Abstract
The number and nature of the mechanisms mediating color appearance remain poorly understood. We used factor analysis to examine the patterns of individual variation in a previous study of hue scaling measurements reported for 59 color-normal observers (Malkoc et al. JOSA A 2005). The settings involved judging the strength of red, green, blue, and yellow percepts for 24 equiluminant stimuli spanning a circle at 15 deg steps in a version of MacLeod-Boynton color space (scaled to equate sensitivity along the LM and S axes). Three observers were excluded based on inconsistent settings. A factor analysis with Varimax rotation was performed for the remaining 56, and revealed a large number of systematic but narrowly tuned and unimodal factors, spaced roughly evenly across hue angle. Together these accounted for a large proportion of the variance (80% for the first 8), while the variance attributable to any individual factor was relatively low (8-15%). Separate factors emerged for reddish and greenish hue angles, and for blue and yellow, indicating that the two poles of the putative opponent axes varied independently. Moreover, additional independent factors occurred for intermediate hue angles corresponding to purple, cyan, or orange regions. Our analysis confirms previous evidence that inter-observer differences in color percepts vary independently not only for the unique hues but also for the intermediate or putative binary mixtures. The pattern is inconsistent with the predictions for individual variations in conventional opponent mechanisms (e.g. in their preferred color axis or sensitivity), and instead suggests the presence of multiple, higher-order mechanisms underlying color percepts, even when observers are restricted to describing these percepts only in terms of the four unique hues. Supported by EY-10834
Meeting abstract presented at VSS 2016