AUTISM - CURRENT ISSUES 4

The issues covered in this review of recently published research evidence include support for the significance of social perception as a characteristic area of weakness among children with Autism rather than an overall attention deficit; the linkage between Theory of Mind skills and language maturity; patterns of language development; family studies of autistic probands; and a reminder of the range of needs that may all fall within an autistic spectrum with implications for access to an equally wide-ranging series of (educational) provisions

M.J.Connor February 1998

Social Perception and Attention

When Autism was first examined as a specific syndrome (see Kanner 1943), there was reference to social weaknesses, and non-responsivity to other people or events in the environment, as defining characteristics; and Pierce et al (1997) comment that deficits in social behaviour have been regarded as primary in all clinical descriptions of Autism since its initial identification.

Meanwhile, converging evidence indicates that people with Autism commonly show limitations in attention, and it may be the case that this type of weakness underlies the characteristic functioning in other areas, notably social perception. The individual with Autism may have difficulty in the orienting, shifting, or sustaining of attention. For example, Courchesne et al (1994) found that autistic adults detected fewer visual or auditory stimuli, compared to control subjects, when required to shift attention rapidly between the two sensory modalities.

Further research evidence reviewed by Pierce et al (opp. cit) has indicated that other attentional abnormalities may arise during social interaction. In particular, the person with Autism may be unable to attend to both a partner and to environmental stimuli; but the ability to shift attention to various elements of social situations is achieved by normal children by around 15 months of age and is regarded as a critical element of social development.

One notes, too, the concept of "stimulus overselectivity" to describe the way in which children with Autism appear to respond to only a part of a stimulus array, whether the stimulus is visual, tactile, verbal, or social. The significant point is that this overselectivity may affect the behaviour of autistic children in day to day settings in that responsivity may decrease as the number or complexity of stimuli increase.

Studies of the ability to process social information (see, for example, Baron-Cohen 1989) have demonstrated how children with Autism have difficulty in recognising facial, body-language, or vocal cues about other people's feelings; in matching or coordinating cues from different sources; and in their ability both to express emotion or to attribute mental states in other people. The Theory of Mind approach holds that the communicative, social, and imaginative deficits result from the inability to perceive and attribute feelings, opinions, points of view, knowledge, etc.. However, an alternative hypothesis is that the poor performance in such areas reflects a critical weakness in "multiple" attention.

Accordingly, the research completed by Pierce et al themselves was designed to explore this attentional deficit, how it might have an impact upon social perception, by means of a task involving social cues ranging number from one to four. The hypothesis was that better performance among the autistic sample would be associated with the single cue condition.

In the light of this dichotomy, they argue that the finding of equal performance among the 3 groups during task conditions of low attentional demand (single cues), but unequal performance during task conditions of high attentional demand (multi-cues), would suggest that the level of complexity in the environment (the number of cues available) is significant for the social perception and interpretation skills of the child with Autism.

In other words, when attentional requirements are high, individuals with Autism are likely to have difficulty in responding to what is happening around them.

An alternative hypothesis (see Dawson and Lewy 1989) is that autistic individuals have difficulties in maintaining the optimal level of arousal and this, in turn, negatively influences their ability to attend to social information. When stimulus conditions are complex (as in social interactions), there is an over-arousal, and the individual may attempt to reduce this uncomfortable state by reducing attention to that stimulus situation and focusing instead upon an alternative, narrow, and less "threatening" one.

Further, and in respect of the Theory of Mind, it is held that the characteristic failure among children with Autism to cope with the test tasks such as "false belief") may be at least partially a matter of exposure to attentional demands which are excessive.

Pierce et al (opp.cit) conclude by underlining the social - attentional deficit linkage, while acknowledging that the attentional limitation may interact with other types of deficit (notably in the affective domain) to produce a multi-component model of Autism. In any event, they suggest that there may be considerable significance for intervention strategies in Autism from further examination of patterns of attention in day to day settings, with a view to determining how attention can be maximised (or the setting modified to minimise attentional strain or over-arousal).

