J Autism Dev Disord (2008) 38:988994 DOI 10.

1007/s10803-007-0468-z

BRIEF REPORT

Brief Report: Human Figure Drawings by Children with Aspergers Syndrome

Hui Keow Lim Virginia Slaughter

Published online: 9 October 2007 Springer Science+Business Media, LLC 2007

Abstract Twenty-nine children with Aspergers syndrome and 28 typically developing children, matched on gender, chronological age and nonverbal IQ, were asked to produce a free drawing, then requested to draw a person, a house and a tree. The drawings were scored using standardized procedures for assessing accuracy, detail and complexity. There were no differences between the diagnostic groups on the tree or house drawing scores. The human gure drawing scores of children with Aspergers syndrome were signicantly lower than those of the typically developing children, and there was a positive correlation between human gure drawing scores and communication sub-scores on the Vineland Adaptive Behaviour Scales, for the Aspergers group. These results suggest that the selective decit in generating human gure representations may derive from a relative lack of interest in the social world, and/or limited practice in drawing people. Keywords Aspergers syndrome Human gure drawing Pictorial representation Draw-a-person test Vineland adaptive behaviour scales

Introduction Childrens drawings, especially those of the human gure, have been the focus of attention for decades because they

H. K. Lim V. Slaughter (&) School of Psychology, University of Queensland, Brisbane, QLD 4072, Australia e-mail: vps@psy.uq.edu.au

are believed to reect a number of psychological variables including the drawers intellectual maturity, personality, thought and emotion (Goodenough 1926; Koppitz 1968; Machover 1949). There is a good deal of debate about the reliability and validity of drawings as clinical assessment tools (see Motta et al. 1993). There is less controversy about the value of childrens drawings as measures of intellectual maturity. Several standardized and externally validated systems for scoring childrens drawings are in existence. The majority of these focus, partly or exclusively, on childrens drawings of the human gure. Human gure drawings are rst produced in the toddler period, and then follow a reasonably consistent developmental trajectory, increasing in accuracy, detail and sophistication with age (Cox 1993). The developmental progression of childrens human gure drawings is sufciently standard to allow estimates of IQ based on human gure drawings alone (Goodenough 1926; Harris 1963; Koppitz 1968; Naglieri 1988). Researchers debate the extent to which the developmental progression in childrens human gure drawings reects changes in childrens knowledge of the human form, as opposed to the acquisition of drawing skill and accumulated practice in generating human gures (Cox 1993; Freeman 1987; Motta et al. 1993). However most agree that childrens drawings, particularly those of the human gure, reect developmental stages that are likely to be linked to both cognitive and socialization processes. Human gure drawings by children with developmental disorders typically reveal delays that are concomitant with any intellectual impairment (Cox and Maynard 1998; Dykens et al. 2000; Golomb and Barr-Grossman 1977). In addition, human gure drawings of children with developmental disorders may reect elements consistent with their specic diagnosis. For instance Dykens et al. (2000)

123

J Autism Dev Disord (2008) 38:988994

989

investigated human gure drawings by children with Williams syndrome and found that those childrens standardized scores on the Draw-a-person test (Koppitz 1968) were higher than their scores on a comparable developmental test of visual and motor integration. Dykens et al. (2000) interpreted this selective difference across drawing topics as reecting the interest in people, and relative strength in reading facial expressions and remembering faces, that are typically exhibited by children with Williams syndrome (Tager-Flusberg et al. 1998). In a similar vein, Lewis and Boucher (1991) hypothesized that children with autism would possess less mature representations of people compared to typically developing children and this would be reected in their drawings. They scored the spontaneous human gure drawings generated by children in both diagnostic groups using Harriss (1963) system. The children with autism did receive lower human gure drawing scores than the comparison group, but the difference was not statistically signicant. Their ndings were based on only 11 children with autism, which may mean that their analyses lacked sufcient power to detect differences between the groups. Lee and Hobson (2006) investigated the distinctiveness of human gure drawings by 14 children with autism, compared to a matched group of children with learning difculties but not autism. They asked children to draw three different human gures and three different houses. Like Lewis and Boucher (1991) they reported slightly lower global scores for human gure drawings by the children with autism compared to the control group, although the difference was not statistically signicant. The human gure drawings of the children with autism were signicantly less individually distinctive than the human gure drawings of children without autism, whereas the drawings of houses produced did not differ in distinctiveness across diagnostic groups. There is some indication, therefore, that the human gure drawings of children with autism may be generally less sophisticated and detailed than those of nonautistic children, but previous studies addressing this specic question have lacked sufcient power to reveal reliable group differences. In the current study we tested a relatively large sample of children with Aspergers syndrome (N = 29) and a matched group of typically developing children. We systematically compared the drawings of human gures, houses and trees across the two groups. Two hypotheses were investigated, based on the fact that children with Aspergers syndrome are relatively disinterested in the social world: (a) the human gure drawings of children with Aspergers syndrome may be selectively less sophisticated than those of the typical children but the drawings of other topics will be equivalent, and (b) in the Aspergers syndrome group, the level of sophistication of

