The main hypothesis was that the children's performances on EF and theory of mind tasks would be positively correlated, and the results supported this view. With age-related effects partialled out, significant correlations were found between working memory and false belief prediction, and between inhibitory control and deceit / false belief explanation. It was not simply the case that associations between EF and theory of mind could be explained in terms of common associations with verbal and nonverbal ability.

There was support for the view that improvements in theory of mind skills across the pre-school years reflect growing strategic abilities, with the results suggesting that age-related improvements in deception skill could be explained by co-occurring improvements in inhibitory control. It is held that this finding might explain some of the conflicting outcomes in studies concerning the age at which children can be successful "deceivers"... i.e. the intent to deceive may develop quite early, but children may take considerably longer to acquire the skill to carry out the intent.

There was no support for the suggestion that improvements on false belief tasks are mediated via enhanced executive functioning; rather, it appeared that deception, and false belief skills, show different patterns of association with aspects of executive functioning. For example, inhibitory control was significantly more correlated with deception than with false belief prediction, and it may prove significant that deception (unlike false belief prediction) involves concealing information or providing misleading information, so that deception does not conform to the usual pattern of communication whereby one provides useful information believed to be correct.

Hughes' conclusion is that children's theories of mind are multi-facetted and involve different types of skill. Individual differences in theory of mind development are not simply a matter of rate but of style. Some children may gain an awareness of mental states by a route which is closely related to language development; others may reach the same awareness by means of more practical routes, such as active engagement in deception tasks. The possibility of multiple routes to the acquisition of an adequate theory of mind has already been suggested in research evidence concerning children with Autism .... for example, Frith et al (1991) have found that a large minority of individuals with autism succeed on false belief tasks despite having otherwise very poor social interaction skills.

In sum, the study suggests that several precursors may be involved in the development of theory of mind; and that both executive function and theory of mind are complex constructs with specific rather than general relationships between the two areas.

Eye Gaze and Mind Reading

Continuing with the theme of developing a theory of mind, one may refer to the work of Lee et al (1998) who introduce their work by describing how increasing significance has been attached to the role of (the direction of) eye gaze in social interaction. In animals, eye gaze may establish dominance, initiate or terminate aggressive behaviours, indicate the source of food, and signal approaching danger.

The role of eye gaze has been extended in human interactions, and it has been suggested (Baron-Cohen 1994) that a uniquely human function of eye gaze and eye gaze monitoring is the revealing of mental state ... mind reading, or provision of clues about the focus of another person's attention, desire, or belief. The hypothesis continues that some lack of sensitivity to eye gaze is related to impairments in social and cognitive abilities, such as are observed in Autism.

The review of studies completed by Lee et al (opp.cit) demonstrates that the development of theory of mind undergoes a shift at about 4 years of age, when children rapidly recognise the beliefs of other people (and appreciate that other people may not immediately recognise their perspectives). 3 year olds have problems with such understanding.

One hypothesis is that appropriate development of understanding about beliefs (and false beliefs) depends upon a more basic awareness, namely the recognition of the wants of another person. It is held that a shared attention mechanism enables children to work out what other people want through the use of eye gaze information. According to Baron-Cohen, this shared attention mechanism evolves from a developing awareness of intentionality, and gaze direction monitoring, both of which appear by the end of the first year of life. The net effect is to enable children to identity the object of a person's attention, and infer why the person is attending to it.

Existing studies indicate that dyadic eye gaze, whose main function is to regulate face to face social interaction, (to determine if the other person is looking at the child or elsewhere) can be observed from as young as 2-3 months. However, triadic gaze, which involves identifying at whom or what another person is looking, emerges rather later, at around 6 months.

Infants of around 12 months begin to use an adult's eye gaze to establish joint attention, and by about 18 months are accurate in identifying the other person's focus of attention. By this age, children use not only triadic gaze, but also various directional cues (such as pointing, or head orientation) for enhancing communication, such as in labelling objects and learning new words.

However, there is little data concerning when or how children develop triadic gaze into mind reading skills.

The existing study of Baron-Cohen (1995) compared autistic 4 year olds with normal controls in their ability to detect what a cartoon character wanted by means of using his eye gaze to select one of 4 sweets. The ordinary 4 year olds had no difficulty with this task, but the children with Autism could not identify which sweet the character wanted.

Lee et al's own study replicated and extended this initial work. For example, their first study, using samples of children in various age groups from 3+ to 6+, involved a similar task of identifying what a cartoon character wanted, but this time, there were 6 different items, to avoid response bias, and the children were asked "what does Larry want ?" as well as "where is he looking?" and "what is he looking at?"

