Abstract
The "Rotating Snakes" illusion is based on patterns with repeated asymmetric luminance steps, arranged in concentric bands forming a "snake wheel". In the underlying luminance sequence "black – darker grey (g1) – white – lighter grey (g2)" we varied the luminances of g1 and g2, and measured illusion strength via a nulling task: On a gamma-corrected VDU, repetitive saccades were performed passing a "snake wheel" that physically rotated, and the participants adjusted rotation direction and speed until a stationary percept obtained. We found that our participants performed perceptual nulling (in the range of ± 1°/s) of the seeming rotation quite reliably and convincingly; when the gaze was directed to the center of the wheel after null-adjustment, the actual rotation could be quite surprising (depending on the levels [g1, g2]). Typical settings for [g1, g2] as measured from images by A. Kitaoka, are around [20%, 60%]. Indeed, we found a marked illusion in an "island" region [g1≈{0–25%}, g2≈{20–75%}]. Unexpectedly, we detected a second island around [60%, 90%] with opposite rotation direction, although the relative order of g1 and g2 stayed the same. Furthermore, there was also some rotation illusion when altogether only 3 luminance levels were visible, e.g.: 0–0–50–100%. These quantitative measurements of illusion strength have revealed new phenomena in this illusion, which challenge the two explanatory models put forth so far.
Meeting abstract presented at VSS 2014