Meaning without a subject: transcendental structures of cognition in I. Kant’s philosophy and the architecture of neural networks
The light and shadows of digital reality. Living intelligence and its avatars
DOI:
https://doi.org/10.17072/2078-7898/2026-3-382-393Keywords:
transcendental philosophy, Immanuel Kant, artificial intelligence, neural networks, epistemology, AGIAbstract
The article undertakes a systematic comparison between the epistemological architecture of Kant’s transcendental philosophy and the principles of design and operation of contemporary neural networks. Unlike existing studies, which address this parallel only fragmentarily, the article offers a comprehensive, element-by-element comparison of all four levels of Kant’s theory of cognition — the a priori forms of sensibility, the categories of understanding, schematism, and the doctrine of the three syntheses (apprehension, reproduction, and recognition) — with specific architectural and algorithmic solutions found in neural networks: input data representations, aggregation operations and activation functions, regularization mechanisms that ensure generalization, and the stages of information processing from low-level features to final classification. The method applied is one of structural rather than ontological analogy, which avoids a metaphorical identification of thinking with computation. It is shown that the structural isomorphism thus revealed is not accidental: it reflects a fundamental property of any cognitive system, namely the need to actively construct a model of the world on the basis of incoming data. At the same time, the article establishes a principled limit to this analogy: the transcendental unity of apperception — the reflexive self-consciousness that makes genuine meaning possible – has, and in principle can have, no analogue in the distributed architectures of contemporary AI. This conclusion is related to J. Searle’s Chinese Room argument and to the debate over predictive processing in the philosophy of mind. From this follows the central thesis of the article: a neural network reproduces the mechanism of meaning-generation but is not a bearer of meaning. The conclusions bear on discussions on the prospects for artificial general intelligence (AGI) and its ethical dimensions.
References
Беттони М. Кант и кризис программного обеспечения. Предложения по построению программных систем, ориентированных на человека // Кантовский сборник. 1995. Т. 1. № 19. С. 131–137. EDN: WBASGL
Брюшинкин В.Н. Кант и «искусственный интеллект»: модели мира // Кантовский сборник: Межвузовский тематический сборник научных трудов. 1990. № 1(15). С. 80–89. EDN: YUQZSD
Брюшинкин В.Н. Кант и искусственный интеллект: трансцендентальный анализ моделей мира // Кантовский сборник: Межвузовский тематический сборник научных трудов. 1991. № 1(16). С. 84–89. EDN: YUQZYS
Григорова Я.В., Комаров С.В. Анализ проблемы искусственного интеллекта через призму философии И. Канта // Вестник Пермского университета. Философия. Психология. Социология. 2024. № 4. С. 463–470. DOI: https://doi.org/10.17072/2078-7898/2024-4-463-470 EDN: SECXZV
Кант И. Критика чистого разума // Кант И. Сочинения: в 6 т. М.: Мысль, 1964. Т. 3. 799 с.
Кожокару Н.И. Философские концепции И. Канта как предпосылки создания искусственного интеллекта // Философия науки. 2024. № 5S. С. 49–62. DOI: https://doi.org/10.15372/PS20240505
Козолупенко Д.П. Между сознанием и интеллектом: трансцендентальная философия И. Канта как теоретическое основание для исследований в области искусственного интеллекта // Трансцендентальный поворот в современной философии – 8: Метафизика, эпистемология, когнитивистика и искусственный интеллект: сб. тезисов. М.: РГГУ, 2023. С. 117–121. EDN: FRHPMK
Пушкарский А.Г. Значение философии сознания Канта для современных исследований по искусственному интеллекту // Вестник Российского университета дружбы народов. Серия: Философия. 2025. Т. 29, № 2. С. 473–490. DOI: https://doi.org/10.22363/2313-2302-2025-29-2-473-490 EDN: THULXV
Beim Graben P. A neural network account to Kant's philosophical aesthetics. URL: https://arxiv.org/html/2404.12395 (дата обращения: 14.07.2026). DOI: https://doi.org/10.53765/mm2024.227
Buckner C. Empiricism without magic: transformational abstraction in deep convolutional neural networks // Synthese. 2018. Vol. 195, № 12. P. 5339–5372. DOI: https://doi.org/10.1007/s11229-018-01949-1
Buckner C. From Deep Learning to Rational Machines: What the History of Philosophy Can Teach Us about the Future of Artificial Intelligence. New York: Oxford University Press, 2024. 328 p. DOI: https://doi.org/10.1093/oso/9780197653302.001.0001
Cho K., van Merriënboer B., Gulcehre C. et al. Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation // Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), October 25–29, Doha. Qatar: Doha, 2014. P. 1724–1734. DOI: https://doi.org/10.3115/v1/d14-1179
Clark A. Surfing Uncertainty: Prediction, Action, and the Embodied Mind. Oxford: Oxford University Press, 2016. 320 p. DOI: https://doi.org/10.1093/acprof:oso/9780190217013.001.0001
Dreyfus H.L. What Computers Still Can't Do: A Critique of Artificial Reason. Cambridge, MA: MIT Press, 1992. 354 p.
