THE ASSORTATIVE MATING THEORY

Conversation : MIND

THE ASSORTATIVE MATING THEORY

A Talk With Simon Baron-Cohen [4.5.05]

My thesis with regard to sex differences is quite moderate, in that I do not discount environmental factors; I'm just saying, don't forget about biology. To me that sounds very moderate. But for some people in the field of gender studies, even that is too extreme. They want it to be all environment and no biology. You can understand that politically that was an important position in the 1960s, in an effort to try to change society. But is it a true description, scientifically, of what goes on? It's time to distinguish politics and science, and just look at the evidence.

Introduction by John Brockman

Simon Baron-Cohen is Professor of Developmental Psychopathology and Director of the Autism Research Centre at Cambridge University. In this Edge feature, he presents his new Assortative Mating Theory which connects his two fields of research: the characteristics of autism in terms of understanding what's going on in the brain and the causes of the condition; and understanding the differences between males and females.

"My new theory is that it's not just a genetic condition," he says, "but it might be the result of two particular types of parents, who are both contributing genes. This might be controversially received. This is because there are a number of different theories out there — one of which is an environmental theory, such as autism being caused by vaccine damage — the MMR vaccine (the measles, mumps, and rubella combination vaccine). Another environmental theory is that autism is due to toxic levels of mercury building up in the child's brain. But the genetic theory has a lot of evidence, and what we are now testing is that if two "systemizers" have a child, this will increase the risk of the child having autism. That's it in a nutshell.

Baron-Cohen realizes that his theory might raise anxieties. "Just because it's potentially controversial," he says, "doesn't mean that we shouldn't investigate it. And there are ways that you can investigate it empirically."

He also expects controversy. Given the continuing public discussing in the US about innate sex differences, he will, no doubt, be challenged when he says "It was interesting for me to discover that there's been a sleight of hand, mostly in the States, such that the word 'sex' has been replaced by the word 'gender'. Baron-Cohen believes that it's time "to distinguish politics and science, and just look at the evidence". Others will feel differently.

— JB

SIMON BARON-COHEN is Professor of Developmental Psychopathology and Director of the Autism Research Centre at Cambridge University. He is also a Fellow of Trinity College, Cambridge. His books include Mindblindness; and The Essential Difference: The Truth about the Male and Female Brain.

SIMON BARON-COHEN's Edge Bio Page

THE REALITY CLUB: Marc D. Hauser, Steven Pinker, Armand Leroi, Carole Hooven, Elizabeth Spelke, Alison Gopnik, David C. Geary, Helena Cronin, Linda S. Gottfredson. NEW Baron-Cohen responds.

THE ASSORTATIVE MATING THEORY

[BARON COHEN:] I've been working on the question of autism, trying to understand what characterizes autism from a psychological perspective and ultimately aiming to understand what's going on in the brain and the causes of the condition. My new theory is that it's not just a genetic condition, but it might be the result of two particular types of parents, who are both contributing genes. This might be controversially received. This is because there are a number of different theories out there — one of which is an environmental theory, such as autism being caused by vaccine damage — the MMR vaccine (the measles, mumps, and rubella combination vaccine). Another environmental theory is that autism is due to toxic levels of mercury building up in the child's brain. But the genetic theory has a lot of evidence, and what we are now testing is that if two "systemizers" have a child, this will increase the risk of the child having autism. That's it in a nutshell.

A systemizer is somebody whose style of thinking is predominantly in terms of understanding things according to rules or laws. You can think of lots of different kinds of systems: mathematical systems (algebra, computer programs), or mechanical systems (computers or cars); natural systems (weather, or rocks, geology); and social systems (businesses, or the military).

In each case, when you systemize what you do, you try to understand the system in terms of the laws that govern the system. Economics would be an example of a system, where people are trying to predict a crash, or predict what's going to happen in terms of stock markets. They are trying to understand things according to laws or rules. The theory we are testing is that if you have a mother and a father who are both systemizers, the risk of the child having autism increases.

Systemizing is expressed in behavior, so, for example, if your hobby is playing with computers, that's the behavior that you see. But obviously such activity reflects your interests, which is what's going on in your mind, not just in your behavior. The mind of a systemizer is drawn to understand systems. Steven Pinker has a nice phrase about spiders, that spiders are just programmed to spin webs. He uses that as an analogy for the way in which a typically developing child is programmed to learn language. These programs are not a hundred percent deterministic; you can intervene, you can change. There's obviously plasticity in the system. In the same way, systemizing isn't going to turn out to be a hundred percent genetic. There are few if any behavioral characteristics in humans that are one hundred percent genetic .

There are five steps for testing this theory. First, we need to establish whether or not systemizing runs in families. Secondly, we need to find out if there are there any genes associated with systemizing. Thirdly, are the parents of children with autism systemizers, defined according to their cognition? Fourthly, do they both carry the genes for systemizing? Finally, when these genes combine, does this raise the risk of their child having autism?

This theory will be controversial, and it might raise anxieties. But just because it's potentially controversial doesn't mean that we shouldn't investigate it. And there are ways that you can investigate it empirically.

How would you investigate it? Well first thing is to look at families where there's already an autistic child, and look at the parent directly. We've already conducted some of those studies, and found that whereas in the general population systemizing is more common among males, in the case of parents of a child with autism, the mother of such a child is also very likely to be a systemizer, with male-typical interests.

One example of how we test this is to give them a task where you have to analyze a visual pattern as quickly as you can to find a component part. In the general population males are quicker at this kind of analytic task, but in the case of parents of children with autism, the mothers are just as fast as typical males. The mothers are showing a typical male profile, and that's counter-intuitive since you would expect them to be showing a more typical female profile. That's just one clue that this theory is worth exploring.

A second clue is that we've looked at the rates of engineering in both fathers and the grandfathers of children with autism. Engineering is an occupation where you have to be a good systemizer, for example, understanding mechanical systems. We found that fathers of children with autism are over-represented in the field of engineering. And what was interesting was that we found exactly that same pattern in the grandparents too.

You start with the child with autism; he or she is the end result of this experiment of nature. And you work backwards to see if there were there clues in the previous generation — or previous two generations. This new theory is called "the assortative mating theory", The clue that both sides of the family are contributing similar genes is that in our study of occupations, grandfathers on the maternal and the paternal sides were both more likely to be working in the field of engineering. So the strong systemizing wasn't coming down just one side of the family. It's called assortative mating because it describes the idea that two individuals might end up in a union because of having similar characteristics. They're selecting each other on the basis of having similar characteristics.

The assortative mating theory connects with the field of sex differences — my other big area of interest . I've been trying to understand the differences between males and females. It was interesting for me to discover that there's been a sleight of hand, mostly in the States, such that the word 'sex' has been replaced by the word 'gender'.

