A recent theory about the nature
of autism, the Theory of Mind hypothesis, suggests that autistic
people are seriously impaired in the capacity to think about one's
own and other persons' minds and to explain and predict behaviour
on the basis of these reflections. These problems in Theory of
Mind (henceforth: ToM) formation and use are suggested to be due
to a deficient metarepresentational mechanism (e.g. Leslie 1987).
This very mechanism is a cognitive data structure which allows
a person to think of and treat representations as mere representations
(e.g. Leslie and Roth 1993), an ability which is, among others,
needed in pretence and attributions of beliefs and thoughts to
other people. More in particular, the metarepresentational mechanism
is assumed to allow a person to form structures like in (1), (2)
and (3),
(1) 1 - pretend - this banana - "it is a telephone"
= the underlying data structure
needed for solitary pretence
(2) John - pretends- this banana "it is a telephone"
= the underlying data structure
needed for the understanding of pretence by others
(3) John - thinks - this banana - "it is a telephone"
= the underlying structure needed
for the belief attribution "John thinks that this banana
is a telephone"
where the verbs pretend
and think are interpretad as expressing a particular "informational
relation" or "attitude" towards the truth of the
description given between quotes.
In order to test this assumption
of a deficient metarepresentational mechanism, (and consequently
ToM mechanism ) in autism, one has run a number of false belief
attrlbution experiments, varying from tests where only first order
metarepresentation is needed as in (3) (e. g. Baron-Cohen et al.
1985), to tests which require more complex, second order metarepresentation
as in (4), where a subject has to be able to think about someone
else's thoughts about a third person's thoughts (e. g. Baron-Cohen
1989).
(4) John thinks that Mary thinks
that this banana is a telephone
In contrast to mentally retarded,
language-impaired and normal controls, most autistic people are
reported to fail these first and second order ToM tests (e. g.
Baron-Cohen et al. 1985, Baron-Cohen 1989), and only a small subset
of more intelligent and often older autistic subjects succeed
on both tasks (e.g. Bowler 1992,, Happé 1993). This profile
has given raise to two subhypotheses within the ToM framework:
a "strong" version which claims that autistic people
will never acquire metarepresentational and ToM skills (their
success on ToM tasks is ascribed to the use of alternative, non-ToM
strategies which result in a fake ToM) (e.g. Leshe 1994), and
a "weak" version which claims that some autistic people
may develop these skills on condition that they have a considerable
verbal mental age, although this development wjll always be different
and delayed (e. g. Baron-Cohen 1989). Both versions implicitly
suggest that the resulting ToM and metarepresentational skills
will be different.
Since the use of mental state
terms such as think plays an important role in the
discussion about autistic people's deficiency or ability in metarepresentational
and ToM skills, I have focussed on the spontaneous use in conversation
of to think and related items by four higher functioning
autistic young adults as compared to normal adult controls. More
in particular, I have looked at two different usages of the verb
to think, viz. one where the verb denotes (aspects
and characteristics of) being in the mental state or performing
the mental process of 'thinking', as in (5), and one where it
expresses an epistemic qualífication, as in (6), which
actually reports a person's evaluation of the probability of a
given State of Affairs (henceforth: SoA). This may be the speaker's
present evaluation (as in (6a)), in which case we speak of a 'performative'
epistemic evaluation, or someone else's evaluation (as in (6b)),
which will be referred to as a 'descriptive' epístemic
evaluation (see Nuyts 1994 for more information on this distinction).
(5) a. John is thinking about how to fix his car
b. Shut up, I am thinking
(6) a. I think that Mary will be home tomorrow
b. John thinks that Mary will
be home tomorrow
Note that in the literature on
metarepresentation and ToM in autism, only the so-called belief
attributions, in our terminology 'descriptive' epistemic qualifications,
have received attention so far. This study has, however, also
included 'performative' epistemic qualifications, since they too
require intact metarepresentational skills (see the resemblance
between (6a) and (l)) (for an extensiva discussion see De Roeck
and Nuyts (1994) and De Roeck (1995)). In addition, two other
expression types of epistemic modality have been looked at, viz.
epistemic modal adverbs as in (7) and auxiliaries as in (8).
