Delusions of gender and technological ability

Delusions of gender and technological ability

When I tell people about my PhD—that I research ways to make technology more engaging for women—I usually brace myself for a reply like this:

“Why would you bother to do that? Everyone knows men are just better at technology. It’s science!”

It happens more often than you would think, and more often than I’d like. And then I get into a whole conversation about why this isn’t true. So I thought I’d lay out the arguments against biological determinism—the idea that ‘male’ and ‘female’ bodies are in some way built to have different competencies in doing maths, designing bridges or operating computers—so I can point people here instead. And then there turned out to be a lot of arguments to cover and this turned into a pretty long read.

Most of this post is based on Cordelia Fine’s 2010 book Delusions of Gender: The Real Science Behind Sex Differences, but applies her work specifically to ideas about scientific and technological ability and expands on it to discuss intersex and transgender identities. If you’re interested in how biological determinism has become popularly accepted (and why it’s scientifically dubious) then I really recommend reading the whole book (it’s brilliant).

Here we go.

Men’s Work, Women’s Work

In the UK, things have changed a lot in the workplace over the past few decades. Changing attitudes towards the concept of working women has led to vast numbers of women entering the workforce and becoming economically independent. Workplace discrimination on the basis of sex, race, disability, religion, sexual orientation and age has been outlawed. On paper, women now share equal rights with men in many areas of employment. In practice, things aren’t as equitable as they seem.

Despite the introduction of the Equal Pay Act in 1970 that enshrined the right to equal pay for both male and female workers doing “equal work or work of equal value”, data from the Office of National Statistics released in October 2016 shows that female full-time workers continue to earn 9.4% less than male full-time workers on average. There’s been a dangerous backlash recently against evidence of this gender pay gap, which mainly rests on the claim that the pay difference is because men and women aren’t doing ‘equal work’ (meaning ‘the same work’). This is a valid point, but one that fails to take account of the other condition in the wording of the original Equal Pay Act: doing “work of equal value”. Is it really the case that the types of job primarily done by women—such as teaching, caring and nursing—are of less value than jobs primarily done by men?

It also fails to question why men and women aren’t doing the same types of work. Now that official barriers preventing women from working have been reduced, there’s a belief that there’s nothing stopping us from doing any job we put our mind to. The lack of women in specific male-dominated fields is therefore assumed to either be because we just aren’t interested in doing it (so any effort to improve gender diversity would just be working against what women actually want), or because we’re not as good at it as men (and therefore a sort of survival of the fittest effect is happening in the workforce). The same goes for the lack of men in female-dominated fields: gendered differences become naturalised.

We’ve already seen how these differences impact negatively on women when it comes to pay, but there’s also a personal issue here. Assumptions about what males and females are interested in—and what they’re good at—shapes how people live their lives. A girl who would have been happy as a computer scientist may never see this as a valid career option because all the programmers she sees in films and on TV are male. A woman who applies for a laboratory manager position is turned down in favour of an equally qualified male candidate who’s seen as more competent for the role.

This is a particularly serious problem in the STEMM (science, technology, engineering, maths and medicine) industries. A 2014 report by the Royal Society found that although women make up 50% of the STEMM workforce in the UK overall, when you remove health-related industries this figure drops to 40%. In primary science roles, referring to “workers in occupations that are purely science based and require the consistent application of scientific knowledge and skills in order to execute the role effectively”, the figure drops down further to only 25%. On top of this, within the STEMM industries 67% of workers in the highest earning income bracket are male. While workers in the science industries earn more on average than workers in non-science industries, female STEMM workers are still less likely to earn as much as their male counterparts.

The report also identified gender segregation across different STEMM disciplines, finding that women are massively under-represented in engineering, computing and physics, but massively over-represented in nursing, paramedic studies, psychology and veterinary science. However, even in these fields women’s representation drops the higher up the career ladder you go, with women making up 90% of first degree studies in Nursing and Paramedic Studies but only 58% of professors. While we may see these figures changing as current female students progress up the academic career ladder, they still pose issues for female academics in terms of failing to provide adequate female role models and for continuing stereotypes of academia as a ‘male’ profession. In addition, research suggests that gender-based hiring biases also exist in academia and that male academic candidates are more likely to be hired than identical female candidates (though these differences disappear when selecting candidates at tenure level).

