Selective color constancy deficits after circumscribed unilateral brain lesions

Abstract

The color of an object, when part of a complex scene, is determined not only by its spectral reflectance but also by the colors of all other objects in the scene (von Helmholtz, 1886; Ives, 1912; Land, 1959). By taking global color information into account, the visual system is able to maintain constancy of the color appearance of the object, despite large variations in the light incident on the retina arising from changes in the spectral content of the illuminating light (Hurlbert, 1998; Maloney, 1999). The neural basis of this color constancy is, however, poorly understood. Although there seems to be a prominent role for retinal, cone-specific adaptation mechanisms (von Kries, 1902; Pöppel, 1986; Foster and Nascimento, 1994), the contribution of cortical mechanisms to color constancy is still unclear (Land et al., 1983; D'Zmura and Lennie, 1986). We examined the color perception of 27 patients with defined unilateral lesions mainly located in the parieto-temporo-occipital and fronto-parieto-temporal cortex. With a battery of clinical and specially designed color vision tests we tried to detect and differentiate between possible deficits in central color processing. Our results show that color constancy can be selectively impaired after circumscribed unilateral lesions in parieto-temporal cortex of the left or right hemisphere. Five of 27 patients exhibited significant deficits in a color constancy task, but all of the 5 performed well in color discrimination or higher-level visual tasks, such as the association of colors with familiar objects. These results indicate that the computations underlying color constancy are mediated by specialized cortical circuitry, which is independent of the neural substrate for color discrimination and for assigning colors to objects.

Fig. 1.

A, Illustration of the color constancy display under the neutral illuminant C (daylight). Each of the Mondrian patterns consisted of a random arrangement of small, colored rectangles and a central larger, horizontal gray rectangle.B, Same reflectances under a bluish simulated illuminant. The colored patches above the Mondrian illustrate settings at different levels of color constancy (not shown during the experiment). C, CIE coordinates of the simulated surface colors of 226 different Munsell chips illuminated by five different illuminants. The corresponding transformation in color space for each Munsell chip under the four colored illuminants is indicated by the appropriate color; the neutral setting under CIE standard illuminant C is indicated in black.D, The observer had to adjust the color of the central rectangle by changing its color value along one of the cardinal directions of color space, as illustrated by the colored lines, until it appeared as a neutral gray under each of the five illuminants.

Fig. 2.

A, Stimulus for the color discrimination task. A matrix of 8 × 8 squares was presented: 4 of the 64 squares were arranged as a 2 × 2 quadratic subset, which differed either slightly in color (red, green, blue, or yellow) or in luminance (lighter or darker) from the other 60 neutral gray squares. The subject had to report the location (quadrant) of the subset. B, Illustration of the color directions used in this study. Note that the appearances of these “cardinal direction” stimuli are distinctly different from perceptually determined “unique hues” (Krauskopf et al., 1982).

Fig. 3.

Examples of the stimuli used in the color-naming, object-naming, and color association tests. A, In the color-naming test, the subjects were required to name the color of familiar objects. B, In the object-naming test, they were required to name the familiar objects and their typical colors.C, In the color association test, they were required to say whether the object was appropriately colored.

Fig. 4.

Histograms of the color constancy results of 27 patients and 8 control subjects. The ordinate gives the number of cases as a function of the color constancy indices in percent for neutral gray settings along both cardinal color directions. An index value of 0 indicates a setting of the gray rectangle irrespective of the change in illuminant (no color constancy); an index value of 100 indicates perfect adjustment to the illuminant (absolute color constancy). For comparison, the black line shows the distribution of settings for all unimpaired observers, including the control subjects; significant deviations of >2.28 SD are indicated by the shaded area. A, Color constancy indices for neutral gray settings along the red–green axis. The average of settings in both directions, toward red and toward green, is plotted. In this task, the five patients, AK, AS, GZ, RM, and GK, matched the gray value outside the statistically normal range. Patients HR and GR were just barely within the normal range. B, Color constancy indices for neutral gray settings along the blue–yellow axis. The average of settings in both directions along the blue–yellow axis is plotted. In this task, the six patients, AK, AS, HR, GZ, RB, and MH, matched the gray value outside the statistically normal range. The match of patient RM was just barely within the normal range.

Fig. 5.

Scatterplot of the color constancy indices from Figure 4 along the red–green and blue–yellow color axes. Theshaded area indicates the range of average color constancy impairment. Slightly negative matches have been set to a value of zero. The five patients showing color constancy deficits along both color axes are AK, AS, HR, GZ, and RM. The other patients with deficits along only one color axis (GK, RB, and MH) showed normal performance along the other color axis, so that their average performance fell within the normal range.

Fig. 6.

Histograms of the luminance and color discrimination thresholds for 27 patients and 9 control subjects. In each histogram, the number of cases is given as a function of the discrimination threshold, specified as luminance or cone contrast. Theblack line corresponds to the distribution of thresholds for all unimpaired observers, including the control subjects; significant deviations of >2.28 SD are indicated by the shaded area. A, Luminance discrimination thresholds for stimuli differing along the light–dark axis. The discrimination threshold is given as the standard luminance contrast (ΔLum/Lum) in percent. Patients MS, KB, and GZ had significant deficits for discriminating darker or lighter stimuli. B, Color discrimination thresholds for stimuli differing along the red–green axis. The discrimination threshold is given as root mean squared L and M cone contrast in percent. Only patient MH showed a small but significant deficit for discriminating stimuli along this axis.C, Color discrimination thresholds for stimuli differing along the blue–yellow axis. The discrimination threshold is given as S cone contrast in percent. Patients IL and MS had small but significant deficits for discriminating stimuli differing along this color axis. Control subject WR had a strong deficit along that color axis, probably because of an age-related yellowing of the lens.

Fig. 7.

Scatterplot of the average color constancy index versus the average color discrimination index. Thex-axis gives the average of the color constancy indices for the red–green and blue–yellow axes; the y-axis gives the average normalized (mean = 50; SD = 25) discrimination thresholds for the red–green and blue–yellow axes (the lower the index, the better the discrimination). The shaded area indicates deficits in each one of these two tasks. The five patients with deficits in color constancy (AK, AS, HR, GZ, and RM) showed normal color discrimination.

Fig. 8.

Axial lesion template reconstructions and tilt planes of the five patients with color constancy deficits: AK, AS, HR, GZ, and RM. The chosen tilt plane is indicated above each row. The templates are shown from the top (left) to the bottom (right) of the brain. The lesioned brain tissue is depicted in black. Note that according to current convention in neuroradiology, the left brain hemisphere is plotted on the right.

Fig. 9.

Comparison of the lesion locations (overlap) of patients AK, AS, and HR (magenta) and RM (yellow) with the approximate location of lesions reported to cause achromatopsia (red squares). Sagittal, axial, and coronal views are shown in a Talairach coordinate frame (Talairach and Tournoux, 1988). The large, green area indicates the union of the lesions of patients AK, AS, and HR after lateralizing them all to the left hemisphere;blue, overlap of the lesions for any two of the patients; magenta, intersection of the lesions of all three, centered at Talairach coordinates ±c, E1, and 9 (±50, 7, and −8 mm for the standard Talairach brain). The yellow region occipitally shows the lesion of patient RM, which was centered on area V2. The red regions associated with achromatopsia have been termed the “color center” of the brain (Lueck et al., 1989). Most likely, they correspond to area TEO in monkeys (Heywood and Cowey, 1998).