Results indicated that 3 year olds could cope easily with what and where questions, but had difficulty with the want questions. They could not draw the inference from the direction of gaze, while 4 year olds performed significantly better than chance level in this task. It was hypothesised that the task may have been over-abstract, and that it was not made clear to the children why "Larry" wanted anything.

Therefore, a further study was completed in which Larry is part of a story in which he goes shopping with his mum for a birthday present. In a given shop, Larry is described as deciding what he wants, and is pictured signaling by means of eye gaze at 1 of 4 toys or 1 of 4 pets.

The expectation was that the younger children would perform better in answering questions about what Larry wanted. However, despite the provision of a meaningful context, the majority of 3 year olds still failed to answer correctly the "want" questions. They were more successful at following the direction of Larry's gaze, but still did not draw an inference about what he wanted.

Subsequently, there was an exploration whether the 3 year olds were experiencing a more general mind reading difficulty and could not link any non verbal cues displayed by an individual to what was wanted. In this task, the children were required to identity the direction of pointing or head direction or eye gaze and infer what was wanted. (Further, there was an additional condition in which the cartoon character was linked to an object by an arrow.)

The results here indicated that most children correctly used one of the direction cues, and that 3 year olds were more accurate than chance in using the pointing or head direction cues in inferring what was wanted. Therefore, there was no general difficulty in liking nonverbal cues to an individual's mental state... rather, the 3 year olds failed to use the gaze or arrow cues, showing a preference for a gestural cue instead of an abstract or symbolic cue.

It was then hypothesised that the 3 year olds' weakness was not so much in making inferences but in making use of eye gaze information in static or "impoverished" contexts. Accordingly, a study was completed in which a TV character demonstrated some desired object by various nonverbal cues, initially using pointing and head direction and eye gaze, then the eye cue and one other, and finally only the eye cue.

Results here indicated that, unlike in the previous tasks, the 3 year olds could use eye gaze to infer what wanted. However, when cues were conflicting, it was found that different levels of importance appeared to be attached to the 3 forms of cues pointing was more salient than eye gaze, but eye gaze was more salient than head direction. In any event, it was concluded that eye contact could be used by 3 year olds to infer wishes when the cue was put into an "enriched" context.

The overall summary highlighted how the use of nonverbal directional cues to infer another person's wishes is present in very young children... as young as 2+ if the stimulus conditions are sufficiently dynamic. There is support for the onset of a mind reading capacity at around 3 years of age.

However, the shared attention mechanism postulated by Baron-Cohen is held to be incomplete. The current findings suggest that young children can make use of eye gaze for desire inference, but that they are reliant upon other nonverbal cues such as pointing or head direction, the use of which emerge rather earlier than that of eye gaze.

The implication is for the importance of contextual information in mind reading of other people's wishes. Desire inference is held to be achieved by means of multiple sources of information, both verbal and nonverbal. Using attentional cues, particularly eye gaze, is only one part of the strategy adopted by young children.

The suggestion continues that initially all attentional cues are grouped together but, with experience, cues are differentiated. The most reliable cue... pointing... is the first to be used on its own; while a cue such as eye gaze comes later because it is less reliable or may carry different meanings according to context. For example, a person gazing upward towards no one item may be considered to be deep in thought, or a downward gaze while talking to someone might be taken to indicate guilt or a lack of self confidence.

In respect of the relationship between theory of mind and attentional cues, two possibilities exist. It may be that the use of attentional cues leads to an awareness of belief and false belief. Alternatively, the understanding of belief and false belief may rest on a developing awareness of intention and desire, and knowledge. Attentional cues are only used to infer a certain mental state by children who are aware of the existence of the mental state in the first place.... i.e. the use of attentional cues is not a precursor of a developing theory of mind, but an outcome.

Communication Breakdowns in (Verbal) Autistic Children

On a different aspect of Autism, Geller (1998) describes how the analysis of pragmatic skills have contributed to the understanding of the particular language capacities and deficits that are typically observed among children with Autism.

In particular, attention is directed towards the nature of linguistic and social-cognitive skills that may be inferred from an examination of the way in which communication breaks down and in which it may be re-established.

When conversation proceeds normally, the speaker will be making appropriate inferences about the listener... i.e. that the topic is understood, that the context is recognised, etc.. Meanwhile the listener will seek to respond in a meaningful and co-operative way.

