发表机构
China University of Geosciences (Beijing); Northwestern Polytechnical University; The Hong Kong University of Science and Technology; Chinese Academy of Sciences; The Hong Kong Polytechnic University; École polytechnique fédérale de Lausanne; National University of Singapore（中国地质大学（北京）; 西北工业大学; 香港科技大学; 中国科学院; 香港理工大学; 洛桑联邦理工学院; 新加坡国立大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究提出可控记忆干扰（CMI）框架，揭示记忆进化受经验相互作用影响，为持续大型语言模型智能体的记忆系统提供诊断与学习方法，提升记忆更新区分度与任务性能。
AI 中文摘要
长期记忆使AI智能体能够跨会话保持连续性、个性化行为，并通过积累的经验进化。然而记忆进化并非简单地存储更多信息的过程：新经验可能强化、修正或干扰现有记忆状态。现有系统主要强调记忆构建和基于相关性的检索，但部分记忆可能同时具有相关性，却在状态、时间有效性或权威性上存在差异。我们提出可控记忆干扰（Controlled Memory Interference，CMI），这是一种用于研究智能体记忆在不同记忆关系下如何进化的可控诊断与数据生成框架。在受控记忆进化过程中，良性积累的影响有限，而特定关系的干扰会通过阻断目标记忆暴露或破坏其下游使用，显著抑制更新可塑性且几乎无稳定性增益。词汇检索和密集检索表现出不同的干扰路径，且中毒对更新权威线索的敏感性高于单纯的新近性。除诊断外，CMI还为感知干扰的记忆学习提供针对性示例，提升有效更新与干扰诱导记忆的区分度，同时保留原始记忆任务的性能。这些发现表明，记忆进化不仅受记忆规模影响，还受积累经验间的相互作用塑造。更广泛而言，记忆干扰是可靠持续智能体记忆系统的重要影响因素。
英文摘要
Long-term memory enables AI agents to maintain continuity across sessions, personalize behavior, and evolve through accumulated experience. Yet memory evolution is not simply a process of storing more information: new experiences may reinforce, revise, or interfere with existing memory states. Existing systems mainly emphasize memory construction and relevance-based retrieval, but several memories may remain simultaneously relevant while differing in state, temporal validity, or authority. We introduce Controlled Memory Interference (CMI), a controlled diagnostic and data-generation framework for studying how agent memory evolves under different memory relationships. Across controlled memory evolution, benign accumulation has limited effects, whereas relationship-specific interference sharply suppresses update plasticity with little stability gain, either by blocking target-memory exposure or by disrupting its downstream use. Lexical and Dense retrieval exhibit distinct interference pathways, while poisoning is more sensitive to update-authority cues than to recency alone. Beyond diagnosis, CMI provides targeted examples for interference-aware memory learning, improving the distinction between valid updates and interference-inducing memories while preserving performance on original memory tasks. These findings show that memory evolution is shaped not only by memory scale, but also by interactions among accumulated experiences. More broadly, memory interference emerges as an important factor for reliable continual agent memory systems.
