Short-term memory and paired-associate learning

Abstract

Summary This study deals with the relationship between STM and PA learning. The first three experiments were an attempt to determine if traditional variables of PA learning had the expected effect on STM for individual pairs. The general procedure was such that a list of pairs composed of common English words paired at random was presented once with a presentation time of 2 sec. per pair; then, one of the pairs was tested for recall. In Exp. I lists were two, three, four, or five pairs long. In Exp. II presentation time was 1, 2, or 3 sec. per pair. In Exp. III one, two, or three presentations of each list preceded the recall test. Results suggested that the retention curve leveled out after approximately two subsequent pairs at a value which increased with presentation time, number of presentations, and, possibly, shorter lists. Experiment IV was essentially the first trial of a PA learning task; all pairs were tested for recall and lists were 8, 16, 24, or 32 pairs long. As predicted, number of words recalled increased with list length. Also, during the first block of eight recalls, recall probability for the 8 pair list was higher than for the 16-, 24-, and 32-pair lists; the latter three lengths did not differ. In general the results lent support to the notion that there might be marked intratrial learning and forgetting of individual pairs during PA learning, that some of the phenomena of PA learning might be predictable from studies of STM, and the PA “learning” might basically be increased resistance to intertrial forgetting.