Modality following serves as the capacity of multimodal large language models (MLLMs) to selectively utilize multimodal contexts based on user instructions. It is fundamental to ensuring safety and reliability in real-world deployments. However, the underlying mechanisms governing this decision-making process remain poorly understood. In this paper, we investigate its working mechanism through an information flow lens. Our findings reveal that instruction tokens function as structural anchors for modality arbitration: Shallow attention layers perform non-selective information transfer, routing multimodal cues to these anchors as a latent buffer; Modality competition is resolved within deep attention layers guided by the instruction intent, while MLP layers exhibit semantic inertia, acting as an adversarial force. Furthermore, we identify a sparse set of specialized attention heads that drive this arbitration. Causal interventions demonstrate that manipulating a mere $5\%$ of these critical heads can decrease the modality-following ratio by $60\%$ through blocking, or increase it by $60\%$ through targeted amplification of failed samples. Our work provides a substantial step toward model transparency and offers a principled framework for the orchestration of multimodal information in MLLMs.

翻译：模态跟随能力是多模态大语言模型（MLLMs）根据用户指令选择性利用多模态上下文的核心功能，对于确保实际部署中的安全性与可靠性至关重要。然而，支配这一决策过程的底层机制仍不甚明晰。本文通过信息流视角探究其工作机制。研究发现，指令标记在模态仲裁中发挥着结构性锚点作用：浅层注意力层执行非选择性信息传递，将多模态线索路由至这些锚点作为潜在缓冲区；深层注意力层在指令意图引导下解决模态竞争，而MLP层则表现出语义惯性，充当对抗性力量。此外，我们识别出驱动该仲裁过程的稀疏化专用注意力头集合。因果干预实验表明，仅操纵这些关键注意力头中的$5\%$，即可通过阻断机制使模态跟随率降低$60\%$，或通过对失败样本的定向增强使其提升$60\%$。本研究为提升模型透明度迈出重要一步，并为MLLMs中多模态信息的协调提供了原理性框架。