The effects of meaningful similarity and practice in the transfer of training

Abstract

This study investigated the effects of two variables within a transfer of training paradigm. The findings of this research are interpreted with a view to assessing Osgood's theory of transfer for Paradigm which predicts that the more nearly similar are two responses, the less would be the interference in learning. The two variables, Meaningful Similarity relations and amount of original list practice, were introduced in a transfer paradigm, Paradigm B, where the responses of two lists are varied and the stimuli are the same, i.e., Original list (OL): Sl--Rl ; Transfer list (TL): SI--R2. The stimuli were letter pairs, and the responses, adjectives, e.g., OL: w.k.--soft, TL: w.k.--relaxed. Three levels of Meaningful Similarity relations between responses, similar (S), neutral or unrelated (N), and opposed (0) were presented. Each OL and TL list consisted of 15 items, 5 Ss, 5 Ns, and 5 0s all counterbalanced and mixed at random; thus, for the Meaningful Similarity variable, each Subject was his own control. Each ~ learned, by the anticipation method, an OL to either a Low or High Practice criterion, and then a TL. There were 36 Ss in the Low Practice group and 36 in the High. All Ss were University of Hawaii students. Three dependent variables were scored on the TL trials: (1) (a) number of trials for each item to be obtained correct once, and (b) number of trials for each item to be obtained correct twice, (2) latency of correct responses, and (3) intrusions and blanks. The statistical analyses performed on the data of the trials with the transfer list (TL) indicated no significant differences for any of the dependent variables with respect to Meaningful Similarity, i.e., significant differences between S, N, and 0 were not found with regard to trials-to-criterion, latency, or intrusions. Significant differences, however, were found on the TL trials between the Low and High practice groups with respect to some of the intrusion and blank measures. An analysis was made to determine whether those items coded for the S, N, and 0 relations had been differentially learned on the OL trials. The finding of a non-significant Meaningful Similarity effect indicates that it is not likely that the findings of this stUdy are biased as a result of initial inequities with respect to ease of learning on the original list. Osgood has theorized that the more nearly similar are two responses, the less would be the interference in a Paradigm transfer situation. He predicts an S N 0 order, where the S relation between responses (the most similar) results in the least amount of interference and the 0 relation (the least similar) results in the most amount of interference, i. e., S