Michelle Dawson on Autism and Atypicality (Ep. 46)

Subscribe: Apple Podcasts / SoundCloudRSS

Perhaps no one else in the world more appreciates the challenges facing a better understanding of autism than Michelle Dawson. An autistic herself, she began researching her condition after experiencing discrimination at her job. “Because I had to address these legal issues and questions,” she tells Tyler, “I did actually look at the autism literature, and suddenly I had information I could really work with. Suddenly there it was, this information that I was supposed to be too stupid to work with.” And so she continued reading papers — lots and lots of papers — and is now an influential researcher in her own right.

For Michelle, the best way to understand autism is to think of it as atypical information processing. Autistic brains function differently, and these highly varied divergences lead to biases and misunderstanding among typical thinkers, including autism researchers.

In her conversation with Tyler, she outlines the current thinking on autism, including her ideas about cognitive versatility and optionality, hyperlexia and other autistic strengths, why different tests yield wildly different measures of IQ among autistics, her ‘massive bias’ against segregating autistics, how autistic memory is different, why sometimes a triangle is just a freaking triangle, and more.

Listen to the full conversation

Read the full transcript

TYLER COWEN: Today we’re up in Montreal, and I’m very honored to be here with Michelle Dawson. Michelle is one of the people I admire most, actually. She has become a very well-known and very influential autism researcher. Her background: she is herself autistic. In some regards, you could describe her as self-taught, but she has become a kind of one-woman force advocating for more science, and . . . You’re shaking your head “no.”

MICHELLE DAWSON: I’m not an advocate.

COWEN: She is arguing for science and ethics being brought into the autism discourse and discourse more generally. Would you accept that description of what you do?

DAWSON: Good enough.

COWEN: Good enough. Okay.

COWEN: So let me start with a very general question. If you ask what would be the most underrated, nonaccountable, and mysterious force controlling people’s lives, and I said that right now, it was actually something called the DSM, Diagnostic and Statistical Manual of Mental Disorders, which influences a lot of issues in the law — what insurance companies do, what hospitals do, when people are institutionalized — what would your reaction be to that statement?

DAWSON: I think that the DSM could be a very bad big deal to people who, for instance, are denied basic human rights or standards in science.

For people who are not situated like that — and that would be most people — the shortcomings of the DSM-5 will be offset by people in research and practice who have good standards, who understand that their job is to benefit people more than they harm them, who know how to read the scientific literature, and so on. So, it depends who you are and how you’re situated.

Also, you have to consider the larger context of what people see as influential or not, which might be very arbitrary at any moment.

COWEN: What would be the best understanding of autism, from your perspective?

DAWSON: The best understanding is seeing autism as atypical brain functioning, resulting in atypical processing of all information. So that’s information across domains — social, nonsocial; across modalities — visual, auditory; whatever its source, whether it’s information from your memory, information coming from the outside world, that is atypical. So that is very domain-general atypicality.

What autistic brains do with information is atypical. How it’s atypical, in my view, involves what I’ve called cognitive versatility and less mandatory hierarchies in how the brain works, such that, for example, an autistic brain will consider more possibilities, will nonstrategically combine information across levels and scales without losing large parts of it, and so on. And that applies to all information.

That is strictly my view. I’m not sure anyone would agree with me.

What autistic brains do with information is atypical. How it’s atypical, in my view, involves what I’ve called cognitive versatility and less mandatory hierarchies in how the brain works, such that, for example, an autistic brain will consider more possibilities, will nonstrategically combine information across levels and scales without losing large parts of it, and so on. And that applies to all information.
That is strictly my view. I’m not sure anyone would agree with me.

COWEN: Do you think it’s true that autistic brains somehow take in or process more information?

DAWSON: They can. That depends on what is available in the environment. Not all autistics are well situated when it comes to access to information that they can process well. Autistics are different in the kinds, quantities, and arrangements of information they process well. Just having a huge quantity of information is not necessarily all that an autistic person needs.

The atypicality is more complex than that. But yes, an autistic person can process more information across levels and scales without losing it, or without a lot of the information being edited in sort of cognitive editing — various ways that information is made manageable in a typical brain. Autistic people don’t need to do that to get the information in a way that it can be processed and useful, that they can learn from it and use it.

On hyperlexia and other autistic strengths

COWEN: What’s an example of something an autistic person might be better at doing than a nonautistic person? Again, with a lot of variation across autistics.

DAWSON: You know, the problem here is that there’s a really long list of things that autistic people can do well, and given the opportunity . . .

COWEN: Give us, say, half a dozen. Just bullet points.

DAWSON: The almost classic example is hyperlexia, which you see starting with very early development. You see this in very young autistic children, who have a spontaneous interest in printed materials. It can be just text, but the information can include other kinds of printed materials: numbers, combinations of letters and numbers.

