Childhood Autism and Assortative Mating

This paper (pdf) is from Hays Golden, who is currently on the economics job market from Chicago:

Diagnosed rates of autism spectrum disorders have grown tremendously over the last few decades. I find that assortative mating may have meaningfully contributed to the rise. I develop a general model of genes and assortative mating which shows that small changes in sorting could have large impacts on the extremes of genetic distributions. I apply my theory to autism, which I model as the extreme right tail of a genetic formal thinking ability distribution (systemizing). Using large sample data from the Centers for Disease Control and Prevention, I find strong support for theories that autism is connected to systemizing. My mating model shows that increases in the returns to systemizing, particularly for women, can contribute significantly to rising autism rates. I provide evidence that mating on systemizing has actually shifted, and conclude with a rough calculation suggesting that despite the increase in autism, increased sorting on systemizing has been socially beneficial.

This is an important paper, though I would stress the generality of the result; autism and systematizing may or may not be the best applications.  If you are in some way genetically “extreme,” and suddenly better at finding/pairing with similar extremists, the numbers of that type in a population can rise relatively rapidly.  We now have a very clear and useful model of how that works.  One way to interpret this is to believe that the internet will, over time, increase human genetic diversity.


Being a first adopter of computer dating services and the "nerd" culture might make this the best application.

E Yudkowsky, come on down.

Err, isn't this simply Simon Baron Cohen's thesis?

Probably, but did Baron Cohen build the model and check economic data?

"Do you have a flag?"

Uh, why is specialization in a match better than diversification? The specialization of couple may be socially beneficial but privately disastrous. I have started looking around for a mentor with my newly acquired brain label but the last thing I need is another me around the house. I suspect that the improvement in diagnostic tools and public views on the functionality of persons on the autism spectrum has done the most to expand the prevalence of autism than any behavior. Autism is almost a cool kid condition in economics and that gives me some pause. The march from extreme to trendy to mundane can be fast and furious. That said, I am all for a respect of diverse cognitive styles and recognizing that your braining wiring is an important consideration in your match quality (in many markets).

Maybe I don't see the obvious, but I didn't understand why assortive mating based on autism is concluded to be socially beneficial. Are increasing autism rates socially beneficial?

It's not based on autism it's based on systematizing. Increased systematizers is good, autism is a bad side effect.

Well, first off: I would guess that increasing the number of autistics in our society would be extremely beneficial at least up to some point quite a lot higher than it is now. In general, autistics and non-autistics have highly complementary skill sets, and in particular I note that autistics are highly resistant to a number of systematic flaws in the non-autistic human brain. Having one person in a group who can say "guys, no, you're having halo effects, you don't have an actual basis for that" can be of immense value -- assuming that people know to listen. Similarly, in a group of autistics, a single person who can point out that something is going to hurt people's feelings can be of immense value -- assuming people know to listen.

Secondly, if it were a side effect of strong selection for other traits which are related, that'd work too. (Think sickle-cell anemia and malaria.)

I am truly shocked by the number of people who seem to think autistics are mostly perfectly functional human beings. They are largely not. They need tons of care and support to survive.

Yes, this is true. I went to high school in the SF bay area (lots of tech companies with systematizer employees), and one of the most common ways high schoolers would get their community service hours was by volunteering for this autism treatment program: I volunteered for this organization, and yes, it's a lot more demanding than traditional childcare.

I am the truly blessed parent of a "mostly perfectly functional" son with an autistic spectrum disorder. I know five families of other autistic children very well - socially, not because I met them because their kids have autism. Two of them are profoundly autistic and can never be left without care, one is less functional than my own son, but will be able to function on his own eventually and two are "mostly perfectly functional". I like that description, by the way, because my son is perfectly functional at least 98% of the time. He doesn't like to socialize, but he does well at school and I don't force him to otherwise. However, when he gets frustrated or angry, the ensuing meltdown can be destructive and frightening. There are days that he just isn't capable of going to school or that he needs to come home early. Those days are fewer and fewer the older he gets and I praise God for that. However, I know that I and my friends are blessed because I know the parents who are not as blessed. I try to be there to provide care when they need help because I know how and that's the best friend I can be to them, but even in my own circle where half of the autistic kids are "mostly perfectly functional", half of them are not. Autism is a spectrum. People in general don't see the lowest 75% or so of that spectrum because they can't function at all in public (too many stimuli) and that is sad, but true.

