Reasoning Like Humans: Enhancing Multi-Image Reasoning Via Cognition-Inspired Meta-Action Framework
Its authors withdrew this paper, so there is no full text to read or summarize.
Lists
Only you see this
Abstract
While Multimodal Large Language Models (MLLMs) excel at single-image understanding, they exhibit significantly degraded performance in multi-image reasoning scenarios. Multi-image reasoning presents fundamental challenges including complex inter-relationships between images and scattered critical information across image sets. Inspired by human cognitive processes, we propose the Cognition-Inspired Meta-Action Framework (CINEMA), a novel approach that decomposes multi-image reasoning into five structured meta-actions: Global, Focus, Hint, Think, and Answer which explicitly modeling the sequential cognitive steps humans naturally employ. For cold-start training, we introduce a Retrieval-Based Tree Sampling strategy that generates high-quality meta-action trajectories to bootstrap the model with reasoning patterns. During reinforcement learning, we adopt a two-stage paradigm: an exploration phase with Diversity-Preserving Policy Optimization (DiPO) to avoid entropy collapse, followed by an annealed exploitation phase with DAPO to to gradually strengthen exploitation. To train our model,We construct a dataset of 57k cold-start and 58k reinforcement learning instances spanning multi-image, multi-frame, and single-image tasks. We conduct extensive evaluations on multi-image reasoning benchmarks, video understanding benchmarks, and single-image benchmarks, achieving competitive state-of-the-art performance on several key benchmarks. Our model surpasses GPT-4o on the MUIR and MVMath benchmarks and notably outperforms specialized video reasoning models on video understanding benchmarks, demonstrating the effectiveness and generalizability of our human cognition-inspired reasoning framework.
BibTeX
@misc{yin2026reasoning,
author = {Jianghao Yin and Qingbin Li and KUN SUN and chengding and Jie Wang and Qin Chen and Jie Zhou and Nan Wang and Changqing Li and wupei and Jian Xu and Zheming Yang and Liang He},
title = {{Reasoning Like Humans: Enhancing Multi-Image Reasoning Via Cognition-Inspired Meta-Action Framework}},
year = {2026},
url = {https://openreview.net/forum?id=l4adVGdetr},
howpublished = {OpenReview},
note = {Submitted to ICLR 2026},
}Loading related content...