And they use this information across different scales. So from individual letters like the alphabet, given the opportunity, they will order individual letters into the order of the alphabet, for instance. But they don’t just use single letters. They will also use complex arrays of texts, like newspapers.

These are very young children — that’s going to start at age two. If they’re in a book-rich, text-rich environment and have free access to this, they will figure out how language is structured, and that happens even with very difficult orthographies like English. English orthography is so difficult that typical children take an extra year or so to learn it because it’s wacky. In a lot of ways it doesn’t make sense; the rules don’t make sense.

But autistic children, even starting, these very young children can work this out, and they figure out the irregularities. They figure out the irregularities using the sort of array of information from letters to words to phrases to sentences to paragraphs to entire text and arrays, and so on. They’ll figure it out, and typical children don’t really do this spontaneously.

COWEN: My mother used to tell me that when I was two, I just taught myself how to read. For a very long time, I never took her seriously. I just thought, “Well, mothers say things like this.” But when I learned about hyperlexia, it occurred to me that she had an actual, literal point.

So, if you’re a hyperlexic, and then you grow up, you keep some kind of special reading ability typically?

DAWSON: We don’t know nearly enough about how any autistic strength in processing information evolves over time. And again, you have to keep in mind that we start looking at this in children. You don’t know what kind of information is going to be available to them. We are not paying attention to that at all.

At the same time, now, in autism, one of the most common aspects of popular interventions is to really ration information, keep it away from autistic people. So you have to consider that in how an ability develops. What kind of access to information do they have in whatever context they’re in, whether it’s in their home, in their school, and so on?

But it is not a dead end, that is for sure. It’s been thought of wrongly as a dead end, and that’s been true of autistic abilities in general — that they are thought of as dead ends. But it’s clearly not the case, even in situations which are very far from good for the autistic person as far as availability of information and their overall situation with respect to how resourceful they can be, how they can add to and build on their abilities, and so on.

COWEN: Now often, in popular discourse, you’ll hear autism or Asperger’s associated with a series of personality traits or features of personality psychology — a kind of introversion or people being nerdy in some regard. In your approach, do you see any connection between personality traits and autism at all?

DAWSON: There is a small literature that shows some connection. I think it’s very weak, and I say no, I don’t think autism is about personality. Autism is sort of orthogonal to personality. The two are not related. Whatever relation there is does not . . . arises from some third factor, let’s say. If there is one — and again, the evidence is, I think, very weak connecting autism to personality — so just say that maybe, if there’s something, let’s say that personality in autistics might be more high variance. That would be my totally wild guess, but I don’t think autism itself is about personality.

On biases affecting autistics

COWEN: One thing, if I think about your work in general, one thing it’s caused me to do is to see a new bias in how humans make judgment, and that is a bias to underrate atypical kinds of intelligence. I view this as a common thread running through your work. Maybe this is what I’m reading in, rather than anything you’ve said explicitly.

But if you look at, say, issues of animal intelligence, there’s a growing body of evidence that animals feel pain, they’re intelligent, they lead social lives, they’re complex in ways maybe we hadn’t understood earlier. Still, a lot of people don’t absorb this knowledge very readily.

If you consider Down Syndrome individuals, when I was growing up, there was a common sense of a kind of hopelessness. Now there’s a Down Syndrome individual who’s a lobbyist, people playing roles on TV, and so on. This would not have been expected by most people in the 1970s.

Why do you think it is we have this bias to underrate atypical kinds of intelligence, or do you even agree with that characterization?

DAWSON: I would be careful about the sort of people in general. I would say that maybe it does seem to be something that if you take the median person, or you do this as a society, sort of collectively, not necessarily as individuals . . . it seems to be something that people are very bad at, to detect abilities and worth in highly atypical individuals. For sure, you see this very much in autism with the whole notion of severity, and this is not defined. There is no workable definition of autism severity.

People use that word all the time. They use it to mean that somebody’s had a bad outcome, or that someone is badly behaved, or that somebody performs badly on an IQ test, and so on. There is something bad going on here, and it’s sort of the equation that the more atypical a person is, the worse they must be. They must not be intelligent. There must be something extra wrong, more wrong about their behavior, and so on.

Now, this is where I’m very familiar with the whole history of this, and it’s obviously problematic. We don’t — I hope we don’t look at a blind person who is a successful lawyer and assume that he is only very mildly blind or barely blind at all, and then look at a blind person who has a very bad outcome and assume that they must be very severely blind.

We do make those kinds of judgments in autism, saying, “The more atypical the person is, the worse they must be in some sense.” That kind of bias has not only harmed a lot of autistic people, it really has impeded research.

