RASD-90; No of Pages 14

Research in Autism Spectrum Disorders xxx (2008) xxx–xxx

Contents lists available at ScienceDirect

Research in Autism Spectrum Disorders Journal homepage: http://ees.elsevier.com/RASD/default.asp

Non-superior disembedding performance in children with high-functioning autism and its cognitive style account Fei Chen a,*, Eric Lemonnier a, Alain Lazartigues a, Pascale Planche b a b

Service de Pe´dopsychiatrie, CHU de Brest, Hoˆpital de Bohars, BP17, 29820 Bohars, France De´partement de Psychologie, Faculte´ des Lettres et Sciences Sociales, Universite´ de Bretagne Occidentale, 29200 Brest, France

A R T I C L E I N F O

A B S T R A C T

Article history: Received 29 February 2008 Accepted 4 March 2008

Some early studies showed a superior disembedding performance in autistic people while other studies found no difference between autistic and controls. The present study aimed to assess such disembedding ability in 14 boys with high-functioning autism (HFA) and 14 chronological age and non-verbal IQ matched typically developed boys using an Informatized Kohs’ Cube Test (‘‘Samuel’’) and a modified Children’s embedded figures test (CEFT). No statistically significant group difference was found between paired subjects (although more control subjects succeeded in the Samuel Test) and the HFA subjects showed in the Samuel Test as much ‘‘flexibility’’ in strategy adoption as the control subjects, which is not in accordance with some early reports of superior visuo-spatial performance nor with the classical weak coherence theory. Results in the present study, the discrepancies in early findings as well as the symptomatic and cognitive heterogeneity of autism are discussed in the light of cognitive style account. ß 2008 Elsevier Ltd All rights reserved.

Keywords: High-functioning autism Disembedding performance CEFT Kohs’ Cube Test Cognitive style

1. Introduction Autistic people are ‘‘well known’’ to have some preserved skills such as the superior performance in the embedded figures test (EFT; Witkin, Oltman, Raskin, & Karp, 1971) and the block design subtest (BDT). They have been found to be either faster or more accurate in these tests than mental age

* Corresponding author. Present address: Laboratory of Prof. Hadjikhani, EPFL, SV-BMI-GRHAD AAB 023, Station 15, CH-1015 Lausanne, Switzerland. Tel.: +41 216931806. E-mail address: [email protected] (F. Chen). 1750-9467/$ – see front matter ß 2008 Elsevier Ltd All rights reserved. doi:10.1016/j.rasd.2008.03.003

Please cite this article in press as: Chen, F, et al., Non-superior disembedding performance in children with high-functioning autism and its cognitive style account, Res Autism Spectr Disord (2008), doi:10.1016/j.rasd.2008.03.003

RASD-90; No of Pages 14

2

F. Chen et al. / Research in Autism Spectrum Disorders xxx (2008) xxx–xxx

matched or/and chronological age matched non-autistic people do (De Jonge, Kemner, & Van Engeland, 2006; Jolliffe & Baron-Cohen, 1997; Ropar & Mitchell, 2001; Shah & Frith, 1983) while some other studies found no group difference (Brian & Bryson, 1996; Kaland, Mortensen, & Smith, 2007). How to explain such a wide behavioral variation in autistic people? The autistic ‘‘cognitive style’’ account (Happe, 1999) seems to be a very interesting and illuminating approach. Cognitive style, ‘‘can be defined broadly as a preferred approach to problem solving that characterizes an individual’s behavior across a variety of situations and content domains’’ (Waber, 1989). Happe´, who has initially proposed the autistic ‘‘cognitive style’’ hypothesis, suggested that the ‘‘weak central coherence theory’’ (Frith, 1989), one of the main accounts for autism, which could ‘‘explain patterns of excellent and poor performance with one cognitive postulate’’, seems to reveal rather a ‘‘cognitive style than a cognitive deficit’’ (Happe, 1999). That is, the autistic superiority in local processing revealed in some studies ‘‘may be a processing bias, rather than a deficit’’ (Happe & Frith, 2006). The global processing that was originally suggested to be deficient by ‘‘weak central coherence theory’’ has been subsequently found to be intact in some autistic individuals by the Navon Hierarchical Figures task (Mottron, Burack, Iarocci, Belleville, & Enns, 2003). Another study (Rondan & Deruelle, 2007) suggested a configural than a global processing deficit for people with autism and Asperger’s syndrome (AS). All these results demonstrated that the visuo-spatial processing and the perceptive organization might be fairly ‘‘contextualized’’ in people with autism. Thus, the cognitive style account seems to reveal how their visuo-spatial resources are organized for problem solving. The question became could we define an autistic cognitive style? From the point of view of the differential psychology, this superior ability in EFT represents a tendency of Witkin’s ‘‘field independent (FI)’’ cognitive style (Witkin et al., 1954). Based on Werner’s organismic theory of development, the field-dependence–independence is one of the ‘‘most prominent and well-researched dimension of cognitive style’’ (Waber, 1989). This theory has been used to reveal the individual difference on their perceptual organization in visual contexts with embedded targets (the EFT) and conflicting visual cues (the EFT, the Rod and Frame Test, etc.). Thus, two different styles have been defined: the field dependent style and the field independent style. Compared to field dependent participants, those defined as field independent were found to be more analytic in perceptive behavior, less distracted by the confusing content field and more competent in restructuring skills. However, it has been argued that it is not appropriate ‘‘to equate FI with weak central coherence’’ since ‘‘FI people are conceptualized as succeeding on EFT because of their ability to see, but resist, the gestalt’’, while ‘‘people with weak coherence are postulated to be good at this test precisely because they do not spontaneously attend to the gestalt, instead seeing the figure first in terms of its parts’’ (Happe & Frith, 2006). There is also evidence that some children with autism are capable of segmenting a complex image and restructuring conjoint elements by mobilizing some particular strategies (Planche, Lemonnier, Moalic, Labous, & Lazartigues, 2002). Hence, there seems to be some limitations in the application of field-dependence–independence theory for accounting the autistic perceptive behaviors. So, could there be any variant between the field independent and the field dependent cognitive styles? Pascual-Leone (1989) has proposed a variant that he called FM (for ‘‘Mobile FI’’) cognitive style containing a ‘‘rich perceptual learning’’ as well as ‘‘a strong executive control’’. This more flexible style seems to bridge the continuity between the two original ‘‘extreme’’ cognitive styles of Witkin’s theory. It has been also under such consideration that three cognitive strategies (analytic, synthetic and global) have been proposed to differentiate individual performance and behavioral specificities in the Kohs’ Cube Test (in which people with autism have also shown some preserved skills). The author suggested: people who adopt mostly an analytic strategy show a good segmentation and anticipation ability in completing this task, following a linear order; people adopt mostly a synthetic strategy will also show a good segmentation and anticipation ability while their completion of task will follow the gestalt order; people with a global strategy (not efficient in this task) show deficiency in segmentation and anticipation and their completion of task is often characterized by trials and errors following neither the linear order nor the gestalt order (Rozencwajg, Corroyer, & Altman, 2002). Could people with autism show one or more of these cognitive strategies? The present study aimed to investigate the cognitive style account in HFA. Fourteen highfunctioning autistic boys and 14 boys with typical development aged from 8 to 12 years were tested Please cite this article in press as: Chen, F, et al., Non-superior disembedding performance in children with high-functioning autism and its cognitive style account, Res Autism Spectr Disord (2008), doi:10.1016/j.rasd.2008.03.003

