The Non-Unitary Nature of Autism and ASD


This set of reviews provides evidence that autism is a condition (or a range of conditions?) for which there is no single and simple explanation and aetiology.

The first summary concerns the inability to encompass the core characteristics within a single explanatory framework. Further, it appears likely that a number of genes and genetic loci will prove to be implicated in the phenotype of autism. 

The subsequent sections offer examples of work concerned with individual elements of autistic symptomatology or “style”; and illustrations of the complexity of the way in which the autism is expressed or moderated.

The implication is for the lack of utility in maintaining a quest for some single cause and, therefore, for avoiding the placing of trust in one all-embracing intervention.


M.J.Connor                                                                                            February 2008


No Single Explanation for Autism


The notes by Happé et al (2006) start by quoting the “triad” of impairments characteristic of autism, namely deficient social interaction, limited communication, and restricted and repetitive interests and activities.

The authors describe these diagnostic elements as reflecting Kanner’s  emphasis in his conceptualisation of autism as a matter of “aloneness” and “insistence upon sameness”, but the individuals who share this diagnostic label may present as both highly recognisable and widely heterogeneous.


They also argue that despite the long term research into autism and ASD, there is a lack of evidence by which logically to link the three core areas.  There seems always to have been an assumption among researchers, working at genetic, behavioural, or cognitive levels, that these impairments must be explained together; but these current authors cast doubt on this assumption and suggest, instead, that the impairments are separable and should be studied as distinct entities.


In respect of behavioural features, reference is made to the way in which circular arguments may arise in that samples of individuals with autism for assessment and analysis will be recruited on the basis of the formal diagnostic criteria which require sets of symptoms covering all three areas of impairment. 


Meanwhile, it has become established that autistic-like traits can be observed and measured in the general population as in the case of a child who may have problem on one or more of the areas of sharing play with other children, maintaining a two-way conversation, and shifting from one preferred type of behaviour or activity. 


The concept of an autistic spectrum or continuum has arisen from the observed similarities between individuals who meet the full criteria for a formal diagnosis and members of the general population who show some anomalous behaviours but above the criterion level for formal identification.


What is seen as significant is the absence of a “bulge” in the overall distribution of such behaviours which would clearly separate clinically-significant from non-clinical levels of difficulty.

The question is raised whether impairments in social interaction and communication, alongside restricted interests and activities, significantly co-occur in the population at large, with implications for some common cause.


The authors’ own previous studies have involved an analysis of large numbers of twin pairs, assessed between the ages of 7 and 9 years, and the derivation of only modest or low correlations among these three areas of impairment. 


This applied even to correlations between social and communication difficulties which seem commonly to co-exist according to day to day observations with the suggested implication of some shared cognitive deficit.  However, the correlations here were still limited, suggesting that the levels of social impairment, communication difficulty and rigidity of behaviour, are not strongly related. 


Further, a considerable number of children in their population-based samples showed difficulty in only one element of the autistic triad.  There appeared to be some enhanced risk for difficulties in the other two elements when one impairment was observed, but the risk was relatively low, supporting the growing view that the three types of impairment are separable. 


The same conclusion is indicated from the observation that there are different developmental trajectories for the different elements.  For example, rigid behaviours and restricted interests are usually evident later than social or communicative difficulties, improve less over the long term, and respond less well to intervention.


The next question explored by these authors is whether there are separate genes contributing to the social, communicative and behavioural impairments ... and their evidence would suggest that this is the case. 


Based upon studies of 7-8 year old twins, the findings suggest that more than half the genes that contribute, for example, to variation in social competences are independent from those that contribute to the communication and behaviour elements of the triad.   In other words, most of the genetic effects (at least in middle childhood) are specific to one trait characteristic of autism. 


Reference is made to other work, such as that of Pickles et al (2000), which has shown that relatives of individuals formally diagnosed with autism demonstrate increased rates of autistic or autistic-like behaviours and traits.  Some of the relatives may show only isolated symptoms, such as poor communication without problems concerned with social weaknesses or with rigidity of behaviour. 

The implication again is for differentiable effects of genes across the various behavioural characteristics of autism.


