Abstract

In a new objective task to measure the efficacy of cues to structure from motion, one of a large lexicon of shapes defined by a multidot display is presented on each trial, and the subject attempts to identify it. The cue of changing dot density is neither necessary nor sufficient for successful performance in this task. This was shown by either eliminating the cue or presenting the cue in isolation. When the maximum dot lifetime is reduced to two frames, performance remains high, showing that it is not necessary to track identifiable points for an effective kinetic depth effect (KDE)—KDE requires optic flow. Stimulus manipulations such as dot flicker and alternation of contrast sign, which impair the computation of optic flow, impair both the KDE task performance and performance in a motion-based texture segregation task (“Where is the odd motion?”). Finally, performance on the KDE task is reduced with shorter stimulus durations, but performance is still surprisingly good when only two frames are presented.

© 1987 Optical Society of America