I hope we don’t look at a blind person who is a successful lawyer and assume that he is only very mildly blind or barely blind at all, and then look at a blind person who has a very bad outcome and assume that they must be very severely blind.
We do make those kinds of judgments in autism, saying, “The more atypical the person is, the worse they must be in some sense.” That kind of bias has not only harmed a lot of autistic people, it really has impeded research.

We do not have a definition for autism severity, but it gets at that idea, that bias — very unhelpful and harmful bias that the more atypical an individual is, the more there must be wrong with them. And autism is almost the paradigmatic case of that happening, so if you want to make a case for that, autism would be where to do it.

COWEN: Are there biases that nonautistic individuals have that autistics might have less, or maybe not at all?

DAWSON: Well, there’s a literature testing some of these biases, and yes, the short answer is yes. I think in all cases you would want a bigger literature; you’d want to wait for replications and so on.

COWEN: What’s an example of a possible bias that’s weaker in autistics?

DAWSON: Well, noticing that two offers which are posed in different ways are actually identical, or being more likely to notice that the outcome will be identical, however it’s posed. Those are framing effects. That’s been replicated once, where, in effect, the autistics are less likely to be fooled by the framing. So, they detect when two situations are the same, whereas the typical group are biased such that they are fooled by the framing. It’s not 100 percent, but autistics are just . . . They will still consider it.

COWEN: What about optical illusions? Is there evidence there?

DAWSON: It’s very mixed, and it’s very picky about the task and the instruction. Some more than others. I’m trying to remember the Shepard’s table, or have I got the name of the table right? The table illusion has been replicated. The one with the line — now I’ve forgotten its name — has been replicated. Others, not so much. It’s very mixed.

There’s a question of, can autistics see an illusion? They probably can. Do they consider other possibilities? Yes, they do, and they will consider other possibilities that the typical population might not. So that produces some differences in results, depending on the task and so on.

COWEN: Are there examples of confirmed biases in autistics, where the bias is stronger in autistics?

DAWSON: I would say no. Keep in mind what I think here is not that autistics don’t have biases. Because they can . . . Well, they will process information atypically. Within that range is the possibility of checking, you know, what is the typical hierarchy, even if it’s not . . . It won’t be mandatory, though. That won’t be it. They will consider other possibilities. So, that will attenuate biases as opposed to strengthen them.

And that, I think, is what you find in the literature, again, depending whether the task is sufficiently complex for that to show up, what the task demands are. My favorite of these is . . . It’s a bit dodgy because it comes from a couple of different papers, but it shows where . . .

A Penrose triangle. Credit: illusionsindex.org

A bit of my background, going way back in perception and finding that autistic people were better at drawing impossible figures. Impossible figures are like Penrose triangles. They are things that you can’t build. So this was interpreted as a couple of possible deficits. They are not — failing to see the global structure, for instance. Yet, if you ask autistic people whether the impossible figure is impossible, they know it’s impossible. So autistics can have both the possibilities. They understand that an impossible figure is a sort of possible drawing.

On social intelligence

COWEN: There are a variety of claims that autistics either have difficulties processing social information or lack certain kinds of social intelligence — why they’re awkward in particular social situations. In your framework, how do you make sense of these claims? And how would you describe or redescribe the phenomena people are talking about?

DAWSON: Actually, how autistic people perform in social tasks is high variance. It is very hard to pull out specific deficits where you can’t point out areas where autistics have performed at least as well as typical people on the task.

There is an exception, which is the most replicated autistic deficit. I believe there’s one finding in the contrary direction. But all the rest — and there’s a lot, especially for the autism literature — show what is considered an autistic deficit.

This is the task of attributing deep mental states to geometric shapes. Usually they are triangles. They will be animated. They sort of move around, and it goes beyond simply attributing agency to the triangles, which autistics can do. You must attribute mental states to them, profound mental states, like they are jealous or they’re flirting. And autistics are very bad at this task. And this has been replicated quite a lot, even at a population-based level, which is not something you necessarily see a lot of.

So, is this a social deficit? Autistics are definitely bad at this task. Now, why might that be if you look at things in my own biased way? It may not necessarily be that autistics don’t notice that these interactions between geometric shapes, or among geometric shapes, resemble something social that involves mental states.

But they are less likely, and they are not going to totally get rid of the accurate information that these are just freaking triangles. They will only go so far. They are not going to lose that possibly important information that these are just triangles. And you would think that this would quite reasonably produce a bit of caution in assuming that you know quite a lot about their profound mental states.

COWEN: Is the following a fair description? Let’s say that many autistics are taking in more information than would be typical. So sometimes those autistics will end up confused, or they will feel confused or act as if they appear they’re confused.

