Journal of Autism and Developmental Disorders, Vol. 28, No. 6, 1998 Attentional Capacities in Children with Autism: Is There a General Deficit in Shifting Focus? Daisy M. Pascualvaca,1,3 Bryan D. Fantie,1,2 Maria Papageorgiou,1 and Allan F. Mirsky1 Twenty-three children with autism and two control groups completed an attention battery comprising three versions of the continuous performance test (CPT), a digit cancellation task, the Wisconsin Card Sorting Test (WCST), and two novel, computerized tests of shifting attention (i.e., the Same-Different Computerized Task and the Computerized Matching Task). Children with autism could focus on a particular stimulus and sustain this focus as indicated by their performance on the digit cancellation task and the CPT Their performance on the WCST suggested problems in some aspects of shifting attention (i.e., disengaging attention). The autism group performed as well as controls on the Same-Different Computerized Task, however, that required successive comparisons between stimuli. This implies that they could, in fact, shift their attention continuously. In addition, they did not differ from controls on the Computerized Matching Task, an analog of the WCST, suggesting that they do not have a general deficit in shifting attention. KEY WORDS: Attention deficit; autism; shifting focus. nated by certain objects or qualities of objects, and may spend an inordinate amount of time repeating an activity. These characteristics have led to a number of theories that implicate disordered attention as the core deficit of autism. For example, several investigators have suggested that altered arousal mechanisms and an aberrant response to novelty underlie the behavioral peculiarities of this disorder (Dawson & Lewy, 1989; Hutt, Hutt, Lee, & Ounsted, 1964). More recently, researchers have found evidence of an inefficient attentional lens (Burack, 1994) and a deficit in shifting attention (Courchesne et al, 1994b). In this study, we wanted to assess the attentional capacities of children with autism using a comprehensive battery that included the measures used by Mirsky and colleagues (Mirsky, 1987; Mirsky, Anthony, Duncan, Ahearn, & Kellam, 1991). These investigators found that different measures of attention assess unique attentional processes including the ability to focus on a particular stimulus, sustain this focus over time, and shift it flexibly and INTRODUCTION The attentional capacities of children with autism have puzzled investigators and clinicians for many years. These children may be unresponsive to approaches by adults and other children, or fail to respond to an environmental stimulus such as an unexpected noise (Kanner, 1943, 1944; Rimland, 1964). Despite their difficulties in attending to a particular stimulus on demand and their overall appearance as aloof or distant, some children with autism appear to have an unusual ability to focus on particular aspects of their environment (Lovaas, Koegel, & Schreibman, 1979). For instance, they may be fasci1 Section on Clinical and Experimental Neuropsychology, Laboratory of Brain and Cognition, National Institute of Mental Health, Bethesda, Maryland. 2 Human Neuropsychology Laboratory, Department of Psychology, The American University, Washington, DC. 3 Address all correspondence to Daisy M. Pascualvaca, Section on Clinical and Experimental Neuropsychology, Laboratory of Brain and Cognition, National Institute of Mental Health, Building 15K, Room 103, Bethesda, Maryland 20892-1366. 467 0162-3257/98/1200-0467$15.00/0 © 1998 Plenum Publishing Corporation 468 adaptively. These processes, although integrated into a single functional entity, are mediated by distinct neuroanatomical regions (see Mirsky, Fantie, & Tatman, 1995). By using this comprehensive battery, we expected to isolate specific strengths and weaknesses in the attentional capacities of children with autism. Results of published studies suggest that children with autism do not have any marked difficulties in sustaining their focus of attention to repetitive, predictable stimuli (Garretson, Fein, & Waterhouse, 1990). In contrast, these children often have difficulty shifting their focus of attention. This difficulty has been demonstrated with neuropsychological measures such as the Wisconsin Card Sorting Test (WCST; Ozonoff, 1995; Ozonoff, Pennington, & Rogers, 1991; Prior & Hoffmann, 1990) as well as with reaction time tasks (Courchesne, 1995; Courchesne, Akshoomoff, Townsend, & Saitoh, 1995; Courchesne et al, 1994a, 1994b). To clarify this deficit in shifting attention, we devised two computerized tasks that make demands upon two different aspects of shifting attention. In the first task, the children had to decide whether some feature of three targets was either the same or different, and therefore, had to change their focus of attention continuously between the features of these stimuli. We designed the second task to resemble the WCST whereby the children had to shift their attention after having focused on a specific target feature for some time. Based on the studies conducted by Courchesne and colleagues (Courchesne, 1995; Courchesne et al., 1994a, 1994b, 1995), who found that children with autism had difficulty establishing a new focus of attention, we expected these children to have particular difficulties with the second task. We were also interested in determining whether the children's difficulties in attention tests were secondary to decreased motivation. There is considerable evidence to indicate that tangible rewards facilitate learning (Lovaas, 1977, 1993) and attentional performance (Garretson et al., 1990) in children with