Evans R. A Kantian Cognitive Architecture // On the Cognitive, Ethical, and Scientific Dimensions of Artificial Intelligence. Philosophical Studies Series, vol. 134. Cham: Springer, 2019. P. 233–262. DOI: https://doi.org/10.1007/978-3-030-01800-9_13
Hochreiter S., Schmidhuber J. Long Short-Term Memory // Neural Computation. 1997. Vol. 9, № 8. P. 1735–1780. DOI: https://doi.org/10.1162/neco.1997.9.8.1735
Hui Y. Kant Machine: Critical Philosophy After AI. Political Theory and Contemporary Philosophy. London: Bloomsbury Academic, 2026. 142 p. DOI: https://doi.org/10.5040/9781350563230
Ioffe S., Szegedy C. Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift // Proceedings of the 32nd International Conference on Machine Learning. 2015. Vol. 37. P. 448–456.
LeCun Y., Bottou L., Bengio Y., Haffner P. Gradient-Based Learning Applied to Document Recognition // Proceedings of the IEEE. 1998. Vol. 86, № 11. P. 2278–2324. DOI: https://doi.org/10.1109/5.726791
LeCun Y., Bengio Y., Hinton G. Deep learning // Nature. 2015. Vol. 521. P. 436–444. DOI: https://doi.org/10.1038/nature14539
Nair V., Hinton G.E. Rectified Linear Units Improve Restricted Boltzmann Machines // Proceedings of the 27th International Conference on Machine Learning (ICML-10). 2010. P. 807–814.
Rumelhart D.E., Hinton G.E., Williams R.J. Learning Representations by Back-Propagating Errors // Nature. 1986. Vol. 323, № 6088. P. 533–536. DOI: https://doi.org/10.1038/323533a0
Schlicht T. Minds, Brains, and Deep Learning: The Development of Cognitive Science Through the Lens of Kant's Approach to Cognition // Kant and Artificial Intelligence. Berlin; Boston: De Gruyter, 2022. P. 1–38. DOI: https://doi.org/10.1515/9783110706611-001
Schlicht T. Predictive processing's flirt with transcendental idealism // Noûs. 2026. Vol. 60, № 1. P. 87–109. DOI: https://doi.org/10.1111/nous.12552
Searle J.R. Minds, Brains, and Programs // Behavioral and Brain Sciences. 1980. Vol. 3, № 3. P. 417–424. DOI: https://doi.org/10.1017/s0140525x00005756
Srivastava N., Hinton G., Krizhevsky A., Sutskever I., Salakhutdinov R. Dropout: A Simple Way to Prevent Neural Networks from Overfitting // Journal of Machine Learning Research. 2014. Vol. 15, № 1. P. 1929–1958.
Vaswani A., Shazeer N., Parmar N. et al. Attention Is All You Need // Advances in Neural Information Processing Systems. 2017. Vol. 30. P. 5999–6008. DOI: https://doi.org/10.65215/r5bs2d54
Zahavi D. Brain, Mind, World: Predictive Coding, Neo-Kantianism, and Transcendental Idealism // Husserl Studies. 2018. Vol. 34, № 1. P. 47–61. DOI: https://doi.org/10.1007/s10743-017-9218-z
References
Beim Graben, P. (2024). A neural network account to Kant's philosophical aesthetics. Available at: https://arxiv.org/html/2404.12395 (accessed 14.07.2026). DOI: https://doi.org/10.53765/mm2024.227
Bettoni, M. (1995). [Kant and the crisis of software engineering. Proposals for the design of human-oriented software systems]. Kantovskiy sbornik [Kant Digest]. Vol. 1, no. 19, pp. 131–137.
Bryushinkin, V.N. (1990). [Kant and "artificial intelligence". Models of the world]. Kantovskiy sbornik [Kant Digest]. No. 1(15), pp. 80–89.
Bryushinkin, V.N. (1991). [Kant and artificial intelligence. A transcendental analysis of models of the world]. Kantovskiy sbornik [Kant Digest]. No. 1(16), pp. 84–89.
Buckner, C. (2018). Empiricism without magic. Transformational abstraction in deep convolutional neural networks. Synthese. Vol. 195, no. 12, pp. 5339–5372. DOI: https://doi.org/10.1007/s11229-018-01949-1
Buckner, C. (2024). From deep learning to rational machines. What the history of philosophy can teach us about the future of artificial intelligence. New York: Oxford University Press, 328 p. DOI: https://doi.org/10.1093/oso/9780197653302.001.0001
Cho, K., van Merriënboer, B., Gulcehre, C. et al. (2014). Learning phrase representations using RNN encoder-decoder for statistical machine translation. Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP). Qatar: Doha, pp. 1724–1734. DOI: https://doi.org/10.3115/v1/d14-1179
Clark, A. (2016). Surfing uncertainty. Prediction, action, and the embodied mind. Oxford: Oxford University Press, 320 p. DOI: https://doi.org/10.1093/acprof:oso/9780190217013.001.0001
Dreyfus, H.L. (1992). What computers still can't do. A critique of artificial reason. Cambridge: MIT Press, 354 p.
