http://asm.sagepub.com/Assessment

http://asm.sagepub.com/content/18/2/192The online version of this article can be found at:

 DOI: 10.1177/1073191110394771

2011 18: 192 originally published online 10 January 2011AssessmentJames Holdnack, Gerald Goldstein and Lisa Drozdick

Syndrome and AutismSocial Perception and WAIS-IV Performance in Adolescents and Adults Diagnosed With Asperger's

  

Published by:

http://www.sagepublications.com

can be found at:AssessmentAdditional services and information for    

  http://asm.sagepub.com/cgi/alertsEmail Alerts:

 

http://asm.sagepub.com/subscriptionsSubscriptions:  

http://www.sagepub.com/journalsReprints.navReprints:  

http://www.sagepub.com/journalsPermissions.navPermissions:  

http://asm.sagepub.com/content/18/2/192.refs.htmlCitations:  

What is This? 

- Jan 10, 2011 OnlineFirst Version of Record 

- May 11, 2011Version of Record >>

at UNIV OF BRAZIL on May 9, 2014asm.sagepub.comDownloaded from at UNIV OF BRAZIL on May 9, 2014asm.sagepub.comDownloaded from

Assessment18(2) 192 –200© The Author(s) 2011Reprints and permission: http://www.sagepub.com/journalsPermissions.navDOI: 10.1177/1073191110394771http://asm.sagepub.com

Disorders in the autism spectrum, including autism and Asperger’s syndrome (AS), are characterized by deficits in social interaction, but autism is more strongly associated with cognitive deficits, varying in severity, than is Asperger’s disorder. In addition, autism is associated with impairment of verbal and nonverbal communication, stereotyped movements, repetitive activities, and resistance to change (American Psychiatric Association [APA], 2000). Even individuals with high-functioning autism (HFA), defined as having an intelligence quotient (IQ) of 70 or higher (Rutter & Schopler, 1987), typically demonstrate cognitive deficits. Neuropsy-chological studies of children and adults with autism have included both an experimental cognitive and a psychomet-ric literature. The psychometric literature largely involving the various Wechsler scales has provided substantial evidence of a characteristic profile marked by prominent variability in cognitive skills with some impaired and some preserved abilities. On the Wechsler intelligence scales, both children and adults with HFA show better perceptual reasoning versus verbal skills, have higher scores on Digit Span and Block Design compared with Comprehension (Goldstein & Saklofske, 2010), and have an overall strength on the Block Design subtest (Gilchrist et al., 2001). Individuals with HFA typically perform worse on Comprehension than on other verbal subtests from the Wechsler Adult Intelligence Scale–Third Edition (WAIS-III) or Wechsler Intelligence

Scale for Children–Third Edition (WISC-III; Mayes & Cal-houn, 2003). Low scores are also found for Picture Comple-tion and Coding subtests (Goldstein & Saklofske, 2010). Their cognitive skills have been described as strengths in short-term auditory memory, visual analysis, and visuomo-tor integration with deficits in verbal reasoning, concept formation, inferential reasoning, and representational capac-ity (Lincoln, Courchesne, Kilman, Elmasian, & Allen, 1988). Comparisons of intellectual function between individuals with HFA and AS have generally demonstrated higher abil-ity levels in AS (Emerich, Creaghead, Grether, Murray, & Grasha, 2003; Ghaziuddin & Mountain-Kimchi, 2004; Hayashi, Kato, Igarashi, & Kashima, 2008; Rinehart, Bradshaw, Tonge, Bereton, & Bellgrove, 2002; Spek, Scholte, & van Berckelaer-Onnes, 2008) with some preliminary evi-dence of possibly associated differences in brain structure (Jou, Minshew, Keshavan, & Hardan, 2010). However, the evidence is not robust and is sometimes equivocal, creating controversy concerning whether AS is a distinct syndrome separate from HFA (Macintosh & Dissanayake, 2004). This

394771 ASM18210.1177/1073191110394771Holdnack et al.Assessment© The Author(s) 2011

Reprints and permission: http://www.sagepub.com/journalsPermissions.nav

1Pearson Inc., San Antonio, TX, USA2VA Pittsburgh Healthcare Center, Pittsburgh, PA, USA

Corresponding Author:James Holdnack, 5 Rose Hill Drive, Bear, DE 19701, USA Email: james.holdnack@pearson.com

Social Perception and WAIS-IV Performance in Adolescents and Adults Diagnosed With Asperger’s Syndrome and Autism

James Holdnack1, Gerald Goldstein2, and Lisa Drozdick1

Abstract

Previous research using the Wechsler scales has identified areas of cognitive weaknesses in children, adolescents, and adults diagnosed with Autism or Asperger’s syndrome. The current study evaluates cognitive functioning in adolescents and adults diagnosed with Autism or Asperger’s syndrome using the Wechsler Adult Intelligence Scale–Fourth Edition (WAIS-IV) and the Social Perception subtest from the Advanced Clinical Solutions. Deficits in social perception, verbal comprehension, and processing speed were found in the Autism sample. Additionally, they exhibited inconsistent performance on auditory working memory and perceptual reasoning tasks. The Asperger’s syndrome group had better overall cognitive skills than the Autism group, but compared with controls, they had weaknesses in processing speed, social perception, and components of auditory working memory. Both groups had relatively low scores on the WAIS-IV Comprehension subtest compared with the other verbal comprehension subtests. Clinical application and utility of the WAIS-IV and Social Perception in Autism Spectrum Disorders are discussed.

Keywords

autism, Asperger’s syndrome, WAIS-IV, social perception, ACS

at UNIV OF BRAZIL on May 9, 2014asm.sagepub.comDownloaded from

Holdnack et al. 193

controversy has become an active source of discussion reg-arding the Autism Spectrum Disorders to appear in Diag-nostic and Statistical Manual of Mental Disorders–Fifth Edition (DSM-V; APA, 2010). The introduction of a new version of the WAIS (WAIS-IV) containing new norms and some new subtests raises the question of whether the find-ings of res earch with the previous versions apply to these new procedures. Of particular concern is the research related to cognitive differences between individuals with HFA and AS. New norms and new subtests may well alter conclusions based on the current literature, particularly because that lit-erature is equivocal.

