Libby Et Al., 1998 Autism

11
Journal of Autism and Developmental Disorders, Vol. 28, No. 6, 1998 Spontaneous Play in Children with Autism: A Reappraisal Sarah Libby, 1,5 Stuart Powell, 2 David Messer, 3 and Rita Jordan 4 Much controversy remains regarding the ability of children with autism to engage in spon- taneous play. In this study children with autism, Down syndrome and typical development with verbal mental ages of approximately 2 years were assessed for play abilities at three data points. Even in this group of children with autism, who had relatively low verbal mental ages, symbolic play skills were not totally absent. However, it was possible to distinguish their pattern of play behaviors from the other two groups. Consequentially, it is argued that there are unusual features in early spontaneous play in children with autism and these atypi- cal patterns are not restricted to their difficulties in the production of symbolic play. Such differences in early spontaneous play raise interesting questions about the etiology of autism, the direction of future research, and the theoretical models that can account for the condi- tion. INTRODUCTION Interest in the pretend play of children with autism has burgeoned over the past decade (e.g., Baron-Cohen, 1987; Jarrold, Boucher, & Smith, 1993, 1994, 1996; Leslie, 1987, 1994; Lewis & Boucher, 1988, 1995). Despite this research the ex- tent of the difficulties in pretend play experienced by this population and the ontogenesis of this deficit remain enigmatic. Furthermore, the focus on the pre- tend play of children with autism has resulted in a failure to investigate more elementary play skills. Pretend play is used here as an umbrella terms for two subcategories of play behavior, functional 1 Department of Clinical Psychology, University of East Anglia, Norwich, NR4 7TJ, United Kingdom. 2 Division of Education, University of Hertfordshire, Watford Campus, Aldenham, Watford, Hertfordshire, WD2 8AT, U.K. 3 Division of Psychology, University of Hertfordshire, Hatfield Campus, Hertfordshire, AL10 9AB, U.K. 4 School of Education, University of Birmingham, Birmingham, B30 1TQ, U.K. 5 Address all correspondence to Sarah Libby, University of East Anglia, Norwich, NR4 7TJ, United Kingdom. and symbolic play. Some authors have used pretend play and symbolic play as synonymous (e.g., Leslie, 1987). However we believe that pretend play may in- volve processes that are not symbolic and as a result we use the term to describe both functional and sym- bolic play. Functional play can be defined as using an object as its function denotes, even if it is a mini- aturized version of this object, for example, pushing a toy car along the carpet making a "brmmm" noise (Ungerer & Sigman, 1981). Symbolic play appears later, from 20 months, and involves treating an object or situation as if it is something else. Leslie (1987) described three forms of symbolic play: (a) object substitution—using an object as if it is something else (e.g., using a brick as some soap), (b) attribution of false properties—attributing properties to an object as if they exist (e.g., pretending a doll is ill), and (c) reference to an absent object—making a reference to something as if it is present (e.g., driving a truck over an invisible bridge). Children with autism tend to produce rigid stereotyped behavior toward toys and objects (Atlas, 1990), and there is a paucity of pretend play. How- ever, the extent of this deficit remains unclear. While 487 0162-3257/98/1200-0487$15.00/0 C 1998 Plenum Publishing Corporation KEY WORDS: Spontaneous play; autism; symbolic play.

description

A peer reviewed article-psychology

Transcript of Libby Et Al., 1998 Autism

Page 1: Libby Et Al., 1998 Autism

Journal of Autism and Developmental Disorders, Vol. 28, No. 6, 1998

Spontaneous Play in Children with Autism: AReappraisal

Sarah Libby,1,5 Stuart Powell,2 David Messer,3 and Rita Jordan4

Much controversy remains regarding the ability of children with autism to engage in spon-taneous play. In this study children with autism, Down syndrome and typical developmentwith verbal mental ages of approximately 2 years were assessed for play abilities at threedata points. Even in this group of children with autism, who had relatively low verbal mentalages, symbolic play skills were not totally absent. However, it was possible to distinguishtheir pattern of play behaviors from the other two groups. Consequentially, it is argued thatthere are unusual features in early spontaneous play in children with autism and these atypi-cal patterns are not restricted to their difficulties in the production of symbolic play. Suchdifferences in early spontaneous play raise interesting questions about the etiology of autism,the direction of future research, and the theoretical models that can account for the condi-tion.

INTRODUCTION

Interest in the pretend play of children withautism has burgeoned over the past decade (e.g.,Baron-Cohen, 1987; Jarrold, Boucher, & Smith,1993, 1994, 1996; Leslie, 1987, 1994; Lewis &Boucher, 1988, 1995). Despite this research the ex-tent of the difficulties in pretend play experiencedby this population and the ontogenesis of this deficitremain enigmatic. Furthermore, the focus on the pre-tend play of children with autism has resulted in afailure to investigate more elementary play skills.

Pretend play is used here as an umbrella termsfor two subcategories of play behavior, functional1Department of Clinical Psychology, University of East Anglia,Norwich, NR4 7TJ, United Kingdom.

2Division of Education, University of Hertfordshire, WatfordCampus, Aldenham, Watford, Hertfordshire, WD2 8AT, U.K.

3Division of Psychology, University of Hertfordshire, HatfieldCampus, Hertfordshire, AL10 9AB, U.K.

4School of Education, University of Birmingham, Birmingham,B30 1TQ, U.K.

