Peek-a-Boo Reasoning: Contrastive Region Masking in MLLMs

• URL: http://arxiv.org/abs/2512.08976v1
• Date: Wed, 03 Dec 2025 16:05:32 GMT
• Title: Peek-a-Boo Reasoning: Contrastive Region Masking in MLLMs
• Abstract summary: We introduce Contrastive Region Masking, a training free diagnostic.<n>It reveals how multimodal large language models (MLLMs) depend on specific visual regions at each step of chain-of-thought (CoT) reasoning.
• Score: 8.669514778798222
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We introduce Contrastive Region Masking (CRM), a training free diagnostic that reveals how multimodal large language models (MLLMs) depend on specific visual regions at each step of chain-of-thought (CoT) reasoning. Unlike prior approaches limited to final answers or attention maps, CRM provides causal, step-level attri- bution by systematically masking annotated regions and contrasting the resulting reasoning traces with unmasked baselines. Applied to datasets such as VisArgs, CRM reveals distinct failure modes: some models preserve reasoning structure, but hallucinate when evidence is missing, while others ground tightly to visual cues yet collapse under perturbations. By shifting the evaluation from correctness of an- swers to faithfulness of reasoning, CRM reframes visual benchmarks as diagnostic tools, highlighting the need for multimodal evaluation frameworks that measure not just performance, but also robustness and fidelity of reasoning.