Men’s Brains, Women’s Brains

So are these ideas about ‘natural’ differences in male and female interests, abilities and aptitudes correct? In Delusions of Gender, Fine summarises and critiques dozens of psychological and neurological studies that have been cited as evidence for essential intellectual differences between men and women. These studies put forward two theories:

  1. Male brains are on average built more for abstract and systematic thinking while female brains are built more for empathetic thinking
  2. Men are generally more interested in careers that involve systematic thinking, while women are generally more interested in careers that involve caring

The alliance here between the physiological and the personal is not accidental, and provides the basis for many popular science books that claim that biological differences explain and underpin the high proportion of men in fields like physics and engineering and the high proportion of women in the caring professions.

If it’s true that men and women do possess different aptitudes then this claim would make sense: we all want to be successful in our careers, so it would be natural to choose a vocation that makes the most of our natural talents. However, throughout Delusions of Gender Fine consistently debunks the assumptions and misinterpretations that lead to this conclusion.

She doesn’t deny that there are physiological differences (on average) between male and female brains, but she does question whether these differences support stereotypical claims about differences in men’s and women’s abilities. She highlights historical examples where differences between male and female bodies—in brain size, spinal cord size, facial angles etc—were used as evidence of women’s intellectual inferiority and, just like today, were used to argue that women could therefore never be great scientists or engineers. We now recognise that women’s lack of intellectual achievements were the result of their lack of access to education and professional opportunities rather than being based in any of these physiological differences. (Similarly phrenology, once considered a legitimate scientific practice, is now recognised as a method of supporting claims for white racial superiority by providing anatomical ‘evidence’ of superior intellectual ability.)

What sets the current deterministic claims about gender differences apart from historical pseudo-science, Fine argues, is their utilisation of modern medical technologies such as neuroimaging. Brain imaging techniques like MRI and fMRI let us ‘see’ the brain’s responses to stimuli in real time: when someone is presented with a difficult maths problem, for example, we can see which areas in the brain are activated, and find out if there are any differences between men’s and women’s neural reactions. Proponents of neuroimaging technology argue that social constructionist ideas about the cultural basis of gender differences were only feasible in the past because we didn’t have access to the technology that would let us truly test those ideas. Neuroimaging has therefore been tauted as the nail in the coffin for the theory of the social construction of gender.

However, neuroimaging isn’t the magic bullet that it’s been made out to be—at least not in its current form. Firstly, it’s much more complicated than it sounds to map specific parts of the brain to specific behaviours and aptitudes. Our neural activity is extremely complex: a particular part of the brain can do lots of different things in different contexts. And while neuroimaging research has linked specific areas of the brain to involvement in specific functions (such as the amygdalae being involved in the processing of fear responses), it’s much more difficult to link abstract concepts of human behaviour, such as our capacity for empathy, with specific parts of the brain (though there has been some headway with this in recent years).

Secondly, the conclusions of many neuroimaging studies are based on a logical fallacy of reverse inference. Reverse inference involves inferring findings about a person’s psychological state from observations of their brain activity: for example, if activity is detected in the amygdalae when a participant is exposed to a potentially frightening situation then an inference may be drawn that a fear response has been triggered. However, because the amygdalae is involved with many other functions besides processing fear responses, this inference is not deductively valid: the person may not have been scared at all. It’s therefore essential to remember when interpreting neuroimaging data that correlation does not necessarily imply causation.

Thirdly, there are issues with neuroimaging technology itself. Technical restrictions on both image resolution and frame rate mean that imaging technologies are currently incapable of capturing the fine details of brain activity. The application of different data analysis techniques lead to different results about sex differences in brain activity, and also find differences amongst randomly selected participant groupings. There are debates around whether the current significance thresholds in neuroimaging data are accurate enough: as an example, Fine quotes a study where ‘significant’ activity was found in one part of a subject’s brain when it was presented with emotional images. The only problem is that the ‘subject’ was a dead fish. The absurdity of this example highlights the danger of relying on uncontextualised data from fledgling neuroimaging technologies as a basis for inferring conclusions about brain functionality. A final issue is that because of the cost of carrying out neuroimaging studies they tend to be conducted with a very small number of participants, so their findings are likely to be statistically unreliable and are susceptible to chance results and false positives as a result of statistical variations.

