A color hierarchy for automatic target selection

Abstract

Visual processing of color starts at the cones in the retina and continues through ventral stream visual areas, called the parvocellular pathway. Motion processing also starts in the retina but continues through dorsal stream visual areas, called the magnocellular system. Color and motion processing are functionally and anatomically discrete. Previously, motion processing areas MT and MST have been shown to have no color selectivity to a moving stimulus; the neurons were colorblind whenever color was presented along with motion. This occurs when the stimuli are luminance-defined versus the background and is considered achromatic motion processing. Is motion processing independent of color processing? We find that motion processing is intrinsically modulated by color. Color modulated smooth pursuit eye movements produced upon saccading to an aperture containing a surface of coherently moving dots upon a black background. Furthermore, when two surfaces that differed in color were present, one surface was automatically selected based upon a color hierarchy. The strength of that selection depended upon the distance between the two colors in color space. A quantifiable color hierarchy for automatic target selection has wide-ranging implications from sports to advertising to human-computer interfaces.

Figure 1. Experimental paradigm.

Subjects fixated (eye position – black dot) on the white fixation cross (0.5°) at which point a 5° aperture appeared at 7° eccentricity in either the lower right or lower left quadrants. After a random period of time (500–1500 ms) the cross disappeared which was the signal to saccade to the aperture. The surface(s) in the aperture continued to move. Smooth eye movements were measured from 50–200 ms post-saccade. In Experiment 1, the aperture contained a single surface moving at either left or right (shown) and was 1 of 4 possible isoluminant colors (red, blue, green, or yellow). In Experiment 2, the aperture contained two superimposed surfaces moving in opposite directions. The surfaces were comprised of 1 of 6 possible color combinations: red-green, red-blue, red-yellow, green-blue, green-yellow and blue-yellow.

Figure 2. Intrinsic color modulation of pursuit to a single surface.

Each subject's data was normalized to the maximum pursuit produced to any color by that subject, to control for between subject variability. The error bars depict SEMs. Subjects exhibited different pursuit levels for different isoluminant colors. A one-way repeated measures analysis of variance showed a significant main effect of color on pursuit. (F _(3,12) = 4.83, p = 0.0198). This result indicates that colors modulate motion processing which drives the pursuit system.

Figure 3. Preferential selection between superimposed surfaces due to color differences.

A. Open circles represent mean normalized pursuit (bars: SEM) for each distance between the two colors of each of the six color pairs tested. For each color pair, there was a dominant, selected color. The pattern of results suggests a color hierarchy for selection going from blue (weakest) to yellow to green to red (strongest). B. Pursuit increases as the color space distance for the color pair increases (order of pairs same as in A). A regression analysis revealed a significant and strong relationship between the distance in color space of the two surfaces and pursuit (R² = 0.9428, F = 65.96, p = 0.0013). The equation for the regression line is: y = 1.78(x) – 0.079.