Purchase this article with an account.
Timothy G. Shepard, Emily A. Swanson, Rhea T. Eskew; Color mechanisms revealed by measuring detection and discrimination together.. Journal of Vision 2013;13(9):1013. doi: https://doi.org/10.1167/13.9.1013.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Much research in human color vision focuses on the psychophysical mechanisms required for the detection and discrimination of color. However, there have been conflicting reports of how many such mechanisms exist. Here, we report results of a study in which detection and discrimination were measured under identical conditions. In the detection task, forced-choice detection thresholds for a 1cpd Gabor patch were measured for many test color angles in the (ͧ0;L/L, ͧ0;M/M) plane. In the separate discrimination task, two stimuli at detection threshold (one "standard" color and another test color, the angles of which were varied between runs) were presented in random order, and observers were asked to select the "standard" angle. Test and standard angles were mostly selected from angles near the ends of the detection contour, near where Hansen & Gegenfurtner (2010, J Vis 10: 388) found highly selective masking, which they interpreted as showing multiple, "higher order" color mechanisms. Qualitatively, (i) if both stimuli are detected by the same mechanism, performance should be at chance level (50%); (ii) if the stimuli are always detected by two different mechanisms, discrimination should be as good as detection (82% or higher); however, (iii) if one or both stimuli are detected by multiple mechanisms, performance should be intermediate (55-65%) since, on some trials, only one mechanism will detect a given stimulus, whereas on other trials two or more will. A Bayesian Color Classifier, threshold-level color categorization model, based upon six unipolar detection mechanisms fit to the detection thresholds, was applied to the discrimination data (Eskew et al., 2001 Vision Research 41, 893). The classifier does a good job of describing discrimination, especially given that there are no free parameters in the model. The discrimination results are consistent with a very limited number of detection mechanisms, not with multiple, "higher order" mechanisms.
Meeting abstract presented at VSS 2013
This PDF is available to Subscribers Only