This has happened in a very subtle way over the last century, so that in the States, nobody talks about sex differences; they talk about gender differences. Whenever you want to refer to somebody's sex you refer to their gender. I call it a sleight of hand, because actually 'sex' is the older word. Your sex is either male or female, and in biology your sex is defined by whether you have 2 X chromosomes or an X and Y chromosome. There's been a subtle shift into talking about gender, to whitewash the word sex.

Why has this happened? Presumably, because your sex is determined by your chromosomes. And in the States the ideology is that we shouldn't be determined by anything; we should be able to be anything we choose. The blank slate. Gender refers to how you think of yourself: as masculine, or feminine, It's much more subjective, and is commonly believed to be culturally constructed. Italian male gender behavior is expressed differently from English male gender behavior. This gives the impression that people's gender behaviour can change as they change culture, even if their biological sex is fixed.

Talking about gender is therefore much more optimistic than talking about sex. It's the rags to riches idea — you can become anything. But I've been very interested to go back to the original notion of sex, as a biological characteristic, and to ask if there are there any essential differences between males and females in the mind. And to understand that if there are psychological differences, what are the biological mechanisms that give rise to these? Are they genes, are they hormones?

In our own work, we have been focused very much on fetal testosterone — the hormone that the fetus is producing in the womb, to see whether that has any effect on later behavior. We had a perfectly good word, which was sex. But it's become almost a profane word in the U.S. I recently wrote a journal article on sex differences in the mind. Everywhere I'd written the word sex, the copy editor changed it to the word gender. A systematic change had been introduced, and I asked that the original word be used. The editors asked me to give them a good reason, because they explained in the States the preferred word is gender. I had to explain, a person's gender is different to their sex. It's a distinction that seems to have got lost. It's hard to know whether it was deliberate, or whether it just happened without anybody noticing.

Back to hormones. We've been conducting laboratory studies on the amniotic fluid in the womb — the fetus is effectively swimming in this amniotic fluid. We analyze how much testosterone, the so-called male hormone, is in the amniotic fluid. It's not actually a male hormone, because both sexes produce it, it's just that males produce a lot more than females. That's because it comes from the testes. Females also produce it in the adrenal glands. And even within the boys, or within the girls, you see individual differences in how much is produced.

The question is, does this translate into anything psychological if you follow up those children? We measured the amniotic fluid testosterone, then waited until the baby was born, and then looked at the baby's at 12 months old, 18 months old, two years old. It's a longitudinal prospective study.

What we found is that the higher the baby's level of fetal testosterone, the less eye contact the child makes at 12 months old. And also the slower they are to develop language at 18 months old. To me these are really fascinating results, because we're looking at something biological, in this case a hormone which presumably is influencing brain development to produce these quite marked differences in behavior. We always knew that girls talked earlier than boys — that there is sex difference in language development — and we also knew that there's huge variability at 18 months: some kids have no words at all, and other kids have huge vocabularies, about 600 words. No one's really been able to explain this variability. Why should one kid be almost mute and another kid be very verbal?

People have identified some factors, such as that first-born children talk earlier than later-born children. Obviously there are environmental factors that are relevant. Presumably that's because first-born children get much more attention from their parents. But over and above your birth order, it looks like hormones also explain some of the variability. We've now followed up these kids into school, they're four years old, and we're still finding that the prenatal hormone production levels are influencing behavior in middle childhood. This is just one example of why we shouldn't ignore biology in explaining differences in how the mind works.

I don't argue it's all biology. But for a long time social behavior and language development were seen as purely environmental or learned experience. These hormone studies suggest hormones are also part of the explanation. We also know from medical conditions that if, for example, for genetic reasons you have an overproduction of testosterone, this condition can change your behavior. So if you look at girls with a condition called Congenital Adrenal Hyperplasia (CAH), where they are producing too much testosterone for genetic reasons, they look like tomboys. Their interests are very male-typical interests; they like playing with little toy cars, they like building tree houses, and they perform very quickly on spatial tests, unlike typical girls.

The evidence for my assertions comes from experiments. And in all of these areas I'm not interested in beliefs without evidence.

One experiment we conducted here in Cambridge was at the local maternity hospital. Essentially we wanted to find out whether sex differences that you observe later in life could be traced back to birth, to see if such differences are present at birth. In this experiment we looked at just over one hundred newborn babies, 24 hours old, which was the youngest we could see them, and we presented each baby with a human face to look at, and then a mechanical mobile suspended above the crib. Each baby got to see both objects.

Obviously these objects are different in interesting ways, because the human face is alive, and it can express emotion, it's a natural object. The mechanical mobile is man-made, it's not alive, and obviously it doesn't have emotions. We tried to make the two objects equivalent in some important ways. One is that they were both the same size; another was that they were a similar colour, in order to try and control features that might be grabbing the child's attention. But effectively what we did was film how long each baby looked at each of these two objects.

We asked the mothers not to tell us the sex of their babies, so that we could remain blind to whether this was a boy or a girl. And for the most part that was possible. Sometimes it was possible to guess that this was a boy or a girl, because there would be cards around the bed saying, "Congratulations, it's a boy." That potentially could have undermined the experiment, although we then gave the videotapes to a panel of judges to simply measure how long the baby looked at the face or the mobile. By the time the judges were looking at these videotapes they didn't have any of these potential clues to the sex of the baby, because all you could see was the eyes of the baby.

The results of the experiment were that we found more boys than girls looked longer at the mechanical mobile. And more girls than boys looked longer at the human face. Given that it was a sex difference that emerged at birth, it means that you can't attribute the difference to experience or culture. Twenty-four hours old. Now you might say, well, they're not exactly new-born, it would have been better to get them at 24 minutes old — or even younger. But obviously we had to respect the wishes of the parents and the doctors to let the baby relax after the trauma of being born. And let the parents get to know their baby. So strictly speaking, it might have been one day of social experience. But nonetheless, this difference is emerging so early that suggests it's at least partly biological.

The results were published in 2001 in a scientific journal and the experiment hasn't yet been replicated, and obviously in science what is needed is independent replication. I'll be interested in other labs to attempt to do this. As far as I know there hasn't been any attempt. This may be because it's quite hard work.

To test a hundred babies, you have to hang around hospitals waiting for babies to be born. That sounds pretty straight-forward, because babies are being born every day. In a city like Cambridge there are about five new babies born a day. For some reason babies tend to be born in the middle of the night, about two or three o'clock in the morning. You have to have a very dedicated research team who are willing to wait. In Cambridge, mothers only stay in hospital for one day. Maybe one night. Then they are sent home, in order to vacate the bed for another expectant mother. In terms of your window of opportunity for testing babies, you therefore have to be there at the right time. We had two very hard-working master's students who approached mothers to ask for parental consent — maybe that was easier in a city like Cambridge, because parents know that in a university town, research is going on.

