Symmetry of control by exclusion in humans' arbitrary matching to sample.

Normally capable children and adults were taught arbitrary matching of visual sample stimuli and nonidentical visual comparison stimuli: if Sample A1, selecting comparison B1 was reinforced; if A2, selecting B2 was reinforced. Unreinforced tests included (1) those that assessed preferences between novel comparisons when samples were also novel and (2) those that assessed selections of the least-preferred novel comparisons when the alternate comparisons were familiar B or A stimuli. Subjects during the latter tests tended to select the novel comparisons and not the B or A stimuli; these performances supported an inference of control by exclusion. The finding that subjects excluded the A stimuli when they were displayed as comparisons is contrary to previous research and suggests that control by exclusion was symmetrical under these conditions. Preference tests given after exclusion testing suggested that four of six subjects learned new arbitrary matching performances; their selections of the novel comparisons persisted when the basis of exclusion (B or A stimuli) was removed.