Modeling reflexive processes as information processing tasks in intelligent computers

Context and relevance. Reflexive processes, that is, the ability of a system to reflect on its own actions and states, have long been studied in philosophy, psychology, and cybernetics. In modern intelligent systems, there is a need to model reflection to improve the adaptability, autonomy, and explainability of their behavior. Objective. To develop and explore a cognitive architecture with explicit support for reflective thinking, considering reflection as a special information processing task within artificial intelligence. Hypothesis. The introduction of a reflection layer (metacognitive control) into the architecture of an intelligent system will allow it to identify and correct its own errors, adapt to uncertainty, and justify its decisions better than a system without such a layer. Methods and materials. A hybrid cognitive architecture was constructed, including a reactive object layer (S1) and a reflexive planning layer (S2) with a metacontroller that evaluates the confidence of decisions and monitors system resources. Scenarios of agent interaction with the external environment were simulated, in which the behavior of the system with reflection enabled and disabled was compared. Results. It was demonstrated that the metacontroller initiates a reflexive cycle when uncertainty about a decision is insufficient or when anomalies are detected, leading to a revision of the agent's goals and plans. In test tasks, the reflexive agent successfully avoided pitfalls and corrected erroneous actions, whereas the non-reflexive agent made more errors. Conclusions. Reflexive processes can be effectively formulated as an internal information processing task in a cognitive system. Modeling them within the architecture of an intelligent agent increases the reliability and explainability of its operation, bringing the system's behavior closer to conscious and goal-oriented, confirming the proposed hypothesis.

1. Анохин, П.К. (1955). Особенности афферентного аппарата условного рефлекса и их значение для психологии. Вопросы психологии, 6, 16—38. http://lib.mgppu.ru/opacunicode/index.php?url=/notices/index/IdNotice:23835/Source:default (дата обращения: 20.01.2026)
   Anokhin, P.K. (1955). Features of the afferent apparatus of the conditioned reflex and their significance for psychology. Psychology issues, 6, 16—38. http://lib.mgppu.ru/opacunicode/index.php?url=/notices/index/IdNotice:23835/Source:default (дата обращения: 20.01.2026)
2. Гапонов, В.Р., Ляпунцова, Е.В., (2024). Использование когнитивных моделей с целью повышения качества образовательных технологий. Известия тульского государственного университета. Технические науки, (10), 397— https://doi.org/10.24412/2071-6168-2024-10-397-398
   Gaponov, V.R., Lyapuntsova, E.V., (2024). Using cognitive models to improve the quality of educational technologies. Bulletin of the Tula State University. Technical sciences, (10), 397–400. https://doi.org/10.24412/2071-6168-2024-10-397-398
3. Лефевр, В.А. (1973). Конфликтующие структуры. М.: Советское радио.
   Lefebvre, V.A. (1973). Conflicting structures. Moscow: Soviet Radio
4. Маркова, Г.М., Барцев, С.И., (2025) Предельно просто не значит предельно ясно: некоторые контринтуитивные результаты нейросетевого моделирования рефлексии. Информационные и математические технологии в науке и управлении, (1), 5— https://doi.org/10.25729/ESI.2025.37.1.001
   Markova, G.M., Bartsev, S.I., (2025) Extremely simple does not mean extremely clear: some counterintuitive results of neural network modeling of reflection. Information and Mathematical Technologies in Science and Management, (1), 5–15. (In Russ) https://doi.org/10.25729/ESI.2025.37.1.001
5. Нагоев, З.В., Нагоева, О.В., Пшенокова, И.А., (2017) Формальная модель семантики естественно-языковых высказываний на основе мультиагентных рекурсивных когнитивных архитектур. Известия Кабардино-Балкарского научного центра РАН, (4), 19—31. https://www.mathnet.ru/php/archive.phtml?wshow=paper&jrnid=izkab&paperid=160&option_lang=rus (дата обращения 28.01.2026)
   Nagoev, Z.V., Nagoeva, O.V., Pshenokova, I.A., (2017) A Formal Model of Semantics of Natural-Language Statements Based on Multi-Agent Recursive Cognitive Architectures. Bulletin of the Kabardino-Balkarian Scientific Center of the Russian Academy of Sciences, (4), 19–31. https://www.mathnet.ru/php/archive.phtml?wshow=paper&jrnid=izkab&paperid=160&option_lang=rus (Accessed 28.01.2026)
6. Разумникова, О.М. (2023). Когнитивная архитектура познавательной деятельности при ее моделировании и психофизиологической оценке. Успехи физиологических наук, 54(3), 90— https://doi.org/10.31857/S0301179823030074
   Razumnikova, O.M. (2023). Cognitive Architecture of Cognitive Activity: Modeling and Psychophysiological Assessment. Achievements of Physiological Sciences, 54(3), 90—104. (In Russ). https://doi.org/10.31857/S0301179823030074
7. Сизикова, Т.Э., Волошина, Т.В., Повещенко, А.Ф. (2016). Обзор исследований рефлексии в психологии. Педагогическая рефлексия. Научное обозрение. Педагогические науки. (2), 89—102. https://science-pedagogy.ru/ru/article/view?id=1488 (дата обращения 29.01.2026).
