Chapter 8
Sex in the Valley
Are hyper-systemizing couples, like those talented individuals who work in STEM fields, more likely to have an autistic child? Given that we now live in a world where individuals with good technology skills move to work in technology hubs, where they are more likely to meet each other, could these couples be contributing to the rising rates of autism? Increasing autism rates of course reflect greater awareness of autism, better diagnosis, and changing diagnostic criteria, but is there more to it than this?1 Answering this question is important both to test the central theory of this book—that the genes for autism drove the evolution of human invention—and to plan for how best to support autistic people.
To find out if hyper-systemizing couples are more likely to have an autistic child, we conducted the Parents’ Occupations Study, way back in 1997. It was the first large-scale survey of the occupations of parents of autistic children.2 We sent a questionnaire to 1,000 parents of autistic kids and to a control group, asking about the parents’ and the grandparents’ occupations. We predicted that fathers and grandfathers of autistic children would be more likely to work in the field of engineering, a clear example of a hyper-systemizing occupation. This was clearly confirmed: they were more than twice as likely to be engineers, compared to the fathers and grandfathers of kids with no diagnosis of autism or of kids with a different disability. And this was true of the grandfathers on both the father’s and the mother’s side. At that time, too few mothers were working outside the home for us to know what kinds of jobs they might have chosen, but a separate study found that mothers of autistic children score higher on tests of systemizing.3 So, genes from hyper-systemizing parents and grandparents seemed to be contributing to the likelihood of autism in a child or grandchild.
The Parents’ Occupations Study was retrospective: it started with a child’s known autism diagnosis and then worked backwards to see if the parents were hyper-systemizers. But could we find any prospective evidence? If you start with hyper-systemizing couples, are they more likely to have an autistic child?
I meet couples who have an autistic child all the time. Take couples like Jim Simons, the gifted mathematician and hedge fund entrepreneur, and his wife Marilyn Hawrys, a quantitative economist, who have an autistic daughter. Like many parents of an autistic child, Jim has used his hyper-systemizing to great advantage. He hired mathematicians and computer scientists to create computer models to predict the behavior of the financial markets. Jim’s personal wealth is estimated at over $15 billion.4 Or take Steve Shirley, who arrived penniless in England on the Kindertransport from Nazi Germany, studied math, and together with her husband Derek, a physicist, also had an autistic son. She found that, by separating software from hardware in the early days of computers when these only came as a bundle, she could create a business that made her a multimillionaire.5
Such couples are consistent with the idea of a genetic link between hyper-systemizing talent in parents and a higher likelihood that they will have an autistic child. Anecdotally, of the 330 wealthiest families in the United States, 27 (8 percent) have an autistic child.6 Given that autism is found in 1 to 2 percent of the general population, this suggests a fourfold increase in the rate of autism in families in which one or both parents have been hugely successful at making money. Making money is usually the result of running a business, and a business is a system, so a successful business man or woman is likely to be a strong systemizer.
To see how key if-and-then thinking is to business, consider this example: “if I make $50 selling one unit, and I scale up production to one million units, then I will make $50 million.” A business, in taking something and turning it into a product for sale, is itself also a system, and if the if-and-then process is a new business pipeline, it would meet the criterion of being an invention.
So, a hyper-systemizing brain type, which is a massive asset when it comes to invention and wealth creation, could be linked to a greater likelihood of producing an autistic child. Anytime one comes across a highly successful entrepreneur—someone who has clearly systemized their business and who also, having the technical knowledge central to their business product, has therefore systemized their product—they might be expected to be more likely to have an autistic child or grandchild.
Of course, hyper-systemizing in a parent or grandparent may not just manifest as financial wealth derived from business acumen: it could also show itself as scientific, academic, technical, literary, or musical expertise. Consider the remarkably successful physicist Stephen Hawking, who has an autistic grandchild.7 Or Elon Musk, perhaps the world’s most famous innovator and inventor, who has an autistic child.8 These anecdotes hint that hyper-systemizing grandparents and parents are more likely to have an autistic child or grandchild. But to move beyond anecdote to evidence, to test if this link is genetic and not due to chance, we need to look at rates of autism in a large population of hyper-systemizing parents.
