Last we showed (Geier et al, 2004 Perception 33 53) that the Hermann grid illusion disappears by applying certainly distortions on the lines. Because there is no sufficient explanation for this new phenomenon, we try to come near to the solution by psychophysical measurements.
The dependent variable was the „distortion tolerance”, that means: at a given distortion type, starting from straight lines and increasing the distortion step by step, what is that measure when the illusion disappears for the given subject. The independent variables was the distortion type and the line width.
The applied distortion types: sinusoid lines (the lines are sinus curves while the intersections rotated, but remains right angled), waved lines (in the middles of the line parts are sine curves with one period), humped lines (“humps” in the middles of the line parts), knotted lines (thickening the middles of the line parts), and half sided knotted lines (the middles of the line parts are thickened only in one side, other side remains straight).
Stimuli were showed in random order by a computer monitor: d=15”, 1024×768 pixels, viewing distance 60cm, gazing with free eye movements. Distorted Hermann grids include 7×5 white lines in a black background, with 3 different line width: 11, 17 and 23 pixels, with constant line spacing: 102 pixels. Subjects n=22.
Results of two way ANOVA: main effect of distortion type is high significant (p<0,01); main effect of line width is no significant; interaction is no significant.
Results of pairwise comparison: differences between the half sided knotted line and every other line types are significant; differences between all other pairs are not significant.
Our conclusion is that the main cause of the Hermann grid illusion is the straightness of the black-white edges; the line width plays no significant role!