I had my eye on Jack. He was ten years old, and he had taken only three skiing lessons in his life. I was in high school, and I’d been taking skiing lessons for three years. Yet I would watch Jack pass me on the slope, and I would see him execute these gorgeous stem christie turns, and, man, he could handle the four-foot ski jump with no problem. Meanwhile, I was still working my way up to one good christie, and every single time I tried the ski jump, I fell, until I was scared to use it.
What was so special about Jack?
Nothing, it turns out. What was so special, instead, was me — me and my autism. The connection between my autism and my poor athletic performance is pretty obvious in retrospect. At the time, though, I didn’t see it. Not until I was in my forties and I had the brain scan showing that my cerebellum — the part of the brain that helps control motor coordination — is 20 percent smaller than normal did I put two and two together. Now it all made sense! I couldn’t keep my skis together without falling because —
Because what? Because I’m autistic? Or because I have a small cerebellum?
Both answers are correct. Which, however, is more useful? That depends on what you want to know. If you’re looking for a label, something that will help you understand who I am in a general sense, then “because I’m autistic” is probably good enough. But if you’re looking for how I got that way specifically — if you’re looking for the biological source of the symptom — then the better answer is definitely “because I have a small cerebellum.”
The difference is important. It’s the difference between a diagnosis and a cause.
My research on subtypes of sensory problems got me thinking about the limitations of labels. I realized that two different labels — underresponsive to sensory input, and overresponsive to sensory input — can describe the same experience: too much information! The labels might be useful, but, as in the skiing example, their usefulness depends on what you want to know. Do you want to know what the behavior looks like from the outside? Or do you want to know what the experience feels like from the inside? Do you want a description for a set of symptoms — a diagnosis? Or do you want a source for a particular symptom — a cause?
Parents come up to me all the time and say things like, “First my kid was diagnosed with high-end autism. Then he was diagnosed with ADHD. Then he was diagnosed with Asperger’s. What is he?” I understand their frustration. They’re at the mercy of a medical system that’s full of label-locked thinkers. But the parents are part of that system, too. They’ll ask me, “What’s the single most important thing to do for an autistic kid?” Or “What do I do about a kid who misbehaves?” What does that even mean?
I call this kind of thinking label-locked because people get so invested in what the word for the thing is that they no longer see the thing itself. I’ve encountered the same kind of label-locked thinking elsewhere as well. A livestock handler might say to me, “My horse is wild. What should I do?” Or someone who has read my books on animal behavior will say to me, “My dog’s crazy. What should I do?” Well, first you have to tell me what wild or crazy even means in each case. I don’t have a clue unless you give me one. Does the dog try to bite the hands of strangers? Or does he jump on people because he’s really happy?
What I say in all these cases is the same: Don’t worry about the label. Tell me what the problem is. Let’s talk about the specific symptoms.
First question I ask parents who want me to advise them is “How old is the kid?” What I might recommend for a three year old is going to be completely different from what I might recommend for a sixteen year old.
Next question is “Does the kid talk?” If he’s nonverbal, that’s one thing. Let’s start trying to teach him and see what happens. If he’s verbal, I’ll say, “How good is his speech?” If the description is too vague, I’ll say, “Give me an example.” I want to know if the child is speaking in complete and grammatically correct sentences. Does he speak only in single words? Does he pronounce words accurately or does he say, as I did, buh for ball?
Can the kid hold a conversation? Can he place an order at a fast-food counter? If not, then the first thing you want to do is teach the kid social skills, starting with taking turns and saying “Please” and “Thank you.”
Does she have trouble making friends? Is she in school? Does she have a favorite subject?
The questions can go on and on, of course, just as they can for anybody — autistic or not. We’re all individuals. We all have a range of skills, habits, preferences, limitations. What would a totally normal brain even be like? A brain that is average in every way, that has the average number of neural connections, the average size of amygdalae and cerebellum, the average length of corpus callosum?
It would probably be pretty boring.