autism. Despite their susceptibility to various reinforcement modalities, we hypothesized that if these children had a real core deficit in attention, increasing their motivation should not be sufficient to compensate for such deficit. Pascualvaca, Fantie, Papageorgiou, and Mirsky METHOD Subjects Twenty-three children with autism (15 boys and 8 girls) participated in the study. They were recruited from several chapters of the Autism Society of America and all had been diagnosed with autism by local practitioners. Their diagnoses were corroborated through interviews with the parents, reviews of the records, and observations of the children during testing. All children met DSM-III-R criteria for Autistic Disorder (American Psychiatric Association, 1987) and scored above the criterion for autism on the Childhood Autism Rating Scale (CARS; Schopler, Reichler, Rochen-Renner, 1988). None of the children had a history of seizures, fragile X, or were taking medication at the time of the study. The children ranged in age from 6 to 12 years, with a mean age of 8 years 7 months (SD = 1.1). As can be seen in Table I, their level of functioning was fairly high. Their full-scale IQ scores, as assessed by the WISC-III (Wechsler, 1991), ranged from 55 to 107. Their performance on the Woodcock-Johnson Psycho-Educational Battery-Revised (Woodcock & Johnson, 1990) was within the average range in reading and lower than average in mathematics. The severity of their symptoms was also mild and all the children were rated as mildly to moderately autistic on the CARS (range = 30-36; M = 33.6). We included two control groups to differentiate the attentional problems specific to autism from those associated with global cognitive deficits or developmental delays. We matched the groups by skill level rather than degree of intellectual impairment. One of the control groups was matched to the autism group by verbal mental age, which usually represents the area of weakness in the cognitive profile of these children. The other control group was matched by performance mental age, typically their area of strength. We decided to use two control groups rather than a single group matched by IQ because we believe that full-scale IQ does not capture the disparate cognitive skills of children with autism. We recruited children for the control groups from local schools, and screened them with a brief parental interview and the WISC-III. We then identified those children who did not have any medical, behavioral, or academic difficulties, and matched them with the children with autism by either verbal mental age or performance mental age. Each child Attention in Children with Autism 469 Table I. Characteristics of the Sample Children with autism (n = 23) Age (years) Chronological age Verbal MA Performance MA WISC-III Verbal IQ Performance IQ Full-scale IQ Woodcock-Johnson Battery Broad reading Broad mathematics Verbal MA controls (n = 23) Performance MA controls (n = 23) M SD M SD M SD 8.7 6.4 7.3 1.1 2.0 2.0 5.11° 1.7 2.1 2.2 6.7° 6.4 6.5 7.2 7.3 1.8 2.2 2.0 74.4 84.7 77.6 16.4 17.9 14.5 106.1° 107.5a 107.6° 11.6 12.2 12.0 108.3° 109.7° 110.2° 10.2 94.7 79.3 14.6 21.6 108.5a 108.9° 11.4 17.1 111.1° 112.9a 10.2 19.4 10.9 8.9 a Significantly different from the autism group (p < .001). with autism was matched individually with two controls by sex and mental age within 6 months. As expected, children in the verbal-matched group were the youngest, although children in both control groups were significantly younger than those in the autism group. They also had higher IQ scores and performed better academically than children with autism (see Table I). Measures As indicated previously, we included the measures Mirsky and colleagues have used to assess distinct attentional processes (Mirsky, 1987; Mirsky et al, 1991). In addition, we devised two computerized tests to measure different aspects of shifting attention. We made every effort to make these tasks interesting and motivating for the children with autism given their difficulty engaging in adult-directed activities and their sensitivity to primary reinforcers. A description of all the measures follows: The Continuous Performance Test (CPT). The CPT (Rosvold, Mirsky, Saranson, Bransome, & Beck, 1956) has been traditionally used to assess an individual's ability to sustain attention over time. We used the Sunrise Systems version of the CPT operated by a Toshiba 1000 laptop computer, and included three versions of the test: two visual and one auditory. In the first visual version, the children saw individual letters presented on the screen of the stimulus response unit (SRU) and had to respond as quickly as possible whenever they saw the letter X. Each of the letters appeared for 200 ms; they were separated by an interstimulus interval of 1,500 ms. The second visual CPT was identical to the simple X task except that the children had to respond to an AX sequence (i.e., respond to the letter X only when it was preceded by an A). In the auditory version, each child listened through earphones to digitized letters (recorded by a male speaker) and had to respond to the letter O when preceded by an L. Each stimulus lasted 500 ms. To increase motivation, we attached a Gerbrands pellet dispenser to the SRU to deliver one M&M chocolate candy whenever the child made a correct response. The children received the accumulated M&Ms at the termination of the session. All children were given practice trials until they detected three consecutive targets and it was clear that they understood the instructions. All CPT versions comprised a total of 76 targets that appeared in a pseudorandom order (i.e., limited by the target probability of .20) and each lasted approximately 15 minutes. In each task, we recorded the number of correct responses (hits), omission errors (misses), commission errors (false alarms), reaction time for hits, and reaction time variability. We also recorded the individual responses in order to evaluate specific types of commission errors as well as changes in response characteristics over time. The Digit Cancellation Task. The Digit Cancellation Task, modeled after the distractibility test described by Lifshitz, Kugelmass, and Karov (1985), measures the ability to focus on a particular stimulus, ignore extraneous stimuli, and make a rapid response. In this task, the child has to scan a 15 x 10 cm array of numbers and cross out the numbers 3 470 Pascualvaca, Fantie, Papageorgiou, and Mirsky Fig. 1. and 7 as quickly as possible. We recorded the time to scan the array (completion time), the number of omission errors, and the number of commission errors. The Wisconsin Card Sorting Test. The WCST (Heaton, 1981) is commonly used to assess problem solving, concept formation, and the ability to shift set. In this test, the child has to match a series of 128 cards to four targets using one of three sorting principles (color, form, and number). The examiner gives feedback (right or wrong) after each card placement, but does not give any further instructions. The examiner changes the correct sorting criterion after ten consecutive correct sorts, and the task continues until the subject has completed six categories or has sorted all the cards. We administered and scored the WCST according to the published guidelines (Heaton, 1981). The scores derived from this test include the number of categories completed, the number of perseverative errors, the number of trials to the first category, and the number of failures to maintain set. Computerized Matching Task. The first computerized test that we designed resembled the WCST On this test, the child saw four spaceships on a computer screen, one in the top half and the other three in the bottom, and had to match the top spaceship to one of the bottom spaceships (see Fig. 1); the possible matching principles were color, size, and chassis type. If the child used the correct matching principle, the spaceships blew up and this explosion produced a variety of visual and sound effects that were clearly attractive to the children; if the child did not match the stimuli correctly, a digitized voice simply said "no." The task consisted of a total of 128 stimuli. We simplified this task relative to the WCST by eliminating ambiguous choices (i.e., those that matched the stimulus cards by more than one attribute) and reduced the number of possible choices to three. In addition to the types of scores derived from the WCST, we also recorded reaction times for individual responses. Same-Different Computerized Task. The second computerized test assessed a different aspect of shifting attention. In this test, the child saw three spaceships on the computer screen and had to press a key on a response box if the three spaceships were the same, and another key if they were different. The spaceships varied in terms of color, size, and chassis type. This test had three levels of difficulty. In the first, easiest level, the three spaceships were either identical or completely different (i.e., differed in color, size, and type). In the second level, the spaceships were considered to be the same if they shared two features. For example, they could be of the same color and size, but a different type. In the most difficult level, the same responses only had one feature Attention in Children with Autism 471 Fig. 2. in common (e.g., same color, but different size and type) (see Fig. 2). The children got auditory feedback after each response; a brief music melody when they made a correct response and a buzzer when the response was incorrect. The task consisted of a total 90 stimuli and lasted approximately 3 minutes. Prior to each part of the task, we gave the children specific instructions (regarding the various matching principles) and five practice trials with feedback. It was clear that all children understood the task requirements during the practice trials. The measures derived from this task included the number of correct responses, errors, and reaction time. We designed this task to elicit continuous comparisons among the stimuli; these comparisons, we believe, involve a type of shifting performance not assessed by sorting tasks such as the WCST Procedure The initial visit of the children in the autism group consisted of an interview with the parents and the administration of the WISC-III at the outpatient clinic of the National Institute of Mental Health. If the children were suitable for the study, they were scheduled for two additional visits, each lasting approximately 2 hours. For the children in the control groups, we administered the WISC-III and the parent interview at their school. If the children did not have any significant difficulties (as outlined earlier) and their mental ages matched those of any of the children in the autism group, they were asked to come to the clinic to complete the evaluation. Examiners trained in the administration of the various procedures presented the tests in random order with the exception of the Same-Different Computerized Task, which always followed the WCST and the Computerized Matching Task to ensure that the children did not know the three sorting principles in advance. All children had the consent of their par- ents and were paid for their participation in the study. RESULTS The autism group was