But other times, if they develop procedures for processing that information — rules — they become better at interpreting the information. Maybe they have to become better because they face this more difficult task. So, you will have autistics show unusual mastery of social situations. Even if they appear awkward, they’ll see much more detail than nonautistics would in some cases.

But then in other cases, they’ll be more confused. And that’s where this high variance comes from that you mentioned earlier.

DAWSON: Well, there’s a huge literature in autism about how autistics judge facial expressions of emotion in other people. And what you have in the autism literature is, you haven’t only just turned autistic people into stereotypes and cartoons, you’ve done that to the typical population.

This is really at odds with the nonautism literature on facial expressions, which is much more complicated. In the autism literature, it’s assumed that you can just read people’s inner emotions and mental states. Mental states are not necessarily well defined, that it’s a simple matter, that it is sort of written all over somebody’s face, or even you can read it just from looking at a photo of their eyes.

Smiles of frustration or delight? Credit: Hoque et al.

And things are far more complicated than that in the literature, in the nonautism literature. For example, MIT — their affective computing group, Rosalind Picard did these fantastic studies showing that people smile in frustration, and those are real honest-to-goodness Ekman-type smiles. You have the whole facial action coding thing going on. Those are real, genuine smiles that people smile in frustration when they are genuinely frustrated. They don’t do it when they’re acting out frustration. And there are many other examples like that.

People smile for many different reasons, and that is acknowledged to some degree in the literature in the typical population, not in the autism literature, where things are completely simple. They’re just very caricatured and cartoonish. Now, what you find is that the typical population can decipher their way through this. They know what these facial expressions are supposed to represent, even if they don’t look like that in real life.

Autistics are — maybe because their experiences are quite complex with how people respond to them starting early in life, and I’m just wildly speculating here — but autistics are going to notice that things are more complex and uncertain than that. Again, it’s the considering more possibilities, and that will very much hamper their task performance if what you are looking for is this automatic certainty that these acted expressions are all there is, which is not accurate.

And that leads to many problems because we’re actually training autistic people to ignore the complex, real, important information in favor of the caricatured, stereotyped, simplified, probably wrong information, and we should really think about that. But that gives you an idea of looking at social deficits, thinking about how autistics process information, and also actually looking at the literature itself.

On chord disembedding and other cognitive skills

COWEN: Let me mention a few cognitive skills, and we’ll just run through them briefly, but tell me what you think we know about autistics and nonautistics. Chord disaggregation: you hear a chord and you want to figure out what are the notes inside the chord. What do we know?

DAWSON: Autistics are good at chord disembedding, and that’s what that ability is called. Now, this is a great way to bring in the importance of the literature on autistic savants. Autistic savants are . . . The best way to think of them is autistics who are very obvious.

COWEN: They can multiply large numbers in their head, they can do the so-called . . .

DAWSON: It’s very obvious that they are processing information atypically. That makes them very obvious autistics. They are very obviously high variance in their range of abilities. There can be quite drastic contrast between things that they are bad at and things that they are excellent at, and this, again, makes them very, very obvious people.

Their abilities are extraordinary for anybody. They have difficulty doing quite basic things. So, they’re very obvious. They’ve been extremely important in research in spite of the literature on them being tiny. It’s disproportionately contributed to what we know. Yet this extraordinary contribution of these individuals gets denigrated all the time. That’s awful to see.

So, this is how we know. Because autistic savants can disaggregate chords, this was then studied in less obviously talented autistics, who it turns out are better at disaggregating chords, chord disembedding. Pamela Heaton did work in this area.

My favorite bit in this literature is in a book chapter by Linda Pring, where they tested Derek Paravicini. He is blind and autistic, outstanding musician. He’s considered intellectually disabled, developmentally disabled, however you want to put it. So, as they do it in the savant literature, they test these savants against people who are experts, so expert musicians, and they conk out pretty quick after five or six notes. They’re getting in a lot of trouble, they can’t disembed these chords, they can’t tell you which notes they consist of. And Derek Paravicini kept going.

COWEN: He could do the first chord in “Hard Day’s Night,” perhaps, right?

DAWSON: Well, exactly, but you know what? What they said in this, what the researchers here didn’t anticipate — and it’s just awful that this isn’t reported in a formal paper, this is in a book chapter — but that it didn’t occur to them that they would run out of fingers. They attempted a [laughs] 10-note chord, and he could still disembed it.

And they had not anticipated that they would need someone with more than 10 fingers to properly test the limits of this autistic person’s ability. This is somebody who is a creative, fantastic musician in every sense. But to know that he can do this, it’s almost by accident that we know this, and we have not managed to get to the limit of his abilities here because they just didn’t anticipate.

COWEN: What do you think of the hypothesis that virtually all autistics are in some logical way savants, but there’s a socially constructed category of savant that some things, such as multiplying large numbers or doing the calendar trick, whether a particular day far in the future is a Monday or a Wednesday, that is regarded as being a savant.