In their general summary, Happé et al argue that attempts to explain all of the core diagnostic features of autism (the triad) through one hypothesised cognitive pathway are unlikely to succeed.

Rather, the behavioural and genetic evidence points to a fractionalisation of the triad, and a more realistic goal for research is to account separately for each element.


Existing theories can be grouped into those which regard social cognition (theory of mind) and emotional processing as primary, and those which focus upon deficits in non-social forms of processing (executive dysfunction) and anomalous attentional and perceptual processing.


However, no primary deficit has been demonstrated as being able to account for the whole triad.  Social deficits could plausibly explain the development of communication impairment, but can hardly explain rigid and repetitive behaviours. Correspondingly, the non-social accounts of autism would not readily explain the marked impairments in communication and social understanding which may co-exist with adequate or better intellectual ability and with good reasoning capacities in non-social domains.


Neuro-imaging studies can also be cited as supporting the independence of the cognitive substrates for social, communicative, and behavioural impairments in that images derived from individuals with autism and from controls suggest that each element calls upon brain activation in different brain regions/networks so that, for example, social cognition involves one specific network including the medial frontal cortex and superior temporal sulcus, while rigid and repetitive behaviour has been linked to caudate anomalies. 


The recommendation, therefore, is to cease the search for an all-encompassing explanation for all the aspects of autism.  Meanwhile, the question still remains why there should be this enhanced probability among individuals with autism of the co-existence of the three core features, and reference is made to at least partial overlap between domains of some genes and of interactions between cognitive domains ... but the overall evidence still highlights the fractionalisation of these core features.


They repeat the implication for measuring and assessing separately the different behavioural manifestations of autism rather than relying upon ratings of the global severity of the condition or ratings.

Similarly, genetic studies would usefully explore how genes contribute specifically to particular traits and behaviours.

Heterogeneity among individuals all validly diagnosed with autism may be explicable in terms of the variation along at least three largely independent (albeit interacting) dimensions of impairment.


One practical concern is that there may be a significant number of people who show significant impairments in one aspect of the triad so that they do not receive a formal diagnosis of autism/ASD (or of any other condition) but whose needs are still quite severe and for whom intervention would be required but not necessarily available. 


Continuing this theme of intervention, the authors’ final point is that, if different features of autism are linked to different genes and different brain regions and to separable cognitive impairments, it is likely that they will not respond equally to the same form of intervention, and a range of approaches will be appropriate if the range of needs are to be met.  If one needs to give up the quest for some single cause of autism, and to cease to regard autism as a single entity, it is no longer logical, either, to seek some single “cure” or means of intervention.


Genetic and Other Risk Factors for Autism   


As a brief reinforcement of the above-summarised arguments, one can cite studies which highlight the “jigsaw” nature of the genetics of autism, or which suggest the complexity of familial or “physical” factors determining the extent of individual risk for autism.


For example, a study completed by a genetic research team, and summarised by Busko (2008), has indicated the likely significance of micro-deletions and corresponding micro-duplications in chromosome 16 and in chromosome 7.


The authors acknowledge that, while an important finding, this evidence is but a small step; and that the genetics of autism remain a puzzle with many pieces to be identified if there is to be coherent insight into the aetiological mechanisms associated with autism.

They go on to note that autism is a matter of significant heritability. However, only a small percentage of cases are explicable in terms of clearly-defined genetic syndromes and chromosomal abnormalities; and the majority of cases are idiopathic and not yet explained by ongoing (genetic) research. 


A second study, also summarised by Busko (2008), confirmed the likely significance of chromosome 7 ... notably a variant in the structure or function of gene CNTNAP2 which appears to increase the risk for autism.

Individual variance and complexity are highlighted again by the observation that not everyone who carries this variant goes on to develop the traits and symptoms of autism.  Why this is so remains unclear, but one commonly expressed view is that the  variant is a necessary but not sufficient causal factor in autism, and that the development of the condition would require the operation of further factors such as the influence of other genes, and specific developmental or environmental factors.


Individual variance in the aetiology of autism is also an implication from the work of Williams et al (2008) who examined population-based data in New South Wales for risk factors for autism associated with maternal characteristics or with the circumstances of birth.


