In his original paper, Kanner (1943) commented on the intelligent appearance of children with autism and observed that they did well on some parts of tests of intelligence. This view led to the impression that children with autism did not suffer from cognitive delay. Observed difficulties in assessment and low test scores were attributed to “negativism” or “untestability” (Brown and Pace, 1969; Clark and Rutter, 1977). As time went on, it became apparent that, although some areas of intellectual development were often relatively strong, many other areas were significantly delayed or deviant in their development and that probably a majority of children with autism functioned in the mentally retarded range. Various investigators (e.g., Rutter, 1983) began to emphasize the centrality of cognitive-communicative dysfunction.
As noted by Sigman and colleagues (1997), studies of normal cognitive development have generally focused either on the process of acquisition of knowledge (emphasizing theories of learning and information processing) or on symbolic development, concept acquisition, and skill acquisition (a combined line of work often based on the theories of Piaget), as well as questions concerning the nature of intelligence. Various authors have summarized the large and growing literature on these topics in autism (e.g. DeMyer et al., 1981; Fein et al., 1984; Prior and Ozonoff, 1998). The interpretation of this literature is complicated by the association of autism with mental retardation in many individuals, by developmental changes in the expression of autism, and by the strong interdependence of various lines of development. For example, deficits in aspects of
symbolic functioning may be manifest in problems with play at one time, and in language at a later time. In addition, individuals with autism may attempt compensatory strategies, either spontaneously or through instruction, so profiles of ability may also change over time.
Children with autistic spectrum disorders have unique patterns of development, both as a group and as individuals. Many children with autistic spectrum disorders have relative strengths that can be used to buttress their learning in areas that they find difficult. For example, a child with strong visual-spatial skills may learn to read words to cue social behavior. A child with strong nonverbal problem-solving skills may be motivated easily by tasks that have a clear endpoint or that require thinking about how to move from one point to another. A child with good auditory memory may develop a repertoire of socially appropriate phrases from which to select for specific situations.
Autistic spectrum disorders are disorders that affect many aspects of thinking and learning. Cognitive deficits, including mental retardation, are interwoven with social and communication difficulties, and many of the theoretical accounts of autistic spectrum disorders emphasize concepts, such as joint attention and theory of mind, that involve components of cognition, communication, and social understanding. Thus, educational interventions cannot assume a typical sequence of learning; they must be individualized, with attention paid to the contribution of each of the component factors to the goals most relevant for an individual child.
COGNITIVE ABILITIES IN INFANTS AND VERY YOUNG CHILDREN
Early studies on development in autism focused on basic capacities of perception and sensory abilities. Although children with autistic spectrum disorders appear to be able to perceive sensory stimuli, their responses to such stimuli may be abnormal (Prior and Ozonoff, 1998). For example, brainstem auditory evoked response hearing testing may demonstrate that the peripheral hearing pathway is intact, although the child’s behavioral response to auditory stimuli is abnormal. In infants and very young children, the use of infant developmental scales is somewhat limited, since such tests have, in general, relatively less predictive value for subsequent intelligence. Indeed, the nature of “intelligence” in this period may be qualitatively different than in later years (Piaget, 1952).
Several studies have investigated sensorimotor intelligence in children with autism. The ability to learn material by rote may be less impaired than that involved in the manipulation of more symbolic materials (Klin and Shepard, 1994; Losche, 1990). Attempts made to employ traditional Piagetian notions of sensorimotor development have revealed generally normal development of object permanence, although the capacities
to imitate gesture or vocalization may be deficient (Sigman and Ungerer, 1984b; Sigman et al., 1997). The difficulties in imitation begin early (Prior et al., 1975) and are persistent (Sigman and Ungerer, 1984b; Otah, 1987). The specificity of these difficulties has been the topic of some debate (Smith and Bryson, 1994), although it is clear that children with autism usually have major difficulties in combining and integrating different kinds of information and their responses (Rogers, 1998).