human gure drawings may correlate positively with communication and social functioning scores on the Vineland Adaptive Behaviour Scales (Sparrow et al. 1984), reecting individual differences in engagement with, and knowledge of, other people.

Methods Participants The sample consisted of 29 children with Aspergers syndrome (25 boys and 5 girls) and 28 typically developing children (24 boys and 5 girls). The mean chronological age of children with Aspergers syndrome was 9.6 years (range = 4.913.6 years). The mean chronological age of typically developing children was 9.4 years (range = 5.3 13.6 years). The groups were matched on nonverbal IQ as assessed by the Standard Progressive Matrices (de Lemos 1989): The mean IQ of the typically developing children was 114.4 (range = 85137) and the mean IQ of the children with Aspergers syndrome was 109.5 (range = 78 139). The participants were primarily Caucasian. The children with Aspergers syndrome were diagnosed either by a psychologist or by a psychiatrist prior to acceptance into the study. Two of the children with Aspergers syndrome were home schooled while the rest attended mainstream schools in Brisbane, Australia. One child in the Aspergers syndrome group attended art lessons outside school curriculum and none of the typical children did so. Parents and children gave written consent for participation in the study.

Measures IQ The Standard Progressive Matrices (de Lemos 1989), using Australian norms, was administered to estimate nonverbal intelligence. This test was chosen for this study because it is not a language-based assessment and the focus of this study was on a nonverbal drawing task.

Vineland Adaptive Behaviour Scales The parents or the main caregivers of all children in the study were interviewed on the communication and socialization sub-domains of this instrument, as these were considered the most relevant to childrens engagement with and knowledge about other people.

123

990

J Autism Dev Disord (2008) 38:988994

Drawing Materials For the drawing task children were provided with a few pencils, a pencil sharpener, an eraser and four pieces of paper (37.5 27.7 cm and presented in landscape orientation). Large paper was chosen to alleviate any constraints that smaller paper may place on drawing strategy (Clements and Barrett 1994).

scores, total good scores and global scores of house, tree and person drawings were transformed into standardized (z) scores.

Koppitzs Human Figure Drawing Scoring System The human gure drawings alone were scored using an independent technique developed by Koppitz (1968). This provided a convergent measure for the childrens human gure drawings. Koppitzs (1968) system includes both developmental and emotional indicators. For the purpose of this study, only the developmental indicators were used as they reect the accuracy, detail and complexity of the human gures portrayed. The presence of each developmental indicator (e.g. digits on the hand) gained one point from a checklist of 30 developmental indicators.

Procedure Children were visited in their homes and tested individually without time limit imposed for the completion of tasks. All children were rst asked to draw a picture of anything they liked, as a warm up task. Then they were asked to draw a house, a tree and a person. The order of the house, tree and person requests was randomized across participants. Next, the children were assessed on the Standard Progressive Matrices. Finally, the parents or caregivers of the children were interviewed. Most of the children completed all the tasks in one session. Three children with Aspergers syndrome were tested over two sessions on different days.

Inter-rater Reliability The drawings were scored by the rst author, who was not blind to the diagnostic status of the children. The complete set (100%) was then re-scored by a second rater who was blind to the age, IQ, gender and diagnosis of the children. For Bucks scoring system, inter-rater reliability across all three drawing topics achieved 91.1% agreement, with a Cohens Kappa of 0.83. For Koppitzs scoring system, both raters agreed on 95.3% of the developmental indicators. Inter-rater disagreements were resolved by discussion.

Scoring The free drawings were not scored. The directed drawings were scored using two different standardized systems. The rst generated scores for all three of the directed drawings (house, tree and person) and the second generated detailed scores for the person drawings only.