The test is not invasive — the baby just has to lie on their back and look up. They were presented with each object for only one minute, because babies tend to get very restless very quickly. It's a difficult experiment to conduct, because babies spend most of their time sleeping, or feeding, or crying. You have to wait until they're not doing any of those three things. When they're awake and calm, you have a couple of minutes to present the stimuli.

The camera is well-hidden off to one side. Babies can't see very far — the depth of vision of a newborn baby is only between 15 and 20 centimeters. So it is unlikely that the presence of the camera itself affected how the baby responded.

I was expecting the experiment to be received more controversially, because as far as I know it is one of the first demonstrations of a sex difference in the mind at birth. In fact it was published without any fuss. It may be simply that the climate has now changed, and that people are much willing to accept that there are sex differences in the mind, and that these might even be partly biological. If that's true, then this is good news for scientists who are interested in how the mind works.

My thesis with regard to sex differences is quite moderate, in that I do not discount environmental factors; I'm just saying, don't forget about biology. To me that sounds very moderate. But for some people in the field of gender studies, even that is too extreme. They want it to be all environment and no biology. You can understand that politically that was an important position in the 1960s, in an effort to try to change society. But is it a true description, scientifically, of what goes on? It's time to distinguish politics and science, and just look at the evidence.

Reality Club Discussion

simon_baron_cohen's picture
Simon Baron-Cohen
Professor of Developmental Psychopathology, University of Cambridge; Fellow, Trinity College, Cambridge; Director, Autism Research Centre, Cambridge; Author, The Pattern Seekers

NEW Baron-Cohen responds to Marc D. Hauser, Steven Pinker, Armand Leroi, Carole Hooven, Elizabeth Spelke, Alison Gopnik, David C. Geary, Helena Cronin, Linda S. Gottfredson on sex differences, linguistic ambiguity, systemizers, empathizers, politics in science, and the assortative mating theory

Marc Hauser raises some deep questions about how one can disentangle a genetic (particularly imprinted genetic) theory from an endocrine (particularly fetal androgen) theory of psychological sex differences. Of course, it needn't be the case that these two theories are in competition with each other. There may be both fetal androgenic effects (after the surge in testosterone produced by the testes in the male fetus at about 12-16 weeks gestation) and genetic effects that predate the production of androgens. Certainly in the animal literature there is evidence of both kinds of effects on sexual dimorphism in the brain and behaviour. It remains a challenge to conduct the right sorts of experiments that go to to the heart of which biological mechanisms underlie the observed sex differences in behaviour from birth.

It is refreshing that Hauser reacts with an open mind. He effectively says: Suppose it were the case that the two sexes differ in the mind from birth: how can we take such an idea to the next level of depth and complexity? How can we isolate the possible biological and environmental causes of such sex differences?

Steven Pinker picks up the amusing confusion that has occurred in language in using the word 'gender' and 'sex' interchangeably when they mean quite different things. He says that we have 3 different meanings (intercourse, dimorphism, and gender roles) to talk about, but only 2 words (sex and gender) available to us.

Pinker is on to something here, in that we are just emerging from decades of linguistic ambiguity and linguistic inadequacy. I would go one step further and argue that we probably have at least 7 different meanings, and only 2 words available to us. So, when we think about someone's sex, we can think about their chromosomal sex (how many X or Y chromosomes they have); their gonadal sex (do they have testes or ovaries); the sex of their brain (do they have the brain structure and function that is typical for people with their chromosomal sex); their sex-typical cognition (do they have the profile of strengths and weaknesses that is typical for people with their chromosomal sex); their sex-typical behaviour (do they act in ways that are typical for people with their chromosomal sex); their gender identity (do they identify with others who share their chromosomal sex); and finally, their sexual orientation.

The scientific studies that have been conducted are now beginning to show that whilst some of these 7 ways to think about sex may be linked. For example, if you have a Y chromosome then you are likely to develop testes, which in turn produce fetal testosterone which affected brain structure and function, cognition and behaviour. But some of these 7 ways to think about sex may be independent. For example, a girl with the genetic condition of congenital adrenal hyperplasia (CAH), which leads to over-production of fetal testosterone, is more likely to resemble a typical boy in her behaviour. Clearly we are going to need finer distinctions in our language to talk about someone's sex at all these different levels (and more), but it is interesting that even given the poverty of our vocabulary, this is not holding the science back. Key experiments (such as those with girls with CAH), are teaching us how to think about sex.

Armand Leroi wonders if children with autism are necessarily hyper-males. In particular, he asks, if this hypermasculinization is correct, why we don't see characteristics such as excessive aggression? I think Armand is right that some predictions flow from the theory. For example, do women with autism have delays in menarche, or have more male-typical interests? Experiments to look at these outcomes are underway in our lab. We should be careful though not to assume that all male characteristics are under the control of fetal rather than post-natal testosterone. Aggression for example may not be, whilst language development may be. We will need to differentiate the 'organizational' effects that testosterone may have during brain development from the 'activational' effects it may have at later points in development.

And we will need to keep in mind that if autism involves elevated levels of fetal testosterone, these may not be at the extreme levels as to cause clear changes to the sex organs, as in the female spotted hyena. Rather, these may simply be subtly higher levels that may affect brain and cognitive development. The effects may be subtle too, such as slowing language or social development. But these are empirical questions that I for one regard as important laboratory work to be undertaken.

Carol Hooven's comments are exactly on target. She raises the question: why do boys with CAH not have superior abilities in spatial tasks, if it is simply a function of how much fetal testosterone they produce? But the work of Grimshaw in Canada suggests that the relationship between fetal testosterone and spatial ability such as performance on the Mental Rotation Task is not linear; it may be an inverse U shape. Optimal performance may be in the low average male range, with either too much or too little fetal testosterone being correlated with worse performance. To me, this makes the mechanism all the more intriguing. I agree with her that we need more research into these delicate dose-dependent effects. Whether mating effort (the behaviour she has studied) has anything to do with fetal rather than current testosterone is an empirical question.

Elizabeth Spelke talks of 'systematizers' where I prefer the economy of syllables ('systemizers')! More seriously, she assumes I am discussing binary categories: either you're an empathizer or a systemizer. Nothing could be further from the claims in my theory. For me, these are continuous dimensions, not categories. For this reason, we developed metrics like the Empathy Quotient (EQ), or the Systemizing Quotient (SQ). It turns out that there are statistically significant sex differences on both of these measures. In my theory, I argue that it is the difference score that defines a typical male or female profile. It is not that female = empathizer (E) or male = systemizer (S). The claim is rather subtler than that. Rather it is that among females we find more individuals with a difference score such that E > S, and that among males we find more individuals with a difference score such that S > E.

