Monday, October 6, 2014

Is Autism a Failure of Prediction?




This afternoon a group of MIT and Mass General researchers released a study called “Autism as a disorder of prediction.” In this paper, they argue that autistic people “experience things they don’t understand,” because our predictive ability is impaired.


Interesting as this sounds, a close reading reveals the premise as totally at odds with my lived experience.  I think of myself as a friend to those engaged in autism research, and I hate to come out in criticism of a newly released piece of work, but in this case I feel their conclusion are just wrong.

Anyone who has observed the prowess of a young Asperger video gamer would realize what a fool he'd be to bet against a kid like that's predictive ability.  But that's not all.  The hypothesis of this study does not hold up any better in my "real world" experience.

I do have some social disability, even now. My problem is that I cannot read the unspoken cues from people around me.  My ability to evaluate what I do know – and to predict from it – is not weak at all.  In fact, as a logical thinker it might even be stronger than average.  In other words, I have a weakness in data input, when it comes to human-on-human engagements.  Too much of the wrong data and not enough of the right data equals trouble, even with the best predictor in the world.

No wonder we stim and compensate.  They nailed that. 

Otherwise, with all due respect, this paper seems to be a perfect example of what happens with autistic behavior is interpreted by neurotypicals, as opposed to having the behavior explained by those who live it.  The study’s authors spend many pages expounding on an explanation of behaviors such as I describe in my own books and essays (Look Me in the Eye is one of their citations) when a conversation with an intelligent, insightful autistic adult could have set the whole thing straight.

Allow me to offer a comparison to put this in perspective.  Imagine that an alien social scientist observes a human population and notes some puzzling and different behaviors.  Some of the people eat some meat, but no pork.  They call themselves Jews.  Some of the people don’t eat meat at all, and they call themselves Vegetarians.  Some of the people don’t eat bread, and they call themselves Gluten Intolerant.  Some Gluten Intolerants eat meat, and some don’t.  Then there are the ones who call themselves Catholics, and their strange seasonal aversion to meats.  They refer to the aversion as Lent, and researchers scratch their heads to determine who’s the borrower and exactly what was loaned, to render Catholics unable to eat like the others.

Why the different behaviors?  After careful observation, the researchers concluded that the Gluten Intolerants had the answer.  They admitted to a biological deficiency; they cannot eat the foods others consume with gusto.  Researchers hypothesized that Vegetarians and Jews were similarly affected but their food limitations were subtler.  They even suggested some meats may be toxic. After reading about empathy, one researcher concluded the Catholics lent some un-discovered digestive process to their fellow men for a period, so they too could be healthy.  Their paper describing these discoveries was published to wide acclaim on Alpha Centari, but the humans mocked its conclusions when they read it back on Earth. The humans in the Alpha Centari zoo just snarled. The researchers wondered why.

An old Gluten Intolerant offered them a piece of wisdom.  “Did you ever ask one of those Jews or Vegetarians about eating meat?  I’ll bet they could give you the answer, and it isn’t what you think.  It may be a mystery to you, but it’s no secret at all to them.”

The thing is, as aliens, they had absolutely no concept of religion.  And the only thing they could conceive for Vegetarianism was the general concept of disability. The idea of a life choice was too strange to consider.

To an Autistic like me, this news is much the same. What it shows most of all is not insight, but the obliviousness of the researchers. I do not have a disorder of predictive ability. I've met many other autistics and I can't think of a one with predictive disability. These researchers cited a line from my book as support for their hypothesis, when in fact the whole book expanded on my thoughts at considerable length, and made amply clear why I have trouble in social settings, and it's not poor prediction capability.  How about you?

Having said that, I concede that there may be differences in how I predict things as compared to how neurotypicals predict.  But this study does not answer that possibility, nor does it present any new evidence for what a difference might be and how it might happen.  The autistic narratives the researchers cite don't distinguish input problems from processing problems in most cases.  In any case, their interpretation takes those writings quite far from the context in which they were intended. 

As an autistic person I don’t perceive the same things as neurotypicals.  I make my decisions based on different incoming data.  It stands to reason that my predictions will be different because the inputs to my predictor are not the same.

What’s the takeaway here?  Bring the members of a community you want to study into your process at the beginning.  Be guided by their knowledge, culture, and wisdom. Don’t let ignorance of another culture lead you down a wrong path.  It’s wasteful at best, and can make you look like a fool.


