, ,

I am an autistic parent, which means I have a relatively high chance of having an autistic child, which means I’ve started thinking about early-intervention treatments for autism. (After all, autism is often diagnosed as early as eighteen months, and I’m not exactly going to have time to do research when I have an infant.)

Many of my friends disapprove of applied behavioral analysis (ABA), one of the most commonly used treatments for autism. I am inclined to be more sympathetic, I think, for several reasons. First, it claims to be the most evidence-based treatment for autism. (As far as I can tell, this is true but complicated; more on that later.) If a treatment is evidence-based I’m always going to take a second look. Second, many of the groups that advocate against ABA support discredited pseudoscience such as facilitated communication, which makes me not trust their judgment very much. Third, I have a natural tendency not to want to admit beliefs that will make my friends mad at me, and “I think this therapy you think is abusive is actually fine” might make people mad at me.

That said, setting those worries aside, I have some concerns about ABA. I welcome opinions from people with more knowledge about ABA.

The evidence is not that great. It’s true that ABA is the most evidence-based autism treatment. As far as I can tell, it gets this status because most autism treatments fall into the category “untested but plausible,” with occasional excursions into the land of “what, no, there is literally no reason to think that would work, the fuck is wrong with you.” (This is a somewhat puzzling state of affairs, because we do spend lots of money funding autism research, and we can’t spend that much money trying to give fruit flies autism.)

This is a fairly representative list of studies of ABA as an autism treatment. Note that no studies follow the children until adulthood. Most have sample sizes of twenty to fifty children. And that’s not even including the really embarrassing stuff: the study of the Early Start Denver Model used parent report for many of its outcome measures even though the parents knew which group their children were in.

I understand that an appropriately blinded randomized controlled trial with a reasonable sample size that follows children to adulthood is really expensive. However, if you’re going to talk about how your treatment is evidence-based, this is the sort of thing that is necessary. The evidence for ABA at this point is less “gold standard” and more “plausible in early, exploratory trials.”

So I feel justified in having other qualms.

Response to adverse event reports. Many autistic adults report that they have PTSD and depression from their experience of ABA. That doesn’t mean that ABA necessarily causes PTSD and depression– perhaps it’s a coincidence, or perhaps they had unusually bad therapists. But I think if lots of people are saying “your therapy gave me depression,” the only ethical thing to do is go “wait, holy shit, we need to study this.”

In my anecdotal experience, I have not seen ABA supporters say this. Instead, I have seen many ABA supporters say that people reporting adverse mental health consequences are pseudoscientists who hate evidence-based medicine, which is not the way you respond to people saying your therapy gave them PTSD.

I have seen some ABA supporters argue that perhaps it causes PTSD and depression among the “high-functioning,” which seems like a really dumb argument. First of all, the whole point of your therapy is that it turns autistic people with high support needs into autistic people with lower support needs; maybe those people have low support needs because your therapy worked. Second, a lot of people with very high support needs can’t communicate in sign, speech, written language, etc., so of course they’re not going to be saying “ABA is terrible and I got depression from it,” even if it gave literally 100% of them depression.

Normalization. I don’t mean to say that faking nonautistic is not a useful skill. It opens more options to autistic people: while many people (including myself) find that faking nonautistic is not worth the cost, I’m not going to impose that on everyone. Many people want to work jobs or have relationships that require them to fake nonautistic, and it’s good to give them that option.

However, in my opinion, it is also not a skill it makes sense to teach people who presently can’t function very well. If a person does not have use of language (whether spoken, written, sign, or through use of AAC), has a severely limited diet that may cause them nutritional deficiencies, regularly experiences meltdowns, self-injures or harms others, or experiences one of the many other severe impairments that can be caused by autism, those need to be the #1 priority. For that matter, if a person is anxious or depressed, has low self-esteem, has executive function issues that mean they can’t meet their goals, or can’t understand nonautistic behavior, then those issues need to be the priority. Only once all of those have been sorted out does it make sense to concentrate on eye contact.