5Address all correspondence to Sarah Libby, University of EastAnglia, Norwich, NR4 7TJ, United Kingdom.

and symbolic play. Some authors have used pretendplay and symbolic play as synonymous (e.g., Leslie,1987). However we believe that pretend play may in-volve processes that are not symbolic and as a resultwe use the term to describe both functional and sym-bolic play. Functional play can be defined as usingan object as its function denotes, even if it is a mini-aturized version of this object, for example, pushinga toy car along the carpet making a "brmmm" noise(Ungerer & Sigman, 1981). Symbolic play appearslater, from 20 months, and involves treating an objector situation as if it is something else. Leslie (1987)described three forms of symbolic play: (a) objectsubstitution—using an object as if it is something else(e.g., using a brick as some soap), (b) attribution offalse properties—attributing properties to an object asif they exist (e.g., pretending a doll is ill), and (c)reference to an absent object—making a reference tosomething as if it is present (e.g., driving a truck overan invisible bridge).

Children with autism tend to produce rigidstereotyped behavior toward toys and objects (Atlas,1990), and there is a paucity of pretend play. How-ever, the extent of this deficit remains unclear. While

4870162-3257/98/1200-0487$15.00/0 C 1998 Plenum Publishing Corporation

KEY WORDS: Spontaneous play; autism; symbolic play.

Page 2: Libby Et Al., 1998 Autism

488 Libby, Powell, Messer, and Jordan

some researchers, including Jarrold et al, (1996), Le-wis and Boucher (1988), Sigman and Ungerer (1984),Stone, Lemanek, Fischel, Fernandez, and Altemeier(1990), and Whyte and Owens (1989), have revealeddeficits in both functional and symbolic play. Otherstudies have only found a deficiency in the ability toproduce symbolic play acts (e.g., Baron-Cohen, 1987;Gould, 1986; Mundy, Sigman, Ungerer, & Sherman,1986).

A number of theories have been put forward toexplain these difficulties in the production of pretendplay. The most influential has been the suggestionthat individuals with autism have a metarepresenta-tional deficit. This account was originally proposedby Leslie (1987). His theory suggested that to under-stand reality and engage in functional play we requireprimary representations of the world. However, insymbolic play a child transgresses reality. To avoidconfusion about the properties of objects during suchsymbolic play another level of representation is re-quired. Leslie suggested that primary representationsare decoupled to create metarepresentations whichare separated from their true properties. Leslie pro-posed that this same mechanism is employed withprepositional attitudes which are believed to be in-volved in the development of a theory of mind. Chil-dren with autism have difficulties generatingsymbolic play and developing theory of mind skills,therefore a general deficit in the process of produc-ing metarepresentations was proposed.

There are, however, a number of problems withthe metarepresentational account. Some individualswith autism are capable of producing limited sym-bolic play, especially those children with higher ver-bal mental ages. Although these behaviors are oftenrepetitive and stereotyped they appear to require de-coupling if one accepts Leslie's theoretical model.Furthermore, there is increasing evidence that chil-dren with autism can produce symbolic play whenthey receive appropriate prompts (Gould, 1986; Le-wis & Boucher, 1988; Whyte & Owens, 1989) andcan comprehend pretend play acts (Jarrold, Smith,Boucher, & Harris, 1994; Kavanaugh and Harris,1994). These findings cannot be reconciled with ametarepresentational account. This approach alsofails to account for reports of a functional play defi-cit.

Leslie's metarepresentational theory is furtherconfounded by conceptual issues (see Jarrold, Car-ruthers, Smith, & Boucher, 1994c for a review). Forexample, his theory fails to take account of the de-

velopmental sequence observed in the normal emer-gence of symbolic play, assuming all three forms ofsymbolic play emerge simultaneously. Researchershave found that object substitutions emerge prior toreference to absent objects and attribution of falseproperties (Corrigan, 1987).

Other theories have suggested that children withautism possess the cognitive capacity to perform pre-tend play but have difficulties demonstrating thisability. For example, executive dysfunction (Ozonoff,Pennington, & Rogers, 1991) and generativity deficits(Jarrold et al, 1993, 1996; Lewis & Boucher, 1995)have been proposed as accounting for many aspectsof the behaviors demonstrated by individuals withautism, including then- difficulties with pretend play.These theories are capable of explaining many of thefindings that fail to be accounted for by themetarepresentational account. However, like Leslie'saccount these theories have failed to considerwhether or not individuals with autism have difficul-ties with more basic exploratory and play behaviors.

Before the emergence of pretend play, infantsnormally demonstrate a range of other play behav-iors. Sensorimotor play emerges in the infants, firstyear and can be defined as playing with an objectwithout accounting for its functional characteristics(e.g., banging, oral exploration, spinning). These be-haviors are often repetitive and the child seems toderive low level stimulation from these acts. From 12months, the child begins to combine objects in playduring relational play, for example, placing objectsinside each other or piling objects up (Fein & Apfel,1979).

Children with autism have been found to persistin their production of sensorimotor play (De Myer,Mann, Tilton, & Loew, 1967; Tilton & Ottinger,1964). However, many studies reporting this pre-dominance are confounded by methodological prob-lems. These include the use of inappropriatediagnostic criteria for recognizing autism and a fail-ure to use control groups matched on appropriatemeasures of verbal functioning (Jarrold et al., 1993).Many studies without such methodological flaws havenot closely examined sensorimotor or relational play(Gould, 1986; Riguet, Taylor, Benaroya, & Klein,1981; Whyte & Owens, 1989), whereas others havefailed to find evidence of a preponderance of sen-sorimotor play in individuals with autism (Lewis &Boucher, 1988; Stone et al, 1990).

