Neural Procedural Memory: Empowering LLM Agents with Implicit Activation Steering
2026-06-29 • Computation and Language
Computation and LanguageArtificial Intelligence
AI summaryⓘ
The authors explain that while large language models work well when given fixed tasks, making them operate independently is harder because they need memory of past actions. Current methods use text instructions to guide the models, but this can confuse the models and lead to mistakes. They propose Neural Procedural Memory (NPM), which helps models remember by directly steering their internal neural activity instead of relying on text. Their tests show NPM works as well as instruction-based methods, and combining both approaches improves performance. They also found that these internal steering signals represent consistent task logic in the model's brain-like activations.
Large Language ModelsAutonomous AgentsProcedural MemoryRetrieval-Augmented GenerationActivation SteeringContrastive LearningNeural RepresentationsImplicit MemoryTask ExecutionRepresentational Analysis
Authors
Chengfeng Zhao, Yuqiao Tan, Shizhu He, Yequan Wang, Jun Zhao, Kang Liu
Abstract
While Large Language Models (LLMs) excel as static solvers, transforming them into autonomous agents remains challenging. This transition requires continuous environmental interaction, yet current agents lack the necessary persistent procedural memory. Existing approaches predominantly employ Retrieval-Augmented Generation (RAG) to inject explicit textual guidelines into model contexts. However, relying solely on symbolic instructions can introduce a text-action disconnect, frequently failing to activate the internal representations necessary for correct task execution. To address this, the paper introduces Neural Procedural Memory (NPM), a training-free framework that represents agent memory through implicit activation steering rather than explicit instructions. By distilling procedural skills from historical contrastive experiences into steering vectors in the activation space, NPM directly activates the task-relevant neural mechanisms to guide task execution. Evaluations across four agent benchmarks show that NPM performs comparably to baselines using explicit textual instructions. Furthermore, the results show that combining implicit steering with explicit workflows provides complementary advantages, leading to more robust task execution. Representational analyses indicate that these steering vectors encode consistent task logic, forming organized structures within the activation space. These findings suggest that implicit activation steering provides a promising approach for managing agent memory.