compared to each of the control groups using ANOVAs. The significance level for all comparisons was set at a = .05. Focused Attention Table II shows the results of these comparisons for the Digit Cancellation Task, which assesses the child's ability to focus on a particular stimulus, ignore irrelevant stimuli, and make a rapid response. As can be seen in Table II, children with autism did not differ from either control group on any of the measures derived from this task. They took the same time to scan the array, and made comparable numbers of omission and commission errors. Sustained Attention Three children, one in the autism group and two in the control groups, made over 100 incorrect responses on the CPT These children, as well as their matched pairs, were eliminated from these analyses because their performance was clearly outside the normal range. This reduced the total number of pairs of children with valid CPT data to 20. Our results indicated that none of the CPT versions differentiated between the groups. Children with autism did not differ from either control group in the number of correct responses, commission errors, reaction time, or reaction tune variability on any of the CPT versions (Table III). Moreover, the performance of individual children in the various groups was very similar. For example, on the AX version of the CPT, 15 children with autism identified more than 80% of the targets and only 1 child recognized less than 25% of the targets. This compared to 18 472 Pascualvaca, Fantie, Papageorgiou, and Mirsky Table II. Performance on the Digit Cancellation Task Children with autism (n = 23) Digit Cancellation Task Completion time Commission errors Omission errors Verbal MA controls (n = 23) Performance MA controls (n = 23) M SD M SD M SD 125.1 78.6 117.1 62.6 108.3 63.2 0.2 6.4 0.6 6.8 0.8 6.7 2.8 6.4 0.7 4.7 2.7 6.3 Table III. Performance on the Continuous Performance Tests Children with autism (n = 20) Continuous Performance Tests CPT (Visual X version) No. correct No. incorrect Reaction time Reaction time SD CPT (Visual AX version) No. correct No. incorrect Reaction time Reaction time SD CPT (Auditory LO version) No. correct No. incorrect Reaction time Reaction time SD Verbal MA controls (n = 20) Performance MA controls (n = 20) M SD M SD M SD 65.7 15.3 655.7 200.4 12.7 21.8 162.3 57.9 67.1 13.6 672.8 170.8 12.5 23.8 144.2 51.2 69.2 11.0 647.8 162.9 12.2 23.4 140.5 52.1 63.2 26.2 571.9 221.8 11.2 29.7 169.6 75.9 64.8 20.0 538.6 200.4 9.2 27.8 135.4 62.8 65.8 21.3 525.9 188.1 31.8 138.9 69.5 56.6 28.1 740.4 306.2 17.5 31.4 246.6 86.9 56.7 31.1 732.9 343.5 15.7 35.8 156.6 88.6 59.5 17.1 715.0 320.2 16.7 24.5 143.1 102.5 children with excellent scores and one child with below average performance in each of the control groups. The children with autism performed differently in the various tasks, obtaining the highest scores in the visual X version and the lowest scores in the auditory version. This pattern of performance was not unique to the autism group, however, but characteristic of all children; this was evident in the lack of significant interactions between CPT version and group membership. An examination of the different types of commission errors revealed that the autism group made significantly more A-not-X errors than the performance mental age (MA) group in the auditory CPT (i.e., in the auditory version, this would mean responses to a letter other than an O following an L). This type of error is suggestive of impulsivity (Halperin et al., 1988). There were no significant dif- 9.6 ferences among the groups in other types of incorrect responses. We then examined changes in performance over time with repeated measures ANOVAs with group as a between factor and time on task (divided by quarters) as a within factor. These analyses revealed that reaction time increased directly with time spent on the task for all the CPT versions and groups. There were no differences in the number of correct or incorrect responses, and no significant interactions between group and time on task. Tests of Shifting Attention As the program for the computerized tests of shifting attention had not been completed when we started testing the children with autism, not all sub- Attention in Children with Autism 473 Table IV. Performance on the Tests of Shifting Attention Children with autism Same-different Task (n = 17) Part A No. of errors Reaction time (ms) Part B No. of errors Reaction time (ms) Part C No. of errors Reaction time (ms) WCST (n = 23) No. of categories No. correct Nonperseverative errors Perseverative errors Trials to first category Failures to maintain set Computerized Matching Test (n = 15) No. of categories No. correct Nonperseverative errors Perseverative errors Trials to first category Failures to maintain set Verbal MA controls Performance MA controls M SD M SD M SD 4.6 1639.8 5.4 1000.7 3.7 1758.3 5.9 928.9 1.7 1178.0 2.5 416.8 6.2 2201.4 4.9 1241.3 5.8 2576.3 4.2 1919.3 4.1 1948.4 3.4 1165.3 9.8 2575.1 4.4 1729.9 9.1 3319.4 4.9 2711.9 7.6 2305.3 4.1 1302.9 2.1 49.9 12.2 60.7 12.5 0.4 2.1 16.2 10.9 28.9 14.7 0.6 3.9b 68.2C 24.5b 25.1C 16.8 1.1b 1.9 13.4 15.4 19.1 13.9 0.9 4.3C 69.6C 20.7a 24.0C 16.3 0.8a 1.9 11.7 13.1 19.1 13.8 0.8 3.2 61.2 21.1 41.9 15.5 1.3 1.7 14.1 10.7 15.7 5.1 1.0 3.4 58.1 22.3 33.7 31.7a 1.2 2.1 12.9 10.2 16.8 26.6 1.2 3.5 61.9 23.4 33.4 25.3 1.2 2.1 9.5 11.1 16.6 18.3 1.0 a p < .05. p < .01. c p < .001. b jects received these tests. The number of children who took each test is shown in Table IV As can be seen in Table IV, the three groups were indistinguishable in the Same-Different Computerized Task. The children's performance declined