RASD-90; No of Pages 14

F. Chen et al. / Research in Autism Spectrum Disorders xxx (2008) xxx–xxx

3

using the Children’s embedded figures test (CEFT) and an Informatized Kohs’ Cube Test (Samuel). The CEFT requires a disembedding operation while the Samuel is a restructuring task (in which a disembedding operation could be an advantage): both are related to the characteristic definition of cognitive style. We wanted to verify firstly whether or not our high-functioning autistic participants will perform as well as persons in the control group in these tests and secondly in which way they would perform these tests (could they show one or more of these above-mentioned cognitive strategies – synthetic, analytic and global – in these tests?). Could we define a unique autistic cognitive style for these individuals? 2. Participants Participants in the HFA group were recruited from the Regional Autism Center of Brittany in the Department of Child and Adolescent Psychiatry, University Hospital of Brest (CRA de Brest, Service de Pe´dopsychiatrie, CHU de Brest). All of these HFA children have been assessed by one or two psychiatrists on ADI-R (Lord, Rutter, & Le Couteur) and have met the ICD-10 and DSM-IV criteria for autistic disorder. Five of these individuals had no ‘‘delay’’ in early language development thus might have met criteria for AS, while their social and communicative impairments as well as their restricted and repetitive behaviors and interests were in accordance with the DSM-IV diagnostic of autistic disorder. Since there is no consensus to date for separating AS people from HFA ones (see Mayes, Calhoun, & Crites, 2001), our clinical group included both (see Table 1). All (but one: subject A4) of these clinical participants had a WISC-III full-scale IQ equal to or above 70 (range: 71–145). A4 was an 8.5 years old autistic boy with a WISC-III full-scale IQ of 68, near to our criteria for HFA and he has a superior non-verbal score in the Raven’s Test (situates at 90th percentile), so we decided to include him in our HFA group. The mean full-scale IQ of HFA group was 96.36 (S.D. = 21.85). Participants in the control group were recruited from an ordinary school in Brest. None of the participants had known history of neurological or psychiatric disorders or developmental retardation. None of their first-degree family members had a pervasive developmental disorder. They were matched individually with the HFA group on chronological age and on non-verbal psychometric measure (Raven’s Test). As shown in Table 2, there was no significant difference between the control group and the HFA group on these measures. All of our participants had normal or corrected to normal vision. Informed and written consents were obtained from their parents. Note that our participants were limited to males in order to exclude female participants who are much rarer in HFA (even rarer in AS) and may represent a different phenotype from male subjects in visuo-spatial performance.

Table 1 Clinical aspects of HFA group Subject

Chronological age

ADI-1

ADI-2

ADI-3

ADI-4

Diagnostic

Al A2 A3 A4 A3 A6 A7 A5 A9 A10 A11 A12 A13 A14

7; 11 8; 06 8; 03 8; 04 9; 09 9; 03 9; 08 9; 08 10; 02 10; 03 10; 07 10; 09 10; 11 11; 02

21 17 14 17 15 14 11 15 12 17 15 14 24 10

16 (6) /(8) 11 (5) 20 (11) 8 (2) 10 (4) 15 (9) 11 (5) 14 (6) 14 (6) 16 (11) 14 (7) 15 (10) 16 (7)

6 5 3 4 3 4 4 6 6 3 4 5 3 6

4 5 1 3 5 3 3 3 5 1 2 3 3 5

Autistic disorder Autistic disorder Autistic disorder Asperger’s syndrome Autistic disorder Asperger’s syndrome Asperger’s syndrome Asperger’s syndrome Autistic disorder Autistic disorder Autistic disorder Asperger’s syndrome Autistic disorder Autistic disorder

ADI-1, -2, -3 and -4 represent total algorithm scores of ADI-R. The cut-off point is respectively 10, 8 (7), 3 and 1, for ‘‘anomalies in reciprocal social interaction’’, for deficits in ‘‘verbal communication (non-verbal communication)’’, for ‘‘Restricted and Repetitive Behavior’’ and for ‘‘anomalies of development by 3 years old’’.

Please cite this article in press as: Chen, F, et al., Non-superior disembedding performance in children with high-functioning autism and its cognitive style account, Res Autism Spectr Disord (2008), doi:10.1016/j.rasd.2008.03.003

RASD-90; No of Pages 14

F. Chen et al. / Research in Autism Spectrum Disorders xxx (2008) xxx–xxx

4 Table 2 Group pairing Group

N

Chronological age

Score in Raven’s Test

HFA Mean (S.D.) Range

14

9; 07(1; 01) 7; 11–11; 02

31.93 (2.92) 25–35

Control Mean (S.D.) Range

14

9; 05 (0; 11) 8; 0–10; 09

32.00 (2.63) 25–35

p value a

Wilcoxon matched pairs signed-ranks test.

b

Matched pairs t-test.