In addition to intellectual function, individuals with autism spectrum disorders show deficits in the development of interpretation of affect, including face recognition and other aspects of emotional recognition (Golarai, Grill-Spector, & Reiss, 2006; Korkman, Kirk, & Kemp, 2007). Adolescents and adults with autism have difficulties accurately identifying fearful, disgusted, happy (Humphreys, Minshew, Leonard, & Behrmann, 2007), and sad (Boraston, Blakemore, Chilvers, & Skuse, 2007) facial expressions. Additionally, they do not show normal brain activation in the amygdala and orbito-frontal regions in response to fearful faces but activate the cingulate and superior temporal lobes instead (Ashwin, Baron-Cohen, Wheelwright, O’Riordan, & Bullmore, 2007). Impaired processing of facial expressions of emotions is not attributable to general impairments in visual–perceptual processing (Humphreys et al., 2007). Deficits in emotion have also been reported for determining emotions from pros-ody (Hollander et al., 2007).

The Advanced Clinical Solutions (ACS) for the WAIS-IV and the Wechsler Memory Scale–Fourth Edition (WMS-IV; Pearson, 2009) contains a Social Perception subtest. It mea-sures the examinee’s ability to recognize and label facial expressions of emotion, to identify emotion from prosody with facial expressions, and to match prosody with pairs of interacting people. The ACS and the WAIS-IV are co-normed, therefore, we now have the advantage of com-paring general intellectual function and affect cognition with the same instrument without the necessity of making infer-ences from the traditional Wechsler subtests that only indi-rectly assess social cognition. The current study compares the cognitive profiles in adolescents and adults with HFA or AS compared with matched controls using the new WAIS-IV and Social Perception subtest. Comparisons are directed toward testing hypotheses regarding whether specific defi-cits or general level of performance differentiate the two clinical samples. The specific cognitive measures used in the study are the WAIS-IV core subtest and index scores and the Social Perception subtest scaled scores.

It is hypothesized that both the HFA and AS groups will show deficits in processing speed, but the HFA group will also have low scores on all verbal tests compared with a

matched control sample, with the lowest score obtained on Comprehension, whereas the AS group will have normal language functions with slightly lower scores on Compre-hension. Both groups will differ from controls on measures of social perception, and the HFA sample will be more impaired than the AS sample. WAIS-IV scores are expected to reliably differentiate HFA but not AS cases from the gen-eral population using multiple classification methods.

MethodParticipants

Diagnostic procedures. Diagnoses were made by qualified professionals in accordance with procedure established for acquisition of the WAIS-IV and WMS-IV normative and clinical procedures. In the case of diagnosis of mental disor-ders, diagnoses had to be made by qualified psychiatrists and psychologists who were asked to submit information for cases they examined using appropriate assessment instru-ments and were found to meet DSM-IV-TR diagnostic pro-cedures. In the case of autistic disorder, there had to be assurance that there was a delay or abnormal functioning in language, social interaction, or play with onset before the age of 3 years. The diagnostic criteria for AS have a good deal of overlap, but there had to be assurances that there was no significant language delay by the age of 3 years or cognitive delay in childhood. Since the data were collected from across the nation, it was not possible to use specified assessment instruments such as the Autism Diagnostic Interview (ADI) or Autism Diagnostic Observation Sched-ule (ADOS), but assurances were obtained that a sufficient assessment was made, including interviews with parents and observation of patients to determine that DSM-IV-TR criteria were met.

High-Functioning AutismA sample of 16 examinees, aged 16 to 34 years (mean = 21.7 ± 6.4 years), who were diagnosed with autistic disor-der according to DSM-IV-TR criteria were assessed with the WAIS-IV and Social Perception from ACS. The sample was collected as part of the standardization of the WAIS-IV/WMS-IV/ACS. Individuals diagnosed with autistic disorder were excluded from this study if they obtained a WAIS-IV General Ability Index (GAI) score lower than 60 (75% having ≥ 70). The demographic characteristics of the sample were as follows: sex—87.5% male, 12.5% female; ethnicity—75% Caucasian, 18.8% African American, 6.2% other; and parent education level—37.5% 12 or more years, 25.0% 13 to 15 years, 37.5% 16 or more years. This over-representation of males in the sample most likely reflects the combined effects of a higher prevalence rate for autistic

at UNIV OF BRAZIL on May 9, 2014asm.sagepub.comDownloaded from

194 Assessment 18(2)

disorder in males than in females (four times higher) and the greater prevalence of intellectual disability in females with autistic disorder (APA, 2000).

Asperger’s SyndromeA sample of 27 examinees, aged 16 to 40 years (mean = 22.1 ± 7.7 years), also drawn from the WAIS-IV/WMS-IV standardization study who were diagnosed with AS accord-ing to DSM-IV-TR criteria (i.e., had no history of language or cognitive delays) were assessed with the WAIS-IV and Social Perception from the ACS. Individuals diagnosed with AS were excluded from this study if they obtained a WAIS-IV GAI score lower than 65 (89% having GAI ≥ 80). Although there has been some discussion of diagnosing AS without a family history (Bennett et al., 2008), we elected to adhere to use of DSM-IV-TR criteria because it contains the most widely used diagnostic system. The demographic characteristics of the sample were as fol-lows: sex—77.8% male, 22.2% female; ethnicity—81.5% Caucasian, 3.7% African American, 7.4% Hispanic American, 7.4% other; and education level—3.7% 8 years or less, 33.3% 12 or more years, 25.9% 13 to 15 years, 37.5% 16 or more years.

