Mind or Message? Auditing Theory of Mind in Multi-Agent Social Simulation
作者: Cong Li, Cheng Chen, Thomas Fung, Alex Rossi, Yi Li
分类: cs.LG
发布日期: 2026-09-21
💡 一句话要点
通过多代理社会模拟审计心智理论以解决理解不足问题
🎯 匹配领域: 支柱六:视频提取与匹配 (Video Extraction)
关键词: 多代理系统 社会模拟 心智理论 语言模型 反事实探测 谈判策略 人机交互
📋 核心要点
- 现有的语言模型代理在模拟社会互动时,虽然表现出良好的交流能力,但未能有效理解合作伙伴的真实意图和心理状态。
- 本文提出了一种新的社会模拟框架,通过构建已知偏好权重的多议题谈判,系统性地审计代理的心智理解能力。
- 实验结果表明,尽管代理在对话中表现出高达96.2%的协议达成率,但在识别双方利益一致性方面存在显著不足,揭示了其理解的局限性。
📝 摘要(中文)
随着语言模型代理在社会互动模拟中的应用日益增加,产生的对话记录似乎表明代理之间相互理解。本文探讨这种表象是否基于对合作伙伴心智的模型,还是仅仅依赖于合作伙伴所说内容的表面记录。研究构建了一个社会模拟,包含40个多议题谈判,隐藏的偏好权重和完整的帕累托前沿已知。通过160对代理进行谈判,记录在任何测量之前被冻结,并进行2880个反事实探测,逐步调整不同因素。结果显示,代理在96.2%的对话中达成协议,但仅有0.7%的交易落在帕累托前沿,且76.6%的交易未能识别双方利益完全一致的问题。探测结果揭示了代理在理解对话内容方面的能力远超对心智的理解。
🔬 方法详解
问题定义:本文旨在解决语言模型代理在多代理社会模拟中对合作伙伴心智理解不足的问题。现有方法往往依赖于表面对话记录,缺乏对真实意图的深入理解。
核心思路:研究通过构建一个包含已知偏好权重的多议题谈判环境,系统地审计代理的心智理解能力,探讨其理解的来源是基于对话内容还是对心智的推测。
技术框架:整体架构包括40个多议题谈判,160对代理进行交互,记录在任何测量之前被冻结,并通过2880个反事实探测逐步调整不同因素,如自身利益、合作伙伴的语气等。
关键创新:最重要的创新在于通过反事实探测方法揭示了代理在理解对话内容与心智状态之间的差异,特别是其在自我中心投射方面的表现。
关键设计:研究中设置了多个参数,包括代理的利益表、语气重写和身份标签等,确保在探测过程中保持其他因素不变,以准确评估不同因素对理解的影响。
🖼️ 关键图片
📊 实验亮点
实验结果显示,代理在96.2%的对话中达成协议,但仅有0.7%的交易落在帕累托前沿,且76.6%的交易未能识别双方利益一致性的问题。探测结果表明,代理在预测合作伙伴对自身的看法时的准确率为72.5%,而合作伙伴对其的看法正确率仅为51.2%。
🎯 应用场景
该研究的潜在应用领域包括智能对话系统、社交机器人和多代理系统的设计与优化。通过深入理解代理的心智模型,可以提升人机交互的自然性和有效性,推动智能系统在复杂社交场景中的应用。未来,这一研究可能对社交网络分析和人类行为预测等领域产生深远影响。
📄 摘要(原文)
Language model agents are increasingly used to simulate social interaction, and the resulting transcripts read as though the agents understand one another. We ask whether that appearance rests on a model of the partner's mind or on the surface record of what the partner said. We build a social simulation in which both questions have exact answers: 40 multi-issue negotiations whose hidden preference weights and whose full Pareto frontier are known by construction. Two model families negotiate across 160 dyads, every transcript is frozen before any measurement, and 2880 counterfactual probes then hold the evidence byte identical while moving one factor at a time: the reader's own stake, the partner's tone, an identity label, and the order of recursion. The agents are socially fluent and economically poor. They reach agreement in 96.2% of dyads with 0 protocol failures, yet only 0.7% of deals land on the Pareto frontier, they leave 20.5% of the available joint value unclaimed, and they miss the one issue on which their interests are perfectly aligned in 76.6% of deals; on the frontier and on that aligned issue, a package drawn at random from the set both sides would accept does as well. The probes locate the failure. Swapping only the reader's own payoff sheet, while the partner's words and offers stay identical, moves the inferred top priority by 15.0 percentage points, which is egocentric projection rather than inference, while a tone rewrite moves it by 5.3 percentage points and an identity label by 0.0. Most tellingly, an agent predicts what its partner believes about it 72.5% of the time while that partner's belief is itself correct only 51.2% of the time: the agents track the conversation far better than they track the mind behind it.