C01 Principal Investigator  Oizumi has published a preprint of a research paper co-authored by Ryota Kanai, President of Araya, and Chanseok Lim, a postdoctoral fellow in Oizumi's lab. We believe that this paper will offer theoretical insights into the nature of qualia structures and inform future research. We propose that qualia structures are fully characterized by principal bundle geometry, which naturally arises from the symmetries that the brain learns from the world. 

 Please click here to read the full text. Please take a moment to read it. Please see Prof. Oizumi's X (formerly: Twitter) for more information.

Paper Information

Authors：Oizumi, Masafumi, Chanseok Lim, and Ryota Kanai

Title：Principal Bundle Geometry of Qualia: Understanding the Quality of Consciousness from Symmetry

Journal： PsyArXiv. July 15. (2025)

Preprint DOI：https://doi.org/10.31234/osf.io/agupq_v2

Abstract: 

 Qualia, the subjective qualities of experience, pose a fundamental challenge to scientific explanation. A promising recent approach is to characterize qualia by their relational structures rather than their intrinsic nature. Assuming the structure of qualia is characterized by the geometry of neural representations, we posit that this geometry is fundamentally organized by symmetry. We propose that the principal bundle, induced by a symmetry group $G$ acting on a neural network's state space, is the essential mathematical object that provides a complete and rigorous characterization of this geometric structure. The core mathematical principle is that, for a $G$-equivariant network, its state space partitions into a quotient space $Q$ and orbits induced by the group $G$. This framework reveals a three-level hierarchy for characterizing qualia. First, the choice of the symmetry group $G$ itself defines the fundamental nature of the qualia modality (e.g., vision versus audition). Second, we identify a qualia signature with a point in the quotient space $Q$ (e.g., perceiving the signature ``cat''), representing the invariant identity of a percept. Third, we identify a qualia attribute with a position on a specific orbit (e.g., the cat's location), representing its continuous variations. This framework uncovers a powerful duality: the geometry of the orbits is rigid, inheriting the structure of the group $G$, which explains the stable relational structure of qualia attributes. In contrast, the geometry of the quotient space is plastic, allowing the relationships between qualia signatures to be shaped by learning. This geometric hierarchy provides a unified language for explaining the structure of perceptual experience. Furthermore, our framework generates empirical predictions and a clear research strategy for characterizing qualia, offering a path to a deeper scientific understanding of subjective experience.