Finally—and this is the real kicker—even when neuroimaging data does identify functional differences between male and female brains, this still doesn’t actually tell us anything about differences in male and female abilities. I mentioned earlier how physiological differences such as men having bigger brains have historically been used to support arguments for men’s intellectual superiority. Unfortunately it was later found that intelligence does not correlate to brain weight, and there’s no reason to assume that other sexual dimorphisms in the brain affect men and women’s intelligence or abilities either. Fine argues that, rather than different brain types developing as a result of sex-based differences in genes or hormone levels, different brain types may actually develop to compensate for physical differences in the brain itself, such as size (which tends to correlate with sex, so it’s easy to see why the differences were assumed to be sex-based). As she puts it, “larger brains create different sorts of engineering problems and so—to minimise energy demands, wiring costs and communication times—there are physical reasons for different arrangements in differently sized brains” (143). These different arrangements aren’t necessarily any better or worse than each other, since they are each optimised to function as efficiently as possible given the wetware available.

Unfortunately, interpretations of neuroimaging data do tend to be constructed to fit around pre-existing gendered assumptions. A good example is the theory that men tend to have more lateralised brain structures, ie. that brain activity while performing a task is more likely to take place in mainly the left or the right hemisphere rather than being spread across both. This leads to the interpretation that men are better systematisers, because more localised brain activity mirrors the narrow focus required to understand complex systems. However, you could just as easily say that women are better systematisers because more lateralised activity brings in broader parts of the brain to deal with the concentration and processing power required to understand complex systems. The tautological idea that male or female brain structures explain our preconceptions about male and female abilities really provides no support for them at all. In fact, research into brain activity in mathematically gifted adolescents actually shows that they tend to have more interhemispheric interaction than less gifted teenagers. (As a side note, we also see this reasoning mirrored in arguments that women are ‘too emotional’ to take on responsible or high-pressure jobs because of their high estrogen levels and hormonal reproductive cycles. The fact that men aren’t considered ‘too aggressive’ to take on these roles due to their high testosterone levels is conveniently ignored.)

Even the data on whether men’s brains really do tend to have more lateralised activity than women’s brains is debatable. Several meta-analyses of neuroimaging studies of language lateralisation found no sex differences overall, but did find that the studies that reported sex differences in language lateralisation tended to have smaller sample sizes than studies that reported no sex differences: this is a good reminder to be aware of statistical variances and false positives in neuroimaging studies with small sample sizes.

But even if we were to allow, after all this, that men—on average—tend to have brain structures that give them an advantage at systematising tasks, this still wouldn’t necessarily make them better scientists or engineers. While logical thinking is an essential component of the problem-solving involved in doing science, engineering or maths, a certain amount of imagination and creative thinking is also required. As Fine notes, several Nobel Prize winners have reiterated the importance of intuition in their work: “Albert Einstein, for example, described his breakthroughs as being the result of ‘intuition, supported by being sympathetically in touch with experience’ rather than the end point of a ‘logical path’” (109). Again, the argument that men have a physiologically distinct brain structure that makes them fundamentally ‘better’ at science or maths than women therefore just doesn’t hold water at any level.

It’s also worth noting here that arguments based in theories of ‘average’ brain differences between men and women are also worthless when it comes to knowing anything about individuals. As any scientist will tell you, it’s not possible to extrapolate average differences to an individual case: neuroscientists can’t even tell whether a brain is from a man or a woman at the individual level. If we really wanted to treat people differently based on brain differences we’d therefore have to scan each person’s brain structure instead of crudely separating them based on sex. Even the foremost proponents of brain sex differences note that stereotyping people based on theories of average sex differences in the brain is discriminatory and harmful.

