Purchase this article with an account.
Shin'ya Nishida, Kaoru Amano, Mark Edwards, David R. Badcock; Global motion with multiple Gabors - A tool to investigate motion integration across orientation and space. Journal of Vision 2006;6(6):1084. doi: 10.1167/6.6.1084.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
There are two streams of motion integration research. One, using plaid stimuli, examines how the visual system integrates ambiguous 1D motions across orientations to solve the aperture problem. The other examines how the visual system integrates local motions over space into coherent global motion using stimuli composed of independently moving 2D dots. Both types of motion integration are suggested to take place somewhere between V1 and MT/V5, but it remains open whether 1D motion signals are integrated over space (one-stage integration), or 1D motion signals are integrated locally, then the resulting 2D signals are integrated over space (2-stage integration). To test this, we introduced a novel global motion stimulus comprised of numerous Gabor elements (drifting sine-waves windowed by stationary Gaussians). While the orientations of Gabors were randomly determined, their drifting velocities were made consistent with a given 2D velocity: observers saw a rigid motion of the whole pattern. The global motion remained detectable even when more than half of the Gabor elements were given random 2D velocities (noise). These observations appear to support one-stage integration. However, we also found that superposition of noise 1D motions on 1D signal motions (which turned the local elements into 2D plaids) severely impaired global motion perception. This implies that local 1D motion integration has priority over global integration, as expected from two-stage integration. Our results suggest that motion integration process has great flexibility, presumably for the purpose of correctly estimating complex optic flows in natural scenes.
This PDF is available to Subscribers Only