COLOR MECHANISMS AND COLOR APPEARANCE
A major thrust of our work has been to model the color mechanisms that underlie chromatic detection and discrimination. These mechanisms are relatively simple circuits that combine signals from the cone photoreceptors of the retina; the mechanism outputs are then sent to other brain centers for further processing before contributing to observable behavior (such as making detection responses).
The picture above shows a recent version of our model, modified from Shepard et al., 2016. There are three pairs of quasi-linear combinations of signals from the L and M cone photo-receptors. This model accounts for detection of colored stimuli by the L and M cones, and correctly predicts the effects of adding masking noise to the stimuli.
More recently, (Shepard et al, 2017), we have measured asymmetric color matches for the same stimuli, and with the same observers, that we used in measuring the detection thresholds. The dots in the figure below represent color matches to threshold stimuli in the LM plane, and show that these matches fall into six, non-overlapping clusters, corresponding to the six mechanisms in our model.
Moreover, the matches map onto the color detection mechanisms nearly perfectly. An example is shown below. The points represent detection thresholds, and the colored lines are the threshold lines of the six mechanisms in the model. The six mechanisms account for detection nearly perfectly. The color of the points is an exaggerated representation of the color matches to the threshold stimuli. The correspondence between clusters of colors and the mechanism lines is nearly perfect; for example, the thresholds detected by the “G” mechanism are all matched with green, the Y-detected stimuli are matched with yellow, and so on. When chromatic masking noise changes the pattern of detection (that is, which tests are detected by which mechanisms), the color matches change appropriately, as predicted by our model.
We are currently studying whether this model can also account for discrimination between two threshold-level stimuli, using our Bayesian Classifier model (Eskew et al. 2001).
This work is supported by the National Science Foundation (Grant BCS- 1921771, Eskew PI).
There is currently an epidemic of myopia (nearsightedness), especially in Asia but also in other developed countries. Myopia is not just an inconvenience and cost: high degrees of myopia are associated with significantly elevated risk of serious eye disease. The major cause of myopia is that the eye grows too long during child development; the question we are studying is why this incorrect regulation of eye growth occurs.
With Dr. Frances Rucker, of the New England College of Optometry, we are studying the relationship between myopia and chromatic sensitivity. It is known that the difference in focus for different wavelengths of light (longitudinal chromatic aberration or LCA) can be a stimulus for accommodation (focusing) of the eye, and perhaps can serve as well as a cue for regulating eye grown during development. If so, it may be that some children are less sensitive to the regulatory influence of LCA, and that may contribute to the development of myopia. This difference would most likely manifest itself as a relative lack of sensitivity to S cone stimulation, since short-wavelength light has the largest LCA.
If the chromatic sensitivity difference persists into adulthood, we should find a correlation between S cone sensitivity and refractive error (degree of myopia) in adult subjects. We have found just that— adults with greater degrees of myopia have higher thresholds for S cone isolating patterns (Taylor et al., 2018).
The lab is currently exploring exciting new methods for studying human sensitivity to the effects of LCA using psychophysics.
This work is supported by the National Eye Institute (NIH Grant R01EY023281, PI Rucker, co-PI Eskew).