Learning of Identity-Difference Relations in Children (I)

The purpose of the present study was to review theories and experiments on learning of identity-difference relations in oddity (nonmatching) and matching tasks, and to suggest further research programs in this field. Four theories and models of identity-difference learning were summarized with several comments: repression theory of inappropriate responses (Harlow, 1958), attention model (House, 1964; House, et al., 1974), rule model (Bowers, 1976), and information-processing model (Fellows, 1968). In addition, three developmental theories of children's learning relating to this field were introduced: verbal mediation theory (Kendler & Kendler, 1962), hierarchical theory of learning processes (White, 1965), and developmental theory of cognitive modes (Gollin & Saravo, 1971). Experimental studies were reviewed under the following headings: developmental studies, stimulus and task variables, and effects of pretraining and instruction. In the first section it was indicated that percents of the subjects who could learn identity-difference tasks increased with age and there was a transition period about at age five. In further studies Age × Experimental variable design should be used, in which the variables used must be closely related to theories or models. In the second section effects of stimulus similarity, stimulus repetition, number of dimensions, and number of identical stimuli were examined and performances in matching and nonmatching tasks were compared. The finding that identity-difference learning became easier with increasing numbers of stimulus repetition may be explained by several positions: attention model, repression theory, and hypothesis testing theory (Small, 1970). Which one is most plausible should be determined by further experiments. Oddity learning is facilitated by increasing numbers of identical stimuli but the mechanism remains undiscovered. Detailed analyses are needed for inconsistent findings on relative ease of learning in matching and nonmatching tasks. In the third sectionit is suggested that types of pretraining and instruction must be closely related to components in identity-difference learning and that further experiments should be directed to specify the substance of relations in identity-difference learning