Abstract
Cognitive deficits in working memory (WM) are characteristic features of Attention-Deficit/Hyperactivity Disorder (ADHD) and autism. However, few studies have investigated cognitive deficits using a wide range of cognitive measures. We compared children with ADHD (n = 49) and autism (n = 33) with a demographically matched control group (n = 79) on a multidimensional battery of cognitive ability. Results confirmed previous research that both groups were characterized by deficits in WM. However, results also suggest verbal WM measures were better predictors than nonverbal WM measures. In addition, measures of visual-motor integration are equally discriminating of children with ADHD and autism from a matched control group. In all, 81% discrimination accuracy was obtained using only WM and visual-motor integration measures. Demonstrated shared deficits in WM and visual-motor integration are explained based on proposed neurological mechanisms common across the two disorders. Clinical implications are discussed.
Attention-Deficit/Hyperactivity Disorder (ADHD) and autism are two neurodevelopmental disorders characterized by cognitive deficits. However, research characterizing the cognitive deficits in ADHD and autism are confounded by the measures used to determine impairment. Often specific cognitive measures are aggregated to form an overall intellectual ability (IQ) score (Allen, Robins, & Decker, 2008; Barkley & Murphy, 2006; Decker, 2008). Research studies frequently report children with ADHD and autism typically have lower than average IQ scores than typically developing children (Barkley & Murphy, 2006). Similarly, research studies typically report that children with autism have lower overall IQ scores than control groups without clarifying which specific cognitive abilities account for this difference (Charman et al., 2011). However, the gross indicator of IQ may mask specific cognitive deficits—reflected on particular measures, or constructs, used to estimate overall IQ—rather than of globally low cognitive functioning. Understanding specific cognitive weaknesses, rather than overall IQ, may not only inform diagnostic assessment, but also intervention planning for children with ADHD and autism.
Several models of ADHD have emphasized the importance of specific cognitive abilities, rather than IQ composite measures, for understanding ADHD. For example, Barkley’s (1997) model of ADHD was one of the first to implicate working memory (WM) as a core executive functioning deficit to explain core symptoms of ADHD. Indeed, children evaluated for special education are frequently administered measures of WM as part of a comprehensive evaluation; however, scores from these specific measures are typically only used to provide a composite estimate of overall IQ score (Decker, 2008). Thus, group differences in overall IQ score may be driven by differences in specific cognitive abilities, such as WM, rather than overall intellectual functioning, and these deficits may be overlooked in clinical practice.
Similarly, autism is a neurodevelopmental condition caused by atypical brain development and characterized by cognitive deficits (Allen et al. 2008; Smalley & Collins 1996). Although there are clear behavioral differences that distinguish children with autism from children with ADHD, such as repetitive behaviors, social communication deficits, and restricted interests Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.; DSM-IV-TR; American Psychiatric Association [APA], 2000), recent research in autism has also focused on specific cognitive deficits, rather than overall IQ, to explain many of the symptoms of the disorder. For example, Lind and Williams (2011) have emphasized the importance of executive function deficits, including WM deficits, as a core contributor to some of the underlying problems in social deficits and repetitive behaviors of autistic children. Lopez, Lincoln, Ozonoff, and Lai (2005) showed that executive processes including cognitive flexibility, WM, and response inhibition are highly related to repetitive behaviors and restricted interests in autism and proposed a theory to predict core autism symptoms based on deficits in executive functioning.
The similarity in cognitive deficits between children with ADHD and those with autism may be explained by the similarity in frontal lobe abnormalities found in both groups. Abnormal connectivity in the frontal lobes is frequently found in individuals with autism (Griebling et al., 2010; Kumar et al., 2010); whereas frontal lobe abnormalities in dopamine pathways are found in individuals with ADHD (Levy, 1991; Levy & Swanson, 2001; Stein, Fan, Fossella, & Russell, 2007; Swanson et al., 2000). These findings are important because abnormalities in the frontal lobes are also linked to observed cognitive deficits in WM and executive functions (Griebling et al., 2010; Langen et al., 2012). As such, the common abnormalities in the frontal lobe may result in the common cognitive deficits of WM found in ADHD and autism.
