What is it that autism therapies do (if they do not cure autism)?

Gil Eyal and Brendan Hart

(Talk given at the Netzwerks Entresol conference at the University of Zurich, 6/14-15/2013)

Abstract: There are many different ways to think about or evaluate the therapies, interventions and
treatments given to children with autism. In evaluation studies, they would be judged in terms of their
effectiveness in reducing certain symptoms and/or in terms of their costs. More critical research seeks to
assess therapies in terms of how normalizing they are or tolerant of cognitive difference. In this talk, I will
propose several other ways to think about the significance of these therapies, and understand them as
constitutive features of the contemporary experience of autism. First, therapies may be analyzed as
mechanisms for translating and aligning the interests (broadly construed) of the different parties involved
in the therapeutic act - patients, parents, therapists, teachers, doctors, etc. - thereby creating stable
networks of expertise. Second, they may be analyzed as grids of specification that target, differentiate and
name certain specific aspects of behavior, communication and comportment, thereby giving shape to the
specific way in which we understand and diagnose autism today. Finally, autism therapies may also be
understood as engaging in “ontological politics” (Mol 2002) and elaborating an infrastructure supporting
autistic personhood.

Our topic is the role that autism therapies have played in the process of autism’s
transformation into a widely diagnosed and prevalent disorder. Before we get to the therapies,
however, we need to clarify our approach, which requires summarizing rather briefly and
schematically the argument of our book – The autism Matrix (Polity 2010).
The question that we tackled in the book was how to understand and explain the
precipitous rise in autism diagnoses that has taken place over the last 25 years. Currently, the
CDC estimates that the prevalence of autism has increased from 1 in 2,500 in 1989 to 1 in 88 in
2008. Some people talk now about 1 in 55. How can this be explained?

Caveat: We do not have time in this lecture to discuss the other explanations currently offered. They
typically fall into either “realist” or “social constructionist” versions and suffer from the problems and
limitations characteristic of these approaches when it comes to dealing with – as Ian Hacking calls them –
“human kinds”. At the Q&A we will be happy to explain what other explanations exist; what are their
advantages and limitations; as well as in what ways the account offered here seeks to go beyond them.

In the book, we suggested that a better way to proceed would be to invert the question.
Instead of asking why autism is more prevalent now, one should ask first why was autism rare in
the past? This question leads one beyond “realist” accounts of the rise of autism, since it directs
attention to the conditions that are necessary for autism to be identified, differentiated and
diagnosed on a large scale.
Posed in this way, one such condition immediately comes to mind. It is so obvious. All of
the 11 kids about whom Kanner (1943) wrote his original article were brought to him because
their parents disputed a diagnosis of “feeblemindedness” given to their children. Indeed, Kanner
(1971) did a follow-up study of the 11 children he diagnosed. 4 of them have spent their lives, as
it turns out, in institutions for the retarded. So without engaging in retroactive diagnosis, we
think it is fairly safe to assume that many kids, who would today qualify for an autism diagnosis,
were in the past diagnosed as retarded and institutionalized. Kanner (1949, 11) claimed he knew
several such cases, and Bernard Rimland (1964, 13) after him simply asserted this as fact. The
crucial point is that in a large state residential institution for the retarded, it would have been
impossible to identify and differentiate autism on a large scale. In the institution, autism was
simply meaningless, below the radar, because the institution produced autistic symptoms on a
massive scale – withdrawal, indifference and self-stimulatory behaviors - these were reactions to
institutional environment that already in 1945 Rene Spitz identified as “hospitalism,” and as akin
to autism.

That’s why in the book we emphasize that the deinstitutionalization of mental retardation
– a lengthy process which begun in the early 1970s and lasted at least two decades - and which
created a new institutional matrix composed of community treatment, special education, and
early intervention programs has been crucial to creating the conditions wherein autism could be
identified, differentiated and multiplied.
Deinstitutionalization played a crucial in the expansion of autism, however, not simply
because the children were now at home and could be observed and diagnosed. There was a
deeper reason. From the moment that autism was first named by Kanner (1943), it represented a
blurring of the boundaries between mental retardation and mental illness. The children referred
to Kanner were originally diagnosed as feebleminded. He himself declared the condition was
inborn, perhaps inherited. Yet, the parents brought the kids to Kanner because they disputed the
diagnosis, and Kanner (1943, 39) agreed with them: “Even though most of these children were at
one time or another looked upon as feebleminded, they are all unquestionably endowed with
good cognitive potentialities.” How did he know that? Kanner (1943, 39-41) pointed out that the
parents of autistic children were typically middle class, in intellectual professions, and highly
intelligent. The kids also did not look feebleminded. They were well-formed, even pretty. We
may snicker, but contemporaries, especially the parents, took this as fairly convincing proof.
Rimland (1964, 28-29) still cited this as evidence that the children were not feebleminded. At the
time, there was an institutional and diagnostic division of labor between child guidance clinics
which treated the behavior problems of middle class children, and the institutions for the retarded
which held mostly lower class adolescent “morons”. The fact that the children were middle class
and that there was no evident genetic defect in the family was taken as proof that this was not
feeblemindedness. Despite the class bias, I would argue that this was a profoundly liberating
moment. It would not be the first time in history that a middle class sense of entitlement paved
the way for a more universal extension of rights. To stay with Kanner (1949), he suggested that
there was a form of feeblemindedness that was only “apparent”, not real, and that often it was
really autism underneath the appearance. In such cases, he said, the children should be
institutionalized, but should be given intensive therapy to realize their potential.
But if they were not feebleminded, what were they? And how to make sense of certain
similarities between them and their parents’ intellectual and cold demeanor? Kanner (1943)
defined autism as involving a “disturbance of affective contact”, i.e. something that at the time
was understood along the lines of mental illness or neurosis. Indeed, two of the 11 kids were at
one point hospitalized as schizophrenic. At other times, Kanner (1949, 27) toyed with the idea
that maternal coldness might have caused mental illness in the children. He quickly abandoned
this idea, but his equivocation is highly instructive. One can say that he was ambivalent, or
changed his mind, or shrewd, but what is more significant is that this ambivalence meant that
autism served as a “wedge” with which a space between illness and retardation was pried open, a
“zone of indiscernibility” (Foucault 2003 [1974-1975]) where it would be impossible to
determine – within the understanding of the period - who is mentally ill, who is feeble-minded,
or what have you.
From then onwards the history of autism can be told as a series of attempts to bring this
in-between domain into being. Deinstitutionalization finally created the conditions to do so. It
formalized this zone of indiscernibility as the new category of the “developmentally disabled”.
The developmentally disabled were no longer to be held in custodial institutions, but would
receive treatment. Yet, their treatment was not oriented to a cure, as in illness, but to
“habilitation,” that is, enabling, making able.
The irony, however, is that it was not Kanner or child psychiatry who finally succeeded
in this task, but a completely different actor – the parents of children with autism. The second
reason, therefore, why autism was rare in the past, is because it was impossible to bring into
being this in-between domain without enlisting the support and collaboration of middle-class
parents. To put it in a nutshell, the blurring of boundaries between mental illness and mental
retardation could not be achieved without a corresponding blurring of boundaries between
experts and laypeople. Just think about it. In the previous arrangement, children’s destinies were
sorted by the institutional distinction between child guidance (mental illness/neurosis) and the
residential institution (retardation), which neatly corresponded to and confirmed class
distinctions. Without this institutional presorting, it becomes the task of the clinician to diagnose
and decide social destiny: who has potential and should get intensive therapy and who doesn’t.
This form of power – medical diagnosis of social destiny – is much too formidable and fraught to
be wielded by any one person or profession. It cannot be exercised without striking a coalition
with middle-class parents. That’s why another set of conditions necessary for the expansion of
autism was a thorough rearrangement of the relations between experts and laypeople, in the
course of which the parents of children with autism came to play the leading role in autism
diagnosis and treatment.