Awareness of Mental States in Self and Others, and Verbal Ability

The work of Kazak et al (1997) continues the theme of social perception and is based upon the well established finding that many individuals with Autism have a specific impairment in their ability to recognise mental states .... feelings, knowledge, etc. ... themselves as well as in other people.

However, it has also been demonstrated (see, for example, Happe 1995) that some people with Autism are able to attribute mental states, and it has been argued that this ability is a characteristic shown by individuals with relatively highly developed language skills.

The kind of tasks used to test the ability to recognise and attribute mental states commonly involve false belief tasks and the prediction of what another person will do. The concern, as expressed by Hogrefe et al (1986), is that such tasks underestimate this ability in people with Autism since normal children may be unable to deal correctly with false belief tasks even if they are shown to be well able to understand what another person knows or does not know.

An additional element involved the comparison of children with Autism with control samples of children with mental handicap and children with no learning or other difficulties.

14 children were included in each group, and were matched on verbal mental age as assessed by the Peabody Picture Vocabulary Test. The chronological age range therefore, was very wide, with means of 114 months, 120 months and 63 months respectively in the autistic, mentally handicapped, and normal groups.

The task involved watching a series of videotaped "vignettes" of child - child interactions, either positive or negative, in which some observable action took place. For example, one child gives a present to the other. Each vignette contained from 1 to 4 cues to demonstrate the extent to which the first child is engaging in positive or negative behaviour, and the feelings of the other child.

The cues were of 4 types ..... verbal content (where the child might say I like that toy"); verbal tone or level of animation; nonverbal (e.g. smiling); and non-verbal with object (e.g. the offering of a present).

Subsequently, the participants were asked a series of 6 questions to highlight the extent to which they had attended to, and could interpret, the material. The first two were simple and factual such as "How many people were there?" "Were they boys or girls?" and showed whether or not the children had paid attention. Subsequently, the videotape was shown again to ensure that memory would not be a confusing variable, and was paused on the final frame.

Social perception was examined by questions about the first child "Was that a good way to make friends?"; "Was that child nice or mean?". Then, in respect of the other child, the questions were "How does that child feel?"; " Why does he feel that way?".

The overall results showed that the autistic participants performed more poorly than either of the other two groups on questions regarding social interactions. However, more detailed analysis revealed that the number of cues was significant in that the autistic group performance was no different when dealing with questions 3 to 5 on single cue vignettes, but significantly worse than either comparison group on questions 3 to 5 in multiple cue vignettes. The autistic group also performed significantly worse in question 6 ("Why does that child feel that way?") irrespective of the number of cues.

The authors draw a number of implications, including the support gained for the view of poor social perception/interpretation as a characteristic autistic feature, albeit with evidence for the significance of the number of cues ( and range of attention).

It is noted that poor performance on questions relating to social cues was not simply a matter of poor overall attention to the stimulus material.

The authors then restate the views that

1. If children with Autism have a social deficit, then they would interpret the social environment the better when there are multiple cues present, because the additional and redundant information should make the task easier.

2. If; however, children with Autism have an attentional deficit, then interpretation of their social environment would be easier when there are minimal requirements for shifting attention or integrating cues from different sources.

Kazak et al (opp.cit) highlight this lack of a normally developing group of children as controls in the Theory of Mind experiments as a significant weakness. Typically, tasks assessing the child's understanding of "know" have only involved children identified as autistic; or, where there have been included samples of children from mainstream schools, there is insufficient information about the IQ scores of the children. They also focus upon the typical requirement of the experimental tasks as that of expecting the children to predict some action which goes beyond simply recognising mental states. For example, reference is made to one study (Leslie and Frith 1988) which showed that 61% of their sample of children with Autism could refer to another person's state of knowledge, but only 41% could make any prediction about that person's behaviour.

