Decoupling Thought from Speech: Knowledge-Grounded Counterfactual Reasoning for Resilient Multi-Agent Argumentation
作者: Jakub Masłowski, Jarosław A. Chudziak
分类: cs.MA, cs.AI, cs.CL
发布日期: 2026-06-09
备注: Accepted for publication in the Proceedings of the 30th International Conference on Knowledge-Based and Intelligent Information & Engineering Systems (KES 2026)
💡 一句话要点
提出知识基础的反事实推理以解决多智能体辩论中的稳定性问题
🎯 匹配领域: 支柱九：具身大模型 (Embodied Foundation Models)
关键词: 多智能体辩论 反事实推理 系统稳定性 动态资源分配 知识基础 逻辑一致性 论点质量
📋 核心要点
- 现有多智能体辩论框架在长时间交流中容易出现逻辑退化和角色漂移，影响系统稳定性。
- 提出知识基础的反事实推理（KG-CFR），通过双阶段架构分离规划和执行，增强系统的稳定性。
- KG-CFR在动态资源分配环境中表现优异，论点质量显著提升，且在95%以上的扰动情况下保持稳定。
📝 摘要（中文）
多智能体辩论框架已被证明能提升大型语言模型在收敛任务中的表现，但目前的优化方式过于偏重最终输出的准确性，而忽视了过程的稳定性。在长时间的交流中，反应系统在持续扰动下常常会出现逻辑退化、论点重复和角色漂移。为结构性地防止身份丧失并维持过程的保真性，本文提出了知识基础的反事实推理（KG-CFR），一种双阶段架构，严格区分了私有的检索增强规划缓冲区和公共执行层。在不确定的动态资源分配环境中，KG-CFR在超过95%的扰动运行中防止了评判者检测到的关键后震退化，论点质量从0.694提升至0.822。
🔬 方法详解
问题定义：本文旨在解决多智能体辩论中因持续扰动导致的逻辑退化和角色漂移问题，现有方法在长时间交流中表现不佳。
核心思路：KG-CFR通过将规划和执行阶段严格分离，确保系统在面对扰动时能够保持稳定性和一致性。
技术框架：KG-CFR包含两个主要模块：私有的检索增强规划缓冲区和公共执行层，前者负责信息检索和计划生成，后者负责执行和论证。
关键创新：KG-CFR的核心创新在于架构的解耦设计，这一设计显著提升了系统在持续压力下的韧性，而不损失输出质量。
关键设计：在实验中，采用了自定义向量度量来评估话语的多样性和计划执行的一致性，确保系统在不同情境下的稳定性。
🖼️ 关键图片
📊 实验亮点
实验结果显示，KG-CFR在动态资源分配环境中有效防止了超过95%的扰动情况下的逻辑退化，论点质量从0.694提升至0.822，展现出显著的性能提升和系统稳定性。
🎯 应用场景
该研究的潜在应用领域包括智能对话系统、自动辩论生成和多智能体协作任务等。通过增强系统的稳定性和一致性，KG-CFR能够在复杂环境中提供更高质量的交互体验，未来可能对人机交互和智能决策支持系统产生深远影响。
📄 摘要（原文）
Multi-agent debate frameworks have been shown to improve large language model performance in convergent tasks, but they are currently optimized in a way that heavily favors final output accuracy rather than stability of the process. During long-horizon exchanges reactive systems under sustained perturbations often experience logic degradation, argument repetition, and role drift. To structurally prevent the identity loss and maintain the process fidelity, we introduce Knowledge-Grounded Counterfactual Reasoning (KG-CFR), a dual-stage architecture that enforces a strict separation of concerns between a private, retrieval-augmented planning buffer, and a public execution layer. We assess this system in Dynamic Resource Allocation under Uncertainty (DRAU), a dedicated 1v1v1 environment, introducing diversity as distinct from standard debate settings. Over 270 completely factorial crisis simulation trajectories with stochastic environmental shocks, KG-CFR prevents judge-detected critical post-shock degradation (defined as a quality shift, $Δ\le -0.20$) in more than 95% of perturbed runs, increasing the overall argument quality from 0.694 to 0.822. Our primary contribution is the demonstration of architectural decoupling being an important factor of systemic resilience enhancement under sustained pressure without quality loss. Furthermore, we introduce custom vector metrics for discourse divergence and plan-execution alignment that provide strong, directionally consistent evidence of operational stability. Our ablation experiments suggest that the proper doctrinal grounding can be an equally important factor for argument quality, as the prospective planning. KG-CFR, according to our initial metric evaluations, reduces semantic looping, by preserving the agent's consistency with the original plan.