One-trial versus incremental-paired associate learning

Abstract

Two experiments were conducted to test several implications of incremental and all-or-none theories of learning. In both experiments, S s were given alternate learning and test trials on lists of word responses paired with consonant-syllable stimuli. The major findings were: (1) items learned late in training (difficult items) showed more shifts from correct on one trial to non-correct on the next than did those learned earlier (easy items); (2) the proportion of items never previously correct which became correct on each trial increased with training; (3) overlearning increased both immediate and delayed retention; (4) the mean number of correct responses gradually increased over a series of five successive recall trials in spite of the fact that no learning trials were given. These findings were interpreted as compelling evidence for an incremental theory of learning, but incompatible with present all-or-none formulations.