The differences are what makes us individuals — the departures from the norm, the variations in the brain. Take the corpus callosum, which is the collection of neural cables that stretch the length ￼￼￼￼￼￼￼￼￼of the brain and connect the left and right hemispheres. I have more of those cables than normal, but obviously someone can have even more than I do, or fewer than I do, or the normal amount, or fewer than normal. And my brain’s language circuit branches more than a normal brain’s, but, again, the extent to which language circuits branch exists on a continuum. The cerebellum size that probably affects my skiing — another continuum. The number of de novo copy number variations in someone’s DNA? The particular position of those CNVs on the chromosome? Continuum and continuum. I have often thought that eventually we’re going to be asking ourselves at what point this or that autism-related genetic variation is just a normal variation. Everything in the brain, everything in genetics — they’re all one big continuum.
The addition of Asperger’s to the DSM-IV in 1994 validated the idea of an autistic spectrum, but the meaning of “on the spectrum” itself has changed over the years. “In scientific circles,” a 2011 article in Nature reported, “many accept that certain autistic traits — social difficulties, narrow interests, problems with communication — form a continuum across the general population with autism at one extreme.”
In other words, you don’t have to have an autism spectrum disorder diagnosis to be “on the spectrum.”
This notion was popularized by the psychologist Simon Baron-Cohen. In 2001 he and his colleagues at the Autism Research Centre in Cambridge, England, introduced the autism-spectrum quotient questionnaire. People often take the AQ test online just to see whether they fall on the autistic spectrum. They might be wondering if they have Asperger’s or high-functioning autism. Or they might want to see what traits they have that, if amplified, would qualify them for one of those labels.
If nothing else, the AQ test got a lot of people thinking about behavior in a new way — the behavior of autistics, certainly, but the be￼￼￼￼￼￼￼￼￼havior of nonautistics, too. Their own behavior. The behavior of a neighbor, or a coworker, or oddball Uncle Ned with his disturbingly thorough stamp collection. Behavior that previously had seemed peculiar or perhaps aggressively strange now made a kind of sense.
The test consists of fifty statements. For each statement, you choose from four responses, ranging from “definitely agree” to “definitely disagree.” Definitely agreeing with the statement “I would rather go to a library than a party” might indicate that a person has an autistic bent. Definitely agreeing with the statement “I find myself drawn more strongly to people than to things” would suggest a more neurotypical person. When Baron-Cohen and his colleagues administered the test in a clinical setting, the average score in the control group was 16.4 out of 50 while 80 percent of those diagnosed with autism or a related disorder scored 32 or higher. But if you scored 33 would you be autistic? Not necessarily. What about 36 Or 39 What is the cutoff point?
Label-locked thinkers want answers.
This kind of thinking can do a lot of damage. For some people, a label can become the thing that defines them. It can easily lead to what I call a handicapped mentality. When a person gets a diagnosis of Asperger’s, for instance, he might start to think, What’s the point? or I’ll never hold down a job. His whole life starts to revolve around what he can’t do instead of what he can do, or at least what he can try to improve.
Label-locked thinking goes the other way, too. You might be comfortable with your diagnosis but worry that it will define you in the eyes of others. What will your boss think? Your coworkers? Your loved ones? Half the employees at Silicon Valley tech companies would be diagnosed with Asperger’s if they allowed themselves to be diagnosed, which they avoid like the proverbial plague. I’ve been to their offices; I’ve seen the work force up close. Many of the hits on my home page come from Silicon Valley and other areas with a high concentration of tech industries. A generation ago, a lot of these people would have been seen simply as gifted. Now that there’s a diagnosis, however, they’ll do anything to avoid being ghettoized.
Label-locked thinking can affect treatment. For instance, I heard a doctor say about a kid with gastrointestinal issues, “Oh, he has autism. That’s the problem” — and then he didn’t treat the GI problem. That’s absurd. Just because gastrointestinal problems are common in people with autism doesn’t mean that the GI problems are untreatable on their own. If you want to help the kid with GI issues, talk about his diet, not his autism.
And label-locked thinking can affect research. “One of the curses in this field,” a study on vision in autism concluded, “is the size of the error bars, which always seem to be at least twice as large in the ASD data compared to the controls.” Error bars twice as large as the controls’ error bars? Right there, that should tell you that you have a huge variation in the sample — that you have subgroups in the population that need to be identified and separated out. You throw people with Irlen syndrome and people who look out of the sides of their eyes into the same sample and you’ll end up comparing apples and oranges. The error bars aren’t a curse. They’re an obstacle that the researchers have created for themselves and then placed in their own path.