Therefore, the first aim of the study completed by these current authors was to investigate the ability of young people with Autism to refer to knowledge states (i.e. does someone know the situation or does that person have to guess?) without any need to infer that ability from the subjects' predictions of behaviour.

A further aim was that of investigating the role of language ability (verbal mental age) as a predictor of the ability to attribute mental states. Existing studies suggest that some language abilities (such as vocabulary knowledge, comprehension skill, and pragmatic skill) may differentiate children who have a higher probability of succeeding in false belief tasks from those with a low probability. However, the vocabulary subtest of the Wechsler Intelligence Scale appears not to differentiate in this way.

Accordingly, the participants were tested on the British Picture Vocabulary Test to gain a measure of simple comprehension of words; Renfrew Action Picture Test to measure grammatical skill and information access; and the Test of Reception of Grammar.

5 groups were formed; 2 groups of children with Autism, divided according to higher (6+ to 15+ vocabulary age) or lower ( 3+ to 5+) scores on the BPVS. Nl=N2=14. The mean actual age of the groups were 13.0 and 10.0 respectively.

2 groups of mainstream school children, 4 year olds, and 5 year olds. NI=17, N2=22.

1 group of Down's Syndrome children. N=17. Mean age 11.0.

In the first task, the experimenter showed the children boxes of red, green, and blue marbles, and then chose one to put into an individual box to give to a friend. The children either watched this action or were asked to close their eyes while the experimenter chose a marble. They were then asked if they knew the colour chosen or had to guess.

In the second task, the child observed while the experimenter worked with a third person (who either followed the action, or closed her eyes) and the child was asked whether this third person knew the colour of the marble chosen or had to guess.

The first prediction was that the autistic group would be poorer than controls at discriminating between "Know" and "Guess" when referring either to themselves or to others (the third person); it was also predicted that the autistic children with the higher language abilities would perform better than those with the lower language level.

The data did indeed show that the "lower language" autistic sample did perform in a limited way ... no better than at a chance level. Meanwhile, the autistics with the better language scores performed well above chance and comparably with the 5 year old mainstream children. The second prediction was supported, but not the first.

The authors posed the question as to which language test should be used as a basis for matching groups since the high language autistics were similar to the 5 year old mainstream children on the RAPT tests but very different on the BPVS.

The issue is whether the use of the BPVS as a discriminator is valid or simplistic given the implication that individuals with Autism have a different profile from normal individuals on a wide range of language tests which afford the opportunity to measure different aspects of language. In other words, results or their interpretation may vary according to which language tests are used as matching measures.

The observed results from this current study also fall to support the view that young people with Autism can be differentiated from young people with Down's Syndrome in respect of a deficit or delay in their understanding of mental states. No differences were observed in the understanding of "know" and "guess" among the lower-language autistic group, the 4 year old mainstream group, and the Down's group.

However, it was speculated that the results observed may be due to some task-specific features affecting the performance of the Down's children and the mainstream 4 year olds in different ways to the impact upon the performance of the low-level autistic group. For example, it may be that a significant factor is the precise nature of the questions posed to the subjects simple or complex, single- or double-barrelled. One needs further to explore whether the children actually possess the skills under consideration, but cannot demonstrate those skills because of problems in the language used in the questions or in the level of language skills likely to be necessary in producing responses to the questions.

In any event, Kazak et al (opp. cit) interpret similarities in performance among these groups as indicating that there may exist a general relationship between language skills and Theory of Mind ability. Language skill and Theory of Mind ability may have some underlying common mechanism, or there may be two separate mechanisms that develop simultaneously but independently.

However, it was found that the correlation between the subjects' ability to refer to their own and to other people's knowledge state, and their performance on the different language tests, was highly significant for the autistic children, but not for mainstream or Down's controls. The conclusion was that the role of language skill in Theory of Mind tasks appears specific to Autism rather than a reflection of overall cognitive ability.. and this. supports the view that language level is an important factor in determining the performance of young people with Autism on Theory of Mind types of tasks.

