Abstract
A number of studies have examined the role of response feedback (informing an observer of performance accuracy) during training in perceptual learning and found that feedback increases the magnitude and speed of learning (Herzog and Fahle, 1998, 1999). Here we present evidence for a new role of feedback: feedback stabilizes perceptual learning, making it more resilient against deterioration/adaptation that occurs due to an excessive amount of continuous trial performance (Mednick et al., 2005, Censor et al., 2009). In the experiment, 12 subjects were trained on a motion detection task. Subjects were trained on two randomly interleaved motion directions for 3 days. One direction was always paired with response feedback, while the other was not. Pre-tests and post-tests were conducted in which performance was measured not only for the two trained directions but also a spread of surrounding directions. No feedback was provided during these tests. Performance changes after training were calculated for the first half (early) trials and second half (late) trials of the test stages for each direction. For the direction trained with feedback and its vicinity, significant learning was observed in both the early and late trials. On the other hand, for the direction trained without feedback and its vicinity, while significant learning was observed for the early trials, this learning effect completely deteriorated in the late trials. We conclude that feedback not only increases the magnitude of learning, as previously found, but also stabilizes learning to be resilient to deterioration or adaptation.
NIH R01 (EY015980, EY019466, AG031941, MH091801).