Abstract
The brain integrates a range of depth cues to estimate depth. Psychophysical results show this improves discrimination; e.g., judgments are more precise when disparity specifies depth in combination with cues such as texture, motion and shading. Here, we test for the cortical circuits involved in representing depth structure from integrated cues. We presented participants (n=15) with random dot stereograms in which convex (‘bump’) and concave (‘crater’) surfaces were depicted using (i) binocular disparity, (ii) Blinn-Phong shading and (iii) disparity and shading in combination. Further, we included a control condition in which convexities and concavities were specified by binocular disparity, and dots in the upper and lower portions of the stimuli were given different luminance levels (binary shading). This ensured differences in the spatial distribution of luminance in the stimuli, but did not contribute to the impression of depth structure. We measured psychophysical judgments of these shapes, finding evidence of enhanced depth estimates when disparity and shading simultaneously signalled depth. We then measured fMRI responses to these stimuli, using a blocked design and measuring responses in independently-localized regions of interest. We analysed our data by measuring the prediction performance of a support vector machine trained to discriminate surface shape (convex vs. concave). In cortical area V3B/KO we find that prediction performance improves significantly when depth structure is concurrently specified by disparity and shading, and this improvement exceeds the minimum bound expected for integration (quadratic summation prediction from component cue accuracies). Moreover, this result is specific to combined cue stimuli, with binary shaded stimuli supporting prediction performance comparable to that of disparity-defined surfaces. Our results are consistent with recent work showing the integration of disparity and motion in area V3B/KO and suggest that this area is a crucial cortical locus integrating depth cues in the human brain.
Meeting abstract presented at VSS 2012