# Perception-Consistency Multimodal Large Language Models Reasoning via Caption-Regularized Policy Optimization

arXiv: https://arxiv.org/abs/2509.21854 · HTML: https://mlsanity.com/paper/2509.21854/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- Placed in Multimodal reasoning, one of 17 directions in LLM reasoning.
- Builds on 31 of its 50 references that are in the map.
- First posted on arXiv 26 Sep 2025.

## Placed in

- LLM reasoning / [Multimodal reasoning](https://mlsanity.com/lr/multimodal-reasoning/)
- Multimodal LLMs / vision-language models / [Embodied and 3D vision-language models](https://mlsanity.com/mllm/embodied-and-3d-vision-language-models/)

## Builds on (most cited first, 30 of 31)

- [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025)
- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [DAPO: An Open-Source LLM Reinforcement Learning System at Scale](https://mlsanity.com/paper/2503.14476/) (2025)
- [SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Features](https://mlsanity.com/paper/2502.14786/) (2025)
- [Vision-R1: Incentivizing Reasoning Capability in Multimodal Large Language Models](https://mlsanity.com/paper/2503.06749/) (2025)
- [MMMU-Pro: A More Robust Multi-discipline Multimodal Understanding Benchmark](https://mlsanity.com/paper/2409.02813/) (2024)
- [The Entropy Mechanism of Reinforcement Learning for Reasoning Language Models](https://mlsanity.com/paper/2505.22617/) (2025)
- [A Survey on Hallucination in Large Vision-Language Models](https://mlsanity.com/paper/2402.00253/) (2024)
- [VL-Rethinker: Incentivizing Self-Reflection of Vision-Language Models with Reinforcement Learning](https://mlsanity.com/paper/2504.08837/) (2025)
- [MM-Eureka: Exploring the Frontiers of Multimodal Reasoning with Rule-based Reinforcement Learning](https://mlsanity.com/paper/2503.07365/) (2025)
- [R1-VL: Learning to Reason with Multimodal Large Language Models via Step-wise Group Relative Policy Optimization](https://mlsanity.com/paper/2503.12937/) (2025)
- [We-Math: Does Your Large Multimodal Model Achieve Human-like Mathematical Reasoning?](https://mlsanity.com/paper/2407.01284/) (2024)
- [Logic-RL: Unleashing LLM Reasoning with Rule-Based Reinforcement Learning](https://mlsanity.com/paper/2502.14768/) (2025)
- [LogicVista: Multimodal LLM Logical Reasoning Benchmark in Visual Contexts](https://mlsanity.com/paper/2407.04973/) (2024)
- [GRIT: Teaching MLLMs to Think with Images](https://mlsanity.com/paper/2505.15879/) (2025)
- [OpenVLThinker: Complex Vision-Language Reasoning via Iterative SFT-RL Cycles](https://mlsanity.com/paper/2503.17352/) (2025)
- [Perception-R1: Pioneering Perception Policy with Reinforcement Learning](https://mlsanity.com/paper/2504.07954/) (2025)
- [Perception-Aware Policy Optimization for Multimodal Reasoning](https://mlsanity.com/paper/2507.06448/) (2025)
- [SRFT: A Single-Stage Method with Supervised and Reinforcement Fine-Tuning for Reasoning](https://mlsanity.com/paper/2506.19767/) (2025)
- [Grounded Chain-of-Thought for Multimodal Large Language Models](https://mlsanity.com/paper/2503.12799/) (2025)
- [R1-ShareVL: Incentivizing Reasoning Capability of Multimodal Large Language Models via Share-GRPO](https://mlsanity.com/paper/2505.16673/) (2025)
- [GRPO-CARE: Consistency-Aware Reinforcement Learning for Multimodal Reasoning](https://mlsanity.com/paper/2506.16141/) (2025)
- [SophiaVL-R1: Reinforcing MLLMs Reasoning with Thinking Reward](https://mlsanity.com/paper/2505.17018/) (2025)
- [Reinforced MLLM: A Survey on RL-Based Reasoning in Multimodal Large Language Models](https://mlsanity.com/paper/2504.21277/) (2025)
- [PlanAgent: A Multi-modal Large Language Agent for Closed-loop Vehicle Motion Planning](https://mlsanity.com/paper/2406.01587/) (2024)
- [Fast-Slow Thinking GRPO for Large Vision-Language Model Reasoning](https://mlsanity.com/paper/2504.18458/) (2025)
- [Shuffle-R1: Efficient RL framework for Multimodal Large Language Models via Data-centric Dynamic Shuffle](https://mlsanity.com/paper/2508.05612/) (2025)
- [GThinker: Towards General Multimodal Reasoning via Cue-Guided Rethinking](https://mlsanity.com/paper/2506.01078/) (2025)
- [Revisiting Visual Understanding in Multimodal Reasoning through a Lens of Image Perturbation](https://mlsanity.com/paper/2506.09736/) (2025)
- [Observe-R1: Unlocking Reasoning Abilities of MLLMs with Dynamic Progressive Reinforcement Learning](https://mlsanity.com/paper/2505.12432/) (2025)

