The Effect of Feedback on Perceptual Learning in Autistic Adults

Background:  Understanding how autistic individuals learn is very important for developing and implementing efficient educational and behavioural interventions. Given their relatively stronger perceptually-based cognitive abilities (relative to those in the verbal domain), it cannot be assumed that autistics learn using the same rules and strategies as non-autistics. Of particular interest, perceptual learning (PL) is a class of learning that is based upon changes induced by the repeated exposition and response to specific types of perceptual information. Such learning often includes feedback, indicating whether or not a response was correct during a trial within a PL task. A previous study assessing PL using feedback-based training in autism suggested that feedback did not benefit autistics, who nevertheless outperformed typical controls when asked to discriminate novel highly-similar stimuli (Plaisted et al., 1998).

Objectives:  To investigate PL in autism using a low-level perceptual task, in the presence and absence of trial-by-trial feedback.

Methods:  Ten autistic and ten non-autistic adults, matched for full-scale IQ and age (18-31 years), performed a low-level PL task. They were asked to indicate whether a grating was tilted to the left (i.e., counter-clockwise) or to the right (i.e., clockwise) relative to an oblique 45-degree reference orientation. Thresholds, defined by the minimal deviation in degrees needed to discriminate tilt orientation, were measured for each participant every 25 minutes, with each session consisting of 420 trials. To assess baseline performance, all participants completed one baseline session with no feedback. During six subsequent testing sessions, half (n=5) the participants in each group were then provided feedback and half were not.

Results:  PL was defined as the percent change in tilt discrimination threshold in the six testing sessions compared to the baseline session. For the autistic group, there was no evidence for PL. Thresholds remained equal to baseline across testing sessions, whether or not trial-by-trial feedback was present. In contrast, in the non-autistic control group, thresholds decreased with training sessions and thus there was evidence for PL, but only when feedback was present.

Conclusions:  Within the context of the experimental conditions and paradigm, evidence for PL was found only in the non-autistic group when feedback was present during testing sessions. Importantly, trial-by-trial feedback did not result in PL in the autistic group, suggesting that the neuromodulatory effects of feedback are different in autism. This preliminary finding raises questions regarding the value of feedback during interventions, and at a more basic level, during studies defining perceptual and cognitive processes in autism.