These authors identified four risk factors significantly associated with autism :


The level of risk becomes greater where two or more of these factors co-occur. 


(Again, there appears to be the implied question why the risk is realised in some cases and not in others .... perhaps further supporting the concept of idiosyncratic interactions of some [genetic?] susceptibility to autism and the impact of some triggering event or circumstance.)


Determinants of “Engagement” in Autism 


The focus of the study completed by Ruble and Robson (2007) provides an example of one specific and common aspect of autism ... the atypical engagement between a child and significant others,  notably between the child and the teacher(s), which may underlie their behavioural relationships and the outcomes in terms of academic and developmental progress. 


Their initial research review demonstrates that the emphasis has been upon the effects of particular instructional styles and contexts upon engagement, with increased task or academic engagement linked to specialised techniques such as visual timetables, peer tutoring, pivotal response training, etc.


In respect also of (behavioural) programmes, it has been recommended by the National Research Council (2001) that a minimum of 25 hours per week of engaged time is necessary for children with autism.  What matters is the amount of time during which the child is actually engaged, not simply the hours set aside for the intervention.


The council defines engagement as “ sustained attention to an activity or a person ”; and other definitions vary according to the age or developmental level of the children.

For older children, engagement is discussed in terms of behaviours associated with classroom responses (attention to the task, answering questions, asking questions, task completion, etc).


Meanwhile, it is noted that qualitative features of engagement have significance in that useful engagement is not just a matter of what the child is doing but how the child is doing it.  A simple measure of time spent on an activity may fail to capture important components of engagement (and learning).

Accordingly, a definition such as that provided by McWilliam and Bailey (1992) is cited as appropriately reflecting the trait-construct nature of engagement ... “ the amount of time that children spend interacting appropriately with the environment at different levels of competence. ”


The current authors summarise by describing engagement as a stable construct that relates to internal factors (temperament or the nature/level of a diagnosed condition), to child behaviours (such as play skills), and environmental factors (classroom activity).


In the case of children with autism, questions concerning the what and the how of behaviour need to be supplemented by questions concerning the why of behaviour.  Seeking to interpret behaviour from the child’s perspective has highlighted key behaviours or “style” such as limitations in non-verbal (social) communication such as joint attention, or the use of stereotyped utterances and echolalia whose function may be that of putting some order into events or enabling the child to make requests or to rehearse interaction or to seek help. 


Engagement is seen by Ruble and Robson as a critical element behind the effectiveness of intervention so that they regard it as curious that it has received little attention in the research literature.  In particular, they regret the lack of information concerning the interaction between levels of engagement and in-child factors relating to diagnostic criteria such as communicative and social impairments.


Therefore, their own study set out further to expand the definition of engagement by focusing upon two types of engagement that have been used as part of outcome measures in research.


The first is compliant engagement which refers to the quality of the child’s responses to the actions and stimuli of the teacher or facilitator (as in applied behavioural analysis).  Compliance refers to the child’s shift in behaviour towards a target activity and goal.


The second is congruent engagement which concerns the extent to which the child’s goal-directed behaviour matches the behaviour of other children in the class.

(Congruence is achieved when the child is working effectively on the same task as the other children, such as completing a number work-sheet; and not achieved when, for example, the child appears to be on task but is actually drawing a repeated series of doodles on the sheet.)


The authors note that it has remained unclear why some children with autism respond well to a specialised intervention and others do not; and they suggest that the process of seeking most effectively to match the style of intervention to characteristics of the child would usefully include an evaluation of qualitative aspects of engagement such as compliance and congruence.


Their participants were samples of children with autism and children with Down syndrome.  In each sample there were 4 boys whose parents agreed to have them observed at school. 


The ages of the children with autism ranged from 8·6 to 10·7.  The two youngest were in kindergarten classes, one was in special education, and the oldest was in a mainstream class.


The ages of the Down syndrome children ranged from 6·8 to 9·9.  Two were in special education and two in mainstream classes. 


Adaptive behaviour scores were comparable between and within groups.

