According to scale expansion theory two distinct sources of angular information are used to compute distance along the ground and shape on the ground: (1) angular (or gaze) declination is used to estimate egocentric ground distance, and (2) optical slant information is used to compute shape (Li & Durgin, 2010, 2012). Texture and binocular information both contribute to optical slant estimation when gaze is forward. Here we ask whether gaze declination also contributes to shape estimation on the ground plane. On the one hand gaze declination is an excellent estimate of the optical slant of a fixated surface on level ground. On the other hand, estimates of gaze declination and of optical slant need to be used conjointly to estimate geographical slant, so using one to estimate the other would seem inadvisable. To test this experimentally, we assumed that if gaze declination is used to help recover shape on level ground, then shape constancy should be enhanced for objects on level rather than slanted ground. Participants judged the aspect ratios of L-shaped configurations of white balls presented either on level ground (gaze declination informative) or on one of three hill surfaces viewed at eye level (gaze declination not informative) in an outdoor environment. The optical slants (6째, 22째 and 35째) and viewing distances used were matched in the two conditions. The results indicated partial shape constancy failure in all conditions. Shape constancy was better for smaller viewing distances and for larger optical slants, but there were no differences in mean judged aspect ratios between matched level ground and hill conditions. Nor was the variance of shape estimation reduced on level ground. These results suggest that the gaze declination information that is available when observing configurations on level ground does not contribute to the perception of shape.
Meeting abstract presented at VSS 2014