It is not clear what relationship, if any, theseearly emerging play skills have with the later capacity

Page 3: Libby Et Al., 1998 Autism

Spontaneous Play in Children with Autism 489

to pretend. Nevertheless, some authors argue thatthe first is antecedent to the second. Gorlitz (1987)claimed that establishing familiarity with an objectthrough exploration is a precursor to the capacity togenerate new perspectives on an object or situation,as is evidenced in symbolic play. He cited evidencethat levels of early exploratory behavior have beenfound to correlate with the later emergence of pre-tend play in some children. Roeyers and vanBerkelaer-Onnes (1994) have extended these viewsto autism and suggested that the deficit in pretendplay is a product of children with autism lacking asense of curiosity and exploratory behavior. Theseproposals are clearly in contrast to Leslie's view ofsymbolic play as a qualitatively distinct behavior thatis not related to previously established abilities. Al-though, some hypotheses have concentrated on therelation between early emerging social skills (Hob-son, 1993) or initial theory of mind abilities (Baron-Cohen & Ring, 1994; Charman, 1997) and the lateremergence of pretense, they have overlooked the po-tential role of early play and exploratory behavior.The further examination of these early play skills isnecessary to enhance our understanding of pretencein autism.

This study was conducted to clarify the problemsthat individuals with autism have in the productionof spontaneous play. Attention is given not only topretend play behaviors but also to the production ofrelational and sensorimotor play as well as explora-tory behavior. The work reviewed above reveals thatcareful examination of these early forms of play havebeen neglected in many studies. All the children inthis study had relatively low verbal mental ages com-pared to many previous studies (e.g., Gould, 1986;Jarrold, Boucher, & Smith, 1994a, 1996; Lewis &Boucher, 1988). This allowed the comparison of typi-cally developing children who were just beginning toengage in pretend play with children with autism andchildren with Down syndrome of comparable mentalages. The children engaged in play with a range ofconventional toys and junk objects. It was hoped that

this range of objects would stimulate different typesof play behavior, maximizing the opportunities forboth functional and symbolic play.

From previous work it was expected that thechildren with autism would experience difficulties inthe production of symbolic play. It is less clear if thechildren with autism would have specific difficultieswith the other types of play behavior.

METHOD

Subjects

Twenty-seven children participated in this study,9 with autism, 9 with Down syndrome, and 9 withtypical development. The characteristics of the chil-dren are displayed in Table I. It should be noted thatthe groups of children were not matched for sex orchronological age, only on language expression andcomprehension, as described later.

The children with autism displayed the triad ofimpairments (Wing & Gould, 1979) that are a req-uisite for a diagnosis of autism and attended schoolsfor children with autism. One of the children withDown syndrome attended a school for children withmoderate learning difficulties (MLD), the otherswere in preschool at the onset of this study but hadaccess to some nursery provision. Six of the childrenwith typical development attended private day nurs-eries, the remainder were cared for at home by a par-ent.

The children were matched on their languageproduction and comprehension using the ReynellDevelopmental Language Scale (Reynell & Huntley,1987) (see Table II). It was not possible to collectdata on the expressive language scores using theReynell for all the children. The MacArthur Com-municative Inventory (Fenson et al., 1993) was alsogiven to a parent or teacher to be completed. Forthe two children with Down syndrome who did nothave Reynell expressive language scores, their pro-

Table I. Sex and Chronological Age of the Children

Chronological age

Group

AutisticDown syndromeTypical development

No. of girls

152

No. ofboys

847

M (years;months)

10;034;042;01

SD(months)

44.0313.592.08

Range (years;months)

5;01-16;053;00-6;051;11-2;04

Page 4: Libby Et Al., 1998 Autism

490 Libby, Powell, Messer, and Jordan

ductive vocabulary levels, as measured by the MacArthur, were taken. The vocabulary scores obtainedfor these two children fell at 22 months and 25months on the 50th percentile.

Using a one-way analysis of variance (ANOVA)there was no significant difference between the threegroups on the scores obtained on the Reynell meas-ures of language comprehension, F(2, 26) = 0.56, p= .58, or expression, F(2, 24) = 0.16, p = .85. If thetwo MacArthur scores given above are used in theanalysis of expressive language there remains no sig-nificant difference between the three groups, F(2, 26)= 0.54, p = .59.

Materials

Three groups of ten objects were presented tothe children; these were (a) conventional toys, (b) amixture of conventional toys and junk objects, and(c) an array of junk objects. These objects are de-scribed in the Appendix.

Design and Procedure

During an initial visit to see the children the ex-perimenter administered the Reynell and had an op-portunity to become familiar to the children. Thetesting took place in a quiet room at school, at thenursery, or at the home of the child, depending onwhat was most convenient or appropriate.

On the second visit the children were recordedplaying with the toys described in the Appendix.There was an initial warm-up session where the chil-dren were given a toy giraffe, a miniature plastic tree,a toy racing car, a doll's bottle, and a foil pie dishand encouraged to play with them. This allowed thechildren to become familiar with the procedure be-fore the experimental objects were presented. Noplay behaviors were modeled by the experimenter.The three object sets were given to the children for5 minutes each, the order in which these objects were

presented was randomized, so any effects of orderwere counterbalanced. The experimenter only gavethe children nondirective prompts (e.g., "Look at thethings on the table"). The prompts gave the childrenno indication of what to do with the objects and onlyacted as a means to redirect a child's attention tothe objects if it strayed. If any other adult was pre-sent they were instructed to do the same.

The children were revisited at two further datapoints, the second visit was 3 to 4 months after thefirst assessment of spontaneous play and the thirdwas 3 to 4 months after data point two. This test-retest design was chosen to explore the consistencyof the type of play produced by the groups of chil-dren across time.

