Atypical Visual Processing in ASD As A Global Deficit or Local Bias: A Meta-Analysis

Background: Atypical visual processing in autism spectrum disorders (ASD), such as superior processing of local details or substandard processing of global structures, has been investigated repeatedly, but research findings vary widely and are often contradictory. In spite of influential theories like the Weak Central Coherence (WCC) and the Enhanced Perceptual Functioning (EPF) account and a great deal of research inspired by them, the atypical visual processing profile of individuals with ASD remains only partly understood. The debate focuses in particular on whether differences in autistic visual perception should be conceptualized as either a local bias or a global deficit. Proper quantitative reviews of the available data, in which the overall effect size of each study is assessed and the influence of potential moderators is investigated, are lacking.  

Objectives: We combined all available experimental data on local and global visual processing in ASD from 1983 till 2011 through meta-analytic techniques, and (a) assessed whether the available data are in favor of the hypothesis of superior local processing or the hypothesis of inferior global processing, in individuals with ASD compared to typically developing (TD) controls, (b) evaluated the conceptualization of local and global visual processing in different paradigms and measures, and (c) evaluated which moderator variables rule the diversity within the literature.

Methods: A systematic literature search was performed. All data of the 44 selected articles were coded and transformed to the standardized mean difference metric d and used in a linear three-level random effects model, accounting for random sampling fluctuation, between-outcome variance and between-study variance. Further analyses carefully extended the model by including one or more of nine moderator variables, comprising ASD subtype, type of control group, sample size, gender, age, IQ, experimental paradigm or task, dependent variable, and assumed direction of group difference (e.g. enhanced local versus reduced global).

Results: Overall, we found no significant performance difference in visual processing between individuals with ASD and controls. Yet, analyses investigating the individual or combined influence of moderator variables showed significantly reduced performance of the ASD group for tasks inducing a global processing advantage (especially as measured by reaction times), but no effect for task elements inducing a local processing advantage (nor when accuracy measures were used). The moderator variables age, gender and IQ yielded mixed results, whereas the specific subtype of ASD did not have any effect.

Conclusions: The results of this meta-analysis favor the evidence for a deficit in global visual processing in ASD rather than an advantage in local processing in ASD. Though several important moderator variables have been included in the study, further research is necessary to identify more clearly which (other) moderator variables are at play and how they influence visual processing abilities in individuals with ASD versus controls.