A hard illusion to break: the robustness of the Jastrow illusion across geometry, objects, and numerosity

Abstract Visual illusions are typically defined by specific geometric configurations, raising the question of how far their perceptual effects persist when those configurations are transformed or embedded in different representations. Here, we investigated the robustness of the Jastrow illusion across changes in geometry, object representation, and numerosity. Participants judged the relative size or numerosity of paired stimuli. In control trials, the two alternatives differed physically in area or numerosity, whereas in illusory trials they were physically identical and arranged according to the Jastrow illusion, in which the lower of two vertically displaced identical stimuli is typically perceived as larger. Illusion-consistent responding was observed across several geometric, object-based, and numerical configurations, although its prevalence varied considerably across stimulus types. Systematic manipulations of curvature and symmetry suggested that geometric properties modulated the probability of illusion-consistent responses without necessarily eliminating the bias. Likewise, the effects of removing internal details varied across object categories, with illusion-consistent responding remaining evident for birds and fish but not significantly exceeding chance for outline faces. However, these differences may partly reflect the distinct visual information preserved by each category’s external contours. Response times complemented the accuracy findings, showing that response-time differences between control and illusory trials also varied systematically with stimulus properties. Together, these findings suggest that illusion-consistent responding can persist across substantial transformations of the original configuration, while its occurrence and consistency depend on geometric and representational properties.