Trial-by-trial feedback does not improve performance or metacognition in a large-sample perceptual task

Abstract

While some studies report that trial-by-trial feedback improves performance on perceptual tasks, others have suggested that it has little to no effect. In the present study, we examined the effect of trial-by-trial feedback on both the perceptual decision and confidence in the accuracy of that decision (metacognition) in a large sample of participants (n = 450). Participants were randomly assigned to a trial-by-trial feedback group or no-feedback group and completed two perceptual tasks. In Task 1, participants were required to make a perceptual judgment concerning whether the letter X or O occurred more frequently in a 7 × 7 grid, and the assigned group received feedback on the accuracy of each response. Task 2 included a similar 7 × 7 grid with a red/blue discrimination. Both groups were given no feedback on Task 2 in order to determine whether any putative improvements would generalize to a new task. All perceptual decisions were followed by a 4-point confidence rating. Although a steady improvement in performance over time was observed in both tasks, results demonstrated that trial-by-trial feedback had no effect on either task performance or confidence ratings. Specifically, in Task 1, participants in the feedback and no-feedback groups had similar perceptual sensitivity (d’_feedback= 1.86, d’_no-feedback = 1.86; p= .97) and metacognitive ability (Mratio_feedback = 0.70, Mratio_no-feedback = 0.76; p= .11). The lack of difference between the feedback and no-feedback groups extended to Task 2 as well (d’_feedback = 1.76, d’_no-feedback = 1.70; p= .28, Mratio_feedback = 0.71, Mratio_no-feedback = 0.77; p= .16). Considering that the sample used was much larger than previous reports, these data indicate the possibility that trial-by-trial feedback may not meaningfully impact either perceptual performance or metacognition.