Although sensorimotor skills may not appear to be highly deviant in some younger children with autism, aspects of symbolic play and imagination, which typically develop during the preoperational period, are clearly impaired (Wing et al., 1977; Riquet et al., 1981). Children with autism are less likely to explore objects in unstructured situations (Kasari et al., 1993; Sigman et al., 1986). Younger children with autism do exhibit a range of various play activities, but the play is less symbolic, less developmentally sophisticated, and less varied than that of other children (Sigman and Ungerer, 1984a). These problems may be the earliest manifestations of what later will be seen to be difficulties in organization and planning (“executive functioning”) (Rogers and Pennington, 1991). Thus, younger children with autism exhibit specific areas of deficiency that primarily involve representational knowledge. These problems are often most dramatically apparent in the areas of play and social imitation. As Leslie has noted (e.g., Leslie, 1987), the capacity to engage in more representational play, especially shared symbolic play, involves some ability for metarepresentation. Shared symbolic play also involves capacities for social attention, orientation, and knowledge, which are areas of difficulty for children with autism.
STABILITY AND USES OF TESTS OF INTELLIGENCE
IQ scores have been important in the study of autism and autistic spectrum disorders. To date, scores on intelligence tests, particularly verbal IQ, have been the most consistent predictors of adult independence and functioning (Howlin, 1997). IQ scores have generally been as stable for children with autism as for children with other disabilities or with typical development (Venter et al., 1992). Though fluctuations of 10– 20 points within tests (and even more between tests) are common, within a broad range, nonverbal IQ scores are relatively stable, especially after children with autism enter school. Thus, nonverbal intelligence serves, along with the presence of communicative language, as an important prognostic factor. Epidemiological studies typically estimate that about 70 percent of children with autism score within the range of mental retardation, although there is some suggestion in several recent studies that this proportion has decreased (Fombonne, 1997). This change may be a function of more complete identification of children with autism who are
not mentally retarded, a broader definition of autism that includes less impaired individuals, and greater educational opportunities for children with autism in the past two decades in many countries. It will be important to consider the effects of these possible shifts on interventions.
In school-age children, traditional measures of intelligence are more readily applicable than in younger (and lower functioning) individuals. Such tests have generally shown that children with autism exhibit problems both in aspects of information processing and in acquired knowledge, with major difficulties in more verbally mediated skills (Gillies, 1965; McDonald et al., 1989; Lockyer and Rutter, 1970; Wolf et al., 1972; Tymchuk et al., 1977). In general, abilities that are less verbally mediated are more preserved, so that such tasks as block design may be areas of relative strength. Tasks that involve spatial understanding, perceptual organization, and short-term memory are often less impaired (Hermelin and O’Connor, 1970; Maltz, 1981) unless they involve more symbolic tasks (Minshew et al. 1992). There may be limitations in abilities to sequence information cross modally, particularly in auditory-visual processing (Frith, 1970, 1972; Hermelin and Frith, 1971). There is also some suggestion that in other autistic spectrum disorders (e.g., Asperger’s syndrome) different patterns may be noted (Klin et al., 1995). In addition, the ability to generalize and broadly apply concepts may be much more limited in children with autism than other children (Tager-Flusberg, 1981; Schreibman and Lovaas, 1973). As for other aspects of development, programs have been implemented to maximize generalization of learning (Koegel et al., 1999), but this process cannot be assumed to occur naturally.
In autism research, IQ scores are generally required by the highest quality journals in descriptions of participants. These scores are important in characterizing samples and allowing independent investigators to replicate specific findings, given the wide variability of intelligence within the autism spectrum. IQ is associated with a number of other factors, including a child’s sex, the incidence of seizures, and the presence of other medical disorders, such as tuberous sclerosis. Several diagnostic measures for autism, including the Autism Diagnostic Interview-Revised, are less valid with children whose IQ scores are less than 35 than with children with higher IQs (Lord et al., 1994). Diagnostic instruments often involve quantifying behaviors that are not developing normally. This means that it is difficult to know if the frequency of autism is truly high in severely to profoundly mentally retarded individuals, or if the high scores on diagnostic instruments occur as the result of “floor” effects due to the general absence of more mature, organized behaviors (Nordin and Gillberg, 1996; Wing and Gould, 1979).
IQ scores have been used as outcome measures in several studies of treatment of young children with autism (Lovaas, 1993; Sheinkopf and
Siegel, 1998; Smith et al., 2000). IQ is an important variable, particularly for approaches that claim “recovery,” because “recovery” implies intellectual functioning within the average range. However, these results are difficult to interpret for a number of reasons. First, variability among children and variability within an individual child over time make it nearly impossible to assess a large group of children with autism using the same test on numerous occasions. Within a representative sample of children with autism, some children will not have the requisite skills to take the test at all, and some will make such large gains that the test is no longer sufficient to measure their skills. This is a difficulty inherent in studying such a heterogeneous population as children with autistic spectrum disorders.