0.091a

0.850b

3. Study 1: CEFT 3.1. Materials and procedure In this study, we used the original CEFT colorful materials in which participants should search for a ‘‘Tent’’ form in 11 complex figures and a ‘‘House’’ form in 14 complex figures. We did a modification in the ‘‘Tent’’ series: we did not introduce the notion of ‘‘Tent’’ but just asked subjects to search in complex figures for this ‘‘triangle form’’ with the same shape, size and orientation as the model. We supposed that the verbal term ‘‘triangle’’ contained a semantically simpler and clearer target to search for, which may lead to a similar performance for both groups, while the verbal term ‘‘House’’ contained a more semantically (than geometrically) salient ‘‘target in memory’’ which may lead to a performance discrepancy between the figures containing a house context and others non-containing a house context, more significant in control group if autistic people were less ‘‘contamined’’ by the meaning of context as suggested by the classical weak central coherence theory, or to be equal in both groups if autistic people were also influenced by the meaning of context as much as people with a typical development. The presentation of simple forms, discrimination series, demonstration series, practice series and test series were preceded in original order (Witkin et al., 1971) except that we asked all our participants to complete all the items. As described in the original manual, responses were scored ‘‘1 or 0’’ and a score was given ‘‘only when the first choice is correct and verified’’ or when ‘‘an incorrect choice is spontaneously corrected before the child sees the cut-out model’’. We adopted an ‘‘open’’ procedure (no time limit) as in the original test because many children in our clinical group were very sensible to the utilization of a stopwatch that could make them disconcentrated and destabilized from the test (as they have shown during other clinical evaluation). However, none of our participants (in both groups) had surpassed 1 h for completing the whole test, that is to say, the mean response time for each figure was less than 180 s (which has been used as time limit in some studies). Data were analyzed using the Statistical Package for the Social Sciences Program (SPSS Version 14.0). The t-test was used to analyze within-group differences of task resolution ratio between Triangle condition and House condition, the between-group differences of scores in each condition and the between-group difference of the total CEFT scores. Pearson’s r was used to examine correlations between the total CEFT scores and other parameters: chronological age (for both groups); the full-scale IQ, the ‘‘Picture completion’’ and the ‘‘Object assembly’’ subtest scores of WISC-III, the ADI-R scores (for HFA group only). 4. Results As seen in Table 3, our clinical group got a higher score in the ‘‘Triangle (Tent)’’ condition while control group got a higher score in the ‘‘House’’ condition, but these differences were not statistically significant ( p = 0.151 and p = 0.336, respectively). No significant difference ( p = 0.762) was found for the total score between the two groups. ‘‘House’’ condition seemed to be more difficult than the Please cite this article in press as: Chen, F, et al., Non-superior disembedding performance in children with high-functioning autism and its cognitive style account, Res Autism Spectr Disord (2008), doi:10.1016/j.rasd.2008.03.003

RASD-90; No of Pages 14

F. Chen et al. / Research in Autism Spectrum Disorders xxx (2008) xxx–xxx

5

Table 3 Results on the CEFT HFA group and the control group CEFT

HFA group

Control group

p value of t-test

Triangle condition Score (S.D.) % correct

10 (1.18) 90.91

9.5 (0.85) 86.37

t = 1.528, p = 0.151 *

House condition Score (S.D.) % correct

10.1 (2.64) 71.94

10.9(2.13) 78.06

t=

1.104, p = 0.290**

Total score (S.D.)

20.1 (2.81)

20.4(2.47)

t=

0.389, p = 0.703***

*

Difference between groups in triangle condition.

**

Difference between groups in house condition.

***

Difference between groups for total CEFT score. The within-group differences of resolution ratio between Triangle condition and House condition were also conducted. The p value was 0.007 ( p < 0.01) for HFA group and 0.064 (NS) for the control group.

‘‘Triangle (Tent)’’ condition in both groups ( p = 0.007 for the HFA group and p = 0.064 for the control group). We compared the ratio of errors (see in Fig. 1) for each figure of both conditions (‘‘Triangle’’ and ‘‘House’’) which showed a similar ratio for most figures except that the control group made apparently more errors in Figs. 7–10 in ‘‘Triangle’’ series while the HFA group made more errors in Figs. 7, 8 and 11 in ‘‘House’’ series. The correlational analyses (Pearson’s r) showed a positive association between scores of Raven’s Test and the total CEFT scores (r = 0.721, p = 0.004) in the control group. This result suggests that a better disembedding ability is related to a better non-verbal reasoning ability in our control subjects. We found no statistically significant correlation between the chronological age and the total CEFT scores for both groups. For the HFA group, we failed to find any statistically significant correlation between scores of Raven’s Test and the total CEFT scores, between the full-scale IQ and the total CEFT scores (as well as some subtests concerning disembedding and restructuring performance: between the total CEFT scores and the ‘‘Picture completion’’ scores of WISC-III, between the total CEFT scores

Fig. 1. Ratio of errors in CEFT, HFA group vs. control group, Triangle (Tent) series (from figure T1 to T11), House series (from figure Hl to H14).

Please cite this article in press as: Chen, F, et al., Non-superior disembedding performance in children with high-functioning autism and its cognitive style account, Res Autism Spectr Disord (2008), doi:10.1016/j.rasd.2008.03.003

RASD-90; No of Pages 14

6

F. Chen et al. / Research in Autism Spectrum Disorders xxx (2008) xxx–xxx

and the ‘‘Object assembly’’ scores of WISC-III), and between the total CEFT scores and the ADI-R scores. These results suggest that the development of disembedding and restructuring abilities in autistic persons might be an issue more complex than in control children with typical development.

5. Study 2: Samuel (an Informatized Kohs’ Cube Test) 5.1. Materials and procedure ‘‘Samuel’’, which consists totally 10 figures (six simple figures of 4-blocks and four complex figures of 9-blocks), is an informatized variant of Kohs’ Cube Test devised by Rozencwajg et al. (2002). One of the main differences between the classical version of Cube Test and this informatized one is the absence of continuous vision of the target figure, so that more working memory resources were needed in the present study compared to the classical version. The same participants (14 pairs) participated in this test. Stimuli were presented on a 17-in. computer screen. Sat in front of the computer, subjects should use the mouse to process throughout this test. To be familiar with the test processing way, every subject received a training session (consists of four different amounts of colorful flower-series in the place of cubes). All of our participants had understood the processing way in the training session. The photo shown depicts the test session procedure of Samuel.