Note: We cannot tell this story here. The crucial point, however, is that the parents did not simply create an
association to advocate for their children and de-stigmatize themselves, or even to provide services not
provided by the state, but they created an alternative network of expertise, in which they played the leading
 role. They joined in the contemporary attack on psychiatry – of which deinstitutionalization was an
important effect and catalyst – by demonstrating that psychiatric diagnosis was unreliable and biased
(Rimland 1971), and that it obfuscated the considerable contribution made by parents (Sullivan 1984).
They developed their own networks for collecting data and their own checklists for diagnosing autism
(Rimland 1964; 1968). In due course, the National Association for Autistic Children played a leading role
in rewriting autism diagnostic criteria for DSM-III and DSM-III-R. (Ritvo and Freeman 1978; Waterhouse
et al 1992) Most importantly, and this is part of the story I will detail below, they formed a coalition with
psychologists, occupational and speech therapists to develop and adapt effective interventions for their
children, conducted, directed and managed by parents in their homes. (Rimland 1972; Park 1974)

What kind of explanation is this? The key concept is taken from Ian Hacking (1995a;
1995b): “the looping effects of human kinds”. It means that the very attempt to name, classify
and diagnose certain human conditions sets in motion processes that act back upon the medical
classification and continuously change it, so that the condition – “autism” – is a moving target.
(Hacking 2007) The autism epidemic, therefore, is neither a naturally occurring event, nor a
socially constructed fiction, but another spiral, perhaps the penultimate one, in an increasingly
widening vortex of looping processes:
 The first spiral was initiated in the 1960s, as the newly invented autism diagnosis became a focal
point for parents to organize around the stigmatized (because of the implication of parental
causation) identity of “autism parents,” and create their own network of expertise.
 The second spiral followed in the 1970s with deinstitutionalization. It meant that a large group of
developmentally disabled children were not to be treated at home, at school and the community.
In this context, the behavioral and sensory therapies adopted by the parents’ network gradually
reshaped the autism prototype from a rare syndrome, the object of clinical observation, into a
spectrum cobbled together mainly on the basis of response to therapy. Thereby they changed what
clinicians were most likely to see in their offices, and the range of possibilities of how to be an
autistic person.
 The third loop came in the 1980s, with the emergence of self-advocates, autistic individuals who
were high-functioning and capable of representing themselves in the media. As Ian Hacking
(2009) argued, they elaborated a language of neuro-diversity with which to represent a thick
autistic person (that is, one with emotional and cognitive depth, whose actions are meaningful and
intentional if hard to decipher). They were now providing parents of even lower-functioning
children with a language to represent and enforce claims on behalf of their children.
 Hence, the final chapter in the looping dynamic came in the 1990s with massive diagnostic
substitution from mental retardation to autism.

We cannot possibly discuss all the elements in this decidedly non-parsimonious and
admittedly ungainly explanation in the space we have left. Readers are referred to our co-
authored book (Eyal et al 2010), where they would find also some clues about the story of
autism’s rise outside the US and the UK, which serve as the focus for the present discussion. The
rest of this paper, however, is dedicated to the role played by therapies, i.e. the second looping
spiral, though without limiting myself to the time period of the 1970s. We want to argue that
autism therapies have played a crucial role in creating the autism epidemic, precisely because
they are an important and largely overlooked mechanism through which the “looping effects of
human kinds” proliferate.
The first order of business, however, is to explain what we mean by autism therapies and
what is included or not in the discussion. Quite deliberately, we are using a very eclectic
understanding of therapy. Our corpus is composed of whatever interventions the parents of
children with autism report they use with their children. Green et al. (2006) found that parents in
the U.S., Canada, and Australia used 108 different types of treatments in total, with a mean of
seven per child. Looking at only the top ten used interventions,1 it is clear that it would not be
useful to impose a pre-existing definition of what is therapy and what is not, since what
characterizes this list is precisely its eclecticism and the blurring of boundaries between what is
medically approved and what is not. Bona fide medications like Risperdal and Melatonin appear
only on nos.24-25 on this list, and they are clearly meant to treat specific symptoms, not the
condition itself. Nutritional supplements and alternative medicine, on the other hand, are clearly
widely used. A survey of 176 families in Alberta, Canada found that 93.8% of children with
autism in that area were being treated with complementary or alternative medicine (Gibbard
2005). No less importantly, it is clear that it would not be useful to reserve the term “therapy” for
a medical or psychological intervention, and exclude educational tool like visual schedules and
PECs). Indeed, we argue that this blurring of the boundaries between medicine and alternative
medicine, between psychiatry and special education, between medication and assistive devices,
between therapeutics and prosthetics, is what makes autism therapies socially innovative and
successful.

