ELICITING SELF-EXPLANATIONS IMPROVES PROBLEM-SOLVING - WHAT PROCESSESARE INVOLVED

The present experiment was designed to determine the role played by th e self-explanation of worked examples in the construction of more gene ral knowledge structures. We chose the domain of algebraic calculation . Four experimental conditions were set up. In the ''public'' conditio n, subjects had to explain the content of the worked examples to the e xperimenter. In the ''private'' condition, they had to analyze the wor ked examples on their own. In the control conditions, subjects were ei ther simply asked to solve the problems presented in the two preceding conditions, or did nothing that had to do with factoring between the pre-test and the post-test. Large differences were observed between bo th control groups and the other two conditions (in favor of the latter ). Subjects' verbalizations indicated that the analysis of the worked examples enabled certain subjects to re-elaborate their initial knowle dge. The results showed that the benefit some subjects drew from the a nalysis of the worked examples led to the construction of a new abstra ct solving schema which allowed them to solve all of the proposed fact oring problems. Furthermore, it appears that for certain subjects, the worked examples were annotated and stored, thereby facilitating the s olving of new problems by case-based reasoning. Two populations of sub jects were thus identified, ''abstract schema'' subjects, and ''case-b ased reasoning'' subjects.