Reply to Barris

We are grateful to Dr. Michael C. Barris for pointing out a resemblance between the subtle color gradients visible in Wassily Kandinsky s 1916 painting (Fig. 1, below) and the assimilative watercolor spreading described by Pinna, 1987. It is true that there is an apparent similarity regarding the veil of color in both. However, the two effects differ in strength and extent as well as in the way they are produced. Kandinsky uses black contours accompanied by a (yellow, red, blue) chromatic gradient that extends far onto the adjoining area. Typically the gradient is wide and shallow not unlike a sawtooth for eliciting the Craik–O Brien–Cornsweet illusion (Wachtler & Wehrhahn, 1997). It has no counterpart on the opposite side, but instead includes a fringe of different color that is detrimental to watercolor spreading (Pinna, Brelstaff, & Spillmann, 2001). In comparison, Pinna et al. (2001) use for their watercolor patterns a thin double contour of a dark (e.g. purple) and a lighter (e.g. orange) color. The lighter fringe runs alongside the darker contour over its full length thereby imparting its color to the enclosed surface area. This induced surface color is so striking that most observers take it for real. It also elicits a slight depth effect and a strong figure–ground segregation not present in Kandinsky s painting. The reader is referred to Pinna et al. (2001, Table 2) for a listing of the main features of the watercolor effect. Fig. 2 demonstrates the coloration and figural property of the watercolor effect in a pattern consisting of a large circle, a square, a hexagon, and a small circle, all arranged concentrically. When the figure is rendered only by purple contours (not illustrated here), the outlines of the four shapes are perceived clearly, but surface organization is lacking. When orange fringes are added to one side of the purple contour of each shape that region is perceived as a figure with a uniformly colored surface. The region without the fringe appears as empty space. If now the orange fringes are added to the other side of the purple contours, the coloration and figure–ground segregation are reversed: what was first a figure now appears as empty space and vice versa. There is no such link between coloration and figural properties in the Kandinsky painting. In fact, from the reproduction in Fig. 1, we would be hesitant to state that there is watercolor. We may safely assume that the weak