Spelke also seems to have a concern that a single concept like systemizing can capture why individuals end up in such diverse occupations as cathedral building, ballroom dancing, or seal hunting. Interestingly, all 3 may be very good examples of systemizing, because they all involve understanding lawful or rule-based systems.

Spelke considers the newborn baby study that is currently a key piece of evidence for a stronger social interest in females and a stronger interest in objects among males. She worries that the experiment betrays something of my politics, but of course she knows nothing about my politics. If she assumes that because I tested sex differences at birth I must have conservative political leanings, or be against the equality of the sexes, she will have to take it on trust that the opposite is true. But more importantly, it is irrelevant. Hence my plea to keep politics and science clearly distinct, so that we can explore if and how and why the sexes are different, even if we are feminists or democrats.

I respect her for focusing on the experimental method of the newborn baby study. We want to know which methodological variables may have produced these results. She is right that all we know from this experiment is that more boys than girls looked for longer at the mechanical mobile, and more girls than boys looked for longer at the human face. But does this mean girls have a stronger interest in people than things, and boys have the opposite pattern of interests?

Possibly, but not necessarily. The infants' attention was being lured by two very moving objects. It could be that on average, more girls were attracted by the type of motion that faces have and, on average, more boys were attracted by the type of motion that mobiles show. Of course, re-describing the results in terms of motion-types may still be very important. For example, faces have 'biological' motion (they are self-propelled, with animacy) whilst mobiles have 'mechanical' or physical-causal motion. Are male and female brains on average tuned to react to these two types of motions differently? Further experiments would be needed to isolate if motion-type is the critical variable. In a first study we could not run additional controls to explore such possibilities, as babies will not tolerate a long experiment where such parameters can be varied systematically. But such additional experiments do need to be tried.

She worries if the experimenter was blind to the hypothesis being tested, or if the results could simply reflect experimenter-bias. I am glad I can reassure her on this point. The experimenters (there were two of them) asked mothers not to reveal the sex of their baby until after the experiment, precisely to avoid such experimenter bias. In 101 babies, there were I recall only 3 cases where the experimenter may not have been fully blind. The results were analysed with and without these 3 cases, and this did not change the pattern of results. More importantly, all that was filmed of each infant was the eye region of their face, so that later an independent panel of judges could look at each video film to measure the infants' looking time to each object. So the opportunity for experimenter-bias at the stage of coding the videos was zero. The experimenters were not the judges, and it was impossible for the judges to tell the sex of the baby from the eye-region of the face alone.

Since for some people, results of such experiments will be controversial, it is of paramount importance to control for such biases, which we took all possible human steps to do. Readers interested in the methodology can find the original article published in the journal Infant Behaviour and Development, to see how we attempted to put in place as many such controls as possible.

Spelke calls this a 'single' experiment, tested with a 'single' person and a 'single' object. If the implication is that following this novel experiment there is not yet a series of independent replication studies, then I agree that we must wait for these to be conducted. It is not that there is a set of failures to replicate. There are simply no attempts yet to replicate. I should point out that in terms of data collection, one is looking at perhaps 3-6 months of testing on a daily basis to complete a sample of about 100 babies, and I hope that other labs undertake this kind of research effort.

If the implication is whether the same results would be obtained with a different person, or a different object, these are good empirical questions. We ascertained that even testing newborns with one object compared to one person was an achievement, given a neonate's attention span. But I look forward to other scientists taking up the challenge of conducting different kinds of experimental manipulations with this age-group, so that we can start to examine issues of reliability and specificity of the results.

Spelke suggests that in 3 decades of infancy research, sex differences have not been observed. One has to remember that subtle sex differences if they exist are not going to emerge from small samples. A typical research project includes 20 babies, 10 of each sex, and researchers wisely do not test for sex differences when the power to detect such differences is not sufficient in such small samples. To demonstrate if a 3 month old infant understands a principle of folk physics, one does not need to test a sample of 100 babies. 20 may be enough. But to test for subtle sex differences, one just might need 100 (50 males and 50 females) in a single study.

And besides, it is not true that infancy studies have not reported significant sex differences. The MacArthur infant vocabulary scales report different norms for boys and girls at age 12 months precisely because girls vocabularies are bigger than boys from that age, and remain bigger over the next 24 months. Many independent studies show that girls on average also make more eye contact, and play with different kinds of toys to boys, from as early as 12 months old. What makes the neonatal study of sex differences new is the age at which they were tested.

Finally, Spelke concludes that "infants don't choose whether to systemize or empathize; they do both, and so do we". That's correct. We all do both. It is only a straw man who might be thought to argue that infants choose one or the other. Recall that we are interested in subtle discrepancies in the amount of a person's attention to the physical and social environment. Whilst in this experiment we may have categorized infants according to whether they looked longer at the face or longer at the mobile, we were still measuring differences in looking times at the two types of stimuli. No one is going to suppose for a moment that an infant is solely interested in one of these objects and not at all interested in the other. The world is not that black and white.

Alison Gopnik asks why on the false belief test, a test of 'theory of mind', no obvious sex differences have emerged. My answer to her excellent question is that 'theory of mind', and the false belief test in particular, is designed as a pass-fail kind of test. To pick up subtle individual differences you need a test with a fine-grained scale. To test for differences in height between the sexes you need a ruler with a resolution of centimeters. You can't just have a test such as 'can both sexes see over the table?' To test for a difference in head circumference between the sexes you need a ruler with a resolution of millimeters. Not just a test like 'can both sexes get the hat on their head?'

But there is another reason why theory of mind might not be the right place to look for sex differences, and that is because theory of mind is not the same as empathy. To empathize, you may need a theory of mind (to attribute or compute a person's mental states) but to empathize you have to do a lot more than simply attribute or compute a person's mental states. Psychopaths can attribute mental states but they don't have great empathy. And on tests of empathy, many studies find strongly significant sex differences.

David Geary proposes that sex differences in folk physics and folk biology may not equate with sex differences in systemizing. Why not? When we engage in folk physics or folk biology, surely we are trying to predict the behaviour of a system (a mechanical system, or a taxonomic system, or a digestive system, etc) in terms of its rules? That is systemizing. The concept of systemizing grew out of the older concept of folk physics, but I found the latter too restrictive because there are systems that are abstract (such as mathematics and music) that are as lawful as tools and machines, but are not well encompassed by the concept of folk physics.