The challenge to find appropriate measures and to use them wisely has direct consequences in measuring response to treatment. For example, there is predictable variation in how children perform on different tests (Lord and Schopler, 1989a). Children with autism tend to have the greatest difficulty on tests in which both social and language components are heavily weighted and least difficulty with nonverbal tests that have minimum demands for speed and motor skills (e.g., the Raven’s Coloured Progressive Matrices [Raven, 1989]). Comparing the same child’s performance on two tests, given at different times—particularly a test that combines social, language, and nonverbal skills, or a completely nonverbal test—does not provide a meaningful measure of improvement. Even within a single test that spans infant to school-age abilities, there is still variation in tasks across age that may differentially affect children with autism; this variation is exemplified in many standard instruments such as the Stanford-Binet Intelligence Scales (Thorndike et al., 1986) or Mullen Scales of Early Development (Mullen, 1995).
Generally, IQ scores are less stable for children first tested in early preschool years (ages 2 and 3) than for those tested later, particularly when different tests are used at different times. In one study (Lord and Schopler, 1989a), mean differences between test scores at 3 years or younger and 8 years and older were greater than 23 points. These findings have been replicated in other populations (Sigman et al., 1997). Thus, even without special treatment, children first assessed in early preschool years are likely to show marked increases in IQ score by school age (Lord and Schopler, 1989b), also presumably reflecting difficulties in assessing the children and limitations of assessment instruments for younger children.
Studies with normally developing children have indicated that there can be practice effects with developmental and IQ tests, particularly if the administration is witnessed by parents who may then, not surprisingly, subsequently teach their children some of the test items (Bagley and McGeein, 1989). Examiners can also increase scores by varying breaks,
motivation, and order of assessment (Koegel et al., 1997). There are difficulties analyzing age equivalents across different tests because of lack of equality in intervals (Mervis and Robinson, 1999). Deviation IQ scores may not extend low enough for some children with autism, and low normative scores may be generated from inferences based on very few subjects. In the most extreme case, a young child tested with the Bayley Scales at 2 years and a Leiter Scale at 7 years might show an IQ score gain of over 30 points. This change might be accounted for by the change in test (i.e., its emphasis and structure), the skill of the examiner, familiarity with the testing situation, and practice on test measures—all important aspects of the measurement before response to an intervention can be interpreted.
Because researchers are generally expected to collect IQ scores as descriptive data for their samples, the shift to reporting IQ scores as outcome measures is a subtle one. For researchers to claim full “recovery,” measurement of a posttreatment IQ within the average range is crucial and easier to measure than the absence of autism-related deficits in social behavior or play. IQ scores, at least very broadly, can predict school success and academic achievement, though they are not intended to be used in isolation. Indeed, adaptive behavior may be a more robust predictor of some aspects of later outcomes (Lord and Schopler, 1989b; Sparrow, 1997). Furthermore, an IQ score is a composite measure that is not always easily dissected into consistent components. Because of the many sources for their variability and the lack of specific relationship between IQ scores and intervention methods, IQ scores on their own provide important information but are not sufficient measures of progress in response to treatment and certainly should not be used as the sole outcome measure.
Similar to findings with typically developing children, tests of intellectual ability yield more stable scores as children with autistic spectrum disorders become older and more varied areas of intellectual development can be evaluated. Although the process of assessment can be difficult (Sparrow, 1997), various studies have reported on the reliability and validity of appropriately obtained intelligence test scores (Lord and Schopler, 1989a). Clinicians should be aware that the larger the sampling of intellectual skills (i.e., comprehensiveness of the test or combination of tests), the higher will be the validity and accuracy of the estimate of intellectual functioning (Sparrow, 1997).
GENERAL ISSUES IN COGNITIVE ASSESSMENT
There are several important problems commonly encountered in the assessment of children with autism and related conditions. First, it is common to observe significant scatter, so that, in autism, verbal abilities
may be much lower than nonverbal ones, particularly in preschool and school-age children. As a result, overall indices of intellectual functioning may be misleading (Ozonoff and Miller, 1995). Second, correlations reported in test manuals between various assessment batteries may not readily apply, although scores often become more stable and predictive over time (Lord and Schopler, 1989a; Sparrow, 1997). Third, for some older children with autism standard scores may fall over time, reflecting the fact that while gains are made, they tend to be at a slower rate than expected given the increase in chronological age. This drop may be particularly obvious in tests of intelligence that emphasize aspects of reasoning, conceptualization, and generalization.