As the neurodiversity activists say, nothing about us, without us.

John Elder Robison is an autistic adult and advocate for people with neurological differences.  He's the author of Look Me in the Eye, Be Different, Raising Cubby, and the forthcoming Switched On. He serves on the Interagency Autism Coordinating Committee of the US Dept of Health and Human Services and many other autism-related boards. He's co-founder of the TCS Auto Program (A school for teens with developmental challenges) and he’s Neurodiversity Scholar in Residence at the College of William & Mary.  The opinions expressed here are his own.

21 comments:

Florencia Ardon said...

Great analogy. By the way, I think NTs have the same issue 'reading' cues from an autistic person. I cannot tell when my daughter (6) is paying attention or how much she's understanding, I just know it's way more than I think. Her teachers and therapists have the same problem. They don't believe my daughter is paying attention, but cannot explain how it is she ends up doing the school work correctly. I suspect she can either need to look only briefly at something and then she's able to remember it, or that she can take it all from the corner of her eye... or something ;-) At any rate it is that she's paying more attention than we think, because she's no magician!

jonathan said...

I wonder if you actually read the study in its entirety and not just the abstract. You can read the entire study since its open source and not behind a paywall. Always best to read primary sources when you can if you blog about it or state in the blog post you did not read the primary source for whatever you reason you didn't or could not read it.

jonathan said...

actually the reason I asked is that the gist of your beefs with the study is that they did not consult actual autistic people, yet they actually quoted from the experiences of four autistic people to back up their theoretical perspectives, Deborah Lipsky, Dora Raymaker, Temple Grandin and Jon Evans. It's true the ratio of females was 3:1 and the actual ratio amongst high functioning persons is 10:1, so I'm wondering if you actually read the primary source and not just the abstract. Again, it's always best to read the primary source when writing a blog post.

John Elder Robison said...

Yes, Jonathan, I did read the whole study. That's how I came to note my own writings in the citations at the end. You are right, they do quote autistics, but they do so from our writing which in this case may be out of context. I think they should have actually spoken to autistics in person for this.

1LINER said...

John this is gorgeous and spot on. I agree that anything produced by neurotypicals and their representation of autism, regardless of function, is and should be highly suspect. Who or what funded the research, Autism MISspeaks (ha)-just curious. The abstract made me sick to my stomach and accompanies a long line of research that does same. And it took decades no less. Just gross. Joanne Secky
I am undiagnosed Aspergers and my son is diagnosed Aspergers, prefers autistic. He just decided yesterday MIT isn't in the mix for him - too much group work! Great irony.

intellectualizing.net said...

I think the paper is sort of circling around a useful insight, but it's really the same "intellect vs. instinct" point that Hans Asperger made 70 years ago. And I bet it only applies to some "flavors" of autism.

I buy Lynn Waterhouse's argument that trying to create a "unified theory of autism" is harmful to the research agenda.

More elaboration at
http://intellectualizing.net/2014/10/07/autism-and-prediction/

Sara Gardner said...

I think you're spot on, John. One look at the title of the study had me thinking, "did they talk to any autistics to design and carry out this study?" Sadly, the answer seems to be "no" as is all to often the case. What is true is that "they" do not understand "us" anymore than "we" understand them, and perhaps less so.

Tammi said...

People that don't eat meat know why they don't eat meat. What do I know about why I am the way I am, etc. I don't want to debate it. Just know, you lost me at the analogy.

Rochelle said...

I am a newly diagnosed "Aspie" as of 5/2013, and I also happen to be a pediatrician (yeah, way harder in every way despite colleagues that should "know" about it)...I caught this article title in a pediatric daily update email from the AAP (amer. acad. peds). I emailed the author and I read Mr. Robison's comments above. I think that it is difficult to 'predict" in that the way I talk to some people (parents, especially), I can't predict how my message will be taken. I admit I did not read the whole MIT article, but I wonder, your points (JER) are good, too, but there is a semblance of "prediction" issues by failing to read people correctly. Primary prediction issues or difficulty predicting due to getting false cues or failing to read cues...Isn't this just an issue of semantics? I certainly get upset by some colleagues who just don't get it, especially when they have to "deal" with a colleague with "aspie" and are less than friendly or compassionate about it. I HATE the parent of autistics extremism of no vaccines, weird diets, and "cures." That makes me very angry!! But this article's summary, it doesn't seem like a big deal in the overall scheme of things.
Oh, and btw, to shamelessly promote my own book, geared towards kids to adults, those with and those caring for those with ASD, I also include a section of ASD from my unique perspective (giving an "insider's perspective)... Floppy Lop-Ears Tries to Get "Off the Spectrum."

