Comparison of S-R and R-S Learning of Paired-Associates

Abstract

Several recent studies have been concerned with backward associations in paired-associate learning and, after the usual S-R learning, have tested for R-S learning. The cest for R-S learning consists of presenting the responses from the S-R list and requiring Ss to give the stimulus which had been paired with each response. The usual procedure is to require Ss to learn the S-R list, cest for R-S learning, and then compare the R-S learning with the S-R learning. If the S-R and R-S pairs are not of equal difficulty then this comparison confounds the direction of learning with the difficulry of the pairs. The backward learning of the S-R list should be compared with the forward learning of an R-S list so that the same pairs are learned in each case and the only difference is the direction of learning. For example, a list consisting of nonsense syllables as stimuli and words as responses is used for S-R learning. The test for R-S learning would consist of presenting the words and requiring Ss to give the syllable which had been paired with each word. It would be expected thac there would be less R-S than S-R learning because increased meaningfulness of responses facilitates learning more than does the increased meaningfulness of stimuli (Sheffield, 1946; Hunt, 1959). If the R-S learning is in a backward direction, from response to stimulus of the S-R list, then the amount of R-S learning may be primarily a function of the difficulty of the R-S pairs and not directly related to the degree of S-R learning. This study is concerned with the comparison of backward and forward learning of paired-associate lists which have different types of items as stimuli and as responses. Words, numbers, and nonsense syllables were selected because it was expected thac the difficulty of a list would depend on whether particular items were used as stimuli or as responses. If the usual pairedassociate (S-R) learning of a list which has nonsense syllables as stimuli and words as responses is faster than the S-R learning of the same pairs except that the words are stimuli and the syllables are responses, then the R-S learning should be inversely related to the S-R learning, i.e., the R-S learning should be faster for the list with nonsense syllables as responses and words as stimuli. The same relationship between S-R and R-S learning should hold whenever reversal of the stimulus and the response of pairs in a list results in a difference in the difficulty of the S-R learning.