal risks: shortcut learning, bias, and fairness issues especially in sensitive domains like mental health and education.
title Reading Smiles: Proxy Bias in Foundation Models for Facial Emotion Recognition
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Human-Computer Interaction
url https://arxiv.org/abs/2506.19079