The overall conclusion was that, on the basis of current findings, one might hypothesize that, in Autism, language and Theory of Mind mechanisms have a common, or very similar, organic basis that is quite different from that in normal and Down's Syndrome individuals. Further, this may explain why a language deficit in general is not sufficient to account for Autism.

Autism and Language Profiles

In the light of the questions raised about language skills in children with Autism (and the nature of language measures to be used for comparison purposes), it is timely to have access to the work of Jarrold et al (1997) who examined theoretical and methodological implications in respect of observed language performance among autistic subjects.

They refer to impaired language use as one of the essential and diagnostic characteristics of Autism; and note that those more able children with Autism who do acquire language do so at a slow rate and may only achieve some (plateau) level of language achievement below what might be expected on the basis of estimates of non-verbal ability. Particular problems are noted with more subtle language usage and there may be an inability to look beyond literal meanings. Further, specific developmental language disorders may coexist with Autism thus creating even more complex patterns of linguistic acquisition or non-acquisition.

Nevertheless, the authors argue that, even with the information that is available about language systems, there has not been set down any typical profile of language attainment among autistic children.

They suggest that there are at least 3 ways in which the language profile of those autistic children with speech has been found to be abnormal.

1. There is an imbalance in that articulation skill is advanced compared to other linguistic components (see, for example, Bartolucci and Pierce 1977).

2. Expressive language may be more developed than comprehension.... even if one must recognise the frequency of echolalic and repetitive/formulaic speech.

3. Understanding of vocabulary may be superior to the comprehension of grammatical structures and of sentences. (High functioning autistics may perform well on tests of vocabulary comprehension, and Paul [1987] has noted that children with Autism may develop large vocabularies, and even take an obsessive interest in words).

Theoretical implications from observed patterns of language attainment include ways in which types of Autism might be differentiated. For example, the primary deficits underlying Asperger Syndrome, where language attainment are not significantly impaired, may not be the same as those linked to "classic" Autism. Does Autism, in at least some cases, involve a deficit linked to the acquisition of language, notably grammatical structures? Alternatively, do non-linguistic factors underlie the commonly observed delays in acquiring a meaningful language system in Autism?

What matters is that language measures are very frequently used in Autism research to produce experimental and control groups matched for language ability; and it may be assumed that a single measure, such as the British Picture Vocabulary Scale, will suffice to control for spoken language ability. However, if the autistic sample show marked differences between their understanding of vocabulary and their understanding of grammar or whole sentences, then the use of the BPVS will put the autistic sample at a disadvantage if they are exposed to instructions which are expressed in whole sentences or if they are required to respond verbally at any length.

In their own study, Jarrold et al (opp. cit) compared the performance of children with Autism on 4 standardized tests.... the BPVS, the Test of Reception of Grammar (TROG), the Renfrew Word Finding Vocabulary Scale, and the Action Picture Test Information and Grammar Scales.

Given their view of the significance of generative language, they predicted that expressive language would not be found to be superior in any meaningful way to comprehension, that comprehension of vocabulary would be superior to the comprehension of grammar, and that production of vocabulary would be superior to production of grammatical sentences.

The results indicated that language attainment profiles, excluding measures of articulation, were very "flat". Minimal differences were observed among scores on the BPVS (comprehension of single words), TROG ( comprehension of morphology and syntax), and APTI/APTG (measures of productive use of substantive words and of morphology/syntax).

Therefore, it was confirmed that the generative use of expressive language (as measured by the APTG) was not superior to language comprehension (as measured by TROG). However, the relationship between the production and the comprehension of single words was contrary to the prediction. Renfrew Word Finding scores were higher than those on the BPVS scores and on both aspects of the Action Picture Test. It is noted that the Word Finding Test assesses only object naming, while both the BPVS and APT assess knowledge of a wider range of vocabulary items.

