Memory Control Signals Emerge Before Action in Long Horizon Agents

2026-09-22

AI Summary

研究解决了长时序代理中记忆控制信号问题，通过分析模型内部表示发现压缩和回忆需求，并提出PaMER方法结合状态引导压缩与外部证据检索来优化。

Abstract

Long horizon language model agents continuously accumulate interaction history, increasing computational cost while making relevant information harder to preserve and reuse. Existing context management methods mainly focus on how to compress or retrieve history, but largely leave open whether the model itself already represents the need for these memory operations before they occur. We study the hidden state immediately before each agent action and find that compression and recall needs are already encoded in the model's internal representations. These signals cannot be explained by simple context length or interaction progress, and they exhibit distinct formation patterns across model depth. We further show that most memory decision information is preserved in a compact recent context, while selectively restored historical evidence complements the long range dependencies that recent context misses. Based on these findings, we propose Preaction Memory with Evidence Retrieval (PaMER), which combines state guided compression with external evidence retrieval. PaMER+ further introduces step level evidence selection to recover only the historical information required by the current task. Experiments on WorkBuddyBench, across multiple context management baselines and model backbones, show that our framework substantially reduces context consumption while maintaining competitive task performance.

Problem

• long horizon agents
• memory control signals
• context management

Innovation

• Preaction Memory
• Evidence Retrieval
• State Guided Compression