Abstract
A continuously moving object is mislocalized in the forward direction when another object is flashed briefly in the same location (flash-lag effect: Nijhawan, 1994). We asked whether abrupt transitions, not flashes, lead to the same effect. Two identical balls (0.53 deg. dia., nearly 100% contrast) start moving simultaneously (17.8 deg./s), the top following an upright V drawn left to right (the legs are orthogonal), the bottom an inverted V, drawn right to left. The trajectories never cross and the turn points are 5.7 deg. from the screen center. Observers judge the horizontal misalignment of the turn points whose physical positions are randomly varied using the method of constant stimuli. a) When the legs of the V are of the same length (3.7 deg.), 2/5 observers perceive the post-turn legs misaligned forward, 3/5 perceive them misaligned behind. The overall group mean misalignment is 0.23 deg. behind. b) When the pre-turn leg length is reduced (0.75 deg.) and the post-turn leg length increased (12.0 deg.), the backward misalignment increases (p < 0.0001; binomial test) to 0.50 deg. c) In a control experiment, we simultaneously flash both complete trajectories. There is no bias in observers' alignment judgments, implying that the effect does not have a static analog and must therefore be motion based. d) Degrading the post-turn trajectory display (fading in from black to white) diminishes the backward bias, implying that the Frohlich effect is unlikely to explain the mechanism behind the elbow illusion. Moreover, the direction of the bias—behind the turn-point—indicates a qualitatively different mechanism from other motion-based perceptual effects such as the flash-lag effect and the Frohlich effect.
B.R.S. was supported by a Division of Biology, Caltech fellowship.