John Elder Robison said...

Thanks for your comments Rochelle. A weakness in prediction because your sensory ability is limited is not a prediction failure. It's a sensory limitation.

Those are separately testable things that were confused by these researchers, I think

Florencia Ardon said...

In regards to Rochelle's comment... I agree that sometimes it sounds like people are arguing about semantics. But if researchers are going to use this paper as basis to look for the neurological mechanism, it is important to know exactly what part of the chain is not "typical." It can make a huge difference in what protein/gene one is looking for. And I agree with John that they should have spoken directly to autistic people, not only have gotten information from books, which are written differently than scientific articles and are easier to be taken out of context.

1LINER said...

Rochelle - I like what you wrote and I think your point valid. If someone "does" something, whatever it may be a small percentage of the time, sure it "may" be validated through research as most things can be - and taking select excerpts from autistic authors a plus for the researcher. (I have an analogy in mind, but I think more than a little off color!) However - it should not take over what is more prevalent, and that is why autism research is so frustrating (for me). The article presents as a negative (you made it more positive/viable) but where's the research on our insight, our enhanced empathy, our contributions despite sometimes just "missing the neurotypical obvious" - which has been dissected? Rhetorical. I like the research that considers the upside, what is profound about autism (of course I do) and it doesn't seem to prevail. I fundamentally find the title off putting and just one more example of making something positive negative.
I hope Floppy Lop decides she likes where she is on the spectrum.

Daniel Butterfield said...

Like Jon E.R., I don't find this passage compelling, when they proffer the interpretations of the Heider and Simmel films of geometric shapes: "Having access to a probabilistic conditional relationship between observed movement patterns and potential histories, neurotypical individuals are able to ascribe causes to the former. The fact that these causes happen to be social in nature is likely happenstance; the social account is simply a better predictor of the observations than other historical accounts. An autistic individual, not having access to such predictive relationships, is constrained to interpret behaviors without any motivating history, social or otherwise....In this way, the PIA hypothesis provides a simple account of the difficulties autistic individual experience with theory of mind and other abstractions, which require the use of learned interevent relationships to read more into an observation than the basic sensory signal offers."

I don't think it's correct (or will be shown to be correct, given that this is a hypothesis) to posit that the social factor is "happenstance". To see what they were talking about, I watched the Heider-Simmel movie on YouTube. (Scary stuff!) Detail- and object-orientation--yes, absolutely makes sense that these would influence a greater percentage of spectrumites to not ascribe a social story than that of neurotypicals. Impaired predictive abilities? Irrelevant to this example.

On top of that, their account of Theory of Mind (which to me is a misguided relic of old-school black-box-psychology) as "inherently a prediction task" seems like a reach. Is seeing things from someone else's point of view a matter of being able to go from antecedent state to subsequent state?

There's more about this hypothesis that I disagree with (basically, how it's constructed--they've got an abstraction that they've fit to these selected attributes the way Erich von Daniken fit his story to his evidence), but I'd prefer to sit down and extract that from this paper before expounding.

But at this point, my gut leans towards Rochelle's comment, "Primary prediction issues or difficulty predicting due to getting false cues or failing to read cues...Isn't this just an issue of semantics?", and Florencia Ardon's, "But if researchers are going to use this paper as basis to look for the neurological mechanism, it is important to know exactly what part of the chain is not 'typical.'" I have a pet "hypothesis" that neurotypicals don't actually "have" a "Theory of Mind" that those with Asperger's "lack". Being empathic and able to accurately assess the past and making educated guesses about the future are things you learn from experience, and some are better than others. (Took me a while to even want to learn these, I'll admit!) And I'm all for helping people on the spectrum with these challenges.

It's easier to put yourself in the shoes of someone who's like you. If Theory of Mind was the norm, men would know how women think and vice versa, rich people would know how poor people think and vice versa, and cultures would know how other cultures think.

Daniel Butterfield said...