Helena Cronin is of course right that we need to understand sexual dimorphism in a Darwinian framework, and she has done more than most academics to explore this framework for understanding human behaviour. The tough part of course is to test predictions from a Darwinian framework in a contemporary human population, when the relevant selection pressures may be ancient. But there are impressive examples of how this can be done.

Linda Gottfredson is the commentator who takes seriously the new idea about assortative mating of two strong systemizers as a cause of autism. Whilst this idea remains speculative, it is testable, and I am glad she pointed out that mating patterns in one generation can change rates of different characteristics in offspring in the next generation(s) with remarkable speed, and that mating patterns effectively create different environments that can affect child development.

I look forward to seeing the tests of the assortative mating theory of autism in the coming years. And I thank those who provided commentaries for their stimulating debate.

linda_s_gottfredson's picture
Linda S. Gottfredson
Project for the Study of Intelligence and Society

Simon Baron-Cohen's work joins the search for causal mechanisms linking genes, brain, and behavior. The patterned variation by sex at all three levels of analysis provides clues to what those mechanisms might be (e.g., testosterone). Baron-Cohen employs those patterns to better understand, in particular, the etiology of autism and its much higher prevalence among males.

Variation is the raw material for much scientific analysis and for evolution itself, but public discussion of human variation seems mostly off-limits today. We are called upon to celebrate diversity but not notice difference; to observe a new etiquette that forbids utterance of supposedly tactless knowledge. Good feeling compels public ignorance. But bewildered or bemused, outraged or apprehensive, most scientists soldier on.

Baron-Cohen continues to investigate the nature of sex differences. His research on babies only 24 hours old, while needing replication, fits the larger pattern of sex differences in interests, personality, and abilities across the lifespan. For instance, at all ages and worldwide, females tend to be more interested in people and males in inanimate objects. As noted in earlier commentaries, humans are not the only primates showing this pattern. It would have been an exception to the rule had Baron-Cohen's team not found boys gazing more at the mechanical object and girls more at the human face. Like the habituation research now used to assay differences in cognitive ability among infants, his results provide prima facie evidence that socialization cannot be the sole cause of variation in social behavior. The interesting question is not whether meaningful innate sex differences exist, but how anyone could construe the preponderance of evidence otherwise.

Baron-Cohen argues that the distinction between "systematizers" (disproportionately male) and "empathizers" (disproportionately female) is especially important in the etiology of autism. He theorizes that genetic risk of autism rises when both parents are systematizers. While onto something important, his work might advance faster and persuade better if, rather than proposing a new distinction, it exploited existing evidence on the dimensionality and relatedness of human psychological traits (all of them heritable), particularly interests ("Holland's hexagon" of six modal types), personality (the "big five"—or three or seven), and abilities (the "3-stratum hierarchical model").

Researchers in vocational interest measurement, personality assessment, personnel testing, and differential psychology have spent a century parsing, cataloguing, and correlating these differences among individuals. They find a regular pattern of sex differences regardless of age, time, or place. It is not clear where Baron-Cohen's systematizer-empathizer distinction fits in this much-explored territory, but it would seem to map best onto dimensions in the non-cognitive realm: sympathetic vs. cold ("agreeableness" personality dimension), "realistic" vs. "social" vocational interests, or valuing "ideas" vs. "feeling."

Prevalence of autism has increased so much recently that some label it epidemic. Baron-Cohen must explain how this increase is consistent with evidence that autism has strong genetic roots. The two facts are not inconsistent, but many people assume they are. They fallaciously reason that if prevalence jumps within only decades (e.g., more violent crime, more women getting BAs in math), then the behavior in question must not be genetically influenced because the gene pool could not have changed during that time. But it need not have.

First, non-random mating can change the distribution of phenotypes in the next generation of the same gene pool. Baron-Cohen's theory would predict that more assortative mating for "systematizing" will lead to more autism in the offspring generation. It would operate like inbreeding, which increases the odds of offspring inheriting the same deleterious recessive allele from both parents. This explanation doesn't work well, however, for rates of socially important behaviors that fluctuate within a generation (e.g., criminal behavior, which has a heritable component), and perhaps not even for autism.

Second, environmental change matters, but not equally for all genotypes. When environments become more deleterious, the more susceptible genotypes are the first casualties. Environmental toxins may be one such factor in autism. Conversely, as environments become more favorable, some genotypes are better able to exploit the new opportunities. So, as barriers to women in education and work have fallen, the most talented and ambitious women have been the best placed to advance. In neither case has the gene pool changed—only the environments favoring some genotypes over others.

Third, different genotypes seek and evoke different experiences and environments. Outgoing, agreeable, feelings-oriented personalities prefer (and are preferred for) dealing with people; non-social, pragmatic, things-oriented personalities find a better fit working with mechanical objects and processes. When free to choose, the two types will gravitate toward different careers. They will also create different personal environments for themselves and their children. So, just as low-IQ parents don't create the most propitious environments for their genetically at-risk children, perhaps two systematizers provide non-optimal rearing for theirs too.

helena_cronin's picture
Helena Cronin
Co-director of LSE's Centre for Philosophy of Natural and Social Science; Author, The Ant and the Peacock: Altruism and Sexual Selection from Darwin to Today

If Simon Baron-Cohen's theory is true, it must be compatible with the workings of natural selection ­ in particular with our Darwinian understanding of the evolution of sex differences in all species.

Baron-Cohen's view is that psychological sex differences in our species can be characterized as male minds tending more towards systemizing than empathizing, female minds vice versa.

The Darwinian understanding of sex differences is that they arise from a trade-off between two kinds of reproductive investment: competing for mates and caring for offspring. Systematic sex differences in investment have been the rule almost since the dawn of high-tech sexual reproduction about 800 million years ago. This is because sexual reproduction of that kind began with one sex specializing slightly more in competing and the other slightly more in caring; that divergence then rapidly became self-reinforcing; and so it widened over evolutionary time, with the differences proliferating and amplifying, down the generations, in every sexually reproducing species that has ever existed. Thus it is that, throughout the living world, males and females tend to differ in much the same characteristic ways. By and large, males are competitive, promiscuous, risk-taking, status-seeking, tenacious, flamboyant; females are less so, for all of the above.

How does Baron-Cohen's view of sexual dimorphism in our species fit within this larger framework? The task of integrating his perspective has not yet been tackled. But this unfinished business is clearly important; and the result could benefit Simon's thesis in several ways. Consider, for example, questions to do with adaptation.

The Darwinian view of sex differences is adaptationist; it explains the differences as resulting from the very different adaptive problems that males and females faced because of their very different reproductive strategies. This raises the question: Are systemizing and empathizing also adaptations? And, if they are, what adaptive problems did they solve and how did they solve them? Alternatively, if they are not adaptations, which adaptations are they the products of? And how accurately do the categories 'systemizing' and 'empathizing' capture a relevant division in these underlying adaptations?

