← 研踪 PaperTrailarXiv:2611.25593 · v1
Characterizing Age-Conditioned Affective Reasoning in Vision-Language Models
首次提交 2026-11-19 14:39
版本更新 2026-11-19 14:39
雷达观测 2026-11-20 19:58
摘要
An age prompt can change how a vision-language model (VLM) rates the emotion evoked by an image. Does the change predict how people of different ages rate that same image? We evaluate 17 VLMs on 900 OASIS images using young-, middle-, and older-adult prompts and estimate image-specific age differences from ratings by 822 participants. Without an age cue, models closely track average human valence (median Pearson ). Under an older-adult prompt, 16 of 17 models assign lower valence on average than under a young-adult prompt. Yet after adjusting for image pleasantness, correlations between model and human age effects range from to , against an estimated ceiling of . The discrepancy is clearest on 54 sensitive images rated as pleasant. Across models, the older prompt reduces valence by points on average, while older participants rate these images points higher than younger participants. Providing human ratings from other images improves agreement on the direction of age differences, although image-specific fidelity remains modest on 600 held-out images (). These results show why evaluations of demographic personalization in foundation models need human comparisons at the level of individual stimuli. Project page available at https://shahriyar-zaman.github.io/age-conditioned-emotion-reasoning/
作者备注:Preprint. Under review
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v1 · 2026-11-19 14:39 · 最新观测版本
Characterizing Age-Conditioned Affective Reasoning in Vision-Language Models
An age prompt can change how a vision-language model (VLM) rates the emotion evoked by an image. Does the change predict how people of different ages rate that same image? We evaluate 17 VLMs on 900 OASIS images using young-, middle-, and older-adult prompts and estimate image-specific age differences from ratings by 822 participants. Without an age cue, models closely track average human valence (median Pearson ). Under an older-adult prompt, 16 of 17 models assign lower valence on average than under a young-adult prompt. Yet after adjusting for image pleasantness, correlations between model and human age effects range from to , against an estimated ceiling of . The discrepancy is clearest on 54 sensitive images rated as pleasant. Across models, the older prompt reduces valence by points on average, while older participants rate these images points higher than younger participants. Providing human ratings from other images improves agreement on the direction of age differences, although image-specific fidelity remains modest on 600 held-out images (). These results show why evaluations of demographic personalization in foundation models need human comparisons at the level of individual stimuli. Project page available at https://shahriyar-zaman.github.io/age-conditioned-emotion-reasoning/
查看此版本解读 →查看此版本原文 ↗