A study was made to assess the extent to which social interaction is beneficial to cognitive development. A total of 154 boys and girls, 5 through 9 years of age, initially participated individually in a conservation-like pretest requiring prediction of the way a beam would tip when different numbers of weights were placed at differing distances from a fulcrum. Seven increasingly sophisticated rules for prediction have reliably been differentiated. On the basis of pretest results, children were assigned to (1) a control group in which children were again tested individually; (2) an "equal rule" group pairing same-age, same-sex, same-class children who had used the same prediction rule on the pretest; or (3) an "unequal rule" group pairing same-age, same-sex, same-class children who had used different prediction rules on the pretest. During treatment, disagreeing subjects resolved their disagreement in discussion. Subjects were individually post-tested twice and improvement in rule use was recorded. Findings indicated that interaction with a partner was not conducive to cognitive development. No significant improvement was found when equal rule and unequal rule groups were compared. Partners using a lower rule in the unequal rule group were the only children who improved; performance of higher partners worsened. Significant sex, age, and age by sex differences were found, with boys consistently benefitting more than did girls from the process of interaction. (RH)