steadily from the first part to the third part of this test, and showed an increasing number of errors and longer reaction times. There were no significant differences, however, between the autism and either of the control groups, suggesting that all children were equally proficient at making successive comparisons between stimuli. We did find group differences in the Wisconsin Card Sorting Test (see Table IV). Children with autism completed fewer categories and made more perseverative and nonperseverative errors than children in both control groups. These children understood the rules, as it is evident by the number of trials they took to complete the first category. They were also able to maintain a particular strategy and, in fact, had fewer failures to maintain set than children in the control groups. Yet, most of them were unable to shift strategies and only 3 children with autism completed six categories; this compared to 8 children in the verbal MA group and 10 children in the performance MA group who completed six categories. Conversely, 7 children with autism did not complete a single category (compared to only 1 child in each of the control groups). We obtained different results with the Computerized Matching Task, the computerized analog of the WCST (see Table IV). On the Computerized Matching Task, all children completed the same number of categories and made a comparable number of perseverative and nonperseverative errors. They also did not differ in the number of failures to maintain set. In fact, the only significant difference we found favored the children with autism in that they took fewer trials to complete the first category than children matched for verbal MA. In addition to their equivalent test scores, all groups exhibited a similar profile in their reaction times on the Computerized Matching Task. Figure 3 shows the reaction times of the three groups in the 474 Pascualvaca, Fantie, Papageorgiou, and Mirsky Fig. 3 trials composing the first category. As can be seen in this figure, all children responded slowly (>4 seconds) in the initial trials, but much faster (<3 seconds) after they had made two correct responses. Their reaction times again increased immediately after the completion of the category (i.e., when they were told that their response was incorrect). This finding indicates that children with autism understood the feedback as well as their peers, showed a similar learning curve, and took more time to respond in the trials immediately following the first incorrect match. We did not find any significant differences in reaction time between the groups either immediately before or after the sorting criterion was changed. We then examined the children's responses on the WCST and the Computerized Matching Task to assess their ability to disengage and reengage their focus of attention. We measured their ability to disengage attention by counting the number of trials they took to use a different matching principle. As an index of reengaging attention, we counted the number of trials the children took to match three consecutive stimuli (or five consecutive stimuli on the WCST) using the same principle, regardless of whether the principle was correct or incorrect. The children with autism did not differ from the controls on the number of trials they took to engage their focus of attention on the WCST (autism group = 7.0 trials, verbal MA group = 8.6, and performance MA group = 8.3), but took significantly longer to disengage their attention (39.2 trials vs. 7.2 and 12.3 trials for the verbal MA and performance MA groups, respectively). In contrast, all groups took the same number of trials to engage (autism = 10.6, verbal = 13.9, and performance = 8.2) and disengage (autism = 4.3, verbal = 3.7, and performance = 3.1) their focus of attention on the Computerized Matching Task (p > .10). We expected all children to perform better on the Computerized Matching Task relative to the WCST because this task comprised meaningful objects and we simplified it by eliminating the ambiguous choices and reducing the possible response options to three. Nevertheless, our findings did not support this prediction. When we compared these two tests using repeated measures ANOVAs (with group as the between variable and test version as the within variable), we found a significant interaction between group membership and test version on number of categories, correct responses, and total number of errors. Post hoc comparisons revealed that children with autism tended to perform better on the computerized test, whereas the reverse was true for children in both control groups. These comparisons did not reach statistical significance, we believe, because of lack of statistical power. Although only 15 children with autism took the Computerized Matching Task, the present results do not appear to be secondary to differences in sample composition. Children who took both tests did not differ from those who only completed the computerized test on any of the WCST measures. Both groups were also equivalent in terms of IQ, age, and CARS Attention in Children with Autism scores. Furthermore, we obtained the same results when we restricted our analyses to the 15 pairs of children who took all the tests. The robustness of this finding was strengthened when we compared the performance of the autism group to control children who were the same age (13 boys and 7 girls). Even though this group of children had significantly higher IQ scores, the groups did not differ in the number of trials to complete the first category (15.5 trials in the autism group vs. 20.2 in the same-age controls), number of failures