Healthy ControlsParticipants for this study included 600 healthy people, between 16 and 69 years of age, from the ACS Social Per-ception standardization sample. The sample was based on and is consistent with the 2005 U.S. Census results. For a complete list of exclusion criteria for the standardization sample, the reader is referred to the ACS: Clinical and Interpretive Manual (Pearson, 2009). A subsample of nor-mal controls, consisting of 43 examinees, was selected to form a comparison group matched for age, ethnicity, and education with the HFA and AS groups (Table 1).

MeasuresThe WAIS-IV (Wechsler, 2008) is a recently revised bat-tery of tests designed to evaluate intellectual abilities. It is composed of 10 core subtests (Vocabulary, Information, Similarities, Digit Span, Arithmetic, Block Design, Matrix Reasoning, Visual Puzzles, Coding, and Symbol Search) and five optional subtests (Comprehension, Letter–Number Sequencing, Figure Weights, Picture Completion, and Cancellation). For purposes of this study, Digit Span was divided into Digits Forward, Digits Backward, and Digit Sequencing components. The primary subtests yield four Index scores (Verbal Comprehension [VCI], Perceptual Rea soning [PRI], Working Memory [WMI], and Processing Speed [PSI]) and two overall indicators of cognitive functioning:

GAI and Full-Scale IQ (FSIQ). The age-adjusted scaled and standard scores for each of the subtest and index scores were used in the between-group analyses.

The Social Perception subtest is a screening tool used to identify deficits in social perception. It contains three tasks: Affect Naming (affect labeling), Prosody–Face Matching (linking prosody to a facial expression), and Prosody–Pair Matching (linking prosody to an interacting pair of people and explaining the intent of the speaker). The Social Perce-ption subtest requires integration of different modalities of emotion (i.e., auditory and visual), dissociation of tone of voice from content (i.e., congruent and incongruent con-tent), and dissociation of sex of a recorded voice from visual expression.

In the Affect Naming task, the examinee sees pictures of faces expressing emotions (i.e., happy, sad, angry, fear, disgust, surprise, and neutral) and names the emotion being expressed. In the Prosody–Face Matching task, the examinee hears an audio recording of a voice saying a single state-ment such as “He looks good today.” Then, the examinee views a page with six faces displaying different facial exp-ressions. The facial expressions include the seven emotions used in the Affect Naming task with the addition of sarcasm and confusion. The examinee selects the face that matches the emotional tone of the speaker (e.g., sarcastic expression if the voice sounded sarcastic). In the Prosody–Pair Matching task, the examinee listens to a recorded statement and must

Table 1. Demographic Characteristics of the Clinical and Control Samples

HFA AS Control

Age (years) 22.2 ± 6.4 22.6 ± 7.6 22.6 ± 7.1Sex (%) Male 87.5 77.8 81.4 Female 12.5 22.2 18.6Years of educationa (%) ≤8 0.0 3.7 0.0 9-11 0.0 0.0 2.3 12 37.5 33.3 37.2 13-15 25.0 25.9 30.2 ≥16 37.5 37.0 30.2Ethnicity (%) White 75.0 81.5 79.0 Black 18.7 3.7 9.3 Hispanic 0.0 7.4 4.7 Other 6.3 7.4 7.0

Note. The clinical samples did not differ significantly from their matched controls on any of the background variables. HFA = high-functioning autism; AS = Asperger’s syndrome.a. All subjects ages 16 to 19 years, reported education level is the average of mother and father’s education level, for Autism sample all reported education is the average of parents’ education level.

at UNIV OF BRAZIL on May 9, 2014asm.sagepub.comDownloaded from

Holdnack et al. 195

use the prosody from the audio to select the picture that best matches the expressed statement from a page with four pic-tures. In each picture, different facial expressions and body positions are demonstrated by two people interacting. To get the correct answer, the examinee must look at both faces in each pair to determine if the prosody makes sense in that interaction. After the examinee points to the picture, he or she is asked to identify the emotion being expressed by the voice and whether the tone of voice changed the meaning of what was said. If the tone of voice changed the meaning, the examinee is asked to indicate the real intent of the speaker.

Social Perception yields four scores: Affect Naming, Prosody, Pairs, and a total Social Perception score. The Affect Naming score is based on total correct responses to the Affect Naming task. The Prosody score is based on the number of pictures correctly selected in both the Prosody–Face Matching and Prosody–Pair Matching tasks. The Pairs score is com-puted from the selection of the pictures in the Prosody–Pair Matching task plus the scores on the verbal responses for Prosody–Pair Matching. The Social Perception total score is calculated by summing the number of correctly labeled faces from Affect Naming and correctly selected pictures from Prosody–Face Matching and Prosody–Pair Matching. The age-adjusted scaled scores for each of these total scores were used in the data analyses.

Data AnalysisAnalyses were performed using SPSS® version 14.0. Multivariate analysis of variance (MANOVA) and analysis of variance (ANOVA) were used to compare the clinical samples with the matched control group. ANOVAs were used for higher order measures (FSIQ and GAI. At the Index level, MANOVAs were computed for the four index scores. Accuracy of classification was evaluated with a stepwise discriminant function analysis that provided infor-mation about the contribution of the individual measures to classification and overall rate of accuracy. Cohen’s kappa (Cohen, 1988) was used to assess the extent of agreement between actual group membership and membership predicted by the discriminant function.

