Abstract:
This paper introduces the Holo-Manifold Theory of Vision, a novel framework positing that the brain does not reconstruct a veridical three-dimensional model of the external world but instead constructs a seven-dimensional information manifold. This manifold serves as the fundamental computational space for visual perception. We define the seven orthogonal dimensions of this manifold: two spatial coordinates (X,Y), chrominance-luminance (C), form-orientation (Φ), temporal dynamics (T), stereopsis-depth (Δ), and a novel seventh dimension, the predictive prior (Ψ). The Ψ-dimension, derived from principles of Bayesian inference and predictive coding, represents the brain’s internal generative model, encoding expectation, context, and attention. Its inclusion transforms the manifold from a purely sensory representation into a complete perceptual state-space. We propose that neural computations, such as object recognition and tracking, are executed as geometric operations (e.g., calculating geodesic paths) within this 7D space. The theory offers a unified explanation for disparate phenomena, including bottom-up feature detection, top-down attentional modulation, and perceptual completion. It provides a conceptual bridge between the distributed, parallel processing observed in the visual cortex and the coherent, holistic nature of conscious visual experience, drawing analogies to the holographic principle in physics.
Yıldırım, E. (2025). The Holo-Manifold Theory of Vision: A Seven-Dimensional Framework for Neural Processing. Zenodo. https://doi.org/10.5281/zenodo.17049740
Leave a Reply