If the conversation breaks down, the speaker may try to repair the situation by recognising that the listener has not understood and clarifying what was meant, by shifting the topic, or by ending the exchange.

It is normally assumed that the (child) speaker genuinely wishes to communicate, is monitoring the listener's comprehension and trying to ease understanding. In normal language acquisition, the 4 year old will be able to respond appropriately to questions, and can modify what (s)he says according to the different type of response observed.

Among children with Autism, communication breakdowns and (non)repair are significant areas of difficulty in that, when an exchange falters, such children appear to lack strategies to bring the conversation back on track.... and one notes the common situation where the autistic child will seek to initiate a conversation by reference to some topic or event, and will assume (given an inadequate theory of mind) that the listener will immediately be able to "tune in " without any confusion.

A question raised (for example, by Paul and Cohen 1984) is whether the child is unwilling or unable to work towards repairing the exchange. These authors observed adults with Autism in conversation with a confederate who made frequent requests for clarification. Compared with subjects with learning difficulties, the autistic sample were no less responsive to another person's query, but were uncertain how to continue. They appeared unable to work out which piece of information was not clear, and it was concluded that the deficit was linked to social-cognitive rather than to purely linguistic impairments.

Geller (opp.cit ) reviews existing studies which appeared to confirm that children with Autism are rarely able to modify messages to fit the needs of the listener, even when they were able to work out what the listener's understanding might be. Meanwhile, other research has led to some further question whether the problem reflects a lack of awareness of social "obligations" when holding a conversation, or simply an immaturity of language development such that there is an inability to make adaptations according to listener needs.

Geller's own study set out to investigate the behaviour of children with Autism during spontaneous conversations with an adult partner; and, in particular, looked for the nature of communicative style leading to breakdown, whether the children could make modifications to their language and non-language behaviours to bring about repairs, and whether they could recognise the communicative needs or intentions behind the listener's questions. A small sample of children with Autism, ranging in age from 7+ to 12+ years was involved and interchanges with an adult, involving free play activities in 30 minute sessions, were videotaped and all the communicative behaviours were transcribed for analysis.

The results indicated a clear trend towards unsuccessful resolutions of communication breakdowns with a general lack of appreciation of the needs of the listener. In respect of the original attempts to communicate, it was found that problems arose from impairments in the basic form of language. For example, despite no evidence for phonological problems, the children were commonly found to demonstrate poor intelligibility, or over rapid rate of speech and/or poor audibility. Listener confusion did not arise from problems in syntax or language structure.

In respect of content, the greatest number of breakdowns were caused by a lack of clarity about the subject matter.... a launching into a conversation when the listener has insufficient awareness about the object or event in question. It is held that this relates to a general developmental issue in respect of the establishing of joint attention. Further, confusion was increased by frequent pragmatic errors, notably a failure to indicate a shift of topic.

Meanwhile, the children did not ignore the adult's questions, appeared to have some sense of their social obligation to respond, and appeared willing to respond. However, their ability to repair messages by increasing the informativeness of utterances was limited. The children appeared to have a repertoire of revisions, including elaborations, or vocabulary substitutions, but no consistent pattern was observed either within or between subjects.

In sum, the language "performance" among the children appeared poorer than would have been predicted by observations of their stage of language development.. i.e. a lack of balance was noted between language structure and semantic-pragmatic use of the language. The implication drawn refers to the risk that children with Autism are too often simply the recipients of communications, and that breakdowns are attributed to an inability to engage in genuine discourse on their part of the children. Instead, the particular forms of weakness can be tackled by direct teaching, modelling, and practice. Children will benefit from all the greater communicative experiences and responsibilities because the emphasis upon adult control of communications will limit their learning of how to be more effective partners.

Developmental Patterns among Children with Autism

The general theme of the paper by Sigman (1998) is that each developmental phase has both its own particular characteristics and requirements which will guide interventions, and consequences for subsequent development.

Reference is made to 3 basic approaches to the study of developmental trends:

* The behaviour of children who share a particular diagnostic label is compared with that of ordinary children matched for mental age in order to explore ways in which the disorder has caused deviation from the normal progression.

* Stability in the observable characteristics is examined by comparing the diagnostic groups at different ages, or by examining an individual's performance in a given area over time.

* Precursors of characteristics or (dis)abilities are investigated by examining how the ranking of a child within the group on some element is correlated with the child's ranking on a different element measured at a later time.

