Margot Prior - University of Melbourne, Melbourne,
Australia
INTRODUCTION
There has been much debate in the psychiatric literature
over the classification of subtypes in the Pervasive Developmental
Disorders (PDD) (APA, 1994), or increasingly referred to as the
Autism Spectrum Disorders (Szatmari, 1992; Wing, 1991). 'The
inclusion of categories such as Aspergers Disorder (AsD) and the
"waste basket" category of PDDNOS in classification
systems such as DSM IV and ICD-10, has occurred without fully
determining the taxonomic validity of these groups. For a category
to have taxonomic validity, it needs to show internal validity,
ie., it can be measured and found to be different to another category;
and second, it needs to have extemal validity, that is be different
to other disorders on some measure that was not used to define
the category. A number of studies have attempted to compare clinically
diagnosed groups on extemal grounds such as: developmental indices
(Szatmari et al., 1989), IQ and neuropsychological profiles (eg.,
Ozonoff et al., 1991; Szatmari et al., 1990), "clumsiness"
(Manjiviona and Prior, 1995) and Theory of Mind (ToM) abilities
(Ozonoff et al., 1991), with mixed results. A criticism made
of these studies is the lack of sufficiently validated diagnostic
criteria required to differentiate the comparison groups. In
other words, the search for external validity has occurred at
the expense of intemal validity issues. In a review of the literature,
Szatmari (1992) suggested that a new generation of studies are
needed to establish differences between potential subgroups and
Autism.
In order to circumvent clinical diagnostic approaches
to form subgroups, a handful of studies (eg., Siegel, et al.,
1986; Eaves et al., 1994) have used a statistical method known
as cluster analysis. This form of analysis empirically derives
groups from the data itself (in this case, autism symptomatology),
and therefore is not reliant on clinicians diagnosis. This approach
was used to derive comparison subgroups for the present study.
The cluster analysis is presentes in a companion paper, "Subgroups
versus the spectrum: A cluster analysis of a sample of autistic
children" (Prior, Eisenmajer, Leekam, Wing & Gould).
To summarize the findings, the study found that a sample of 135
autistic spectrum and 22 normal children were best clustered on
autism symptomatology into three groups, an "Autistic-like"
group, an "Asperger-like" group, and a "Mild/Control"
group. In general, the "Autistic" group were more severely
impaired in terms of clinical symptomatology than the "Asperger"
group, and least impaired was the "Mild/Control" group.
Age and verbal mental age factors also played a part in differentiation.
In the main the "Asperger" group were more likely to
have a friend with similar interests, use long winded pedantic
speech and show more joint attention behaviour than the "Autistic"
group. lt was concluded that the subgroups were best explained
in terms of severity and developmental characteristics, rather
than by distinctive symptom patterns. lt is suggested that the
results are best understood as evidence of an autistic spectrum,
rather than the existence of separate diagnostic groups.
Of interest is the comparison between the cluster groups on tests of Theory of Mind (ToM). ToM is defined as the ability to attribute mental states to others and it is claimed that we make use of this ability to make sense of everyday behaviour. Many studies have reported that Autistic children tested on experimental ToM tasks perform at lower levels than normal and intellectually disabled controls. It has been proposed (Baron-Cohen, Leslie and Frith, 1985; Frith, 1989) that a deficit in ToM ability may explain the distinctive triad (ie., the social, communication and imagination) of impairments found in autistic individuals.
However, the theory is not foolproof, as studies investigating ToM ability have consistently reported a proportion of autistic children that indeed do pass these tasks. Although originally ToM ability was believed to be independent of age and IQ, evidence now suggests that autistic children who pass these tasks tend to be older and have a higher verbal mental age (Eisenmajer and Prior, 1990; Frith, Happe & Siddons, 1995).
It has been suggested that subgroups on the autistic spectrum may differ on ToM ability, and the talented subgroup who pass the experimental tasks may indeed possess rudimentary ToM skills (Frith, et al., 1995). Our study is the first to confirm this notion, finding the "Asperger"-Iike group more likely to pass first- and second-order ToM tasks than the other two groups. These children are also older and have a higher verbal mental age, suggesting an association between ToM ability and verbal skills.
Of further interest is whether the "Asperger-like" group's better performance on the experimental tasks translates into everyday ToM behaviour. Do the children who pass ToM tasks show more real-life ToM dependent behaviours (such as lying, joking, empathy) than those who fail? If this occurred it would provide further support for the use of ToM as a possible marker for subgroups within the autistic spectrum.
An important related issue is the extent to which
the ToM concept has construct validity. Do the ToM tasks actually
reflect real-life ToM competence in autistic children? A recent
study by Frith, Happe and Siddons (1995) found differences between
first-order ToM passers and failers on naturalistic ToM behaviours.
It is of interest therefore, to determine whether this difference
also occurs in a higher functioning autistic sample of 2nd order
ToM passers and failers?
In summary, our aims are twofold:
First, does passing experimental ToM tasks translate into competence with everyday ToM behaviours such as lying, joking, and showing empathy?
