🤖 AI Summary
This work addresses the limitations of existing conversational agents, whose memory systems typically rely on passive, one-shot retrieval mechanisms that decouple reasoning from memory and fail to align memory structure with navigational needs. To bridge this gap, the authors propose a Memory-as-Cognition paradigm, which constructs a navigable memory repository grounded in an associative graph structure. This framework features a multi-step reasoning–driven, cross-dimensional traversal interface and an active memory triggering mechanism, enabling deep integration between memory and cognition. The study introduces ProactiveMemBench, the first benchmark for evaluating proactive memory capabilities, and demonstrates state-of-the-art performance on LoCoMo (92.98) and LongMemEval (95.8), significantly outperforming existing approaches on ProactiveMemBench.
📝 Abstract
Existing agent memory systems universally follow what we term a Memory-as-Tool paradigm where a single query triggers one-shot retrieval of flat passage lists, suffering from passive invocation, reasoning-retrieval decoupling, and structural mismatch between retrieved fragments and the agent's navigational needs. We propose MemCog, a Memory-as-Cognition system that makes memory access an integral part of the reasoning process. MemCog organizes user knowledge as Navigable Memory Store with associative link graphs, exposes Cross-Dimensional Navigation Interface for multi-step reasoning-driven traversal, and employs Proactive Reasoning Protocol that drives agents to spontaneously initiate memory exploration from conversational context. We additionally construct ProactiveMemBench, the first benchmark for evaluating proactive memory triggering. Experiments show that MemCog achieves state-of-the-art on passive QA benchmarks (92.98 on LoCoMo, 95.8 on LongMemEval) while substantially outperforming baselines on ProactiveMemBench, demonstrating the advantage of Memory-as-Cognition.