Why Do Some People Take a Long Time to See Complex Random-Dot Stereograms?

Abstract

Some observers take a long time to see the appropriate surface shapes depicted in complex random-dot stereograms (RDSs) whereas others report seeing them quickly. We have hypothesised that this may be due in part to some observers being more affected than others by the cue conflicts that are typically present in RDSs, for which disparity is often at odds with accommodation, texture, and outline shape cues. We report two experiments in which we investigated whether a measure of dominance of texture/outline cues over disparity cues in stereograms of ridges could predict latencies to see RDSs of ‘wedding cakes’ viewed from above. Latencies were measured for correctly reporting the number of layers in the wedding cakes. The ridge cue dominance index was obtained by using amplitude judgements of convex parabolic vertical and horizontal ridge stereograms embodying texture/outline vs disparity cue conflicts (Buckley and Frisby, 1993 Vision Research33 919 – 934). In the first experiment ( N=38) disparity cues in the wedding cakes were depicted with inconsistent dot texture elements, ie the dots signalled flatness as there were no density differences between the cake layers. Observers giving greater weight to texture/outline in the vertical-ridge amplitude task showed significantly longer latencies to the wedding cakes. In a second experiment with different observers ( N=35), 18 were shown wedding cakes with textures consistent with disparity (diamonds drawn with correct perspective and density) and 17 saw the same disparity cues carried by inconsistent textures (diamonds signalling flatness). Large and significant correlations were found between the cue dominance index from vertical ridges and the latencies to wedding cakes depicted with inconsistent textures: observers with higher texture/outline cue dominance showed longer latencies. This result was remarkable given the lack of any readily apparent differences between the consistent and inconsistent textures in the wedding-cake stimuli. Equivalent correlations for wedding cakes depicted with consistent textures were small and nonsignificant. We conclude that individual differences in latencies to complex RDSs can be partly explained by differences in observers' susceptibility to texture/disparity cue conflicts embodied in those RDSs.