Abstract
In the classic Ponzo illusion the perceived size of a stimulus is affected by lines in the background suggesting a linear perspective in depth. A stimulus or object close to the vanishing point appears further away and therefore larger than an object of the same size placed in the foreground. However, introducing a static or moving shadow cast by the object creates the impression that the object floats above the ground, thereby changing perceived size (Yonas, Goldsmith & Hallstrom, 1978; Kersten, Mamassian & Knill, 1997). Here we investigate how strong a cast shadow of a stimulus ball affects perceived size. A total of N=38 observers participated in an online study with a 4 by 2 within-subjects design. We tested the effect of four vertical shadow offsets (0, 46, 92, and 184 pixels) and two positions of the reference ball (64, and 164 pixels). An analysis of variance (ANOVA) with repeated measurement revealed that shadow offset (F(3,111)=13.17, p<.0001, partial eta squared=0.263) as well as reference position (F(1,37)=30.08, p<.0001, partial eta squared=0.448) had a statistically significant effect on perceived size. The interaction was not significant (F(3,111)<1). The average observer seems to base their size judgments entirely on the relative position of the cast shadows because this disambiguates the location of the stimulus and reference ball in the 3D scene. A linear mixed model (package lme4 in R) showed considerable individual variability for both effects but confirmed that vertical offsets of cast shadows systematically changed the Ponzo size illusion. We conclude that the relative position of stimulus and reference shadow rather than the relative position of stimulus and reference ball determines perceived size.