And yet in many ABA programs eye contact is one of the first things worked on, even in autistic children with severe difficulties functioning. It seems to me that this is less about giving autistic people options and more about saying that autistic ways of being are inherently worse than nonautistic ways of being. That is not a therapy I can get behind.

“Effective” is one of those words that depends on your values. Effective for what? Regularly beating your children is a very effective way of making them so scared of you that they instantly and quietly obey; I don’t want to beat my kids in part because I think that’s a terrible fucking goal. And turning autistic children into facsimiles of nonautistic children is also a terrible fucking goal.

Aversives. Fortunately, the use of aversives has been phased out in ABA treatment. In addition to the physical abuse of children (I think it should be fairly obvious why this is objectionable), aversives sometimes included things like taste aversives, which don’t seem that bad to nonautistics. However, as an autistic person, being forced to eat something I have a severe taste aversion to is literal torture. It is wrong to do that to a child.

However, Lovaas’s original randomized controlled trial of ABA for autism did include the use of aversives. How do you know that aversives weren’t the active ingredient? Therefore, this change (which is very positive) makes the evidence base for ABA even more limited.

Prompt dependence. As far as I know, prompt dependence has not been studied in autistic people, and any claims about it should be taken with appropriate grains of salt. But, anecdotally, it is a very common experience among autistic people to find yourself doing things because other people or the environment prompts you to, without actually intending to do it. Sometimes prompt dependence is helpful (I personally use it to get work done). Sometimes it is very unhelpful, as when you find yourself doing something that you don’t want to do or that is even harmful to you.

I am concerned that many forms of ABA, by rewarding specific behaviors in response to prompts, would increase prompt dependence in autistic children. It would certainly not teach the essential life skill of noticing that you’re doing something because you’re being prompted to do so and being able to do something else.

Amount of time spent in therapy. ABA often involves twenty to forty hours a week of therapy. This seems to me to be an excessive workload for a toddler, particularly when you consider that therapy is stressful and requires a lot of energy, so the remaining hours are unlikely to be high-quality hours. When does the child get to play? I don’t just mean this as a “play is fun and it is mean to deprive developmentally disabled children of the opportunity to play.” The current scientific consensus is that play improves social skills and executive function. (See, for instance, this report from the American Academy of Pediatrics.) Those are exactly the things autistic children are impaired in. Without strong evidence of efficacy, it seems ill-advised to give children so much therapy that their play time is limited.

Behaviorism. Unlike many autism advocates, I do think that there’s an appropriate role for behaviorism as one component of therapy for mental disorders. However, pure behaviorism has been rejected in most areas of psychology. Patients with mood disorders and personality disorders receive therapy with both a cognitive and a behavioral element (and often with other elements, such as mindfulness). Experts no longer advise not comforting crying infants for fear that it would incentivize the child to cry; indeed, comforting crying infants is often recommended as a way to build attachment. Many people are concerned that extrinsic rewards of the sort promoted by behaviorists may decrease intrinsic motivation and ultimately lead to lower performance. We understand the importance of social learning and attachment in children’s psychology.

And yet here we are treating autism with pure Skinnerian behaviorism. I have no doubt that pure Skinnerian behaviorism is effective for some things– it works quite well for phobias– but I would be really really surprised if it were the correct treatment for a complex condition like autism. Teaching children to say “hi, what’s your name?” in response to someone saying “hi” is not actually teaching any useful social cognition. There is, in fact, a difference between people and chatbots.

I don’t know what my ideal autism early-intervention program would look like. Certainly it would be customized to the child. Maybe there’d be a big play component. Maybe there’d be work on building attachment between the child and caregivers. Maybe there’d be age-appropriate cognitive therapy (despite the difficulties in providing cognitive therapy to children without language). Maybe children could play at a play group with both autistic children and neurotypical children who have been taught how to play with autistic children, so they could build social skills in an easier environment. (The neurotypical children seem like they ought to be easier to teach than the autistic children, anyway, since they’re the ones without social impairments.) Maybe there would be a lot of occupational therapy to help with motor skills, feeding, speech, sensory sensitivities, and other common areas of impairment. Maybe children would be taught to identify their needs and self-advocate. It is hard to know without more study. But I feel like pure behaviorism is not it.