Herbert Wertheim School of Optometry and Vision Science and the Helen Wills Neuroscience Institute, University of California, Berkeley, Berkeley, CA, 94720-2020, USA.
Sci Rep. 2024 Mar 22;14(1):6863. doi: 10.1038/s41598-024-57406-2.
The precision of stereopsis and vergence are ultimately limited by internal binocular disparity noise. Here we propose an equivalent noise model with both global and local internal disparity noises to provide a unified explanation of both absolute and relative disparity thresholds. To test this model, we developed a psychophysical procedure to measure the equivalent internal disparity noise by adding external disparity noise to random-Gabor-patch stereograms. We used the method of constant stimuli to measure the minimum and maximum disparity thresholds (Dmin and Dmax) for both absolute and relative disparity. Consistent with previous studies, we found that Dmin thresholds are substantially worse for absolute disparity than for relative disparity. We tested three relative disparity mechanisms: (1) the difference between the monocular separations of targets projecting to the two eyes; (2) the direct measurement of relative disparity; and (3) the difference of absolute disparities of targets. Computing the difference of absolute disparities when detecting relative disparity, Mechanism 3 cancels global noise, resulting in a much lower relative Dmin threshold, and provides a reasonable fit to the experimental data. We also found that the presence of as much as 2400 arcsec of external disparity noise does not appear to affect the Dmax threshold. This observation suggests that Dmax is implicated in a mechanism that disregards the disparity variance of individual items, relying instead on the average disparity across all items, supporting the depth model proposed in our previous study (Ding & Levi, 2021), which posits distinct mechanisms governing Dmin and Dmax thresholds.
双眼视差的精确性最终受到内部双眼视差噪声的限制。在这里，我们提出了一个具有全局和局部内部视差噪声的等效噪声模型，为绝对和相对视差阈值提供了统一的解释。为了测试这个模型，我们开发了一种心理物理学程序，通过向随机 Gabor 补丁立体图像添加外部视差噪声来测量等效的内部视差噪声。我们使用恒定刺激法来测量绝对和相对视差的最小和最大视差阈值（Dmin 和 Dmax）。与之前的研究一致，我们发现 Dmin 阈值对于绝对视差明显比相对视差差。我们测试了三种相对视差机制：（1）投射到两只眼睛的目标的单眼分离之间的差异；（2）相对视差的直接测量；和（3）目标的绝对视差的差异。在检测相对视差时计算绝对视差的差异，机制 3 会消除全局噪声，从而导致相对 Dmin 阈值大大降低，并与实验数据合理吻合。我们还发现，即使存在高达 2400 弧秒的外部视差噪声，也似乎不会影响 Dmax 阈值。这一观察结果表明，Dmax 涉及一种机制，该机制忽略了单个项目的视差变化，而是依赖于所有项目的平均视差，这支持了我们之前研究中提出的深度模型（Ding & Levi，2021），该模型假设了控制 Dmin 和 Dmax 阈值的不同机制。
