Abstract

Discriminations between highly similar, complex spatial patterns often show configuration effects: two patterns that differ from each other by a single transformation (such as rotation or magnification of all components in one pattern) are more accurately discriminated than two patterns in which transformations of the same magnitude have been applied differentially to the components of one pattern (such as one component magnified and another minified). One hypothesis is that contrast related nonlinearities produce distortion products that aid discrimination in the first case, but not in the second case. This hypothesis was tested by measuring discrimination for the two types of pattern, plus appropriate single component and masking controls, as the contrast of the individual components was varied from 2 to 20 times the detection threshold of the component when viewed alone. Comparison of performance on the complex patterns with performance on single- component control patterns indicates that sum or difference products do not aid discrimination.

© 1991 Optical Society of America