Gross misperceptions in the perceived trajectories of moving dots.

We report illusions we observed while investigating thresholds for detecting a deviation in one out of several trajectories of dots moving in straight lines (Tripathy and Barrett 2003). The deviating trajectory consisted of two straight-line segments, with the deviation occurring in the middle of the trajectory, at the intersection of the two segments. Robust distortions were noticed in the shape of the deviating trajectory, in spite of observers knowing exactly when and where the deviation occurred, and the actual shape of the trajectory. The stimulus consisted of several dots undergoing apparent motion along linear trajectories (figure 1). All dots moved with equal average speed (typically 16 deg s À1 , though speeds from 1 to 16 deg s À1 were tested) and reached the midline of the screen (indicated by vertical markers) at exactly the same instant of time. One of the dots/ trajectories (target dot/trajectory) deviated at the midline, while the others continued along straight lines without deviating (distractor dots/trajectories). Observers reported what the perceived shape of the target trajectory was, and the location on the screen where the deviation was perceived. All observers perceived gross distortions in the shape of the target trajectory. When the angle of deviation of the target trajectory was much greater than the threshold for detecting the deviation, the following distortions were perceived: (i) The perceived location of the deviation was shifted spatially in the direction of motion. (ii) Immediately following the deviation, observers overestimated the angle of deviation. Figure 1. Schematic representation of the typical stimulus arrangement which leads to the illusory percepts described in the text. Three left-to-right dot trajectories are shown, one with a deviation (the target trajectory) and two without a deviation (distractor trajectories). The path of the target dot/trajectory was not perceived veridically (see figure 2 and text for details). At a viewing distance of 1.29 m the screen subtended 13.3 deg69.6 deg. Path lengths were varied between 1 and 12 deg. The angle of deviation was varied between threshold deviation and 788, while the speed of the dot trajectories ranged from 1 to 16 deg s À1 , and the number of distractor dots/trajectories ranged from 0 to 9. Each dot had a square profile and subtended 5 min. Regardless of the angle of the trajectory, the edges of the squares were oriented horizontally and vertically (unlike in the figure).