Subjects in a rule-learning task learned either Conjunction, Exclusion, or Joint denial concepts with the relevant attributes presented in either their positive form (“red” and “square”) or in the form in which they are expressed in the rule to be learned (“red” and “not square” for Exclusion; “not red” and “not square” for Joint denial). Presenting relevant attributes in their rule-relevant form significantly improved rule learning. In addition, when rule-relevant attributes were used the Exclusion and Joint-denial rules were not significantly different in difficulty from the Conjunction rule, as predicted by a model of rule learning developed by Salatas and Bourne (1974).
