Abstract
Purpose: To study the effects of coherence and disparity on motion transparency processing. Previous work has shown that for small direction differences in transparent displays, only a single motion vector is perceived. Further, disparity has been suggested to enhance the ability to represent two motions independently. We examine the ability of disparity to segregate transparent motion as a function of motion coherence and direction difference. Methods: In two psychophysical tasks we use random dot stereograms to quantify the ability of observers to discriminate one vs. two directions of overlapping dot motions. Stimuli were presented for 481ms in a 6deg aperture at a speed of 4deg/sec. In both tasks, discrimination thresholds (79% correct) were obtained using a 2TAFC adaptive staircase paradigm, requiring observers to select the stimulus containing transparent motion. In the first test, coherence thresholds were obtained as a function of angle difference [0deg,180deg] and disparity [-12′,12′]. In the second test, angle difference thresholds were obtained over the same disparity range for fixed coherence levels. Results: In the first test, we show that performance is not significantly different for angles greater than 45deg (p=.44), but that disparity does cause a significant difference over this range (p=.034). For angles less than 45deg, thresholds rise rapidly, and the benefit gained from disparity is less important relative to direction differences. This effect is illustrated in the second task, where for large signal strengths (e.g., 56% coherence), angle difference thresholds do not show a significant trend across disparities (p=.068). Conclusion: Our results show an ability to represent motions differing by less than 10deg, smaller than previous suggestions. Further, broad disparity (relative to direction) tuning may be sufficient to account for these results, without requiring inhibition among similar detectors.
Supported by NIH grant EY-2R01-07861-13 to LMV.