Should feedback confirm success, correct errors, or both? An empirical test of the sandwich approach to delivering feedback

In motor learning, feedback has long been viewed as a mechanism for correcting errors present in a skill. Recent research has suggested that feedback that confirms the success of a movement can also be valuable for learning (Chiviacowsky & Wulf, 2007). The purpose of the present study was to test the relative merits of knowledge of performance feedback that confirms success (CONF), corrects errors (CORR), or does both through the method commonly referred to as the sandwich approach (SAND). Participants (36), were randomly assigned to one of the three feedback groups, and practiced a soccer throw-in task. The acquisition phase consisted of 30 trials with feedback after each trial about the most critical error (CORR), the most critical component of form performed correctly (CONF), or a combination of the two (SAND). Approximately 24 hrs after the conclusion of acquisition, participants completed a retention test, and two transfer tests. Participants completed self-efficacy assessments prior to the start of each experimental phase, and then completed a post-experiment questionnaire after the conclusion of the study. Results revealed that the CORR group and the SAND group received higher form scores than the CONF group across acquisition, retention, and both transfer phases (p < .05). The SAND and CORR groups, however, did not differ with regard to form. No differences between groups emerged for self-efficacy, but the post-experiment questionnaire revealed the SAND and CORR groups found the feedback to be more useful than the CONF group (p < .05). The results of this study suggest that for learning complex tasks, feedback including corrections leads to better form than feedback only confirming success. There was, however, no documented advantage to receiving both corrective and confirmatory feedback as opposed to corrective feedback only. The findings here contrast with recent studies suggesting a benefit of feedback after “good” trials, and support the notion that when learning complex tasks corrective information is needed for learning.