arXiv cs.AI· Ido Haber, Giulio Tononi·· 12 小时前AI 评分34
智能体需要睡眠吗？论文提出用睡眠机制解决 LLM 智能体记忆漂移问题
Do agents need to sleep?
AI 导读
一篇 arXiv 论文提出，持续跨会话运行、通过交互积累记忆的 LLM 智能体已出现语义漂移、检索精度下降、存储错误传播和记忆容量饱和等症状，并借鉴睡眠的"振铃、共振与重整化"（RRR）选择主义理论，将睡眠三阶段映射为数字记忆架构中可执行的操作。作者由此给出可证伪预测：随着记忆存储增大，选择主义式整合将优于指令主义方案。核心主张是，在有限容量下持续学习并与环境耦合的系统，其可塑性代价就是睡眠。
正文 · 原文
Abstract:LLM-based agents, which increasingly operate persistently across sessions and accumulate memory through interaction, already exhibit the familiar symptoms of unchecked plasticity: semantic drift, degraded retrieval precision, propagation of stored errors, and saturation of memory capacity. Drawing on the selectionist "ringing, resonance and renormalization" (RRR) account of sleep, we propose that each of its three phases maps onto an operation feasible in digital memory architectures. Engineers building stateful agents, meanwhile, are independently rediscovering the problems sleep evolved to solve: what to forget, and when to do the forgetting. From this we derive the falsifiable prediction that selectionist consolidation will outperform instructionist alternatives as memory stores grow large. The core claim concerns the nature of continuous learning itself: plasticity has a price, and for any system that learns continuously under finite capacity while coupled to its environment, that price is sleep.
| Subjects: | Neurons and Cognition (q-bio.NC); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.02216 [q-bio.NC] |
| (or arXiv:2611.02216v2 [q-bio.NC] for this version) | |
| https://doi.org/10.48550/arXiv.2611.02216 arXiv-issued DOI via DataCite |
Submission history
From: Ido Haber [view email]
[v1]
Sun, 1 Nov 2026 21:24:59 UTC (235 KB)
[v2]
Tue, 3 Nov 2026 17:52:05 UTC (235 KB)
来源：arXiv cs.AI · arxiv.org