to maintain set (1.3 vs. 1.1), number of categories (3.2 vs. 4.0), or number of nonperseverative responses (21.1 vs. 19.8). In fact, the autism group differed from controls only in the number of perseverative errors (41.9 vs. 25.9). DISCUSSION Children with autism have very unusual attentional capacities. They can focus, for example, on minute details and carry out certain activities repetitively without seemingly getting bored or tired. At the same time, however, they can be very difficult to engage and often fail to respond to cues in their environment. In this study, we administered a comprehensive attentional battery to a group of children with autism to define with more precision their attention problems, and shed some light into the nature of their disorder. Our results indicate that these children have no difficulty focusing on a particular stimulus and sustaining this focus over time. They were able, for instance, to ignore irrelevant stimuli and detect infrequently presented targets rapidly and accurately on the CPT and the Digit Cancellation Task. Similarly, they did not show any appreciable decline in their performance over time relative to the controls. They were more skillful at processing visual than auditory information. All groups exhibited the same profile, however, suggesting that processing information aurally is generally more difficult for young children, not only for those with autism. This difficulty processing auditory information is consistent with the results of published studies (Lincoln, Dickstein, Courchesne, Elmasian, & Tallal, 1992). The only significant difference we found in the focus and sustained attention tasks suggested that children in the autism group were more impulsive in the auditory CPT than the performance MA group. 475 Since impulsivity typically declines as the child matures (Halperin, Sharma, Greenblatt, & Schwartz, 1991), this finding may signal a developmental delay rather than a true deficit. This interpretation was supported by the lack of significant differences between the autism group and the verbal MA group who were younger than the performance MA controls. Some studies have shown that children (Courchesne et al, 1994a, 1994b, 1995; Garretson et al, 1990) and adults (Casey, Gordon, Mannheim, & Rumsey, 1993) with autism and other pervasive developmental disorders have an adequate ability to focus and sustain attention. These abilities are consistent with the repetitive behaviors and tendency to focus on a particular stimulus characteristic of people with autism. On the other hand, their typical behavioral presentation also suggests problems in shifting attention. Our data, however, do not support fully this interpretation of the deficit. Children in the autism group were able to make comparisons and shift their attention continuously on the Same-Different Task. They did have difficulty in the WCST a task that assesses several cognitive processes including problem-solving skills and the ability to shift set. On this task, they clearly understood the instructions and, at least initially, performed like controls. Nevertheless, once they started to use a particular strategy, they could not change it, and failed to benefit from the examiner's feedback. Similar results have been reported in other studies of children (Ozonoff, 1995; Ozonoff & McEvoy, 1994; Ozonoff et al, 1991) and adults (Rumsey, 1985; Rumsey & Hamburger, 1990) with autism. The children's adequate performance in the Same-Different Task and their difficulties in the WCST suggest that they could shift then: attention continuously, but could not shift it when they were already engaged in a particular activity. The results of the studies conducted by Courchesne et al. (1994a, 1994b, 1995) are also consistent with this interpretation. In the tasks these investigators used, the children have to focus, for example, on information presented visually until they detect a target; once they detect a target, they have to shift their attention to the material presented aurally until they make a correct detection, and so on. Therefore, in these tasks the children have to engage their attention for some time before they have to shift it. The problems that children with autism have in these tasks have been attributed to a difficulty in disengaging or rees- 476 tablishing their focus of attention (Casey et al., 1993; Courchesne et al., 1994a, 1994b, 1995). We found the same difficulty in disengaging attention on the WCST However, we also found these children to be as skillful as controls in their ability to disengage and reestablish their focus of attention on the Computerized Matching Task, an analog of the WCST On the computerized test, children with autism did not differ from controls, both in terms of the number of correct responses and errors, as well as reaction time. There are several plausible explanations for this finding. First, it is possible that the computerized task was easier than the WCST because we eliminated ambiguous choices and reduced the sorting options from four to three. Children with autism, in fact, seemed to perform slightly better on this task than on the WCST but, given the results of Prior and Hoffmann's study (1990), it is unlikely that they benefitted from this change. Prior and Hoffmann also used a modified version of the WCST in which they eliminated ambiguous responses and told the children when the sorting rules had changed. Despite these changes, the children in the autism group still performed worse on this task than the chronologicaland MA-matched controls. Another possibility is that children with autism were not receptive to the feedback they