 Firstly, participants should use the mouse to click on a button that is shown on the upper left corner of the screen to see the target figure. The button is marked with words ‘‘See the model’’. But once the click is released, the target figure will disappear. If the subject wants to see the figure again, he just needs to click again.  Then he should choose from six blocks (four red-white two-color blocks, one red block and one white block) on the left foot corner of the screen and move the chosen blocks one by one into a frame on the right part of the screen to construct the target figure.  A wrong displace can be removed by a simple clicking on.  When a target figure has been constructed, subject can click on a button marked with the words ‘‘I have finished!’’ on the upper right corner of the screen to begin with the next one. ‘‘Samuel’’ test registers automatically behavioral data (for example, frequency of consulting the model, mean time in consulting the model, total time in consulting the model, resolution time, segmentation index, index of construction following the Gestalt order, etc.). This automatic recording by the computer begins with every first click on the target figure. From these behavioral data, their dominant cognitive strategy for each 9-blocks figure and for the whole 9-blocks condition can be ruled out by the computer. As has been mentioned in the introduction: an analytic strategy would be defined by a good segmentation and anticipation ability in processing this task following a linear order; a synthetic strategy will show also a good segmentation and anticipation ability while their task resolution will Please cite this article in press as: Chen, F, et al., Non-superior disembedding performance in children with high-functioning autism and its cognitive style account, Res Autism Spectr Disord (2008), doi:10.1016/j.rasd.2008.03.003

RASD-90; No of Pages 14

F. Chen et al. / Research in Autism Spectrum Disorders xxx (2008) xxx–xxx

7

Table 4 Behavioral results of ‘‘Samuel’’ in 8 paired succeeded HFA vs. control subjects (4-blocks condition and 9-blocks condition) Variable

HFA vs. control 4-Blocks condition Group mean (S.D.)

Fa

0.4325(0.174) 0.4625 (0.201)

9-Blocks condition p value 0.796 (t-test)

Group mean (S.D.) 0.449 (0.182) 0.488 (0.193)

p value 0.732 (t-test)

T (total-see)b

18.125(8.408) 16.875 (14.759)

0.313 (Wilcoxcon)

63.000 (32.654) 45.625 (44.877)

0.313 (Wilcoxon)

T (mean-see)c

4.125 (1.458) 3.875 (2.167)

0.383 (Wilcoxcon)

6.375 (2.669) 4.500 (2.879)

0.250 (Wilcoxon)

T (completion)d

68.013 (41.692) 57.675 (21.217)

0.576 (t-test)

Segmentation index

0.880 (0.106) 0.864 (0.232)

0.864 (t-test)

0.790 (0.233) 0.841 (0.134)

0.654 (t-test)

Anticipation index

0.779 (0.166) 0.746 (0.178)

0.744 (t-test)

0.614 (0.238) 0.683 (0.124)

0.501 (t-test)

Completion following linear orders

–

–

0.428 (0.115) 0.395 (0.124)

0.626 (t-test)

Completion following Gestalt order

–

–

0.690 (0.152) 0.543 (0.146)

0.013 (t-test)

a

176.750 (73.420) 143.488 (53.835)

0.385 (t-test)

Frequency of consulting the model.

b

Total time (s) of consulting the model.

c

Mean time (s) for every consulting of the model.

d

Total resolution time (s).

follow the gestalt order; a global strategy (not efficient in this task) would be defined by deficiency in segmentation and anticipation and by a characteristic ‘‘trials and errors’’ style of task resolution following neither the linear order nor the gestalt order. The same statistical method was contacted for the Samuel Test as described above for the CEFT. Group differences in behavioral indices between HFA group and control group were analyzed using the t-test. Differences in total time and mean time of model consulting (for the 4-blocks and the 9blocks conditions) between HFA group and control group were analyzed non-parametrically using Wilcoxon’s test, due to the uneven distribution of the data. Pearson’s r was used to examine correlations between Samuel behavioral indices and the total CEFT scores, chronological age, ADI scores (for HFA group only). 6. Results Nine participants in the HFA group (64.3%) successfully completed this test. Of the 5 participants (A1, A4, A5, A7 and A10) who failed to finish this test, 4 had given up when encountering difficulties in complex figures (e.g. 9-blocks), one had made a mistake in manipulating the mouse that had led to a loss of his data. Moreover, we lost the data of one successful participant (A2) because of unexpected technical problem. Thus, analyzable data was available for 8 participants from the HFA group.1 In the control group, 12 participants (85.7%) had successfully finished this test. Behavioral results of the 8 HFA–control matched pairs revealed, as shown in Table 4, no statistically significant difference between these two groups except the variable ‘‘construction following the 1 As in the study of Kaland et al. (2007), some of our HFA subjects were also very sensitive to, and afraid of, test failure and they gave up easily during the test. Some of them showed ‘‘motor clumsiness’’ in mouse manipulation. Besides, the absence of continuous vision of the target figures in our informatized cube test (Samuel Test) requests more working memory resources. All of these may partly explain why there was more control subjects succeeded in the Samuel Test.

Please cite this article in press as: Chen, F, et al., Non-superior disembedding performance in children with high-functioning autism and its cognitive style account, Res Autism Spectr Disord (2008), doi:10.1016/j.rasd.2008.03.003

RASD-90; No of Pages 14

F. Chen et al. / Research in Autism Spectrum Disorders xxx (2008) xxx–xxx

8

Table 5 Dominant cognitive strategies (by subject and by figure) in Samuel (‘‘S’’: synthetic; ‘‘A: analytic; ‘‘G’’: global; N* = sum of strategies recruited in the 9-blocks condition) Subject

Fig. 7

Fig. 8

Fig. 10

Dominant

N*

S G G A G S G G

S G S A S G G S

2 1 2 2 2 2 2 2

Paired succeeded controls (mean ratio in the 8 controls = 17/32 (items) = 53.13%) T3 S G S G T6 G G G G T8 A G A A T9 G G S G T11 S S S G T12 A A G A T13 S G A A T14 A G S A

S G G G S G A G

2 1 2 2 2 2 3 3

Other succeeded controls (mean ratio in the total of 12 controls = 26/48 (items) = 54.17%) T1 S G S G T2 S G A G T4 G G S G T5 S G S G

S S G S

2 3 2 2

Paired succeeded HFAs (mean ratio in the 8 HFA = 15/32 A3 S G A6 G G A8 S G A9 A A A 11 S G A 12 S G A 13 G S A 14 S G

Fig. 9 (items) = 46.88%) S G S S S G G S

For the sum of strategies recruited in the 9-block condition, no significant difference was found neither between paired succeeded (8 subjects) HFA and control group ( p = 0.471, Wicoxon) nor between paired succeeded HFA group (8 subjects) and total succeeded control group (12 subjects) ( p = 0.360, Wilcoxon).