In her review, Sigman chose to focus upon social competence because of its significance in shaping the children's day to day experiences; and her general theme tends to echo that of the implication in the preceding section... viz, that early interventions should be directed towards improving basic communication skills given their significance in influencing later verbal and social skills.

Social Competence Deficient language acquisition is seen as a predominant symptom of Autism; and even those individuals who achieve productive speech are still likely to show deficits in prosody and pragmatics. Observations indicate that nonverbal communication and representational play are similarly deficient. Videotaped interactions show how children with Autism initiate and respond to bids for joint attention less than Down's Syndrome children, developmentally delayed children, or normally developing children.

Requesting behaviour is also less among the autistic group whose functional or symbolic play also appears limited compared to that of the other groups.

It is held that the deficit in joint attention is of particular significance given its observedly high correlation with language skills. In contrast to other children, autistic children tend to look less at other people whatever their behaviour, and this in itself will have a negative impact upon pretend play since, in order to pretend, children need to observe the actions and reactions of others. Further, the autistic sample observed tended to be less concerned in the presence of somebody showing distress, and were generally less empathic.... possibly to be attributed to a limitation in the ability to differentiate among others' facial expressions.

An alternative hypothesis is that the autistic children do not respond to the emotions of other people because such emotions produce an over-arousal.

In sum, the autistic child is less interested in the responses of others, and this may be linked with deficits in joint attention or imaginative play. It may equally be that the children have an innate defect in the mechanism underlying attention and affect, or in the processing of relational information.

Stability of Development Many children may appear to outgrow difficulties because they acquire compensatory skills or because the environmental circumstances or demands change. However, the reverse may come to pass in that new developmental stages may be so challenging that the children seek to withdraw from what is an over stressful setting. The question is begged to what extent do children diagnosed with Autism remain affected during the whole of their lives.

A review of studies completed by Sigman (opp.cit) suggests that most individuals with Autism will be as severely affected in adolescence as in early childhood. Estimates indicate that adult outcomes are generally poor. What is less clear is whether particular symptoms, or individual diagnoses, are stable over time. The reason for this seems to concern the range of diagnostic criteria adopted by various clinicians with implications for difficulty in creating valid comparison groups.

In Sigman's own sample, a large percentage of the children identified at 3 to 5 years of age were still available for reassessment at adolescence, and the immediate observation was that all the symptoms underlying the original diagnosis were still evident. This was all the more striking in the light of the access of the children to (early) intervention programmes for at least some of the time.

Other studies cited suggest that the nature of some symptoms may change. For example, some older children may show decreasing stereotypic behaviour, and this may apply particularly to the higher functioning cases. Also, if a child develops language, a lack of verbal skill may be replaced by language that is stilted, prosodically odd, or showing pronoun confusion ( commonly "you" or "he" instead of "I").

Another question concerns the nature or stability of scores on tests of intellectual functioning ... with the necessary caveat about the possible difficulties of gaining meaningful intelligence scores in the first place given the problems of eliciting the children's attention, and of interpreting responses. Sigman's own study suggested a retest reliability with the Cattell Developmental Scale, or the Stanford Binet Scale, to be in the order of 0.6, only slightly lower than for the samples of Down's Syndrome children and normally developing children. It is noted that samples of autistic children often show higher nonverbal scores because of particular problems with language development, such that some researchers will use measures of performance IQ rather than verbal or combined IQs. Nevertheless, a similar level of stability of scores has been observed.

On the other hand, while there is stability in group means in IQ results, individual scores may show wide variation over time, with increases or decreases evenly distributed in the sample. In particular, it was noted that "a surprising number of children who tested in the mentally retarded range at intake had scores above that range at follow up some 8 or 9 years later".

(The present writer -MJC- would simply wonder whether this finding ought to be borne in mind when one is presented with claims about IQ gains as a result of certain forms of intervention. It is not indicated whether the gains quoted by Sigman were linked to any one form of intervention, but one might need to consider the possibility that test-retest differences are a function of the age and developmental level of the child when initially assessed, and the nature of the tests used, as much as of the precise form of treatment).

Stability at both individual and group level was observed in respect of communication skills, including nonverbal behaviours. Meanwhile, all the children showed some increase in mean language age (as measured, for example, by the Reynell test) although the amount of gain was greatest among the developmentally delayed group, and the gains among autistic, DD, and Down's groups were very much less than among typically developing children. In the autistic sample, it appeared that those children who gained at least a 2 year level of understanding of language did not differ in initial intelligence scores from those children who did not achieve this level. Thus, the early assessment of intelligence did not predict language skill (but improvement in measured intelligence over time was linked to increases in language skills).

