Is there evidence of generalized slowing in Autism Spectrum Disorder? A meta-analytic investigation

• In Event: Poster Session 04
  In Poster Session: PS 04 Topic - Developmental Disabilities

Abstract

Various studies have asked whether individuals with Autism Spectrum Disorder (ASD) exhibit slower processing as compared to age-matched peers—an association with potential ramifications for cognitive and social functioning (Kail & Hall, 1994; Haigh et al., 2018). Two recent reviews synthesized the literature on processing speed in ASD and drew opposite conclusions. Ferraro (2016) used Brinley Plot analyses of reaction time (RT) tasks and failed to find evidence of slowing in ASD (k = 32). However, the Brinley Plot methodology is flawed and outdated, e.g., it does not assign weights in accordance with sample size in contrast to newer meta-analytic approaches. Ferraro included simple and choice RT tasks that ranged widely in complexity in the same Brinley Plot, which precluded analysis of how task-specific features might contribute to slowing (Windsor et al., 2001). Ferraro also failed to limit studies to those with age-matched controls, which further confounded the analysis. More recently, Vekikonja et al. (2019) investigated impairments in individuals with ASD across cognitive domains including processing speed. Notably, the authors operationalized processing speed differently from Ferraro (2016) by including non-RT-based measures, and found evidence of slower processing in ASD (k = 21). In effort to reconcile these findings, the current meta-analysis used Robust Variance Estimation (RVE; Fisher & Tipton, 2015) to address the question of whether ASD is associated with generalized slowing. We limited our search criteria to RT tasks to facilitate comparison with Ferraro (2016).
We searched ERIC ProQuest using Boolean combinations of terms: Autism, ASD, Asperger, response time, reaction time, RT, interference control. Studies had to be in English and include an ASD group and an age-matched neurotypical group. To minimize task-complexity, we limited study selection to simple and interference-control RT tasks with neutral, baseline, or congruent conditions; Figure 1 for PRISMA table. The final analysis included 32 studies (k) with 75 effects (m); median age = 11 years; 7 months, range = 8 to 32 years; 85% male. Inter-rater reliability (92%) was calculated as a proportion based on mutual effects extracted by both raters across variables included in the analysis. Effects (g) were calculated using RVE in R. The overall effect indicated significant slowing in ASD groups (g = .43 p < .001, CI = .27; .60). Moderation analyses used RVE meta-regression to test whether stimulus type (linguistic v. nonlinguistic; social v. nonsocial) or task-category (simple v. interference-control) influenced effect-size estimates. All moderators were nonsignificant; Table 1. Age (grand-mean centered) was entered as a covariate in a separate model and was also nonsignificant. These results suggest that individuals with ASD exhibit slower processing than neurocognitive peers, contrasting with Ferraro (2016) while concurring with Vekikonja et al. (2019), albeit with a different set of tasks. We found little evidence of task-specific influences on effect-size estimates, which fits with findings suggesting links between domain-general processing in ASD and other skills, including social-communication abilities (Haigh et al. 2018). Slower processing might be expected to have cascading effects on development of automaticity across domains and should be investigated using longitudinal designs.

Authors

• Nicole M. Zapparrata, The Graduate Center
  Presenting Author

• Patricia J. Brooks, City University of New York College of Staten Island
  Non-Presenting Author

• Teresa Marie Ober, University of Notre Dame
  Non-Presenting Author