Abstract
Purpose. In the staircase Gelb illusion, five adjacent and coplanar squares, ranging from white to black (30:1 reflectance ratio), are suspended in midair and illuminated by a spotlight. The perceived lightness ranges from white to light-middle gray (3:1 perceived ratio), which is a dramatic compression of the actual range. This compression decreases (5:1 perceived ratio) when the squares are rearranged into a Mondrian pattern. According to the anchoring theory (Gilchrist et al, 1999), this decrease is due to a configuration effect that would make Mondrians stronger local frameworks with respect to the global framework, with the result of better lightness constancy. Here, we tested this claim by maintaining the original linear arrangement but rearranged the positions of the squares. Our hypothesis is that the difference in perceived ratio's between Mondrian and linear configurations may depend on local interactions among luminances rather than on a global configuration effect.
Methods. The display was five Munsell Neutral Value Scale papers (7×7 cm each), in which the luminance of the black square was 217 cd/m2. The position of the white square was systematically varied in each condition. In a between-subjects design, matches were made using an extended 31-step Munsell scale.
Results. i) The white square darkened its immediate neighbors. ii) The position of the white square, relative to the black, was highly critical. When the two squares were adjacent to each other the perceived lightness ratio (10:1) was greater than that of the Mondrian pattern (5:1) above.
Conclusions. Our results challenge the hypothesis that full range Mondrian patterns produce better lightness constancy. We discuss our findings in the context of current lightness theories (Gilchrist et al., 1999; Kingdom et al., 1996; Bressan, in press).