Visual Thoughts: A Unified Perspective of Understanding Multimodal Chain-of-Thought

TL;DR

This paper investigates how visual thoughts enhance multimodal chain-of-thought reasoning in large vision-language models, revealing their role as intermediaries that improve interpretability and performance across different MCoT formats.

Contribution

It systematically analyzes the forms and functions of visual thoughts, clarifies their mechanism in boosting MCoT, and introduces a comprehensive framework for understanding visual thought expressions.

Findings

Visual thoughts improve MCoT performance regardless of format.

Four distinct visual thought expressions differ in clarity and effectiveness.

Visual thoughts act as intermediaries transmitting visual information to deeper model layers.

Abstract

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.