A new diffusion computational model predicts both the positive and the negative short afterimage effects

The goal of this research was to develop a compound computational model able to predict the "opposite" effects of the "color dove illusion" [1], [2] and the "filling in the afterimage after the image" effect [3]. Until now, only one study [4], based on a previous model, FACADE [5], had attempted to model the negative filling-in effect. However, this model cannot predict the positive effect and was not designed for this purpose. Our proposed model is based on the diffusion equation with boundary conditions that take the location of the remaining edges into account and assumes that both illusions derive from the same visual mechanism. Using the model, we were able, for the first time, to obtain accurate perceptual predictions of the core properties of both the positive and negative effects. The suggested model supports the idea that both "conflicting" phenomena stem from the same visual mechanism. In addition, the model was able to predict additional "conflicting" phenomena from different locations of remaining edges, or the direction of the chromatic gradient.