The same is true for studies that conclude that some solutions to sensory problems, like weighted vests or Irlen glasses, don’t work for people with autism. I used to read these studies, and I would tell myself, But I’ve seen weighted vests work, again and again! The problem with the research, I’ve realized, is that autistic people don’t all have the same sensory problems. If you have twenty people with autism, shaded glasses or weighted vests will help maybe three or four. So researchers say, “Well, look — these devices work on only 15 or 20 percent of the autistic population!” So what? That result doesn’t mean that colored glasses don’t work for autism; it means that colored glasses do work for autistics with certain specific visual problems. They work on a subgroup of the autistic population.
I’m not saying that we shouldn’t use labels. Of course we should. Without the label that Leo Kanner gave it, autism might have gone undiagnosed, untreated, and just plain ignored. Labels have been incredibly important, and they will continue to be incredibly important. For the purposes of medicine, educational benefits, insurance reimbursements, social programs, and so on, they’re necessary. And if you’re a researcher looking into autism, sometimes it makes sense to test only autistic subjects against controls.
But sometimes it doesn’t, because autism is not a one-size-fits-all diagnosis.
However the APA defines autism, the diagnosis is going to be imprecise. That’s the nature of a spectrum. The lack of a diagnosis was what the first formal set of standards in the DSM-III tried to correct, and the lack of precision in the diagnoses for autism and autism-related disorders was what subsequent editions tried to correct. Unfortunately, I don’t think the latest effort — the DSM-5 — is going to be much help in clearing up the confusion, and in some ways, it’s only going to complicate the situation.
In the DSM-IV, a diagnosis of autism depended on three criteria, called the triad model. Those criteria were:
• Impairment in social interaction.
• Impairment in social communication.
• Restricted, repetitive, and stereotyped patterns of behavior, interests, and activities.
The first two might sound similar to each other in that they both involve issues of socializing. In fact, that’s the official justification for collapsing them into one criterion for the DSM-5. In a 2010 presentation before the federal Interagency Autism Coordinating Committee, the chair of the DSM-5 Neurodevelopmental Workgroup said, “Deficits in communication are intimately related to social deficits. The two are ‘manifestations’ of a single set of symptoms that are often present in differing contexts.” As a result, the DSM-5 uses a two-criteria, or dyad, model:
• Persistent deficits in social communication and social interaction.
• Restricted, repetitive patterns of behavior, interests, or activities.
I understand why the APA might consider changing from a triad to a dyad model. The idea of separating the social from the behavioral does have a basis in science; the two domains are in fact biologically different. In lab tests on mice, researchers have shown that risperidone, an antipsychotic drug, does not affect social behaviors but does affect fixated behaviors — possibly because it sedates the mice. Conversely, researchers have shown that the social behavior of mice was improved by training, but the fixated behavior was not. Those results alone tell us that repetitive behaviors and social problems operate in separate systems in the brain. So a dyad system that recognizes the distinction between those two systems does make sense.
What isn’t scientific about the DSM-5’s handling of the diagnostic criteria, however, is its collapsing together social interaction and social communication. Social interaction covers nonverbal behavior that involves being with another person — making eye contact, smiling, and so on. Social communication covers the verbal or nonverbal ability to converse — share ideas and interests, for example. Do impairments in social communication and impairments in social interaction actually belong to one single domain? Does an inability to get words out and master grammar and syntax (known as specific language impairment or syntactic-semantic disorder) really come from the same place in the brain as a tendency to speak with abnormal intonation and give conversational responses that are socially inappropriate (known as pragmatic language impairment or semantic-pragmatic disorder)? Are language mechanics and social awareness closely related, neurologically speaking? I doubt it — and I’m not alone in that doubt.
