Controllable Memory Usage: Balancing Anchoring and Innovation in Long-Term Human-Agent Interaction

arXiv:2601.05107v2 Announce Type: replace Abstract: As LLM-based agents are increasingly used in long-term interactions, cumulative memory is critical for enabling personalization and maintaining stylistic consistency. However, most existing systems adopt an ``all-or-nothing'' approach to memory usage: incorporating all relevant past information can lead to extit{Memory Anchoring}, where the agent is trapped by past interactions, while excluding memory entirely results in under-utilization and the loss of important interaction history. We show that an agent's reliance on memory can be modeled as an explicit and user-controllable dimension. We first introduce a behavioral metric of memory dependence to quantify the influence of past interactions on current outputs. We then propose extbf{Stee}rable extbf{M}emory Agent, exttt{SteeM}, a framework that allows users to dynamically regulate memory reliance, ranging from a fresh-start mode that promotes innovation to a high-fidelity mode that closely follows interaction history. Experiments across different scenarios demonstrate that our approach consistently outperforms conventional prompting and rigid memory masking strategies, yielding a more nuanced and effective control for personalized human-agent collaboration.

Related

• AMA: Adaptive Memory via Multi-Agent Collaboration
• MindMemOS: A Portable and Self-Evolving Memory Operating Layer for AI Agents
• RecMem: Recurrence-based Memory Consolidation for Efficient and Effective Long-Running LLM Agents
• The Retriever Should Remember: Experience-Amortized Reranking for Long-Term Agent Memory
• A Multi-Memory Segment System for Generating High-Quality Long-Term Memory Content in Agents

Source: arXiv cs.AI | 2026-09-01