ResultsTable 2 presents the means and standard deviations for the various WAIS-IV scores, including general and factor score indices and subtests by group. The results comparing the HFA, AS, and the control group indicate a significant differ-ence between groups on FSIQ, F(2, 80) = 18. 63, p < .001, η2 = .32. A Scheffe test indicated that the AS group and the control group did not differ from each other, with both scoring higher than the HFA group. In the same manner, the groups also differed significantly on the GAI, F(2, 81) = 13.82, p < .001, η2 = .25. On the index scores, the MANOVA was

significant, confirming poorer cognitive function in the HFA group across domains, Wilks’s Λ = .53, F(4, 77) = 7.28, p < .001, η2 = .99. Greater effect sizes (both of which would be interpreted as clearly large) were observed for VCI—F(1, 27) = 14.06, p < .001, η2 = .26 and PSI—F(1, 27) = 23.54, p = .001, η2 = .37, than for PRI—F(1, 27) = 7.71, p < .001, η2 = .16 and WMI—F(1, 27) = 8.05, p < .05, η2 = .17—whose effect sizes were closer to the medium range. Scheffe tests indicated that in the cases of VCI, PRI, and WMI, the AS and control group did not differ from each other; however, both performed better than the HFA group. For PSI, all three groups were different from each other with the poorest performance in the HFA group and the best performance in the control group.

These findings are supported in the subtest level analysis where the multivariate analysis was statistically significant—Wilks’s Λ = .319, F(32, 130) = 2.87, p < .001, η2 = .44—with a large effect size. Analysis of the individual subtests

Table 2. Mean WAIS-IV Subtest and Index Scores by Clinical and Control Samples

HFA AS Control

VC 5.8 (2.0) 10.4 (3.1) 10.7 (2.7)SI 6.6 (2.9) 11.3 (3.5) 10.4 (3.0)IN 6.6 (2.9) 9.7 (2.7) 11.2 (2.7)CO 5.0 (2.1) 9.1 (2.8) 10.9 (3.1)BD 7.2 (2.40 9.7 (3.2) 10.7 (3.2)MR 8.8 (3.6) 10.0 (2.7) 10.5 (2.9)VP 7.2 (3.3) 9.6 (3.0) 10.8 (3.3)AR 6.3 (7.1) 9.3 (3.5) 11.0 (3.2)DSF 8.0 (3.5) 9.1 (3.6) 10.3 (3.0)DSB 9.0 (3.0) 9.6 (2.8) 10.5 (3.4)DSS 6.4 (3.6) 9.2 (3.0) 10.3 (3.3)DS 7.1 (3.5) 9.2 (3.4) 10.3 (3.3)SS 5.4 (2.0) 8.0 (2.8) 9.7 (2.8)CD 4.9 (2.1) 7.2 (2.5) 9.3 (2.3)VCI 79.4 (11.4) 102.4 (15.8) 103.8 (14.2)PRI 86.6 (15.4) 98.2 (13.9) 103.6 (16.1)WMI 81.5 (15.3) 95.8 (17.3) 103.4 (16.9)PSI 74.0 (9.7) 87.2 (12.1) 96.9 (11.8)FSIQ 76.3 (12.2) 95.8 (13.7) 102.6 (14.0)GAI 80.0 (12.7) 100.4 (14.8) 104.1 (15.8)

Note. WAIS-IV = Wechsler Adult Intelligence Scale–Fourth Edition; HFA = high-functioning autism; AS = Asperger’s syndrome; VC = Vocabulary; SI = Similarities; IN = Information; CO = Comprehension; BD = Block Design; MR = Matrix Reasoning; VP = Visual Puzzles; AR = Arithmetic; DSF = Digit Span Forward; DSB = Digit Span Backward; DSS = Digit Span Sequencing; DS = Digit Span; SS = Symbol Span; CD = Coding; VCI = Verbal Comprehension; PRI = Perceptual Reasoning; WMI = Working Memory Index; PSI = Processing Speed Index; FSIQ = Full-Scale Intelligence Quotient; GAI = General Ability Index. Standardization data from the Wechsler Adult Intelligence Scale–Fourth Edition (WAIS-IV) copyright © 2008 NCS Pearson, Inc. Used with permission. All rights reserved.

at UNIV OF BRAZIL on May 9, 2014asm.sagepub.comDownloaded from

196 Assessment 18(2)

yielded significant differences among the groups with the exception of Matrix Reasoning (p = .09) and Digits Forward (p = .11). Scheffe tests indicated that in most cases, the AS group and control group did not differ from each other, and both groups performed significantly better than the HFA group. The exceptions were Matrix Reasoning and Digits Forward on which no significant differences were observed, Coding where all three groups differed significantly from each other, and Visual Puzzles where the HFA group per-formed significantly more poorly than the control group, but the AS group did not differ from either the HFA or the control group.

The means and standard deviations for the Social Perce-ption subtest measures are presented in Table 3 by group. Differences were significant at the multivariate level, Wilks’s Λ = .60, F(6, 158) = 7.81, p < .001. The HFA sample per-formed significantly worse than controls at the multivariate level—Wilks’s Λ = .427, F(4, 26) = 8.71, p < .001, η2 = .23—with a large effect size. At the univariate level, Affect Naming was not statistically significant, but performance on Prosody and Pairs differed among the groups. Scheffe tests indicated that on both Prosody and Pairs, the HFA group did significantly worse than the AS and control groups, which did not differ from each other. Controlling for general intel-lectual function through performing a MANCOVA with FSIQ as the covariate did not substantially alter the results, Wilks’s Λ = .650, F(6, 98) = 3.92, p < .001, η2 = .19.

Classification of the HFA, AS, and Control Groups by Discriminant FunctionIndividual scores were combined to determine if the differ-ing cognitive profiles among the three groups could be used to classify them with acceptable levels of sensitivity and spe-cificity. The SPSS Discriminant program was used for that purpose, taking the stepwise option following application of the direct method to determine overall significance. The stepwise method provides level of classificatory accuracy and ranks the variable combinations with regard to their pre-dictive values. Two such analyses were performed, one for

the WAIS-IV subtests and the other for the Social Perception Test measures. Standard default options were used, an F to enter of 3.54 and an F to remove of 2.71. Using these stan-dard default options, only Comprehension and Coding were entered into the prediction equation. Cohen’s kappa was applied to the classification matrices to evaluate strength of agreement between actual and predicted group member-ship. The classification matrix for the WAIS-IV subtests is presented in Table 4. There were 70.2% correct classifica-tions, with most of the misclassifications occurring in the AS group with approximately equally high numbers of mis-classifications in the HFA and control groups. Cohen’s kappa was equal to .52 (p < .001), which according to the Landis and Koch criteria would be rated as reflecting mod-erate agreement.