gestalt order’’ in 9-blocks condition which was significantly higher in the HFA group than in the control group ( p < 0.05). However, when we compared these 8 HFA subjects with all the 12 control subjects who had succeeded in Samuel, difference became non-significant (t = 1.246, p = 0.229). Our HFA subjects seemed to be slower than control subjects in completing both the 4-blocks and the 9blocks condition of this test (as indexed by the ‘‘resolution time’’ variable in Table 4) but these differences were not statistically significant ( p = 0.576 and p = 0.385, respectively). An advantage of this informatized test is that the dominant cognitive strategy (‘‘S’’: synthetic; ‘‘A’’: analytic; ‘‘G’’: global) for each figure in the 9-blocks condition can be analyzed according to their behavioral variables. As shown in Table 5, most subjects in HFA group recruited also various strategies according to the characteristic of each item as subjects in control group did. We calculated the total sum of strategies recruited in the 9-blocks condition, no significant difference has been found, neither between paired succeeded (8 participants) HFA and control group ( p = 0.471, Wilcoxon), nor between succeeded HFA group (8 participants) and total succeeded control group (12 participants) ( p = 0.360, Wilcoxon). We compared also the constituent ratio of dominant strategy between HFA group (8 participants) and control group (12 subjects): a similar ‘‘global–local’’ processing ‘‘preference’’, we mean a ‘‘distribution’’ between Global and (Synthetic or Analytic), was observed in these two groups. Participants in our HFA group seemed to have a tendency to recruit a little more frequently synthetic strategy than subjects in control group did. The correlational analyses (Pearson’s r) showed (1) in the control group a negative association between the total CEFT scores and the total resolution time in 9-blocks condition of Samuel (r = 0.786, p < 0.01), a positive association between the total CEFT scores and the variable ‘‘anticipation index’’ in 9-blocks condition of Samuel (r = 0.582, p = 0.047), a negative association between the chronological age and the variable ‘‘construction following the gestalt order’’ in 9-blocks condition (r = 0.603, p < 0.05), a positive association between the chronological age and the variable ‘‘anticipation index’’ in 9-blocks condition (r = 0.861, p < 0.001) and a negative association between Please cite this article in press as: Chen, F, et al., Non-superior disembedding performance in children with high-functioning autism and its cognitive style account, Res Autism Spectr Disord (2008), doi:10.1016/j.rasd.2008.03.003

RASD-90; No of Pages 14

F. Chen et al. / Research in Autism Spectrum Disorders xxx (2008) xxx–xxx

9

the variable ‘‘anticipation index’’ and the total resolution time in 9-blocks condition (r = 0.620, p = 0.031). These results suggest for our controls that a better disembedding ability (CEFT) is related to a better anticipation and resolution in rather difficult condition (9-blocks) of Samuel Test. Elder control subjects are less dependent on the Gestalt context of this task and have better anticipation capacity which can facilitate the task resolution. While in HFA group, we found a negative association between the chronological age and the total resolution time in 4-blocks condition (r = 0.738, p < 0.05), a negative association between the variable ‘‘anticipation index’’ and the total resolution time in 4-blocks condition (r = 0.763, p = 0.028) and a negative association between the variable ‘‘segmentation index’’ and the total resolution time in 4-blocks condition (r = 0.845, p = 0.008). These results suggest that in our HFA group, older participants are faster in task resolution and that anticipation and segmentation abilities can facilitate the task resolution, but only in rather easy condition (4-blocks). We failed to find any statistically significant correlation between the CEFT scores and Samuel behavioral indices for our HFA group. Furthermore, we found no correlation for both groups, between the chronological age and the variable ‘‘segmentation index’’ (4-blocks and 9-blocks conditions), between the variable ‘‘segmentation index’’ (9-blocks condition) and the variable ‘‘construction following the gestalt order’’ (9-blocks condition). 7. Conclusion and discussion In the present study, the disembedding performance was assessed in a modified version of two classical tests in which many studies had revealed a superior performance in autistic people. In accordance with the finding of Kaland et al. (2007), our data also failed to show such performance superiority in paired HFA boys compared to chronological age and non-verbal IQ matched controls (8 pairs in the Samuel Test and 14 pairs in the CEFT). The modification we did in the ‘‘Triangle (Tent)’’ series of CEFT (introducing a semantically simpler target to search for) could be a factor, in our opinion, to make the inter-group performance discrepancy be smaller in ‘‘Triangle (Tent)’’ series than in ‘‘House’’ series, if the semantic (or contextual) factor did not give as much impact to autistic subjects as to control subjects (as suggested by the classical weak coherence hypothesis). However, we found rather a slightly greater inter-group discrepancy in ‘‘Triangle (Tent)’’ series ( p = 0.151) than in ‘‘House’’ series ( p = 0.290). Besides, our data of errors’ analysis did not support the hypothesis that autistic people were less ‘‘contamined’’ by the contextual meaning, just like the ‘‘susceptibility to illusions’’ revealed by Ropar and Mitchell (2001): our HFA subjects made more errors than control subjects in ‘‘House’’ series (Figs. 7, 8 and 11). It may not be surprising that our autistic subjects made more errors than control subjects in item 11 (a fairly difficult one that contains not only global Gestalt distractor but also a lot of swimmy points as texture distractor), while more errors in item 8 which contains, we suppose, not a complicated but a ‘‘misleading’’ house context, could be interpreted by their sensibility to the contextual meaning (the Gestalt form), which does not seem to be in accordance with the classical weak coherence theory of autism. However, the nature of the errors could be complicated for every subject and for every item. Our modification to CEFT did not seem to have changed apparently the total performance of autistic participants (% correct: 80.4), which is very similar to the result of Kaland et al. (2007), while their control subjects had a better performance (92%) than ours (81.6%). However, their controls were much older (mean age = 15:9) than ours (mean age = 9:5) and it is known that better disembedding ability can be observed with increasing age (Witkin et al., 1971). This had been proposed in terms of an evolutional change tendency in cognitive style, and is partly in accordance with the negative association that we had observed between the chronological age and the variable ‘‘construction following the gestalt order’’ in 9-blocks condition of Samuel Test in our control group, which suggests that in our control group, older subjects ‘‘benefit’’ less from the Gestalt form (more ‘‘independent’’). We did not observe such correlation in our HFA group. It has been proposed that visuo-spatial peaks are a developmental ongoing process in autism (Joseph, Tager-Flusberg, & Lord, 2002), while another study showed that children with autism have better performance than controls on the EFT only at younger ages (Edgin & Pennington, 2005). However, the mean ages of clinical groups in the first study were 5:5 and 8:11 (years:months), while the mean age of clinical group in the second study was 11.43 Please cite this article in press as: Chen, F, et al., Non-superior disembedding performance in children with high-functioning autism and its cognitive style account, Res Autism Spectr Disord (2008), doi:10.1016/j.rasd.2008.03.003

