Purchase this article with an account.
Javid Sadr; An Alarming and Practical Case of Compromised Color Constancy. Invest. Ophthalmol. Vis. Sci. 2014;55(13):6241.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
High-level color perception isn't a simple product of cone activity; it derives from broader analyses concerning the illuminant, surface reflectance properties, and context. The perceptual phenomenon of color constancy nicely demonstrates our ability to negotiate variable lighting and maintain "correct" color percepts, even when "correct" does not mean veridical in terms of received wavelengths. Certain manipulations, however, can disrupt color constancy, and in an important and potentially dangerous real-world application we hypothesized that the extreme filtering properties of colored translucent containers (specifically, common medical prescription bottles) would result in gross misperception of colored items within (i.e., prescription pills).
Subjects performed a series of color perception tasks not limited to basic verbal color identification; these included: free color-naming of individually-presented colored item samples in clear (control) vs yellow-tinted containers; free color-naming of multiple samples of colored items in a gallery (allowing simultaneous comparison across containers); color-matching, pairing colored samples in yellow-tinted containers with a gallery of reference samples in clear containers; "same"/"different" comparison between pairs of samples in yellow-tinted containers and reference samples in clear containers; and a combined color-naming / color-counting task given containers with multiple/mixed color samples (allowing simultaneous comparison within each container).
As predicted, compared to colored items in clear (control/reference) containers, for which all forms of testing yielded at-ceiling correct performance, the same colored items presented in yellow-tinted containers were identified, matched, compared, and counted incorrectly with incredible frequency. For example: in free color-naming, correct identification fell as low as 15% for the pink samples; in paired same/different comparisons, incorrect matching of, e.g., purple and green samples occurred in 73% of trials -- 93% for yellow and white; with mixed-color samples, the mixture of orange and pink items, or of yellow and white items, was almost universally perceived as a homogenous sample.
Our results reveal profound deficits in color perception caused by common translucent containers, with direct and alarming translation to the real-world potential for very hazardous pharmaceutical errors.
This PDF is available to Subscribers Only