Abstract

In a complex spatial-frequency discrimination task, the observer distinguishes between two stimuli, each of which contains two grating patches superimposed at orthogonal orientations. When the frequencies of both components increase (or both decrease) the combined change is more discriminable than when one increases and the other decreases. The result challenges vector models because the magnitude of the difference vector at each orientation is the same in both cases, yet performance is different. The result cannot easily be explained on the basis of overlapping bandwidths, pathway interactions, or noise or on the grounds that the viewer uses one component as a standard against which the other component is compared. Thus, the configurational effect probably does not reflect properties of the V1 representation by spatially tuned pathways; rather, it reflects properties of a later stage of processing.

© 1990 Optical Society of America