The classification matrix for the Social Perception Test is presented in Table 5. Pairs and Prosody were entered into the prediction equation. The classification matrix indicates that there were 56.1% correct classifications. Again, most of the misclassifications occurred in the AS group with sub-stantial misclassification into both of the other groups but with a slightly higher percentage in the HFA group. Cohen’s kappa was equal to .33 (p < .001), indicating fair agreement according to the Landis and Koch criteria. Thus, the classifi-cation analyses of both the WAIS-IV subtests and the Social Perception Test generated reasonably good classificatory accuracy for the HFA and control groups, but substantial classification error in the AS group, many members of which were misclassified as HFA or control cases.

Classification of the HFA, AS, and Control Groups by Number of Low ScoresAnother model for identifying atypical cognitive functioning is the application of multivariate base rates (Brooks, Iverson, Feldman, & Holdnack, 2009). Since it is normal for healthy controls to obtain one or more low scores on cognitive test-ing (Brooks, Iverson, Holdnack, & Feldman, 2008), it is exp ected that a matched-control sample would also have indi viduals performing at very low levels on cognitive tests of interest. Therefore, the difference between clinical sam-ples and controls may be associated with the number of low

Table 3. Mean Social Perception Scores by Clinical and Control Samples

HFA AS Control

SP Total 4.4 (2.7) 8.2 (3.0) 9.2 (2.9)SP Affect Naming 6.4 (4.1) 7.8 (2.6) 9.6 (3.1)SP Prosody 4.9 (3.6) 9.0 (3.2) 9.4 (2.9)SP Pairs 4.9 (3.4) 8.3 (3.2) 10.2 (2.3)

Note. HFA = high-functioning autism; AS = Asperger’s syndrome; SP = social perception subtest. Standardization data from the Advanced Clinical Solutions for the WAIS-IV/WMS-IV (ACS) copyright © 2008 NCS Pearson, Inc used with permission. All rights reserved.

Table 4. WAIS-IV Discriminant Analysis Classification Table

Predicted Group

Actual Group HFA AS Control

HFA 92.9% 0.0% 7.1%AS 25.9% 44.4% 29.6%Control 2.3% 18.6% 79.1%

Note. WAIS-IV = Wechsler Adult Intelligence Scale–Fourth Edition; HFA = high-functioning autism; AS = Asperger’s syndrome.

at UNIV OF BRAZIL on May 9, 2014asm.sagepub.comDownloaded from

Holdnack et al. 197

scores attained in cognitive domains of interest rather than based on those that show the largest differences between the groups. A low score on social perception may not be atypical, but having low scores on multiple indicators of social perception and other relevant cognitive variables might indicate atypicality.

As seen in the data in Tables 2 and 3, HFA and AS group performance varies across the subtest and index scores, and low performance is not uniformly low, particularly in the AS group. A subset of scores from the WAIS-IV and Social Perception tests, which are relatively lower in the clinical samples, although not necessarily statistically significant were used to evaluate the utility of using a multivariate cut score in classifying individuals. The scores used were VCI, Comprehension, PSI, Digit Span Sequencing, Arithmetic, Social Perception Total, Prosody, and Pairs. Cut scores of ≤85 for index scores and ≤7 for scaled scores were desig-nated as a low score (i.e., 1 standard deviation below the mean) on a specific variable. For index scores, a standard score of 85 or lower was assigned a value of 1 point, whereas scores above 85 were assigned a value of 0. Similarly, sub-test scores were assigned a value of 1 point if the scaled score was 7 or lower and 0 points if the scaled scores were 8 or higher. One point was also assigned when Matrix Rea-soning was greater than Comprehension by more than 3 points; otherwise, 0 points were assigned for differences of less than 4 points or if Comprehension exceeded Matrix Reasoning. The total number of possible points was 9.

Table 4 presents the average number of low scores obtained on the WAIS-IV and Social Perception scores by the HFA, AS, matched control, and standardization sam-ples. The HFA group had significantly more scores below the cutoff score than both the matched-control sample and the standardization sample, F(2, 584) = 50.39, p < .001. On average, the HFA sample had almost 7 scores below the cutoff score compared with roughly 1.5 for the matched control and standardization samples. Large effect sizes were obtained for both comparisons. Using a multivariate cutoff of 5 low scores, 91.7% of the HFA sample was correctly identified (sensi-tivity), whereas 90.7% of the matched controls (specificity) and 90.5% (specificity) of the standardization sample were correctly identified. The classification rates of the HFA versus

the matched-control sample are statistically significant— χ2(1) = 32.08, p < .001—with a kappa agreement rate of .76 (p < .001).

The variables and cut scores selected for this analysis were designed to identify individuals with autism versus controls and the general population. These scores were not intended to identify individuals with AS. The AS group had an average of 2 low scores, which was similar to the control samples. In the AS group, 23.1% were identified as having an atypical profile. When the AUT (autism group), AS, and control group were considered together, the classification was significant, χ2(2) = 33.03, p < .001. To identify more cases of AS having an atypical cognitive profile, a different set of variables and cut scores were analyzed.

