Benchmarking and Mitigate Sycophancy in Medical Vision-Language Models

• URL: http://arxiv.org/abs/2509.21979v2
• Date: Fri, 10 Oct 2025 12:35:53 GMT
• Title: Benchmarking and Mitigate Sycophancy in Medical Vision-Language Models
• Authors: Zikun Guo, Xinyue Xu, Pei Xiang, Shu Yang, Xin Han, Di Wang, Lijie Hu,
• Abstract summary: Vision language models often exhibit sycophantic behavior prioritizing alignment with user phrasing social cues or perceived authority over evidence based reasoning.<n>This study evaluate clinical sycophancy in medical visual question answering through a novel clinically grounded benchmark.
• Score: 21.353225217216252
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Vision language models(VLMs) are increasingly integrated into clinical workflows, but they often exhibit sycophantic behavior prioritizing alignment with user phrasing social cues or perceived authority over evidence based reasoning. This study evaluate clinical sycophancy in medical visual question answering through a novel clinically grounded benchmark. We propose a medical sycophancy dataset construct from PathVQA, SLAKE, and VQA-RAD stratified by different type organ system and modality. Using psychologically motivated pressure templates including various sycophancy. In our adversarial experiments on various VLMs, we found that these models are generally vulnerable, exhibiting significant variations in the occurrence of adversarial responses, with weak correlations to the model accuracy or size. Imitation and expert provided corrections were found to be the most effective triggers, suggesting that the models possess a bias mechanism independent of visual evidence. To address this, we propose Visual Information Purification for Evidence based Response (VIPER) a lightweight mitigation strategy that filters non evidentiary content for example social pressures and then generates constrained evidence first answers. This framework reduces sycophancy by an average amount outperforming baselines while maintaining interpretability. Our benchmark analysis and mitigation framework lay the groundwork for robust deployment of medical VLMs in real world clinician interactions emphasizing the need for evidence anchored defenses.