But if you have a quite ordinary skill that you’re very good at, like making managerial decisions, no one will call you a savant, but in essence, you’re doing the same thing. Agree or disagree?

DAWSON: I’m not going to answer that. I’m going to totally answer a different question, which is . . . what people should notice first off is that savant skills are sort of conglomerated in specific areas. Why might that be? Say, music, hyperlexia, drawing in three dimensions, calculation, calender calculation. Why? Because this is information that tends to be available in most environments, and that’s often quite hard to take out of those environments. Now, you have to think about that . . .

COWEN: So the people can’t take it away from autistics. You can always look at calendars, think about calendar . . .

DAWSON: Well, they do now.

COWEN: They try.

DAWSON: This taking that information away is a feature of many, if not most, popular autism interventions. That’s awful. So I’m answering the question, why are some people considered savants and some not? We don’t know anything! We are not interested in what kinds of information autistics have available to them starting development.

I’m trying to get other people interested in that. It’s very uphill. Because autistics may be specific on how they start to figure out how to work with the kinds, quantities, and arrangements of information they process well.

Let’s say an autistic will have access to, say, music. And they will have access to good music and so on. And they will figure that out. And they have access to a piano, which seems a particularly good array of information for autistics. The musicians aren’t all pianists, but an awful lot of them are. And the piano — all the information is there, right? It’s in front of you. [laughs] It’s complex. It’s all there. So this tends to be the kinds of information that autistics will start with.

Now, let’s say you have somebody, and they figure out music. They figure it out at all levels, from a single note to complex compositions, and this is what savants do. What if they realize then, and it’s in their environment, that they can apply this to something else? And you do get savants that have multiple abilities. That used to be considered — wrongly — not what happened, but it is what happens.

The idea of an autistic polymath has just not been explored at all. So you get a paper where the autistic person . . . You’re just exploring specific aspects of their extraordinary music ability. You will only mention in passing that they are an outstanding chess player. And we don’t know . . . [laughs] We don’t know anything more about that. That’s just an example that is actually in the literature.

COWEN: True or false? Autistics often have a weaker episodic memory, and they’re less likely to think in terms of stories. And when they watch movies or read books, they will pick out very different pieces of information and have very unorthodox readings relative to the mainstream.

DAWSON: I would say that’s very well established, though it is not interpreted that way. The way to interpret it is that autistics do not organize information in memory typically — that should not surprise anybody. They do not use narratives to edit things, to organize things. And that means that they perform very differently on episodic memory tasks and in related tests that require people to organize information into common or popular narratives.

On measuring IQ

COWEN: There are at least two kinds of IQ tests, and autistics do better on one than on the other. And some of your best-known research is in this area. Tell us the basic finding.

DAWSON: There are very different approaches to measuring intelligence. One is a battery of tests, as in Wechsler Scales of Intelligence. You are trying to get at general abilities by using an array of different specific tests, so you’re trying to get at something latent. These are specific abilities, but through those, you come up with some kind of latent general ability that we call intelligence or full-scale intelligence.

Or you can take one test with one format that is sufficiently complex and difficult that it gets at the same thing, just with one test. That would be Raven’s Progressive Matrices, a test that has been extremely influential in the entire intelligence literature. It’s a very . . .

COWEN: What do those questions look like? Just give us a very direct intuitive sense.

DAWSON: It’s a matrix of entries. The number varies, and you have the last one in the matrix — the last item in the matrix is missing, and you choose the right one from an array of offered answers. So, that’s it. It sounds too simple or easy, but it is a very complex test and a very difficult one. And, in effect, it’s played a large role in defining what general intelligence is.

COWEN: And which test do autistics tend to do better on and why?

DAWSON: Autistics do better on Raven, quite a bit better. In some individuals, the discrepancy is spectacular — it’s a lot. So, how they would be judged — again, go to the savant literature, and this was what influenced me to pursue this.

You’ll have an individual on the savant literature who really cannot perform other tests, would be judged as very intellectually disabled on other tests, effectively can’t perform a receptive language test, much less produce language. And the score is very high on Raven — I mean, better than most of the typical population, all but 1 percent or 2 percent. And again, he had to have the opportunity to do that for us to find that out.

There’s a lot of interesting things about that because if you look at the kinds of deficits that have been assumed that autistics have in processing complex information, in executive functioning, in working memory, in all kinds of complex processes, this is what Raven actually demands.

COWEN: And what was your role in discovering this differential between the two kinds of IQ tests? Your personal role as a researcher?

DAWSON: There was a paper in 2004, and it was one of the first ones I actually looked at in an editing capacity after it was accepted provisionally, and it did find, but very tentatively, that autistics were possibly performing better on Raven than these other tests. The interpretation in the paper was that Raven wasn’t a good test for autistics. It must overestimate autistic intelligence.

