# FairReason: Balancing Reasoning and Social Bias in MLLMs

arXiv: https://arxiv.org/abs/2507.23067 · HTML: https://mlsanity.com/paper/2507.23067/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Multimodal reasoning, one of 17 directions in LLM reasoning.
- Builds on 9 of its 31 references that are in the map.
- First posted on arXiv 30 Jul 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, 9 of 9)

- [Let's Verify Step by Step](https://mlsanity.com/paper/2305.20050/) (2023)
- [MultiTrust: A Comprehensive Benchmark Towards Trustworthy Multimodal Large Language Models](https://mlsanity.com/paper/2406.07057/) (2024)
- [MetaSpatial: Reinforcing 3D Spatial Reasoning in VLMs for the Metaverse](https://mlsanity.com/paper/2503.18470/) (2025)
- [MMCode: Benchmarking Multimodal Large Language Models for Code Generation with Visually Rich Programming Problems](https://mlsanity.com/paper/2404.09486/) (2024)
- [Chain-of-Action: Faithful and Multimodal Question Answering through Large Language Models](https://mlsanity.com/paper/2403.17359/) (2024)
- [SciSciGPT: Advancing Human-AI Collaboration in the Science of Science](https://mlsanity.com/paper/2504.05559/) (2025)
- [VLBiasBench: A Comprehensive Benchmark for Evaluating Bias in Large Vision-Language Model](https://mlsanity.com/paper/2406.14194/) (2024)
- [Conv-CoA: Improving Open-domain Question Answering in Large Language Models via Conversational Chain-of-Action](https://mlsanity.com/paper/2405.17822/) (2024)
- [Codev-Bench: How Do LLMs Understand Developer-Centric Code Completion?](https://mlsanity.com/paper/2410.01353/) (2024)

## Built on by (most cited first, 3 of 3)

- [Evo-MARL: Co-Evolutionary Multi-Agent Reinforcement Learning for Internalized Safety](https://mlsanity.com/paper/2508.03864/) (2025)
- [MetaFind: Scene-Aware 3D Asset Retrieval for Coherent Metaverse Scene Generation](https://mlsanity.com/paper/2510.04057/) (2025)
- [MagicSim: A Unified Infrastructure for Executable Embodied Interaction](https://mlsanity.com/paper/2606.17511/) (2026)

## Abstract (arXiv)

Multimodal Large Language Models (MLLMs) already achieve state-of-the-art results across a wide range of tasks and modalities. To push their reasoning ability further, recent studies explore advanced prompting schemes and post-training fine-tuning. Although these techniques improve logical accuracy, they frequently leave the models' outputs burdened with pronounced social biases. Clarifying how reasoning gains interact with bias mitigation-and whether the two objectives inherently trade off-therefore remains an open and pressing research problem. Our study begins by benchmarking three bias-mitigation strategies-supervised fine-uning (SFT), knowledge distillation (KD), and rule-based reinforcement learning (RL)-under identical conditions, establishing their baseline strengths and weaknesses. Building on these results, we vary the proportion of debias-focused and reasoning-centric samples within each paradigm to chart the reasoning-versus-bias trade-off. Our sweeps reveal a consistent sweet spot: a roughly 1:4 mix trained with reinforcement learning cuts stereotype scores by 10% while retaining 88% of the model's original reasoning accuracy, offering concrete guidance for balancing fairness and capability in MLLMs.