A Metric for Three-Dimensional Color Discrimination Derived from V1 Population Fisher Information
A new 17-parameter model from neural population codes fits four major color discrimination datasets spanning 84 years.
Neuroscience researcher Michael Menke has published a novel computational model that derives a geometric framework for human color discrimination directly from the brain's visual processing. The work, titled 'A Metric for Three-Dimensional Color Discrimination Derived from V1 Population Fisher Information,' constructs a Riemannian metric on color space using Fisher information—a statistical measure of precision—from neural population codes in the primary visual cortex (V1). This approach maps biological stages like photoreceptor adaptation and retinal opponent channels onto specific geometric constructions, resulting in a 17-parameter metric tensor whose components correspond to measurable neural quantities.
Menke validated the model by performing a joint fit to four foundational but historically separate experimental datasets on color discrimination thresholds. These include MacAdam's (1942) classic chromaticity ellipses, the recent three-dimensional ellipsoids from Koenderink et al. (2026), Wright's (1941) wavelength discrimination function, and the threshold color difference ellipses from Huang et al. (2012). Together, these datasets represent 96 independently measured conditions across varied chromaticities and luminances, spanning over eight decades of research. The model achieved standardized STRESS fit values of 23.9, 20.8, 30.1, and 30.8 on the respective datasets, demonstrating its ability to unify disparate perceptual phenomena under a single neuro-geometric theory.
The research, available as a preprint on arXiv (2603.24356), provides a rigorous mathematical bridge between low-level neural mechanics and high-level perceptual experience. By showing how the brain's encoding strategy geometrically warps color space, it offers a predictive framework that could refine color models used in computer vision, display technology, and visual neuroscience.
- Derives a 17-parameter Riemannian metric for color space from V1 neural population Fisher information, linking geometry to biology.
- Unifies four major experimental datasets from 1941 to 2026, covering 96 discrimination conditions, with STRESS fits between 20.8 and 30.8.
- Provides a predictive framework that maps photoreceptor adaptation, retinal channels, and cortical encoding onto geometric constructions.
Why It Matters
This model bridges neural mechanics and perception, offering a unified theory to improve color science in displays, computer vision, and visual prosthetics.