Scoring

The videos were analyzed for instances of sen-sorimotor, relational, functional, and symbolic play,as well as nonplay behaviors. The behavior categoriesin the analysis, expanded on previous schemes de-vised by Leslie (1987), Lewis, Boucher, and Astell(1992), and Ungerer and Sigman (1981), are summa-rised in Table III. A time-interval analysis was con-ducted, so at every 15 seconds the play in behaviorthat the child was currently engaged in was coded.This gave a total of 20 coded acts for each of theobject sets, 60 acts in all for each child at each datapoint. The objects that the children acted upon andany language produced were also noted. The videoswere examined again using event sampling coding sothat all symbolic play acts produced could be ana-lysed.

Interrater agreement was calculated for the playbehaviors of three children at each data point, onein each of the groups. Cohen's kappa's of .84 forcategorizing behavior, .92 for noting the objects used,and .90 for event sampling the symbolic play actswere achieved (Cohen, 1960).

Table II. Language Abilites of the Children at Data Point One

Reynell Score

Comprehension(months)

Expression(months)

Autistic

M

29

27

SD

4.80

3.52

n

9

9

Down syndrome

M

27

26

SD

3.43

1.95

n

9

7

Typically developing

M

29

26

SD

4.82

2.29

n

9

9

ANOVA

F

0.56

0.16

P.58

.85

Page 5: Libby Et Al., 1998 Autism

Spontaneous Play in Children with Autism 491

RESULTS

Any change found across time is described first,followed by a consideration of the occurrence of playand nonplay behaviors, and the total number of sym-bolic play acts produced. Correlations were con-ducted to examine any possible relations between thedifferent types of behavior. As the three groups ofchildren differed on their chronological age the as-sociation between age and the production of differ-ent play behaviors was also assessed.

Change Across Time

A Wilcoxon matched-pairs signed-rank test re-vealed no consistent significant changes across all

three data points in the type of play displayed by anyof the three groups of children. Over the 6 monthsthe children with typical development and the chil-dren with Down syndrome showed few changes inthe types of play they were demonstrating. The onlysignificant change for the children with Down syn-drome was from data point one to data point threewhere sensorimotor play was significantly reduced (t= -2.24, p = .03). The children with typical devel-opment showed a reduction in sensorimotor play (t= -2.07, p = .04) and functional play (t = -2.19, p= .03) between data points two and three. Therewere no changes in the play behaviors produced bythe children with autism although exploratory behav-ior increased significantly between data points oneand two (t = -2.67, p = .01) but fell again at datapoint three. These data suggest that the type of play

Table III. Description of the Coding Scheme

Code

1AIB2

34

5

6

7

8

9A

9B

10A

10B

10C

Behavior

Not attendingAttending to objects not actingUnrelated behavior

LabelingGiving/showing

Attempt to terminate session

Exploration

Sensorimotor play

Relational play

Functional play (conventional objects)

Functional play (junk objects)

Symbolic play (object substitution)

Symbolic play (attribution of falseproperties)

Symbolic play (reference to an absentobject)

Description and examples

Looking elsewhere.Looking at objects without acting on them.Self-stimulation that does not involve objects, asking irrelevantquestions, talking to self.Labeling or telling the adult about the objects.Giving or showing the objects to the adult without commentingon them, e.g., trying to initiate a game with the adult.E.g., Throwing the objects off the table or saying 'put away' or'finish.'E.g., Transferring an object from one place to another orturning object over in hands.Objects are acted on without the child taking account of thefunctional features of the object. This is often repetitive. E.g.,oral exploration, banging, spinningRelating two or more objects in a way that does not indicatefunctional or symbolic play, e.g., piling objects up, puttingobjects in a box, put car on top of carwash.Using a toy as its function denotes, e.g., brush doll's hair, pushcar through carwash.Using a junk object as its function denotes, e.g., wiping facewith cotton wool, placing straw in bottle and putting in mouth.Using an object as if it was something else, e.g., using brick asa car, plug as a baby's dummy.Attributing properties to an objects as if they were present,e.g., claiming the toy stove is 'hot,' cleaning the doll's facesaying 'dirty.'Making reference to something that is absent as if it is present,e.g., driving the truck over an imaginary bridge, eatingimaginary food.

Page 6: Libby Et Al., 1998 Autism

492 Libby, Powell, Messer, and Jordan

produced by the children, especially those withautism, was generally consistent across time. In therest of this paper data from all three data points areaveraged to provide a mean score.

Occurrence of Play and Nonplay behaviors

This analysis explores the number of play andnonplay behaviors produced by the three groups. Thepercentage of each behavior was calculated by divid-ing the frequency by 60 (the total number of acts thatwere observed). Table IV provides details of thesedata and the results of statistical analyses using aKruskal-Wallis one-way ANOVA.

There were no significant differences in theamount of nonplay behavior that the three groupsproduced. Although, the children with Down syn-drome produced more behavior that could be classi-fied as nonplay than the other two groups. Theproportion of time that the children spent engagedin exploratory behavior was not significantly differentacross the three groups. This category was used todescribe much of the observed activities of all threegroups, each group spent more than 29% of theirtime in exploratory behavior.

The children with autism produced significantlymore sensorimotor play when compared to the othertwo groups (H = 10.77, df = 2, p < .01). A Mann-Whitney U test on these data revealed that the chil-dren with autism produced significantly moresensorimotor play than children with Down syn-drome (U = 8, p < .01) and children with typicaldevelopment (U = 11, p < .01). There was no sig-nificant difference in the production of relationalplay by the groups. However, the children withautism produced less of this behavior than the chil-dren with Down syndrome and typical development.

There was no significant difference in the pro-portion of functional play using conventional objectsproduced by the three groups of children. Although,the children with typical development producedmore functional play than the other two groups.