A second analysis was conducted to identify individuals in the AS sample. The following variables were applied: PSI, Comprehension, Arithmetic, Digit Span Forward, Affect Naming, and Pairs. The maximum number of low scores in this evaluation was 6. Given that the AS sample does not show the degree of deficits found in the HFA sample, the cut-offs used for identifying low scores based on the AS weaknesses were higher. For index scores, the cut-off used was 90 rather than 85, and for subtest scaled scores, a cut-off of eight rather than seven was used. Table 6 presents the mean scores for the AS, matched-control, and standardiza-tion samples. The average rate of low scores in the AS sam-ple was 3.5 scores, whereas the control and standardization samples had 2 low scores on average. The control groups had significantly fewer low scores than the clinical sample, F(2, 598) = 32.58, p < .001. Using a cut-off of 4 or more low scores, 42.3% of the AS sample was correctly identified; whereas 95.3% of the matched controls and 91.1% of the standardization sample were correctly identified. Applying a cut-off of 3 low scores increases identification of the clini-cal sample to 57.7%, but also increases the false-positive rate from 8.9% to 30.8%. The classification rate for a cut-off of 4 low scores was statistically significant—χ2 (1) = 15.02, p < .001—with a kappa of .42 (p < .001).

DiscussionThe participants with HFA clearly displayed lower general cognitive function as measured by the WAIS-IV FSIQ and GAI measures, than normal controls, but that was not the case for the AS group. Although all the cognitive domains in the HFA group were significantly lower than controls, the largest effects were observed for Verbal Comprehension and Processing Speed. Among Verbal Comprehension measures, the Comprehension subtest was significantly lower in the HFA sample with a η2 of .34, a large effect size. The percep-tual reasoning and working memory domains showed more variability in the abilities affected in HFA. In the perceptual reasoning domain, Matrix Reasoning was not significantly different among the three groups; whereas Block Design

Table 5. Social Perception Discriminant Analysis Classification Table

Predicted Group

Actual Group HFA AS Control

HFA 66.7% 26.7% 6.7%AS 26.9% 34.6% 38.5%Control 7.0% 25.6% 67.4%

Note. HFA = high-functioning autism; AS = Asperger’s syndrome.

at UNIV OF BRAZIL on May 9, 2014asm.sagepub.comDownloaded from

198 Assessment 18(2)

and Visual Puzzles were areas of relative weakness for this group. In multiple comparisons, the most common pattern was the absence of a significant difference between the AS and control samples, with both doing better than the HFA sample. Significant differences among the three groups were not found for Digit Span Forward or Backward and Matrix Reasoning. Thus, individuals with HFA or AS and normal individuals were found to do equally well on certain atten-tional and visual perceptual tasks.

We did not find the particularly intact performance of individuals with HFA reported in other studies on Block Design, with performance on that subtest in line with per-formance on the other subtests. Perhaps that is because some studies had samples of participants with HFA that had sub-stantially higher intellectual function than was found here. In their review, Siegel, Minshew, and Goldstein (1996) reported Full-Scale WAIS mean IQs of as high as 104 in HFA samples. In their own data, the mean FSIQ was 96 (SD = 14.5), with a mean Block Design subtest score of 11.3 (SD = 3.72). These results are closer to what was found for the AS sample in the present study. However, the present findings are consistent with previous studies using the Wechsler intelligence scales, which found significant deficits on the Comprehension subtest in particular and tests of pro-cessing speed with variable results on perceptual reasoning and auditory working memory tasks (Goldstein & Saklofske, 2010). The primary difference between the results for the cur-rent study and previous studies is relative sparing of perfor-mance on Matrix Reasoning rather than on Block Design, and the presence of deficits on Digit Span Sequencing.

The cognitive profile of WAIS-IV scores in HFA indi-cates global deficits in language, particularly involving social judgment and in speed of information processing. It is noted

that the Comprehension subtest, unlike the other verbal sub-tests, requires extended speech production, which may also be a deficit area. In addition, slowed speed of visual pro-cessing does not appear to be limited to fine-motor control but is also related to inspection and comparison time and implicit learning, as assessed with the Cancellation and Coding subtests. Auditory working memory skills are intact for items that are simply repeated or repeated in reverse order; however, when information needed to be operated on in a series of steps and transformations, as assessed with Digit Sequencing, performance of the HFA group was substan-tially below that of healthy controls. In general, if one com-pares the WAIS-IV profile of the HFA group with that of the AS group, they are substantially different. The AS cogni-tive profile more closely resembles the healthy control group than it resembles the HFA sample in the present study.

Participants with AS did not show pronounced deficits in general cognitive functioning, only performing worse than controls on processing speed tasks, with mean scores that fell between those of the HFA and control groups on Coding and Cancellation. The cognitive findings therefore suggest a pronounced difference between individuals diagnosed with AS and those with autism, even HFA. The mean FSIQ and GAI scores were in the average range and differed only slightly from the equivalent scores for the control group. The factor index scores were in the low average to average range and did not differ substantially from the control group with the exception of the Processing Speed Index where the mean score of the AS group was substantially lower than that of the control group. These results are confirmatory of the existing literature that reports relatively average intelli-gence in AS.

On the social perception tests, the HFA group performed particularly poorly, but the pattern of scores was different among the three subscores. On Affect Naming, the three groups did not differ significantly from each other. On Prosody and Pairs, the HFA group did more poorly than the other two groups, which did not differ from each other. Thus, the HFA group demonstrated substantial impairment of social perception even when an adjustment for general inte-lligence was made; with the AS group showing only mild impairment that did not reach the level of statistical signifi-cance in comparison with the control group. These results indicate that individuals with HFA have difficulty integrat-ing prosody with facial expressions. Also, they have diffi-culty analyzing the body language and facial expressions of pairs of people interacting. Previous research has identi-fied basic deficits in face processing and affect recognition (Boraston et al., 2007; Golarai et al., 2006; Humphreys et al., 2007; Korkman et al., 2007), as well as understanding emotion from prosody (Hollander et al., 2007), affect label-ing, and interpreting pairs of people interacting. Individuals with AS demonstrated this difficulty on the Social Perception Test only to a limited extent, as indicated by a statistically

Table 6. Total Number of Scores Identified as Cognitive Weaknesses by Clinical, Control, and Standardization Samples

HFA Control Standardization AS

WAIS-IV/Social Perceptiona,c

6.9 (1.9) 1.5 (1.7) 1.5 (1.8) 2.8 (2.1)

