Abstract
Anchoring theory claims that simultaneous lightness contrast occurs because the gray target on the black background is the highest luminance in its local framework. Maniatis (2015) challenged this account merely by placing an additional white square on the black background and showing that the illusion is not eliminated. Clearly if the black background were replaced by white, there would be no illusion. Such a white background would differ from the white square Maniatis added in at least three ways: The white background (1) is larger in area, (2) surrounds the target, and (3) shares a border with the target. To determine which property (or combination) might kill the illusion, we created 5 stimulus displays that varied in these properties. Each was viewed on a computer monitor by a separate group of 15 observers who made matches both using a Munsell chart and by adjusting a variable square patch in a checkerboard background on the same computer screen. Surrounding the target with white (without increasing its area) did not reduce the illusion, even though 100% of the target border was in contact with white. Making half of the target overlap the white square reduced the illusion by about half. Only one display killed the illusion: the added white square was enlarged to fill much of the black background but made no contact with the target. This implies that in order to fully anchor the gray target on the black background (thus killing the illusion), the white region must have a sufficient relative area. This is consistent with several publications (Gilchrist & Radonjic, 2009) showing that anchoring depends on both relative luminance and relative area. Our results provide absolutely no support for the traditional lateral inhibition explanation of this illusion.
Meeting abstract presented at VSS 2017