J. Child Psychol. Psychiat. Vol. 40, No. 5, pp. 743–755, 1999 Cambridge University Press  1999 Association for Child Psychology and Psychiatry Printed in Great Britain. All rights reserved 0021–963099 $15.000.00 Local Bias in Autistic Subjects as Evidenced by Graphic Tasks : Perceptual Hierarchization or Working Memory Deficit ? Laurent Mottron, Sylvie Belleville, and Edith Me  nard Universite  de Montre  al and Ho  pital Rivie  re-des-Prairies, Montre  al, Canada In the present study, copying tasks were used to assess hierarchical aspects of visual perception in a group of 10 nonsavant autistic individuals with normal intelligence. In Experiment 1, the hierarchical order of graphic construction and the constancy of this order were measured for the copying of objects and nonobjects. In comparison to control participants, autistic individuals produced more local features at the start of the copying. However, they did not differ from controls with respect to graphic constancy. Experiment 2 measured the effect of geometrical impossibility on the copying of figures. Results revealed that autistic individuals were less affected by figure impossibility than were controls. Therefore, these experiments seem to support the notion of a local bias for visual information processing in individuals with autism. Two interpretations are proposed to account for this effect. According to the hierarchical deficit hypothesis, individuals with autism do not manifest the normal global bias in perceiving scenes and objects. Alternatively, the executive function hypothesis suggests that autism brings about limitations in the complexity of information that can be manipulated in short-term visual memory during graphic planning. Keywords : Autistic disorder, drawing, executive function, hierarchical processing, neuro- psychology, visual perception, working memory. Abbreviations : ADI-R : Autism Diagnosis Interview-Revised. Introduction Individuals with autism are known to engage in atypical behaviors in the visual domain, including hand- flapping in front of their eyes, lateral vision, gaze avoidance, and preferences for certain visual patterns, such as flickering stimuli and spinning objects. In ad- dition, the regions and types of features used as cues for face and emotion identification appear to be abnormal in these individuals (Baron-Cohen, Wheelwright, & Jolliffe, 1997 ; Langdell, 1978). Finally, individuals with autism exhibit special abilities in the visuospatial domain (Frith & Happe  , 1994). Although these behaviors are not entirely specific to autism, they are by far more common in this condition than in any other developmental syndrome. Despite the fact that these clinical signs suggest perceptual anomalies (Young & Ellis, 1992), it has been proposed that low-level perceptual processes are intact in autism because autistic individuals are usually able to reproduce auditory pat- terns and label visual patterns (Frith, Baron-Cohen, & Paul, 1987). The conclusion that perception is intact in individuals with autism might depend on the degree of elaboration in the perceptual models in the literature. Indeed, the last 20 Requests for reprints to: Dr Laurent Mottron, Clinique spe  cialise  e des troubles envahissants du de  veloppement, Ho  pi- tal Rivie  re-des-Prairies, 7070 Boul. Perras, Montre  al (Qc), CANADA, H1E 1A4 (E-mail : mottron1istar.ca). years have seen a marked increase in the knowledge of visual perception (Ellis & Young, 1988 ; Humphreys & Bruce, 1989), which has resulted in the fragmentation of visual perception into numerous distinct operations, each of which is susceptible to impairment by neurological diseases. One area in which the refinement of knowledge was the most crucial was in the study of high-level perceptual processes (i.e. the recognition of the perceptual charac- teristics of objects). This is now known to involve several subcomponents, such as the construction of perceptual 2- D and 3-D representations, selective visual attention, feature grouping, and the hierarchization of local and global parts. Only a fraction of these mechanisms has been empirically investigated in the autistic population. In addition, it is now recognized that a number of perceptual deficits may leave the naming of objects unimpaired while severely altering other abilities such as face perception (Farah, 1990). As visual perception has become better defined, the hypothesis of a deficit in visual perceptual processes in autism has received increased empirical support (for a review, see Mottron & Belleville, 1998). Indeed, indivi- duals with autism present several peculiarities in the processing of complex visual stimuli. For example, they do not benefit from the segmentation of visual stimuli (Shah & Frith, 1993), nor do they perceive visual illusions to the same extent as controls (Happe  , 1996). Moreover, they detect hidden figures more easily than normal participants (Jolliffe & Baron-Cohen, 1997), although 743