AS Control Standardization HFA

WAIS-IV/Social Perceptionb,c

3.5 (1.7) 1.6 (1.2) 1.9 (1.6) 5.3 (1.2)

Note. Samples sizes vary because of missing data: WAIS-IV/Social Perception; Autism (12), Matched Controls (16), Standardization (559); WAIS-IV/Social Perception; Asperger’s (26), Matched Controls (27), Standardization (548); HFA = high-functioning autism; AS = Asperger’s syndrome.a. Scores based on weaknesses identified in the Autism sample.b. Score based on weaknesses identified in the Asperger’s syndrome sample.c. Clinical group is significantly different from control and standardization sample at p < .01 applying analysis of variance with Scheffe post hoc comparison.

at UNIV OF BRAZIL on May 9, 2014asm.sagepub.comDownloaded from

Holdnack et al. 199

significant difference on the multivariate analysis. They also performed substantially worse than the controls on the Affect Naming component of the Social Perception test, their mean performance level resembling that found for the HFA group. It would appear that social cognition deficits associated with AS are found primarily at less complex lev-els of identifying and labeling expressions of emotion.

The classification analyses are supportive of these find-ings. With regard to the WAIS-IV subtests, the classifica-tion matrix shows almost perfect classification of the HFA group with only one case misclassified. The control group also showed a reasonably high classification rate with 79% of the cases correctly classified. However, only 44% of the AS cases were correctly classified with approxi-mately equal numbers of classifications into the other groups. This finding suggests the presence of cognitive heterogeneity associated with AS, with some individuals performing like normal people, whereas others perform like individuals with HFA. The discriminant analysis of the Social Perception test yielded a lower classification rate, with only 56% of the cases correctly classified. Percentage of correct classification was about the same for all three groups. However, only 2 (12.5%) of the control cases were classified into the HFA group. The major source of error appears to be that almost half of the AS cases were classified into either the normal or HFA group, again suggesting heterogeneity.

Applying a multivariate base rate model yielded results similar to those obtained through discriminant analysis. The advantage of the multivariate base rate model is that it does not rely on a few indicators being significantly dif-ferent between clinical and control samples. Rather, it indicates whether or not a set of variables is atypical from the general population. It is well established that healthy controls obtained low scores on cognitive tests for a vari-ety of reasons; it may be that clinical samples are identified by a pervasiveness of mild to moderate cognitive weak-nesses rather than one or two severe cognitive limitations. In the current study, both models were able to readily identify atypical cognitive performance in the HFA group but only identified cognitive weaknesses in a subset of individuals with AS.

Cognitive profiles are rarely diagnostic of a specific con-dition (Siegel et al., 1996). Thus, we are not suggesting that the Wechsler scales and Social Perception tests should be used to diagnose autism spectrum disorders. Cognitive tests can provide a reasonable degree of accuracy in identifying nonnormal cognitive functioning rather than diagnosing a specific condition. Although having the test profiles reported here could indicate nonnormal status, this does not mean that the examinee has autism or AS as there were not a suf-ficient number of additional patient populations in the study to allow differential diagnosis (e.g., schizophrenia, traumatic brain injury, epilepsy, dementia). Further research is required

to determine if the specific set of variables identified here would be useful in differential diagnosis.

With regard to the construct validity of the diagnostic distinction between HFA and AS, the findings regarding heterogeneity in the AS group are of particular interest. Apparently, there are some individuals with AS who have a level and pattern of cognitive function involving general intelligence and social perception that resemble what is found in HFA, whereas others are indistinguishable from normal with the exception of reduced processing speed. The limitations of this study may limit the generalizeability of this and other findings. The small sample sizes, particularly in the HFA sample, significantly limit the statistical power to identify cognitive deficits. Neither the HFA nor the AS sample were randomly selected from the general population of individuals with these conditions, and therefore, results may not represent the cognitive functioning of these groups more broadly. The HFA sample also had relatively low general cognitive ability, and individuals with higher ability may not show the same pattern of deficits observed here. Finally, the diagnosis of the patients was based on the indi-vidual examiner or clinical sites and application of DSM-IV-TR criteria and consistency in diagnosis across sites was not controlled. Subsequently, some of the patients, particu-larly in the AS group, may have been misidentified. Further studies using larger, more well-defined, and more represen-tative samples of adolescents and adults with HFA and AS that include control samples that are more closely matched to the Autism Spectrum Disorders samples on relevant demo-graphic characteristics are needed to verify the findings in this study.

Acknowledgment

Dr. Goldstein acknowledges the Medical Research Service of the Department of Veterans Affairs and the MIRECC, VA Pittsburgh Healthcare System for support of this research.

Authors’ Note

Standardization data from the Wechsler Adult Intelligence Scale–Fourth Edition (WAIS-IV) copyright © 2008 NCS Pearson, Inc. and Advanced Clinical Solutions for the WAIS-IV and WMS-IV (ACS) copyright © 2009 NCS Pearson, Inc. Used with permis-sion. All rights reserved.

Declaration of Conflicting Interests

The authors declared a potential conflict of interest (e.g., a finan-cial relationship with the commercial organizations or products discussed in this article) as follows:

Drs. Holdnack and Drozdick are Research Directors with Pearson. Dr. Goldstein does not have a conflict of interest to declare.

Funding

The authors received no financial support for the research and/or authorship of this article.

at UNIV OF BRAZIL on May 9, 2014asm.sagepub.comDownloaded from

200 Assessment 18(2)

References

American Psychiatric Association. (2000). Diagnostic and sta-tistical manual of mental disorders (4th ed., text revision). Washington, DC: Author.

American Psychiatric Association. (2010). Diagnostic and statistical manual of mental disorders (5th ed.). Manuscript in preparation.

