Abstract
Abstract: :
Purpose: It is commonly reported that the perceived direction of moving Type 2 plaids is biased away from the veridical intersection-of-constraints (IOC) direction and toward the vector-sum direction. Spatial frequency (SF) modulates the strength of this bias. One explanation is that high SF creates more local features in the stimulus for a putative feature tracker to process, and that the feature tracker is responsible for veridical direction perception mimicked by IOC predictions (Alais et al., 1994, 1997). Another explanation is that high SF creates more reliable motion signals, thus yielding more veridical perception. A Bayesian computational model is consistent with this latter explanation (Weiss & Adelson, 1998; Cobo-Lewis & Smallwood, 2002). Studies in our laboratory have previously confirmed the predictions of the Bayesian model in the context of missing-fundamental plaids, with results that confirmed the effect of SF while ruling out feature-tracking explanations. To establish converging evidence, we now seek an SF effect while silencing feature trackers via a static pedestal instead of via the missing-fundamental illusion. Methods: Naive subjects viewed sinusoidal Type 2 plaids whose component gratings drifted smoothly in directions separated by 15° and whose speeds differed by a factor of √1.5. On half the trials, the drifting image was linearly superimposed on a higher-contrast static version of the same image. The presence of the pedestal causes local feature information not to yield useful motion signals. Subjects pointed an arrow in the direction of perceived drift. Results: Perceived motion was intermediate between vector-sum and veridical IOC directions. Increasing SF decreased the bias toward the vector-sum direction. The presence of the pedestal had no effect on perceived direction, not did it modulate the effect of SF. Conclusions: A Bayesian computational model of motion extraction suggests that increasing SF can decrease the bias toward the vector-sum direction by increasing the reliability of the motion signals. A feature-tracking mechanism is not necessary to explain our results.
Keywords: motion-2D • temporal vision • computational modeling