RASD-90; No of Pages 14

10

F. Chen et al. / Research in Autism Spectrum Disorders xxx (2008) xxx–xxx

years. It is possible that a proportion of autistic children had some early superior perceptual abilities as proposed by Mottron and Burack (2001) in terms of neurobiological compensation, which could, in our opinion, contribute to, or interact with, an earlier establishment of specific cognitive styles (some of them could show fairly efficient abilities in some special daily-life behaviors – for example, in spatial memory – to ‘‘live with’’ their cognitive specificity), while this early superior performance in autistic children could be ‘‘caught up’’ by their non-autistic pairs in a later life stage with a more rapid cognitive development (Edgin & Pennington, 2005). Although some studies have found differences between autistic persons and controls when looking at performance time rather than accuracy in EFT or CEFT (De Jonge et al., 2006; Jolliffe & Baron-Cohen, 1997), we did not register the performance time in the present study for two reasons: firstly, as abovementioned, most of our subjects are sensitive and ‘‘distractible’’ autistic children, so the noise of a stopwatch or our action of time recording could easily slow down their task processing (some children would stop the test and ask: ‘‘what’s happened? What did you do? Did I make a mistake?’’ and so on. . .); secondly, this study aimed mainly to characterize the autistic cognitive style in these two tests. The idea was to give them enough time to perform at their best and to show their dominant strategy (cognitive style), so we did not want to impose a time limit upon our participants. In this way they would not be ‘‘forced’’ to use a particular cognitive strategy because of time limit. Another interesting finding was the performance heterogeneity and ‘‘strategic flexibility’’ of our HFA group. They adopted as many strategies in different items of Samuel Test as our controls did. Our HFA participants showed strategy changes in different conditions of the Samuel Test (from analytic or synthetic dominant strategy to global dominant strategy) similar to control subjects (and also similar to the standard population presented in ‘‘Samuel’’ test manual) when the task turned more difficult (in Figs. 8 and 10 compared to in Figs. 7 and 9 of the Samuel Test) (see in Fig. 2). This is in accordance with the finding of Baillargeon, Pascual-Leone, and Roncadin (1998) in a sample of 239 school-age children with typical development, suggesting that the ‘‘contextual perception driven strategy’’ becomes dominant when the mental-attentional demand is above the mental-attentional capacity. Overall, our informatized Samuel Test succeeded to reveal some cognitive strategic nature more than the behaviorally ‘‘no difference’’ result and confirmed that each of the three above-mentioned strategies can be recruited by our HFA children. They might not be as ‘‘rigid’’ in strategy mobilization as we ‘‘thought’’ before, at least in the visuo-spatial domain. Furthermore, we observed a tendency to use more frequently the synthetic strategy in the HFA group than for controls, which is in accordance with the finding of Caron, Mottron, Berthiaume, and Dawson (2006). They observed visuo-spatial peaks in both local and global processing for high-functioning autism (HFA) persons. Their result also showed that only up to half of their autistic participants evinced a visuo-spatial peak (Caron et al., 2006) which suggests a phenomena of the heterogeneity in autistic population, even in the visuo-spatial domain that was not only limited to able autistic participants. More studies are needed on this interesting path since such heterogeneity in visuo-spatial peak may lead to some difference in individual treatment consideration. The fact that only 1 out of 8 HFA subjects had adopted an analytic dominant strategy in Samuel Test suggests that the theory of a global processing impairment needs to be re-examined. The hypothesis of an autistic configural deficit also seems to be plausible in our HFA subjects. The configural operation is needed in restructuring processing in Samuel Test but 9 out of 14 autistic participants still succeeded throughout the test and they did not show a general tendency to focus more on details (analytic) when processing complicated configurations as has been suggested by Rondan and Deruelle (2007). Other 5 autistic subjects encountered most of their difficulties in a rather complex condition (the 9-blocks condition): an experimental condition with more complicated configural relation among stimuli (than in the 4-blocks condition). This suggests perhaps a difficulty in processing complex configural stimuli rather than an absolute deficit in configural processing. Our correlational analyses in HFA group also suggest that visuo-spatial performance amelioration with increasing age might be limited to relatively simpler figures or tasks. This is in accordance with the proposition of Bertone, Mottron, Jelenic, and Faubert (2005) that in autistic people the performance in visuo-spatial processing depends on stimulus complexity. Besides, no statistically significant correlation between ADI-R scores and CEFT scores suggests that disembedding ability may not be a sensible index for the typology of autistic symptoms at least in HFA subjects. Please cite this article in press as: Chen, F, et al., Non-superior disembedding performance in children with high-functioning autism and its cognitive style account, Res Autism Spectr Disord (2008), doi:10.1016/j.rasd.2008.03.003

RASD-90; No of Pages 14

F. Chen et al. / Research in Autism Spectrum Disorders xxx (2008) xxx–xxx

11

Fig. 2. Constituent ratio of dominant strategy in Samuel (8 succeeded HFA vs. 12 succeeded control subjects), (1) generally dominant strategy, (2) dominant strategy changes from Fig. 7 to 8 (Samuel) and (3) dominant strategy changes from Fig. 9 to 10 (Samuel).