Rommelse, Geurts, Franke, Buitelaar, and Hartman (2011) have presented evidence that shared cognitive endophenotypes—defined as an intermediate marker linking low-level genetics to higher level symptom presentation (Gottesman & Gould, 2003)—may explain the common cognitive deficits found in individuals with ADHD and autism. In addition to informing assessment and intervention practice, elucidating common, specific cognitive deficits in ADHD and autism can contribute to the evidence base for shared cognitive endophenotypes that Rommelse et al. suggest may hold the key to finding shared etiological factors for the two disorders.
The Current Study
The shared cognitive deficits across ADHD and autism provide interesting insights in the complexities of these disorders. Despite the consistent finding of WM deficits, few studies have investigated other possible common cognitive deficits across the two disorders. As previously mentioned, many studies collapse specific cognitive scores into a composite IQ score. The current study extends previous research by investigating specific cognitive deficits in children with ADHD and autism using a multidimensional battery of cognitive tests that included various WM measures. Guided by previous studies, we hypothesized that in using a standardized test of cognitive functioning, children with ADHD and autism would exhibit specific deficits in WM when compared with typically developing children. This hypothesis was based on previous research implicating abnormalities in frontal lobe functioning in conditions and the association of WM performance with the frontal lobes (D’Esposito, Detre, Alsop, & Shine, 1995).
Although findings of WM deficits in ADHD have been more consistent than those in children with autism, it is possible that this is due to the matching of children with autism to children without autism based on overall IQ. Matching based on overall IQ may have obscured differences in specific cognitive abilities, such as WM. As such, we focused on difference in specific cognitive constructs rather than overall IQ scores. We used the Stanford–Binet Intelligence Scales, Fifth Edition (SB5; Roid, 2003), because it measures a wide range of cognitive skills. In addition, the Bender–Gestalt II (BG-II; Brannigan & Decker, 2003), which was conormed with the SB5, provided a further extension of the skills measured. Together, these tests provided a broad array of measures to sample most domains of cognition. The conormed measures share the same standardization sample, making them ideal for statistical comparisons. Using both measures provides representations of a broad range of cognition, extending the visual processing and memory domains with Copy (visual-motor integration) and Recall (delayed recall) scores (Decker, Allen, & Choca, 2006). Both measures are frequently used in comprehensive psychoeducational evaluations of children in schools, which strengthen the ecological validity of the study.
We hypothesized children with ADHD and autism would demonstrate the greatest deficits on WM measures in comparison with typically developing children, given the research previously reviewed. To determine the degree of sensitivity in the measures to discriminate typically developing children from atypically developing children, we used Discriminant Function Analysis (DFA) to analyze the data. In using this methodology, we are not suggesting cognitive measures should solely be used to diagnose either condition. Rather, DFA provides beta weights (i.e., standardized canonical discriminant function coefficients) equivalent to those calculated in regression analysis (allowing us to identify areas of greatest deficit), but also provides information about classification accuracy, allowing the results to have clear clinical applications. These statistics are useful for determining the degree to which specific measures are sensitive to different clinical conditions.
Method
Participants
Participants in this study were recruited for the clinical (ADHD and autism) samples as part of the standardization and validity research for the SB5 and BG-II. For all participants, tests were administered by individuals with training in psychological assessment, and specific training on the SB5 and BG-II measures. Only participants conormed on both instruments were included. This sample was based on a stratified, random sampling plan to match the percentages of stratification variables from the U.S. 2000 census. Specifically, demographic matching during standardization and validity studies for the SB5 and BG-II included the variables: age, sex, ethnicity, geographic region, and socioeconomic level (see Brannigan & Decker, 2003, for details).