Related papers

• CARE: Towards Clinical Accountability in Multi-Modal Medical Reasoning with an Evidence-Grounded Agentic Framework [29.22693846221723]
  We introduce CARE, advancing Clinical Accountability in multi-modal medical Reasoning with an Evidence-grounded agentic framework.<n> CARE decomposes the task into coordinated sub-modules to reduce shortcut learning and hallucination.<n>Our CARE-Flow improves average accuracy by 10.9% over the same size (10B) state-of-the-art (SOTA)
  arXiv  Detail & Related papers  (2026-03-02T08:38:37Z)
• Strong Reasoning Isn't Enough: Evaluating Evidence Elicitation in Interactive Diagnosis [29.630872344186873]
  Interactive medical consultation requires an agent to proactively elicit missing clinical evidence under uncertainty.<n>Existing evaluations largely remain static or outcome-centric, neglecting the evidence-gathering process.<n>We propose an interactive evaluation framework that explicitly models the consultation process using a simulated patient and a revsimulated reporter grounded in atomic evidences.
  arXiv  Detail & Related papers  (2026-01-27T16:36:35Z)
• Making medical vision-language models think causally across modalities with retrieval-augmented cross-modal reasoning [16.243806723551454]
  Medical vision-language models (VLMs) achieve strong performance in diagnostic reporting and image-text alignment.<n>Their underlying reasoning mechanisms remain fundamentally correlational, exhibiting reliance on superficial statistical associations.<n>We propose Multimodal Causal Retrieval-Augmented Generation, a framework that integrates causal inference principles with multimodal retrieval.
  arXiv  Detail & Related papers  (2026-01-26T11:03:00Z)
• Overalignment in Frontier LLMs: An Empirical Study of Sycophantic Behaviour in Healthcare [1.9010852820067994]
  We propose the Adjusted Sycophancy Score, a novel metric that isolates alignment bias by accounting for model instability, or "confusability"<n>Our results suggest that benchmark performance is not a proxy for clinical reliability, and that simplified reasoning structures may offer superior robustness against expert-driven sycophancy.
  arXiv  Detail & Related papers  (2026-01-26T10:21:34Z)
• Benchmarking Egocentric Clinical Intent Understanding Capability for Medical Multimodal Large Language Models [48.95516224614331]
  We introduce MedGaze-Bench, the first benchmark leveraging clinician gaze as a Cognitive Cursor to assess intent understanding across surgery, emergency simulation, and diagnostic interpretation.<n>Our benchmark addresses three fundamental challenges: visual homogeneity of anatomical structures, strict temporal-causal dependencies in clinical, and implicit adherence to safety protocols.
  arXiv  Detail & Related papers  (2026-01-11T02:20:40Z)
• S-Chain: Structured Visual Chain-of-Thought For Medicine [81.97605645734741]
  We introduce S-Chain, the first large-scale dataset of 12,000 expert-annotated medical images with bounding boxes and structured visual CoT (SV-CoT)<n>The dataset further supports 16 languages, totaling over 700k VQA pairs for broad multilingual applicability.<n>S-Chain establishes a new benchmark for grounded medical reasoning and paves the way toward more trustworthy and explainable medical vision-language models.
  arXiv  Detail & Related papers  (2025-10-26T15:57:14Z)
• MedAlign: A Synergistic Framework of Multimodal Preference Optimization and Federated Meta-Cognitive Reasoning [52.064286116035134]
  We develop MedAlign, a framework to ensure visually accurate LVLM responses for Medical Visual Question Answering (Med-VQA)<n>We first propose a multimodal Direct Preference Optimization (mDPO) objective to align preference learning with visual context.<n>We then design a Retrieval-Aware Mixture-of-Experts (RA-MoE) architecture that utilizes image and text similarity to route queries to a specialized and context-augmented LVLM.
  arXiv  Detail & Related papers  (2025-10-24T02:11:05Z)
• EchoBench: Benchmarking Sycophancy in Medical Large Vision-Language Models [82.43729208063468]
  Recent benchmarks for medical Large Vision-Language Models (LVLMs) emphasize leaderboard accuracy, overlooking reliability and safety.<n>We study sycophancy -- models' tendency to uncritically echo user-provided information.<n>We introduce EchoBench, a benchmark to systematically evaluate sycophancy in medical LVLMs.
  arXiv  Detail & Related papers  (2025-09-24T14:09:55Z)
• Automated Clinical Problem Detection from SOAP Notes using a Collaborative Multi-Agent LLM Architecture [8.072932739333309]
  We introduce a collaborative multi-agent system (MAS) that models a clinical consultation team to address this gap.<n>The system is tasked with identifying clinical problems by analyzing only the Subjective (S) and Objective (O) sections of SOAP notes.<n>A Manager agent orchestrates a dynamically assigned team of specialist agents who engage in a hierarchical, iterative debate to reach a consensus.
  arXiv  Detail & Related papers  (2025-08-29T17:31:24Z)
• GEMeX-ThinkVG: Towards Thinking with Visual Grounding in Medical VQA via Reinforcement Learning [50.94508930739623]
  Medical visual question answering aims to support clinical decision-making by enabling models to answer natural language questions based on medical images.<n>Current methods still suffer from limited answer reliability and poor interpretability, impairing the ability of clinicians and patients to understand and trust model-generated answers.<n>This work first proposes a Thinking with Visual Grounding dataset wherein the answer generation is decomposed into intermediate reasoning steps.<n>We introduce a novel verifiable reward mechanism for reinforcement learning to guide post-training, improving the alignment between the model's reasoning process and its final answer.
  arXiv  Detail & Related papers  (2025-06-22T08:09:58Z)
• DeVisE: Behavioral Testing of Medical Large Language Models [14.832083455439749]
  DeVisE is a behavioral testing framework for probing fine-grained clinical understanding.<n>We construct a dataset of ICU discharge notes from MIMIC-IV.<n>We evaluate five LLMs spanning general-purpose and medically fine-tuned variants.
  arXiv  Detail & Related papers  (2025-06-18T10:42:22Z)
• SURE-VQA: Systematic Understanding of Robustness Evaluation in Medical VQA Tasks [2.033441577169909]
  Vision-Language Models (VLMs) have great potential in medical tasks, like Visual Question Answering (VQA)<n>Their robustness to distribution shifts on unseen data remains a key concern for safe deployment.<n>We introduce a novel framework, called SURE-VQA, centered around three key requirements to overcome current pitfalls.
  arXiv  Detail & Related papers  (2024-11-29T13:22:52Z)
• Sycophancy in Vision-Language Models: A Systematic Analysis and an Inference-Time Mitigation Framework [18.54098084470481]
  We analyze sycophancy across vision-language benchmarks and propose an inference-time mitigation framework.<n>Our framework effectively mitigates sycophancy across all evaluated models, while maintaining performance on neutral prompts.
  arXiv  Detail & Related papers  (2024-08-21T01:03:21Z)
• Robust and Interpretable Medical Image Classifiers via Concept Bottleneck Models [49.95603725998561]
  We propose a new paradigm to build robust and interpretable medical image classifiers with natural language concepts. Specifically, we first query clinical concepts from GPT-4, then transform latent image features into explicit concepts with a vision-language model.
  arXiv  Detail & Related papers  (2023-10-04T21:57:09Z)
• Bayesian Networks for the robust and unbiased prediction of depression and its symptoms utilizing speech and multimodal data [65.28160163774274]
  We apply a Bayesian framework to capture the relationships between depression, depression symptoms, and features derived from speech, facial expression and cognitive game data collected at thymia.
  arXiv  Detail & Related papers  (2022-11-09T14:48:13Z)
• Proactive Pseudo-Intervention: Causally Informed Contrastive Learning For Interpretable Vision Models [103.64435911083432]
  We present a novel contrastive learning strategy called it Proactive Pseudo-Intervention (PPI) PPI leverages proactive interventions to guard against image features with no causal relevance. We also devise a novel causally informed salience mapping module to identify key image pixels to intervene, and show it greatly facilitates model interpretability.
  arXiv  Detail & Related papers  (2020-12-06T20:30:26Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.