We should mention that there was a large IQ range in our HFA group, which may be a factor partly leading to the performance heterogeneity. But it is also the true profile of this population just like the heterogeneity in their clinical symptoms. We have decided to include them in this study also because that the cognitive style has been defined to be ‘‘independent of intellectual competence’’ (Waber, 1989). This has also been shown in our study: the distribution of dominant cognitive strategies in Samuel Test was independent of their full-scale IQ level. For example, participant A13 and participant A14 of the HFA group had both a full-scale IQ of 88 but they showed totally different dominant cognitive strategy adoption in the Samuel Test (global vs. synthetic). On the other side, IQ level may probably limit the flexibility in cognitive strategy adoption because the mental-attentional demand could more easily surpass the mental-attentional capacity in subjects with a relatively lower IQ. Thus the ‘‘contextual perception driven strategy’’ could be more frequently dominant in these participants, according to the above-mentioned proposition of Baillargeon et al. (1998). And it is really the case in Please cite this article in press as: Chen, F, et al., Non-superior disembedding performance in children with high-functioning autism and its cognitive style account, Res Autism Spectr Disord (2008), doi:10.1016/j.rasd.2008.03.003

RASD-90; No of Pages 14

12

F. Chen et al. / Research in Autism Spectrum Disorders xxx (2008) xxx–xxx

our study: Out of totally 8 HFA subjects, all the 3 who adopted a global dominant strategy in Samuel Test had a full-scale IQ level below 90 (71, 85 and 88, respectively). Altogether, these facts suggest that such heterogeneity could have two facets: one is due to the mental capacity in nature that could (but not necessarily) influence the flexibility of cognitive strategy adoption; the other is due to a strategy preference related to autism, which has also been shown in this study to be heterogeneous in HFA subjects. One of the limitations of this study is the small sample size which limits its interpretation of results especially in the Samuel Test with just 8 ‘‘valid’’ participants. Actually, this study was only a preliminary investigation. We tried to introduce such an informatized test for the exploration of cognitive style account in autism and to show that the cognitive strategy which settles the aspect of ‘‘how’’ may be as well interesting as the performance which settles the aspect of ‘‘what’’, which is true not only in the visuo-spatial processing but also in other cognitive dimensions. And in this study, we did have shown in our HFA participants (although the sample size was small) that Gestalt effect and strategy mobilization ‘‘flexibility’’ were not ‘‘absent’’ in some HFA autistic subjects but just depended on their ‘‘cognitive style’’. It seems that we could not define a unique cognitive style for our subjects with HFA. Surely that such ‘‘flexibility’’ of strategy adoption that we reported in the present study in these relatively more able autistic children does not necessarily signify neither a flexibility of strategy adoption in social interaction nor an efficacy of executive function in processing more complicated visual or multimodal information (for example, in processing faces with emotional and social information, in audio–visual integration, etc.). The advantage of recruiting more frequently a synthetic strategy in perceptive tasks may be demonstrated in social interaction as a compensation strategy to categorize and to better anticipate the ‘‘unexpected’’ exterior world. Further studies with larger sample size and longitudinal perspective are needed to reveal the role of the cognitive style account in social cognition and learning procedure in different sub-groups of autism. Another open question is that the similar style of strategy adoption and mobilization in these tests may be achieved via different multifactorial ‘‘mosaic’’ cognitive development. What is the role of development in the evolution of visuo-spatial performance in children with autism? It is possible that the fade of autistic visuo-spatial ‘‘superiority’’ in an elder age reported in some studies is accompanied by an increasing global contextual learning (Rondan & Deruelle, 2007) while children with typical development could ameliorate their performance in these tests with a more and more powerful executive function (inhibition) during their cognitive development (Houde, 2004). A recent study using fMRI (Kana, Keller, Minshew, & Just, 2007) has shown in their adult autism group that the inhibition circuitry ‘‘is activated atypically and is less synchronized’’ which seemed to be accomplished by strategic control rather than automatically. The evolution of general problem solving capacity (as the evolution in visuo-spatial processing) in autistic children may be somewhat characteristic and be helpful to renew the criteria of diagnostic and intervention. And we may also ask: ‘‘Could the autistic processing bias be ‘reconstructed’ by an early intervention in language and in social abilities (and how could be. . .)?’’ From 1999 to present, Happe´’s cognitive style account for autism has given us a very interesting and perhaps also a more appropriate way to understand the discrepancies of findings in this domain of research as well as the symptomatic and cognitive heterogeneity of autism. However, more research is needed to clarify the complex nature of cognitive development in autistic people. A combination of various research methods (Brown, Gruber, Boucher, Rippon, & Brock, 2005; Grice et al., 2001; Goode, Goddard, & Pascual-Leone, 2002; Kana, Keller, Cherkassky, Minshew, & Just, 2006; Ring et al., 1999; Rippon, Brock, Brown, & Boucher, 2007) in future studies may establish us a bridge to better understand the nature of autism. For some of these people, this might be just registered in a continuity of cognitive style variance as you and me.

Acknowledgements Special thanks are given to all the children, especially the HFA (AS) children, who participated in this study as well as their families, to the school directors and teachers who kindly helped us recruit the control subjects and to Josiane Rozec for her invaluable help as well as all the workmates in the Please cite this article in press as: Chen, F, et al., Non-superior disembedding performance in children with high-functioning autism and its cognitive style account, Res Autism Spectr Disord (2008), doi:10.1016/j.rasd.2008.03.003