A 2011 paper in the Journal of Autism and Developmental Disorders surveyed more than two hundred fMRI and DTI studies in an effort to determine whether the dyad model has a basis in neuroimaging data. The authors’ conclusion: “only partially.” They found that neuroimaging supports the separation of behavior and communication into two categories. No surprise there. But they also found that neuroimaging supports the division of communication into two further categories, just like the DSM-IV said — though not necessarily the two categories the DSM-IV described!
The DSM-5 is also changing the scope of the diagnosis itself. In the DSM-IV, the autism-related category was pervasive developmental disorders, and it included these diagnoses:
• Autistic disorder (also called “classic” autism)
• Asperger syndrome
• Pervasive developmental disorder not otherwise specified (or atypical autism)
The DSM-5 lists one:
• Autism spectrum disorder
So, you might ask, what happened to Asperger’s and PDD-NOS? Let’s take them one at a time.
The big change regarding Asperger’s and autism is speech delay.
Previously, if you had speech delay as a kid, as I did, then you fell on the autistic side of the diagnostic divide (assuming you met the other necessary criteria, of course). If you didn’t, then you fell on the Asperger’s side. Now some of the former Aspies will get an ASD diagnosis, just by virtue of meeting all the criteria for that diagnosis but not having speech delay.
The APA says that those already diagnosed with autism will keep the diagnosis. But what about the previously undiagnosed Aspies who meet only the social half of the new dyad criteria — deficits in social communication and interaction but not in repetitive behaviors and fixated interests? They could find themselves in another subcategory altogether: communication disorder. Specifically, they’ll find themselves receiving a diagnosis that’s new to the DSM: social communication disorder. Which is, basically, autism without the repetitive behaviors and fixated interests. Which is, basically, rubbish. (To my way of thinking, social impairments are the very core of autism — more so than the repetitive behaviors.) So having a diagnosis of social impairment that’s distinct from the diagnosis of autism is the same as having a diagnosis of autism that’s distinct from the diagnosis of autism!
Those who previously would have been diagnosed with Asperger’s might learn that they don’t belong in the neurodevelopmental-disorders category at all, at least not officially. They could find themselves in a whole other diagnostic category: disruptive, impulse-control, and conduct disorders. The decision ultimately comes down to an individual doctor’s opinion — and if you say that that doesn’t sound like science, I wouldn’t disagree.
First, as a biologist, I find just about this whole diagnostic category scientifically suspect. The category includes six diagnoses. As far as I can see, only one has any basis in science: intermittent explosive disorder. Neuroimaging shows that if you lack top-down control from the frontal cortex to the amygdala, you’ll be prone to outbursts that will get you fired or arrested. But as for the other diagnoses in the dis￼￼￼￼￼￼￼￼￼ruptive, impulse-control, and conduct disorders category? I smell a strong case of “If we label them that, then we don’t have to give them ASD services and we can just let the police deal with them.” The DSM might as well call this category Throw ’Em in Jail.
Second, these diagnoses overlook the gifted but frustrated — the typical Aspie or high-functioning autistic who is laboring in a non-sympathetic environment. Consider the oppositional defiant disorder diagnosis: “The disturbance in behavior causes clinically significant impairment in social, educational, or vocational activities.” I guarantee you that if you take a third-grader who can read high-school math texts and make him do baby-math drills over and over and over, he will turn oppositional defiant — because he’s bored out of his mind.
How do I know? Because I’ve seen these cases — kids who are considered to have severe behavior problems at school until you give them math lessons that meet them where their brains are. Then their behavior normalizes, and they become productive and engaged — maybe even model students.
And here, again, we see the limitations, and even dangers, of label-locked thinking—the difference between what behavior looks like from the outside and what it feels like from the inside.
As for PDD-NOS, DSM-IV used this catchall diagnosis to describe several scenarios, including atypical autism, defined as “presentations that do not meet the criteria for autistic disorder because of late age of onset, atypical symptomatology, or subthreshold symptomatology, or all of these.” In the DSM-5 though, people with that diagnosis might find themselves jettisoned from autism altogether and put into another neurodevelopmental-disorder subcategory, intellectual-development disorders — specifically, intellectual or global developmental delay not elsewhere classified. No wonder so many parents feel like they’re in the Diagnosis of the Year club.
