Mixed Signals: Decoding VLMs' Reasoning and Underlying Bias in Vision-Language Conflict

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Pezeshkpour, Pouya, Aminnaseri, Moin, Hruschka, Estevam
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916686246969344
author Pezeshkpour, Pouya
Aminnaseri, Moin
Hruschka, Estevam
author_facet Pezeshkpour, Pouya
Aminnaseri, Moin
Hruschka, Estevam
contents Vision-language models (VLMs) have demonstrated impressive performance by effectively integrating visual and textual information to solve complex tasks. However, it is not clear how these models reason over the visual and textual data together, nor how the flow of information between modalities is structured. In this paper, we examine how VLMs reason by analyzing their biases when confronted with scenarios that present conflicting image and text cues, a common occurrence in real-world applications. To uncover the extent and nature of these biases, we build upon existing benchmarks to create five datasets containing mismatched image-text pairs, covering topics in mathematics, science, and visual descriptions. Our analysis shows that VLMs favor text in simpler queries but shift toward images as query complexity increases. This bias correlates with model scale, with the difference between the percentage of image- and text-preferred responses ranging from +56.8% (image favored) to -74.4% (text favored), depending on the task and model. In addition, we explore three mitigation strategies: simple prompt modifications, modifications that explicitly instruct models on how to handle conflicting information (akin to chain-of-thought prompting), and a task decomposition strategy that analyzes each modality separately before combining their results. Our findings indicate that the effectiveness of these strategies in identifying and mitigating bias varies significantly and is closely linked to the model's overall performance on the task and the specific modality in question.
format Preprint
id arxiv_https___arxiv_org_abs_2504_08974
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Mixed Signals: Decoding VLMs' Reasoning and Underlying Bias in Vision-Language Conflict
Pezeshkpour, Pouya
Aminnaseri, Moin
Hruschka, Estevam
Artificial Intelligence
Computer Vision and Pattern Recognition
Vision-language models (VLMs) have demonstrated impressive performance by effectively integrating visual and textual information to solve complex tasks. However, it is not clear how these models reason over the visual and textual data together, nor how the flow of information between modalities is structured. In this paper, we examine how VLMs reason by analyzing their biases when confronted with scenarios that present conflicting image and text cues, a common occurrence in real-world applications. To uncover the extent and nature of these biases, we build upon existing benchmarks to create five datasets containing mismatched image-text pairs, covering topics in mathematics, science, and visual descriptions. Our analysis shows that VLMs favor text in simpler queries but shift toward images as query complexity increases. This bias correlates with model scale, with the difference between the percentage of image- and text-preferred responses ranging from +56.8% (image favored) to -74.4% (text favored), depending on the task and model. In addition, we explore three mitigation strategies: simple prompt modifications, modifications that explicitly instruct models on how to handle conflicting information (akin to chain-of-thought prompting), and a task decomposition strategy that analyzes each modality separately before combining their results. Our findings indicate that the effectiveness of these strategies in identifying and mitigating bias varies significantly and is closely linked to the model's overall performance on the task and the specific modality in question.
title Mixed Signals: Decoding VLMs' Reasoning and Underlying Bias in Vision-Language Conflict
topic Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2504.08974