RASD-90; No of Pages 14

F. Chen et al. / Research in Autism Spectrum Disorders xxx (2008) xxx–xxx

13

Department of Child Psychiatry, University Hospital of Brest (Service de Pe´dopsychiatrie, CHU de Brest). This work was supported by Regional Autism Centre of Brittany (‘‘Centre de ressource pour l’autisme de Bretagne’’). This study has been presented as poster communication at the 8th International Congress Autism Europe (Oslo, August 31–September 2, 2007) and at the 9th Summer School of ARAPI (Le Croisic, France, October 9–13, 2007). Some results will be published in short form in French on Bulletin Scientifique de l’ARAPI (non-indexed publication of the French parents-and-professionals’ association for autism). References Baillargeon, R., Pascual-Leone, J., & Roncadin, C. (1998). Mental-attentional capacity: Does cognitive style make a difference? Journal of Experimental Child Psychology, 70, 143–166. Bertone, A., Mottron, L., Jelenic, P., & Faubert, J. (2005). Enhanced and diminished visuo-spatial information processing in autism depends on stimulus complexity. Brain, 128, 2430–2441. Brian, A. J., & Bryson, S. E. (1996). Disembedding performance and recognition memory in autism/PDD. Journal of Child Psychology and Psychiatry, 37, 865–872. Brown, C., Gruber, T., Boucher, J., Rippon, G., & Brock, J. (2005). Gamma abnormalities during perception of illusory figures in autism. Cortex, 41, 364–376. Caron, M. J., Mottron, L., Berthiaume, C., & Dawson, M. (2006). Cognitive mechanisms, specificity and neural underpinnings of visuospatial peaks in autism. Brain, 129, 1789–1802. De Jonge, M. V., Kemner, C., & Van Engeland, H. (2006). Superior disembedding performance of high-functioning individuals with autism spectrum disorders and their parents: The need for subtle measures. Journal of Autism and Developmental Disorders, 36, 677–683. Edgin, J. O., & Pennington, B. F. (2005). Spatial cognition in autism spectrum disorders: Superior, impaired, or just intact? Journal of Autism and Developmental Disorders, 35, 729–745. Frith, U. (1989). Autism: Explaining the enigma. Oxford: Basil Blackwell. Goode, P. E., Goddard, P. H., & Pascual-Leone, J. (2002). Event-related potentials index cognitive style differences during a serial-order recall task. International Journal of Psychophysiology, 43, 123–140. Grice, S. J., Spratling, M. W., Karmiloff-Smith, A., Halt, H., Csibra, G., De Haan, M., et al. (2001). Disordered visual processing and oscillatory brain activity in autism and Williams syndrome. Neuroreport, 12, 2697–2700. Happe, F. (1999). Autism: Cognitive deficit or cognitive style? Trends in Cognitive Science, 3, 216–222. Happe, F., & Frith, U. (2006). The weak coherence account: Detail-focused cognitive style in autism spectrum disorders. Journal of Autism and Developmental Disorders, 36, 5–25. Houde, O. (2004). La psychologie de l’enfant. Paris: Presses Universitaires de France. Jolliffe, T., & Baron-Cohen, S. (1997). Are people with autism and Asperger syndrome faster than normals on embedded figures test? Journal of Child Psychology and Psychiatry, 38, 527–534. Joseph, R. M., Tager-Flusberg, H., & Lord, C. (2002). Cognitive profiles and social-communicative functioning in children with autism spectrum disorder. Journal of Child Psychology and Psychiatry, 43, 807–821. Kaland, N., Mortensen, E. L., & Smith, L. (2007). Disembedding performance in children and adolescents with Asperger syndrome or high-functioning autism. Autism, 11, 81–92. Kana, R. K., Keller, T. A., Cherkassky, V. L., Minshew, N. J., & Just, M. A. (2006). Sentence comprehension in autism: Thinking in pictures with decreased functional connectivity. Brain, 129, 2484–2493. Kana, R. K., Keller, T. A., Minshew, N. J., & Just, M. A. (2007). Inhibitory control in high-functioning autism: Decreased activation and underconnectivity in inhibition networks. Biological Psychiatry, 62, 198–206. Lord, C., Rutter, M., & Le Couteur, A. (1994). Autism diagnostic interview-revised: A revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders. Journal of Autism and Developmental Disorders, 24, 659–685. Mayes, S. D., Calhoun, S. L., & Crites, D. L. (2001). Does DSM-IV Asperger’s disorder exist? Journal of Abnormal Child Psychology, 29, 263–271. Mottron, L., & Burack, J. A. (2001). Enhanced perceptual functioning in the development of autism. In J. A. Burack, T. Charman, N. Yirmiya, & P. R. Zelazo (Eds.), The development of autism: Perspectives from theory and research (pp. 131–148). New York: Erlbaum. Mottron, L., Burack, J. A., Iarocci, G., Belleville, S., & Enns, J. T. (2003). Locally oriented perception with intact global processing among adolescents with high-functioning autism: Evidence from multiple paradigms. Journal of Child Psychology and Psychiatry, 44, 904–913. Pascual-Leone, J. (1989). An organismic process model of Witkin’s field-dependence–independence. In T. Globerson & T. Zelniker (Eds.), Cognitive style and cognitive development. New Jersey: Ablex Publishing Corporation. Planche, P., Lemonnier, E., Moalic, K., Labous, C., & Lazartigues, A. (2002). Modalities of information processing in autistic children. Annales Me´dico-Psychologique, Revue Psychiatrique, 160, 559–564. Ring, H. A., Baron-Cohen, S., Wheelwright, S., Williams, S. C., Brammer, M., Andrew, C., et al. (1999). Cerebral correlates of preserved cognitive skills in autism: A functional MRI study of embedded figures task performance. Brain, 122, 1305–1315. Rippon, G., Brock, J., Brown, C., & Boucher, J. (2007). Disordered connectivity in the autistic brain: Challenges for the new psychophysiology. International Journal of Psychophysiology, 63, 164–172. Rondan, C., & Deruelle, C. (2007). Global and configural visual processing in adults with autism and Asperger syndrome. Research in Developmental Disabilities, 28, 197–206. Ropar, D., & Mitchell, P. (2001). Susceptibility to illusions and performance on visuospatial tasks in individuals with autism. Journal of Child Psychology and Psychiatry, 42, 539–549.

Please cite this article in press as: Chen, F, et al., Non-superior disembedding performance in children with high-functioning autism and its cognitive style account, Res Autism Spectr Disord (2008), doi:10.1016/j.rasd.2008.03.003

RASD-90; No of Pages 14

14

F. Chen et al. / Research in Autism Spectrum Disorders xxx (2008) xxx–xxx

Rozencwajg, P., Corroyer, D., & Altman, P. (2002). Samuel: Diagnostic du fonctionnement cognitive (A Psychological Test). Paris: Delta Expert. Shah, A., & Frith, U. (1983). An islet of ability in autistic children. Journal of Child Psychology and Psychiatry, 24, 613–620. Waber, D. (1989). The biological boundaries of cognitive styles: A neuropsychological analysis. In T. Globerson & T. Zelniker (Eds.), Cognitive style and cognitive development. New Jersey: Ablex Publishing Corporation. Witkin, H. A., Lewis, H. B., Hertman, M., Manchover, K., Meissner, P. B., & Wapner, S. A. (1954). Personality through perception. New York: Harpes and Brothers. Witkin, H. A., Oltman, P. K., Raskin, E., & Karp, S. A. (1971). A manual for the embedded figures tests. California: Consulting Psychologists Press.

Please cite this article in press as: Chen, F, et al., Non-superior disembedding performance in children with high-functioning autism and its cognitive style account, Res Autism Spectr Disord (2008), doi:10.1016/j.rasd.2008.03.003