Ashwin, C., Baron-Cohen, S., Wheelwright, S., O’Riordan, M., & Bullmore, E. T. (2007). Differential activation of the amyg-dala and the “social brain” during fearful face-processing in Asperger syndrome. Neuropsychologia, 45, 2-14.

Bennett, T., Szatmari, P., Bryson, S., Volden, J., Zwaigenbaum, L., Vaccarella, L., . . . Boyle, M. (2008). Differentiating autism and Asperger syndrome on the basis of language delay or impair-ment. Journal of Autism and Developmental Disorders, 38, 616-625.

Boraston, Z., Blakemore, S. J., Chilvers, R., & Skuse, D. (2007). Impaired sadness recognition is linked to social interaction deficit in autism. Neuropsychologia, 45, 1501-1510.

Brooks, B. L., Iverson, G., Feldman, H., & Holdnack, J. A. (2009). Minimizing misdiagnosis: Psychometric guidelines for determin-ing possible and probable memory impairment in older adults. Dementia and Geriatric Cognitive Disorders, 27, 439-450.

Brooks, B. L., Iverson, G. L., Holdnack, J. A., & Feldman, H. H. (2008). The potential for misclassification of mild cognitive impairment: A study of memory scores on the Wechsler Mem-ory Scale-III in healthy older adults. Journal of the International Neuropsychological Society, 14, 463-478.

Cohen, J. (1988). Statistical power analysis for the behavioral sci-ences (2nd ed.). Hillsdale, NJ: Erlbaum.

Emerich, D. M., Creaghead, N. A., Grether, S. M., Murray, D., & Grasha, C. (2003). The comprehension of humorous materials by adolescents with high-functioning autism and Asperger’s syndrome. Journal of Autism and Developmental Disorders, 33, 253–258.

Ghaziuddin, M., & Mountain-Kimchi, K. (2004). Defining the intellectual profile of Asperger syndrome: Comparison with high functioning autism. Journal of Autism and Developmen-tal Disorders, 34, 279-284.

Gilchrist, A., Green, J., Cox, A., Burton, D., Rutter, M., & le Couteur, A. (2001). Development and current functioning in adolescents with Asperger syndrome: A comparative study. Journal of Child Psychology and Psychiatry, 42, 227-240.

Golarai, G., Grill-Spector, K., & Reiss, A. L. (2006). Autism and the development of face processing. Clinical Neuroscience Research, 6, 145-160. doi:10.1016/j.cnr.2006.08.001

Goldstein, G., & Saklofske, D. (2010). The Wechsler Intelligence Scales in the Assessment of psychopathology. In L. G. Weiss, D. H. Saklofske, D. Coalson, & S. E. Raiford (Eds.), WAIS-IV clinical use and interpretation: Scientist-practitioner perspec-tives. Burlington, MA: Academic Press.

Hayashi, M., Kato, M., Igarashi, K., & Kashima, H. (2008). Supe-rior fluid intelligence in children with Asperger’s disorder. Brain and Cognition, 66, 306-310.

Hollander, E., Bartz, J., Chaplin, W., Phillips, A., Sumner, J., Soorya, L., . . . Wasserman, S. (2007). Oxytocin increases retention of social cognition in autism. Biological Psychiatry, 61, 498-503.

Humphreys, K., Minshew, N., Leonard, G. L., & Behrmann, M. (2007). A fine-grained analysis of facial expression processing in high-functioning adolescents and adults with autism. Neu-ropsychologia, 45, 685-695.

Jou, R. J., Minshew, N. J., Keshavan, M. S., & Hardan, A. Y. (2010). Cortical gyrification in autistic and Asperger dis-orders: A preliminary magnetic resonance imaging study. Journal of Child Neurology. Advance online publication. doi:10.1177/0883073810368311

Korkman, M., Kirk, U., & Kemp, S. (2007). NEPSY (2nd ed.). San Antonio, TX: Pearson.

Lincoln, A. J., Courchesne, E., Kilman, B. A., Elmasian, R., & Allen, M. (1988). A study of intellectual abilities in high-functioning people with autism. Journal of Autism and Deve-lopmental Disorders, 18, 505-524.

Macintosh, K. E., & Dissanayake, C. (2004). Annotation: The similarities and differences between autistic disorder and Asperger’s disorder: A review of the empirical evidence. Journal of Child Psychology and Psychiatry, 45, 421-434.

Mayes, S. D., & Calhoun, S. L. (2003). Analysis of WISC-III, Stanford–Binet: IV, and academic achievement test scores in children with Autism. Journal of Autism and Developmental Disorders, 38, 428-439.

Pearson. (2009). Advanced clinical solutions for the WAIS-IV and WMS-IV. San Antonio, TX: Author.

Rinehart, N. J., Bradshaw, J. L., Tonge, B. J., Bereton, A. V., & Bellgrove, M. A. (2002). A neurobehavioral examination of individuals with high-functioning autism and Asperger’s dis-order using a fronto-striatal model of dysfunction. Behavioral and Cognitive Neuroscience Review, 1, 164-177.

Rutter, M., & Schopler, E. (1987). Autism and pervasive develop-mental disorders: Concepts and diagnostic issues. Journal of Autism and Developmental Disorders, 2, 159-186.

Siegel D. J., Minshew, N. J., & Goldstein, G. (1996). Wechsler IQ profiles in diagnosis of high-functioning autism. Journal of Autism and Developmental Disorders, 26, 387-388.

Spek, A. A., Scholte, E. M., & van Berckelaer-Onnes, I. A. (2008). Brief report: The use of WAIS-III in adolescents and adults with HFA and Asperger syndrome. Journal of Autism and Deve-lopmental Disorders, 38, 782-787.

Wechsler, D. (2003). Wechsler Intelligence Scale for Children (4th ed.). San Antonio, TX: Pearson.

Wechsler, D. (2008). Wechsler Adult Intelligence Scale (4th ed.). San Antonio, TX: Pearson.

at UNIV OF BRAZIL on May 9, 2014asm.sagepub.comDownloaded from