When it came to response to (apparent) distress on the part of the adult, it was found that the children looked at the adult's face for longer when they were older compared to when they were younger, although the autistics typically maintained this looking for less time that the other groups of children, and were rated as less concerned than the other groups ( as illustrated by, for example, any change in their facial expressions).

Predictive Indices of Nonverbal Communication and Play It appears logical to equate positive skills in communication and play, which involve joint attention, with likely gains in language. The functional and symbolic use of objects in play is an indicator of the existence of the kind of conceptual understanding that is necessary for the use and understanding of language.

Among the autistic children, the language skills observed were correlated with the nonverbal and play skills. Long term gains in language usage 8 or 9 years later were predicted by the children's responsiveness to another person's bid for joint attention as well as the diversity of use of play objects. It is held that these basic skills are not only prerequisites for the acquisition of language, but are needed for its consolidation.

Meanwhile, it has been suggested that prosocial behaviours among autistic samples are limited because the children cannot readily understand the needs of others. It was further anticipated that weaknesses in nonverbal communication, play, and recognising stress in others, would inhibit relationships. However, what empirical evidence there is, in Sigman's review, would suggest otherwise in that young children with Autism are seen to use their caregivers as a secure base, becoming distressed when the caregivers leave them in an unfamiliar setting and showing increased verbal or physical contact on their return.

In respect of social interaction, the long term follow-up suggested that the pattern of limited interaction and a preference for solitary play observed initially was maintained, even among high functioning individuals. This pattern appeared to be the preference of the children with Autism who initiated fewer social bids, and rejected more of the bids directed to them, than the developmentally delayed or Down's children. Nevertheless, school environment was a factor in that children with Autism who had some contact during the school day with normally developing children were more socially involved than comparable others who had no such contact.

The implications set down by Sigman include the need for maximally early identification not only of the autistic symptoms but also of those environmental factors, in the home and in the school, which underpin the characteristics in the children which are associated with positive developmental gains.... for example, how best to encourage joint attention and pretend play; or how to focus more upon the development of social competence in the mid-school years, such as ways of helping children with Autism to acquire skills for entering play situations in the playground. The observations by Sigman suggest that school staff may tend towards an emphasis upon scholastic development and an underplaying of the significance of social capacities and social involvement for the children with Autism.

Further, it is recommended that intensive language interventions are continued beyond the preschool period to well into the mid school years (to avoid the risk that any plateau in language growth is simply a function of a reduction of language input).

Autism and the Use of "Circles of Friends"

In respect of classroom practice, Whitaker et al (1998) describe the outcome of the use of the Circle of Friends' technique with autistic children in mainstream schools as the focus. This technique seeks to promote inclusion of pupils with disabilities, and provides one means of including (socially) vulnerable children within a peer group which is seen as having considerable potential influence upon the behaviour of individuals.

In Autism, the core issue is that of impaired social interactions compounded by problems in communication and imagination. Lord and Magill-Evans (1995) held that individuals with Autism had particular difficulty in seeking to coordinate eye contact with smiling or other communicative behaviours, and that other children tended to make little response to the attempts to initiate interactions that were made... or did not even recognise them as attempts to communicate!

There is a risk, therefore, that the social problems become self perpetuating ; and reference is made to Roeyers (1995) who found that benefits were gained from putting socially competent children into contact with children with Autism and asking them to interact .... without offering any specific advice as to how to do this. It was argued that informal groupings may have longer lasting effects than more formal situations where peers are explicitly trained and used as social skill instructors.

In any event, without some kind of intervention, children with Autism are likely to experience major difficulties in gaining any meaningful peer interaction. Additional support may be at a high level in the classroom, but significant problems are likely to arise in social interchange, which may be all the more marked outside the classroom setting. Accordingly, Whitaker et al (opp.cit) set up a series of 6 Circles of Friends' groups to support children with Autism in years 3 to 10 of mainstream schools.

The observations of the circle leaders referred to enhanced quality and quantity of contact between the target child and the wider group (beyond the circle) which was attributed to changes in the target child in terms of the desire to make contact and reduced anxiety about making contact. Behaviours showed improvement, e.g. in reducing obsessionality or compulsive behaviour.

There was some problem in the case of 2 of the children with Autism in respect of some increased egocentricity... a development of an enhanced sense of self importance and a failure to show any sensitivity to the needs or feelings of others ... with implications for a need for the leaders to intervene and work towards reducing negative feelings. Meanwhile, in other groups, the circle members appeared to have worked towards this goal without outside prompting. Further, the benefits of the circles were demonstrated through the increased levels of empathy commonly seen in circle members, and a reduction in any tendency to interpret personally certain of the target child's behaviours.

