# MemTrapBench：为 LLM 记忆使用中的认知陷阱设立基准

- 来源：HuggingFace Daily Papers（社区热门论文）
- 发布时间：2026-08-20 08:00
- AIHOT 分数：57
- AIHOT 链接：https://aihot.virxact.com/items/cmt2daxf802ejroc9phpowlbz
- 原文链接：https://arxiv.org/abs/2608.20202

## AI 摘要

现有记忆基准多只评估信息提取与检索，却忽视了检索到的记忆如何重塑模型推理并影响当前任务表现。为此研究者推出 MemTrapBench，覆盖推理固着与信念扭曲两类认知陷阱。实验显示，所有被测记忆策略均不如无记忆设置，最强方法性能下降也超过 10%；新提出的 AdaptiveMem 推理期方法可缓解这些陷阱，同时保持或提升标准记忆基准上的表现。

## 正文

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.