Special Skills in Autism Spectrum Disorders Are Associated with Unique Phenotypic Characteristics

Background: Autism spectrum disorders (ASD), more so than other neurodevelopmental disorders, are associated with ‘islets of ability’ ranging from relative strengths to so-called ‘savant’ skills (Heaton & Wallace, 2004).  In spite of this well-established connection, very little is known about how individuals with ASD may differ from one another based on the presence of special skills, which may constitute an informative phenotype within ASD.

Objectives: Examine basic characteristics (i.e., age, sex ratio, IQ, autism symptomatology) associated with elevated special skills in a large sample of individuals with ASD (n>500). 

Methods: Using Autism Diagnostic Interview-Revised parent ratings of special skills from several sites within the Collaborative Programs of Excellence in Autism and Studies to Advance Autism Research and Treatment, we categorized participants into two groups: individuals without a reported special skill (n=245) and individuals with a skill above both their general ability level and population norms, regardless of its functionality (n=274).  We assessed whether the presence of special skills is associated with differences in terms of age, sex ratio, IQ, and autism symptomatology.  For the sake of comparability across sites and instruments, we restricted our sample to those individuals with available Wechsler IQ data (range=39-150) and participants 60 months and older (range=60-351 months).  

Results: Individuals with ASD rated as possessing special skills (Ns: memory=183, reading=102, visual-spatial=75, computational=58, drawing=49, music=45; multiple skills=138) were older (M=140.02+/-53.37 vs. 119.53+/-49.41 months), had a higher male to female ratio (244:30 vs. 184:61), and had a higher mean IQ (M=91.79+/-26.08 vs. 77.52+/-25.10) than those without special skills (ps<.001).  Importantly, males and females in this sample did not differ in terms of age and IQ (ps>.3), ruling out these possibly confounding factors in the observed sex ratio difference.  Repetitive behavior symptoms, unlike social and communication symptoms, were elevated in the special skills group (p=.004) even though these participants on average had higher IQs and were older.  Therefore, controlling for age and IQ only served to enhance the group difference in repetitive behavior symptoms (p<.001).  Preliminary item analysis within the repetitive behavior domain revealed significantly greater endorsement of circumscribed interests for the special skills group (p<.001).

Conclusions: The presence of special skills in the context of ASD is associated with unique phenotypic characteristics.  We find that within a sample of children, adolescents and young adults with ASD, individuals rated as possessing special skills were on average older, and consistent with Howlin, Goode, Hutton, and Rutter (2009) had higher mean IQs and a greater male to female ratio.  Extending previously documented subclinical associations (Vital, Ronald, Wallace, & Happé, 2009), repetitive behavior was the only symptom cluster to be more elevated among individuals with special skills.  Howlin et al. (2009) did not find differences in Autism Diagnostic Interview (ADI) repetitive behavior ratings in their savant group, but the ADI version used in their study did not include assessment of circumscribed interests, which we have shown to receive greater endorsement in the special skills group.