Approximately 10 percent of children with autism show unusual is-lets of ability or splinter skills. These abilities are unusual either in relation to those expected, given the child’s overall developmental level, or, more strikingly, in relation to normally developing children. The kinds of talents observed include drawing, block design tasks, musical skill, and other abilities, such as calendar calculation (Treffert, 1989; Shah and Frith, 1993; Prior and Ozonoff, 1998). Hermelin and colleagues (e.g., Hermelin and Frith, 1991) noted that these unusual abilities may be related to particular preoccupations or obsessions. Such abilities do not seem to be based just on memory skills; they may reflect other aspects of information processing (Pring et al., 1995).
In summary, general measures of intellectual functioning, such as IQ scores, are as stable and predictive in children with autistic spectrum disorders as in children with other developmental disorders, but this does not mean that these measures do not show individual and systematic variation over time. Because IQ scores provide limited information and there are complex implications of test selection across ages and developmental levels, IQ scores should not be considered a primary measure of outcome, though they may be one informative measure of the development of the children who participate in an intervention program. Specific cognitive goals, often including social, communicative, and adaptive domains, are necessary to evaluate progress effectively. Direct evaluations of academic skills are also important if children are learning to read or are participating in other academic activities.
THEORETICAL MODELS OF COGNITIVE DYSFUNCTION IN AUTISM
Various theoretical notions have been advanced to account for the cognitive difficulties encountered in autism. The “theory of mind” hypothesis proposes that individuals with autism are not able to perceive or understand the thoughts, feelings, or intentions of others; i.e., they lack a theory of mind and suffer from “mind blindness” (Leslie and Frith, 1987;
Leslie, 1992; Frith et al., 1994). Various experimental tasks and procedures used to investigate this capacity generally indicate that many somewhat more able (e.g., verbal) children with autism do indeed lack the capacity to infer mental states. This capacity is viewed as one aspect of a more general difficulty in “metarepresentation” (Leslie, 1987) that is presumed to be expressed in younger children by difficulties with understanding communicative gesture and joint attention (Baron-Cohen, 1991). While not all children with autistic spectrum disorders entirely lack a theory of mind (Klin et al., 1992), they may be impaired to some degree (Happe, 1994). There appear to be strong relationships between verbal ability and theory of mind capacities in autism (e.g., Ozonoff et al., 1991), though many language-impaired non-autistic children can normally acquire these skills (Frith et al., 1991). The theory of mind hypothesis has been a highly productive one in terms of generation of research, and in focusing increased attention on the social aspects of autism, including deficits in joint attention, communication, and pretense play (see Happe, 1995, for a summary). However, specific behaviors that evidence a deficit in theory of mind are not by themselves sufficient to yield a diagnosis of autism, which can be associated with other cognitive deficits. In addition, research in which theory of mind concepts were taught to individuals with autism did not result in general changes in social behavior, suggesting that links between theory of mind and sociability are not simple (Hadwin et al., 1997).
A second body of work has focused on deficits in executive functioning, that is, in forward planning and cognitive flexibility. Such deficits are reflected in difficulties with perseveration and lack of use of strategies (see Prior and Ozonoff, 1998). Tests such as the Wisconsin Card Sort (Heaton, 1981) and the Tower of Hanoi (Simon, 1975) have been used to document these difficulties. In preschool children, the data on executive functioning deficits are more limited. McEvoy and colleagues (1993) used tasks that required flexibility and response set shifting, and noted that younger children with autism tended to exhibit more errors in perseveration than either mentally or chronologically age-matched control children. More recently, others did not find that the executive functioning in preschoolers with autistic spectrum disorders differed from that in other children (Griffith et al., 1999; Green et al., 1995).
A third area of theoretical interest has centered on central coherence theory, in which the core difficulties in autism are viewed as arising from a basic impairment in observing meaning in whole arrays or contexts (Frith, 1996; Jarrold et al., 2000). As Frith (1996) has noted, it is likely that a number of separate cognitive deficits will be ultimately identified and related to the basic neurobiological abnormalities in autism.