Therefore, it may be that children with Autism have object naming ability somewhat in advance of other language skills, especially productive grammer.... although it is acknowledged that the evidence is barely significant, and the results might be an artefact of test standarization problems (given that significant results with regard to the APT were gained on the earlier edition but not on a later edition).

In any event, the authors accept that there was no support for the predictions that comprehension of vocabulary would be better than comprehension of grammar, or that production of vocabulary would be better than the production of sentences. To explain why these findings are inconsistent with other research data, they suggest that there are differences between the tests of comprehension of grammar used. The TROG test may be easier in terms of pragmatic language requirements than tests typically used in previous research (such as the Test for Auditory Comprehension).

It is also speculated that differences may emerge from the unequal regard to other interfering variables, such as memory load.

Finally, the authors suggest that the failure to find the expected discrepancy between attainment in vocabulary and grammar may be explained in terms of the heterogeneous nature of their group, and that this very heterogeneity masked the fact that a sub-sample of children with Autism did indeed have vocabulary significantly in advance of grammar.

Therefore, it is accepted that the observed results must be interpreted with caution, but it appears that language acquisition across the whole of their identified autistic group develops in a broadly similar way.

This was considered worthy of attention (and further investigation) given that one of the key distinguishing factors between Autism and Asperger Syndrome is the presence or absence of early language.

In summary, Jarrold et al (opp.cit) describe the implications of their findings that Autism and Asperger appear not to be differentiable in terms of language profile or of the presence of a discrepancy between verbal and non-verbal abilities. This is not to say that language abilities do not differ across these two groups, as there may be a readily observable range of levels of delay, but the manner of language acquisition appears similar.

There is also support for the commonly held view that impairments within the acquisition of language systems which are characteristic of Autism or Asperger Syndrome are the result of cognitive or social deficits rather than specifically linguistic deficits. Had specifically linguistic deficits been implicated, one would have anticipated more deviant language attainment profiles.

A practical implication is for the use of any well-standardized test of vocabulary or grammar comprehension or production since the results from one such test may reliably predict the results from another.

So, it is acknowledged that the use of such a test as the BPVS for producing equitable groups is justified, although it must be recognised that no language attainment test can control for all the factors which may interact in contributing to the understanding and use of speech among children with Autism... notably knowledge of pragmatics or the recognition of non-verbal signals and prosody to supplement comprehension.

Finally, the authors stress that while their findings might be taken to suggest that no single aspect of language is in particular need of remediation compared to any other, it is still the case that not all individuals will show a flat language profile, and it is necessary to base any intervention upon an individual assessment of the child's language strengths and weaknesses.

Family Study of Autism - Parents and Siblings

Over the recent past, evidence has consistently suggested that there is a strong genetic influence in the development of Autism. For example, twin studies such as that of Bailey et al (l995) have shown that there is a far higher concordance rate for Autism for monozygotic pairs than for dizygotic pairs.

The review of studies by Fombonne et al (1997) provides further support for this genetic aetiology although it is noted that there is some uncertainty about precisely what it is that is inherited. Within families with an autistic proband, there have been deficits reported in respect of personality functioning in the parents, in Theory of Mind skills or central coherence also among the parents, and in executive functioning among siblings.

In studies of intellectual functioning, reference is made to Minton et al (1982 ) who reported that 7% of full scale IQ scores in a sample of siblings of autistic children were within the mild mental retardation range, while subtest discrepancies tended to be in favour of non-verbal items. However, the small sample size meant that results had to be treated with caution.

Larger scale studies have indicated an increased rate of pervasive developmental disorders among siblings of autistic children, but not a raised frequency of learning difficulties. Results in respect of more subtle cognitive abnormalities such as verbal-performance discrepancies, IQ subtest patterns, and scores on literacy tests, appear more equivocal.

The study completed by Fombonne et al (opp.cit) investigated these issues by means of interviewing and psychometric testing of parents and siblings of a large sample of children with Autism (omitting from the sample those with associated medical disorders such as Tuberous Sclerosis, and those with very low measured IQs), with parents and siblings of Down's probands acting as a control group.