I thought first that there are other possibilities, and also, I looked around at how Raven was being . . . First off, I had to look at what Raven was and how it was being interpreted in the autism literature, which is that when autistics performed well, it was almost denigrated as a test: “This is just visual-spatial abilities,” and so on. And you look in the intelligence literature, and you have this test that is really central in defining human intelligence that is very important and requires all these abilities that autistics supposedly don’t have.

And I thought, “Well, isn’t this worth pursuing?” And it did take a lot of persuasion to go in that direction, to sort of question the commonplace and typical. And this is typical of the autism literature. Autistics perform well on the test? There must be something wrong with the test.

On how autistics process information

COWEN: Before, you offered an understanding or conception of what autism is, a definition. But if we get to the nitty-gritty question of what’s the core mechanism or the core difference in autistic brains, some people have suggested there’s some kind of weaker or disabled or maybe more optional form of top-down processing. So there’s maybe more direct access to smaller bits of information or greater optionality, as to the level at which information is accessed.

If you’re characterizing what you think autism actually is, in terms of how the information is processed, what do you think of that hypothesis? And how would you best articulate how you understand it?

DAWSON: Well, it’s my hypothesis, the optionality anyway, that’s me. So I’m maybe not the person you should ask.


COWEN: No, you are the person I should ask.

DAWSON: Now, how do I answer that? These things don’t happen in the proper order if you look at the literature. So this came out of, actually, a lovely study done by Isabelle Soulières, who is a PhD student at the time, and she was looking at categorization. Her field is categorization, and she was looking at how autistic people categorize ellipses.

The interesting thing about categories is that they distort perception. People will form a category boundary, let’s say wide ellipses versus skinny ellipses, and where they put the boundary will mean that they distort how they see those things that are just one side or the other of that boundary, even though the difference isn’t any greater than those who are both on one side and both on the other. The difference between the two is actually the same, but because there’s been this division into two different categories.

This is called a discrimination peak. And you notice that, because it’s found in the typical population, it’s not considered a deficit or biased, it’s considered a peak — something good.

So the question was, “Do autistics have this peak?” They’re asked to do two things: Classify these ellipses, which is, in effect, categorize them. So can they categorize these ellipses as wide or skinny? And you can look and see whether they’re doing this. Then, do they produce the discrimination peak?

The autistics produce this completely typical classification curve. They formed typical categories when this was the task. But when they’re asked to discriminate one from another — tell them apart — they did not show a discrimination peak. That is to say, they were not biased by the category boundary.

I leapt in on this [laughs] because this was considered to be a deficit in category formation. I said, you can’t say that because the autistics, when you ask them to form categories, formed typical categories. You can’t form more typical categories than that. The classification curve was identical. Then you asked them to discriminate. They performed that task. That doesn’t mean that they forgot how to categorize. It means that bias was not useful in performing the discrimination task accurately.

So there came the idea of autonomy. This means that discrimination was autonomous in the autistics. It was autonomous from categorization. In the typical population, even when it wasn’t useful, they formed categories — they had to. So there you have these non-optional categories. They form categories even when it produced biases, even when it wasn’t useful for the task. They had to; they had no choice.

In the autistics, it was optional. The processes were more autonomous. I’ve also called this cognitive versatility. It doesn’t mean that autistics can’t do this. It doesn’t mean that they can’t do what might be thought of as top-down processing, what might be thought of as using priors, or what might be thought of as predictive processing, or whatever. It’s not that they can’t do it. It’s that it is not so mandatory or automatic. It is more optional and more versatile.

COWEN: With this optionality, does that mean that, on average, autistics are somehow more different from each other than nonautistics are?

DAWSON: You see this fairly striking in the literature. You can see this in the whole predictive processing literature, where there’s one optimal way to process everything, which is amazing. This is how . . . supposing, not only within one person but across the entire population, there’s the premise that there is this optimal predictive inference, that there is one optimal way to process everything.

So if you have a hierarchy of information processing that is similar across typical people — and I do believe you have that. There is some variation, but the hierarchy works in about the same way, so you would get much more similarity [in non-autistics].

It doesn’t mean that everyone’s similar. They’re brought up in different places, they have different experiences, they have different genetics, they have different on and on. But they have that greater similarity in what they do with information that will make them more similar in what they do with information and that will give them much more in common.

Autistics do not have that, which makes them not just different from typical people, but quite different from each other. So this is, I think, being slightly recognized now, that autistic people may be more idiosyncratic. This is where it has popped into the literature.

COWEN: There’s a fairly new literature with respect to autism and rare copy variants and de novo copy variants in the genetics of autism. What do you think we’re learning from these articles?

