You don’t fool me! Disclosing visual illusions via Wilson-Cowan-type dynamics

In this talk, I will discuss the reproduction of visual perception phenomena, specifically visual illusions, by means of Wilson-Cowan-type models of neuronal dynamics. In particular, we show that the Wilson-Cowan equations can reproduce a number of brightness and orientation-dependent illusions, and that the latter type of illusions require that the neuronal dynamics equations consider explicitly the orientation, coherently with the architecture of V1. I’ll then focus on a slightly different modification of the Wilson-Cowan equations which makes such model consistent with the efficient representation principle, that is it can be interpreted as the gradient flow of a suitable functional. This shows that this model minimises redundant information, making it capable of replicating more visual illusions than the original Wilson-Cowan formulation.