The Level and Nature of Autistic Intelligence III: Inspection Time Elise B. Barbeau Université de Montréal Isabelle Soulières and Michelle Dawson Centre d’Excellence en Troubles Envahissants du Développement de l’Université de Montréal (CETEDUM), Fernand-Seguin Research Center Thomas A. Zeffiro Massachusetts General Hospital, Boston, Massachusetts Laurent Mottron Université de Montréal Across the autism spectrum, level of intelligence is highly dependent on the psychometric instrument used for assessment, and there are conflicting views concerning which measures best estimate autistic cognitive abilities. Inspection time is a processing speed measure associated with general intelligence in typical individuals. We therefore investigated autism spectrum performance on inspection time in relation to two different general intelligence tests. Autism spectrum individuals were divided into autistic and Asperger subgroups according to speech development history. Compared to a typical control group, mean inspection time for the autistic subgroup but not the Asperger subgroup was significantly shorter (by 31%). However, the shorter mean autistic inspection time was evident only when groups were matched on Wechsler IQ and disappeared when they were matched using Raven’s Progressive Matrices. When autism spectrum abilities are compared to typical abilities, results may be influenced by speech development history as well as by the instrument used for intelligence matching. Keywords: autism, Asperger syndrome, perception, processing speed, Raven Progressive Matrices The observation that intelligence in autism is atypical dates back to its earliest descriptions (Asperger, 1991; Kanner, 1973), with early empirical studies reporting uneven autistic perfor- mance both within (Rutter, 1966) and across (Bartak, Rutter, & Cox, 1975) commonly used intelligence tests. It is now well- established that intelligence estimates in autism vary with the instrument used for assessment (e.g., Magiati & Howlin, 2001; Mottron, 2004), but opinions conflict concerning which mea- surement tool is most accurate. Moreover, general intelligence estimates based on constructs established using typical samples may not be wholly appropriate for autism spectrum individuals, whose neurodevelopmental histories and cognitive architecture are markedly atypical. These issues have obvious implications for determining both the prevalence of intellectual disability in autism and the detailed nature of the autistic cognitive pheno- type. As autistic abilities are routinely assessed using compar- isons with intelligence-matched controls, the choice of an ap- propriate intelligence measure is a crucial procedural decision. General intelligence in autism is commonly estimated with Wechsler intelligence scales, which combine scores from an evolving battery of subtests to estimate full-scale IQ (FSIQ). The third versions of the test (WISC–III, WAIS–III; Wechsler, 1991, 1997) include a verbal IQ (VIQ) estimate derived from subtests that require both verbal comprehension and production, as well as a performance IQ (PIQ) derived from largely non- verbal subtests that, nonetheless, depend on verbal instructions. Wechsler tests are normed such that typical individuals will tend to achieve similar scores and, thus, an even cognitive profile for FSIQ, PIQ, and VIQ estimates. However, autism spectrum individuals exhibit large variations among the Wechsler subtests. These subtests produce different and uneven profiles in subgroups, divided according to speech development history (Soulières, Dawson, Gernsbacher, & Mottron, 2011). Asperger individuals, whose diagnostic definition requires non- delayed speech development, show strengths in verbal subtests such as Vocabulary and Information. Autistics, who present with delays and anomalies in speech development, are charac- terized by good perceptual and visuospatial skills and show a relative peak in the performance subtest Block Design. Overall, it remains unclear which Wechsler subtest scores, or combina- tions thereof, best estimate intelligence in autism spectrum individuals. This article was published Online First October 22, 2012. Elise B. Barbeau and Laurent Mottron, Centre d’Excellence en Troubles Envahissants du Développement de l’Université de Montréal (CET- EDUM), Fernand-Seguin Research Center and Department of Psychiatry, University of Montreal, Montréal, QC, Canada; Isabelle Soulières, Centre d’Excellence en Troubles Envahissants du Développement de l’Université de Montréal (CETEDUM), Fernand-Seguin Research Center and Depart- ment of Psychology, University of Quebec, Montreal; Michelle Dawson, Centre d’Excellence en Troubles Envahissants du Développement de l’Université de Montréal (CETEDUM), Fernand-Seguin Research Center; Thomas A. Zeffiro, Neural Systems Group, Massachusetts General Hos- pital, Boston, Massachusetts. This work was supported by a Canadian Institutes for Health Research grant to Laurent Mottron and a doctoral Award to Elise B. Barbeau. Correspondence concerning this article should be addressed to Laurent Mottron, Hôpital Rivière-des-Prairies, 7070 Boulevard Perras, Montréal, QC, Canada H1E 1A4. E-mail: mottronl@uniserve.com Journal of Abnormal Psychology © 2012 American Psychological Association 2013, Vol. 122, No. 1, 295–301 0021-843X/13/$12.00 DOI: 10.1037/a0029984 295