The children with Down syndrome and typicaldevelopment produced significantly more symbolicplay acts than the children with autism (H = 8.21,df = 2, p = .02). This is confirmed by a Mann-Whit-ney U test conducted on these data, children withautism produced significantly less symbolic play thanthe children with Down syndrome (U = 12.5, p =.01) and the children with typical development (U =13.5, p = .02).

Analysis of All Symbolic Play Acts

Due to the low level of symbolic play productionobserved using the time-interval analysis the videosof the children were reexamined to record every in-cidence of symbolic play. These data and analyses aredisplayed in Table V

There was no significant difference in the totalnumber of symbolic play acts, which were classifiedas object substitution, across the three groups of chil-dren. However, the children with autism produced nosymbolic play acts involving a reference to an absentobject (H = 10.28, df=2,p = .01). This was signifi-cantly less than the children with Down syndrome (U= 9, p < .01) and the children with typical develop-ment (U = 18, p = 0.01). The children with autismalso produced significantly fewer acts that involved at-tributions of false properties (H = 12.04, df = 2, p< .01). This was significantly fewer than the childrenwith Down syndrome (U = 4, p < .01) and the chil-dren with typical development (U = 12, p = .01). Fivechildren with autism produced some symbolic play

Table IV. Average Percentage of Each Play Behavior Produced by the Children in Each Group0

Type of play

NonplayExploratorySensorimotorRelationalFunctionalSymbolic

Autistic

M21.4938.8517.2911.949.370.70

SD7.56

11.217.435.246.911.10

Down syndrome

M30.7929.587.80

15.978.145.05

SD13.545.165.357.575.084.90

Typically developing

M21.2433.6310.6715.6113.333.75

SD

8.717.683.695.235.663.49

Kruskal-Wallis

H2.283.62

10.773.284.658.21

(df = 2)

P.32.16

<.01.19.12.02

an was 9 for each type of play except for symbolic play, n for symbolic play was 5 for the children with autism, 8 for thechildren with Down syndrome, and 8 for the children with typical development.

Page 7: Libby Et Al., 1998 Autism

Spontaneous Play in Children with Autism 493

acts, compared to all 9 children with Down syndromeand 8 of the children with typical development.

Correlations Between Play behaviors

To begin to understand if any specific relation-ships hold between different types of play behaviora Pearson's r was produced for all the pairs of playbehaviors using data from all three groups. A num-ber of negative correlations were significant. Sen-sorimotor play was negatively correlated withsymbolic play (r = -.53, p < .01) and relationalplay (r = -.45, p = .05). Nonplay behaviors werenegatively correlated to functional play (r = -.43,p — .05) and exploration (r = -.66, p < .01).

Whether or not the chronological age (CA) ofa child mediates the type of play behaviors producedwas also of interest. CA was not correlated to anyplay behaviors in the group of children with autism.However, for the children with Down syndrome theproduction of functional play was correlated to CA(r = .69, p = .04). The children with typical devel-opment demonstrated a correlation between CA andthe production of relational play (r = .79, p = .01).

DISCUSSION

Symbolic Play

The children with autism had difficulties in theproduction of symbolic play compared to the other

two groups of children, this is especially evident whenexamining the total number of symbolic play actsproduced. However, symbolic play was demonstratedby five of the children with autism during at leastone of the data points. This finding supports the ar-gument that children with autism do not have ametarepresentational deficit in Leslie's terms. As,even in this group of children with relatively low ver-bal mental ages there is some evidence of a capacityto engage in symbolic play. Interestingly, all but twoof the symbolic play acts that were produced by thechildren with autism were object substitutions.

Other investigators who have detailed the typeof symbolic play acts produced by children withautism have predominantly described object substi-tutions (Baron-Cohen, 1987; Boucher & Lewis,1990). For example, Boucher and Lewis (1990) pre-sented data giving examples of creative symbolic playgenerated by children with autism in their study ofelicited pretence. The majority of the acts they de-scribe are object substitions, for example, a subjectholds a straw upright on the floor, pushes the carpast, and says "lamp-post." However, this tendencyto produce object substitutions rather than otherforms of symbolic play has not been addressed in theliterature. This finding clearly conflicts with Leslie'sview that the three forms of symbolic play emergeas a package.

This preserved capacity to produce object sub-stitutions could be explained in a number of ways. Itmay indicate that the production of object substitu-tions is dependent on a different cognitive mecha-nism than other forms of symbolic play. Currie

Table V. Total Number of Symbolic Play Acts Observed Across All Three Data Points

Symbolic play type

Object substitutionMnAttribution of false propertyMnReference to absent objectMn

TotalMn

Group

Autistic

244

21

00

265

D/S

296

659

877

1819

T/D

347

457

595

1388

Kruskal-Wallis(df = 2)

3.01, p = .22

12.04, p < .01

10.28, p = .01

9.63, p = .01

Mann-Whitney U

Autistic vs. D/S

30, p = .33

4, p < .01

9, p < .01

8, p < .01

Autistic vs. T/D

21.5, p = .08

12, p = .01

18, p = .01

13, p = .01

Page 8: Libby Et Al., 1998 Autism

494 Libby, Powell, Messer, and Jordan

(1996) has argued that a process like metarepresen-tation may not be required for object substitutions,"I do not, in particular, believe that the use of anobject for the purposes of pretence requires theagent to see the object as a symbol for somethingelse" (p. 252). Consequentially, it is possible that ob-ject substitutions can be generated purely throughthe understanding of correspondence or simulationwhich do not involve a symbolic process (Harris,1994; Perner, Baker, & Mutton, 1994; Lillard, 1993).

