Visual Thoughts: A Unified Perspective of Understanding Multimodal Chain-of-Thought
This work provides insights into MCoT mechanisms for researchers in multimodal AI, though it is incremental as it builds on existing MCoT methods without introducing a new paradigm.
The paper tackles the lack of understanding in how multimodal chain-of-thought (MCoT) improves large vision-language models by revealing that visual thoughts, which convey image information to reasoning, drive these gains based on their clarity and conciseness, with analysis showing varying levels of improvement across different forms.
Large Vision-Language Models (LVLMs) have achieved significant success in multimodal tasks, with multimodal chain-of-thought (MCoT) further enhancing performance and interpretability. Recent MCoT methods fall into two categories: (i) Textual-MCoT (T-MCoT), which takes multimodal input and produces textual output; and (ii) Interleaved-MCoT (I-MCoT), which generates interleaved image-text outputs. Despite advances in both approaches, the mechanisms driving these improvements are not fully understood. To fill this gap, we first reveal that MCoT boosts LVLMs by incorporating visual thoughts, which convey image information to the reasoning process regardless of the MCoT format, depending only on clarity and conciseness of expression. Furthermore, to explore visual thoughts systematically, we define four distinct forms of visual thought expressions and analyze them comprehensively. Our findings demonstrate that these forms differ in clarity and conciseness, yielding varying levels of MCoT improvement. Additionally, we explore the internal nature of visual thoughts, finding that visual thoughts serve as intermediaries between the input image and reasoning to deeper transformer layers, enabling more advanced visual information transmission. We hope that the visual thoughts can inspire further breakthroughs for future MCoT research.