Nonetheless, for purposes of brevity and focus, the following discussion is limited to a
shorter list of interventions: Sensory Integration Therapy (SIT), Applied Behavioral Analysis
(ABA), TEACHH (also known as “developmental therapy”), “floortime” (also known as DIR),
Facilitated Communication and Rapid Prompting Method (RPM). There are several reasons why
we’ve chosen to focus on these. First, because autism therapies are not only eclectic, but also
modular. Rather than specific interventions, SIT, ABA , TEACHH and DIR are really
comprehensive treatment programs that incorporate many of the other interventions listed in
Green et al’s (2006) table, as well as borrow liberally from one another. Thus, visual schedules
and PECs are typically employed by ABA therapists, but also as part of TEACCH or “floortime”
programs. Medications, diets and dietary supplements can and are combined with any of these
programs. Part of the eclecticism we noted above is not merely the parents’ willingness to
employ “whatever works”, but also the fact that the various treatment programs pragmatically
borrow from one another. Thus, the TEACCH program employs elements of operant

1
They are, in this order: speech therapy, visual schedules, Sensory Integration therapy (SIT), Applied Behavioral
Analysis (ABA), Social stories, Vitamin C, Vitamin B6, Essential Fatty Acids, Picture Exchange Communication
system (PECs), and a Caesin-free diet. (Green et al 2006)
conditioning together with SIT sensory gym, as well as a “structured educational environment”.
The second reason is that these therapies or programs are historically significant. ABA, SIT and
TEACCH all date from the 1960s. They began the new era in autism treatment and are the most
responsible for the looping spiral we are analyzing here. They are the early pioneers and the
heavyweights. The other therapies are included here mostly in order to illustrate a specific
contrast or point. Facilitated Communication, for its part, generated a great deal of controversy
and represents the point at which autism interventions most clearly shade into self-advocacy.
Floortime is significant because its “child-directed” nature is often contrasted with ABA and
TEACCH, while RPM is significant because it was developed by a parent. Ultimately, however,
the main function of this list is illustrative, and others could be suggested.
Lest we be misunderstood, let us immediately say that we are not arguing that these
therapies are efficacious. That’s not why we think they are important. None of these
interventions “cure” autism, and our reading of the existing evaluation research is that the claims
of “recovery” on behalf of existing therapies – ABA included – are impossible to verify and
likely greatly exaggerated. (Rutter 1983; Howlin 1997; Rogers and Vismara 2008) But this is not
the level at which we would like to analyze autism therapies. We don’t want to measure autism
therapies against an external yardstick, but on the contrary, we want to see how they themselves
define the goal of intervention.

So, what is it that autism therapies do, if they do not cure autism? How to analyze their
significance if their efficacy is unproven? In order to tackle these questions, we propose an
alternative approach to therapies as, first, mechanisms for translating the interests and securing
the cooperation of patients and parents; second, as grids of specification that target, differentiate
and name specific aspects of behavior, and; third, as “ontological politics” and an infrastructure
supporting autistic personhood. Let us begin with the first one, which in a sense is our answer to
the question of the therapies’ success. Instead of looking at whether they are efficacious or not,
we need to analyze the therapeutic exchange on its own terms: What is it that the therapies give
to parents and patients? What is it that they ask in return? And how do they secure cooperation?
This question can be approached by means of a brief contrast between autism and
childhood schizophrenia. Just as we are told today about an epidemic of autism, people in the
1950s and 1960s warned about an impending epidemic of childhood schizophrenia. (Mosse
1958; Bender 1953) Yet, by the early 1970s childhood schizophrenia was thrown to the garbage
bin of history (Rutter 1972), and autism took its place. What can explain this rapid change? At
least part of the answer had to do with the prescribed treatment for childhood schizophrenia: 6
months of hospitalization, in the course of which children got 20 rounds of electric shock
treatment and/or Metrazol, a medication that induced convulsions. (Bender 1953) Now, our point
is not that this doomed childhood schizophrenia because it was a harsh, even cruel, treatment, or
because it was not efficacious. Both are probably correct, but we already saw that it is not clear
in what sense autism therapies are efficacious, and some of them, especially ABA, were also
condemned as rather harsh. Lovaas was notorious for screaming at the children and even
spanking them. Our point is that this treatment failed to translate the interests and secure the
cooperation of the patients and their parents. The parents were excluded from the process of
diagnosing and treating their children, who were literally taken away from them and
hospitalized. The therapy was not only harsh, it was standard issue, one size fits all. To
compound matters, most of the patients were minority adolescents who were referred to
treatment after being labeled “juvenile delinquent” because of suspected gang membership or
truancy. After discharge, they were still tagged with a diagnosis of “behavior disorder” while the
side effects of the treatment were often visible. No wonder, then, that the parents eventually
refused to allow their kids to be treated with ECT, and Bender reported increasing difficulties in
recruiting new patients. Put differently, precisely because the treatment of childhood
schizophrenia attempted to imitate the model of acute medical care – illness-hospitalization-
treatment-cure-discharge – securing the cooperation of parents and patients was completely
staked upon obtaining a “cure”, and when this failed to materialize, the initial begrudging
cooperation turned into open resistance.
Autism therapies contrast, point-by-point, with this medical model. We’d like to discuss
three characteristics of autism therapies that contrast with the medical model and that allow them
to secure the cooperation of parents and patients: generosity, co-production and their elective
affinity with the peculiarities of children with autism.
We take the term “generosity” from Nikolas Rose (1992). Instead of knowledge being
monopolized by the expert, it is shared with laypeople, especially parents, thus permitting them
to become “experts on their own children”. Knowledge is not esoteric, but on the contrary, is
crafted so it can be easily incorporated into the practices of laypeople. There are several
indications of this generosity.