(7) Probably Mary is at home
(8) Mary may be at home
The corpus used in this study
consists of a total of twenty hours recorded one-to-one conversations
between four young autistic adults and myself. These were casual
conversations, not structured on the basis of questions on my
part. In order to be able to evaluate the claim that autistic
people need a high level of verbal inteligence to develop metarepresentational
and ToM skills (see the "weak" version above), I have
chosen to work with four participants that had a different profile
of intelligence: S had a high verbal and nonverbal IQ, G a high
verbal and low nonverbal IQ, M a low verbal and high nonverbal
IQ and T a combined low verbal and nonverbal IQ. If the weak version
of the ToM hypothesís is true then only S and G could possibly
have developed the necessary skills. The control data (Nuyts
1994) were taken from a corpus of seven and a half hours of recorded
television conversations between non-autistic adults.
In short, the results of a fine-grained
analysis of both the epistemic expressions and the descriptions
of the process of thinking boil down to two major observations
(again details of this analysis can be found in De Roeck (1995)
and De Roeck and Nuyts (1994)):
(i) S, G and M showed an adequate
use of the items under consideration and have thus developed metarepresentational
skills which are fully normal, in T's case the ability is only
partially developed: she produces only 'descriptive' epistemic
qualifications, and her descriptions of the process of thinking
revealed a very idiosyncratic view of the thought process.
(ii) All subjects showed a significantly
high use of epistemic modal auxiliaries which results from a(n
extreme) tendency to perseverate. This tendency was also characteristic
of the subjects' conversational abilities in general, and may
suggest an executive functioning-like problem.
The problem with present ToM and
related theories of autism, however, is that they cannot fully
account for the findings in (i) and (ii). Let us start with the
two core ToM hypotheses already mentioned above. Leslie's "strong"
version of the ToM hypothesis is refuted by the results of S,
G and M, whose use of epistemic expressions shows intact, normal
metarepresentational skills: there are no indications whatsoever
that these skills are fake and deviant. T's case would fit the
theory more although the fact that 'performative' epistemic expressions
are more difficult than the 'descriptive' ones can not be explained
in this framework. As for the perseveration tendency, Leslie
could only interpret them as resulting from fake metarepresentational
and ToM skills, but since the latter assumption is not supported
by our results, this line of argumentation does not work.
Baron-Cohen's "weak"
version of the ToM hypothesis, on the other hand, does not rule
out that some autistic people (e.g. S, G, and M) may develop genuine
metarepresentational and ToM skills. But our data do not confirm
his assumptions about the necessity of high verbal abilities:
since T had exactly the same low(er) level of verbal intelligence
as M, and M showed intact metarepresentational abilities, I would
rather suggest that the ability for metarepresentation in autism
is dependent on general level of intelligence and not on level
of verbal intelligence (alone). In addition, the "weak"
ToM version can also not explain the difference between 'performative'
and 'descriptive' epistemic qualifications. The severe perseveration
could be attributed to the effect of later developed metarepresentational
skills, but since I have no information on the development of
these skills in S, G and M, the study cannot confirm nor refute
this claim.
A related though at the same time
distinct position is taken by Frith and Happé (e.g. Frith
and Happé 1994) who suggest that autism is the result of
a combination of ToM problems with a special cognitive style ("having
a weak central coherence"). The postulation of this additional
cognitive style does, however, not enable them to better the explain
the difference between 'performative' and 'descriptive' epistemic
qualifications (T's profile). On the other hand, a "weak
central coherence" might explain why our subjects perseverated
so much, were it not that the conversational abilities of some
of the subjects suggests a "strong" central coherence
as well.
Finally, the recent "executive
functioning" hypotheses (see e.g. Ozonoff et al. 1991a and
b, Hughes and Russell 1993, Harris 1993, Jarrold et al. 1994)
may seem the best candidates to explain the linguistic and cognitive
profiles of my subjects, in particular their extreme perseveration.
How the difference between 'performative' and 'descriptive' evaluations
should be accounted for is difficult to evaluate in terms of an
executive function deficit, however, and the possibility of intact
ToM abilities despite executive function problems is dealt with
differently by the respective proposals.
