Perceptual encoding and decision strategies for integral dimensions.

These three experiments employed rectangles in stimulus identification tasks. Consistent with the stimulus set used by Weintraub, the rectangles were generated by modifying a square. Across experiments, the number of stimulus/response alternatives was varied (two-, three-, and four-choice tasks). In the two-choice task, redundancy gain for the positively correlated set was just as large as for the negatively correlated set. In contrast, reaction time ws faster for the negatively correlated set than for the positively correlated set in the three-choice task (after extended practice) and in the four-choice task. Considered in the context of previous research, the data support two conclusions. First, the initial perceptual processing of rectangles is accomplished by separate dimensional analyzers operating in parallel. Second, observers adopt a different decision strategy for the negatively correlated set than for the positively correlated and the single dimension sets when the number of stimulus/response alternatives is increased.