The children were observed by assistants trained in the recording and coding of behaviours during blocks of time identified by the teachers to ensure both structured and unstructured activities. 


The findings indicate a similar number of goal directed behaviours from the children in both groups, with around 50% of the behaviours congruent, and around 75% compliant.  The disparity between congruent and complaint behaviours had been unexpected, and the implication was for the importance of improving congruent engagement through the programmes designed for children with disabilities.


It was also noted that both internal child factors and external environmental factors influenced types of engagement ... supporting the view of engagement as a state construct influenced by external events but also mediated through trait or internal factors.


The children with autism had over 80% of behaviours coded as compliant but only a little more than 40% coded as congruent. 39% were coded as both.

The children with Down syndrome showed behaviours of which 73% were coded as compliant, 57% congruent, and 38% both.


In other words, while there was some similarity between the two groups in that compliant behaviour was a relative strength across the whole sample (especially among the children with autism), the quality of engagement among the Down syndrome children was significantly better for congruence compared to the children with autism.


The nature of the classroom task and context was shown to be influential in that large-group instruction was associated with most goal-directed behaviour.  Children with autism showed relatively high levels of compliant engagement in this setting, but congruent engagement was a significant challenge.


In contrast, the children with autism produced the fewest compliant units of activity during one to one instruction with an adult.  Also, during independent work, the children showed a high level of compliant engagement, as reflected in sustained attention, but a low level of congruence. 

The question is raised about how to adapt and modify instruction to produce both compliance and congruence. 


It is noted that Education departments are publishing guidance concerning programmes for children with autism and that there is a frequent reference to the significance of engagement for educational and social gains.  However, the reference to the definition of engagement as a matter of sustained attention to an activity or person is seen as restrictive, and the implication is for a more meaningful and objective means of assessing engagement in order to improve outcomes.


In particular, the authors refer to the possibly misleading or superficial use of inclusion or mainstreaming of a child as a marker of effective response to some early intervention ... as in the case of the target sample in the Lovaas (1987) study.

The current data suggest that mainstreaming per se is not a sufficient outcome because a child may have learnt to sit and to display apparently adaptive behaviours, but may still be engaged in activities that are unproductive and unrelated to the overall class goals. 


The authors refer to a more recent study (Howard et al 2005) comparing the efficacy of an intensive ABA programme and an eclectic programme for young children with autism.  Their findings suggested that 25-40 hours of intensive discrete trial training was more effective than 30 hours per week of intensive eclectic provision plus 15 hours of non-intensive educational provision in producing gains across a range of domains. 


Their conclusion was that intensive ABA is more effective than eclectic approaches.

However, Ruble and Robson pose the question of what exactly is meant by “intensive”.  Without clear evidence concerning the amount of time during which the children were actively engaged it may not be possible to make a true comparison of interventions. 


The need is for the number of hours of engaged time to be calculated as a separate variable such that one can more meaningfully explore the relationships among instructional settings/methodologies, engaged behaviours, and developmental outcomes in terms of independence, social and communication skills, and academic achievement. 


Gestalt Perception and Local vs Global Processing             


A further example of apparently autism-specific “style” concerns the way in which (visual) stimuli are processed and interpreted. 


In their discussion of the characteristic impairments among individuals with autism in social and communicative reciprocity, Bölte et al (2007) refer to a lack of theory of mind as a very commonly cited basis; while executive dysfunction and weak central coherence (WCC) are cited as underlying the non-social elements of autism. 


The WCC hypothesis holds that individuals with autism adopt a processing style which involves a piecemeal and local examination of the stimulus rather than an examination of the whole stimulus and its context.  WCC may even explain strengths on certain tasks such as Block Design or identifying figures embedded within a larger pattern since the individual can focus on the component parts and avoid being distracted by the overall picture. 


A slightly modified hypothesis is that of a “ hierarchisation deficit ” which suggests that individuals with autism do not lack the actual ability to process stimuli globally as well as locally, but that, unlike typically developing individual, they do not assign particular importance to certain elements of the stimulus over others (the eyes in a face, for example) but all parts are seen as of similar importance. 


