Higher-Order Color: From Cones to Postreceptoral Mechanisms

Northeastern University, Boston MA

Investigators

Abstract

Color vision has often been considered a model system for understanding brain-behavior relationships: the sensory inputs are well-defined, the initial processing in the eye is fairly well-understood, and for the most part the dimensions of conscious color experience (hue, brightness, and saturation) are agreed-upon. Yet there is no consensus on a theory of how the color visual system accomplishes its most basic tasks: detection, discrimination, and hue appearance. The experiments in this comprehensive study will provide a complete, computational model for the basic operations of human color vision. The model is fully mechanistic, in the sense that it is built from simple circuits that could easily be built into artificial vision systems that use color for detecting objects, recognizing faces, and decoding emotions. The model will offer concrete hypotheses for other neuroscientists to test: it makes predictions for studies of single-cell physiology as well as for functional neuroimaging. The model can even be used in applied domains such as human factors engineering, textiles, and marketing, example fields where understanding color discrimination is vital. A subset of the model has already been tested and shown to provide remarkably successful predictions for detection, color matching and color discrimination in limited conditions. What is needed is the basic data to set the model parameters in all of color space. The proposed project will provide that data, fit the complete model, and test its application to multiple perceptual tasks. Since 1986, when failures of the 'cardinal axis' color model were discovered by Krauskopf and colleagues, the field has lacked an agreed-upon model of the postreceptoral mechanisms of color detection and discrimination. In our previous work, we showed that a simple model, with only 8 mechanisms, can account for detection and color matching in one plane of color space, and pilot data shows it also accounts for forced-choice discrimination at threshold. No other color mechanism model has been shown to account so well for so many different perceptual phenomena. The project will extend our previous studies in two ways: (1) whereas the previous experiments were limited to a single plane in cone space, the new ones will explore the entire 3-dimensional space of color stimuli; (2) two additional procedures, forced-choice chromatic discrimination and hue scaling, will be used to define the properties of the color mechanisms and test the model predictions. All the measurements will take place under three different chromatic noise masking conditions, and all the subjects will be tested in all of the conditions, to allow for individual differences. Because it builds upon our prior successes, the research is highly-likely to solve the central problem in the field, developing a fully-computable model to explain detection and discrimination in the full 3D color space. The Behavioral Systems Cluster in the Division of Integrative Organismal Systems participated in co-funding of the award. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.