Results Comparing Diagnostic Groups on Drawings of House, Tree and Person To evaluate the relative quality of the directed drawings across diagnostic groups, a 2 3 mixed model ANOVA was computed with diagnostic group (children with Aspergers syndrome versus typically developing children) as the between-groups variable and drawing topic (house versus tree versus person) as the within-groups variable. The dependent variable was Bucks standardized global drawing score. There was a main effect for drawing topic, F(2, 54) = 3.96, p \ 0.05, partial g2 = 0.13, subsumed by a signicant interaction of drawing topic and diagnostic group, F(2, 54) = 4.67, p \ 0.05, partial g2 = 0.15. Table 1 provides the standardized global scores for house, tree and person drawings by the two groups of children. Independent samples t-tests comparing the mean standardized global scores of the drawings by the two groups of

Bucks House-tree-person Scoring System Evaluations of the house, tree and person drawings were based on the presence and absence of features, details, proportions and perspectives (Buck 1948). Two scores were generated: A good score that represented positive inclusions, and a aw score that represented omissions and errors. Examples of good features include a house with wall material shown in any recognizable way, a tree with two-dimensional roots tapering into the ground, and a person with recognizable digits on the hand. Examples of awed aspects presented in the drawings were a house without a roof, a tree with one-dimensional branches and a person with a circular two-dimensional trunk. A global score for each drawing was computed by subtracting the total aw score from total good score. In order to compare the relative performance of two groups of children across the three directed drawings, the total aw

123

J Autism Dev Disord (2008) 38:988994 Table 1 Bucks standardized global scores and Koppitzs developmental scores for house, tree and human gure drawings, by diagnostic group

991

Drawing topic

Children with Aspergers syndrome Mean (SD) 0.12 (0.78) 0.20 (0.49) 0.61 (1.54) 18.00 (5.06)

Typically developing children Mean (SD) 0.08 (0.74) 0.21 (0.46) 0.40 (1.23) 21.57 (3.43)

Bucks standardized global scores (z-scores)

House Tree Person

Koppitzs human gure developmental score (max = 30)

Person

children were conducted to explore the signicant interaction. The standardized global scores for house drawings did not differ across groups, t(55) = 0.19, p = 0.85, nor did the those for tree drawings, t(55) = 0.05, p = 0.96. In contrast the standardized global scores for the human gure drawings were signicantly different across groups, t(55) = 2.73, p \ 0.01, though the magnitude of the differences in the means was small (g2 = 0.12). This indicates that the human gure drawings by the children with Aspergers syndrome were less sophisticated than those by the typically developing children. As described above, Bucks scoring system tabulated both good and awed features of human gure drawings. On this system and as shown in Fig. 1, the human gure drawings by children with Aspergers syndrome had signicantly more aws, t(55) = 2.65, p \ 0.005, and fewer good details, t(55) = 2.41, p \ 0.05, than those by typically developing children. The g2-values for these comparisons were 0.11 and 0.10, respectively.

Comparing Diagnostic Groups on Developmental Indicators for the Human Figure Drawings Koppitzs system for scoring the human gure drawings also revealed a signicant difference between the groups, t(55) = 3.13, p \ 0.005, g2 = 0.15. Children with Aspergers syndrome drew fewer indicators of developmental maturity in their human gures, compared to the typically developing children. As a matter of interest, we correlated Bucks standardized global scores with the scores derived form Koppitzs system and found that they were highly consistent, r(56) = 0.88, p \ 0.001. This suggests that the two scoring systems generated valid independent measures of the developmental sophistication of childrens human gure drawings.

Relations between Communication, Social Functioning and Human Figure Drawings To test the hypothesis that there may be a correlation between social and/or communication skills and the ability to draw a person in children with Aspergers syndrome, we correlated the Vineland sub-domain scores with human gure drawing scores obtained with both scoring systems, partialling out age and IQ. The partial correlation between Vineland communication sub-scores and Bucks global human gure drawing scores just missed statistical signicance: Partial r(25) = 0.37, p = 0.061 (two-tailed), but the magnitude of the correlation was nevertheless noteworthy. The partial correlation between Vineland communication sub-scores and Koppitzs human gure developmental indicators was signicant: Partial r(25) = 0.41, p \ 0.05 (two-tailed). There were no signicant correlations between Vineland socialization sub-scores and human gure drawing scores among the children with Aspergers syndrome. Although our hypotheses only concerned the children with Aspergers syndrome, for completeness we ran the

1.4 1.2 1

M=1.31 SD=1.41 M=0.98 SD=1.08

Means

0.8 0.6 0.4 0.2 0 Standardised Global Flaw Score Standardised Global Good Score
M=0.49 SD=.87 M=0.31 SD=1.04

Children with Asperger's syndrome Typically developing children

Fig. 1 Bucks aw and good scores for human gure drawings by diagnostic group

123

992

J Autism Dev Disord (2008) 38:988994

Children with Aspergers syndrome Typically developing children

same correlations between human gure drawing scores and both Vineland sub-domain scores for the typically developing children. No signicant correlations were observed.