An alternative view is that object substitutionsare easier to produce because of the visual similari-ties between the object used and what is being sym-bolized. This could be explained by the generativityhypothesis. Jarrold et al. (1996) and Lewis andBoucher (1988) have demonstrated that individualswith autism produce more symbolic play when theenvironment is structured. The predominance of ob-ject substitutions in the autistic population could bea result of greater contextual support for these sym-bolic acts. Other forms of symbolic play may be moredependent on the child generating ideas for pretensewhich are independent of the current context.

A related view would be to suggest that the pre-tend play difficulties experienced by those withautism were a product of general developmental de-lay. Object substitutions are the first forms of sym-bolic play to emerge in typical development(Corrigan, 1987), so it may be that the play of chil-dren with autism is no different, only delayed in itsmaturation. However, a delay hypothesis fails to ac-count for other aberrations in the pretend play pro-duced by this population. For example, children withautism tend to produce isolated or perseverative sym-bolic play (Atlas, 1990; Wulff, 1985). An examplefrom the children observed in this study is a boy withautism who perseverated in labeling a piece of stringas a "worm" without animating it. Although suchnarrow behavior was not completely absent in thenonautistic groups, the other children tended to showmore flexible and elaborate symbolic play.

Currently, there is little to choose between thedifferent explanations about the capacity of childrenwith autism to produce object substitutions. Furtheranalysis of the type of symbolic play behaviors in in-dividuals with autism who have a higher mental agemay help to distinguish between these various hy-potheses and influence the type of theories which canexplain this condition. Another issue which needs tobe addressed in future work concerns the finding thatsuggests children with autism have some limited ca-

pacity to produce symbolic play but have problemsexercising this in a flexible manner.

Functional Play

The children with autism did not demonstratesignificantly less functional play than the other twogroups of children. Furthermore, functional play wasproduced by all the children with autism. Superfi-cially these findings indicate that there is not a func-tional play deficit in this group. However, it appearsthat both the children with autism and Down syn-drome produce less functional play than the childrenwith typical development. For the children withDown syndrome this limited amount of functionalplay is apparent alongside a normal capacity to pro-duce symbolic play, but for the children with autismthis accompanies their difficulties in the productionof symbolic play.

A possible reason for the production of lessfunctional play by children with autism and childrenwith Down syndrome is that both groups have diffi-culties in the development of conceptual and cate-gorical knowledge that may be required in theproduction of functional play (Klinger & Dawson,1995). However, further work is required to deter-mine if these conceptual difficulties relate to prob-lems with functional play. In addition subtyping offunctional play may be necessary to identify differ-ences between children with autism and other chil-dren.

Other Play behaviors

Relational play did not appear to vary signifi-cantly between the three groups of children. How-ever, in this study the different types of relationalplay were grouped together. Fein and Apfel (1979)described how children first relate two objects to-gether before engaging in more complex combina-tions and relating objects in a socially appropriatemanner. There was a trend for the children withautism to produce less relational play than the othergroups of children and it is possible that if their be-haviors were observed in more detail, significant dif-ferences in the play behavior of children with autismmight be identified.

The children with autism produced significantlymore sensorimotor activities supporting the view thatthis behavior may dominate play in this group. A

Page 9: Libby Et Al., 1998 Autism

Spontaneous Play in Children with Autism 495

dominance of sensorimotor play could prevent an in-dividual with autism from displaying their capacityfor symbolic play. This study provides some supportfor this suggestion as the production of sensorimotorplay was negatively correlated with the production ofsymbolic play. When a child with autism is con-fronted with a set of objects the salience of the sen-sorimotor features of the objects may overpowertheir functional and symbolic potentials. In struc-tured situations such difficulties are less apparent asthe functional and symbolic aspects of an object aremade more accessible to the child by controlling theirprepotent sensorimotor response. This is compatiblewith reports of increased symbolic play abilities instructured settings (Gould, 1986; Lewis & Boucher,1988; Whyte & Owens, 1989).

Alternatively, Roeyers and van Berkelaer-Onnes(1994) have suggested that because of the dominanceof stereotyped, persistent, and self-stimulatory behav-ior in this population, curiosity is not awakened, andthe development of symbolic play muted. They arguethat the dominance of sensorimotor play could re-strict children with autism's experience of objects.This in turn would limit their perceived control andcapacity to explore the objects and delay the devel-opment of stable functional concepts about those ob-jects. This could culminate in difficulties generatingsymbolic play.

Another argument should also be considered. Itis possible that because of the problems developingsymbolic play sensorimotor behavior proliferates.Longitudinal studies following the development ofchildren with autism over time are clearly requiredto understand how a tendency to produce sensori-motor behavior may influence the development ofpretend play.

There was no difference in the amount of ex-ploratory behavior between the three groups of chil-dren. However, the time-interval analysis may nothave been sensitive enough to pick up any subtle dif-ferences in exploratory behavior between the threegroups. By assessing the specific nature of explora-tory behavior produced, a more accurate assessmentof exploration may be obtained. For example,whether a child visually scans the objects when actingon them may be of particular interest as this is oneof the defining feature of exploration in typical de-velopment (Ruff & Salterelli, 1993).

Methodological Considerations

The extent that the behavior recorded here wasspontaneous should be considered. Although theprompts that were given were nondirective andwould not affect the type of behavior that the chil-dren produced on the objects if these cues were notpresent, it is possible that the children with autismmay have spent even less time interacting with theobjects. The children with autism spent more timelooking elsewhere at all three data points and mayhave remained distracted in this behavior if attentionwas not drawn to the objects. Skelly, Lewis, and Col-lis (1993) also found that children with autism spentsignificantly more time "looking away" than the con-trol groups, supporting these findings.