Since none of the above hypotheses
can fully explain the results of this study, I have developed
an alternative hypothesis which has integrated parts of the different
theories describes. In order to explain the profile of my four
autistíc subjects' linguistic behaviour, I have postulated
that a deficit in "x", being one of the processing and
reasoning principles by means of which a 'control mechanism' operates
with information, is probably the core problem in autism. It
is suggested that early damage to "x" affects the development
of both metarepresentational and 'executive function-like' skills,
and that a high level of intelligence helps one to compensate
for problems in metarepresentation but is less successful in remedying
the 'executive function-like' problems. What this implies for
the exact relationship between "x", metarepresentational
skills and 'executive functioning-like' skills is a question for
further study. Anyway, the advantage of a postulation of a deficit
in "x" is that it can easily explain why descriptive
epistemic evaluations are easier to make than performative ones.
Performative evaluations require the 'control mechanism' to actively
select and compare relevant information about a SoA in order to
arrive at a conclusion regarding the probabjlity of that SoA.
Descriptive evaluations, on the other hand, are not the result
of an active operation by the 'control mechanism' at the moment
of speaking, rather, they involve a report on evaluations which
are stored in conceptual knowledge. A deficit in one of the 'control
mechanism's' processing and reasoning principle will thus be less
damaging to those processes which do not require the active involvement
of the mechanism.
References:
Baron-Cohen, Simon
1989 The autistic child's theory of mind: A case for speciflc developmental delay. Journal of Child Psychology and Psychiatrv, 30, 285-297.
Baron-Cohen, S., A.M. Leshe and U. Frith
1985 Does the autistic child have a "theory of mind"? Cognition, 21,
37-46.
Bowler, D. M.
1992 "Theory of mind" in Aspergerls syndrome. Journal of Child Psychology and Psychiatry, 33/5, 877-893.
Frith, U. and F. Happé
1994 Autism: beyond "theory of mind". Cognition, SO, 115-132.
Happé, F.
1993 Communicative competence and theory of mind in autism: A test
of relevance theory. Cognition, 48, 101-119.
Harris, P.
1993 Pretending and planning. In S. Baron-Cohen, H. TagerFlusberg and D.J. Cohen (eds.), Understanding other minds.
Oxford: Oxford Uníversity Press., 228-245.
Hughes, C. and J. Russell
1993 Autistic children's difficulty with mental disengagement from an object: Its implications for theories of autism.
Developmental Psychology, 29, 498-510.
Jarrold, C., P. Smith, J. Boucher and P. Harris
1994 Comprehensíon of pretense in children with autism. Journal of
Autism and Developmental Disorders, 24, 433-455.
Leslie, A. M.
1987 Pretense and representation: The origins of "theory of mind".
PUchological Revíew, 94, 412-426.
1994 Pretending and believing: issues in the theory of ToMM.
Cognition, 50, 211-238.
Leslie, A.M. and D. Roth
1993 What autism teaches us about metarepresentation. In S. BaronCohen, H. Tager-Flusberg and D.J. Cohen (eds.), Understanding other minds. Oxford: Oxford Universíty Press, 83-111.
Nuyts, J.
1994 Epistemic modal qualifications: On their linguistic and conceptual structure. Wilrijk: Antwerp Papers in Linguistics 81.
Ozonoff, S., B.F. Penníngton and S.J. Rogers (referred to as Ozonoff et al. 199la)
1991 Executive function deficits in high-functioning autistic individuals: Relationship to theory of mind. Journal of Child Psychology and Psychiatry, 32/7, 1081-1105.
Ozonoff, S., S.J. Rogers and B.F. Pennington (referred to as Ozonoff et al. 1991b)
1991 Asperger's syndrome: Evidence of an empirical distinction from high-functioning autism. Journal of Child Psychology and Psychiatry, 32/7, 1107-1122.
Roeck, A. de
1995 Metarepresentation in autistic language use: A test case for cognitive theories of autism. PhD-thesís, University of Antwerp.
Roeck, A. de and J. Nuyts
1994 Epistemic modal expressions by high-functioning autistic
adults: A test case for the "theory of mind" hypothesis.
Wilrijk: Antwerp Papers in Linguistics 82.