The most obvious one is that autism therapies outsource the conduct of therapy to
parents. While the treatment of childhood schizophrenia took place in the hospital, away from
the parents, and without their involvement or drawing upon their knowledge, autism therapies
like ABA and TEACCH secure the cooperation of parents by involving them in the conduct of
therapy, and by treating them as “experts on their own children”. This was so from the very
beginning. Not long after developed ABA, he was visited by Bernard Rimland, the father of a
boy with autism. Rimland watched Lovaas working with children. He was so impressed with
what he saw that he tried the method on his own son, and also gave a speech about Lovaas’
method at the founding meeting of the National Society for Autistic Children (NSAC), extoling
it as a “technique that parents could learn with demonstrable success.” Lovaas (1993), for his
part, calls parents his “colleagues” and says that given how time-consuming and labor-intensive
it is to work with autistic children, “this means that we will have to give away our professional
skills to lay people, and the sooner the better.” At the TEACCH program, parents are coached to
be “co-therapists” or even primary therapists of their children. TEACCH introduced a simple
device that gave material form to these new relations of generosity and transparency between
experts and parents – a one-way glass screen that permitted parents to watch clinicians and
therapists working with their children without being seen. The one-way screen, said Schopler
(1971), takes away “the mystique and unfounded authority” of the therapist. Gone were the days
when parents waited anxiously outside the closed door of the clinician’s office. Now they saw
exactly what the therapist did, and they quickly came to surpass the therapists who originally
trained them. They became “experts on their own children”. Similar arrangements obtain in
floortime, where the glass screen is replaced by a video camera. It goes without saying that RPM,
which was developed by the mother of a child with autism, is the logical extension of this
generosity.
Not only do parents play an active role in therapy, but therapy is grafted onto the
everyday work of parenting. ABA “discrete trials” are used to teach everything that parents need
their children to do: toilet training, sleep training, how to hold a spoon, how to speak. They have
a modular structure. They are short, with a clear beginning and end, and can be easily inserted
into the routines of the household, or into the few free “teachable moments” that crop up. Despite
the major differences between them, “floortime” is similar in this respect, because the main
technique of “circles of communication” is similarly modular and can be incorporated into any
interaction or play sequence.
Not only does it become part of the everyday work of parenting, therapy is also infused
with the moral politics of parenting. ABA is like strict parenting: be firm, yell, even slap their
bottom, in order to make sure they are punished for failure. “Floor time,” on the other hand, had
built its cachet by explicitly opposing ABA’s model of strict parenting, enjoining parents to get
down to the floor to the child’s level, capture the child’s interest and mobilize her emotions,
rather than learning by “rote.” The opposition between ABA and floortime, therefore, is a
political one, like the opposition between liberal and conservative styles of child-rearing. (Lakoff
200?)
The second characteristic of autism therapies that allows them to secure the cooperation
of parents and patients is co-production. Just as generosity is the opposite of professional
monopoly over the supply of knowledge, so co-production is the opposite of professional
autonomy, namely control over the demand for knowledge. Instead of the goals of treatment
being defined by the expert, as in the medical model, they are formulated in collaboration with
the clients – in this case the parents. The treatment of childhood schizophrenia was oriented to
goals defined by Bender. Childhood Schizophrenia, in her assessment, was a disease that could
be cured only by “heroic” means. If cure could not be obtained, the alternative was discharge.
Essentially, nothing more could be done for them. Autism therapies, by contrast, formulate
intermediate goals that blur the boundaries between medical cure of illness and special education
and custody over retardation, and do so in collaboration with the parents.
The first step is to deconstruct the large unities of illness or retardation into smaller,
actionable units. Lovaas (1979, p. 315), for example, calls autism a “hypothetical” entity. His
point is that effective treatment cannot be directed at such a speculative entity, but instead must
take as its target concrete behaviors to be modified and skills to be acquired. ABA breaks the
behavior of autistic children “apart into smaller units of self-destruction, imitation, vocalizations,
units of grammar, labeling, etc.” and attempts to shape each one separately. Similarly, SIT was
not originally developed for autism treatment, but for learning disorders and ADHD. It works on
basic postural, vestibular, or perceptual mechanisms so as to “jump start” the faculty of sensory
integration. In TEACCH, as well, the point is not to cure autism but to provide building blocks
for cognitive and psychic growth so that autism “will become sufficiently camouflaged . . . to be
unnoticeable” (Schopler and Reichler 1971; Ayres 1974, pp. 56–62; Lovaas 1979, pp. 320–21;
1981, p. x). The exception that proves the rule is RPM, which does present itself as a therapy
specifically for autism. This is clearly because of its status as a newcomer still needing to secure
a niche for itself.
By claiming that they do not treat autism per se, therapists and the parents are freed from
the compulsion to seek heroic cure or recovery of the children. It’s almost like what happens in
Alcoholics Anonymous. You start by admitting that you have no control over drinking, precisely
so you can begin working on yourself and achieve some measure of control. Similarly, autism
therapies start by admitting that autism itself cannot be changed, only “camouflaged”, precisely
in order that they can begin a process of gradual, piecemeal improvement. Characteristically, this
process of co-producing therapeutic goals requires adjustment of expectations, translating the
interests of parents from a desire for “heroic cure” or rescue for their children, to an appreciation
for tangible improvements in functioning, self-determination, and quality of life. As Lovaas
(1981, p. 3) put it: “Set small goals so both you and your child will be rewarded. Find pleasure in
small steps. Do not struggle for an absolute and unattainable ideal of normalcy.” Accordingly, he
gives instructions on toilet training, sleep training, on how to administer a spanking. Nothing is
too low-tech. Moreover, therapists and parents collaborate in producing assistive devices like
adaptive spoons, with thick, easy-to-grasp handles, and of course the PECs and visual schedules
mentioned earlier.
The third characteristic of autism therapies that allows them to secure the cooperation
specifically of the patients is that instead of the passive “sick role” typical of the medical model
(and as was true for Childhood Schizophrenia), they assign the children an active role in the
conduct of therapy and gain their interest and trust by latching onto their peculiarities. They are
especially innovative in this respect. Let us quote at length here the report by David Park, father
of an autistic girl, about the success that he and his wife had in treating their daughter with
“operant conditioning” at home.

