发表机构
Research Center for Advanced Science and Technology, The University of Tokyo; School of Engineering, The University of Tokyo（东京大学先进科学技术研究中心; 东京大学工学院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究通过循环同步实验发现，状态编码（低阶循环矩vs直方图）会影响语言模型智能体的同步结果，且效应因模型而异，状态编码是依赖模型的有效交互法则的组成部分。
AI 中文摘要
语言模型智能体基于其环境的状态编码行动，但这些状态编码被视为可互换的接口。我们使用预训练语言模型设计了一个循环同步实验，在保持物理系统固定的同时对状态编码进行干预：每个智能体仅能看到其邻居相对相位的摘要，并选择前进、保持或后退。将该状态编码为低阶循环矩而非直方图，会产生不同的集体结果。在GPT中，矩编码使6个随机种子的群体均实现同步，而直方图编码则0个种子实现同步；该效应在Claude中重复出现，但方向相反。在GPT、Claude和Gemini中，重放相同场会使每个智能体的前进/保持/后退概率变化远超过编码内的重复变异；在GPT中，仅呈现信息即可在矩值固定的情况下改变算子。因此，状态编码构成了依赖于模型的有效交互法则的一部分，而非中性接口。
英文摘要
Language-model agents increasingly act in populations, where the outcome that matters is collective: whether they align, split or fail to coordinate. Each acts not on the world but on a text description of it, a choice usually fixed in software. Using synchronization, the canonical probe of how interaction rules produce collective order, we show that this choice can decide the outcome. Agents on a circle chose to advance, stay or move back after reading the others' relative positions, in 507,112 valid responses across matched populations, controlled inputs and three model families. In GPT, numerical summaries aligned every matched population at both positive couplings, whereas histograms aligned none; Claude showed the reverse at the stronger coupling. Re-describing identical states shifted action probabilities in all three families, even between histograms carrying the same information. No single directional coefficient explained the outcome: state descriptions are part of the interaction rule that turns individual responses into collective order.
