Abstract
Transparent motion can be perceived when two objects move within the same region of space simultaneously. Though it is clear that the global-motion stage is crucial to transparent-motion detection, the exact means by which global-motion activity is decoded to yield multiple signal directions remains elusive. An important constraint on this process is the smallest angular separation (Δθ) that gives rise to transparency. However, there is considerable variation in the literature concerning this value. It is also unclear whether Δθ thresholds would be affected by the oblique effect that influences unidirectional processing. Observers in our task were required to discriminate between transparent-motion and global-flow stimuli. The latter contained a range of directions equal to the angle between the comparison transparent-motion signals. This ensured that performance was based on detection of transparency rather than relative motion. Thresholds were lowest around the cardinal axes, particularly horizontal, where the minimum Δθ was around 25deg. This increased to around 35deg for the oblique axes. Thresholds were elevated when signal intensities were lowered, though the oblique effect remained. Direction-specific analysis of thresholds obtained with both signal intensity levels indicates that this oblique effect depends upon the mean direction, rather than the component directions of transparent motion. This suggests that transparent-motion detection relies heavily upon the central aspect of the population activity distribution. That is, global-motion units with preferred directions between the two component directions may offer the most insight into the mechanisms of transparent-motion detection.