“This is how the system finally worked. Every week Elly sat down with her mother and together
they wrote a contract that specified rewards and penalties for certain actions. Actions not on the
contract were neither penalized nor rewarded. The contract was a negotiated agreement and
contained nothing that the child had not approved.” [The contract specified what words and
actions were to be rewarded. The girl gave herself points by using a wristwatch-like counter on
which she clicked a point whenever appropriate. If she made enough points at the end of the day,
as stipulated in the contract, she got a small reward.] “…It soon became clear that there were
several reasons the system worked so well:
1) It concentrated the child’s attention on a few problems at a time.
 2) The contract made no impossible demands…
3) The child was surrounded by success…
4) The child took a responsible part in her own treatment, which is even more important than
engaging the parents in it.
5) The system utilized and turned to advantage the peculiarities of autistic children: their interest
in systems, in counting, in ritual, and their literal honesty.
…The child…welcomes this liberation as much as the parents do. Elly has told us so herself.”
(Park 1974, 189-191)

Compare the report by two occupational therapists on the results of using sensory
integration therapy with a young autistic girl at her school district. Prior to therapy, the girl had
lost “verbal contact with people and reality.” Yet, after six months of tactile and vestibular
stimulation (including rubbing, brushing, spinning in a hammock-net, and riding prone on a
board scooter) there was marked improvement, as the girl began to “direct her own therapy,”
instructing the therapist to emulate the sounds of different animals while she spun or “firmly
directing a great deal of pretend play…involving the therapists, classmates, and the therapeutic
equipment.” (Ayres and Heskett 1972, 178–80).

We use these two stories not as evidence that these therapies were successful in treating
autism. For that purpose, they are purely anecdotal and worthless. The stories are useful,
however, for the purpose of describing the common mechanisms employed by behavioral and
sensory therapies to elicit the cooperation of the patients. First, treatment takes place in the
children’s home or school environment, is continuous with everyday activities, and is conducted
not by medical personnel but by parents or school personnel. Second, the goals of treatment are
modest, achievable, and pertain to everyday functioning, so not only is the child “surrounded by
success” but each success means further integration in the activities of the family, school, and
community. Compared with these modest goals, however, treatment is intensive and intrusive,
calculated to impinge upon the children’s initial resistance and involve them in interaction.
Indeed, if there is one thing that is common to all autism therapies, as Grinker (2007) says, is that
they all involve “being in your child’s face.” Schopler and Reichler (1971, p. 94) called it “the
adult’s non-specific impingement on the child”. This is very different from the leisurely pace of
earlier therapies that sought to slowly draw the child out of the “fortress” of autism he or she
erected. (May and May 1959; Lovatt 1962; Shore 2003) Yet to achieve this intrusion, therapy
seeks to graft itself onto the “peculiarities” and concrete behaviors of the children and turn them
to its advantage. Park’s observations about the intrinsic charm of counters, clickers, schedules,
contracts and rituals for the children have been echoed by others (Siegel 1996, pp. 230–52).
Sensory integration therapy similarly builds on what children find enjoyable. [Slide 25c] “Fun,”
says Ayres (1998, p. 7) is the word that children use for sensory integration. Most important,
these therapies involve the children as responsible participants in their own treatment. Park noted
this feature as crucial to the success he achieved with his daughter. Ayres (1983, pp. 127–28)
emphasized that often the therapist needs to follow the child and not vice versa: “Occasionally, a
child seems to know more about the nature of his dysfunction and what to do about it than the
therapist does.” Finally, common to all these therapies is that they individualize the patient,
tailoring for each child a unique treatment plan - and not the one-size-fits-all of Childhood
Schizophrenia - adapted to his or her deficits, strengths, and sensitivities.
Deinstitutionalization has provided a propitious ecology for these new behavioral,
sensory and educational methods of treatment, and by the end of the 1970s they largely replaced
insight psychotherapy as the main form of autism treatment. They did so despite the fact that
none of them could cure autism or significantly undo severe initial handicaps of intelligence and
language. (Rutter 1983, 210–11) The success and spread of these therapies, therefore, was not
due to their technical superiority but to the social innovativeness with which they were able to
graft themselves onto the practices of parents and patients, translate their interests, and secure
their cooperation. At the same time, however, the success of the therapies has meant that autism
was transformed from a syndrome with “cardinal symptoms,” as Kanner called it, to a list of
“items of autistic behavior” that could be worked upon separately and from a rare illness into a
wide spectrum of social and communicative deficits. This is because the therapies functioned not
only as mechanisms for securing the cooperation of the parents, but also as grids of specification
that target, differentiate and determine the diagnostic value of certain specific aspects of
behavior, communication and comportment, thereby giving shape to the specific way in which
we understand and diagnose autism today.