It is held by these current authors that the stage of perception where either the WCC or hierarchisation deficit has its impact is not clear.  However, work on misperceptions based upon visual illusions has suggested that individuals with autism do tend to make accurate judgements when presented with common two-dimensional illusions ... ie they are not misled by the overall stimulus ... with the implication that local processing rather than global processing operates at a very early stage in perceptual processing.


Meanwhile, there is some issue whether the terms “global” and “gestalt” mean much the same thing and can be used interchangeably, or whether gestalt processing should be subsumed under global perception. 

Gestalt perception refers to a tendency to group together elements of a visual array, which is not really the same as global perception, and the suggestion is that gestalt processing may precede and aid global processing.  There is, in other words, a tendency among typically developing individuals to perceive stimuli as making up a meaningful whole rather than as a collection of separate parts. 


One investigation of this issue in the case of children with autism was completed by Brosnan et al (2004). 

They found that, compared to a matched group of children with learning difficulties, their sample of children with low-functioning autism were less likely to follow the gestalt principles of grouping elements of the stimulus according to proximity, similarity, and closure.

(The principle of proximity states that items close together are perceived as belonging together; the principle of similarity states that items which look similar will be put together; and the principle  of closure states that something missing from an otherwise complete figure will be added automatically during the perceptual processing by the observer.)


Their conclusion was that in addition to the anomalous pattern of local-global processing, there is, in autism, a very basic impairment in these gestalt processes, offering scope for further insight into perceptual mechanisms characteristic of autism.


Accordingly, Bölte et al set out to examine these three gestalt principles among a group of individuals with high functioning autism.

A series of visual “puzzles” were used, including the identification of individual but overlapping shapes; visual illusions; identifying a shape embedded within a larger and more complex figure; block design; and hierarchical letters (where letters are created from many small letters  .....

e.g.  L                 or   S

        L                       S

        L                       S

        L                       S

        L                       S

        LLLLL             SSSSSS .... to determine whether the initial focus is upon local processing so that the small letters are identified, and whether the large letter is identified spontaneously (global processing).


The findings were that the (adult) sample with high functionig autism, compared to all control groups, made significantly less use of the gestalt principles, notably similarity, but also proximity and closure.

This outcome is in line with the findings of Brosnan concerned with the performance of individuals with lower functioning autism. 


The significance of the atypical performance in respect of similarity was said to be striking given that the autistic participants could be differentiated even from those control participants (those identified with schizophrenia and those with depression) who have been shown to have unusual or impaired gestalt perception.


Meanwhile, the authors held that the results on the embedded figure test show that this measure assesses visuo-spatial abilities linked to local-global processing which can reliably differentiate a sample of individuals with autism from other samples.

It was also noted, in line with some existing research, that the participants with autism were less likely to be misled by the visual illusions .. supporting the view that perceptual anomalies rise at a low level in the perceptual processing stages.

(There is some other existing research which did not find this relative immunity to illusions in autism; but Bölte et al suggest that the discrepancy may be explained by  to differences in the chosen procedure ... the type of illusion, the instructions, and the expected manner of response ... or in the particular nature of the sample of participants.)


The authors accept that their study has some weaknesses which could affect the generalisability of the findings.  These include the focus upon adult male participants, the use of a particular selection of visual challenges/illusions which may have produced outcomes that would not be associated with a different selection, and the overall small sample size.

Nevertheless, the results were seen as a useful step forward in ongoing research, and as suggesting decreased gestalt perception in high functioning autism associated with a more general bias towards local visual processing. 


Some Additional Autistic “Anomalies”


Finally, two recent findings may serve further to highlight the complexity and idiosyncratic nature of autism and autistic spectrum disorders.


Firstly, Curran et al (2007) have reported that the behaviours of children with ASD may improve if they develop a fever. 

This conclusion was based upon parental reports of variations in behaviour among their children (aged between 2 and 18 years) during and after episodes of fever where temperature reached or exceeded 100.4F (38C).  Similar findings are cited from existing case studies and anecdotal reports. 


The behavioural change was in terms of a reduction in the amount of maladaptive or aberrant behaviours such as irritability, hyperactivity, stereotyped activity, and inappropriate speech.  Such changes were observable in the great majority of the sample of children, but the improvements were not lasting. 