Second, do empirically derived groups who pass experimental
ToM tasks perfonn more everyday ToM behaviours, and thus provide
further externa validity for ToM as a means of differentiating
autistic subtypes?
METHOD
Subjects:
The details of how subjects were grouped by cluster
analysis is provided in the accompanying paper (cited earlier).
Of the 135 autistic spectrum subjects, 60 took part in the naturalistic
ToM study. Twenty-four subjects from the "Asperger-like"
group, twentyeight "Autistic-like", and eight "Mild"
subjects participated. Due to the small sample size of the "Mild"
group, these subjects were excluded from the analyses.
Table I. Subject characteristics of "Asperger-like" and "Autistic-like" group.
| SUBJECT
CHARACTERISTICS | "ASPERGER-LIKE"
N=24 | "AUTISTIC-LIKE" N=28 | P |
| Chronological Age (years)
(Range) |
|
| |
| Verbal Mental Age (years)
(Range) |
|
| |
| Gender (M/F) |
Table 1 shows that the groups were smilar in age and gender.
The "Asperger-like" group had a significantly higher
verbal mental age compared to the "Autistic-like" group
as measured by the Peabody Picture Vocabulary Test (PPVT).
Measures:
Naturalistic ToM
The main measure used to assess naturalistic ToM behaviours was developed by Australian researchers, Beryl McKenzie and Kirsten Hoogenraad, of La Trobe University. Their questionnaire examined the development of deception abilities in young normal children. A factor analysis indicated two stages in the development of deception in young normal children:
Stage 1 relates to the manipulation of behaviour using learnt strategies. This stage measured behaviours such as lying to avoid punishment, sharing, sabotage, and use of mental state terms such as "know" and "think".
Stage 2 related to the understanding of mental states. This stage measured false belief attribution, understanding intentions, informational access, joking and deceit.
A hypothesized third stage in older children was not confirmed by the factor analysis due to insufficient data, but conceptually Stage 3 is related to the use of deception for altruistic purposes.
lt is argued that these stages reflect development of meta-representational ability, such as is needed to pass ToM tasks.
A further two groups of questions were added to the checklist
from knowledge of the particular deficits observed in autistic
children. It is argued that some of the behaviours are probably
more likely to require ToM ability than others, therefore we composed
high and low ToM relevant categories. Examples of the Low ToM
category ie., behaviours unlikely to necessarily require ToM ability
are eg., behaviour to please parent, informing parent of sickness,
engaging in passive play, offering help, and respecting privacy.
It is possible that these activities could be rote learnt and/or
reinforced and achieved without full appreciation of mental states.
Examples of the High ToM category ie., behaviours likely to necessarily
require ToM ability are eg., understanding social rules, understanding
other people's emotions, understanding importance of keeping promises,
non-verbal emotional manipulation, awareness of "difference"
to other children, and feeling embarrassed
Experimental ToM tasks
The first-order ToM measures included the "'Sally/Anne"
task and the "Box of Smarties" task. Bowler's (1992)
"Shopping" story was a measure of higher order or second-order
ToM. Children were only tested on this task if they passed either
of the first-order ToM tasks. lf they failed either of these tasks,
they were presumed unable to pass the second-order ToM task.
Verbal Mental Age
The Peabody Picture Vocabulary Test - Revised was administered
to the subjects as a measure of verbal mental age.
RESULTS
Only 10 subjects failed the first-order ToM tasks and so no comparisons
were possible due to unequal group sizes. All comparisons were
made using the second-order ToM task. To determine whether passing
the second-order ToM task reflected real-life ToM competence,
the subjects were divided into a 'passers' group and a 'failers'
group. Twenty- five subjects passed and twenty-seven failed.
Table II. Subject characteristics of second order ToM 'passers' and 'failers'.
| SUBJECT
CHARACTERISTICS |
|
| P |
| Chronological Age (years)
(Range) |
|
| |
| Verbal Mental Age (years)
(Range) |
|
|
Table 2 shows that the age of the 'passers' was similar to the 'failers', however the 'passers' had a significantly higher verbal mental age to the 'failers'.
When these groups were compared on the three stages of deception
and High and Low ToM questions, the second-order 'passers' were
clearly more able than the 'failers' for all stages and High and
Low ToM behaviours (See Table 3).
Table III. Secores of second-order ToM 'passers' and 'failers'
in relation to levels of naturalistic ToM competence.
| Levels of ToM |
|
| |
| STAGE 1
(max=11) | |||
| STAGE 2
(max=7) | |||
| STAGE 3
(max=5) | |||
| LOW ToM
(max=8) | |||
| HIGH ToM
(max=9) |
To determine whether cluster groups differed on experimental and
naturalistic ToM abilities, the following analyses were completed.