[citation needed]

I'm autistic. The vast majority of my friends are also autistic. All of us have held jobs. Some have college degrees. And, perhaps most importantly: If you look at the ways in which we've failed, they are nearly always not a result of an intrinsic inability to function, but rather of being in circumstances, or near people, who were hostile and/or unwilling to make even minor accommodations.

I've been autistic for my entire life. I only got diagnosed about 35 years in, because for the most part I can get by -- although now that we know, we know what went wrong in any of a number of major mishaps that could have, with slightly worse luck, resulted in me being unable to function in the workforce.

It is quite likely true that the majority of people who had been diagnosed as autistic by 2005 or so were people who were unable to function well enough to get by in society -- but that is tied to the tendency for people not to get diagnosed unless something is going significantly wrong. As educators and parents get more educated, the chances of healthy and functional autistics also getting diagnosed increase.

There is also the secondary issue, which is that it's easy to render people significantly less functional by mistreating them, and a lot of things other people take for granted are mistreatment for autistics. (And a lot of things that I find pleasant and congenial would be mistreatment for non-autistics, too.)

It's certainly true that most people don't meet the autistics who can't function -- but it's also the case that large numbers of autistics who function reasonably well are never diagnosed. A friend of mine just got a diagnosis in her 50s. She's been holding down jobs, getting things done, and generally surviving this whole time. Her life would have been a lot easier if she'd known why some things were harder than expected, but until a couple of months ago, a hypothetical list of "autistic people in the US" wouldn't have included her, and thus would have left out at least one person who had held down jobs for her entire adult life.

I know some people who do have more severe meltdowns. And you know what? I have found that many of them are fine if you just put a little thought into what they can handle and how they function well, and let them go about things that way. I don't think that's necessarily a big problem if we can get people used to the idea that basic human decency involves some amount of awareness of the people you're dealing with.

Specialization of the couple may be privately beneficial if there are significant complements in consumptions?

At least that is an answer that I remember seeing put forth somewhere. That whole companionate marriage thing. And anyway, people can be alike in one trait, say conscientiousness, and differ in others.

And I thought intermarriage would make us less diverse, over a very long time.

Intermarriage across what lines? Assortative mating decreases intermarriage across the sorting criteria, by definition.

Golden is starting from the assumption that autism rates actually have increased. That may be the case, but it's also possible that many of the children being diagnosed as autistic would not have been so diagnosed a decade ago.

Very much so - this column claims that the entire increase in autism-spectrum disorder diagnoses may be caused by diagnostic substitution.

This is undeniably the case to an extent, but it is also plausible that an actual increase in "full-blown" autism or a shift of the personality spectrum towards that tail of the distribution have increased awareness.

My question has been have we seen the same increase in other disorders?

Also, a while ago people claimed you couldn't see a genetic shift so quickly. Now we accept that premise implicitly through the mechanism of assortative mating.

We need a careful longitudinal study of this important question. It was pointed out about a decade ago that there were high rates of autism among children in Silicon Valley, which supports this theory, but there are major methodological problems involving counting autistic children.

Maybe Silicon Valley just correlates with the sort of parents that are more likely to be able to afford autism testing?

This is what makes me wonder if we are going to seamlessly transition from the "there is no autism epidemic" to the "of course it's increasing due to assortative mating, older parents, etc." And you guys wonder think I am always insanely crazed because of narcissism. No, it's because noone is aware of anything until they think it is self-evident!

Alternately, maybe the real count of autisms is going down, but there are countervailing forces at work, assortive mating acting to raise the frequency and the lower concentration of pollutants acting to lower the frequency? Silicon Valley is suspected of having high concentrations of toxins like lead from the various manufacturing activities that have gone on there over the years. (I have just heard people say this; I have no real evidence.)

Anyway, my point is that assortive mating could act to increase autism while the count of autisms is going up, going down, or remaining stable.

It's just that noone really knows. And people literally use the proliferation of hypotheses as de facto evidence that those hypotheses are unbelievable. If it's just an extreme of the bell curve of personality then it's probably a lot of little things.

We've spent 5 years trying to pin it on increased diagnosis because 'nerds' have the resources, connections, and information to obtain diagnosis and services, but that is just another hypothesis and it does not preclude an actual increase.

And if there is something like assortative mating placing people further on the spectrum, then it will take less environmental impact to further impact an epigenetic component.

Autism testing, at least here in NC, costs nothing.