Discussion In this study we found some systematic differences between the human gure drawings of typically developing children and those diagnosed with Aspergers syndrome. As a group, the children with Aspergers syndrome produced less sophisticated human gure drawings than typically developing children, in contrast to their house and tree drawings which were as good as those by the typically developing children. This pattern indicates that the children with Aspergers syndrome were not decient in their general drawing skill, but rather selectively less likely to produce a developmentally advanced, high-scoring human gure drawing. This nding contrasts with both Lewis and Boucher (1991) and Lee and Hobson (2006). As noted above, sample sizes in those two studies were relatively small compared to that of the current study, and since the group difference, while statistically signicant, is a fairly weak effect, it is perhaps not surprising that this group difference in human gure drawing was not reliable when the sample sizes were smaller. The uneven prole in drawing ability seen in our sample of children with Aspergers syndrome is consistent with previous observational studies of drawing in children with autism (Kellman 2001) and also with the fact that IQ proles of individuals with Aspergers syndrome are often characterized by peaks and troughs (Goldstein et al. 2001). Several factors may account for this selective decit in human gure drawings by children with Aspergers syndrome. First, because children on the autism spectrum spend less time looking at people (Swettenham et al. 1998) they may as a result possess less detailed representations of human beings (as suggested by Lewis and Boucher 1991). Second, compared to typically developing children, those with Aspergers syndrome may be less motivated to generate accurate and detailed drawings when given the instruction to draw a person. That is, if children with Aspergers syndrome are more interested in inanimate objects and/or their special interests, then they may put relatively little effort into producing a human gure drawing. Indeed, in the current study several children with Aspergers syndrome drew perfunctory humans in the context of their special interests (e.g. a human stick gure being squashed under a realistically portrayed elephants foot; see Fig. 2). Third, children with Aspergers syndrome may be less likely to practice drawing humans than their typical peers, and this may result in developmentally less mature
8.33 years old; Female IQ 127 (special interest is ) 7.33 years old; Female IQ 127

8.25 years old; Male IQ 118

8.17 years old; Male IQ 127

8.33 years old; Male IQ 115

8.67 years old; Male IQ 108

9.50 years old; Male IQ 104 (special interest is Egyptian myths)

9.75 years old; Male IQ 110

11.17 years old; Male IQ 111 (special interest is elephants)

11.5 years old; Male IQ 110

Fig. 2 Examples of human gure drawings by diagnostic group

drawings of people. In support of this interpretation, we noted informally during our testing that more than one child with Aspergers syndrome stated that they were poor at drawing people. We also observed that a number of children in the Aspergers group began their human gure drawings at a location other than the head, which is different from the norm for typical children (Koppitz 1983). This difference in drawing procedure could reect group differences in local versus global processing strategies

123

J Autism Dev Disord (2008) 38:988994

993

(Frith 1989) or else may reect lack of practice in human gure drawing by the children with Aspergers syndrome. Finally, it is possible that the request to draw a human gure may have selectively increased the cognitive load for the Aspergers-diagnosed children, exacerbating any ne motor problems and thereby causing them to produce developmentally less advanced, relatively low scoring human gure drawings. Our second hypothesis was that children with Aspergers syndrome who were reported to have relatively good communication and socialization skills, would draw more accurate and detailed human gures than those whose communication and socialization skills were weaker. This hypothesis was partially supported. We found a signicant partial correlation between Vineland communication subscores and person drawing scores derived from Koppitzs scoring system (1968). There was however no correlation between socialization sub-scores on the Vineland and the quality of human gure drawings. The fact that scores on person drawing correlated with communicative functioning, rather than socialization, may reect a link between the Aspergers-diagnosed childrens capacity to understand and follow the direction to draw a human, and to put his or her ideas accurately to paper. However, it is worth noting that there was no correlation between Vineland communication scores and childrens drawings of the house or tree, in either group, so this explanation holds only for the production of human gure drawings. We observed no correlation between Vineland scores and the typically developing childrens drawings, which goes against the idea that there is a general relation between individual childrens communication or socialization skills and their ability to draw human gures. It may be that this study lacked the power to reliably detect those correlations because although the sample size represents an improvement over previous studies, there were fewer than 30 children per diagnostic group. A replication study that includes more children would be valuable, both to replicate the main nding of differences in the sophistication of human gure drawings, and to further explore how these relate to childrens socialization and communication skills. Mention must be made of another limitation associated with the present study. We followed Bucks standard procedure and asked the children to draw a person in the house-tree-person drawing task. It is possible that children with Aspergers syndrome interpreted this instruction differently from the typically developing children. For example, the drawings of Marge Simpson and the Egyptians by children with Aspergers syndrome were quite accurate in the context of those themes (see Fig. 2), but were penalized in both scoring systems because of the built-in expectation that children would draw a human

