Abstract

Adelson and Movshon [Nature London 300, 523 (1982)] have proposed a two-stage model to solve the aperture problem; the velocities of the components of a complex target are separately extracted and then combined to form the percept of a rigidly moving object. Welch [ARVO (1988)] has presented data that support their model by illustrating two separate stages that can limit speed discrimination thresholds. What rules are used to combine motion information at the object-motion stage? We used plaids composed of two sinusoidal gratings: a mask grating which varied randomly in contrast, orientation, or speed and a test grating. Speed discrimination for the test grating was unaffected by contrast variations but was strongly impaired by speed or orientation variations of the masking grating. Since contrast does not affect the speed of the plaid pattern and speed and angle both do, speed discrimination must be based on the plaid pattern speed even when better information is available in the test grating alone. The degree of masking depends on how well the two gratings cohere to make a rigid plaid; the less they cohere the less masking is seen, with the paradoxical result that a high-contrast mask has less effect than a mask of moderate contrast.

© 1988 Optical Society of America