The authors held that the behavioural change was not linked to the severity of the illness given that the observed changes occurred across children showing a range of levels of fever. However, it is not clear to what the change should be attributed, and there was speculation concerning the role of immunological influences, neurobiological pathways, and synaptic modifications. 

Further research is needed both to confirm this pattern, and to gather further data by which to interpret the mechanisms (which may shed further light upon the issue of autistic aetiology generally).


The second concerns the contagious nature of yawning among typically-developing individuals but apparently not among those with autism. 

The study by Senju et al (2008) found that children with ASD, whose characteristic features include a lack of empathy, appear to be immune to the effects of yawning.

Their target sample comprised 24 children, identified with ASD, between the ages of 7 and 15 years.  An age-matched control group also participated.

Each group watched a series of very brief video recordings of people yawning and of people opening their mouths but not yawning. 

Observations of the children while they viewed the recordings showed that the non-autistic children yawned more frequently during and after watching a person yawning than  after watching non-yawning.

The children with autism were observed to show no difference in their responses (in terms of yawning) to the different recordings. They appeared not to be affected by the contagious effect of yawning; and this finding held even after controlling for both age and intellectual ability.


The authors speculated that this finding may reflect the atypical face processing style and focus of gaze of individuals with autism; and also that the contagion observable among typically-developing individuals may share cognitive and neural mechanisms with empathy.  They suggest that a further topic for research is the extent of immunity from contagious yawning among groups of people identified with other disorders which also involve impaired empathy.


                                          *          *          *          *          *


M.J.Connor                                                                                            February 2008 






Bölte S., Holtmann M., Poustka F., Scheurich A., and Schmidt L.  2007   Gestalt perception and local-global processing in high functioning autism.   Journal of Autism and Developmental Disorders  37  1493-1504


Brosnan M., Scott F., Fox S., and Pye J.  2004    Gestalt processing in autism.  Journal of Child Psychology and Psychiatry  45  459-469 


Busko M.  2008   Genetic hot spots for autism probed in 2 studies.   Medscape Medical News  January 10th 2008.

(Original articles summarised : Weiss et al  2008  New England Journal of Medicine. Published on-line 9-1-08; and Arking et al  2008   American Journal of Human Genetics 82[1])


Curran L., Newschaffer C., Li-Ching L. et al  2007   Behaviours associated with fever in children with autistic spectrum disorder.   Pediatrics  120  e1386-e1392 


Happé F., Ronald A., and Plomin R.  2006   Time to give up on a single explanation for autism.   NatureNeuroscience  9(10)  1218-1220


Howard J., Sparkman C., Cohen H., Green G., and Stanislaw H.  2005   A comparison of intensive behaviour analytic and eclectic treatments for young children with autism.  Research in Developmental Disabilities  26  359-383


Lovaas O.  1987   Behavioural treatment and normal educational and intellectual functioning in young autistic children.   Journal of Consulting and clinical Psychology  55  3-9


McWilliam R. and Bailey D.  1992   Promoting engagement and mastery.  In D. Bailey and M. Wolery (Eds) Teaching Infants and Preschoolers with Disabilities.  New York : MacMillan  


National Research Council  2001   Educating Children with Autism.   Washington : National Academy Press  


Pickles A., Starr E., Kazak S. et al  2000   Variable expression of the autism broader phenotype : findings from extended pedigrees.   Journal of Child Psychology and Psychiatry  41  491-502


Ruble L. and Robson D.  2007   Individual and environmental determinants of engagement in autism.   Journal of Autism and Developmental Disorders  37  1457-1468


Senju A., Maeda M., Kikuchi Y., et al  (In Press 2008)  Absence of contagious yawning in children with autistic spectrum disorder.   Biological Letters.


Williams K., Helmer M., Duncan G., Peat J., and Mellis C.  2008   Perinatal and maternal risk factors for autism spectrum disorders in New South Wales, Australia.

Child : Care, Health, and Development  34(2)  249-256

This article is reproduced by kind permission of the author.

© Mike Connor 2008.

Back to NAS Surrey Branch Welcome Page