A developmental study of hypothesis testing and cue selection strategies

Previous investigators who have evaluated M. Levine's hypothesis testing model of learning have employed blank-trial probes in order to determine which hypothesis was sampled on specific trials. A method of determining which hypothesis was sampled that did not require blank trials was administered to children in Grades 1, 3, and 5, undergraduates, and old adults (mean age = 75 years). A 2-choice, multidimensional discrimination task was presented in which the Ss first chose the stimulus thought to be correct and then pointed to a single cue in a complete set of decomposed cues from the learning task. Results are in accord with other studies of hypothesis testing in that developmental differences were observed. Younger and older children differed in the cue selection strategies used during learning. Undergraduates used very efficient strategies that often led to learning, while the old adults were quite inconsistent and often failed to recognize when they had selected the correct reinforced cue.