In summary, the results indicated that the socio-economic status of the families involved (as defined by the occupation of the head of the family) was different from what might have been expected from population norms. There was a higher proportion of non-manual occupations and fewer manual or unskilled occupations than chance would have predicted.... and this is line with the pattern commonly observed among referred samples.

In intellectual functioning, the relatives of the autistic probands showed a distribution of scores closely matching a normal distribution, and parents' scores showed a mean in the high average range. However, among the parents in the autistic group, there were higher scores for 4 of the 5 WISC verbal tests and for verbal IQ as compared with the scores typically gained by the Down's group of parents (with differences indicating a generally depressed mean verbal score among the latter parents as much as enhanced mean scores among the parents of the autistic children). However, the results did not reach a statistically significant level.

Further analysis indicated that, in respect of verbal-nonverbal discrepancies, a significantly higher proportion of parents and siblings of autistic probands, compared to the control relatives, had discrepancies in favour of verbal scores.

In tests of reading and spelling, no significant differences were observed in performance whether using a 10th. or 5th. centile cut-off point as a criterion marker.

The authors argue that, taken in conjunction with the outcome of other similar studies (such as Szatmari et al 1993), one might legitimately give weight to the finding of no increased incidence of mental retardation among first-degree relatives of autistic probands (once siblings with Autism or Atypical-Autism" were excluded). However, they acknowledge that the study did not include autistic probands in the very severely retarded range such that generalisability of findings should not be fully assumed.

Comparisons between the relatives of the autistic sample and control probands did not produce any further results indicative of a raised incidence of mild cognitive deficits among the Autism relatives. In fact, what differences emerged, in respect of WISC verbal scores or literacy tests, were in favour of this group.

Further, no increased rate of verbal-performance discrepancies were found in this group) suggesting that this relatively specific pattern found among autistic individuals is not part of what is transmitted to non-autistic relatives.

The findings on the standardised tests of reading and spelling were also largely negative... so, for example, there was no observed trend towards an increase in literacy difficulties in the families of individuals with Autism.

In the light of the fact that language abnormalities constitute a large part of the broader phenotype of Autism (i.e. abnormality in only 1 or 2 of the areas of communication, social interaction, and behaviour [ restricted and repetitive] ), and because traditionally diagnosed autistics have such a distinctive profile, the authors suggest that those siblings of an autistic proband who could be said to fall within the broader phenotype would be likely to show a characteristic cognitive pattern. However, the current findings do not support this expectation. There may have been lower scores on the WISC verbal and full scale IQs, and a greater incidence of literacy difficulties, but the differences were too small to be of diagnostic significance.

The authors conclude by restating their finding of some mildly superior verbal skills among unaffected relatives of autistic probands. Caveats include the relatively small size of this advantage, and the not fully representative (socio-economically speaking) sample of the Autism families. Nevertheless, there are reasons for believing that this finding is valid. For example, parental group differences on verbal-performance discrepancies, and on verbal IQ, remained even after controlling for parental education and occupation.

Accordingly, it is suggested that this area remains "unproven" and that the possibility of some verbal talent among the relatives of autistic individuals exists. Reference is made to the comments of Kanner (1943) about the apparent cognitive advantages among the parents of autistic children; and to the possibility, described by Rotter and Diamond (1987), that a genetic liability for a disorder may be associated with compensatory benefits.

The final advice is that more research is needed, to confirm or refute the suggestion that there are verbal talents among unaffected relatives, before any further theoretical speculation!

Miscellaneous Issues National Autistic Society

The following brief items were drawn from the NAS journal "Communication of the Winter 1997

2. There has been renewed concern about a possible link between the use of the measles, mumps, and rubella vaccination (MMR) and Autism. Opinions and conclusions drawn from available cases are divided, and a current study at the Royal Free Hospital is examining possible links between MMR and Autism ( and Crohn's Disease or colitis). Further research is to be monitored given the dilemma over possible risk in the vaccination alongside the existing and serious risk associated with measles, mumps, and rubella.