For a lot of people, the changes to the DSM won’t make a difference. For instance, under the DSM-5 guidelines, I would be diag￼￼￼￼￼￼￼￼￼￼￼nosed with autistic spectrum disorder. If you look at the description of what constitutes social impairments and repetitive behaviors, I definitely qualify. Extreme distress at small changes? That was me as a kid. Fixated interests? Boy, I had that. Hypersensitivity to sensory input? Let me tell you about the squeeze machine.
But for a lot of people, these changes will make a huge difference. A 2012 survey of 657 people who had been clinically diagnosed with any one of the three DSM-IV autism spectrum disorders found that 60 percent would continue to receive the ASD diagnosis under DSM-5 criteria but 40 percent would not. Breaking those numbers down into subgroup diagnoses, the researchers discovered that 75 percent of subjects who had received the specific diagnosis of autism according to DSM-IV criteria would also meet the DSM-5 criteria for ASD, but only 28 percent of those diagnosed with Asperger’s would, and only 25 percent of those diagnosed with PDD-NOS would.
A later study that concentrated on only the PDD-NOS diagnosis reached a far more optimistic conclusion: nine out of ten children with a DSM-IV PDD-NOS diagnosis would be eligible for a DSM-5 ASD diagnosis. The disparity between the two reports, however, should give any parent, let alone scientist, pause.
What practical effects will these diagnostic changes have? Will people who were labeled Asperger’s and are now labeled autistics experience a different response from the world? From themselves? How will these changes affect insurance coverage? What about social services? Autistics have more problems than those with Asperger’s; will they still get the same range of help as before? That question will be decided on a state-by-state basis, but these changes have opened a Pandora’s box of possibilities.
And research! Any study of autism that uses DSM-5 criteria for autism is going to be mixing speech-delay apples and non-speech-delay oranges. For instance, we’ve seen in the literature that sensory problems tend to be a whole lot worse among members of the population who have speech delays. How are researchers going to be able to compare DSM-5 sensory-problem studies with pre-DSM-5 studies?
To me, the DSM-5 sounds like diagnosis by committee. It’s a bunch of doctors sitting around a conference table arguing about insurance codes. Thanks to label-locked thinking, we now have a cornucopia of diagnoses — and there simply aren’t enough brain systems for all these names.
Back in 1980, when the DSM-III first tried to codify the diagnosis of autism, nobody knew about brain systems. Nobody knew much about DNA sequencing. But now we do. We may not be able to apply those advances in science to the DSM yet, but what we can do, I feel, is begin to change the way we think about the autistic brain. Instead of talking about sets of symptoms in an attempt to assign them a label, we can begin to talk about one particular symptom and attempt to determine its source. We’ve reached a point in our research that we can match symptoms and biology.
For the first thirty years or so after Leo Kanner introduced the term autism, in 1943, the emphasis in the psychiatric community was on finding a cause, and because psychoanalytic theory dominated the psychiatric thought of the day, the cause was hypothesized to be the behavior of the parents, especially the mother.
Let’s call this period phase one in the history of autism, and let’s say it extended from 1943 to 1980, the year the American Psychiatric Association published the DSM-III.
That edition of the DSM represented a shift in the psychiatric community toward greater scientific rigor in its treatment of mental illnesses, a shift that included the first official diagnosis for autism. Since then, much of the discussion about autism has involved what specific symptoms make up the diagnosis.
Let’s call this period phase two in the history of autism, and let’s say it extended from 1980 to 2013, the publication year for the DSM-5.
The diagnosis can and will continue to change, but now we can ￼￼￼￼￼￼￼￼￼￼shift our emphasis once again. Thanks to advances in neuroscience and genetics, we can begin phase three in the history of autism, an era that returns to the phase-one search for a cause, but this time with three big differences.
One, the search for the cause involves not the mind but the brain — not some phantom refrigerator mom but observable neurological and genetic evidence.
Two, because we realize how extraordinarily complex the brain is, we know this search will lead not to a cause but to causes.
Three, we need to be looking for a cause or multiple causes not of autism but of each symptom along the whole spectrum.
Phase-two thinking says, Maybe I can’t ski well because I’m autistic. Phase-three thinking says, Maybe I can’t ski well because I have a smaller than normal cerebellum.