I've always thought that the evidence for genetic causation was overwhelming, even when I first looked into it in 1961. It was just that there were political reasons inside the psychiatric community, with reinforcement from outside it, for assuming otherwise right into the 1980s. A professional Psychologist I spoke with about my Asperger Syndrome said that when she was doing her graduate work in Autism treatment in the late 1980s, there was still no mention of the genetic causation now so noticeable. By now there are upwards of 25 identified genes associated with one or another ASD trait.

As to assortative mating, it is fairly obvious here in the "Silicon Forest" west of Portland. Combine the genes of people with different mixes of those 25 genes, and a far higher number of autistic traits can be expressed in the next generation. Even those of us at the highly functional end of the Spectrum are often not, though not always, employable. What is of interest is that a mechanism exists whereby Aspies could grow common enough that the relatively mild mechanism of assortative mating would lead to the next generation having far higher rates of autism farther down the Spectrum.

Aspie-level autism was once far less common as well. During the Little Ice Age there were far fewer places in village life for those whose lack of communication skills did not let them "fit in". Villagers lived too close to the edge, and a mistaken communication in spring planting could mean another 5-10 people from the village died in the "hungry times", between the first of January and the middle of March. Young villagers causing such incidents might easily be found dead. Check Parish records for unexplained deaths like that. It's surprising how many times that happened. Not too surprising, either, that it was mostly males.

However, as the industrial revolution opened up niches for people who could concentrate in nearly obssessive detail on topics for hours, Aspies seem to have won enough acceptance inside cities to survive more often to reproductive age. Enough could meet, and marry, that the addition of their ASD gene complements to ther spouses' began to show up in Germany as what we have called autism. From that time onwards, assortative mating had fertile territory once the computer revolution took hold, and many couples met in their Silicon Forest cubicles.

Epigenetics? Maybe. The change in the reactions of the society around them, certainly, IMHO.

BTW, the effort needed to help higher functioning Aspies will be far less once schools start programs among their staff to make it clear the staff *does*not* give permission for violence against people on the Spectrum. In Vancouver, Wash. the schools in the last decade needed a 3 years long staff "education" program to get this "lack of permission", but their next 5 years showed not one assault on people because of their ASD. While those farther down the Spectrum will still have overwhelming sensory issues, I strongly suspect that a lack of fear will help them strongly. I know it would have changed my life in the Vancouver Public Schools completely, especially if family had allowed me to advance in classes at my intellectual rate instead of strictly by age.

Maybe autism is like sickle cell anemia. If both of your alleles are for sickle cell anemia, you have the disorder, which is bad. If you have it in only one of your alleles, you don't have the disorder and you do have significantly increased resistance to malaria, which is good in the part of Africa where it comes from. If there is an autism allele, having one copy of it may confer a useful trait here in the valley.

I am skeptical only because autism has been linked to literally everything that has ever ever ever been created or will be created.

Whatever the merits of the paper, "Golden Hays" is a great name.

Glad to see assortative mating get some love, as assortative mating would have important implications for income equality and SES mobility. However, through a quick scan I couldn't tell to what degree parental age was controlled for in this paper.

The people who are predisposed toward "genetic formal thinking ability" would presumably use more of their younger years on their careers and getting degrees. Both higher paternal and maternal age have been linked to autism.

Probably tangential to your point; but aren't increased parental ages correlated in general with a lot of different genetic disorders? What's the mechanism behind that, anyone know?

Greg Cochran & Henry Harpending have been doing a series of posts on the subject (example). You accumulate random mutations with age.


OTOH, the older you are shouldn't that mean nature has had more of a chance to have weeded you out if you had some bad genes? Wasn't sure if that effect exists and if it does older parents ought to have "better" kids?

Yes, spermatogonium in the testes constantly replicate themselves to provide enough primary spermatocytes that will eventually be sperm. The replication process involves a duplication of the cell's genome before splitting into two cells (mitosis). This replication process is not an exact process and introduces errors into the genome. I think the number is about 1 in 50 million bases but this number is always getting revised by scientists. Anyway, since there are 3 billion DNA bases, and replication of these stem cells happen many, many times in a year, a 40 year old male will be producing sperm with more errors in the genome of the sperm than when he was 20. Most of these errors are in unused stretches of the genome, but some may be in an important gene, and the chances increase over time as I just described.
There are other reasons but this is the big one I think. Females don't have this problem as much because their eggs are in a form of stasis and protected deep in their body. Since females can only get pregnant a set number of times in their lives (~25 year fertility window/9 months of pregnancy) they only need a low number of eggs. They are essentially born with as many as they need.

