Adjustment to a Shifted Set of Stimuli

Two category rating experiments are described. In both experiments, one group of subjects first judged the upper half and another group first judged the lower half of a graded series of 16 squares. After four blocks of trials with the practice series, both groups were shifted to the total series. Specific instructions to modify or not to modify scales were given before the shift. Experiment 1 manipulated the instructions as a between-subjects variable and Experiment 2 manipulated them as a within-subjects variable. The results indicate that (a) presentation period can serve the function of a context cue, that (b) two contexts for the same set of stimuli (squares) can coexist in memory, and that (c) subjects have a great deal of control over the weighting of these two contexts at any moment. The results support the hypothesis that the principles of scale formation are the principles of paired-associate learning. Numerous studies have shown that experience in a first series of stimuli affects the judgment of the second. Figure 1 shows the results of a typical experiment (unpublished) using a series of 16 squares whose widths varied from 12 mm to 102 mm, each step by 6 mm, in arithmetic progression. The participants were asked to judge each square with integers from 1 to 6 according to size. One group (H, n = 19) first saw the eight largest squares whereas the other group (L, n = 18) first saw the eight smallest. Each square appeared twice in a block with its occurrences separated by other squares. After four blocks of trials, both groups were shifted to the total range. The squares appeared, one at a time, in bright lines in the center of a 15-inch computer monitor. The intertrial interval was 3 s. Responses were to be made by pressing the appropriate number key. Response time was measured without the subjects’ knowing. To track the upward and downward movement of the scales, I calculated the neutral point of the series for each block of trials, that is, the average value of the squares assigned to the two middle categories (Categories 3 and 4). Figure 1 shows the resulting curves for Groups H and L. The vertical line marks the point of shift. One can see that the curves quickly approach each other but do not meet to the end of the experiment. After eight blocks of trials (i.e., 128 presentations) with the total series, the levels still differ by 1.6 stimulus units. The results suggest that the old impressions impede adjustment to the extended series. BLOCK 12 11 10 9 8 7 6 5 4 3 2 1 M E A N N P 14