I had been hearing anecdotal reports that autism was much more common in Silicon Valley since at least 1997. I had also read an article anecdotally suggesting that autism was more common among children of alumni of the Massachusetts Institute of Technology (MIT). If these rumors were backed up by data, they could confirm the genetic link between hyper-systemizing and autism.
And then, in 2003, I received an interesting email from the former president of the MIT Alumni Association, Brian Hughes, who told me that among his alumni, autism rates were anecdotally reported as high as 10 percent, not the usual 1 to 2 percent. If the rates of autism really were five-fold higher in this unusual population, this would be super-important. With Brian and psychologist Sally Wheelwright, we set up the MIT Study, a survey that would go out to thousands of MIT alumni.
MIT was an interesting place to choose to stage the largest experiment on the mating patterns of systemizers because, before 1975, MIT only admitted men to study at the university, and the only courses on offer were in the “exact” sciences (STEM). So, if these men later ended up having autistic children, we could assume that, at a minimum, those children had a father who was talented in STEM. In 1975, MIT went coeducational, giving us the opportunity to study couples who had met there after this date and who had started a family. These would be couples in which both parents were talented in STEM, so we could see if they were more likely to have an autistic child. We set out to compare the rates of autism among the children of couples in which both parents were in STEM, one parent was in STEM, and neither parent was in STEM. (And of course there are lots of couples in the general population where neither parent is talented in STEM, effectively the control group.) The study received formal approval by the MIT Institutional Review Board, so we were good to go.9
And then another shocking email from Brian arrived in my inbox. He told me that the president of MIT, Charles Vest, had intervened to say that he wasn’t willing to authorize this study to go ahead. Soon after, Vest retired but Brian still was told the study couldn’t go ahead for fear that, if the hypothesis was confirmed, the reputation of MIT might be harmed. The study had been blocked, from the highest level in the university.
I was shocked for two reasons. First, scientifically, this was a really important question to answer, to help understand the link between autism and systemizing talent. And secondly, MIT’s decision challenged the important governing principle in universities of academic freedom of ideas.10 I have worked in universities for thirty-five years, and I have never ever heard of the president of a university—anywhere in the world—dictating what research could or couldn’t be conducted within their university, unless the research contravened legislation or might have been morally offensive. Other than these kinds of reasonable limits on academic freedom, the principle is inviolable. Nevertheless, I had no wish to embarrass MIT, so I had to drop the plan. But I was not going to just abandon my attempt to test this important hypothesis. I would simply have to find a different location for the study.
Then one day in 2010 my luck changed. I received another email, this time from Patrick Wiercx, a Dutch journalist from the city of Eindhoven in the Netherlands. He was writing a story about the anecdotal reports of very high rates of autism in Eindhoven, the Silicon Valley of the Netherlands. I invited Patrick to come over to discuss this idea, and he jumped on a plane straightaway. He explained that Eindhoven was a hub for hyper-systemizers because of two big magnets that drew them: the Eindhoven Institute of Technology (another MIT) and the Philips factory, which has been there for over a hundred years. We decided to work together.
By good fortune, I had a Dutch master’s student, Martine Roelfsema, who could help us navigate the cultural challenges of such a study. With a team of epidemiologists, we designed the Eindhoven Study to see how many autistic children there were in Eindhoven compared to two other Dutch cities, Utrecht and Haarlem, which had similar-sized populations and were comparable on other demographics, but which, unlike Eindhoven, were not STEM hubs.11
Martine contacted every primary and secondary school in all three cities—over 650 schools—to ask them how many children were already on their special educational needs register as having a diagnosis of autism. More than half of these schools agreed to take part, providing information on over 60,000 children. When the results came in, we were astonished—they exactly fitted our predictions. In Eindhoven, there were 229 cases of autism per 10,000 children, compared to just 84 in Haarlem and 57 in Utrecht. So, autism was more than twice as common in Eindhoven.
