Strength vs. temporal-order information in delayed-matching-to-sample performance by monkeys

The usual procedure in delayed-matching-to-sample (DMTS) experiments is that the most recently presented sample stimulus is correct on choice tests. The present paper reports two experments in which the first of two samples was always correct. Theories based on trace strength, such as the Roberts and Grant trace strength competition model, predict that such a procedure will yield below chance performance, because on the average the more recently presented stimulus will have the greater trace strength. In contrast, D'Amato and others have proposed that DMTS performance is mediated by a temporal discrimination process and, since either stimulus may act as S+ in a discrimination problem, monkeys should be capable of learning to choose either the second or the first of two sample stimuli. Performance did not exceed chance in either experiment. The generality of these results is tentatively explored.