Effects of stimulus complexity on simple spatial discriminations

We have extended previous measurements on the detection of a target, differentiated in orientation or magnification from a number of identical reference elements. In these new experiments, we have used more than one class of reference element from which the target must be discriminated. For example, in the case of orientation discrimination, the reference elements are of two classes, which differ from each other in orientation, but are otherwise identical. Representatives of the two classes are present in equal numbers, and the target to be detected is presented at an orientation angle different from that of either class of reference element. Data for discrimination of target magnification or orientation are given for simple geometric elements, namely lines, squares and triangles. In nearly all cases, the characteristic time required for 50% probability of target detection, T1/2, is greater with two classes of reference elements than with a single class. In nearly all cases, T1/2 values for the mixed reference elements increase with the number of reference elements, N, corresponding to discrimination by serial processing. This remains the case even when detection is parallel (T1/2 independent of N) for either class of reference element used separately. We discuss the properties of the spatial discrimination mechanisms which give rise to these responses.