The first point is that deinstitutionalization acted as a sort of “moral blender” into which
disappeared the old categories that belonged to the custodial institutional matrix, giving rise to an
undifferentiated mass of children with “developmental disabilities”. Suddenly, it was not so clear
who is autistic and who is not, and importantly, it became less meaningful to strongly distinguish
autism from retardation, as Rimland did just a few years earlier. As Lorna Wing, psychiatrist,
mother of an autistic child, and co-founder of the British parents’ association put it in a speech to
the American association in 1973: “It soon becomes clear in the field of communication
problems how pointless it is to draw sharp lines between autistic and ‘not really autistic’ in
educational practice, even if this distinction is of great theoretical interest. The question to ask
about each child is – what are his handicaps, what are his skills and what can we do to help
him?” Note that this pragmatism was not only more suitable to the new realities of
deinstitutionalization, where very young children with disabilities were at home and at school,
but that it also fit together with the modus operandi of the therapies. Recall that Lovaas dubbed
autism a “hypothetical” entity, and that the techniques of ABA, he argued, were applicable
across all different diagnoses. Ayres said the same things about SIT.
Upon this new landscape, now lacking “sharp lines”, the therapies layered a new grid of
specification based on the internal economy of the therapeutic exchange, what it demanded of
the patients and how the patients responded to these demands. This grid of specification grouped
together and foregrounded certain behaviors, especially those pertaining to communicative
rapport, social interaction and sensory sensitivities, while discounting and marginalizing others.
 [Slide 30a] One sort of evidence for this looping dynamic is the sort of things that began
to fall out of the diagnosis of autism, namely self-injurious behaviors. [Slide 30b] When
Rimland wrote his book in 1964, he said “I do not know any one of these children who did not
have the rocking and banging of head.” [Slide 30c] In the DSM-III, self-mutilation was a key
criterion for diagnosing pervasive developmental disorder. [Slide 30d] But in the DSM-III-R
criteria, written by Lorna Wing, self-injurious behaviors were merely an “associated feature.”
[Slide 30e] Finally, in the DSM-IV they are no longer mentioned and psychiatrist Bryna Siegel
can assert – in a book written for parents – that self-injurious behaviors are quite rare in autism.
[Slide 30f] What happened between 1964 and 1996? Siegel goes on to say that “it’s my clinical
impression that there is a lot less hand biting now than twenty years ago.” Why? Because self-
injurious behaviors came under the jurisdiction of behavior modification therapies in the 1970s.
Lovaas certainly made them a prime target. They came to be seen as treatable by behavioral
therapy and so were discounted as outside the core symptoms of autism.
One example of this looping dynamic is the sort of things that began to fall out of the
diagnosis of autism, namely self-injurious behaviors. In his book, Rimland (1964) said that he
did “not know any one of these children who did not have the rocking and banging of head.” In
the DSM-III (1980), self-mutilation was a key criterion for diagnosing pervasive developmental
disorder. But in the DSM-III-R (1987) criteria, written by Lorna Wing, self-injurious behaviors
were merely an “associated feature.” Finally, in the DSM-IV (1994) they are no longer
mentioned and psychiatrist Bryna Siegel (1996, 74) could assert – in a book written for parents –
that self-injurious behaviors are quite rare in autism. What happened between 1964 and 1996?
Siegel went on to say that “it’s my clinical impression that there is a lot less hand biting now
than twenty years ago.” Why? Because self-injurious behaviors came under the jurisdiction of
behavior modification therapies in the 1970s. Lovaas certainly made them a prime target. They
came to be seen as treatable by behavioral therapy and so were discounted as outside the core
symptoms of autism.

Another example is more fundamental and more dramatic, namely the fate of Kanner’s
“cardinal symptom” of autistic aloneness and utter lack of interest in people. “Being in your
child’s face” meant that aloneness was no longer delineated as a singular and essential quality.
The sharp lines that surrounded it faded into the background, and what was foregrounded instead
was a certain equivalency between the multiple ways – from aloofness to oddness – in which
communicative rapport could break down, thereby stretching what was understood to be the core
of the syndrome into a continuum of “impairment of social interaction.” Kanner’s original
concept was still reflected in DSM-III (1980). Autism could only be diagnosed if there was
“pervasive lack of responsiveness to other people.” In DSM-III-R (1987), however, the cardinal
symptom was replaced by a variable of “qualitative impairment in reciprocal social interaction”
of which the aloofness and flat affect noted by Kanner were a possible but no longer necessary
component. In DSM-IV-TR (2000), finally, the diagnostic criteria are so written that a child can
be diagnosed with autism if there is “failure to develop peer relationships appropriate to
developmental level” and “lack of spontaneous seeking to share enjoyment,” which to me sounds
pretty much as what colloquially we would call a “loner”.
The conceptualization in DSM-III-R was based on Wing and Gould’s (1979) classic
article. Wing was the Chair of the committee writing autism diagnostic criteria for DSM-III-R.
Wing and Gould argued that they were able to distinguish five different types of communication
difficulties in a population of children with developmental disabilities (i.e. the undifferentiated
product of deinstitutionalization): children who were aloof and indifferent in all situations (i.e.
Kanner’s original prototype); children who liked merely physical contact with others; children
who were indifferent only towards other children, but not towards adults; children who were
passive but amiable when approached; and children who made active social approaches but in
inappropriate ways. As wing (1981) explained later: “The justification for regarding them as
related is that all the conditions in which the triad of language and social impairments occurs,
whatever the level of severity, are accompanied by similar problems affecting social and
intellectual skills. Furthermore, individuals with the triad of symptoms all require the same kind
of structured, organized educational approach, although the aims and achievements of education
will vary from minimal self-care up to a university degree.” In other words, the justification to
consider all of them as merely points on a single dimension of variance was, once again,
pragmatic: they were all responsive to the same set of therapies and “non-specific impingement”,
and, most importantly, the intensive application of therapy could move a child from one type to
the other.