   Sizikova, T.E., Voloshina, T.V., Poveshchenko, A.F. (2016). Review of research on reflection in psychology. Pedagogical reflection. Scientific review. Pedagogical sciences. (2), 89–102. https://science-pedagogy.ru/ru/article/view?id=1488 (viewed 29.01.2026).
8. Симакин, А.Е. (2024). Эволюция когнитивных процессов и влияние дуального мышления в современной психологии. В: О взаимоотношениях психического и телесного в норме и патологии: Сборник научных трудов по материалам национальной научно-практической конференции (с. 314—324). Санкт-Петербург: Восточно-Европейский институт психоанализа. https://elibrary.ru/item.asp?id=65428980&pff=1 (дата обращения 28.01.2026)
   Simakin, A.E. (2024). The evolution of cognitive processes and the influence of dual thinking in modern psychology. In: On the relationship between the mental and the physical in norm and pathology: Collection of scientific papers based on the materials of the national scientific and practical conference (pp. 314–324). St. Petersburg: East European Institute of Psychoanalysis. (In Russ). https://elibrary.ru/item.asp?id=65428980&pff=1 (viewed 28.01.2026)
9. Ackerman, R., Thompson, V. (2017). Meta-Reasoning: Monitoring and Control of Thinking and Reasoning. Trends in Cognitive Sciences, 21(8), 1—21. https://doi.org/10.1016/j.tics.2017.05.
10. Dong, D., Franklin, St. (2014). The Action Execution Process Implemented in Different Cognitive Architectures: A Review. Journal of Artificial General Intelligence, 5(1), 47—66. https://doi.org/10.2478/jagi-2014-0002.
11. Kahneman, D. (2011). Thinking, Fast and Slow. New York: Farrar, Straus and Giroux.
12. Lepskiy, V. (2021). Reflexivity and Artificial Intelligence in Control (Subjectness-oriented Approach). IFAC-PapersOnLine, 54(13), 221—226. https://doi.org/10.1016/j.ifacol.2021.10.449
13. Lewis, R.W., Sarkadi, S. (2024). Reflective Artificial Intelligence. Minds and Machines, 34(14), 1—30. https://doi.org/10.1007/s11023-024-09664-2
14. Listopad, S., Matsoula, V., Luchko, A. (2024). Modeling reflection in artificial intelligence systems: state of art and prospects. In: ITM Web of Conferences, Volume 59: II International Workshop “Hybrid Methods of Modeling and Optimization in Complex Systems” (HMMOCS-II 2023) (pp. 1–7). St. Petersburg: East European Institute of Psychoanalysis. https://doi.org/10.1051/itmconf/20245904005
15. Novikov, D.A., Chkhartishvili, A.G., (2014). Mathematical Models of Informational and Strategic Reflexion: a Survey. Advances in Systems Science and Applications, 14(3), 254—278. https://ijassa.ipu.ru/index.php/ijassa/article/view/401/339 (viewed: 29.01.2026)
16. Sloman, A. (1997). What sort of architecture is required for a human-like agent? In: Cognitive Modeling Workshop AAAI96. Portland Oregon. https://cogaffarchive.org/Sloman.what.arch.pdf (viewed: 29.01.2026)