From Impasse to Adaptation: An Integrated LLM Supporting Meta-Cognition in a Cognitive Architecture
Proceedings of the AAAI Symposium Series · 06.11.2026 · 10 (1) · 123-132
DOI: 10.1609/aaaiss.v10i1.43399 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Symposium Series, online erschienen am 06.11.2026. Verfasst von Amitis Hamidi und 1 weiteren Autor:innen.
Abstract (Originalsprache)
Autonomous cognitive systems deployed in real-world settings will inevitably encounter unknown and novel scenarios that require knowledge acquisition through exploration, adaptation, and recovery. We argue that successfully navigating such scenarios will require meta-cognition and that meta-cognitive capabilities can be realized in a cognitive architecture by integrating a large language model (LLM) as the engine for generating potential solutions when the agent encounters impasses during task performance that its regular planning and reasoning capabilities cannot solve. We propose a particular such integration that is focused on enabling three meta-cognitive capacities: performance prediction, model introspection, and corrective replanning. We provide a proof-of-concept demonstration of the operation of the integrated architecture on an autonomous robot, showing how it can solve a problem that it would otherwise not have been able to tackle.
Zitieren
APA
Hamidi, A., & Scheutz, M. (2026). From Impasse to Adaptation: An Integrated LLM Supporting Meta-Cognition in a Cognitive Architecture. Proceedings of the AAAI Symposium Series, 10(1), 123-132. https://doi.org/10.1609/aaaiss.v10i1.43399
Vancouver
Hamidi A, Scheutz M. From Impasse to Adaptation: An Integrated LLM Supporting Meta-Cognition in a Cognitive Architecture. Proceedings of the AAAI Symposium Series. 2026;10(1):123-132. doi:10.1609/aaaiss.v10i1.43399
BibTeX
@article{Hamidi2026,
author = {Hamidi, Amitis and Scheutz, Matthias},
title = {From Impasse to Adaptation: An Integrated LLM Supporting Meta-Cognition in a Cognitive Architecture},
journal = {Proceedings of the AAAI Symposium Series},
year = {2026},
volume = {10},
number = {1},
pages = {123--132},
doi = {10.1609/aaaiss.v10i1.43399}
}
EUVIMED