Disability condition (ADHD or autism) was based on DSM criteria as determined by a professional psychologist or a medical doctor using criteria from the Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV; APA, 1994). In addition, participants had to be receiving special education services in their schools for the respective disability at the time of testing. Only individuals with written documentation of an ADHD or autism diagnosis were included in the clinical samples for the SB5 and BG-II standardization study. Participants were also assessed at the time of data collection for the standardization sample for possible comorbid diagnoses (e.g., autism and ADHD).
From the archival data provided from SB5 and BG-II standardization and validity studies, we selected three groups for the current study: A nonclinical control group, an ADHD group, and an autism group. The control group (typically developing group) included individuals who did not meet diagnostic criteria for any DSM-IV disorder and were not receiving special education services for any disability at the time of testing. Only participants without missing data for any of the key variables, between the ages of 4 and 18, were selected for the control group. Participants for the clinical groups (ADHD and autism) were also selected from the standardization sample and excluded if they had missing data for any of the key variables used in the analysis or if they were outside the 4 to 18 years age range. The clinical group and the control group were matched on age and ethnicity.
We chose to focus on specific cognitive abilities in the current study, and we did not match participants on overall IQ score. As previously mentioned, matching children with autism based on overall IQ may obscure differences between clinical and nonclinical groups in specific cognitive abilities, such as WM. For example, when comparing a child with autism and a low overall IQ score with a child without autism and a low overall IQ score, both are likely to have low specific cognitive ability scores, as well. As WM is a specific cognitive ability score that contributes to the calculation of overall IQ, children matched on IQ may have more similar WM—or specific cognitive ability—scores than if they were not matched on IQ.
In addition, for instance, findings of WM deficits in ADHD have been more consistent than those in children with autism, but this may be due to the fact that matching of control participants based on IQ is more common in studies of children with autism than in studies of those with ADHD. For example, in Korkman and Pesonen (1994), children with ADHD were not matched to either the control group or the learning disabilities group based on IQ; in fact, the authors noted that the groups were significantly different in terms of verbal IQ. Results indicated significant differences between groups in specific neuropsychological tasks, including inhibition and control and WM. However, in studies of children with autism, such as that conducted by Ozonoff and Strayer (2001), the evidence is often mixed regarding WM deficits, and subjects are either grouped or matched on the basis of overall IQ score. As the focus of the current study was on determining which specific cognitive abilities best discriminate between groups, we chose not to match participants based on overall IQ scores.
The final sample used for the analysis included 161 total participants, with 61 females and 100 males, ages 4 to 18 years, with 79 (37 females, 42 males) in the typically developing control group, 49 (15 females, 34 males) in the ADHD group, and 33 (9 females, 24 males) in the autism group (see Table 1). The ADHD group included individuals from Combined Type (n = 25), Primarily Inattentive (n = 14), and Primarily Hyperactive (n = 10) subtypes. Parents of participants with ADHD were asked not to administer their child’s medication for the testing session. No individuals in the ADHD group had comorbid autism diagnoses, and no individuals in the autism group had comorbid ADHD diagnoses. In the ADHD group, one individual had an additional Developmental Delay classification, one was receiving services for comorbid Serious Emotional Disturbance, and one was receiving services for a comorbid learning disability in writing. In the autism group, one individual had an additional visual-auditory disability classification, and two were receiving services for comorbid learning disabilities (one in reading, one in reading and writing).
Demographic Information for Typically Developing Control, ADHD, and Autism Groups.
Note. ADHD = Attention-Deficit/Hyperactivity Disorder.
