语言模型中符号接地机制的涌现
The Mechanistic Emergence of Symbol Grounding in Language Models
AI总结：
通过机械因果分析，发现符号接地在语言模型的中层计算中通过注意力头聚合环境信息实现，并在多模态对话和多种架构中复现。
摘要：
符号接地（Harnad, 1990）描述了词语等符号如何通过连接真实世界的感知运动经验来获得意义。最近的研究初步表明，在大规模训练且未使用显式接地目标的（视觉-）语言模型中，接地可能涌现。然而，这种涌现的具体位置及其驱动机制仍 largely 未被探索。为解决这一问题，我们引入了一个受控评估框架，通过机械和因果分析系统地追踪符号接地如何在内部计算中产生。我们的发现表明，接地集中在中层计算中，并通过聚合机制实现，其中注意力头聚合环境接地以支持语言形式的预测。这种现象在多模态对话和跨架构（Transformer 和状态空间模型）中复现，但在单向 LSTM 中未出现。我们的结果提供了行为和机械证据，表明符号接地可以在语言模型中涌现，并对预测和潜在控制生成的可靠性具有实际意义。
英文摘要：
Symbol grounding (Harnad, 1990) describes how symbols such as words acquire their meanings by connecting to real-world sensorimotor experiences. Recent work has shown preliminary evidence that grounding may emerge in (vision-)language models trained at scale without using explicit grounding objectives. Yet, the specific loci of this emergence and the mechanisms that drive it remain largely unexplored. To address this problem, we introduce a controlled evaluation framework that systematically traces how symbol grounding arises within the internal computations through mechanistic and causal analysis. Our findings show that grounding concentrates in middle-layer computations and is implemented through the aggregate mechanism, where attention heads aggregate the environmental ground to support the prediction of linguistic forms. This phenomenon replicates in multimodal dialogue and across architectures (Transformers and state-space models), but not in unidirectional LSTMs. Our results provide behavioral and mechanistic evidence that symbol grounding can emerge in language models, with practical implications for predicting and potentially controlling the reliability of generation.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
