Mediated transfer in paired-associate learning

The present study attempts to demonstrate both mediated facilitation and interference in the learning and retention of single lists of paired-associates under conditions analogous to the standard retroaction paradigms. Conditions for intra-list facilitation (Positive Condition) are established by pairing identical responses with pairs of associatively related stimuli (Lists A-B/A’-B). Conditions for intra-list interference (Negative Condition) are established by pairing different responses with pairs of associatively related stimuli (List; A-B/A’-C). Two experiments were conducted. In Experiment I, homogeneous lists of eight paired-associates with pairs of stimulus terms having three degrees of associative relatedness were used. The three degrees of associative relatedness were: (1) first associates, (2) second associates, and (3) a non- related or List-Control level. This resulted in a 3 x 2 factorial experiment: two conditions (Positive and Negative) by three degrees of stimulus pair relatedness. In Experiment I, seventy-two undergraduate females received one study- trial and six learning trials. All subjects (Ss) returned forty-eight hours later and received ten recall trials and then relearned the lists to a criterion of one errorless trial. The Immediate-Recall Method for estimating the level of original learning was used in Experiment I. In Experiment II, eighty undergraduate males received the same procedures as Ss in Experiment I, except that five learning trials were administered so that the Projection Method for estimating the level of original learning used in Experiment II could be compared directly with the Immediate-Recall Method employed in Experiment I. In Experiment II, only the first association Positive and first association Negative Conditions were used. Facilitation and interference were demonstrated in the learning of A-B/A’-B and A-B/A’-C lists, respectively. Degree of association was not an effective factor in the production of either facilitation or interference. The effectiveness of experimental manipulations was further demonstrated by the finding that S-R pairs which appeared in the lists of the Positive Conditions were learned significantly faster than identical S-R pairs which appeared in the lists of the Negative Conditions. A direct index of mediated interference was provided by the greater than chance frequency of predicted errors (i.e., in List A-B/A’-C, predicted errors are of the type A-C or A’-B) found in the learning of the lists of the Negative Conditions. The experimental manipulations responsible for facilitation and interference in the learning of the lists did not affect retention of the lists. Recall was high in all conditions of both experiments. Relearning to a criterion of one errorless trial did result in significantly faster relearning in the Positive Conditions of both experiments. Significant differences in relearning were not obtained when relearning to the level of original learning was the criterion. An analysis of original learning, recall and relearning in terms of the Two-Stage analysis of verbal learning indicated that the primary focus of facilitation and interference was the second or associative stage of learning. Comparisons of the Immediate-Recall and Projection methods showed that the two methods resulted in almost identical estimates of original learning. The mediation hypothesis of intra-list transfer is discussed and compared with the stimulus-generalisation hypothesis of transfer effects. The mediational-linking hypothesis is presented as the most adequate explanation of intra-list interference. The common-concept hypothesis is offered as the most adequate explanation of intra-list facilitation in the present study.