Neuropsychological assessments are sometimes of help in documenting sensory-perceptual, psychomotor, memory, and other skills. The util-
ity of more traditional neuropsychological assessment batteries in children, especially in young children, is more limited than for adults. Extensive neuropsychological assessments may not provide enough useful information to be cost-effective. However, selected instruments may be helpful in answering specific questions, particularly in more able children. Exploring a child’s visual-motor skills or motor functioning can be of value for some children whose learning and adaptation appear to be hindered by deficits in these skills. (Motor and visual motor skills are discussed in detail in Chapter 8.)
ACADEMIC INSTRUCTION AND OUTCOMES
In addition to interventions that have been designed to improve intellectual performance (e.g., scores on IQ tests), there is a small literature on instructional strategies designed to promote the academic performance of young children with autism. Academic performance, for this discussion, refers to tasks related to traditional reading and mathematics skills. This literature consists primarily of single-subject design, quasi-experimental design, and descriptive observational research, rather than randomized clinical trials. The studies have usually included children with autism at the top of the age range covered in this report (i.e., ages 5–8), and the participant samples often include older children with autistic spectrum disorders as well. Notwithstanding these caveats, there is evidence that some young children with autistic spectrum disorders can acquire reading skills as a result of participation in instructional activities. There is very limited research on instructional approaches to promoting mathematics skills.
A range of instructional strategies have involved children with autistic spectrum disorders. In early research, Koegel and Rincover (1974) and Rincover and Koegel (1977) demonstrated that young children with autism could engage in academic tasks and respond to academic instruction as well in small-group instructional settings as they did in one-to-one instruction with an adult. Kamps and colleagues replicated and extended these findings on small-group instruction of academic tasks to a wider range of children within the autism spectrum and other developmental disabilities (Kamps et al., 1990; Kamps et al., 1992).
In another study, Kamps and colleagues (1991) first performed descriptive observational assessment of children with autism in a range of classroom settings. They used these data to identify the following commonly used instructional approaches:
Incorporate naturally occurring procedures into intervention groups across classrooms.
Include three to five students per group.
Use individual sets of materials for each student.
Use combination of verbal interaction (discussion format) and media.
Use five-minute rotations of media/concept presentation.
Use a minimum of three sets of materials to teach each concept.
Use frequent group (choral) responding.
Use fast-paced random responding.
Use serial responding—three to five quick responses per student.
Use frequent student-to-student interactions.
They then conducted a series of single-subject designs that demonstrated experimentally (with treatment fidelity measures documenting implementation) the relationship between the instructional measures and the children’s performance on criterion-referenced assessments of academic tasks. This combination of instructional strategies (choral responding, student-to-student responding, rotation of materials, random student responding) was also found to be effective in teaching language concepts to elementary-aged children with autism in a later study (Kamps et al., 1994a). In their subsequent research, Kamps and colleagues (1994b) have examined the use of classwide peer tutoring (i.e., classmates provide instruction and practice to other classmates) with young children with autistic spectrum disorders. In a single-subject design study, these researchers found increased reading fluency and comprehension for children who received peer tutoring, as compared with those who received traditional reading instruction.
Other strategies have also appeared in the literature. Using an incidental teaching technique, McGee and colleagues (1986) embedded sight-word recognition tasks in toy play activities and found that two children with autism acquired sight-word recognition skills and generalized those skills to other settings. Cooperative learning groups are another instructional approach. Provided tutoring by peers, a group of children with autistic spectrum disorders practiced reading comprehension and planned an academic game; the children increased their academic engagement in reading (Kamps et al., 1995).
There is also some evidence that children with autism might benefit from computer-assisted instruction (CAI) in reading. Using a single-subject design, Chen and Bernard-Optiz (1993) compared delivery of academic tasks by an instructor or through a computer monitor and found higher performance and more interest from children in the CAI than the adult-delivered intervention. In a study conducted in Sweden, Heimann and colleagues (1995) used a CAI program and a traditional instructional approach to present lessons to students. Children with autism made significant gains in the CAI program (compared with traditional instruction), while typically developing children progressed similarly in both
settings. These two studies suggest that a CAI format for presenting instruction to young children with autism may be useful, but the results are far from conclusive and require further study.
FROM RESEARCH TO PRACTICE
There is need for research on the development of more specific measures of important areas of outcome in cognition, including the acquisition and generalization of problem-solving and other cognitive skills in natural contexts (e.g., the classroom and the home) and the effects of these skills on families and other aspects of children’s lives. There is also a need for research to define appropriate sequences of skills that should be taught through educational programs for young children with autistic spectrum disorders, as well as methods for selecting those sequences, while developing programs for individual children.