Global internal disparity noise increases with rising levels of disparity pedestal

Poster Presentation 56.429: Tuesday, May 21, 2024, 2:45 – 6:45 pm, Pavilion
Session: Binocular Vision: Disparity, stereopsis and suppression

Jian Ding1 (), Michelle Y. Ma1, Hilary H. Lu1, Dennis M. Levi1; 1University of California, Berkeley

In our previous study (Ding et al., 2023, VSS), we described an equivalent noise model incorporating both global and local internal disparity noises to provide a unified explanation of absolute and relative minimum disparity thresholds. The model predicts smaller thresholds for relative disparity compared to absolute disparity, as global noise present in absolute disparity detection is canceled when detecting relative disparity. However, the model was only tested within the vicinity of the zero-disparity plane. To extend the model beyond the zero-disparity plane, we developed psychophysical procedures involving the introduction of external disparity noise to random-Gabor-patch (RGP) stereograms to measure absolute and relative minimum disparity thresholds at different disparity pedestals. To measure absolute disparity thresholds, two RGP stereograms with external disparity noise were presented in two temporal intervals, one with a mean disparity greater than the pedestal disparity and the other smaller. Observers discerned which interval appeared closer. For relative disparity thresholds, we presented an RGP stereogram with different mean disparities (one larger and one smaller than the pedestal disparity) in the top and bottom halves, along with external disparity noise. Observers identified whether the top or bottom of the stereogram appeared closer. We measured disparity thresholds across 7 pedestals and 6 external noise conditions for both absolute and relative disparity detection, utilizing the constant stimulus method. Consistent with previous studies, our results demonstrated an elevation in both absolute and relative disparity thresholds with increasing disparity pedestals. The proposed model provides a unified explanation for all data sets. Our modeling shows that global disparity noise increases with rising levels of disparity pedestal, contributing to increased absolute disparity thresholds. In contrast, local disparity noise seems to remain unaffected by disparity pedestals. The increase in relative disparity thresholds with disparity pedestals is more likely attributed to a decline in detection efficiency.

Acknowledgements: NIH RO1EY030544 and NIH RO1EY020976