got from the examiners on the WCST and, therefore, had more difficulty with the social aspects of this task. This possibility is consistent with several lines of investigation. For example, Garretson et al. (1990) found that children with pervasive developmental disorders performed better on the CPT when given tangible reinforcers than when they were reinforced verbally by the examiners. Ozonoff (1995) also questioned the role of social/motivational factors on the WCST She administered the traditional WCST and a computerized version to a group of children with autism and a group of controls matched by IQ and chronological age, and found that the autism group performed less well on the standard test, but did not differ from the controls in the computerized test. Furthermore, all children did better in the computerized version of the WCST relative to the standard version, although both tasks appeared to be equivalent. Our results differ from Ozonoff's in that our control groups tended to perform worse on the Computerized Matching Task than on the WCST This dismay also be secondary to crepancy social/motivational factors. Children in our control groups were much younger than those included in Pascualvaca, Fantie, Papageorgiou, and Mirsky Ozonoff's study, and younger children may be especially susceptible to social interactions, structure, and feedback provided by the examiners. The role of motivational factors has also been implicated to account for the deficits schizophrenic subjects exhibit on the WCST (Bellack, Mueser, Morrison, Tierney, & Podell, 1990; Green, Ganzell, Satz, & Vaclav, 1990; Green, Satz, Ganzell, & Vaclav, 1992). In fact, computerized and experimenter-administered tasks appear to differ in subtle ways, despite their apparent equivalence. For example, Battig, Rosvold, and Mishkin (1960) found that monkeys with lesions in the anterior frontal cortex and the head of the caudate nucleus performed worse than unoperated monkeys on a traditional delayed alternation task. These monkeys, however, were indistinguishable from the controls on an automated version of the task. It is possible that the human examiner provides additional, subtle cues to the subjects. In the WCST for example, facial expressions or gestures may supplement the examiner's verbal feedback. We know that children with autism are less skillful at interpreting nonverbal cues such as facial expressions than normal children (Braverman, Fein, Lucci, & Waterhouse, 1989; Davies, Bishop, Manstead, & Tantam, 1994; Fein, Lucci, Braverman, & Waterhouse, 1992). It is possible, therefore, that this difficulty in recognizing/understanding nonverbal behaviors accounts for the different results we obtained with the WCST and the Computerized Matching Test. A second, not mutually exclusive, explanation is that children with autism are less susceptible to social reinforcement and feedback and that motivational factors are primarily responsible for their deficits. This deficit would be particularly pronounced on tasks they find difficult or complex. To investigate the relative contributions of these factors further, it is necessary to compare performance on the Computerized Matching Test with that on a hand-administered version as well as add primary reinforcers to the standard WCST Our findings, however, do not support a general deficit in shifting attention. If the deficit represented a core inability to shift, then we would expect to see this deficit across tasks, regardless of the specific requirements. In contrast, if children with autism perform adequately in some shifting tasks, then it is specific aspects associated with shifting that are particularly impaired. We know from Courchesne et al.'s studies (1994a, 1994b, 1995) that children with autism Attention in Children with Autism can shift their focus of attention when given additional time. Therefore, their deficit is not in shifting attention per se, but may be secondary to difficulties in the coordination and modulation of attentional resources, as well as in the activating effects of motivation. REFERENCES American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (3rd ed., Rev.). Washington, DC: Author. Battig, K., Rosvold, H. E., & Mishkin, M. (1960). Comparison of the effects of frontal and caudate lesions on delayed response and alternation in monkeys. Journal of Comparative and Physiological Psychology, 53, 400-404. Bellack, A. S., Mueser, K, T, Morrison, R. L., Tierney, A., & Podell, K. (1990). Remediation of cognitive deficits in schizophrenia. American Journal of Psychiatry, 147, 1650-1655. Braverman, M., Fein, D., Lucci, D., & Waterhouse, L. (1989). Affect comprehension in children with pervasive developmental disorders. Journal of Autism and Developmental Disorders, 19, 301-316. Burack, J. A. (1994). Selective attention deficits in persons with autism: Preliminary evidence of an inefficient attentional lens. Journal of Abnormal Psychology, 103, 53 5-543. Casey, B. J., Gordon, C. T, Mannheim, G. B., & Rumsey, J. M. (1993). Dysfunctional attention in autistic savants. Journal of Clinical and Experimental Neuropsychology, 15, 933-946. Courchesne, E. (1995). Infantile autism. Part 1: MR imaging abnormalities and their neurobehavioral correlates. International Pediatrics, 10, 141-154. Courchesne, E., Akshoomoff, N. A., Tbwnsend, J., & Saitoh, O. (1995). A model system for the study of attention and the cerebellum: Infantile autism. Electroencephalography and Clinical Neumphysiology Supplement, 44, 315-325. Courchesne, E., Townsend, J., Akshoomoff, N. A., Saitoh, O., Yeung-Courchesne, R., Lincoln, A. J., James, H. E., Haas, R. H., Schreibman, L, & Lau, L. (1994a). Impairment in shifting attention in autistic and cerebellar patients. Behavioral Neuroscience, 108, 848-65. Courchesne, E., Townsend, J., Akshoomoff, N. A., YeungCourchesne, R., Press, G. A., Murakami, J. W, Lincoln, A. J., James, H. E., Saitoh, O., Eggas, B., Haas, R. H., & Schreibman, L. (1994b). A new finding: impairment in shifting attention in autistic and cerebellar patients. In S. H. Broman & J. Grafinan (Eds.), Atypical cognitive deficits in developmental disorders: Implications for brain function (pp. 101-137). Hillsdale, NJ: Erlbaum. Davies, S., Bishop, D., Manstead, A. S., & Tantam, D. (1994). Face perception in children with autism and Asperger's syndrome. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 35, 1033-1057. Dawson, G., & Lewy, A. (1989). Arousal, attention, and the socioemotional impairments of individuals with autism. In G. Dawson (Ed.), Autism: Nature, diagnosis, and treatment (pp. 49-74). New York: Guilford. Fein, D., Lucci, D., Braverman, M., & Waterhouse, L. (1992). Comprehension of affect in context in children with pervasive developmental disorders. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 33, 1157-1167. Garretson, H. B., Fein, D., & Waterhouse, L. (1990). Sustained attention in children with autism. Journal of Autism and Developmental Disorders, 20, 101-114. 477 Green, M. F., Ganzell, S., Satz, P, & Vaclav, J. F (1990). Teaching the Wisconsin Card Sorting Test to schizophrenic patients (letter). Archives of General Psychiatry, 47, 91-92. Green, M. E, Satz, R, Ganzell, S., & Vaclav, J. E (1992). Wisconsin Card Sorting Test performance in schizophrenia: Remediation of a stubborn deficit. American Journal of Psychiatry, 149, 6267. Halperin, J. M., Wolf, L. E., Pascualvaca, D. M., Newcorn, J. H., Healey, J. M., O'Brien, J. D., Morganstein, A., & Young, J. G. (1988). Differential assessment of attention and impulsivity in children. Journal of the American Academy of Child and Adolescent Psychiatry, 27, 326-329. Halperin, J. M., Sharma, V, Greenblatt, E., & Schwartz, S. T (1991). Assessment of the continuous performance test: Reliability and validity in a nonreferred sample. Psychological Assessment 3, 603-608. Heaton, R. K. (1981). The Wisconsin Card Sorting Test. Odessa: Psychological Assessment Resources. Hutt, S. J., Hutt, C., Lee, D., & Ounsted, C. (1964). Arousal and childhood autism. Nature, 204, 908-909. Kanner, L. (1943). Autistic disturbances of affective contact. Nervous Child, 2, 217-250. Kanner, L. (1944). Early infantile autism. Journal of Pediatrics, 25, 211-217. Lifshitz, M., Kugelmass, S., & Karov, M. (1985). Perceptual-motor and memory performance of high-risk children. Schizophrenia Bulletin, 11, 74-84. Lincoln, A. J., Dickstein, R, Courchesne, E., Elmasian, R., & Tallal, P. (1992). Auditory processing abilities in non-retarded adolescents and young adults with developmental receptive language disorder and autism. Brain and Language, 43, 613622. Lovaas, O. (1977). The autistic child: Language development through behavior modification. New York: Irvington. Lovaas, O. (1993). The development of a treatment-research project for developmentally disabled and autistic children. Journal of Applied Behavioral Analysis, 26, 617-630. Lovaas, O., Koegel, R., & Schreibman, L. (1979). Stimulus overselectivity in autism: A review of research. Psychological Bulletin, 86, 1236-1254. Mirsky, A. F. (1987). Behavioral and psychophysiological markers of disordered attention. Environmental Health Perspectives, 74, 191-199. Mirsky, A. F., Anthony, B. J., Duncan, C. C., Ahearn, M. B., & Kellam, S. G. (1991). Analysis of the elements of attention: A neuropsychological approach. Neuropsychology Review, 2, 109-145. Mirsky, A. E, Fantie, B. D., & Tatman, J. E. (1995). Assessment of attention across the lifespan. In R. L. Mapou & J. Specter (Eds.), Clinical neuropsychological assessment: A cognitive approach (pp. 17-48). New York: Plenum Press. Ozonoff, S. (1995). Reliability and validity of the Wisconsin Card Sorting Test in studies of autism. Neuropsychology, 9, 491-500. Ozonoff, S., & McEvoy, R. E. (1994). A longitudinal study of executive function and theory of mind development in autism. Development and Psychopathology, 6, 415-431. Ozonoff, S., Pennington, B. F, & Rogers, S. J. (1991). Executive function deficits in high-functioning autistic individuals: Relationship to theory of mind. Journal of Child Psychology, Psychiatry, and Allied Disciplines, 32, 1081-1105. Prior, M., & Hoffmann, W (1990). Brief report: Neuropsychological testing of autistic children through an exploration with frontal lobe tests. Journal of Autism and Developmental Disorders, 20, 581-590. Rimland, B. (1964). Infantile autism. New York: Appleton-Century-Crofts. 478 Rosvold, H., Mirsky, A. F., Saranson, I., Bransome, E. D., & Beck, L. H. (1956). A continuous performance test of brain damage. Journal of Consulting Psychology, 20, 343-350. Rumsey, J. M. (1985). Conceptual problem-solving in highly verbal, nonretarded autistic men. Journal of Autism and Developmental Disorders, 15, 23-36. Rumsey, J. M., & Hamburger, S. D. (1990). Neuropsychological divergence of high-level autism and severe dyslexia. Journal of Autism and Developmental Disorders, 20, 159-186. Pascualvaca, Fantie, Papageorgiou, and Mirsky Schopler, E., Reichler, R. J., & Rochen-Renner, B. R. (1988). The Childhood Autism Rating Scale. Los Angeles: Western Psychological Services. Wechsler, D. (1991). Wechsler Intelligence Scale for Children (3rd ed.). New York: Psychological Corp. Woodcock, R. W, & Johnson, M. B. (1990). Woodcock-Johnson Psycho-Educational Battery—Revised. Texas: DLM Teaching Resources.