Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Recent research has identified changes in the strategies used and the shape of learning curves of children learning concepts as a function of developmental level. Two recent models have been proposed to explain these changes. One model views these changes as an improvement of hypothesis-testing strategies. The other model views these changes as a switchover from reliance on one type of learning process to a second type of learning process. The present study compared the two models with respect to the strategies used and the shape of the learning curves of four grade levels of subjects. One hundred twenty-nine children from grades 1, 5, and 7 and 40 college undergraduates were tested using a concept task consisting of a list of items in which a keyword was followed by five choices (e.g. yourself: smoke, else, chief, field, five). The correct choice was the word consisting entirely of letters in the keyword (else in the above example). Each subject was tested individually with one of two instructions: to try to find a rule relating the correct choice word and the keyword, or, to memorize which choice word goes with each keyword. After each block of 10 trials, subjects were asked "Why did you pick this word?". Results showed that there were age-related differences in concept learning among the four grade levels with respect to: (a) steepness of the learning curves (transition phase length), (b) the trial block of last error, and (c) the number of subjects who eventually discovered the concept rule. A high correlation between the trial block where performance began to rise above chance (performance cutpoint) and the point where subjects' verbal response indicated some knowledge of the concept rule (verbal cutpoint) was found in all grade levels. The findings generate some support for both models.