form that closely resembled a real person. This potential misinterpretation of the instruction by some children with Aspergers syndrome might also account for the correlation between communication sub-scores and person drawing scores discussed above. In future, researchers may want to alter the standard instruction for children on the autism spectrum, and ask them to draw a real person, when eliciting human gure drawings.
Acknowledgments This study was carried out by the rst author in partial fullment of the degree of Doctor of Psychology at the University of Queensland, Australia. We are grateful to the children and parents who participated. We also thank Philippa Neary for coding assistance, and two anonymous reviewers for their constructive feedback on an earlier version of the manuscript.

References
Buck, J. N. (1948). The H-T-P technique: A qualitative and quantitative scoring manual. Journal of Clinical Psychology, 4(4), 317396. Clements, W., & Barrett, M. (1994). The drawings of children and young people with Downs syndrome: A case of delay or difference? British Journal of Educational Psychology, 64, 441 452. Cox, M. (1993). Childrens drawings of the human gure. Hove: Erlbaum. Cox, M. V., & Maynard, S. (1998). The human gure drawings of children with Down syndrome. British Journal of Developmental Psychology, 16, 133137. de Lemos, M. M. (1989). Standard progressive matrices Australian manual. VIC, Australia: The Australian Council for Educational Research Ltd. Dykens, E. M., Rosner, B. A., & Tran, M. L. (2000). Drawings by individuals with Williams syndrome: Are people different from shapes? American Journal on Mental Retardation, 106(1), 94 107. Freeman, N. (1987). Current problems in the development of representational picture-production. Archives de Psychologie, 55, 127152. Frith, U. (1989). Autism: Explaining the enigma. London: Blackwell. Goldstein, G., Beers, S. R., Siegel, D. J., & Minshew, N. J. (2001). A comparison of WAIS-R proles in adults with high-functioning autism or differing subtypes of learning disability. Applied Neuropsychology, 8, 148154. Golomb, C., & Barr-Grossman, T. (1977). Representational development of the human gure in familial retardates. Genetic Psychology Monographs, 95, 247266. Goodenough, F. L. (1926). The measurement of intelligence by drawings. New York: World Books. Harris, D. B. (1963). Childrens drawings as measures of intellectual maturity: A revision and extension of the Goodenough Draw-aMan test. New York: Harcourt, Brace and World. Kellman, J. (2001). Autism, art, and children. Westport, CT: Bergin & Garvey. Koppitz, E. M. (1968). Psychological evaluation of childrens human gure drawings. London: Grune & Stratton. Koppitz, E. M. (1983). Projective drawings with children and adolescents. School Psychology Review, 12, 421427. Lee, A., & Hobson, R. P. (2006). Drawing self and others: How do children with autism differ from those with learning difculties? British Journal of Developmental Psychology, 24, 547565.

123

994 Lewis, V., & Boucher, J. (1991). Skill, content and generative strategies in autistic childrens drawings. British Journal of Developmental Psychology, 9, 393416. Machover, K. (1949). Personality projection in the drawing of the human gure. Springeld: C.C. Thomas. Motta, R., Little, S., & Tobin, M. (1993). The use and abuse of human gure drawings. School Psychology Quarterly, 8, 162169. Naglieri, J. (1988). Draw a person: A quantitative scoring system. San Antonio, TX: The Psychological Corporation. Sparrow, S., Balla, D., & Cicchetti, D. (1984). The vineland adaptive behavior scalesInterview edition, survey form manual. Circle Pines, MN: American Guidance Service.

J Autism Dev Disord (2008) 38:988994 Swettenham, J., Baron-Cohen, S., Charman, T., Cox, A., Baird, G., & Drew, A. (1998). The frequency and distribution of spontaneous attention shifts between social and nonsocial stimuli in autistic, typically developing, and nonautistic developmentally delayed infants. Journal of Child Psychology and Psychiatry, 39(5), 747 753. Tager-Flusberg, H., Boshat, J., & Baron-Cohen, S. (1998). Reading the windows to the soul: Evidence of domain-specic sparing in Williams syndrome. Journal of Cognitive Neuroscience, 10, 631639.

123