Table IV. Comparison of "Asperger-like" and "Autistic-like"
groups on first and second order ToM.
| CLUSTER ToM LEVEL | "ASPERGER-LIKE" | "AUTISTIC-LIKE" | P |
| FIRST-ORDER ToM | PASSERS=23 FAILERS=1 96% passed | PASSERS=19 FAILERS=9 68% passed |
|
| SECOND-ORDER ToM | PASSERS=16 FAILERS=8 67% passed | PASSERS=9 FAILERS=19 32% passed |
X2 = 6.17 p<0.01 |
Table 4 shows that almost all of the 'Asperger-like' group passed
the first-order ToM task and two thirds passed the second-order
task. A significantly smaller percentage of the 'autistic-like'
group demonstrated first- or second-order ToM ability compared
to the Asperger group.
Table V. Cluster group in relation to levels of naturalistic
ToM competence.
| Levels of ToM |
|
| |
| STAGE 1
(max=11) | |||
| STAGE 2
(max=7) | |||
| STAGE 3
(max=5) | |||
| LOW ToM
(max=8) | |||
| HIGH ToM
(max=9) |
When the two cluster groups are compared on naturalistic ToM behaviours
the same pattern is found, with the 'Asperger-like' group more
likely to show real-life mentalizing abilities than the 'Autistic-like'
group (See Table 5).
SUMMARY AND DISCUSSION
In summary, the results show firstly, that some autistic children do pass ToM tests and display ToM dependent behaviours in real-life. The extent to which autistic ToM competence compares to that of normal children however is yet to be determined. Second, the results also replicate other recent studies showing that ToM is associated with verbal mental age in autistic individuals. Third, experimental ToM tasks appear to have some construct validity as they appear to correspond with real-life ToM competence in autistic individuals. Fourth, experimental and naturalistic evidence supports the notion that ToM is potentially a useful marker for subtypes on the autistic spectrum.
The ToM hypothesis has offered clinicians and researchers a useful tool to explain the many impairments that we see in autistic people. This study and a recent study by Frith, Happe and Siddons (1995) has now found a talented minority on the spectrum who do display real-life ToM abilities. This group may be sufficiently different in clinical presentation to be clustered into a separate group. It is these individuals who may offer some insight into how an autistic individual with a functioning yet inadequate ToM 'mechanism' may present. It is the autistic individuals who pass ToM who may also offer some insights into the considerable clinical variability observed in individuals over time. It is known that a proportion of children with a classic 'aloof-type' autistic presentation who are largely uninterested in the social world, develop into individuals with an "Asperger-like active-but odd" interest in their social surroundings. Indeed these individuals act as if they have 'partial' use of a ToM mechanism. Long-term studies would discover whether this progression from one autistic 'type' to another corresponds with an increase in mentalizing ability. Work in this area will help establish the validity of the autistic spectrum in which individuals are possibly spread on a ToM/mental age continuum.
REFERENCES
and St
American Psychiatric Association. (1994), Diagnostic and Statistical Manual of Mental Disorders (Revised) - DSM IV 4th Edition. Washington DC: American Psychiatric Association.
Baron-Cohen, S., Leslie, A. M.. & Frith, U. (1985). Does the autistic child have a "theory of mind". Cognition, 21: 37-46.
Eaves, L.C., Ho, H.H., Eaves, D.M. (1994), Subtypes of autism by cluster analysis. J. Autism Dev. Disord., 24: 3-22.
Frith, U. (1 99 l), Asperger and his syndrome. In: Autism and Asperger Syndrome. Frith, U., Ed. Cambridge: C.U.P.
Frith, U., Happe, F., Siddons, F. (1995), Autism and theory of mind in everyday life. Social Development, 3: 108-124.
Manjiviona, J. & Prior, M. (1995). Comparison of Asperger Syndrome and high-functioning autistic children on a test of motor impairment. J. Autism Dev. Disord., 25: 23-39.
Ozonoff, S; Pennington, B. F& Rogers, S.J. (1991), Executive function deficits in high-functioning autistic individuals: Relationship to Theory of Mind. J. Child Psychol. Psychiat., 32: 1081-1105.
Siegel, B., Anders, T.F., Ciaranello, R.D., Bienenstock, B., & Kraemer, H.C. (1986),
Empirically derived subclassification of the autistic syndrome. J. Autism Dev. Disord., 16:
275-294
Szatmari, P. (1992), The validity of autistic spectrum disorders: A literature review. J.
Autism Dev. Disord., 22: 583-600.
Szatmari, P., Bartolucci, G. & Bremner, R. (1989), Asperger's syndrome and autism:
Comparison of early history and outcome. Dev. Med. Child Neurol., 31: 709-720.
Szatmari, P., Tuff, L., Finlayson, A.J., & Bartolucci, G. (1990), Asperger's syndrome and
Autism: Neurocognitive aspects. J. Am. Acad. ChildAdolesc. Psychiatry, 29: 130-136.
Wing, L. (1991), The relationship between Asperger's syndrome and Kanner's autism. In:
Autism and Asperger Syndrome. Frith, U., Ed. Cambridge: C.U.P.