Simple displacement models cannot explain some aspects of optical illusions and figural aftereffects. The orientation-detector interaction model proposed by Blakemore and others is more suitable to explain many aspects of the Zöllner illusion, positive and negative illusions, the effect of gap between the inducing and test lines, and the anisotropy of illusions. If we hypothesize size detectors whose tuning width and distribution steps are proportional to logarithmic size, interactions between them explain well the fact that the Delboeuf illusion and figural aftereffects of circles are determined by the size ratio of the inducing to test circle, not by the absolute distance between the contours of these circles.