Jain Siddharth
EECS Department, University of California, Berkeley, California, United States of America.
PLoS One. 2009;4(2):e4536. doi: 10.1371/journal.pone.0004536. Epub 2009 Feb 19.
The performance of Watson & Ahumada's model of human visual motion sensing is compared against human psychophysical performance. The stimulus consists of random dots undergoing rotary motion, displayed in a circular annulus. The model matches psychophysical observer performance with respect to most parameters. It is able to replicate some key psychophysical findings such as invariance of observer performance to dot density in the display, and decrease of observer performance with frame duration of the display.Associated with the concept of rotary motion is the notion of a center about which rotation occurs. One might think that for accurate estimation of rotary motion in the display, this center must be accurately known. A simple vector analysis reveals that this need not be the case. Numerical simulations confirm this result, and may explain the position invariance of MST(d) cells. Position invariance is the experimental finding that rotary motion sensitive cells are insensitive to where in their receptive field rotation occurs.When all the dots in the display are randomly drawn from a uniform distribution, illusory rotary motion is perceived. This case was investigated by Rose & Blake previously, who termed the illusory rotary motion the omega effect. Two important experimental findings are reported concerning this effect. First, although the display of random dots evokes perception of rotary motion, the direction of motion perceived does not depend on what dot pattern is shown. Second, the time interval between spontaneous flips in perceived direction is lognormally distributed (mode approximately 2 s). These findings suggest the omega effect fits in the category of a typical bistable illusion, and therefore the processes that give rise to this illusion may be the same processes that underlie much of other bistable phenomenon.
将沃森和阿胡马达的人类视觉运动感知模型的性能与人类心理物理学性能进行了比较。刺激由在圆形环带中显示的进行旋转运动的随机点组成。该模型在大多数参数方面与心理物理学观察者的性能相匹配。它能够复制一些关键的心理物理学发现，例如观察者性能对显示中点密度的不变性，以及观察者性能随显示帧持续时间的减少。与旋转运动概念相关的是旋转所围绕的中心的概念。人们可能会认为，为了准确估计显示中的旋转运动，必须准确知道这个中心。简单的矢量分析表明情况并非如此。数值模拟证实了这一结果，并可能解释了MST(d)细胞的位置不变性。位置不变性是指旋转运动敏感细胞对其感受野中旋转发生的位置不敏感这一实验发现。当显示中的所有点都从均匀分布中随机抽取时，会感知到虚幻的旋转运动。罗斯和布莱克之前对这种情况进行了研究，他们将这种虚幻的旋转运动称为欧米伽效应。关于这种效应报告了两个重要的实验发现。首先，尽管随机点的显示会引发旋转运动的感知，但所感知到的运动方向并不取决于显示的点图案。其次，感知方向的自发翻转之间的时间间隔呈对数正态分布（众数约为2秒）。这些发现表明欧米伽效应属于典型的双稳态错觉类别，因此产生这种错觉的过程可能与构成许多其他双稳态现象基础的过程相同。
