Visual Evidence Sensitivity in VLMs

VLM: Vision-Language Model

Follow this topic Ask about this topic

Latest papers

73 papers44 papers in the last four weeks, up 1367% on the four weeks before. 0.3% of all new papers.

Consistent Yet Wrong: Evidence Insensitivity in Spatial Vision-Language Models
Jun 1, 2026S Divakar Bhat, Toshihiko YamasakiVLM Evaluation Multi-View Consistency
The University of Tokyo, Japan
Real Images, Worse Judgments: Evaluating Vision-Language Models on Concreteness and Imagery
May 26, 2026Yifan Jiang, Ruoxi Ning, Sheng Yao +1VLM Evaluation Vision-Language Models
University of Waterloo · Vector Institute
Seeing without Looking: Do Vision-Language Benchmarks Really Test Vision?
May 21, 2026Zixuan Lan, Luzhe Sun, Matthew R. Walter +1VLM Evaluation VLM Interpretability
University of Chicago · Toyota Technological Institute at Chicago · Stony Brook University
Are VLMs Seeing or Just Saying? Uncovering the Illusion of Visual Re-examination
May 15, 2026Chufan Shi, Cheng Yang, Yaokang Wu +4VLM Interpretability VLM Robustness
University of Southern California · University of California San Diego · Carnegie Mellon Univer +1
Cross-Modal Attention Analysis and Optimization in Vision-Language Models: A Study on Visual Reliability
Apr 19, 2026Lijie ZhouCross-Modal Learning VLM Robustness
School of Computer Science University of Nottingham Ningbo China Ningbo, China
Arbitration Failure, Not Perceptual Blindness: How Vision-Language Models Resolve Visual-Linguistic Conflicts
Apr 10, 2026Farhad Nooralahzadeh, Omid Rohanian, Yi Zhang +2Vision-Language Grounding Visual Evidence Sensitivity in VLMs
Institute of Computer Science, Zurich University of Applied Sciences, Switzerland · University of Oxford, UK.
How Well Do Vision-Language Models Understand Sequential Driving Scenes? A Sensitivity Study
Apr 8, 2026Roberto Brusnicki, Mattia Piccinini, Johannes BetzVisual Evidence Sensitivity in VLMs Temporal Reasoning in VLMs
Professorship of Autonomous Vehicle Systems, TUM School of Engineering and Design, Technical University of Munich, Munich, Germany
CDH-Bench: A Commonsense-Driven Hallucination Benchmark for Evaluating Visual Fidelity in Vision-Language Models
Mar 30, 2026Kesheng Chen, Yamin Hu, Qi Zhou +2Emotional Hallucination in VLMs VLM Hallucination
Guangdong Provincial Key Laboratory of Novel Security Intelligence Technologies, Institute of Cyberspace Security, School of Computer Science and Technology, Harbin Institute of Technology, Shenzhen, China
Tinted Frames: Question Framing Blinds Vision-Language Models
Mar 19, 2026Wan-Cyuan Fan, Jiayun Luo, Declan Kutscher +2Vision-Language Models VLM Robustness
University of British Columbia · Vector Institute for AI · University of California, Berkeley
To See or To Please: Uncovering Visual Sycophancy and Split Beliefs in VLMs
Mar 19, 2026Rui Hong, Shuxue QuanSelective Prediction for VLMs Sycophancy in VLMs
George Mason University Fairfax, VA, USA · Independent Researcher
Do MLLMs Really See It: Reinforcing Visual Attention in Multimodal LLMs
Feb 9, 2026Siqu Ou, Tianrui Wan, Zhiyuan Zhao +2Reasoning for VLMs Attention Intervention in VLMs
TeleAI · Shanghai Jiao Tong University · Northwestern Polytechnical University
On the Faithfulness of Visual Thinking: Measurement and Enhancement
Oct 27, 2025Zujing Liu, Junwen Pan, Qi She +2Faithfulness in VLMs Multimodal CoT Reasoning
Wuhan University · ByteDance
How Well Do Vision-Language Models Understand Movies?
Jun 6, 2025Emmanouil Zaranis, António Farinhas, Saul Santos +28Video-Language Model Evaluation Multimodal Understanding
Instituto Superior Técnico, Universidade de Lisboa · Instituto de Telecomunicações · ILLC, University of Amsterdam +12

Jun 1, 2026J/K move · Enter open · S save

Consistent Yet Wrong: Evidence Insensitivity in Spatial Vision-Language Models

S Divakar Bhat, Toshihiko Yamasaki

The University of Tokyo, Japan

VLM Evaluation Multi-View Consistency VLM Robustness Spatial Reasoning Benchmarks Visual Spatial Reasoning Visually Grounded Reasoning+1

Open paper AskPDF arXiv

Spatial reasoning is fundamental to robotics, autonomy, and embodied AI, yet modern vision-language models (VLMs) remain unreliable on metric distance queries. A common assumption is that consistent predictions across viewpoints reflect geometric grounding. We test this assumption and find the opposite: leading VLMs often produce view-invariant and consistent answers even when those answers are incorrect, indicating weak coupling between predictions and viewpoint-specific visual evidence. We introduce ViewDiag, a controlled multi-view evaluation protocol built from Hypersim, ScanNet, and KITTI360, comprising 176 object-pair tracks across 80 scenes with 2--10 views per track. The protocol evaluates models along three axes: metric accuracy, distributional concentration, and internal collapse, the last of which is assessed using a latent feature probe. Across diverse models, we observe a consistent pattern of high prediction stability paired with substantial error, clustering in a regime characterized by strong consistency but low accuracy. \noindent These results challenge the common use of cross-view consistency as a proxy for geometric understanding. Instead, we show that stable predictions may reflect prior-driven collapse rather than evidence-sensitive reasoning. ViewDiag provides a controlled benchmark and diagnostic framework for evaluating whether spatial VLMs are not only accurate, but also meaningfully coupled to visual evidence.