Chi-squared tests indicated that when the typically developing control group was compared with the clinical group (ADHD and autism), there were no significant differences in ethnicity, χ2(4, N = 161) = 1.33, p = .857; mother’s education level, χ2(4, N = 161) = 8.58, p = .073; father’s education level χ2(7, N = 161) = 11.53, p = .117; or school type (parochial, private–nonparochial, or public), χ2(3, N = 161) = 7.69, p = .053. There were significant differences between the groups in proportion of males to females, however, with the clinical (ADHD and autism) group having a higher ratio of males to females than the control group, χ2(1, N = 161) = 5.28, p = .857. This is consistent with the higher incidence of ADHD (Centers for Disease Control and Prevention [CDC], 2012) and autism (Fombonne, 2005) in males than in females. Results of a one-way ANOVA indicated no age differences in the control versus the clinical group, F(1) = 0.74, Mean Square Error (MSE) = 11.64, p = .451.
Measures
The SB5 battery is based on multiple theoretical perspectives of cognitive ability. It is based on the Cattell-Horn-Carroll (CHC) model of intelligence (Carroll, 1993; Cattell, 1963). According to the SB5 Technical Manual (Roid, 2003), construct validity, as determined by confirmatory factor analysis, suggests the presence of five factors which correspond to CHC theory: Fluid Reasoning (Gf), Knowledge (Gc), Quantitative Reasoning (Gq), Visual-Spatial Processing (Gv), and Working Memory (Gsm)—each of which comprise one Verbal and one Nonverbal subtest. However, it also includes Verbal and Nonverbal dimensions of these constructs. Pomplun and Custer (2005) have demonstrated that the SB5 WM subtests have excellent construct validity. The SB5 Verbal WM subtest includes Memory for Sentences, in which examinees repeat orally provided sentences of increasing length, and Last Word, in which examiners read increasingly long sequences of questions and examinees must answer each question and recall the last word in each question at the end of the sequence. Last Word is based on reading span tasks, which have shown to correlate highly with higher level cognition and academic achievement (Daneman & Carpenter, 1980). The SB5 Nonverbal WM subtest includes Delayed Response, in which examinees must identify which cup a toy duck is hidden under after a brief delay from presentation, and Block Span. In Block Span, the examiner taps an increasingly long and complex sequence of blocks and the examinee must imitate the sequence.
The normative sample of the SB5 included a nationally representative sample of 4,800 individuals and reflected the demographic characteristics of the 2000 U.S. Census. Internal reliability coefficients for the Full-Scale IQ ranged from .97 to .98 across 23 age groupings. Average reliabilities across ages for the Nonverbal IQ and Verbal IQ were .95 and .96, respectively. Average reliabilities for Verbal subtests ranged from .84 to .89; for Nonverbal subtests average reliabilities ranged from .85 to .89. In addition, average correlations for the five-factor index scores were as follows: Fluid Reasoning .90, Knowledge .92, Quantitative Reasoning .92, Visual-Spatial Processing .92, and WM .91. Mean reliability coefficients across the age groupings spanned from .84 to .89. Split-half reliability coefficients, corrected using the Spearman–Brown formula, for individual subtests ranged from .85 to .88, for Nonverbal subtests, and from .84 to .89 for Verbal subtests. The test–retest Full Scale IQ coefficients across all age ranges were high, ranging from .93 to .95. Correlations for the Nonverbal IQ were from .89 to .93, and for the Verbal IQ, .92 to .95. The correlations for the Factor Indexes spanned from .79 to .85, with a median of .88. Finally, stability coefficients across the subtests ranged from .66 to .93. Interrater agreement correlations ranged from .74 to .97, with a median of .90.
