Purpose: to determine whether the visual system uses information past 2-frames in determining surface shape from motion information. Using additional frames to (for example) estimate image accelerations would in principle allow richer surface shape representations to be derived. Stimuli and task: simulated ellipsoids were used, defined by dots randomly distributed across the surface. Projection was orthographic. Simulated surface protrusion was varied in six steps; this included a spherical surface, along with surfaces more compressed than a sphere and surfaces more elongated than a sphere. These simulated half-ellipsoids were rotated side to side, five degrees per sweep (object rotation). 2, 4 and 8 frame (per sweep) animations were used. Each sweep lasted 400 ms. Observers then chose which of 8 cross sections most closely matched the cross section of the simulated surface; cross sections 2-7 corresponded to the cross sections of the stimuli shown, while cross sections 1 and 8 were extremes not actually presented. Stimuli were shown through a simulated aperture, spanning 7 degrees. Viewing was monocular. Results: for n=7 observers, there was a highly significant effect of frame number on perceived depth, as indicated by the height of the cross section chosen (p < .01). The frame number X stimulus-protrusion interaction was also highly significant (p <.01). Paired t-tests on slopes revealed no difference between the 4 frame and the 8 frame conditions (both with positive slopes). But there was a significant difference between the 2-frame condition and the 4-frame condition (p <.01), and there was a significant difference between the 2-frame condition and the 8-frame condition (p <.01). Conclusions: subjects used information available past 2 frames. This may include image acceleration information; this hypothesis will be explored in subsequent experiments.
Meeting abstract presented at VSS 2012