# Structured Role-Aware Policy Optimization for Multimodal Reasoning

arXiv: https://arxiv.org/abs/2605.07274 · HTML: https://mlsanity.com/paper/2605.07274/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Multimodal reasoning, one of 17 directions in LLM reasoning.
- Builds on 24 of its 57 references that are in the map.
- First posted on arXiv 8 May 2026.

## 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, 24 of 24)

- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [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)
- [InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models](https://mlsanity.com/paper/2504.10479/) (2025)
- [Decoupled Weight Decay Regularization](https://mlsanity.com/paper/1711.05101/) (2017)
- [MathVerse: Does Your Multi-modal LLM Truly See the Diagrams in Visual Math Problems?](https://mlsanity.com/paper/2403.14624/) (2024)
- [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)
- [Self-Distilled RLVR](https://mlsanity.com/paper/2604.03128/) (2026)
- [Visionary-R1: Mitigating Shortcuts in Visual Reasoning with Reinforcement Learning](https://mlsanity.com/paper/2505.14677/) (2025)
- [Skywork R1V2: Multimodal Hybrid Reinforcement Learning for Reasoning](https://mlsanity.com/paper/2504.16656/) (2025)
- [Ground-R1: Incentivizing Grounded Visual Reasoning via Reinforcement Learning](https://mlsanity.com/paper/2505.20272/) (2025)
- [VisPlay: Self-Evolving Vision-Language Models from Images](https://mlsanity.com/paper/2511.15661/) (2025)
- [Perception-R1: Advancing Multimodal Reasoning Capabilities of MLLMs via Visual Perception Reward](https://mlsanity.com/paper/2506.07218/) (2025)
- [KDRL: Post-Training Reasoning LLMs via Unified Knowledge Distillation and Reinforcement Learning](https://mlsanity.com/paper/2506.02208/) (2025)
- [Reinforcement-aware Knowledge Distillation for LLM Reasoning](https://mlsanity.com/paper/2602.22495/) (2026)
- [Exploiting Tree Structure for Credit Assignment in RL Training of LLMs](https://mlsanity.com/paper/2509.18314/) (2025)
- [Perception Before Reasoning: Two-Stage Reinforcement Learning for Visual Reasoning in Vision-Language Models](https://mlsanity.com/paper/2509.13031/) (2025)
- [Revisiting Visual Understanding in Multimodal Reasoning through a Lens of Image Perturbation](https://mlsanity.com/paper/2506.09736/) (2025)
- [Visually-Guided Policy Optimization for Multimodal Reasoning](https://mlsanity.com/paper/2604.09349/) (2026)
- [Bridging Perception and Reasoning: Token Reweighting for RLVR in Multimodal LLMs](https://mlsanity.com/paper/2603.25077/) (2026)
- [Seeing with You: Perception-Reasoning Coevolution for Multimodal Reasoning](https://mlsanity.com/paper/2603.28618/) (2026)
- [Not All Tokens See Equally: Perception-Grounded Policy Optimization for Large Vision-Language Models](https://mlsanity.com/paper/2604.01840/) (2026)
- [VTPerception-R1: Enhancing Multimodal Reasoning via Explicit Visual and Textual Perceptual Grounding](https://mlsanity.com/paper/2509.24776/) (2025)

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

- [Zone of Proximal Policy Optimization: Teacher in Prompts, Not Gradients](https://mlsanity.com/paper/2606.18216/) (2026)
- [Visual sensitivity is not claim retractability: persistence-aware credit assignment for multimodal reinforcement learning](https://mlsanity.com/paper/2609.36572/) (2026)

## Abstract (arXiv)

Reinforcement learning from verifiable rewards (RLVR), especially with Group Relative Policy Optimization (GRPO), has shown strong potential for improving the reasoning capabilities of large vision-language models (LVLMs). However, in multimodal reasoning, final-answer rewards are typically assigned at the sequence level and do not distinguish the functional roles of different tokens, making it difficult to determine whether a correct answer is supported by task-relevant visual evidence. In this paper, we revisit multimodal RLVR from the perspective of role-aware token-level credit assignment, where structured responses are decomposed into perception tokens for extracting visual evidence and reasoning tokens for deriving answers from that evidence. Based on this perspective, we propose Structured Role-aware Policy Optimization (SRPO), which refines the sequence-level GRPO advantage into role-aware token-level advantages without changing the reward function. Specifically, SRPO assigns role-specific credit by using self-distilled on-policy contrasts: perception tokens are emphasized according to their visual dependency under original versus corrupted visual inputs, while reasoning tokens are emphasized according to their consistency with the generated perception. These role-specific signals are further unified through a shared trajectory-level baseline, yielding positive token weights that adjust relative update magnitudes while preserving the original GRPO reward and optimization direction, without requiring external reward models or separate teachers. Experiments across diverse multimodal reasoning benchmarks show that SRPO improves evidence-grounded reasoning, highlighting the importance of moving beyond uniform sequence-level credit toward role-aware optimization for reliable multimodal reasoning.