Attention Tunnelling and Autism

Dinah KC Murray, 1992. From Living with Autism: The Individual, the Family, and the Professional. Durham Conference Proceedings

  • Editor’s note: this was Dinah’s first presentation on autism, and the first ever publication on monotropism – a term coined on Dinah’s behalf by her friend Jeanette Buirski. It is presented here more-or-less as it appeared in the conference proceedings, aside from reformatting to take advantage of the technology of the web.

When Archimedes ran down the street, naked and shouting – oblivious to the fact that no-one could make any sense of what he was shouting, oblivious to the shock and discomfiture of those who saw him – he was displaying classically autistic behaviour. He was, I believe, in an “attention tunnel”1 so deep that he had nothing left over for modelling other people and their reactlons. A surprising number of the symptoms of autism make sense seen in this light: my basic suggestion is that it’s helpful to see autism as attention tunnelling writ large. We assume that Archimedes came out of his attention tunnel and reverted to normal behaviour: my speculation is that people with autism naturally go from one attention tunnel to the next, and their whole development is radically affected by this2.

My own background in psycholinguistics (especially conversation analysis – see refs) has led me to conclude that shifting and multiple interests play a fundamental role in normal cognitive life. Each of us has an interest system in which each interest is in competition with every other for a share of the scarce resource attention or interest.

A self-structuring, distributed, complex system adaptively results from competition for a scarce resource: it is polytropic, having simultaneous high arousal at a number of different points. (The mathematical model, by Allen, Lesser, and Corliss, was developed for ecological modelling, and consists in a single instruction dynamical equation constantly reapplied to its output. It includes such parameters as birthrate and deathrate,overall quantity of the desired resource, and positive and negative feedback. Unfortunately, the reproduction quality of this volume means I can’t include the highly graphic representation it can be given on a large enough computer. I find it fascinatingly isomorphic with my informal, non-mathematised conclusions about the nature of interest systems (see my rough illustrations). It is a model whose meaning is interpretable within a great many of the “life” disciplines. What the relationship is between these interpretations is a very interesting question3.

In autism, I believe, that distribution somehow goes wrong and arousal is monotropic or focussed in only one interest at a time. (In a small version of the model we find this may happen through overall scarcity of the resource, through restrictions on accessing It, or through excessive positive feedback for high arousal.) Hypersensitive within their attention tunnels and subject to overload, the establishment of connections is vastly slowed down for children with autism, and their capacity to deal with change is severely impaired.

Shattock and Lowdon (in press) argue that the most likely culprit for monotropic development is a malfunction in the metabolization of the opioid peptides. This is found in a high proportion of people with autism.Very speculatively, if there are genuine biochemical analogues of the parameters of our equation, then one might guess that the oploid peptides would be good candidates for the positive feedback parameter. People with autism who do not show any opioid peptide malfunction may perhaps turn out to have different neurochemical disorders which are analogous with, eg.,resource-access monotropism.

A brief account of a normal interest system

In the normal, polytropic, case, we all have a network of interests and desires which are active to varying degrees at different times. They are where all information is stored, they embody the history of their arousal,they are what they are now because of what has been co-attended in the past (arousal = allocation of a scarce attentional resource). Crudely, the more highly aroused an interest is, the more likely it is to be affected by input; but interests which have been highly aroused in the past tend to remain sensitive to relevant input and easily triggered, and input is of variable salience.

Polytropism naturally creates multiple connections between areas of interest/information, so that an immensely complex (mathematically, multidimensional) network of interests is formed. Developmentally, this is a highly dynamic system within which conservative structures of progressive complexity emerge out of interaction with the environment.


actIve awareness of others  – absence
active awareness of context,  – absence
accessible and densely connected information,         
[coarousal and connection]                    
– {+retrieval probs)
 complex value system,{thin meaning
variety of expression,
flexibility of response,– {+rigidity
stereotypy & echolalia
idiot savantry
the capacity for simultaneous awareness of the actual and the possible which is a requisite of imaginative play and serious planning.– poor or absent planning and playing
Hand-drawn illustrations of a polytropic interest system, and a monotropic one. Both show various inputs, and expressions (motor activity), and a peak interest/performer. In the monotropic diagram, there is little else; on the polytropic side, there is a cluster of interests around an 'observer/other', and many other smaller interests aroused to varying degrees.

To have a meaning is to have a place in an interest system: the more diverse and connected the interest system, the more subtle, deep,or ramified will be the meanings it confers through Interpretation or expression.