From this, we concluded that people talented in STEM are more likely than people in the wider population to have an autistic child, and in such communities we should see a spike in the prevalence of autism. This needs to be replicated in other STEM hubs like Silicon Valley itself and may have relevance for understanding the exponential rise in the rate of autism in the digital age, although of course many factors might contribute to this.12 But the evidence from Eindhoven strongly suggests that inheriting a hyper-systemizing mind makes a child more likely to be autistic.
Let’s return to the results of the Parents’ Occupations Study, which showed that, in a family with an autistic child, both the maternal and paternal grandfathers were more likely to work in engineering. This presents a number of puzzles.
First, what attracts these men and women to each other in the first place? When they met, they didn’t know they each had a parent who was more likely to be an engineer. And a related puzzle is this: How do two individuals carrying hyper-systemizing genes end up as a couple? If it is just physical attraction, why should they have similar minds? If it is mental attraction (called sapiophilia), again, why should they have similar minds? I was intrigued by what makes two hyper-systemizers more likely to end up as a couple.
“Like marrying like” is what biologists call “assortative mating.”13 Assortative mating is widespread in nature. For example, in humans, tall people are more likely to marry tall people, extraverts are more likely to marry extraverts, and alcoholics are more likely to marry alcoholics. Based on our Parents’ Occupations Study, it seemed that hyper-systemizers were more likely to marry hyper-systemizers. This fits the “assortative mating” theory of autism, which predicts that autism rates will be higher among hyper-systemizing couples’ children or grandchildren. Interestingly, this has recently been confirmed at the genetic level.14
There are a few ways in which assortative mating among hyper-systemizers might happen. It could be that people who share similar interests are more likely to end up living in the same location, because they have pursued comparable paths in their education and career choices. Or perhaps hyper-systemizers are more likely to become a couple because more socially skilled individuals form couples sooner, and less socially skilled individuals are the ones who, years later, are still available and looking for a mate.15 They find each other because they’re both still single. A third possibility is that assortative mating among hyper-systemizers might be the result of two people being attracted to each other because their minds are similar.16 But how could we test this?
We invited mothers and fathers of autistic children to take a systemizing test. We used the Embedded Figures Test, a pattern recognition test that involves finding a target piece hidden within a design. We had previously given this test to autistic adults, who are faster and more accurate at it than are typical adults, probably because, being hyper-systemizers, they break down information into its component parts as rapidly as possible, to then look for if-and-then patterns.
Figure 8.1. The Embedded Figures Test. Autistic people and their parents are on average faster than typical people at finding the target (the cube below) hidden in the design (above).
Sure enough, we found that both the mothers and the fathers of autistic children were also faster and more accurate on this pattern recognition test.17 They shared a skill with each other, related to hyper-systemizing, that most people don’t have. Of course, they were totally unaware of their shared skill on this test when they began dating and later started a family, but one possibility is that each was attracted to the other’s detail-oriented, laser-sharp mind, or at least wasn’t deterred by it. Perhaps they felt comfortable with a person whose mind was just like theirs, and just like the mind of one of their parents.
Eindhoven was a “natural” experiment that allowed us to observe what happens when you speed up evolution to see how the children of hyper-systemizers develop. And the Eindhoven Study again provides evidence consistent with the idea that the genes for systemizing overlap with the genes for autism. We can therefore assume that the hyper-systemizers who drove the evolution of human invention, starting 70,000 to 100,000 years ago, themselves had a high level of autistic traits and were more likely to have a child with a high number of autistic traits. We can assume this because it is still the case today. But back then, the likelihood of two hyper-systemizers having children was much lower. Today, as more Eindhovens and Silicon Valleys take root and blossom in every country, bringing hyper-systemizers together who make families, what might we expect the future to hold?