Cognitive Extensions for Dual-Process Language Agents: Memory and Self-Reflection in Interactive Environments

arXiv:2609.19128v1 Announce Type: new Abstract: Language agents remain brittle in interactive environments, where success requires long-horizon state tracking, valid action execution, and recovery from failed steps. We extend SwiftSage, a dual-process agent that combines a fast action proposer with a slower planner, using two modular cognitive extensions: an Adaptive Memory Module (AMM) for salience-gated episodic storage and trigger-driven retrieval, and a Self-Reflection Module (SRM) for bounded execution-time validation and corrective intervention. Both modules are implemented as feature-flagged extensions over the same execution substrate, enabling controlled ablations on ScienceWorld. Across four configurations---baseline, baseline+AMM, baseline+SRM, and the full system---the full system achieves the best mean final score (64.62), success rate (43.17%), and successful-step efficiency (19.33 steps), while SRM is the strongest standalone contributor. The results suggest that execution-time control is the dominant bottleneck in this setting, while episodic memory becomes most useful once the runtime loop is stabilized.

Related

• Beyond Semantic Organization: Memory as Execution State Management for Long-Horizon Agents
• AdMem: Advanced Memory for Task-solving Agents
• SAM: State-Adaptive Memory for Long-Horizon Reasoning Agent
• MemToolAgent overview with a simple restaurant booking scenario where the agent retrieves similar memories, receives feedback on an invalid time format, and generates a reflection to update its memory
• From Cognitive Architectures to Language Agents: A Mechanism-Level Review of Lineage, Convergence, and Migration Gaps

Source: arXiv cs.AI | 2026-09-17