Interesting. In which case only paternal age ought to be correlated to such disorders but they ought to be mostly independent of maternal age.

Wonder if that's empirically observed.

Women use their best eggs first. Also pregnancy is much harder on an older mother's body. But there was a recent finding that paternal age is a big risk factor for autism where there is not a hereditary component.

OTOH, he eggs hang around, so they can accumulate damage. The sperm replicate so they can accumulate replication errors. So, I suspect there will be different genetic defects associated with maternal or paternally contributed genes.

This particular subject concerns whether sub-clinical autistic features actually confer a mating advantage.

Indeed, there are studies which claim a correlation between autism and the age of the father at conception. Men over the age of 42 are supposedly more likely to sire autistic children. The age of the mother does not appear to have any correlation.

The diagnosis of autism does seem to cluster in geographic areas and specific time periods. The incidence appears to have increased greatly over the last fifty years.

It is difficult to determine the validity of such correlations as an increasingly large variety of different neurological development problems are diagnosed as "autism". There is a lot of noise in the data.

Maternal age is associated with chromosomal abnormalities, which means miscarriage or Downs Syndrome. There are claims that maternal and paternal age are associated with other diseases, including both miscarriage/Downs and diseases whose genetics are not understood, but the effects are much smaller and probably genetic, or even real.

Paternal age contributes to new mutations, but at a very small rate. This matters not for an individual, but only for a society, where it shifts the equilibrium mutational load.

I mean that the specific claimed paternal age effects are not genetic and probably not even real.

Douglas, our two posts seem to contradict each other. :) I have been out of the research world for a while so you are probably right. I did forget about chromosomal dislocations during the crossing over events associated with meiosis. Ahh, brain cells too fail with age it seems.

Your qualitative description is completely correct.

Wonks Anonymous gives a link to a blog post that cites a very recent measurement of the human mutation rates: 15 from the mother and a variable number from the father, larger than 15, but not 10x larger. Nowhere near 1 per 50 million bases.

You're wrong.

OK guys. How many generations of more assortative breeding before we have a new species?

His dad wrote "Memoirs of a Geisha"

A good read.

Systemizing Paradox: As the ability to find systemizing partners increases, the costs to abandoning them also decreases. The sum total of these two effects may cancel out.

I wonder if the existence of different 'races' is the nature's way to push us to assortative mating?

In 2007 a survey of Autism Spectrum Disorder (ASD) was carried out for the British National Health Service, using a fixed diagnostic tool:
"Autism Spectrum Disorders in adults living in households throughout England", Brugha, et. al. (available on-line). The study found that
the rate of ASD was about 1% for young, middle aged, and old. This suggests that the rate of ASD has not changed significantly.

Asssortive mating!!! How does this theory help explain the considerable number of children with autism found to be suffering from gut and immune system problems and who present with toxins and a host of other health issues?

Unless you have numbers saying that the autistic children suffering from X Y and Z are a greater or lesser fraction of autistic children, it's irrelevant. There are always insults in the environment, both before sorting took place and afterwards.

The same thing as with many other immune system disorders, and formerly ectodermal tissue, Gillian. There is a strong correlation of Type 1 Diabetes (insulin dependent) with inheritance, and it is an immune system disorder. There is no reason that other ectodermal affecting mutations would not also affect the immune system. Likewise, it is not at all odd that a mutation that affects one ectodermal-descended tissue, nerves and brain, will also affect another ectodermal-descended tissue, the gut. Most people do not remember their embryology, but the embryonic alimentary canal forms from the folding of what becomes also the ectodermal layer in the human embryo around a central cavity, which fold becomes the alimentary canal when completed. This is called "gastrulation".

I wonder if there is any connection to the timing and pervasive use of hormonal anticonception.

Really interesting question, Eduardo. The Pill in all its glory... Are you up on recent discussions about the placebos included with monthly cycles of the Pill, and how non-normal it is for women to expedience monthly periods so flipping often? Women on the Pill undergo a hormonal whipsawing unlike anything their systems were ever meant to withstand; it'll be very interesting to see the results of it, if any.

But as this thread indicates, there's still Question 1: is there actually a higher incidence or clustering of autism at all?

Comments for this post are closed