# Verification Mirage: Mapping the Reliability Boundary of Self-Verification in Medical VQA

arXiv: https://arxiv.org/abs/2605.10850 · HTML: https://mlsanity.com/paper/2605.10850/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Medical vision-language models, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 36 of its 49 references that are in the map.
- First posted on arXiv 11 May 2026.

## Placed in

- Multimodal LLMs / vision-language models / [Medical vision-language models](https://mlsanity.com/mllm/medical-vision-language-models/)

## Builds on (most cited first, 30 of 36)

- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [Language Models (Mostly) Know What They Know](https://mlsanity.com/paper/2207.05221/) (2022)
- [Gemma 3 Technical Report](https://mlsanity.com/paper/2503.19786/) (2025)
- [Teaching Large Language Models to Self-Debug](https://mlsanity.com/paper/2304.05128/) (2023)
- [SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Models](https://mlsanity.com/paper/2303.08896/) (2023)
- [LLM Evaluators Recognize and Favor Their Own Generations](https://mlsanity.com/paper/2404.13076/) (2024)
- [Phi-4-Mini Technical Report: Compact yet Powerful Multimodal Language Models via Mixture-of-LoRAs](https://mlsanity.com/paper/2503.01743/) (2025)
- [PathVQA: 30000+ Questions for Medical Visual Question Answering](https://mlsanity.com/paper/2003.10286/) (2020)
- [SLAKE: A Semantically-Labeled Knowledge-Enhanced Dataset for Medical Visual Question Answering](https://mlsanity.com/paper/2102.09542/) (2021)
- [MedGemma Technical Report](https://mlsanity.com/paper/2507.05201/) (2025)
- [PMC-VQA: Visual Instruction Tuning for Medical Visual Question Answering](https://mlsanity.com/paper/2305.10415/) (2023)
- [Large Language Models are Better Reasoners with Self-Verification](https://mlsanity.com/paper/2212.09561/) (2022)
- [Self-Preference Bias in LLM-as-a-Judge](https://mlsanity.com/paper/2410.21819/) (2024)
- [LEVER: Learning to Verify Language-to-Code Generation with Execution](https://mlsanity.com/paper/2302.08468/) (2023)
- [Lingshu: A Generalist Foundation Model for Unified Multimodal Medical Understanding and Reasoning](https://mlsanity.com/paper/2506.07044/) (2025)
- [MedXpertQA: Benchmarking Expert-Level Medical Reasoning and Understanding](https://mlsanity.com/paper/2501.18362/) (2025)
- [VisualPRM: An Effective Process Reward Model for Multimodal Reasoning](https://mlsanity.com/paper/2503.10291/) (2025)
- [Pride and Prejudice: LLM Amplifies Self-Bias in Self-Refinement](https://mlsanity.com/paper/2402.11436/) (2024)
- [GMAI-MMBench: A Comprehensive Multimodal Evaluation Benchmark Towards General Medical AI](https://mlsanity.com/paper/2408.03361/) (2024)
- [Quilt-LLaVA: Visual Instruction Tuning by Extracting Localized Narratives from Open-Source Histopathology Videos](https://mlsanity.com/paper/2312.04746/) (2023)
- [Self-Evaluation Improves Selective Generation in Large Language Models](https://mlsanity.com/paper/2312.09300/) (2023)
- [MM-Verify: Enhancing Multimodal Reasoning with Chain-of-Thought Verification](https://mlsanity.com/paper/2502.13383/) (2025)
- [Trust, But Verify: A Self-Verification Approach to Reinforcement Learning with Verifiable Rewards](https://mlsanity.com/paper/2505.13445/) (2025)
- [Generate, but Verify: Reducing Hallucination in Vision-Language Models with Retrospective Resampling](https://mlsanity.com/paper/2504.13169/) (2025)
- [UniGen: Enhanced Training & Test-Time Strategies for Unified Multimodal Understanding and Generation](https://mlsanity.com/paper/2505.14682/) (2025)
- [Can Large Vision-Language Models Correct Semantic Grounding Errors By Themselves?](https://mlsanity.com/paper/2404.06510/) (2024)
- [ReXVQA: A Large-scale Visual Question Answering Benchmark for Generalist Chest X-ray Understanding](https://mlsanity.com/paper/2506.04353/) (2025)
- [VeriFact: Verifying Facts in LLM-Generated Clinical Text with Electronic Health Records](https://mlsanity.com/paper/2501.16672/) (2025)
- [Generative Universal Verifier as Multimodal Meta-Reasoner](https://mlsanity.com/paper/2510.13804/) (2025)
- [Let's Think in Two Steps: Mitigating Agreement Bias in MLLMs with Self-Grounded Verification](https://mlsanity.com/paper/2507.11662/) (2025)

## Built on by (most cited first, 2 of 2)

- [EVADE: Evidence-Verified Agentic Diagnosis with Escape](https://mlsanity.com/paper/2608.18833/) (2026)
- [Unmentioned Checklist Findings Change How Reinforcement Learning Appears to Improve Chest Radiograph Report Checking](https://mlsanity.com/paper/2610.05425/) (2026)

## Abstract (arXiv)

Self-verification, re-invoking the same vision language model (VLM) in a fresh context to check its own generated answer, is increasingly used as a default safety layer for medical visual question answering (VQA). We argue that this practice is fundamentally unreliable. We introduce [METHOD NAME], a diagnostic framework for mapping the reliability boundary of medical VLM self-verification by decomposing verifier behavior into discrimination capability and agreement bias. Because the verifier and answer generator are capacity-coupled, the verifier can overly agree with the generator, creating a verification mirage: a regime with both high verifier error and high agreement bias, driven by false acceptance of incorrect answers. Evaluating six open-weight VLMs across five medical VQA datasets and seven medical tasks, we find that this boundary is strongly task-conditioned. Knowledge-intensive clinical tasks fall deepest into the mirage, simpler tasks are more resistant, and perceptual tasks lie in between. Verification also fails to provide an independent safety signal: logistic mixed-effects analysis shows that verifier error and agreement bias become more likely when the generator is wrong, while saliency analyses show that verifiers under-attend to image evidence relative to generators, a phenomenon we call the lazy verifier. Cross-verification reduces but does not eliminate the mirage. Moreover, when verification is reused in multi-turn actor-verifier loops, most initially wrong answers become locked in by false verification. Since our experiments use clean benchmarks, the observed reliability boundary likely underestimates failures in real clinical deployment.