One possible criticism of the current study isthat although the children in the different groupswere matched on verbal mental age, the childrenwith autism were substantially older than the othertwo groups of children. Therefore, the IQ of the chil-dren with autism was likely to be lower than the chil-dren in the other groups. Measures of verbal mentalage are imperative when studying pretend play as intypical development the production of pretend playis believed to be related to language level (Ungerer& Sigman, 1981). Although CA was related to theproduction of certain play skills in the other twogroups, CA did not mediate the type of play whichwas produced by the children with autism. The chil-dren with autism had also been exposed to moreyears of schooling than the other two groups of chil-dren. This may have influenced the type of behaviorsthe children with autism demonstrated. Future stud-ies should consider the effects of IQ and CA as wellas verbal mental age on the play behaviors producedby the autistic population.

CONCLUSION

This study focused on the type of early sponta-neous play children with autism produce as well asfunctional and symbolic play behaviors. By observingmany aspects of play behavior a dominance of sen-sorimotor play was identified. Some tentative sugges-tions were made regarding how this predominanceof sensorimotor behavior could have a bearing on theproduction of symbolic play and it is hoped that fu-ture research will explore the relationship between

Page 10: Libby Et Al., 1998 Autism

496 Libby, Powell, Messer, and Jordan

these play behaviors in both typical and atypical de-velopment.

As some symbolic play was observed in the chil-dren with autism, Leslie's claim that this group hasa metarepresentational deficit appears simplistic. Ofparticular interest is the dominance of object substi-tutions over other types of symbolic play producedby children with autism. What the presence of thesesymbolic play behaviors mean for the underlying cog-nitive capacities of individuals with autism and howthis is linked to other difficulties experienced by thisgroup remains to be established. There was also evi-dence of some difficulties in functional play althoughthe amount of functional play produced by the chil-dren with autism did not differ significantly from thecontrol groups. Clearly, the qualitative dimensions offunctional play should be examined in future studies.To conclude, this study provides some new findingsregarding both the pretend play and early play be-haviors produced by children with autism. It is hopedthat future studies will build on the observationsmade here to further reappraise our understandingof play in the autistic population.

ACKNOWLEDGEMENTS

This research would not have been possiblewithout the help of the parents and children whotook part, to whom we are indebted. We are alsograteful to the staff at Radlett Lodge School, Hert-fordshire; Heathermount School, Berkshire; andCopperbeech Nurseries, Hertfordshire for participat-ing. This work was first presented at the British Psy-chological Society Developmental Conference,Glasgow, September 1995.

APPENDIX

Objects Used in the Study

(a) Conventional toys and accessories:CarDollCarwashTraffic lightsPetrol pumpHairbrushSpoonBowl

Toy stovePan

(b) Conventional toys and junk objects:Teddy bearTruckCotton woolCardboard boxPiece of materialPlugRulerToilet rollFlower potWooden brick

(c) An array of junk objects:Washing-up liquid bottleMargarine tubStringPiece of tin foilMatchboxStrawCottonreelLollipop stickClothes pegPiece of cardboard

REFERENCES

Atlas, J. A. (1990). Play in assessment and intervention in thechildhood psychoses. Child Psychiatry and Human Develop-ment, 21, 119-133.

Baron-Cohen, S. (1987). Autism and symbolic play. British Journalof Developmental Psychology, 5, 113-125.

Baron-Cohen, S., & Ring, H. (1994). A model of the mindreadingsystem: Neuropsychological and neurobiological perspectives.In C. Lewis & P Mitchell (Eds.), Children's understanding ofmind. Hove, U.K.: Erlbaum.

Boucher, J, & Lewis, V (1990). Guessing or Creating? A reply toBaron-Cohen. British Journal of Developmental Psychology:8(2), 205-206.

Charman, T. (1997). The relationship between joint attention andpretend play in autism. Development and Psychopathology, 9,1-16.

Cohen, J. (1960). A coefficient of agreement for nominal scales.Educational and Psychological Measurement, 20, 37-46.

Corrigan, R. (1987). A developmental sequence of actor-objectpretend play in young children. Merill-Palmer Quarterly, 33,87-106.

Currie, G. (1996). Simulation-theory, theory-theory, and the evi-dence from autism. In P. Carruthers & P. K. Smith (Eds.),Theories of theories of mind. Cambridge, U.K.: CambridgeUniversity Press.

De Myer, M. K., Mann, M. A., Tilton, J. R., & Loew, L H. (1967).Toy play behavior and use of body by autistic and normal chil-dren as reported by mothers. Psychological Reports, 21, 973-981.

Fein, G. G., & Apfel, N. (1979). Some preliminary observationson knowing and pretending. In N. Smith & M. Franklin

Page 11: Libby Et Al., 1998 Autism

Spontaneous Play in Children with Autism 497

(Eds.), Symbolic functioning in childhood. Hillsdale, NJ:Erlbaum.

Fenson, L., Dale, P. S., Reznick, J. S., Thai, D., Bates, E., Hartung,J. P., Pethick, S., & Reilly, J. S. (1993). MacArthur Commu-nicative Development Inventories. San Diego, CA: Singular.

Gorlitz, D. (1987). Exploration and attribution in a developmentalcontext. In D. Gorlitz & J. F. Wohlwill (Eds.), Curiosity, imagi-nation and play: On the development of spontaneous, cognitiveand motivational processes. Hillsdale, NJ: Erlbaum.

Gould, J. (1986). The Lowe and Costello play test in socially im-paired children. Journal of Autism and Developmental Disor-ders, 16, 199-213.