3. In respect of educational provision, it is argued that many appropriate services exist, contrary to some parental beliefs aired in national media that children on the autistic spectrum have access to little provision. The advice is that a range of options is necessary to match the range of difficulties observed among the children such that places in mainstream schools, units attached to mainstream schools, and specialist schools should coexist. The critical needs are for specialist understanding of the nature of Autism/Asperger Syndrome, regular reviewing of the educational programme, and consistent communication among all staff concerned and parents.

Particular concern is expressed about the child's ability to integrate within a mainstream secondary school where there are increased demands upon the child in terms of generalisation, analysis, etc., alongside the complex social and interactional circumstances which may prove threatening. As Autism is a spectrum disorder affecting children across the ability range, and as each child will have his/her own profile of strengths and weaknesses, one needs to accept that what is appropriate for one child in the way of educational placement may not be appropriate for another. Further, one must recognise that needs may change over time... although what appears to be consistent is the requirement for a learning environment which is regular, predictable, and structured. Over stimulation ( e.g. excess language or over complexity of language) is to be avoided so that the child can anticipate and interpret the environment and not be exposed to high levels of stress.

REFERENCES

Bailey A., Le Couteur A, Gottesman I., et al (1995) Autism as a strongly genetic disorder. Psychological Medicine 25 63-77.

Baron-Cohen S. (1989) Are autistic children behaviourists? Journal of Autism and Developmental Disorders 19 579-600.

Bartolucci G. and Pierce S. (1977) A preliminary comparison of phonological development in autistic, normal, and mentally retarded subjects. British Journal of Disorders of Communication 12 137-147.

Courchesne E., Townsend J., Akshoomoff N., et al (1994) Impairment in shifting attention in autistic and cerebellar patients. Behavioural Neuroscience 108 1-17.

Dawson G. and Lewy A. (1989) Arousal, attention, and the socio-emotional impairments of individuals with Autism. In Dawson G. (Ed) Autism Nature, Diagnosis, and Treatment. New York, Guilford.

Fombonne E., Bolton P., Prior J., Jordan H., and Rutter M. (1997) A family study of Autism. Journal of Child Psychology and Psychiatry 38(6) 667-683.

Happe F. (1995) The role of age and verbal ability in the theory of mind task performance of subjects with Autism. Child Development 66 843-856.

Hogrefe G., Wimmer H., and Perner J. (1986) Ignorance versus false belief Child Development 57 567-582.

Jarrold C., Boucher J., and Russell 3. (1997) Language profiles in children with Autism. Autism 1(1) 57-76.

Kanner L. (1943) Autistic disturbance of affective contact. Nervous Child 3 217-250.

Kazak S., Collis G., and Lewis V. (1997) Can young people with Autism refer to knowledge states? Journal of Child Psychology and Psychiatry 38(8) 1001-1009.

Leslie A. and Frith U. (1988) Autistic children's understanding of seeing, knowing, and believing. British Journal of Developmental Psychology 6 315-324.

Minton J., Campbell M., Green W., Jennings S., and Samit C. (1982) Cognitive assessment of siblings of autistic children. Journal of the Academy of Child Psychiatry 21 256-261.

Paul R. (1987) Communication. In Cohen D., Donnellan A., and Paul R. (Eds) Handbook of Autism and Pervasive Developmental Disorders. New York, Wiley.

Pierce P., Glad K., and Schreibman L. (1997) Social perception in children with Autism an attentional deficit? Journal of Autism and Developmental Disorders 27(3) 265-282.

Rotter J, and Diamond J. (1987) What maintains the frequencies of human genetic disease? Nature 329 289-290.

Szatmati P., Jones M., Tuff L., et al (1 993) Lack of cognitive impairment in first degree relatives of children with pervasive developmental disorder. Journal of the Academy of Child and Adolescent Psychiatry 32 1264-1273.