A labeled line theory accurately predicts individual differences in color experience

Abstract

Purpose: Explaining widespread individual differences in the spectral location of unique green has proven to be a challenge for opponent models of color experience. We have developed a labeled line neurobiological model of hue perception that predicts large individual differences in the spectral location of unique green as a function of middle (M) to long (L) wavelength sensitive cone ratio, but very little corresponding variability in unique yellow. We tested these predictions by collecting data on L:M cone ratio and color appearance from a group of observers.

Methods: L:M ratios were measured for 14 color normal individuals with an electroretinogram flicker-photometric method that accounts for individual differences in the spectral sensitivities of L and M cones deduced from genetics. For color testing, we used a programmable Gooch & Housego OL 490 light source to generate monochromatic stimuli. In two separate sessions, the location of unique yellow and green was measured using a forced choice randomly interleaved double staircase procedure. Each session consisted of 3 trials for each hue. The first session was discarded for practice.

Results: A spectral neutral point of the red-green (RG) system (yellow; µ=574.7, σ=3.78) was not significantly correlated (r²=0.08, p=0.31) with the relative number of L to M cones. Strikingly, however, the neutral point of the blue-yellow (BY) system (green; µ=525.7, σ=11.579) was significantly correlated with cone ratio (r²=0.54, p=0.003). Higher L:M ratios were predictive of shorter unique green values and the relationship followed a function generated from our model.

Conclusions: We discovered a predictive relationship between L:M cone ratio and the location of pure green - lower ratios were associated with unique green closer to yellow, while a high ratio influenced the point towards blue. In stark contrast, L:M ratio did not significantly impact the position of unique yellow. These findings are accurately predicted by a neurobiological model in which color experience is mediated by midget ganglion cells with S-cone input via H2 horizontal cells and a random mix of L and M cone signals. In this model the RG system arises when M and (S+L) signals are compared, while BY percepts are driven by ganglion cells with L vs. (M+S) opponency. The agreement between model and observation implies that a small subset of midget ganglion cells serve as labeled lines carrying hue information.

Keywords: 471 color vision • 641 perception • 649 photoreceptors: visual performance