Evans, R. (2019). A Kantian cognitive architecture. On the cognitive, ethical, and scientific dimensions of artificial intelligence. Philosophical Studies Series. Vol. 134, pp. 233–262. DOI: https://doi.org/10.1007/978-3-030-01800-9_13
Grigorova, Ya.V. and Komarov, S.V. (2024). [Analyzing the problem of artificial intelligence through the prism of Immanuel Kant’s philosophy]. Vestnik Permskogo universiteta. Filosofiya. Psikhologiya. Sotsiologiya [Perm University Bulletin. Philosophy. Psychology. Sociology]. Iss. 4, pp. 463–470. DOI: https://doi.org/10.17072/2078-7898/2024-4-463-470
Hochreiter, S. and Schmidhuber, J. (1997). Long short-term memory. Neural Computation. Vol. 9, no. 8, pp. 1735–1780. DOI: https://doi.org/10.1162/neco.1997.9.8.1735
Hui, Y. (2026). Kant machine. Critical philosophy after AI. Political theory and contemporary philosophy. London: Bloomsbury Academic Publ., 142 p. DOI: https://doi.org/10.5040/9781350563230
Ioffe, S. and Szegedy, C. (2015). Batch normalization. Accelerating Deep Network Training by Reducing Internal Covariate Shift. Proceedings of the 32nd International Conference on Machine Learning. Vol. 37, pp. 448–456.
Kant, I. (1964). [Critique of pure reason]. Sochineniya v 6 t. [Works in 6 vols]. Moscow: Mysl’ Publ., vol. 3, 799 p.
Kozhokaru, N.I. (2024). [Kant's philosophical concepts as prerequisites for the creation of artificial intelligence]. Filosofiya nauki [Philosophy of Science]. No. 5S, pp. 49–62. DOI: https://doi.org/10.15372/PS20240505
Kozolupenko, D.P. (2023). [Between consciousness and intelligence. I. Kant's transcendental philosophy as a theoretical foundation for research in artificial intelligence]. Transtsendental’nyy povorot v sovremennoy filosofii – 8: Metafizika, epistemologiya, kognitivistika i iskusstvennyy intellekt [The Transcendental Turn in Contemporary Philosophy – Metaphysics, Epistemology, Cognitive Science and Artificial Intelligence]. Moscow: RSUH Publ., pp. 117–121.
LeCun, Y., Bengio, Y. and Hinton, G. (2015). Deep learning. Nature. Vol. 521, pp. 436–444. DOI: https://doi.org/10.1038/nature14539
LeCun, Y., Bottou, L., Bengio, Y. and Haffner, P. (1998). Gradient-based learning applied to document recognition. Proceedings of the IEEE. Vol. 86, no. 11, pp. 2278–2324. DOI: https://doi.org/10.1109/5.726791
Nair, V. and Hinton, G.E. (2010). Rectified linear units improve restricted Boltzmann machines. Proceedings of the 27th International Conference on Machine Learning (ICML-10). Madison: Omnipress Publ., pp. 807–814.
Pushkarskiy, A.G. (2025). [The importance of Kant’s philosophy of mind for contemporary research in artificial intelligence]. Vestnik Rossiyskogo universiteta druzhby narodov. Seriya Filosofiya [RUDN Journal of Philosophy]. Vol. 29, no. 2, pp. 473–490. DOI: https://doi.org/10.22363/2313-2302-2025-29-2-473-490
Rumelhart, D.E., Hinton, G.E. and Williams, R.J. (1986). Learning representations by back-propagating errors. Nature. Vol. 323, no. 6088, pp. 533–536. DOI: https://doi.org/10.1038/323533a0
Schlicht, T. (2022). Minds, brains, and deep learning. The development of cognitive science through the lens of Kant's approach to cognition. Kant and Artificial Intelligence. Berlin, Boston: De Gruyter Publ., pp. 1–38. DOI: https://doi.org/10.1515/9783110706611-001
Schlicht, T. (2026). Predictive processing's flirt with transcendental idealism. Noûs. Vol. 60, no. 1, pp. 87–109. DOI: https://doi.org/10.1111/nous.12552
Searle, J.R. (1980). Minds, brains, and programs. Behavioral and Brain Sciences. Vol. 3, no. 3, pp. 417–424. DOI: https://doi.org/10.1017/s0140525x00005756
Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I. and Salakhutdinov, R. (2014). Dropout. A simple way to prevent neural networks from overfitting. Journal of Machine Learning Research. Vol. 15, no. 1, pp. 1929–1958.
Vaswani, A., Shazeer, N., Parmar, N. et al. (2017). Attention is all you need. Advances in Neural Information Processing Systems. Vol. 30, pp. 5999–6008. DOI: https://doi.org/10.65215/r5bs2d54
Zahavi, D. (2018). Brain, mind, world. Predictive coding, neo-kantianism, and transcendental idealism. Husserl Studies. Vol. 34, no. 1, pp. 47–61. DOI: https://doi.org/10.1007/s10743-017-9218-z
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Perm University Herald. Philosophy. Psychology. Sociology
This work is licensed under a Creative Commons Attribution 4.0 International License.
