Reversing the testing effect by feedback is a matter of performance criterion at practice.

Retrieval practice is generally considered to be one of the most effective long-term learning strategies and is presumed to be more favorable than repeated study. However, a few recent studies have demonstrated that repetitive feedback at final recall can reverse the long-term advantage of testing over restudy. The result that feedback at long-term tests can dramatically decrease the relative effectiveness of retrieval-based learning could be important for both theoretical and practical reasons. Considering that these earlier studies administered low retrieval success at retrieval practice, we investigated whether the effect of feedback on the testing effect is modulated by the level of retrieval success during practice. In three experiments the level of success at retrieval practice was manipulated by multiple pre-practice learning trials, and multiple tests with feedback were applied after a 1-week retention interval at final recall. Our results have demonstrated that a feedback-induced reversed testing effect was present only at low retrieval success during practice (Experiment 1), whereas with moderate (Experiment 2) and high retrieval success (Experiment 3) during practice a significant testing effect emerged and no reversed testing effect was found even after repeated cycles of feedback. These results point to the conclusion that the level of retrieval success was the key factor in reversing the testing effect in earlier studies. Application of high retrieval success during practice can produce long-lasting accessible memories even in learning settings applying multiple tests with feedback.