Abstract

The true velocity of a straight contour moving behind an aperture cannot be determined (the aperture problem), but the true velocity of two or more contours with different orientations belonging to the same rigid object can be found. My results support the two-stage motion process proposed by Adelson and Movshon (Nature, 1982), which first extracts oriented-component motion and then combines that information to determine pattern or object velocity. A plaid pattern formed of two moving gratings is an interesting stimulus for velocity processing and coherence experiments because observers may see a rigid, coherent plaid object moving in one direction, or they may see two gratings moving different directions and sliding transparently over one another, i.e., two different solutions to the aperture problem. An observer's ability to respond to grating or plaid speed depends on whether coherence or transparency is seen; when the gratings cohere, only the plaid speed is perceptually available, and when the gratings appear transparent, only the gratingspeeds are available. The gratings must be similar to each other in contrast, spatial frequency, speed, color, and depth to be perceived as coherent. Thus, there must be a decision process based on component similarity that determines whether gratings, and presumably other oriented components, are part of the same pattern or object.

© 1990 Optical Society of America