Savant Skills in a Large Autistic Sample: Prevalence and Relation with Age and Intelligence

Background:  Postal surveys, parent report, and/or inspection of Weschler subtest scores within limited sample sizes variously indicate a 10-30% prevalence of savant skills in the autistic population (e.g., Treffert, 2010; Howlin et al., 2009). Contrary to earlier views, savant skills are not solely found in intellectually disabled individuals but may primarily be found in autistics whose measured intelligence, at least on certain instruments or subtests, is within the normal range. Using the ADI-R to investigate savant skills may be a time- and cost-effective method allowing characterizing of savant skills in large samples of autistic individuals.

Objectives:  We aimed to investigate in a large autistic population (1) the nature and prevalence of reported savant skills, using scores on ADI-R questions 106-111, and (2) the relation of savant skills with age, Wechsler FSIQ and Raven’s Progressive Matrices (RPM) intelligence scores.

Methods:  Our sample included 238 (26 female) autistic participants enrolled in our database and diagnosed with the ADI-R performed by trained clinicians. Participant age range was 2-39 years (mean11.4) at the time of ADI-R administration. Savant skills are indicated by a score superior or equal to 3 on at least one of Questions 106-111 of the ADI-R on either “current” or “ever” shown behavior.

Results:  61.3% (146/238) of our sample were reported to have savant skills: 28% of autistics age 2 to 5 years, 70% of autistics age 6 to13 years, and 76% of autistics age 14 years and up. Reported savant skills were significantly more frequent in older participants than in pre-schoolers (Χ²(2) = 32.64, p<.0005).Among119 participants for whom both Wechsler and RPM scores were available, those with reported savant skills (93/119) had significantly higher RPM percentile scores (68.9 vs 45.2; t(117)=3.423, p<.001) and significantly higher Wechsler FSIQ (91.8 vs 80; t(117)=3.051, p<.003) than those without reported savant skills.

Among 19 participants with Wechsler FSIQ under 70 (range 40-69), 63% were reported to have savant skills, as were 81% among100 participants with FSIQ of 70 or higher. The ratio of participants reported as with versus without savant skills was comparable in low versus normal-range FSIQ subgroups, Fisher p=.126. Among 10 participants with RPM percentile ranks between 1 and 10, 50% were reported to have savant skills, in contrast to 81% among 109 participants with RPM percentile ranks between 11 and 99. The ratio of participants with versus without savant skills was significantly lower in low versus high RPM percentile rank subgroups, Fisher p=.039.

There was an average of 2.3 exceptional abilities reported among the 146 autistic savant participants scoring 3 or 4 on the ADI-R.The most common skills were memory (35%), visuo-spatial (20.5%), followed by reading (13.9%), musical (10.7%), arithmetic(10.1%) and drawing (9.8%).

Conclusions:  These results indicate that savant skills are an intrinsic part of autism. Savant skills are more frequent than previously thought in the autistic population and are mostly evident among older individuals with higher measured intelligence. RPM scores predict savant skills in autism to a greater extent than Wechsler scores.