Implicit learning of perceptual distributions in children with ASD

Abstract

Recent theories of autism spectrum disorder (ASD) based on the Predictive Coding framework hypothesize differences in implicit learning based on the statistical regularities of the environment in ASD. Previous studies indicate that children and adults with ASD are able to represent sets of low-level stimuli by its mean when assessed explicitly. However, it is unknown whether and how children with ASD implicitly represent sets of similar low-level stimuli. Chetverikov and colleagues (2016) found evidence for implicit learning of feature distributions’ parameters (mean, variance and shape) in adults assessed by repetition and role-reversal effects in an odd-one-out search task. In this study, the odd-one-out search task (Chetverikov et al., 2016) was administered in an ASD group and a typically developing (TD) group of children (10–14y). We aimed for two groups of 25 children, matched for age, IQ and gender ratio. The task consisted of prime and test blocks and the children had to search for the odd-one-out in arrays of differently oriented lines. Within blocks, the orientations of the distractors were sampled from a Gaussian distribution with a constant mean and standard deviation of 10° or 5°. In the test blocks, target distance from the mean of the preceding prime distractor distribution (current-target-to-previous-distractors-distance; CT-PD) was manipulated. The internal model of the distractor distribution can be assessed by plotting reaction times (RTs) as a function of CT-PDs. Preliminary analyses on a subsample of the ASD (N=18) and TD groups (N=26) showed no significant difference in RTs or accuracy between groups. In addition, a significant decrease in RTs with repetitions in prime blocks was found in both groups. RTs in test blocks decreased as CT-PD increased in children with and without ASD. These preliminary results suggest that both children with and without ASD are able to implicitly learn the distractor distributions.

Acknowledgement: Lisa Lemmens is a doctoral fellow, and Sander Van de Cruys a postdoctoral fellow, both of the Research Foundation – Flanders (FWO). This work is supported in part by the Methusalem program by the Flemish Government (METH/14/02), awarded to Johan Wagemans.