Effects of sample-set size on delayed-matching-to-sample

Abstract

In a series of experiments the discriminative performance of university students was assessed using a Delayed-Matching-to-Sample task. White disks, presented on computer screens, served as stimuli. For the general DMTS task, used here, a sample stimulus was presented to the participant. After a variable delay (0.05, 4, 8 or 16 s) two comparison stimuli were presented on the computer screen. One of these comparison stimuli was the same size as the sample stimulus (matching comparison) and the other was slightly larger or smaller than the sample stimulus (non-matching comparison). The sample stimuli were not intended to appear as comparison stimuli and for each of the sample stimuli there were two possible non-matching comparison sample stimuli, one larger and one smaller than the sample stimulus. The first three experiments showed that it was possible to use a titration procedure to get similar levels of performance across participants, at zero delay. During the titration procedure the size of the non-matching comparison stimuli was changed, for each participant until they were 100% correct at 0.05 s delay. The first two of these experiments also showed that there were generally only small decreases in performance as the delay between the presentation of the sample stimuli and the comparison stimuli was increased. The fourth experiment showed that performance on the DMTS task did not change when feedback was given for correct responses. The fifth and sixth experiments showed that as the number of sample stimuli used in the DMTS task, was increased there was a systematic decrease in a, a measure of discrimination at zero delay, but no systematic change in b, a measure of the rate of decrement in performance with increasing delay. It was not clear whether this decrease in a was a product of the increase in the sample-set size or of the associated decrease in the correspondence of sample stimuli on consecutive trial pairs. Experiment 7 suggested that, when the number of sample stimuli used for the DMTS was held constant and the ratio of corresponding to non-corresponding trial pairs was varied, there was no systematic change in either a or b. However, the results of this experiment were confounded by a ceiling effect and so a different procedure was required. The procedure used for the remaining experiments was a two-sample DMTS task, where the sample stimuli were also used as comparison stimuli. This type of DMTS task is similar to that typically used in discrimination tasks with animal subjects. The eighth and ninth experiments were designed to investigate whether this type of procedure could be successfully used with human participants and found that it could be. Experiment 10 also used this type of two-sample DMTS task to investigate the effect of decreasing the ratio of corresponding to non-corresponding trial pairs. It was found that, when there was no ceiling effect, varying the ratio of corresponding to non-corresponding trial pairs did not result in a systematic change in either a or b as in Experiment 7. These results are discussed in terms of the effects of range, sample-set size and of the ratio of corresponding to non-corresponding trial pairs on human performance on DMTS.