Errors May Not Cue Recall of Corrective Feedback: Evidence against the Mediation Hypothesis of the Testing Effect

Successfully retrieving information protects it against later forgetting. Failed retrieval attempts are also beneficial if followed by study of corrective feedback. To explain both of these findings, researchers have proposed the "mediation hypothesis." In the case of learning from corrective feedback, initial errors may serve as mediators, becoming associated with the corrective information and assisting with recall of that information during later retrieval attempts. A simple prediction follows: reminding learners of their initial errors should improve their recall of the corrections. We tested this prediction with a set of 3 experiments. In Experiment 1, participants learned cue-target word pairs (e.g., "cactus-point") by reading or by incorrectly guessing target words before viewing corrective feedback. Immediately afterward, participants were equally good at recalling targets from the cue alone, the cue plus their own guess, or the cue plus a guess made by another participant. Experiment 2 produced similar results when the retention test was delayed by 24 hr. In Experiment 3, participants instead learned word triplets having a mediational relationship through preexisting associations (e.g., "film-star-galaxy"). Here, reminding participants of a genuine mediator greatly improved target recall, supporting the validity of our cueing procedure.