Phase-two thinking says, Let’s group people together by diagnosis. Phase-three thinking says, Forget about the diagnosis. Forget about labels. Focus on the symptom.
Instead of — or at least in addition to — assigning human subjects to studies by their autism diagnosis, we should be assigning them by their main symptoms. As I learned from examples like Carly Fleischmann’s description of feeling overstimulated in the coffee shop, I think researchers should stop pooh-poohing self-reports and start looking at them very carefully and, in addition, begin eliciting them from subjects in new ways. Then they should be putting the subjects into studies based on those self-reports.
I once had a graduate student who saw wavy lines between the curved lines in a drawing of a cattle chute, and sometimes she saw only pieces of words. She wasn’t autistic, but these symptoms were notably similar to those described by Donna Williams, who definitely is autistic.
I say, Throw ’em both in a scanner, and let’s see what lights up. Let’s see where the problem is. Is it in the language-output area? Language-meaning?
￼￼￼￼￼￼￼￼￼￼Let’s take the people who can’t ride on an escalator because they can’t figure out how to get on and off. Or let’s take the people who hate driving at night. Let’s take those subgroups and put them against controls who don’t have that problem. Let’s take this secretary over here who can type 180 words a minute. Let’s take another secretary who can type 90 words a minute. Let’s throw them both in a scanner and compare them, motor cortex to motor cortex.
Some researchers, I’m pleased to see, are beginning to recognize the limitations of labels. And they’re beginning to recognize the need for narrower definitions of targets. A 2010 article, “Neuroimaging of Autism,” concluded: “For autism it becomes more and more clear that the possibility to identify one single marker might become very small, just because of the large variability we meet in [this] spectrum. In this view the definition of smaller autism subgroups with very specific characteristics might give a key to further elucidate this complex disease” (emphases added).
Personally, I would go even further and argue that we need to think not just about smaller autism subgroups that are defined by their symptoms but about the symptoms themselves. Because thinking about individual symptoms on a symptom-by-symptom basis will eventually allow us to think about diagnosis and treatment on a patient-by-patient basis.
My friend Walter Schneider, who developed high-definition fiber tracking at the University of Pittsburgh, is already making that argument—probably because he has so vividly seen for himself the potential of this approach.
“We are searching for actionable diagnosis,” he says. “Not just that we say, ‘Yeah, you’re different,’ but, ‘You are different and because of this particular form of difference, we think this is the most likely path for getting you to as much of the outcome as we want you to get.’ We want to go in and in on that individual brain — not a group study but an individual brain — so we can say to a parent, ‘This is what the situation is, this is what we expect the effect to be, and this is how we plan to get around it as efficiently as possible to give you effective communication with your child in the next two years.’”
You can hear the same argument beginning to surface in genetics as well. Yale neurogeneticist Matthew W. State likes to invoke the medical phrase "from the bench to the bedside" — meaning from experiments on groups to treatments for individuals. In a 2012 article in Science, he and collaborator Nenad Šestan suggested that researchers look for inspiration from other areas of medicine that have made this transition. “For example, heart disease and stroke prevention both rely in part on the management of hypertension,” they wrote. “It may well be that ASD and schizophrenia will increasingly be thought of in a similar light” — different behaviors arising from the same genetic source. As a result, Šestan and State anticipated that treatment trials would be organized around “shared mechanisms” rather than “psychiatric diagnostic categories.” They didn’t doubt that this rethinking of the autistic brain would be challenging. But like Schneider, they foresaw the development of therapies that were not only more effective but “more personalized.”
Twenty years from now, I think we’re going to look back on a lot of this diagnostic stuff and say, “That was garbage.” So as I see it, we have a choice. We can wait twenty years and several more editions of the DSM before we start to clean up this mess. Or we can take advantage of the technological resources that are beginning to become available and start phase three right now.
I choose phase three.
Excerpted from "The Autistic Brain: Thinking Across the Spectrum." Copyright (c) 2013 by Temple Grandin and Richard Panek. Reprinted by permission of Houghton Mifflin Harcourt Publishing Company. All rights reserved.