Thus, if Baron-Cohen's categorization reflects adaptive categories — either because it pinpoints specific adaptations or because it maps accurately onto underlying ones — then it can be slotted into its proper place in our wider understanding of psychological sex differences in humans. And we could then assess the robustness and scope of his theory — how reliably and how comprehensively systemizing-versus-empathizing characterizes the entire spectrum of differences.

So, embedding the distinction between systemizing and empathizing within that wider understanding could transform what might otherwise appear to be a merely pragmatic, even arbitrary, distinction into a rich explanatory and predictive theory.

And that would open up exciting new vistas. So, for example, it could shed light on male-female differences in other species. Are there any other species whose ancestors also faced adaptive problems to which systemizing and empathizing are a solution? And if this sex difference is peculiar to our species, why?

For all these reasons and more, I look forward to seeing Baron-Cohen's intriguing insights integrated into our understanding of sexual dimorphism in general.

By the way, moving from the science to its uses... Discussing the claim that sex differences are "all environment and no biology", Baron-Cohen says that one can appreciate "that politically that was an important position in the 1960s, in an effort to try to change society". But no; one most certainly should not appreciate it. If you really want to change society, you need first to understand it. An a priori commitment to 'environment over biology' sounds more like ideological posturing than a serious attempt to solve society's problems.

david_c_geary's picture
David C. Geary
Curators’ Professor, Department of Psychological Sciences, University of Missouri-Columbia

Baron-Cohen's proposals regarding the potential origin of autism and the relation to sex differences in cognition are very interesting and provocative, and at the very least will spur the field to think about these issues in different ways.

I wonder though if 'systematizer' is really the best way to conceptualize the autistic mind in particular and the male mind in general. Systematizing and searching for within category laws suggests explicit, conscious mechanisms that can be focused on more narrow domains, such as knowledge of biology or engineering. There is no doubt that humans are good at this sort of cognitive activity, and there may be more males than females who obsessively engage in systematizing. But in general, there do not appear to be large sex differences in the ability to consciously represent information in working memory and to organize this information in meaningful ways (this is another ways of saying there aren't large differences in general fluid intelligence). Still, boys and men do focus on different aspects of their world than do girls and women and work to organize their worlds in different ways.

More precisely, there are sex differences in more fundamental and implicit cognitive systems that may better capture the male mind and perhaps the mind of many individuals with autism than systematizer. Most broadly, these implicit systems encompass the domains of folk psychology (e.g., theory of mind, face processing), folk biology (e.g., categorizing flora and fauna in the local ecology), and folk physics (e.g., navigation, cognitions but tool use). Each of these folk domains is composed of a number of more specialized systems that in total seem to capture the essence of the evolved human mind.

Girls and women typically outperform boys and men in folk psychological domains — they are better at reading facial expressions, gestures, and at language production, among other differences - and boys and men typically outperform girls and women in folk physical domains — they are better at most spatial tasks, navigating in novel environments, and have a better intuitive grasp of tools. The book is open regarding sex differences in folk biological domains, although I suspect a male advantage in some subareas and a female advantage in others.

In any case, some of the patterns described by Baron-Cohen here and elsewhere (e.g., the results for visualization, over-representation of the relatives of autistic individuals in engineering) suggest that the autistic mind — and that of males — may be better described as being biased toward folk physics, at a cost to folk psychological systems.

Finally, I'd like to point out that many sex differences in general can be conceptualized in terms of Darwin's sexual selection, that is, competition with members of one's own sex for access to desired mates, and mate choice. The advantage of girls and women in many folk psychological domains can be understood in terms of female-female competition over boyfriends, would-be husbands, and other resources. Girls and women compete by gathering as much information on other people as they can get and then using this information to attempt to organize their web of social relationships so as to have better control of these relationships and through this access to what they want. This is called relational aggression. This way of 'fighting' will elaborate folk psychological brain and cognitive systems and eventually result in a sex difference in the same way that male-male physical contests over mates will result in the evolution of bigger and stronger males. Male-male competition may have favored an elaboration of folk physical domains in boys and men, because these support common survival — and reproduction-related activities of men in traditional societies and presumably during human evolution (e.g., navigation in novel terrain to hunt or engage in raiding parties). These types of evolutionary mechanisms provide a very useful big picture framework for hormonal studies that focus on proximate mechanisms.

alison_gopnik's picture
Alison Gopnik
Psychologist, UC, Berkeley; Author, The Gardener and the Carpenter

A quick follow-up to Liz Spelke's point. In hundreds of studies of children's Theory of Mind there is no evidence of sex differences or a female advantage. These include hundreds of studies of the very false belief task that Baron-Cohen, in excellent earlier work, showed was strikingly deficient in people with autism, as well as many other studies of other similar abilities. There are other individual differences, younger siblings, for example, show a consistent advantage over older ones. But not sex differences.

elizabeth_spelke's picture
Elizabeth Spelke
Professor of Psychology and Co-Director of the Mind, Brain, and Behavior Initiative

We humans seem to have an abiding need to reduce the richness, variety, and complexity of our mental lives to categories. In past times, we divided ourselves into the introverts and the extroverts, the field-dependent and the field-independent, the visualizers and the verbalizers. With the advent of neuroscience came the chance to divide our brains into categories as well. That supremely intricate and elusive organ became the left brain and the right brain, the grey matter and the white matter, the male brain and the female brain. Simon Baron-Cohen builds on the last distinction and offers us a new pair of categories by which to sort ourselves: we are systematizers or empathizers.

The categories of the past have a quaint look about them, because none of them has proved very satisfying. Binary categories don't buy us much, because humans are both more and less variable than they suggest. Two categories are far too few to pigeonhole a species whose members can grow from newborn infants to seal-hunters, cathedral builders, or capoeira dancers. Yet two categories are also one too many. In the right circumstances, all of us become introverted or extroverted, swayed by others or steadfast to our principles, visually imaginative or verbose. Global categories tend to obscure the commonalities in human experience and the common capacities, hopes, and failings that define us as a species.

Is there a male brain, and is systematizing its job description? Baron-Cohen asks us to distinguish politics from science and consider the evidence. The evidence for an inborn, male predisposition for systematizing comes from a single experiment on newborn infants, tested with a single person and object. The person was the report's first author, who surely knew the experimental hypotheses and who, we now learn, may have known the sex of the infants whose attention she elicited. The experiment provides no evidence that the basis of infants' preference, if real, had anything to do with the categorical distinction between the displays. Would infants show the same preferences for other face/object pairs? Would they maintain this preference if low-level properties of the two displays, such as their speed of motion, were equated? One need not object to Baron-Cohen's politics to be less than persuaded by his data.