In addition to subtests in the SB5, a measure of visual-motor integration and recall performance was also included in this study. Such tests are routinely given with IQ tests, and there is some empirical support for the use of such tests in identifying neurodevelopmental conditions. In this study, visual-motor integration was operationalized as the participant’s standard score on the Copy (visual-motor), subtest of the BG-II (Brannigan & Decker, 2003). During the Copy subtest, examinees are asked to copy the designs presented on stimulus cards, and during the Recall subtest, they must draw the designs again from memory. In the Bender–Gestalt II Examiner’s Manual, Brannigan and Decker (2003) report interrater reliabilities for the Copy and Recall subtest of .85 and .92, respectively. The overall reliability for the standardization group was .91. Corrected test–retest coefficients for the Copy subtest ranged from .80 to .88, whereas the range for the Recall subtest was from .80 to .86. Criterion validity was established via correlations with the Beery-Buktenica Developmental Test of Visual-Motor Integration (Beery, 1997; Copy r = .65; Recall r = .44) and the Koppitz Developmental Bender Scoring System (Koppitz, 1975; Copy r = .80; Recall r = .51). An exploratory factor analysis yielded high factor loadings on a single factor. Explained variances ranged from 47.51% to 64.70%, with the highest percentage for the 4 to 7 age group (for additional information see the BG-II Technical Manual; Brannigan & Decker, 2003).
Results
A DFA was computed using only the five SB5 factor scores (Knowledge, Quantitative Reasoning, Fluid Reasoning, WM, and Visual-Spatial) and BG-II Copy and Recall scores, with three groups: typically developing, ADHD, and autism. We used DFA to calculate percentages of each group correctly classified by the specified measures using prior probabilities based on group sample sizes. SB5 factor scores, rather than overall IQ, were examined first to obtain a broad picture of which cognitive abilities best predicted group membership. In addition, the factor level is the recommended focus of interpretation for practitioners (Roid, 2003).
Descriptive statistics for each group’s subtest scores are shown in Table 2. Standardized canonical discriminant function coefficients from the DFA indicated that, of the seven SB5 and BG-II variables entered simultaneously in Step 1, the SB5 WM factor best predicted group membership, F(2, 158) = 29.22, p < .001. In Step 2, after accounting for variance explained by SB5 WM, the BG-II Copy scores were the most significant predictor of group membership, F(4, 314) = 20.13, p < .001. Using prior probabilities based on group sample sizes, 89.8% of typically developing individuals, 40.0% of individuals with ADHD, and 44.1% of individuals with autism were correctly classified by the model including only SB5 WM and BG-II Copy scores. On average, this function correctly classified individuals at a rate of 66.3%, which should be compared with a rate of approximately 33% for random classification of 3 groups. Within an applied context, the incremental validity should be compared with the base rates of these conditions, which are approximately 5% to 15% for ADHD and approximately 1% for autism (CDC, 2009, 2012).
Descriptive Statistics for SB5 Subtest Scores a and BG-II Copy and Recall Scores for Typically Developing Control, ADHD, and Autism Groups.
Note. SB5 = Stanford–Binet Intelligence Scales, Fifth Edition; BG-II = Bender–Gestalt II; ADHD = Attention-Deficit/Hyperactivity Disorder; N = Nonverbal; V = Verbal; FR = Fluid Reasoning; KN = Knowledge; QR = Quantitative Reasoning; VS = Visual-Spatial; WM = Working Memory.
SB5 factors each include one verbal and one nonverbal subtest.
n = 79.
n = 49.
n = 33.
n = 161.
We collapsed ADHD and autism groups into one clinical group for further analyses. Results of the DFA are presented in Table 3. Standardized canonical discriminant function coefficients indicated results similar to those in the first analysis: BG-II Copy performance best predicted group membership (typically developing vs. Clinical), F(1, 159) = 52.08, p < .001, followed by SB5 WM, F(2, 158) = 39.23, p < .001. Again, based on prior probabilities due to sample sizes, the model including only SB5 WM and BG-II Copy scores correctly classified 84.1% of typically developing individuals and 78.6% of clinical individuals, with a mean classification accuracy of 81.4%.
Discriminant Function Analysis for Clinical (ADHD and Autism) Versus Typically Developing Control Groups Using SB5 Factor Scores and BG-II.
