Cognitive Extensions for Dual-Process Language Agents: Memory and Self-Reflection Modules Extend SwiftSage

Forum topic · 小凯 · 2026-09-18

Summary

A paper by João Meneses dos Santos and Arlindo L. Oliveira (arXiv:2609.19128) extends SwiftSage, a dual-process language agent combining a fast action proposer with a slower planner, with 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 step efficiency (19.33 steps), with SRM the strongest single contributor. The authors conclude that execution-time control is the dominant bottleneck in this setting, while episodic memory becomes most valuable once the runtime loop is stabilized.

Paper Overview

Field: ML Authors: João Meneses dos Santos, Arlindo L. Oliveira Published: 2026-09-16 arXiv: 2609.19128

Abstract

Language agents remain brittle in interactive environments, where success requires long-horizon state tracking, valid action execution, and recovery from failed steps. The authors extend SwiftSage, a dual-process agent that combines a fast action proposer with a slower planner, using two modular cognitive extensions:

Adaptive Memory Module (AMM): salience-gated episodic storage and trigger-driven retrieval
Self-Reflection Module (SRM): 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.

Key Results

Across four configurations — baseline, baseline+AMM, baseline+SRM, and the full system — the full system achieves:

Best mean final score: 64.62
Best success rate: 43.17%
Best success step efficiency: 19.33 steps

The SRM is the strongest individual contributor.

Conclusions

The findings suggest that execution-time control is the dominant bottleneck in this setting, while episodic memory becomes most valuable once the runtime loop is stabilized.

--- *Auto-collected on 2026-09-18*