The (subjective) relevance of input and its cognitive effects depend on current states of arousal within such a complex system. Being polytropic makes an interest system flexible and adaptive, new information rarely takes it by surprise, or fails to be integrated. Responses in the shape of action and expression are informed and appropriate. Such actions and expressions may be imagined as well as actual. They are often tailored for an audience, modelled as one of many mini interest systems within the whole, from which one can evoke a response – hopefully akin to how the actual person would respond, and distinct from one’s own.

The table [above] sums up and contrasts the features of a polytropic and a monotropic Interest system: it conveys the essential content of my talk. The rest of this paper is a response to some of the thought-provoking questions I was asked in Durham, particularly those of Elizabeth Newson. She aked me how I thought my ideas would help work with people with autism,and remarked that I seemed to be talking about “social empathy”. I agree. I try as hard as I can, below, to imagine what it would be like to be monotropic, since I believe that’s a route to understanding autistic individuals and the difficulties they face. Encouraged by Paul Shattock’s “one of the advantages of a presentation such as this is that it gives one the opportunity to speculate and hopefully promulgate some discussion”(1989, p.137), the rest of this is blatant speculation.

Adaptivity and Expectation

An ordinary, alert, polytropic interest system is ready for almost anything: it is adequately primed over-all. That means appropriate Information (where known) is easily accessed, and multiple waves of cognitive effects will occur in sensitive response to relevant input (see Sperber and Wilson, 1985). Such an adaptive and well-adjusted interest system is the product of a long developmental history of learning. Clearly some people with autism do eventually develop quite densely structured interest systems (especially the more linguistically able), but their connections are much thinner than ours, and the whole system is much less elastic.

If we think about the early days of a monotropic existence, the picture is of intense engagement with a focal object for some period of time, followed by intense engagement with some other focal object. The system lurches from one top-heavy monotrope to another, from one hypersensitised perception to the next. While everything within each attention tunnel will tend to be strongly interconnected, a single connection will be formed between interests: very little in the way of preparing the system to respond to a changing world will have occurred. A series of little modules will form, which may or may not ever be triggered again: they are not enmeshed. Some may sink without trace whilst others assume great importance and get triggered often (see below for Why).

Instead of being a system with relative over-all sensitivity and readiness,a monotropic interest system will be very unevenly primed, and generally rigid. What is interesting will be very interesting indeed, what is merely disturbing will be tolerable through exclusion rather than assimilation.Unpredictable change will be the norm, and coping with it will be a great challenge and a painful effort every time. The world outside the chosen attention tunnel will be unfocussed and nonsensical; complex stimuli which can’t be interpreted without access to information from more than one source will be completely baffling. Panic will always be near.

Predilections for Predictability

In a world of incomprehensible flux, how soothing it must be to know for sure what’s going to happen next. Change in the world is at least as salient for monotropic individuals as it is for us4: everything else being equal, it tends to arouse their interest. But unavoidable unpredictable change will tip them towards panic – hence the longing for guarantees, with which we can all surely empathise.

What are the circumstances in which predictions are reliable?

In general, the more information you can bring to bear on the world, the more predictable you’ll find it. Something is perfectly predictable only when you know everything you need to know to make that prediction.Monotropic individuals will only be able to make safe predictions when all the relevant information is already ‘there’, within the bounds of the current interest. They won’t be able to put two ‘from here’ and two ‘from there* together. And in a monotropic system, interests are sharply distinguished from one another and have no close neighbours, so the usual ‘bleeding in’ of adjacent information which benefits ordinary thinkers is aiso absent. Not only is predictability particularly highly valued by people with autism, it is also scarce.

The occasions when enough is known that a monotropic prediction is safe are of just two kinds as far as I can tell5. The spinning of a wheel, or the spinning of a coin with its utterly reliable teetering decline, or a train running down the track – these are all objects which reliably keep on doing the same thing: responding with interest to them requires no adaptation; instead, it fully satisfies the expectation brought to it.Music, timetables, calendars and snooker balls all have a highly reliable sequence, Invarlable in principle, and almost invariable in practice: you always know where you are with them (NB. music and snooker also characteristically have a fixed, hushed setting).

