Illusion Susceptibility Indicates a Two-Factor Structure for the Systemizing Trait of Autism

Background: Several theories suggest that individuals on the autism spectrum should show a decreased susceptibility to visual illusions, either due to an increased focus on local cues, a decreased focus on global contextual cues, or both. However, evidence for decreased illusion susceptibility in the autism spectrum disorders has been mixed. Recently, though, these disparate results have been reconciled by the finding that the autistic trait of systemizing negatively covaries with susceptibility to visual illusions driven by contextually-induced distortions of an observer’s egocentric reference frame, but not with illusions driven by allocentric distortions (Walter et al., 2009).

Objectives: The current study sought to determine whether the relationship between systemizing and illusion susceptibility can be attributed to a heightened processing of local cues, an attenuated processing of global cues, or some combination of the two.

Methods: Scores on the Systemizing Quotient – Revised (SQ-R, Wheelwright et al., 2006) were compared to measures of susceptibility to the Rod-and-Frame Illusion (RFI) in a large neurotypical population (n = 162). Depending on the size of the illusion-inducing frame, the RFI is thought to be driven by a weighted combination of local, low-level orientation contrast effects (more prominent with small frames) and globally-induced distortions of the observer’s egocentric reference (more prominent with large frames). Susceptibilities to these two components of the illusion were measured using recently-developed techniques designed to isolate the two (Dassonville & Williamson, 2010).

Results: Replicating previous results, higher SQ-R scores were found to be associated with a decreased susceptibility to the large-frame RFI, suggesting a decreased tendency to use global contextual cues. However, SQ-R scores were also found to be associated with an increased susceptibility to the small-frame RFI, indicating an additional increased tendency to rely on local orientation effects. Interestingly, susceptibilities to the large- and small-frame RFI were uncorrelated, suggesting that local processing biases do not necessitate impairments in use of global contextual information and that, while comorbid in high systemizers, these may be two orthogonal perceptual processes. To further examine the relationship between these perceptual processes and systemizing, a principal components analysis was conducted to isolate factors in SQ-R that might correlate with the local and global effects of the RFI. Two factors were extracted that accounted for 92.8% of the variance in the SQ-R. High scores on factor 1, consisting of items measuring ‘analytical tendencies’, were associated with decreased global effects of the RFI, while high scores on factor 2, consisting of items measuring a ‘need for sameness’, were associated with increased local effects. An examination of factor score distributions also indicated that while men scored significantly higher on ‘analytical tendencies’, women scored significantly higher on ‘need-for-sameness’.

Conclusions: These results suggest that the systemizing trait of autism contains a two-factor structure, with components that separately measure analytical tendencies and a need for sameness. While analytical tendencies are associated with a decreased tendency to use global contextual cues, a need for sameness is associated with a heightened processing of local cues.