Abstract
In the Barber-Pole Illusion (BPI), a diagonally moving grating is perceived as moving vertically because of the narrow, vertical, rectangular shape of the window through which it is viewed. This strong shape-motion interaction persists through a wide range of parametric variations in the shape of the window, the spatial and temporal frequencies of the moving grating, the contrast of the moving grating, complex variations in the composition of the grating and window shape, and the duration of viewing. It is widely believed that end-stop-feature (third-order) motion computations determine the BPI, and that Fourier motion-energy (first-order) computations determine failures of the BPI. Here we show that the BPI is more complex: (1) In a wide variety of conditions, weak-feature stimuli (extremely fast, low contrast gratings, 21.5 Hz, 4% contrast) that stimulate only the Fourier (first-order) motion system actually produce a slightly better BPI illusion than classical strong-feature gratings (2.75 Hz, 32% contrast). (2) Reverse-phi barberpole stimuli are seen exclusively in the forward (feature, third-order) BPI direction when presented at 2.75 Hz and exclusively in the opposite (Fourier, first-order) BPI direction at 21.5 Hz. (3) The BPI in barber poles with scalloped edges (Badcock, McKendrick, and Ma-Wyatt, VisRes, 2003) is much weaker than in normal straight-edge barber poles for 2.75 Hz stimuli but not in 21.5 Hz stimuli. Conclusions: Both first-order and third-order stimuli produce strong BPIs. In some stimuli, local Fourier motion-energy (first-order) produces the BPI via a subsequent motion-path-integration computation (Peng, Chubb,and Sperling, JOV, 2014); in other stimuli, the BPI is determined by various feature (third-order) motion inputs; in most stimuli, the BPI involves combinations of both. High temporal and spatial frequencies, low contrast, and peripheral viewing, all favor the first-order motion-path-integration computation; low spatio-temporal frequencies, high contrast, and foveal viewing favor third-order motion computations.
Meeting abstract presented at VSS 2015