(continued) I'm probably on the spectrum myself, but I would fall below the level to be diagnosed because it's pretty "mild" and I do fine. When I'm with people on the spectrum, I don't have a problem knowing social protocol and "predicting" the outcome of my statements and how others will respond to them (I don't actually have trouble with this with NTs, but if it turned out I really do have trouble with this I wouldn't notice the difference). I know when I'm with my kind, and the social norms are pretty easy to observe.

Any predictive impairment I have in social situations with NTs is because I'm different from a lot of NTs unless we have some common interest. Let me get to know you, and I'll get better at "predicting" you. And I had to teach myself to do the standard, generic, impersonal conversations that many NTs prefer to silence and to consider these conversations interpersonally meaningful rather than personally irritating.

Though I'll admit this... That the subway would be out of operation on a Saturday morning in Toronto--in hindsight, I should have been able to predict that. They probably wait until the work week is over to do maintenance. Gonna remember that one.

In general, if I were looking into what's different about those on the spectrum, I wouldn't look for what it is that those on the spectrum "lack" compared to the normative NT. That's kind of like studying gender communication dynamics as a matter of women "lacking" the skills to communicate as the "normal" males do.

Dr Beth Weiner said...

What are your thoughts about the elimination of Asperger's from the DSM 5? As a psychologist, I thought it was useful for my patients to have this diagnosis.

Rebecca Miriam said...

Jonathan, please take some time to examine why you felt so sure that one of the people they cited was mistaken, that you must know better than him and that it was appropriate to scold him based on what could have been (turned out to be) your own reading error.

rob j said...

The new ASD dx is essentially putting an NOS on the end to clump a group of 'sx' with no regard for etiology.
Aspergers appears to be quiet hereditary (my dad, his brother, myself, probably my brother, my brothers 1st child... These people include only one without a degree (by age) and those degrees include a phd from caltec (Fienman gave the cargo cult science lecture at my dad's commencement) I have seen arrivals arguing that having a parent who is a college proffesor should be a rule out for Aspergers, funny, between John Robison, and Tim Page, all hopes I might have had a chance of publishing an autobiography about growing up in New England as the undiagmosed aspergian son of a proffesor. (Hell, Tim even mentions the same streets I have stories

John Elder Robison said...

Rebecca, I have had those thoughts since casting this story out into the world. I think the essence of my objection is that the researchers chose to interpret selected passes of autistic writing when the actual autistics were available for conversation.

To me that is a major failing. I stand by pointing that out.

Moving beyond that, I feel bad if you or the scientists felt I was scolding them because I am a believer in science and a believer in that school. I just took exception with how they did this and I think a flawed methodology led to flawed premises.

The researchers may well feel battered by the backlash of this story. For that I am sorry because that was never my intent.

John Elder Robison said...

Rebecca, I have had those thoughts since casting this story out into the world. I think the essence of my objection is that the researchers chose to interpret selected passes of autistic writing when the actual autistics were available for conversation.

To me that is a major failing. I stand by pointing that out.

Moving beyond that, I feel bad if you or the scientists felt I was scolding them because I am a believer in science and a believer in that school. I just took exception with how they did this and I think a flawed methodology led to flawed premises.

The researchers may well feel battered by the backlash of this story. For that I am sorry because that was never my intent.

Jennifer Wamaling said...

This is a bit off topic, but what do you suggest for NT parents trying to better understand the world for their non-verbal child on the spectrum? I read your book, "Look Me in the Eye" and while I loved it, the phrase if you've met one person with autism, you've met one person with autism rang very true. Do you have suggestions of people to follow online or books to read? I want to know more about how to try and understand my child's world. He is a very lovey, fun-loving guy...we just can't communicate very well, yet.

Ralph Doncaster said...

I completely agree with your statement, "this paper seems to be a perfect example of what happens with autistic behavior is interpreted by neurotypicals".
Many NT's studying ASD seem to treat people on the spectrum more like specimens than human beings; and they say we lack empathy!

However I'm not as dismissive as you about predictive abilities in ASD. In fact my first post in my Aspie Brain blog was about brain prediction circuits.

I think an important distinction needs to be made between intuitive and cognitive abilities. For example you might cognitively understand gravity and wind resistance, but if you can't intuitively predict the motion of a baseball, you're not going to make a hit.

Where I do agree with you is that the problem can be with the data or the program. For example if social problems are due to a lack of instinctive knowledge rather than intuitive predictive ability, then reasonably accurate understanding and prediction of social behavior can be learned.