Challenging the Developmental Substructure Model for Autism: A Perspective From Causal Statistics

Area of Science:

• Psychiatry
• Genetics
• Developmental Psychology

Background:

• The etiology of core autism traits (social interaction, communication, restricted and repetitive behaviors) is debated, with models like the fractionable triad and developmental substructure.
• Molecular genetic data has challenged traditional heritability estimates from twin studies, prompting new theoretical frameworks.

Purpose of the Study:

• To empirically evaluate the developmental substructure model of autism proposed by Constantino et al. (2021).
• To test whether a unitary, latent autism variable adequately explains the observed correlations among core autism phenotype traits.

Main Methods:

• Utilized causal inference principles and analyzed published data on autism phenotype correlations.
• Conducted simulations to assess the impact of a latent autism variable on trait correlations under different model assumptions.

Main Results:

• Correlations among core autism traits remained consistent across general population and clinical (autistic and non-autistic) subsamples.
• Simulations indicated that if a unitary latent autism variable were the primary cause, trait correlations should significantly decrease when conditioned on autism, which was not observed.

Conclusions:

• The developmental substructure model, despite its theoretical appeal, fails empirical validation.
• The findings suggest that core autism traits may not stem from a single latent cause as proposed by the model, necessitating further investigation into alternative etiological frameworks.