A free-association test of the differential unlearning hypothesis of pseudomediation

Three predictions, generated from the Mandler and Earhard postulates of pseudomediation, were considered in a three stage PA transfer situation designed to facilitate a mediatedchaining effect. Predictions of superior control condition performance in the second stage, and superior experimental condition performance in the third stage were supported. Free-association data failed to support a prediction of differential unlearning of A-B associations during the second stage. The Earhard and Mandler evidence for differential unlearning is questioned in terms of relevance to experimental conditions used to show the pseudomediation effect.