Note. Percentage of grouped cases correctly classified = 81.4%. SB5 = Stanford–Binet Intelligence Scales, Fifth Edition; BG-II = Bender–Gestalt II; FR = Fluid Reasoning; KN = Knowledge; QR = Quantitative Reasoning; VS = Visual-Spatial; WM = Working Memory.
SB5 factors each include one verbal and one nonverbal subtest.
Variables not included in the analysis.
Because SB5 WM measures consist of a verbal and a nonverbal test, we conducted a subsequent DFA using the SB5 subtest scores (Verbal and Nonverbal Knowledge, Verbal and Nonverbal Quantitative Reasoning, Verbal and Nonverbal Fluid Reasoning, Verbal and Nonverbal WM, and Verbal and Nonverbal Visual-Spatial), along with BG-II Copy and Recall scores. Focusing on the subtest level helped specify whether verbal or nonverbal subtests from the SB5 factors were driving factor-level results. Results indicated that, again, BG-II Copy score, F(4, 314) = 20.11, p < .001, was the most significant predictor of group membership, followed by the SB5 Verbal WM subtest, F(2, 158) = 27.32, p < .001. This clarifies that the Verbal, rather than Nonverbal, WM measure provides the greatest discrimination of group membership.
Discussion
Results from this study confirm that children with ADHD and autism are distinguishable from typically developing children by specific cognitive impairments in WM. We hypothesized that of all cognitive measures within a standard IQ battery, cognitive measures of WM would accurately distinguish children with ADHD and autism from typically developing children. Results from this study supported this prediction. However, measures of visual-motor integration, as measured by the BG-II Copy score, were also found to be sensitive predictors in discriminating typically developing children from atypically developing children. This was not predicted a priori. Additional analyses of the data at the subtest level suggest, specifically, that verbal WM and visual-motor integration scores best discriminated children with ADHD and autism from typically developing children.
Classification accuracies were much higher for distinguishing the clinical group (autism and ADHD combined) from the typically developing group than for distinguishing three groups—autism, ADHD, and typically developing. This suggests that WM and Visual-Motor Copy measures distinguish clinical from typically developing groups, but do not accurately distinguish between the conditions (ADHD and autism).
WM (Gsm) was a consistent contributor to variance in group membership across multiple analyses, supporting previous research findings of WM deficits in ADHD and autism (Alloway & Archibald, 2011; Barkley, 1997; Bennetto, Pennington, & Rogers, 1996; Boucher & Lewis, 1989; Boucher & Warrington, 1976; Gathercole & Alloway, 2006; Happe, Booth, Charlton, & Hughes, 2006; Levy & Swanson, 2001; Martinussen, Hayden, Hogg-Johnson, & Tannock, 2005; Ozonoff, Pennington, & Rogers, 1991; Shalom, 2003; Sowerby, Seal, & Tripp, 2010; Steele, Minshew, Luna, & Sweeney, 2007; Williams, Goldstein, & Minshew, 2006). We also found WM deficits in both disorders, and further analyses clarified that verbal, rather than nonverbal, measures of WM best predicted group membership. However, the finding that visual-motor integration performance significantly predicts group membership was unexpected.
WM and visual-motor functions involve frontal areas of the brain, in which individuals with autism and ADHD have demonstrated abnormalities (Griebling et al., 2010; Kumar et al., 2010; Langen et al., 2012; Levy, 1991; Levy & Swanson, 2001; Stein et al., 2007; Swanson et al., 2000). Our results also converge with Rommelse and colleagues’ (2011) report of shared deficits in frontal lobe functions such as executive functioning (e.g., WM) and motor coordination (e.g., visual-motor integration) in children with ADHD and autism, and offer support for verbal WM and visual-motor skills as possible common cognitive endophenotypes linking neurobiological and genetic vulnerabilities to behavioral deficits.