DAWSON: That we should be less hasty in deciding things about autistic genetics. I think that’s the main message. We’re doing this seesaw between . . . The de novo rare variant story was just a very big deal. So you have an entire collection. You have the Simons Simplex Collection, which was really designed to find these, which were assumed to . . . If we find these, we will crack autism genetics.

COWEN: And just to clarify, you’re saying the literature has shown that autistics are quite different from each other in their mutations?

DAWSON: No, that’s not what I’m saying. I’m saying autistics will be quite different from each other in . . . Those are two different issues: how autistics process information and the sort of genetics that may lead to brains that work like autistic brains.

I’m not a genetics person. I do cognitive stuff. But I think the genetic stuff is really a cautionary note of leaping to conclusions that . . . For instance, you have these rare de novo variants that are the real big deal, and even looking for them in the most favorable circumstance possible — that is, you’re looking in families that only have one autistic person — they did not find much. But that was the leading story for a long time.

The seesaw is back to common variants, and we are getting slightly more complicated views — that if you have a rare variant, that doesn’t mean you don’t also have common variants that contribute to how atypically your mind works.

So it’s been this long slog, but it is a cautionary note against any simplistic explanation for autistic genetics. I think there has been some progress, but I do think we have real problems in where we’ve looked and how we’ve looked. Some of that mirrors what has happened generally in genetics.

COWEN: In percentage terms, is autism becoming more common or more frequent over time?

DAWSON: No, autism has not become more prevalent, and there is no reason to believe that it has. The idea that it has become more prevalent has been very destructive for us in research and to actual autistic people, since the premise . . . Really, what you’re saying when you say that there’s been a large increase in autism prevalence is, you are saying older autistics don’t exist.

That means we can’t learn anything from them, including from their early development, from their outcomes, from their achievements. We just have totally erased them and write them off.

COWEN: So, maybe autistics are being treated worse over time, at least in their roles as children. Is that possible?

DAWSON: That is possible. It is possible. Again, very hard idea to test. And it’s one that I wouldn’t even think that I would ever say, but I’ve said it, and I’ve even presented about it, is asking whether an autism diagnosis is more beneficial than harmful. Very hard question to investigate.

When there is an actual purpose-built study that does that — obviously has a lot of flaws, but found that the probable autistics who weren’t diagnosed were better off. And in population-based studies like the one in Korea, the autistics who had not been diagnosed — did not have an existing diagnosis — were much better off. You often find that autistics who are diagnosed at older rather than earlier ages are better off. This is contrary to everything you’ve ever heard, probably, and so on.

I don’t think that those things are being taken seriously enough. We definitely are getting data from autistic people who are diagnosed as adults, who seem to be better off than what you find in longitudinal studies of autistics who were diagnosed as children. All very, very hard to interpret, but the question is raised, and the fact that it is out there.

I’m not trying to discourage or encourage anything or give advice in the least, but it speaks to my own concerns — maybe they’ve been futile — about standards, basic standards in research and practice, standards of science, ethics in medicine. And these are things I see as human rights issues. They are, to me, a big deal, and we should be very concerned that we are doing things to autistic people that we know produce bad outcomes in anyone.

COWEN: One of my striking takeaways from your work — and just to be clear, I don’t think you’ve ever said or written this in general terms — but just to see the extent to which parents do not necessarily have the actual interests of their children at heart, that parents seem to have a strong bias for children who will be a lot like they are. And they will take a lot of steps to try to make their children more like them, even if that, in some ways, harms the child from the point of view of the child itself.

I think a diagnosis is a kind of segregation, and in most other settings, we’re morally very reluctant to segregate. But somehow, when it comes to children, we seem really quite willing to segregate. This is a moral split, so to speak, possibly not always for any good reason.

DAWSON: First, I’m going to totally disagree about the parents. Keep in mind, in autism, you have very, very loud people — loud advocates, loud influential advocates. And those are autistics and nonautistics, professional organizations, people selling services, whatever — very loud advocates. They are not necessarily representative.

COWEN: But they are the most influential group, right?

DAWSON: Sure, but you can’t extrapolate from that to parents in general. You just absolutely can’t. The literature is different from what you get from the very loud advocates, and, of course, you have a lot of autistic parents of autistic offspring. This is still very poorly studied, but this definitely happens. The reproductive fitness, as they call it, of autistics has been wildly underestimated, I think.

This is, again, my own views, a nongenetics person. So, you have that, but I think you have to be very careful not to confuse advocacy and autism with any group in particular, like parents, or autistics for that matter, their views. I think that’s a real problem.

