Embedded Figures Test Performance in the Broader Autism Phenotype: A Meta-analysis.

The team pooled every paper that gave neurotypical adults the Embedded Figures Test. This test hides a simple shape inside a busy picture. People who spot the shape fast show a 'local bias' — they see tiny details before the big scene.

They also gathered each adult’s Autism-Spectrum Quotient score. Then they asked: do people with higher autistic traits finish the hidden-shape task faster? They compared two ways of answering: looking at the full spread of scores, or only comparing the highest and lowest groups.

Across many studies, the advantage showed up only when the top AQ group was stacked against the bottom group. Continuous scatterplots looked flat. The edge is real, but you have to slice the data wide to see it.

Cantio et al. (2016) seems to disagree. In people with ASD they found a local bias was rare and added no diagnostic value. The key difference: Cathriona studied diagnosed kids using many tasks; the meta-analysis studied subtle traits in average adults using one task. Both can be true — local bias matters for research groups, not clinic lists.

Green et al. (2020) extend the idea. They showed that relatives who fail faux-pas stories have weak language-based executive skills. Together the papers build a chain: subtle autistic traits → specific EF style → different test profiles.

Ko et al. (2024) push the link younger. Preschoolers with ASD already show large EF gaps. The meta-analysis shows the same cognitive style lives quietly in typical adults; the preschool study shows it can grow into clinical need.

When you test for cognitive style, pick your comparison wisely. A mild AQ rise in one client may not predict blazing EFT speed, but the top-versus-bottom split can still guide research matching. Also, remember that fast hidden-shape finding is part of the broader autism phenotype — it is not unique to diagnosed ASD. Use this insight when explaining neuro-diversity profiles to staff or families, and keep language-based EF on your intervention radar if social reasoning is shaky.