analiza aiqbench
This arXiv study examines how vision-language models handle negation by comparing their attention to human eye-fixation data. The authors find that most models do not follow human-like attention when resolving negated statements, suggesting they rely on mechanisms beyond attention. They also flag a left-side spatial bias in LLaVA-1.5-7B and a consistent performance gap between affirmative and negated questions on downstream tasks. For readers building or evaluating multimodal systems, the takeaway is that negation remains a reliability gap that attention-based interpretability alone does not explain.
Probing Negation Understanding in Vision-Language Models With Object Attention
arXiv:2610.20852v1 Announce Type: cross Abstract: Negation is a fundamental aspect of language understanding and it is important vision-language models (VLMs) are able to process it effectively. We investigate how VLMs process negation by analysing attention patterns of different models. We draw comparisons between human gaze and attention in models when processing affirmative vs negated statements. We compare object level model attention to human eye-fixations in both binary and multary settings, evaluating with both synthetic and realistic datasets. We find models generally do not follow human-like attention patterns when resolving negation, suggesting they rely on mechanisms outside of attention. While layerwise analysis shows that PaliGemma2-3B shifts object attention to the alternative under negation, our headwise ablations suggest this is a byproduct rather than a negation processing mechanism. Additionally, we find that LLaVA-1.5-7B displays a spatial bias towards left-side objects. Our evaluation of models on downstream tasks shows that all models have a discrepancy between affirmative and negated questions highlighting the need to better understand the underlying processing of negation to close this performance gap. Code available at https://github.com/otmive/negation_attention
Abstract:Negation is a fundamental aspect of language understanding and it is important vision-language models (VLMs) are able to process it effectively. We investigate how VLMs process negation by analysing attention patterns of different models. We draw comparisons between human gaze and attention in models when processing affirmative vs negated statements. We compare object level model attention to human eye-fixations in both binary and multary settings, evaluating with both synthetic and realistic datasets. We find models generally do not follow human-like attention patterns when resolving negation, suggesting they rely on mechanisms outside of attention. While layerwise analysis shows that PaliGemma2-3B shifts object attention to the alternative under negation, our headwise ablations suggest this is a byproduct rather than a negation processing mechanism. Additionally, we find that LLaVA-1.5-7B displays a spatial bias towards left-side objects. Our evaluation of models on downstream tasks shows that all models have a discrepancy between affirmative and negated questions highlighting the need to better understand the underlying processing of negation to close this performance gap. Code available at this https URL
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
arXiv:2610.20852 [cs.CV]
(or arXiv:2610.20852v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2610.20852 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 278 cuvinte
Subiecte conexe
[AINews] not much happened today
Latent Space · 18 sept. 2026
Improving CoT Monitorability of Evaluation Awareness via Verbalization Training
LessWrong · 2 oct. 2026
Fable and Mythos: Model Welfare
LessWrong · 16 iun. 2026
[AINews] High Taste Testers Ep. 1
Latent Space · 17 oct. 2026