Factors influencing the ability to detect motion reversals in rotation simulations

Five experiments probed the conditions under which observers fail to report instantaneous reversals in the direction of motion of pixels that define the rotation of a transparent sphere or plane. Our results showed that the extent to which rotation reversals were not reported depended upon whether observers used strict or lax criteria to make their judgments, the degree of perspective present in the rotation simulations, and the percentage of pixels that actually reversed direction. Furthermore, we found failures to report rotation reversals both with stimuli whose pixels were confined to smooth surfaces and scattered within volumes. Reversal detection with planar stimuli, unlike sphere stimuli, depended upon the orientation of the stimulus at the moment of reversal. Treue et al. (1995) postulated a surface-interpolation process as the explanation for the apparent insensitivity of observers to such reversals. However, we suggest that other stages of processing (e.g. a structure-from-motion process) are required to account for these results.