So over the long term, what happened between autistics and therapists began to redefine
what autism really was, both the boundaries and the core of the syndrome. It changed what the
clinicians knew they should be looking for. Then it was reported back to parents in books such as
Siegel’s, which meant that clinicians would indeed see in their offices more or less what they
were looking for. In this way, the movement started by Kanner completed a full 180° loop and
ended up with a substantial revision of how autism looks and feels, is understood, classified and
counted.
The final part of the story is about therapies as ontological politics (Mol 2002) and as an
infrastructure for autistic personhood. The ideas developed here are based on the ethnography
conducted by one of us (Hart) working with autistic children and their families in the US and
Morocco.
What do we mean by “ontological politics”? We mean that often what is at stake in
therapies is the very being of autism: what it is, or can become. An autistic behavior like “hand
flapping”, for example, can be interpreted as a meaningless symptom to be extinguished by a
therapist (remember the self-injurious behaviors?), or it can be interpreted as a form of
communication requiring interpretative effort. Perhaps the child has experienced an “emotional
overload”, and is expressing it in this way? What it means to have autism, and our image of the
autistic person, differ between the two scenarios. Now, as our reference to extinction by a
therapist was calculated to hint, this is not something that is decided merely by an effort of the
imagination or even by the language available to us to represent inner states and other minds.
There is a level of material practices – therapies and assistive devices - that play a crucial role in
these ontological politics because they literally seek to bridge between autistics and non-
autistics, to “impinge” upon the autistic person and turn his “aloneness” into communicative
rapport.
We are trying to extend the argument made by Ian Hacking (2009), who drew attention to
the role played by autistic self-advocates in these ontological politics. His point was that in their
memoirs and blogs, the self-advocates have elaborated a language of neuro-diversity with which
to represent a thick autistic person (that is, one with emotional and cognitive depth, whose
actions are meaningful and intentional if also hard to decipher). The point is not whether these
are really accurate descriptions of what it is like to be autistic; the point is that this sort of
language begins to constitute the experience of what it is to be autistic by telling stories that
connect words from ordinary language, used to describe emotions and mental states, with the
behavior of autistic children, thus making this behavior intelligible to parents, even of “low-
functioning” children who do not speak, to experts and conceivably also to autistics themselves.
It is not inconceivable to argue that this language has played a role in making the autism
diagnosis vastly preferable to mental retardation and thus unleashing the fourth loop of
diagnostic substitution.

If our attempt to connect the therapies with this argument seems farfetched, this is
because in most histories of autism up till now, and in the stories that the self-advocates tell
about themselves, the therapies, especially behavioral ones, and the parents who use them, are
depicted as on precisely the opposite side from the self-advocates in these ontological politics.
While the self-advocates argue that they are not ill and do not need to be cured (Ne’eman 2010;
Sinclair 2005; Bumiller 2008), they depict the parents and the therapists as “cure-bies”, i.e. as
unable to tolerate this difference and attempting to find a cure for it. Instead of recognizing
autism as neurodiversity, they are purportedly trying to “normalize” autistic children, i.e. to cure
them, recover them, camouflage their autism, as we heard from Schopler, or simply make them
conform to how neurotypicals behave. Parents, for their own part, especially the parents of
severely disabled children, resent the neuro-diversity talk as irrelevant to their plight. The
rhetoric on both sides is accusatory and uncompromising.
This is precisely where the “looping” argument is so valuable. Ian Hacking’s (2009)
argument about “autistic autobiographies” has the great merit that it can potentially make some
of the angry parents see the enormous contribution that the self-advocates have made to the way
they understand and represent their own children. But this argument is also clearly incomplete
and one-sided. It leaves the appearance of the self-advocates themselves completely unexplained,
while all the creativity and dynamism seem to flow from them to other parts of the autism world.
It also tend to reinforce the image of the therapies as “normalizing” and as the opposite of the
language of neuro-diversity elaborated by the self-advocates. (Hacking 2007; Bumiller 2008). In
what follows, we would like to balance the accounts, and to persuade you that the role that the
therapies play in these ontological politics is more complicated.