Harris, P. L. (1994). Understanding pretence. In C. Lewis & P.Mitchell (Eds.), Children's early understanding of mind. Hove,U.K.: Erlbaum.

Hobson, R. P. (1993). Autism and the development of mind. Hove,U.K.: Erlbaum. Jarrold, C., Boucher, J., & Smith, P. (1993).Symbolic play in autism: A review. Journal of Autism and De-velopmental Disorders, 23, 281-309.

Jarrold, C., Boucher, J., & Smith, P. K. (1994). Executive func-tioning deficit and the pretend play of children with autism:A research note. Journal of Child Psychology and Psychiatry,35, 1473-1482.

Jarrold, C., Boucher, J., & Smith, P. K. (1996). Generativity deficitsin pretend play in autism. British Journal of DevelopmentalPsychology, 14, 275-300.

Jarrold, C., Carruthers, P., Smith, P. K., & Boucher, J. (1994). Pre-tend play: Is it metarepresentational? Mind and Language, 9,445-468.

Jarrold, C., Smith, P., Boucher, J., & Harris, P. (1994). Childrenwith autism's comprehension of pretense. Journal of Autismand Developmental Disorders, 24, 433-455.

Kavanaugh, R. D., & Harris, P. L. (1994). Imagining the outcomeof pretend transformations: Assessing the competence of nor-mal children and children with autism. Developmental Psychol-ogy, 30, 847-854.

Klinger, L. G., & Dawson, G. (1995). A fresh look at categoriza-tion abilities in persons with autism. In E. Schopler & G. B.Mesibov (Eds.), Learning and cognition in autism. New York:Plenum Press.

Leslie, A. M, (1987). Pretense and representation: The origins of"theory of mind." Psychological Review, 94, 412-426.

Leslie, A. M. (1994). Pretending and believing: Issues in the theoryof TbMM. Cognition, 50, 211-238.

Lewis, V, & Boucher, J. (1988). Spontaneous, instructed and elic-ited play in relatively able autistic children. British Journal ofDevelopmental Psychology, 6, 325-337.

Lewis, V, & Boucher, J. (1995). Generativity in the play of youngpeople with autism. Journal of Autism and Developmental Dis-orders, 25, 105-121.

Lewis, V, Boucher, J., & Astell, A. (1992). The assessment of sym-bolic play in young children: A prototypical test. EuropeanJournal of Disorders Of Communication, 27, 231-245.

Lillard, A. S. (1993). Pretend play skills and the child's theory ofmind. Child Development, 64, 348-371.

Mundy, P., Sigman, M., Ungerer, J., & Sherman, T (1986). Defin-ing the social deficits of autism: The contribution of nonverbalcommunication measures. Journal of Child Psychology and Psy-chiatry, 27, 657-669.

Ozonoff, S., Pennington, B. E, & Rogers, S. J. (1991). Executivefunction deficits in high-functioning autistic individuals: Re-lationship to theory of mind. Journal of Child Psychology andPsychiatry, 32, 1081-1105.

Perner, J., Baker, S., & Hutton, D. (1994). Prelief: The conceptualorigins of belief and pretence. In C. Lewis & P. Mitchell(Eds.), Children's early understanding of mind: Origins and de-velopment. Hove, U.K.: Erlbaum.

Reynell, J. K., & Huntley, M. (1987). Reynell Developmental Lan-guage Scale (3rd ed.). Windsor, U.K.: NFER-Nelson.

Riguet, C. B., Taylor, N. D., Benaroya, S., & Klein, L. S. (1981).Symbolic play in autistic, Down's, and normal children ofequivalent mental age. Journal of Autism and DevelopmentalDisorders, 11, 61-70.

Roeyers, H., & van Berkalaer-Onnes, I. (1994). Play in autisticchildren. Communication and Cognition, 27, 349-359.

Ruff, H. A., & Saltarelli, L. M. (1993). Exploratory play with ob-jects: Basic cognitive processes and individual differences. InM. H. Bornstein & A. W O'Reilly (Eds.), The role of play inthe development of thought: New directions for child develop-ment, 59. San Francisco: Jossey-Bass.

Sigman, M., & Ungerer, J. (1984). Cognitive and language skillsin autistic, mentally retarded, and normal children. Develop-mental Psychology, 20, 293-302.

Skelly, A., Lewis, V, & Collis, G. (1993). Affect and attention dur-ing motivated symbolic play in children with autism. Presentedat the British Psychological Society Developmental SectionAnnual Conference, University of Birmingham, September 10-13.

Stone, W. L., Lemanek, K. L., Fischel, P. T., Fernandez, M. C., &Altemeier, W. A. (1990). Play and imitation skills in the di-agnoses of autism in young children. Pediatrics, 86, 267-272.

Tilton, J. R., & Ottinger, D. R. (1964). Comparison of the toyplay behavior of autistic, retarded and normal children. Psy-chological Reports, 15, 967-975.

Ungerer, J. A., & Sigman, M. (1981). Symbolic play and languagecomprehension in autistic children. Journal of the AmericanAcademy of Child Psychiatry, 20, 318-337.

Whyte, J., & Owens, A. (1989). Language and symbolic objectplay: Some findings from a study of autistic children. IrishJournal of Psychology, 10, 317-332.

Wing, L., & Gould, J. (1979). Severe impairments of social inter-action and associated abnormalities in children: Epidemiologyand classification. Journal of Autism and Developmental Dis-orders, 9, 11-19.

Wulff, S. B. (1985). The symbolic play and object play of childrenwith autism: A review. Journal of Autism and DevelopmentalDisorders, 15, 139-148.