Most of us find our own behaviour highly predictable, but I believe that is not initially the case for monotropic people. Predicting changed behaviour seems to involve simultaneous awareness of two behaviours, current and future. I am guessing that they can’t imagine doing something different whilst they are currently engaged, any more than they can imagine someone else’s interests whilst remaining aware of their own. I believe their attitudes to their own behaviour and their attitudes to the behaviour of objects are closely parallel. Repetition and fixed sequence are safe,anything else tends to be deeply troubling and invite retreat into a familiar, reliable, monotrope.

Even if their own behaviour is hard for them to predict, at least it accords with their own inclinations, and is more or less under their own control. They can make sure that what happens in their life is that, eg.,they keep on biting their thumbs, or following their routines, whatever else may be going on around them. Other people are a completely different kettle of fish: unpredictable and uncontrollable. Worse still, they won’t leave you alone, they keep reaching into your interest system and trying to yank it about.

Other People

I have long argued that ordinary people characteristically use language as a tool for manipulating6* each other’s interest systems. Though it’s a much more primitive tool, working with children with autism has made me see gaze that way, too7. We invade each other’s interest systems with our speech and our gaze, and try to change them.

By definition people resist going against their inclinations. But most of us are fairly elastic, and often yield with barely a struggle. For a monotropic person, however – for whom every shift of interest is abrupt -I believe the force of other people’s interests trying to change theirs (trying to line them up with their own) is like a wrench each time. I believe it actually hurts unless applied with great delicacy – which is not the human norm even between the most loving mother and baby. To a monotropic individual, all human contact will tend to feel like coercion.It’s no wonder, then, that people with autism generally choose to avoid people’s gaze, and make a practice of excluding speech from their attention tunnel wherever possible. Other people hurt.

It’s sometimes suggested that there is a fundamental problem with attachment in people with autism,8 I don’t believe that’s quite right. I think the fundamental problem is that fear of being pushed around makes them shut other people out, yet they need attachment just as much as any child (or human) does. My subjective Impression is of strong attachment to a few trusted people: people who can be relied on for continuity of presence, and of behaviour, and who are “tuned in” to that individual’s interests. They’ll be there, they’ll be as usual, and they won’t feel coercive.

As illustration 1 shows, a polytropic Interest system is peopled with Other models, through which we can simultaneously be aware of other people’s viewpoints whilst retaining our own. I like Daniel Stern’s name for these,ie. ‘evoked companions’9. They play a crucial role in ordinary reflective thinking10, as does the polytropic capacity itself. It may even be that we normally get shifted into a polytropic mode of functioning by other people’s interests, in the first place: that it takes other people to “kick-start” reflection, as both Hobson (in effect) and Jordan and Powell (literally) have suggested11, Children with autism may be caught in a double bind: unless they cease to be monotropic they won’t be able to bear letting other people in; but they won’t cease being monotropic unless they do let other people in.

If we assume – as I am inclined to – that all babies are monotropic to start with, why don’t they just stay that way? It may well be that in our model a small difference in the parameter setting will make all the difference between a rigidly monotropic system and one which can comfortably (smoothly) be pulled about by other people, and so happily receive the stimulation of their interest. It’s also possible that there is a quite narrow band of paramaterisations at one end of which a very delicate input of other’s interest could lead the baby out of its monotropism, while at the other end a thrusting input of other’s interest could drive it into monotropism. Out of range at one end, some people will be monotropic come what may, whilst out of range at the other end only a major injury to the system could push it into permanent monotropism12. This is not a question of good or bad caring; it’s a question of force of engagement in the effort to get the baby involved13. Most babies respond very well to enthusiastic efforts to engage their interest, and such efforts are a normal part of parenting. But for babies with autism it may swiftly become intolerably intrusive once they have learned to recognise what’s going on.

Learning and Being Taught

Most people find they learn better when they learn for themselves – out of autogenous interest – than when they do when they are taught. But some teachers seize your interest and make it their own. They draw you into their interest systems, make you eager for the information there. Their input somehow promotes autogenous interest in a new place. Succeeding in teaching a monotropic person with superhigh autogenous interest and a deep reluctance to shift it, is a great challenge. In fact, given what I’ve been saying, shouldn’t it even be impossible?

Well, the picture isn’t that bleak for most children with autism, as most readers of this will know well. For one thing, there are of course huge differences between such children (a point I have not been emphasising – but it fits our model). For another, once a fairly structured cognitive landscape has eventually been established in a more able child with autism,it facilitates activity outside the peak interest even though the picture goes on being essentially monotropic. Most important, in the right circumstances it seems to be possible to generate in many children with autism a willingness to be taught.