Now, the segregation thing. I have a massive bias against segregation. See, this may be one place where I’m — as an autistic, I have not just a bias, but I have a profound segregation deficit, and anybody who’s tried to segregate me is going to run into that. If you try to segregate me, I’m not going to go there. So, it should not be segregation.

DSM should not be segregation, DSM-5 autism, DSM autism is segregation. Why? Because the diagnosis segregates autistic people into lower standards of science and ethics than would be acceptable for anybody else. That’s what your diagnosis does.

DSM should not be segregation, DSM-5 autism, DSM autism is segregation. Why? Because the diagnosis segregates autistic people into lower standards of science and ethics than would be acceptable for anybody else. That’s what your diagnosis does.
Now, you may be lucky anyway. You may encounter extraordinary individuals who know you’re autistic and who will apply good standards of science and ethics to you. That certainly happens, but you must be lucky.

Now, you may be lucky anyway. You may encounter extraordinary individuals who know you’re autistic and who will apply good standards of science and ethics to you. That certainly happens, but you must be lucky. It will not be the more taken for granted, where the exception is that you’ll be denied these standards, and if you are, there will be recourse. In autism, that is not the case. So, to me, that is a segregation.

I can’t speak to the other diagnoses. I’ve looked very hard and have not found a situation like you have in autism, where you have these unacceptably low standards that are universally advocated for, and that there was no advocacy in the other direction, and the influence of advocacy is just completely determinative.

On autistics in the workforce

COWEN: As you know, there’s a Danish firm called Specialisterne, right? And they make a point of hiring autistic people to do things such as programming. Is that, in your view, an instance of possibly unjust or harmful segregation?

DAWSON: If you’re autistic and you want to be in the workforce in a way that is not very potentially harmful to you, you would be wise to seek out enterprises like this because they are . . . First off, obviously, they shouldn’t be necessary. Obviously, autistics should be able to go into the workforce like everybody else, be respected for their abilities. Not be discriminated against due to their job-unrelated atypicalities, job-irrelevant atypicalities.

That is not the case, so you have autistic people who, as employees, are being very much underappreciated and undervalued. So you have these organizations. And it’s massive. You have these huge efforts to get autistics employed under the umbrella of these. “We will test these autistics. We will hire them out to firms. We will get the firms used to having autistics around. We will coach people through things.” And so on.

This allows autistics to be in the workforce in a much less precarious way than they would if they simply went out in the world and then had to face this sort of problem of being denied basic rights and standards, which is just a very precarious situation. You may be lucky, but you would have to know it’s only luck. No matter how hard you work, no matter how good your work is, your situation is very precarious.

Whereas under them . . . You know what? I’m trying to think of the right term. You’re almost endorsed. You’re an endorsed autistic, or you have this stamp of approval. And that indicates to the employer that they should treat you in a certain way. That will improve your chances greatly and make you in a much less precarious situation. And yes, it is a form of segregation, of saying “Yes, this is what we should do to get autistics employed.” Now, it’s very hard to argue against that in the current situation.

COWEN: If you think of the United States, there has been such a thing as gay identity politics. Those groups have fought for the legalization of gay marriage, and they achieved that. I would say that’s a very good thing. It’s spread to many parts of the world, often before the United States. Should there be some kind of comparable identity politics for autistics, given that in some ways it has helped gay individuals?

DAWSON: Oh my goodness. So, no. No.

COWEN: Why not?

DAWSON: I’m not even sure . . . Well, first off, I’m going to say that what helped gay individuals is having human rights and the benefit and protection of basic standards in science and ethics. I think that’s what autistics are missing, not identity politics.

On the Michelle Dawson production function

COWEN: Michelle and I have a coauthored piece on Alan Turing. Maybe we’ll get to that next time. My last question is on what I call the Michelle Dawson production function. You know a great deal about autism. How is it that, given your history — you were not trained as an autism researcher in the formal sense — but how is it that you learn about autism?

DAWSON: I had to. It came out of my situation being terrible, so I was quite motivated. I had been told over and over again that I was too stupid — no other way to put it — to understand autism research. I tried other avenues. I was in a very difficult situation legally . . . everything, really.

Because I had to address these legal issues and questions, I did actually look at the autism literature, and suddenly I had information I could really work with. Suddenly there it was, this information that I was supposed to be too stupid to work with. And I can tell you that view continues onwards, [laughs] that I’m still too stupid to work with the literature. But what I’ve found is that the information I really needed was in these papers, so I read a lot, a lot of papers.

I read a lot of papers. I have not been taught how or what or where to read or anything. I read a lot of papers. I read associated information, like a trial registration, grant information, to get the information as complex and as useful as possible.

COWEN: Michelle Dawson. Thank you very much.

DAWSON: Thank you.

Like what you read? Give Mercatus Center a round of applause.

From a quick cheer to a standing ovation, clap to show how much you enjoyed this story.