Approaches to understanding visual illusions

Abstract

Visual illusions can be broadly categorized into three types: physiological/pathological, perceptual, and ambiguous (bistable/multistable). Whether seen as malfunctions or natural consequences of the normal operations of the visual system, these deviations from the veridical perception of physical sources of sensory inputs provide rich test beds for the evaluation of theories of perception. Here we consider the strengths and weaknesses of three theoretical frameworks: conventional feature detection, empirical ranking and Bayesian decision, in accounting for each type of illusion.