Within a framework of routine, and in an atmosphere of trust in which they know what’s expected of them, they can risk letting a little interest be drawn by a teacher. If they’re well supplied with this scarce resource,and if they have good language, they may even be able to learn a great deal quite well outside focal attention – thus avoiding the risk of overload (Donna Williams’s autobiography is very interesting on this). However, it remains true that for people with autism, as for us all, what is best learnt is what is chosen. Trying to draw the interest of a monotropic person and generate activity where it is not naturally inclined to be can be painfully laborious for both parties. What is worse, if I am right,is that what they learn in this way is unlikely to have the same sharp-edged connectedness and easy access as what they have learnt through autogenous interest.

Maximising their control over what they learn whilst making sure new relevant information comes their. way, would seem to be the most appealing pedagogic option for chidren with autism. Computers and specially designed computer programs should be uniquely fit for this purpose14:controllable, predictable, potentially limitlessly informative, and capable of clear, easily understood expression if used that way. (‘Facilitated communication’ is, I suspect, facilitated more by the machine itself than by the ‘helping’ person15.) What is more, computers could be used to network people with autism, both with each other and with non-autistic people, and so put them into a comfortable mode of communication. If this is right, then it should be a priority to make sure that as many people with autism as possible have access to computers, from as early an age as is feasible.

My main conclusion from these thoughts is that people with autism deserve a lot of sympathy, and a lot of respect. I believe they deserve a little envy too: having autism has positive aspects as well as negative ones. So long as they are safe, and feel safe, their intensity of awareness must be wonderful rather than terrifying. What is not comprehended, what is not a mere part of a whole, nor an example of a generality, will be extraordinary anyway: an object of wonder not indifference. Furthermore, most ordinary people are spread thin, with dozens of coactive concerns each taking a bit, and little or nothing left over for appreciation of the beauties of the material world or the beauties of pure calculation. People with autism I believe retain a capacity for wonder that most of us lose in infancy – it’s the positive side of bafflement.

Ordinary lives are spent performing and conforming to other people. People with autism don’t do that, and they don’t have the same desperate greed for other people’s attendant energy16. Ordinary people are constantly trying to get into each other’s interest systems and do things there.People with autism are innocent of this, they neither do it nor have it done to them. Ordinary minds are constantly being affected by other people’s, by other people’s expectations and manipulations. People with autism are extraordinarily resistant to those coercive pressures: in a certain sense, in even the worst tyranny they’11 go on being free.

Dinah KC Murray, ex presentation at Durham, April 1992, revised May 1992


For the development and interpretation of the mathematical model I am deeply grateful to Professor Peter Allen and Mike Lesser, as I am to Jack Corliss for his brilliant graphic realisations and to Grant Worrell for his recent amazing labours.

For the development of my understanding of autism since I finished my earlier piece on the topic (in press), I’m grateful to a great many people. I have to thank several of the staff of Harborough School for putting up with my fairly useless presence in their midst once a week since the beginning of 1991, And I have to thank the children there for getting used to me and getting on with me, especially Ferenc, I’ve learnt a lot about autism from being with and talking with all concerned. Discussions (spoken and written) over a period of time with a number of people who know more about autism than I do have been stimulating, illuminating, and encouraging. I’m thinking here particularly of Francesca Happé, Rita Jordan, Stuart Powell, and Paul Shattock – to all of whom I’m most grateful. I also want to thank the participants in The National Autistic Society’s 1992 Durham conference for their interested and helpful questions. Comments from John Morton on the first draft of this were also very helpful.

Lastly I want to thank Donna Williams for her determination to understand herself, and for the extraordinary book (Nobody Nowhere) which resulted.


