A scaling analysis of the snake lightness illusion

Logvinenko and Maloney (2006) measured perceived dissimilarities between achromatic surfaces placed in two scenes illuminated by neutral lights that could differ in intensity. Using a novel scaling method, they found that dissimilarities between light surface pairs could be represented as a weighted linear combination of two dimensions, "surface lightness" (a perceptual correlate of the difference in the logarithm of surface albedo) and "surface brightness" (which corresponded to the differences of the logarithms of light intensity across the scenes). Here we attempt to measure the contributions of these dimensions to a compelling lightness illusion (the "snake illusion"). It is commonly assumed that this illusion is a result of erroneous segmentation of the snake pattern into regions of unequal illumination. We find that the illusory shift in the snake pattern occurs along the surface lightness dimension, with no contribution from surface brightness. Thus, even if an erroneous segmentation of the snake pattern into strips of unequal illumination does happen, it reveals itself, paradoxically, as illusory changes in surface lightness rather than as surface brightness. We conjecture that the illusion strength depends on the balance between two groups of illumination cues signaling the true (uniform) illumination and the pictorial (uneven) illumination.