Abstract
Bent paperclips with variable depth rotations were displayed either with normal stereopsis, reversed-stereopsis, or without binocular disparity in a sequential matching task. In Exp. 1, where display mode was manipulated as a within-subject factor, the results from the first half of the experiment (trials 1–600) replicated previous findings by Burke (2005), with a weaker depth rotation cost on behavioral performance on “same” trials with normal stereoscopic displays than with reversed stereopsis or null binocular disparity. However, this benefit from normal stereopsis turned to a handicap in the second half of the experiment (trials 601–1200). In Exp. 2, display mode was again manipulated in a within-subject manner but the first (trials 1–600) and second (trials 601–1200) halves of the experiment used distinct sets of stimuli. The negative impact of the depth rotation of stimuli on “same” trials remained weaker with normal stereo than with reversed stereo or null binocular disparity through the first and second halves of the experiment. The same findings occurred in Exp. 3, where display mode was manipulated as a between-subjects factor. The present results suggest that by default, stereoscopic information contributes to visual shape perception and to shape constancy. However, when stereo has a poor information value for specific stimuli, such as in Exp. 1, this contribution may be cancelled, turning potentially useful and valid binocular disparity signals into a source of distraction that impairs performance. In this regard, we note that almost all the literature on shape constancy rests on studies using flat computer screens to display stimuli that participants are supposed to interpret that as 3-D despite a lack of binocular disparity. This may constitute a significant artifact altering the validity of these past studies.