  • Argyle, Michael (1975). Bodily Communication, London: Routledge.
  • Frith, Uta (1989). Autism: explaining the enigma. Oxford: Blackwell.
  • Frith, Uta (1989b). “Autism and ‘theory of mind’” in ed Gillberg, C. (1989) Diagnosis and Treatment of Autism. New York: Plenum Press.
  • Hartmann, Hellmut (in press). “Self-organisation of cognitive processes and psychosis – development and test of a theoretical model.” Acta Paedopsychiatrica. See also, this volume.
  • Hobson, R. Peter (1989), “Beyond cognition: a theory of autism” in ed. Dawson, Autism: Nature, Diagnosis and Treatment. New York, Guilford Press.
  • Hobson, R. Peter (1990). “On psychoanalytic approaches to autism”. American Journal of Orthopsychiatry, 60(3), July.
  • Hobson, R. Peter (1991). “What is Autism?” Pervasive Developmental Disorders. 14 (1),March.
  • Jordan, Rita (1990). The Option Approach to Autism: Observer Project Report. Available from the author, Shool of Humanities and Education, Hatfield Polytechnic, Wall Hall Campus, Aldenham, Herts., WD2 8AT UK.
  • Jordan, Rita and Powell, Stuart (1990a). The Special Curricular Needs of Autistic Children: Learning and Thinking Skills. London: Association of Head Teachers of Autistic Children and Adults.
  • Jordan, Rita and Powell, Stuart (1990b) “Improving thinking in autistic children using computer presented activities.” Communication, 24,2.
  • Kanner, L. (1943) “Autistic disturbances of affective contact.” Nervous Child, 2.
  • Leslie, Alan M. (1987). “Pretence and representation: the origins of theory of mind’” in Psychological Review 94, 412-426.
  • Morton, John (1989). “The origin of autism”, in New Scientist, 9 Dec.
  • Murray, Dinah (1987), “Silent speech acts and their cognitive effects.” in ed. Verschueren, J. and Bertuccelli-Papi, M., The Pragmatic Perspective. Amsterdam/Philadelphia
  • Murray, Dinah (in press) “Holy Fools” in Talking, Thinking, and Wanting.(Routledge).
  • Powell, Stuart and Jordan, Rita (1991), “Using pedagogical principles to remediate the thinking of children with autism”, presented at Durham, at the 1991 conference, “Therapeutic Approaches to Autism: research and practice”, published in series with this volume.
  • Shattock, Paul (1990). “Some implications of basic physiological research for the behaviour and treatment of people with autism” presented at the International Conference, Experimental Psychology and the Autistic Syndromes, at Durham University, published in series with this volume.
  • Shattock, Paul and Lowdon, Gillian (1991), “Proteins, Peptides and Autism Part 2: Implications for the Education and Care of People with Autism.” Brain Dysfunction.
  • Sperber, Dan and Wilson, Deirdre (1985). Relevance. Oxford: Blackwell.
  • Stern, Daniel (1985). The Interpersonal World of the Infant. New York:Basic Books.
  • Tannen, Deborah (1991), You Just Don’t Understand! USA, William Morrow London, Virago.
  • Van Engeland, H. (1990). “Information processing deficits in early infantile autism: a psychophysiological approach.” Presented in Durham, see Shattock (1990) for details.
  • Williams, Donna (1992). Nobody Nowhere. London, Australia, New Zealand,Canada. Doubleday.


  1. See Jordan and Powell (1990a). ↩︎
  2. See Frith (1989), and Murray (in press). ↩︎
  3. This is a response to a comment from John Morton. ↩︎
  4. Helmut Hartmann’s work is relevant here, as is H. van Engeland’s. ↩︎
  5. The short film by Tim Webb, A is for Autism, and Elizabeth Newson’s discussion with me after my talk have both helped focus my ideas on this topic. ↩︎
  6. ‘manipulate’ here need have no negative connotations, it just means ‘take hold of and do something with…’ ↩︎
  7. My thoughts on this have also been stimulated by Deborah Tannen’s (1991) perception of gaze avoidance as dominance avoidance – she finds it’s commonplace among males.
    See, eg., Kanner (1943) and Hobson (1991) – a comment by Francesca Happéprovoked my thoughts on this. ↩︎
  8. Anyone who finds the idea of ‘theory of mind’ a bit thin should read not only Hobson (1990. 1991), but also Daniel N Stern’s fascinating account of The Interpersonal World of the Infant. ↩︎
  9. See Stern (1985). ↩︎
  10. See Murray (1987). ↩︎
  11. See Powell and Jordan (1991). ↩︎
  12. See the discussion in Shattock and Lowdon (in press). ↩︎
  13. See Stern’s sensitive account of this. ↩︎
  14. Stuart Powell’s program is the only one I know specifically designed with autism in mind for children with learning difficulties. There will surely be many more in years to come. ↩︎
  15. See Jordan and Powell (1990b) and Carol Rahne, this volume. ↩︎
  16. I have recently learnt the wonderful expression “attendant energy” from Eve Grace. It apparently comes from acting theory. I believe people with autism do like to receive attendant energy, but only when it is strictly non-invasive. Since that’s the exception not the rule, they’d generally rather do without than risk invasion. ↩︎

Skip to content