Self esteem was observably increased among circle members, who appeared to benefit from the opportunity to exchange views and to have their opinions taken seriously by the adults. One or two circle members with some emotional or behavioural difficulties benefited either directly or in terms of enhanced teacher opinion following the observed contribution to the circle; and some quieter children began to contribute more confidently. All the circle leaders felt that the experience was valuable, and were impressed by the sensitivity and commitment shown by the circle members.

In sum, the authors note that the circles were organised on the understanding of some "unequal competence" such that the participating children worked towards helping a less competent peer. The pupil with Autism was more readily credited with positive attributes, less blamed for problems, and had his particular difficulties more recognised and understood. A year on, the circles are said to be still operating among the original target and peer samples of pupils, and other circles have been established in other schools where positive reports are cited.

Therefore, they conclude that, while one cannot be certain how much of the changes in behaviour are directly the result of the Circles of Friends, it is legitimate to regard this approach as a means of mobilising practical support for the children with Autism, while gaining benefits for peers, all with minimal costs, and minimal drawbacks for any participants.

Hormone Action and Autistic Recovery

As a final section in these sets of notes, one would draw attention to the reports, published in the national press on the 14th. and 15th September, of what might prove a significant means of treating Autism.

A child in New Hampshire, described as Autistic, was treated with a hormone - Secretin - as part of investigation and treatment of a gastrointestinal condition. His mother reported that, within days, the then 3+ year old showed rapid signs of improvement in respect of verbal behaviour, responsiveness, concentration span, eye contact, calm behaviour, and consistent sleeping pattern. Facial tics were eliminated. 3 weeks after the use of Secretin, short phrases were being articulated ; and potty training, previously impossible, was achieved within a relatively short time.

There are plans to carry out formal studies in America which will be initiated before the end of 1998. Meanwhile, other informal and anecdotal evidence from both parents and doctors, gathered by Paul Shattock of the Autism Research Unit (Sunderland University), is said to be positive with regard to the use among children with Autism of Secretin, whose normal function as a hormone is to stimulate the pancreas to produce enzymes. However, the lack of any controlled studies, and the non-quantitative nature of the results claimed, are acknowledged.

One will await, with interest, further events in these biochemical and metabolic fields.

M.J.Connor September 1998

REFERENCES

Baron-Cohen S. 1995 Mindblindness. Cambridge MA: MIT Press

Duncan J. 1986 Disorganisation of Behaviour after Frontal Lobe Damage. Cognitive Neuropsychology 3 271-290

Frith U. 1992 The Cognitive Neuropsychology of Schizophrenia. Hove: Eribaum

Frith U., Morton J., and Leslie A. 1991 The cognitive basis of a biological disorder: Autism. Trends in Neuroscience 14 433-438

Geller E. 1998 An investigation of communication breakdowns and repairs in verbal autistic children. The British Journal of Developmental Disabilities 44(2) 71-85

Hughes C. 1998 Executive function in pre-schoolers: links with theory of mind and verbal ability. British Journal of Developmental Psychology 16 233-253

Lee K., Eskritt M., Symons L., and Muir D. 1998 Children's use of triadic eye gaze information for "mindreading". Developmental Psychology 34(3) 525-539

Lord C. and Magill-Evans J. 1995 Peer interaction of autistic children and adolescents. Development and Psychopathology 17 611-626

McEvoy R., Rogers S., and Pennington B. 1993 Executive function and social communication deficits in young autistic children. Journal of Child Psychology and Psychiatry 34 563-578

Paul R. and Cohen D. 1984 Responses to contingent queries in adults with mental retardation and pervasive developmental disorders. Applied Psycholinguistics 5 349-357

Roeyers H. 1995 A peer mediated proximity intervention to facilitate the social interaction of children with a pervasive developmental disorder. British Journal of Special Education 22(4) 161-164

Sigman M. 1998 Change and continuity in the development of children with Autism. Journal of Child Psychology and Psychiatry 39(6) 817-827

Welsh M., Pennington B., and Groisser D. 1991 A normative-developmental study of executive function. Developmental Neuropsychology 7 131 - 149

Whitaker P., Barratt P., Joy H., Potter M., and Thomas G. 1998 Children with Autism and peer group support, using "circles of friends". British Journal of Special Education 25(2) 60-64