Comparison of Associative Learning under Three Methods and Three Directional Forms of Presentation

Summary.-Paired-associate recall was compared under three methods of presentation, prompting, correction, and non-correction arranged in forward, semibackward, and backward forms. The methods were characterized by pairs of events: RT-R for prompting, R-RT for correcuon, and R-KR for non-correction. These ordered pairs determined the number of correct pairings Ss could achieve in each method, the largest for prompting and least for non-correction. They served to predict the over-dl speed of learning for the methods: fastest for prompting, slowest for non-correction. Direcrionaliry forms were characterized by pairs ST-R and ST-RT for forward, R-ST or RT-ST for semibackward, and R-ST and RT-ST for backward training. The number of backward paits sewed to predict the differential speed of learning in each of the three directionaliry forms, fastest for forward, slowest for backward. The hypotheses were wted on a verbal-motor task and were confirmed. The results of previous research indicate that sometimes different methods of presentation of materials lead to different speeds of forward associative learning. Verbal forward paired-associate learning under the method of recall was found superior to that under anticipation (Battig & Brackett, 1961; Battig & Wu, 1965; Schild & Battig, 1966) but with some notable exceptions depending on the materials used (Goss & Nodine, 1965; Cofer, et al., 1967). Prompting, which is similar to recall, gave better learning than confirmation (Cook & Spitzer, 1960), which is similar to anticipation; but this again held only for certain verbal materials (Levine, 1965 ) . Selective psychomotor learning was found to be slower under the method of non-correction than correction (Noble, 1965 ). Although experimental evidence is not unanimous, it can be shown that on theoretical grounds, at least from the contiguity point of view, the prompting method may be expected to be the most efficient, followed by the confirmation-correction and confirmation-without-correction methods in that order. The analysis of formal properties of the methods and the derivation and testing of the predictions derived from this analysis form one part of the study. The other part concerns backward learning. Ekstrand ( 1966) reviewed forward-backward learning comparisons and found that with equally available stimulus and response terms learning sometimes tends to show symmetry but that there are important exceptions to this rule, even with highly available mate'Part of research financed by the National Research Council of Canada Grant APB 106 to V. B. Cervin.