Abstract
Binocular depth perception depends on the matching of features in the two eyes. Matching can be based upon first-order, luminance defined features, or second-order features such as the contrast-envelope. Binocular matching of the contrast envelope allows depth perception to proceed even when the contents of the contrast-envelope (i.e. the carrier) contain many false matches. However, the robustness of this matching process has only been evaluated for stereoacuity. In the present work, we studied the robustness of contrast-envelope matching at larger pedestal disparities. This was done in two stages. In the first stage, we found that depth discrimination using a shifted contrast envelope was relatively easy for a large range of pedestal disparities. By comparison, when the envelope was not shifted and only the carrier had disparity, then depth discrimination was much harder at the same range of pedestal disparities. In all stimuli, the carrier consisted of bandpass filtered noise. In the second stage, we produced a new type of stimulus in which the contrast envelope and the carrier had different disparities. On the hypothesis that the shifted envelope would be used to compute disparity and that the carrier would be ignored, performance should be relatively easy for a large range of pedestal disparities. On the alternative hypothesis, matching of the carrier would be more important in computing disparity, and performance should be relatively lower. We found that the data supported the alternative hypothesis, showing that the carrier could not be ignored in computing disparity. These findings were especially evident for large pedestal disparities.