Abstract
Observers discriminated between two second-order patterns in a) control conditions (vertical contrast-modulated (CM) sinusoids); b) masking conditions (orthogonal CM mask) and c) two two-cue conditions that test for higher-level summing or differencing circuits. Discriminations were based on small differences in spatial frequency, contrast or orientation. Spatial frequency and contrast thresholds were similar to those obtained with first-order stimuli, but orientation thresholds were nearly doubled. Contrast and spatial frequency judgments suffered from the presence of an orthogonal mask; orientation judgments were unaffected. Spatial frequency judgments indicated the presence of a higher-level summing circuit. A summing circuit was indicated in two observers for contrast judgments, but two others showed no sign of one. Preliminary data suggested no summing circuit when judgments are based on orientation. In order to test the quantitative nature of the summation, we tested two meaningful values of p in the Minkowski metric (d1p + d1p)1/p. The first was a value of 1, denoting a linear summing circuit. The second was 2, indicative of optimal integration from two independent sources at the decision stage of processing if added; integration of energy if both added and subtracted. No one model consistently predicted the data.