More important, Baron-Cohen fails to consider the extensive evidence that has accumulated, over the last thirty years, on infants' developing understanding of object mechanics. Hundreds of well-controlled experiments reveal no male advantage for perceiving objects or learning about mechanical systems. In most studies, male and female infants are found to discover the same things at the same times. Both males and females come to see the complete shapes of partly hidden objects under the same conditions and at the same ages. They figure out how objects support one another, through the same series of steps. They reach for objects by extrapolating their motions, with equal accuracy. They make the same errors when they search for hidden objects, and they get over those errors at the same time. Sometimes female infants have an edge: In experiments by Laura Kotovsky and Renee Baillargeon, for example, females start to learn about the relation between force and acceleration (the harder a stationary object is hit, the further it goes) a month earlier than males do. Males catch up, however: by 6 1/2 months, you can't tell them apart.

Whatever the newborn infants in Baron-Cohen's experiment were doing, the male and female participants in three decades of infant research have followed a common path, engaging with objects and people. Infants don't choose whether to systematize or empathize; they do both, and so do we all. Baron-Cohen's categories may seem as quaint as left and right brains by the time his newborn subjects are old enough to read about them.

carole_hooven's picture
Carole Hooven
Anthropologist, Harvard University; Author, T: The Story of Testosterone, the Hormone that Dominates and Divides Us

Simon Baron-Cohen's Assortative Mating Theory of Autism is ambitious, and demonstrates the usefulness of innovative, "big picture" thinking in getting a handle on seemingly intractable problems. The ideas seem to fit together nicely to present a clear picture of the etiology of autism. The flip-side of big-picture thinking, however, is that the authenticity of the full picture relies on the validity of each sub-theory; and in this case, the big-picture is painted with a series of appealing, yet tentative strokes. Baron-Cohen acknowledges that the underlying theories need to be tested, and he presents five sub-theories, in the form of testable hypotheses, which focus primarily on the relationship between genetics and systemizing.

His main idea is as follows: people process the world using a cognitive style that falls somewhere along the spectrum of systemizing (more masculine) to empathizing (more feminine). In systemizing, information is processed with attention to rules and laws, and in empathizing, processing is biased toward social cues. As autism can be characterized by the combination of two extremes — a lack of attention to social cues and a focus on laws and rules — then it follows that, should there be genes for systematizing, two systemizers have a higher chance of producing an autistic child. It further follows that, should systemizing be a predominantly masculine trait, the agent that masculinizes the brain, prenatal testosterone, may be also play a central role in both systemizing and autism.

A wealth of evidence, including Baron-Cohen's own research, links the effects of early testosterone to later masculine behavior in humans and non-human animals. Some of the most robust research has examined the effects of Congenital Adrenal Hyperplasia (CAH), in which too much testosterone is produced by the adrenal gland prenatally. Females with this condition (whose hormonal levels are normalized at birth) are somewhat masculinized—e.g., they engage in relatively high levels of rough and tumble play and perform in the male-typical range on spatial tests (higher than normal females). If early testosterone increases systematic processing and masculine behavior, then it follows that CAH boys should also show increased performance on tests of spatial ability; but among males, the relationship between perinatal testosterone and later spatial performance is equivocal at best. Indeed, there is evidence to suggest that excess testosterone in utero actually hampers male spatial ability.

A big-picture, evolutionary analysis of the Assortative Mating theory reveals somewhat of a paradox between conventional notions of masculinity, and the newer notions of "cognitive masculinity." Testosterone can be thought of as promoting behaviors that are traditionally masculine, preparing males physically and psychologically to bias energetic investment toward mating effort. In adult males, high testosterone levels are associated with status-seeking behaviors and the pursuit of mating opportunities. In men, confidence and social dominance (which would require a relatively high social facility) are predicted by high testosterone. The case of the classic nerdy scientist conjures up images of the stereotype of the low testosterone, but in the current context "cognitively masculine," man — a scrawny male who, although he may be successful in the world of technology, is a miserable failure socially and romantically.

The paradox of the two notions of masculinity raises questions about the role of testosterone in shaping psychological traits, such as status-seeking behavior and spatial ability, in utero and in adulthood. With my colleagues Chris Chabris, Peter Ellison and Steve Kosslyn, I've investigated the role of testosterone in solving spatial problems. We have found that although high testosterone males outperformed low-testosterone males on mental rotation tests, the high performers gained their advantage not because they were better at internally transforming objects, but, as the evidence suggested, because they were more confident in their decisions about the similarity of objects. Perhaps the paradox can be at least partially resolved by furthering our understanding of testosterone's role in affecting performance on cognitive tests.

These findings on individual differences in mental rotation performance, along with a relative lack of robust findings on the effects of perinatal testosterone, remind us that picture of how testosterone affects cognition is still far from complete. While evidence strongly suggests that early and late sex differences in testosterone levels play a central role in shaping traits that are clearly related to mating effort, such as dominance, sexual, and competitive behaviors, we should not take for granted that testosterone affects cognitive ability directly. More research should examine the extent to which the relationship between testosterone and cognitive performance is actually modulated by a third variable, having to do with dispositional factors such as confidence and competitiveness, that are more closely associated with mating effort.

armand_marie_leroi's picture
Armand Marie Leroi
Professor of Evolutionary Developmental Biology, Imperial College; Author, The Lagoon

I take the premise of Simon Baron-Cohen's project — that there are innate sexual differences in behaviour and aptitudes — for granted. As Olivia Judson recently pointed out in a New York Times piece about "1/14", it "s hard for a zoologist to suppose otherwise.

I am not, however, wholly convinced by his argument that autistic children — nearly always boys — are, in effect, hypermales. Baron-Cohen has shown that, relative to girls, boys are good at systematising and poor at empathising, and that autistic boys are exceptionally so. This fascinating result then raises a question, namely, why should these two, seemingly unrelated, attributes should trade-off with each other?

Baron-Cohen's answer seems to be: foetal testosterone. Perhaps autistic boys were exposed to unusually high levels of testosterone in the womb, so developing the systematising part of their brains, and repressing the empathising part. It's an exaggeration of a normal process. This strikes me as perfectly plausible, but it also entails a number of peculiar, if testable, consequences.

Testosterone is a hormone and, as such, affects the entire brain and body of the foetus. This means that one should expect autistic boys to be hypermales not just with respect to systematising and empathising — but in all possible ways.

Are autistic boys, then, exceptionally aggressive at play? When they grow up, are they invariably heterosexual? Do they look exceptionally masculine — that is, are they morphometrically hypermale? And what of those rare autistic girls? If they, too, are hypermale due to heavy doses of foetal testosterone, should this not be reflected in their behaviours, their sexual orientations, their bodies?

