Categorization and Discrimination of Random Forms Following Schema Training

Several investigators have shown that when given a task of categorizing several classes of random patterns, where each class consists of a number of patterns all generated from a common pattern or prototype, Ss learn to abstract a code analogous to or congruent with each class prototype in the process. This learning has been termed schematic abstraction or schema learning. Inferences regarding schematic abstraction have been drawn from the observation of .§s\u27 categorization performance in a transfer task where the prototype is first presented along with both new and previously sorted class patterns. Enhanced performance with class prototypes when compared to new class patterns is then taken as evidence for schema learning. While several investigators have provided evidence as to the validity of the phenomenon of schema learning, there remains a paucity of evidence to show that such learning serves to enhance categorization behavior and also encoding-processing demands, fundamental tenents of the basic notion of schema learning. The effects of schema learning on subsequent categorization behavior were investigated by providing a control group given irrelevant schema pretraining, while changes in discrimination behavior were examined by presenting Ss with a two-stimulus same/ difference RT task. The two stimuli were presented in succession with varying interstimulus delay intervals. An initial experiment shows the RT paradigm to be sensitive to different levels of encoding-processing demands. The results of the present study, generally, support the notion that schematic abstraction serves to facilitate not only categorization of new stimuli, but, in addition, reduces demands placed on encoding processes