Related papers

• UR$^{2}$-MLLM: Uncertainty-aware Revisit Reasoning in Multimodal Large Language Models for Radiology Report Generation [10.99095149424783]
  Existing methods provide static visual evidence without a dynamic revisit mechanism during reasoning.<n>We propose an Uncertainty-aware Revisit Reasoning MLLM framework that dynamically revisits uncertain regions during reasoning.<n>Experiments on MIMIC-CXR and IU-Xray show that UR$2$-MLLM achieves state-of-the-art performance.
  arXiv  Detail & Related papers  (2026-08-23T04:43:45Z)
• Beyond Visual Cues: CoT-Enhanced Reasoning for Semi-supervised Medical Image Segmentation [11.723739388396815]
  CERS (CoT-Enhanced Reasoning) is a framework that integrates Chain-of-Thought (CoT) reasoning to distinguish pathologically distinct cases.<n>A semantic-aware reference selection strategy is introduced to identify historical evidence.<n>A multi-scale coordinate attention module is designed to effectively fuse this reasoning-derived context into the decoding process.
  arXiv  Detail & Related papers  (2026-06-16T14:10:19Z)
• Motion-Guided Causal Disentanglement for Robust Multi-View Cine Cardiac MRI Diagnosis [0.4638331367650009]
  We propose a Motion-Guided View-Disease Disentanglement framework MoViD built upon a ViT-MAE backbone.<n>We evaluate the framework on a private clinical venous thrombosis dataset and two public benchmarks.
  arXiv  Detail & Related papers  (2026-06-03T03:45:11Z)
• MetaRA: Metamorphic Robustness Assessment for Multimodal Large Language Model-based Visual Question Answering Systems [35.86959578973465]
  We propose Metamorphic Robustness Assessment (MetaRA), a testing framework that employs Metamorphic Relations (MRs) to probe vulnerabilities in MLLM-based VQA systems.<n>Applying MetaRA to multiple MLLM-based VQA models across different tasks reveals nuanced failure patterns, including sensitivity to linguistic perturbations, over-reliance on superficial visual cues, and deeper weaknesses in multimodal reasoning.
  arXiv  Detail & Related papers  (2026-05-19T03:37:54Z)
• Enhancing Multimodal In-Context Learning via Inductive-Deductive Reasoning [65.15766304205657]
  In-context learning (ICL) allows large models to adapt to tasks using a few examples, yet its extension to vision-language models (VLMs) remains fragile.<n>Our analysis reveals that the fundamental limitation lies in an inductive gap, models often produce correct answers from flawed reasoning.<n>We introduce a framework that restructures multimodal ICL as a principled inductive-deductive process.
  arXiv  Detail & Related papers  (2026-05-04T09:18:19Z)
• What, Whether and How? Unveiling Process Reward Models for Thinking with Images Reasoning [18.918845075485]
  This work introduces the first comprehensive benchmark specifically designed for evaluating Process Reward Models (PRMs) under the thinking with images paradigm.<n>We define 7 fine-grained error types and demonstrate both the necessity for specialized PRMs and the potential for improvement.<n>Our experimental analysis reveals that current LVLMs fall short as effective PRMs, exhibiting significant performance disparities across error types, positive evaluation bias, and sensitivity to reasoning step positions.
  arXiv  Detail & Related papers  (2026-02-09T07:31:14Z)
• Do MLLMs Really See It: Reinforcing Visual Attention in Multimodal LLMs [55.61018839017648]
  Chain-of-thought (CoT) reasoning has substantially improved multimodal large language models (MLLMs) on complex reasoning tasks.<n>Existing approaches largely rely on long textual reasoning trajectories and provide limited mechanisms for learning stable visual attention policies.<n>We propose SAYO, a visual reasoning model trained with a reinforcement learning framework that introduces a region-level visual attention-based reward.
  arXiv  Detail & Related papers  (2026-02-09T03:33:23Z)
• Analyzing Reasoning Consistency in Large Multimodal Models under Cross-Modal Conflicts [74.47786985522762]
  We identify a critical failure mode termed textual inertia, where models tend to blindly adhere to the erroneous text while neglecting conflicting visual evidence.<n>We propose the LogicGraph Perturbation Protocol that structurally injects perturbations into the reasoning chains of diverse LMMs.<n>Results reveal that models successfully self-correct in less than 10% of cases and predominantly succumb to blind textual error propagation.
  arXiv  Detail & Related papers  (2026-01-07T16:39:34Z)
• MM-CoT:A Benchmark for Probing Visual Chain-of-Thought Reasoning in Multimodal Models [49.32415342913976]
  We introduce MM-CoT, a diagnostic benchmark designed to probe the visual grounding and logical coherence of CoT reasoning in multimodal models.<n>We evaluate leading vision-language models on MM-CoT and find that even the most advanced systems struggle, revealing a sharp discrepancy between generative fluency and true reasoning fidelity.
  arXiv  Detail & Related papers  (2025-12-09T04:13:31Z)
• MRFD: Multi-Region Fusion Decoding with Self-Consistency for Mitigating Hallucinations in LVLMs [57.931366749890906]
  Multi-Region Fusion Decoding (MRFD) is a training-free decoding method that improves factual grounding by modeling inter-region consistency.<n>MRFD identifies salient regions using cross-attention, generates initial responses for each, and computes reliability weights based on Jensen-Shannon Divergence.
  arXiv  Detail & Related papers  (2025-08-14T01:17:39Z)
• Joint Evaluation of Answer and Reasoning Consistency for Hallucination Detection in Large Reasoning Models [12.270274049887298]
  Reasoning traces can be redundant or logically inconsistent, making them a new source of hallucination.<n>Existing hallucination detection methods focus primarily on answer-level uncertainty.<n>We propose RACE, a novel framework specifically tailored for hallucination detection in LRMs.
  arXiv  Detail & Related papers  (2025-06-05T09:54:04Z)
• Semi-structured LLM Reasoners Can Be Rigorously Audited [43.72257027288094]
  We introduce Semi-Structured Reasoning Models (SSRMs), which are trained to produce semi-structured representations of reasoning.<n> SSRMs generate reasoning traces in a non-executable Pythonic syntax that names each reasoning step and marks its inputs and outputs.<n>We show that all of these methods can be used to effectively flag probable reasoning errors.
  arXiv  Detail & Related papers  (2025-05-30T05:06:10Z)
• Socratic-PRMBench: Benchmarking Process Reward Models with Systematic Reasoning Patterns [79.42805969325036]
  Process Reward Models (PRMs) are crucial in complex reasoning and problem-solving tasks.<n>PRMs are required to identify errors under various reasoning patterns during the reasoning process.<n>Existing benchmarks mainly focus on evaluating PRMs with stepwise correctness.<n>We introduce Socratic-PRMBench, a new benchmark to evaluate PRMs systematically under six reasoning patterns.
  arXiv  Detail & Related papers  (2025-05-29T14:26:53Z)
• Retrieval is Not Enough: Enhancing RAG Reasoning through Test-Time Critique and Optimization [58.390885294401066]
  Retrieval-augmented generation (RAG) has become a widely adopted paradigm for enabling knowledge-grounded large language models (LLMs)<n>RAG pipelines often fail to ensure that model reasoning remains consistent with the evidence retrieved, leading to factual inconsistencies or unsupported conclusions.<n>We propose AlignRAG, a novel iterative framework grounded in Critique-Driven Alignment (CDA)<n>We introduce AlignRAG-auto, an autonomous variant that dynamically terminates refinement, removing the need to pre-specify the number of critique iterations.
  arXiv  Detail & Related papers  (2025-04-21T04:56:47Z)
• Spatial Reasoning with Denoising Models [49.83744014336816]
  We introduce a framework to perform reasoning over sets of continuous variables via denoising generative models.<n>For the first time, that order of generation can successfully be predicted by the denoising network itself.<n>Using these findings, we can increase the accuracy of specific reasoning tasks from 1% to >50%.
  arXiv  Detail & Related papers  (2025-02-28T14:08:30Z)

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.