REPORT Preverbal children with autism understand the intentions of others Michelle A. Aldridge, 1 Kari R. Stone, 1 Melissa H. Sweeney 1 and T.G.R. Bower 2 1. Callier Center for Communication Disorders, University of Texas at Dallas, USA 2. University of Texas at Dallas, School of Human Development, USA Abstract There is a `theory of mind' theory of autism. Meltzoff has recently developed a procedure that gives a nonverbal assessment of `theory of mind'. A group of children with autism and a matched control group of normally developing infants were given three of Meltzoff's tasks and three conventional, gestural imitation tasks. The children with autism showed the expected deficits on gestural imitation, but were significantly better than the normally developing infants on the Meltzoff tasks. The implications of these results for a number of theoretical issues are discussed. In recent years, there has been an increasing interest in thedevelopmentofa`theoryofmind'inchildren.Akey direction of research has been to study the development of `theory of mind' in children with autism and to compare their performance with children who are not autistic. It has been concluded by many that children with autism either lack a theory of mind or have a defective theory of mind Baron-Cohen, Leslie & Frith, 1985; see for example Mitchell, 1997, for a review). The focus of this research has been a representational model of mind. The representational model of mind is seen as a developmental successor of a simpler mentalism, a simpler belief that there are unseen desires and inten- tions at the root of observable human behavior e.g. Gopnik, Slaughter & Meltzoff, 1994). Until recently, simple mentalism was understudied due, almost cer- tainly, to the lack of a technique for exploring the relevant issues in children too young to give verbal reports. Recently, Meltzoff 1995) presented us with a solution to this problem. His behavioral re-enactment procedure presented infants with an adult who demon- strated an intention to act in a certain way. The adult never fulfilled this intention. The children were then presentedwiththeobjectorobjectswithwhichtheadult had been dealing. His results showed that 18-month-old infants successfully completed the intended acts of the adults i.e. that they perceived the intention of the adult). Meltzoff proposes that this shows a rudimentary `theory of mind', a primitive `mentalism'. In his paper, heaskswhetherchildrenwithautism,childrenwhohave beenallegedtohaveprofounddeficitsinotheraspectsof understanding the minds of others, would show deficits on the kinds of tasks devised by him. The research to be reported here was designed to answer Meltzoff's question. Our primary subjects were ten children with diag- noses of autistic spectrum disorder DSM IV), aged between 2 years 2 months and 4 years 2 months. Compared to Meltzoff's 1995) subjects, 18-month-old normally developing infants, they were very different. We decided that cognitive developmental level was the most appropriate basis for matching these children with normally developing infants. We therefore as- sessed their level of cognitive functioning using a set of object concept tasks, to be described below. Tasks of this kind provide a robust nonverbal assessment of cognitivedevelopmentWishart&Bower,1984),which has been shown to be a stable differentiator of normal and autistic children Butterworth & MacPherson, 1987). Even with such measures it is difficult to say Address for correspondence: Michelle Aldridge, Callier Center for Communication Disorders, University of Texas at Dallas, 1966 Inwood Road, Dallas, TX 75235, USA; e-mail: aldridge@utdallas.edu # Blackwell Publishers Ltd. 2000, 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA. Developmental Science 3:3 2000), pp 294±301