Relation to Prior Work: On the one hand, the current text is too focused on the contributions of Purves et al. Fantastic papers of Purves et al. are very inspiring, but similar ideas were suggested before and this has to be acknowledged. Specifically, Horace Barlow proposed that matching the sensors to the statistics of the stimuli in order to reduce redundancy in the response (using a sort of linear ICA) could lead to visual illusions [Barlow90]. More generally, redundancy reduction or information maximization is connected to (nonlinear) Gaussianization and uniformization transforms. Therefore, more recent uniformization techniques such as Sequential Principal Curves Analysis (SPCA) have been proposed to explain the emergence of illusions when environment is changed [Lapàrra15]. Nonlinear transforms for error minimization [Twer01,McLeod03] may also be achieved by SPCA, thus giving alternative statistical explanation for illusions [Laparra15].