The motivation for these questions is that we do know something about the consequences of high foetal testosterone — witness the Spotted hyena. In all mammals, the placenta produces a lot of testosterone. This testosterone is, however, broken down into oestrogen by an enzyme called Aromatase. Spotted hyenas foetuses have naturally low levels of Aromatase, and so are exposed to extraordinarily high levels of placental testosterone. The result of this is that females are born with large penis-like clitorises that they can jaunt erect in dominance displays, fused vaginas and very nasty tempers.

Now, clearly we can't ask Spotted hyenas to systematise and empathise. But we can ask their human equivalents. Loss-of-function Aromatase mutations are occasionally found in humans. Like the hyenas, girls homozygous for such mutations are born with masculinised genitalia. Nothing, I think, is known about their temperaments or talents. Are they autistic? Better yet, are their brothers homozygous for the same mutations, autistic as well? If Baron-Cohen is right, they should be.

Of course, even if the hypermale theory of autism is wrong, Baron-Cohen's proposed mechanism of assortative mating among systematisers could still be right. But let us hope not. Surveying my colleagues it seems to me that the reproductive success of gifted female scientists is poor enough as it is. 

steven_pinker's picture
Steven Pinker
Johnstone Family Professor, Department of Psychology; Harvard University; Author, Rationality

I find Simon Baron-Cohen's work admirable in several ways. The systematizer-empathizer dimension is an interesting new way to capture some of the variation between male-typical and female-typical cognitive styles. It cuts across motivation and aptitude (which are often difficult to distinguish) and might subsume some of the long-noted sex differences that have been stated in cruder form, such as orientation to objects versus people. I also am intrigued by the studies of the effects of fetal testosterone, a valuable new way to learn about which psychological sex differences might be consequences of the biological programs that build our brains.

Baron-Cohen wonders why sex is so often referred to these days as "gender." Part of it is a new prissiness — many people today are as squeamish about sexual dimorphism as the Victorians were about sex. But part of it is a limitation of the English language. The word "sex" refers ambiguously to copulation and to sexual dimorphism, and it's often important not to confuse them! The linguistic term "gender" literally means "kind," as in the cognates "genus," "generic," and "genre." Languages often subdivide their nouns into kinds for purposes of inflection, such as human/nonhuman, animate/inanimate, long/flat/round, vowel-final/consonant-final, and male/female. Many Indo-European languages have a gender distinction in their nouns that aligns with a masculine/feminine distinction in their pronouns, and so "gender" was pressed into service as a way to refer to the difference between men and women. Some academics want "gender" to refer specifically to socially defined rather than biologically determined patterns of sex-typical behavior, but this guideline, like most top-down prescriptions about lexical semantics, is rarely obeyed. The basic problem is that we have three concepts to convey — intercourse, dimorphism, and social roles — and at best two words with which to convey them.

I was amused to read that "It may be simply that the climate has now changed, and that people are much willing to accept that there are sex differences in the mind, and that these might even be partly biological." Was this interview conducted before the event that is coming to be known as "1/14"?

[ED. NOTE: The interview took place at Trinity College on 3/12.]

It is hard to imagine anyone living today disagreeing with Baron-Cohen's starting premise that there are biological differences between the sexes. Even the staunchest cultural relativists have to acknowledge that there are differences in the sex chromosomes and hormonal titers that lead directly to differences in our anatomy.

Recent work on imprinted genes a class that fails to follow the classic Mendelian patterns of inheritance shows that maternal contributions are often in complete conflict with paternal contributions. For example, with some imprinted genes, the maternal copy is quiet and the paternal copy is expressed, causing the fetus to extract more from its mother than she would like; these genes often cause pregnancy complications such as gestational diabetes. Studies of the brain using neuroimaging reveal sex differences in structure and function, and work with patient populations reveal differences in vulnerability to mental disorder. And closer to home, there are massive sex differences in the incidence of autism, with studies reporting an 8:1 bias in favor of males.

Where the debate gets interesting is when one attempts to explain how tightly the biology constrains our thoughts, preferences and actions. Baron-Cohen's assortative mating hypothesis is an attempt to grapple with this issue. Much of the evidence hinges on the early appearance of sex-specific signatures of mental function. Early signatures are a tell-tale sign of an innate capacity peaking through, but they are not definitive. One needs to rule out that the experience obtained is insufficient for a learning mechanism to create the capacity.

And here is where Baron-Cohen's observation that newborn boys like to look at mobiles and little girls at faces is fantastic, and just the right kind of start into a serious research program on the biology of sex differences; these results fit nicely with other data showing that for spontaneously generated paintings by young children, little girls almost always draw one or more people into the scene, whereas little boys rarely do, using their canvas as a vehicle for vehicles, from rocket ships to more mundane cars and bicycles.

But now comes the hard work.

What is it about the male genome that sets up a preference for the mechanical or physical whereas the girl genome leans toward faces and the social? How quickly, and with what kind of experience, can these initial biases be exaggerated? Why did these differences evolve? In the language of Darwin, what selected for this kind of preference? Was it our division of labor, with males focused on hunting and therefore technology, while females focused on gathering and the schmoozing that goes on during this kind of activity?

One clue that these are evolved sex differences comes from recent work looking at the incidence of innovation among primates. Across all the primates, including our closest relatives the chimpanzees, males are far more likely than females to take the lead in innovation, and much of the creativity lands in the domain of tool technology. In contrast, for most primate societies, females are for more engaged in the intricacies of social life than are males, largely because females tend to stay in their natal groups for life whereas males emigrate out. If there is a bias toward male folk physics and female folk psychology, there may be traces way back to our primate ancestors.

How are data like Baron-Cohen's reconciled with the fact that for imprinted genes, maternally active copies appear to be largely expressed in the rational frontal lobes whereas the paternally active copies appear to be largely expressed in the emotional limbic lobes? Are there in utero battles that arise over the concentration of testosterone circulating during development, with paternally active genes pushing hard for increased testosterone to push growth and toughness? Are maternal copies pushing in a different direction, attempting to regulate the physiology in such a way that their offspring are social specialists?

What makes work like this so very difficult, especially in terms of selling it to the public, is that more often than we would like to admit, reported sex differences either crumble in the face of follow up work, or for those differences that have been reported and replicated, claims regarding biological underpinnings have fallen prey to more experientially-based accounts. One only need think back to gay genes and gay brains, and the sad fate of those results. Thus, although I am sympathetic to Baron-Cohen's research project and find it odd that anyone would consider this work controversial, there is an obligation to get the story right here that far exceeds the demands in other areas.