Comparison of color and luminance vision on a global shape discrimination task

We compared the performances of the blue-yellow, red-green and luminance systems on a shape discrimination task. Stimuli were radial frequency patterns (radially modulated fourth derivative of a Gaussian) with a peak spatial frequency of 0.75 cpd. Stimuli isolated the chromatic (red-green and blue-yellow) and achromatic post-receptoral mechanisms. We showed that in all cases performance, measured as a radial modulation threshold for discrimination between a circular and non-circular stimulus, improves with contrast. Performance was compared across radial frequencies with contrast matched in multiples of stimulus detection threshold. We find that blue-yellow color system performs the worse on this shape discrimination task, followed by the red-green, with the achromatic system performing best. The average difference is a factor of 2 between achromatic and blue-yellow performance, and a factor of 1.7 between red-green and achromatic. Despite these performance losses, chromatic shape discrimination can still reach hyperacuity performance levels. In a secondary experiment we contrast modulate the radial contour to eliminate either the "corners" or "sides" of an RF4 (square) pattern. We find that for the achromatic system, the sides are more important for the task than the corners. However, for the chromatic system, removal of sides or corners produces similar performance deficits. We conclude that color vision has a selective although relatively mild deficit for two-dimensional form perception.