Abstract
In many types of learning including rule-based learning, feedback concerning the accuracy of participant’s responses on a given task plays an important role. The effect of feedback depends on its polarity: whether the feedback is positive (indicating the correctness of a response) or negative (indicating the incorrectness). While positive feedback facilitates a temporary internal hypothesis associated with a current response, negative feedback inhibits the hypothesis and leads to development of an alternative hypothesis. To explore whether positive and negative feedback play such differential roles in perceptual learning (PL), we directly compared the effects of positive and negative feedback on PL. Through a 7-day experiment, participants conducted a two-interval-forced-choice grating detection task. In each trial, two displays were presented for 200msec in sequence with a blank interval of 400msec. One display consisted of only spatial white noises (noise interval) while the other display consisted of a dim, rotated grating embedded in a noise background (signal interval). Participants were asked to report which interval included the grating. While no feedback was provided in the pre- or post-test, manipulated feedback was given in every trial during a 5-day training period, with an attempt to dissociate positive and negative feedback. When a grating with a certain orientation (positive orientation) was presented, valid feedback was given. However, if a grating had another orientation (negative orientation), participants obtained reverse-feedback: when they made a correct response (choosing the signal interval), negative feedback was provided, whereas positive feedback was given for a wrong response (choosing the noise interval). After training, performance significantly improved, not only for the positive orientation, but also for the negative orientation. These results suggest that, unlike other types of learning that are formed through facilitation or inhibition depending on the polarity of feedback, PL is merely boosted by feedback, irrespective of its polarity.
Meeting abstract presented at VSS 2013