Explaining Cognition and Communication in Cognitive Robots
Proceedings of the AAAI Symposium Series · 06.11.2026 · 10 (1) · 174-181
DOI: 10.1609/aaaiss.v10i1.43404 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Symposium Series, online erschienen am 06.11.2026. Verfasst von Marjorie McShane und 4 weiteren Autor:innen.
Abstract (Originalsprache)
Cognitive robots of the class called LEIAs can explain their cognition and action using a language generation process that is, itself, explainable. This transparency contrasts sharply with the opaque operation of systems whose cognition relies on language models. This paper describes recent advances in the neurosymbolic modeling of LEIAs at the intersection of metacognition, language generation, and social agency. It answers the question: How can we prepare robots to produce the kinds of explanations that people will want using the most generalizable, best-engineered, scalable solutions? We argue that the symbolic modeling of cognition, including the process of language generation itself, is both realistic and scalable when guided by strong methodology, clear objectives, and the support of modern technology — including language models, but used only for carefully selected tasks that are resistant to hallucinations.
Zitieren
APA
McShane, M., English, J., Arndt, C., Nirenburg, S., & Oruganti, S. (2026). Explaining Cognition and Communication in Cognitive Robots. Proceedings of the AAAI Symposium Series, 10(1), 174-181. https://doi.org/10.1609/aaaiss.v10i1.43404
Vancouver
McShane M, English J, Arndt C, Nirenburg S, Oruganti S. Explaining Cognition and Communication in Cognitive Robots. Proceedings of the AAAI Symposium Series. 2026;10(1):174-181. doi:10.1609/aaaiss.v10i1.43404
BibTeX
@article{McShane2026,
author = {McShane, Marjorie and English, Jesse and Arndt, Christian and Nirenburg, Sergei and Oruganti, Sanjay},
title = {Explaining Cognition and Communication in Cognitive Robots},
journal = {Proceedings of the AAAI Symposium Series},
year = {2026},
volume = {10},
number = {1},
pages = {174--181},
doi = {10.1609/aaaiss.v10i1.43404}
}
EUVIMED