In ADHD and autism, impaired abilities to integrate spatial and temporal, verbal and nonverbal, or sensory and motor information and action suggests integration of diverse information in the frontal lobes is one general area of weakness these measures may be detecting (Levy & Swanson, 2001; Oberman & Ramachandran, 2008; Prabhakaran, Narayanan, Zhao, & Gabrieli, 2000). As such, the measures that best predicted group membership in this study may be a consequence of including tasks that focus on the integration of multiple cognitive capabilities (e.g., verbal WM and visual-motor integration). This weakness may be related to impairments in functional connectivity and abnormalities in the frontal areas of the brain, which have been demonstrated in both populations (Griebling et al., 2010; Kumar et al., 2010; Langen et al., 2012; Levy, 1991; Levy & Swanson, 2001; Stein et al., 2007; Swanson et al., 2000).
Implications for Practice
Results from this study extend the literature on cognitive deficits in ADHD and autism and inform neuropsychological practice by suggesting comprehensive assessment for children with these disorders should include specific measures of not only WM, but also visual-motor integration. Specifically, tests of verbal WM should be included. The results of this study suggest that, in addition to evaluating the symptoms and difficulties traditionally assessed in ADHD and autism evaluations to arrive at a clinical diagnosis using DSM criteria, verbal WM and visual-motor integration assessment are also important indicators for determining whether children may need additional intervention to address specific cognitive deficits. The traditional focus on overall IQ in the assessment of children with ADHD and autism may obscure these specific cognitive strengths and weaknesses.
Furthermore, assessment results from these cognitive measures should be linked to intervention planning. Although children with ADHD and autism may exhibit deficits in multiple cognitive abilities, our results suggest differences in verbal WM and visual-motor integration are the largest when compared with typically developing children. These deficits may therefore warrant more attention than overall IQ or other cognitive areas and will need to be addressed in intervention and instructional planning—or at least considered when selecting interventions targeting autism or ADHD symptoms. To date, no intervention has been demonstrated to be effective in directly increasing overall IQ. However, interventions specifically targeting WM and visual-motor integration are available and have demonstrated effectiveness for children with neurodevelopmental disabilities (e.g., Dawson & Watling, 2000; Klingberg, 2010; Klingberg, Forssberg, & Westerberg, 2002, Klingberg et al.,2005). This may be especially important for practitioners working with children with ADHD, as there is no special education category specifically for children with ADHD, but schools can provide targeted supports and interventions specific to children’s strengths and weaknesses in general and special education under other categories, such as Other Health Impairment.
Limitations and Future Directions
The current study could be improved by increasing the sample size for the ADHD and autism groups, which would provide stronger evidence for the generalizability of these findings. In addition, there were limitations in the current study regarding similarities and differences among the autism, ADHD, and nonclinical samples. First, we did not match or control for participant gender. Although the composition of our clinical group reflects prevalence estimates showing a higher incidence of ADHD and autism in boys than in girls (CDC, 2012; Fombonne, 2005), and no consistent evidence has been found in typically developing children for significant gender differences in WM performance, future studies may wish to explore the contributions of gender to differences in specific cognitive abilities across clinical and nonclinical groups. In addition, although there were no statistically significant differences across groups on key demographic variables, differences between the clinical and nonclinical groups approached significance for mother and father education and school type. Future studies with larger samples for each group may benefit from obtaining samples with more closely matched participants to rule out the possibility that any significant results are due to demographic factors.
Although the constructs of WM and visual-motor integration proved valuable in differentiating “typical” from “atypical” development, they did not provide high differential validity specific to the condition (i.e., ADHD or autism). As such, additional measures are needed to provide further discrimination between these conditions on the basis of cognitive abilities. In addition, future studies should explore which other brief, already widely used cognitive measures might differentially predict specific group membership (ADHD vs. autism). Based on the results from the current study, researchers should focus on measures of WM, visual-motor integration, and perhaps on tasks that require integration of different skills (e.g., sensory and motor) that are distributed across brain regions.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.