AI 中文总结
研究人员提出MemTrapBench基准测试，发现现有LLM记忆策略存在认知陷阱致性能下降，进而提出AdaptiveMem方法可缓解该问题并提升标准记忆基准性能。
AI 中文摘要
记忆已成为大语言模型（LLM）的关键组成部分，使其能够保留信息并从长期交互中学习。然而，现有记忆基准主要评估信息是否被正确提取、存储和检索，却在很大程度上忽略了检索到的记忆如何重塑模型推理以及影响当前任务的性能。我们识别出记忆诱导的认知陷阱：即使是忠实地记录且语义相关的记忆，也可能扭曲模型的推理或信念，降低当前任务的性能。为了系统评估这些失效模式，我们引入了MemTrapBench，它涵盖两种形式的认知陷阱：推理固着（Reasoning Fixation）和信念扭曲（Belief Distortion）。在两个模型家族和五个代表性记忆框架上开展的实验表明，MemTrapBench具有挑战性：所有被评估的记忆策略的表现均逊于无记忆设置，即使是最强的方法也出现了超过10%的性能下降。为了缓解这些认知陷阱，我们提出了AdaptiveMem，一种简单但有效的推理时方法，用于指导LLM避免记忆陷阱。AdaptiveMem在MemTrapBench上缓解了认知陷阱，同时在不同记忆框架的标准记忆基准上保持或提升了性能。
英文摘要
Memory has become a key component of large language models, enabling them to retain information and learn from long-term interactions. However, existing memory benchmarks mainly evaluate whether information is correctly extracted, stored, and retrieved, while largely overlooking how retrieved memories reshape model reasoning and affect performance on the current task. We identify memory-induced cognitive traps: even faithfully recorded and semantically relevant memories can distort model reasoning or beliefs and degrade current task performance. To systematically evaluate these failure modes, we introduce MemTrapBench, which covers two forms of cognitive traps: Reasoning Fixation and Belief Distortion. Experiments across two model families and five representative memory frameworks show that MemTrapBench is challenging: all evaluated memory strategies underperform the no-memory setting, with even the strongest methods suffering drops of more than 10%. To mitigate these cognitive traps, we propose AdaptiveMem, a simple yet effective inference-time method that instructs LLMs to avoid memory traps. AdaptiveMem mitigates cognitive traps on MemTrapBench while preserving or improving performance on standard memory benchmarks across diverse memory frameworks.
CommentsWork in progress