Our first point is that the self-advocates and the politics of neuro-diversity they represent,
could not have emerged without the previous looping spirals initiated by the parents’ movement
and the therapies. There is good reason for approaching this question cautiously: anybody who
reads their memoirs cannot fail to recognize the extent to which these are extraordinary
individuals who, against enormous obstacles, managed to compensate for their deficits through
ingenious techniques of the self: diaries, alternative personalities, and highly elaborate cognitive
devices for processing experience and endowing it with meaning (Williams 1992, pp. 18–20, 56;
Grandin 1995, pp. 18–19, 95–97). Nonetheless, even if the self-advocates themselves remain sui
generis, we would argue that the slot into which they stepped was prepared for them by the
processes described thus far. Perhaps the most obvious point is that if autism had remained a
relatively rare disorder and had not been transformed by the therapies into a spectrum
accommodating different levels of severity, it is highly unlikely that the self-advocates would
have been recognized as autistic, and their testimony would not have been taken as relevant. For
significant portions of their lives, indeed, they were considered “neurotic,” “psychotic,”
“retarded,” “partially deaf,” or just plain “weird,” but not autistic (Williams 1992, pp. 29, 64;
Shore 2003, p. 38). Parents and parent organizations played a crucial role in building this slot for
the self-advocates and in certifying them as credible public representatives of autism. In the early
1970s, the British and American parents’ associations collaborated in a project on the “near
normal autistic adolescent.” The project was coordinated by two mothers who wrote an article
suggesting that this group “would benefit…from recognition of their condition as a separate
category of autism” (Dewey and Everard 1974, 355). They also translated a lecture given by
Hans Asperger, thereby leading directly to Lorna Wing’s (1981) seminal article suggesting that
Asperger’s disorder should be included within the autism spectrum. Rimland wrote the preface
for Williams’s book (1992, pp. ix–xii), essentially certifying that this was indeed a report about
autism and not something else. Because Williams herself only learned about autism in 1989,
three years before the book was written, Rimland’s authority was necessary in order to certify
that this is a story about autism and not something else. He also wrote the blurb on the back
cover of Grandin’s book, and served as an adviser to the Rain Man production. Dustin Hoffman
worked with Mark, Rimland’s son, to prepare for his role as “autistic savant”.
Some of the language of the self-advocates, moreover, has unmistakable affinities with
Sensory Integration Therapy. Grandin (1995, pp. 42–48), for example, says that the autism
spectrum is a continuum of “sensory processing disorder” and makes numerous references to
sensory integration, as does Shore (2003, pp. 39–51). This affinity is not accidental. We saw
earlier how sensory integration therapy requires the active participation of the patient in their
own treatment, precisely because like Grandin or Williams, the child knows more about the
nature of his or her problem than the therapist. Autism therapies constitute the patient as agent,
as active participant in the therapeutic exchange. This is our and second and more fundamental
point: autism therapies play a role that is similar in principle to the one played by the self-
advocates. If the self-advocates elaborated a language with which to represent a thick autistic
person, the therapies provided the grammar of this language, or more precisely the infrastructure
for supporting autistic personhood. What do we mean by this?
Autism is depicted by the self-advocates as a problem of translation. The severely
autistic individual has a rich inner world, but lacks the means to participate in the publicly shared
and sanctioned forms of expression that constitute intersubjectivity and therefore personhood, the
capacity to participate in public life. Who would translate? The self-advocates offer themselves,
but how were they able to extract themselves from the autistic idiolect? As Hacking (2009, 56)
notes, “neurotypicals and severely autistic people do not initially share a form of life because the
bedrock is lacking, and so an artificial platform must be constructed”. This artificial platform, we
would argue, was at least in part provided by the therapies.
Understanding autism as a problem of translation, in fact, predated the self-advocates.
Speaking to the American parents’ association in 1973 about the views of her predecessors,
Hermelin and O’Connor, Lorna Wing (1973, 111) said as much: “they came to the conclusion
that autistic, meaning ‘socially withdrawn,’ is a completely inappropriate label to apply to these
children. They believe (as I do) that autistic children have all the normal emotions appropriate
for their mental age, but are severely handicapped in showing them” And she added, “many
parents who have learned the special language of their own autistic child are of the same
opinion.” In this she indicated what should be fairly obvious, namely that the first to shoulder the
task of translation, especially in the wake of deinstitutionalization, were the parents. In this,
moreover, they were assisted by autism therapies and their prosaic technologies to create the
necessary “artificial platform”.

We will give two examples. The first is a staple technique of ABA – prompting. ] To
teach a new skill, therapists or parents will usually begin with “hand-over-hand” physical
prompting, where the therapist’s hands initiate the action by the child, and continue to provide
cues about how to proceed. Over time, as the child becomes more skilled, the prompts are “faded
out.” In reality, however, it is often impossible to completely fade out the prompts. The child
learns some elements of the task or skill, but is never completely independent. The child learns,
however, also how to interact with and perform everyday tasks in concert with teachers or
parents. At the same time, the parent or teacher also learn, through prompting, how to respond to
the subtle cues given by the child. They learn what sensory stimuli are likely to attract him or to
ignite a tantrum; they learn the meaning of “hand flapping” in its various manifestations; they
learn that an averted gaze does not mean inattention; in short, precisely by virtue of the
entwinement of their actions and their bodies, they gain a sense of the child as an autonomous
person. The result is what we call “joint embodiment”, a repertoire of joint behaviors,
expressions, prompts and counter-prompts, scripted dialogues, through which autistics and their
companions can participate together in public life. It is easy to criticize the “robotic-like”
performances that ensue, and to ignore the fact that they also mean that the public sphere is
modified, that it becomes an arena where personhood may be constituted through hybrid
arrangements and non-conventional forms of sociality. If it is “normalization”, than not in the
quasi-Foucauldian sense of discipline and conformity, but in the sense that Wolfensberger (1972)
gave to it, namely arranging the environment so that the disabled could obtain “an existence as
close to the normal as possible”. The controversy around the procedures and claims of
“facilitated communication” demonstrates that this joint embodiment represents such a radical
challenge.

Or consider the very humble technology of the PECs, developed in the 1980s by
American speech therapists and ABA experts. It is extremely low-tech, yet it is perhaps the most
significant technology in the history of autism therapies. It is literally the “artificial platform” on
which the problem of translation is negotiated and routinely solved. There is no universal or
standard PECs. Each PECs has to be adapted to the specific child involved because just as the
child is being taught to exchange the pictograms in order to make requests, so the parent or the
therapist must learn the specific visual idiom that suits the child. The resulting system is sui
generis. By now, PECs have been joined by more high-tech devices like computers and ipads,
yet the essential point remains: just as the therapies and their associated devices and prosaic
technologies work toward the adaptive goal of helping individuals to “pass” as normal or to
“camouflage” their autism, at the same time they also reconfigure what counts as normal and
how someone can pass as such. When autistic people make requests in restaurants by exchanging
pictures for items, typing out words on small computers, tapping out requests and responses on
ipads or letter boards, or having their garbled speech prompted and translated by a parent or
caregiver, they engage in “ontological politics”. Bereft of joint embodiment, of assistive devices
or the “sensory-friendly” environments encouraged by SIT, their demand for personhood would
quickly become the exclusive province of only a select few. Just as autistic self-advocates have
created a language that permits to represent an autistic self, so therapists and parents have created
a repertoire of joint public behaviors, scripts, routines and prosaic technologies that enable
autistic personhood.