P133 A Recursive Stability Model of Qualia: Philosophical Self-reference, Neural Attractor Structures, and Experimental Exploration in LLMs
Chang-Eop Kim
Department of Physiology, College of Korean Medicine, Gachon University, 1342, Seong- namdaero, Seongnam 13120, Republic of Korea
Email:eopchang@gachon.ac.kr
Introduction
Qualia represent a fundamental challenge in consciousness research, defined as inherently subjective experiences that resist objective characterization. Philosophically, qualia have been proposed to possess self-referential characteristics, aligning conceptually with Douglas Hofstadter’s "strange loop" theory, which suggests subjective experience might arise from recursive structures [1]. However, explicit mathematical and empirical formulations of this concept remain scarce.
Methods
We developed a mathematical formalization of qualia using recursive stability, identifying fixed-point states reflecting neural circuits recursively referencing their outputs. Neuroscientific literature was reviewed to identify biological phenomena potentially implementing recursive stability. Additionally, analogous candidate structures were explored within artificial neural networks, particularly focusing on attention mechanisms in large language models (LLMs).
Results
The mathematical formulation effectively captured essential characteristics of subjective conscious experiences, including their inherent immediacy and the necessary equivalence between existence and self-awareness. Neuroscientific literature suggested candidate biological structures, such as hippocampal CA3 attractor networks indirectly supporting self-referential episodic memory, and sustained-activity circuits in prefrontal cortex known for roles in conscious cognition [2,3]. At the cellular level, basic biological feedback loops provided foundational examples of recursive mechanisms. Computationally, Hopfield network-like structures, explicitly self-referential and analogous to Hofstadter's "strange loop," were identified in the attention mechanisms of LLMs, indicating potential attractor-like behaviors and recursive self-reference within these models.
Discussion
This research supports recursive stability as a robust mathematical framework bridging philosophical, neuroscientific, and computational perspectives on qualia. Computational findings suggest LLMs as practical platforms for experimentally exploring self-referential consciousness models. Future research should empirically validate these recursive structures within biological systems and further refine computational implementations to deepen our understanding of consciousness.
Acknowledgements
This work was supported by the National Research Foundation of Korea(NRF) grant funded by the Korea government(MSIT)(RS-2024-00339889).
References
[1] Hofstadter, D. R. (2007). I Am a Strange Loop. Basic Books. ISBN: 978-0465030798.
[2] Dehaene, S., Lau, H., & Kouider, S. (2017). What is consciousness, and could machines have it? Science, 358(6362), 486-492.https://doi.org/10.1126/science.aan8871
[3] Mashour, G. A., Roelfsema, P., Changeux, J. P., & Dehaene, S. (2020). Conscious processing and the global neuronal workspace hypothesis. Neuron, 105(5), 776-798. https://doi.org/10.1016/j.neuron.2020.01.026
