Quantifying color constancy: evidence for nonlinear processing of cone-specific contrast

Area of Science:

• Vision science
• Color perception
• Computational neuroscience

Background:

• Color constancy is the ability to perceive consistent object colors despite changes in illumination.
• Existing models, such as Retinex and von Kries transformation, rely on cone-specific normalization.
• Previous studies often used white illuminants, limiting the understanding of color constancy under complex lighting.

Purpose of the Study:

• To investigate color constancy under a wider range of illuminants, including pairs of colored lights.
• To test the predictive accuracy of current computational models of color constancy.
• To identify the factors influencing deviations from model predictions.

Main Methods:

• Haploscopic matching of color samples presented on a CRT monitor under two different illuminants.
• Stimulus included 35 color samples and 5 neutral samples on a white background.
• Simulation of trichromatic illuminant-object interaction using three monochromatic lights.

Main Results:

• Systematic deviations were observed between perceived colors and model predictions.
• A nonlinear response transformation, dependent on sample/background contrast and illuminant stimulation, better described the data.
• The short-wave-sensitive (SWS) cone system was identified as the primary source of deviations due to its selective stimulation.

Conclusions:

• Current cone-specific normalization models inadequately explain color constancy under diverse illuminants.
• A nonlinear response transformation is necessary to account for observed perceptual data.
• The unique stimulation properties of the SWS system play a critical role in color constancy deviations.