## Built on by (most cited first, 9 of 9)

- [Seeing with You: Perception-Reasoning Coevolution for Multimodal Reasoning](https://mlsanity.com/paper/2603.28618/) (2026)
- [CrystaL: Spontaneous Emergence of Visual Latents in MLLMs](https://mlsanity.com/paper/2602.20980/) (2026)
- [TabSieve: Explicit In-Table Evidence Selection for Tabular Prediction](https://mlsanity.com/paper/2602.11700/) (2026)
- [PaperAudit-Bench: Benchmarking Error Detection in Research Papers for Critical Automated Peer Review](https://mlsanity.com/paper/2601.19916/) (2026)
- [OpenVLThinkerV2: A Generalist Multimodal Reasoning Model for Multi-domain Visual Tasks](https://mlsanity.com/paper/2604.08539/) (2026)
- [PivotMerge: Bridging Heterogeneous Multimodal Pre-training via Post-Alignment Model Merging](https://mlsanity.com/paper/2604.22823/) (2026)
- [Spec-o3: A Tool-Augmented Vision-Language Agent for Rare Celestial Object Candidate Vetting via Automated Spectral Inspection](https://mlsanity.com/paper/2601.06498/) (2026)
- [Saliency-Guided Representation with Consistency Policy Learning for Visual Unsupervised Reinforcement Learning](https://mlsanity.com/paper/2604.05931/) (2026)
- [AutoSearch: Adaptive Search Depth for Efficient Agentic RAG via Reinforcement Learning](https://mlsanity.com/paper/2604.17337/) (2026)

## Abstract (arXiv)

While multimodal large language models excel at tasks that integrate visual perception with symbolic reasoning, their performance is often undermined by a critical vulnerability: perception-induced errors that propagate through the reasoning chain. Current reinforcement learning (RL) fine-tuning methods, while enhancing reasoning abilities, largely fail to address the underlying misalignment between visual grounding and the subsequent reasoning process. To address this challenge, we propose Caption-Regularized Policy Optimization (CapPO), a novel RL framework that explicitly enforces perceptual consistency during policy optimization. CapPO integrates two key mechanisms: (1) a caption-based consistency regularization, which minimizes the divergence between responses conditioned on raw images and those conditioned on captions, thereby anchoring reasoning to semantically faithful visual content; and (2) a KL-weighted advantage estimation scheme, which adaptively scales reinforcement signals to strengthen perceptually consistent trajectories while suppressing spurious correlations. Extensive experiments on five math-focused and five general reasoning benchmarks demonstrate that CapPO achieves competitive performance, yielding gains of +6.0% accuracy on math-related tasks and +2.4% on general reasoning tasks over the base Qwen2.5-VL-7B model. Moreover, ablation studies further confirm the effectiveness of each component, while error analysis reveals that CapPO significantly reduces perception-related mistakes compared with baselines. Overall, CapPO provides a simple yet effective framework for improving multimodal reasoning.