To wrap up this discussion of the dangers of using perceived gendered differences in the brain as evidence of essential intellectual differences between men and women, I should note that even these differences aren’t necessarily ‘natural’. The term ‘neuroplasticity’ has emerged in neuroscience to describe the malleability of the human brain as a response to environmental factors. According to Fine, neuroscientists have argued that terms like ‘hardwired’ which are often used to give the impression that brain structures are fixed and innate “translate poorly to the domain of neural circuits which change and learn throughout life, indeed, in response to life” (178). Even our genes aren’t deterministic: “yes, gene expression gives rise to neural structures, and genetic material is itself impervious to outside influence… But gene activity is another story: genes switch on and off depending on what else is going on. Our environment, our behaviour, even our thinking, can all change what genes are expressed” (177). Since our experiences, education, training, and interactions with others can thereby shape our brain structure itself, would it really be surprising if we saw evidence of differences between male and female brains? And instead of seeing this as an explanation for the differences in interests and aptitudes between men and women, wouldn’t it make just as much sense to see them as a result of how men and women are differently treated?

Men’s Aptitudes, Women’s Aptitudes

Other methods have also been used in studies of aptitudinal sex differences which don’t rely on direct observation of brain activity: for example, self-reporting questionnaires where participants are asked to rate their proficiency at different tasks have been a popular method in psychological aptitude studies.

One of the most popular pieces of evidence used to support the argument that male brains are built more for systematic thinking and female brains are built more for empathetic thinking are the Empathizing Quotient (EQ) and Systemizing Quotient (SQ) self-assessment tests developed by psychologist and cognitive neuroscientist Simon Baron-Cohen at the University of Cambridge. Baron-Cohen argued in his 2003 book The Essential Difference: Men, Women and the Extreme Male Brain that “the female brain is predominantly hard-wired for empathy. The male brain is predominantly hard-wired for understanding and building systems” (1) and that people with a ‘male brain’ make better scientists, mathematicians and engineers while people with a ‘female brain’ make better teachers, nurses and counsellors. This might all make it sound like we might as well just resign ourselves to the current status quo due to the fact that men, with their male brains, are natural scientists and women are natural carers: however, the actual results of the EQ and SQ questionnaires show that only 59% of men report having a ‘male’ or ‘extreme male’ brain type and only 49% of women report having a ‘female’ or ‘extreme female’ brain type. Yes that’s right, not even half the women self-reported as having the ‘hard-wired’ empathetic brain types that Baron-Cohen says makes them poorly suited to STEM careers (and we’ve already seen that ‘hard-wired’ isn’t an appropriate term when discussing brain differences and that being a better systematiser doesn’t necessarily make you better at science). Despite the flimsiness of this evidence Baron-Cohen has become a very public promoter of brain sex differences, and while he always takes pains to reiterate that research about average brain sex differences can’t be used to perpetuate sexist stereotypes about individuals, it’s unlikely that anyone will pay attention to this caveat when he continues to use inaccurate terms like ‘male brains’ and ‘female brains’.

The fact that Baron-Cohen’s EQ and SQ tests have become such a firm basis for theories about essential brain sex differences is also surprising given that self-reporting questionnaires are not an accurate measure of actual ability. Asking somebody to report on how good they are at a particular task doesn’t actually tell you anything about how good they are: in fact, observation of the Dunning-Kruger effect on self-assessment questionnaires tells us that people who are incompetent at a particular skill tend to massively overestimate their abilities, while people who are highly-skilled may tend to underestimate their abilities. Does this mean that the men who are reporting better systemising skills on their SQ questionnaires are actually over-compensating for their lack of ability in this area, and that it’s really women who are the better (but more modest) systemisers? (As a fun exercise, you can take a combined version of Baron-Cohen’s EQ and SQ tests online: I scored 22 out of 80 in the Empathizing Quotient and 18 out of 80 in the Systemizing Quotient, suggesting I have a “lower than average” ability for both understanding how other people feel and analysing and exploring a system. Yay.)

Rather than relying on self-reporting questionnaires, a much more accurate measure of a person’s aptitude at a specific task is to test them under formal or experimental conditions. Studies that have tested men’s and women’s performance at systemising and empathising tasks have produced some very interesting results.

Performance on mental rotation tests is a common measure of general systemising ability. These tests assess a person’s ability to visualise and mentally manipulate 2D and 3D objects, and while some studies have reported differences in male and female performance, others have shown that women perform just as well as men in mental rotation tests… but only under the right conditions, which I’ll get into a moment.

Similarly, results of tests designed to measure empathising abilities—such as correctly detecting which emotion is being experienced by another person—were also contradictory. Some studies found a small advantage for females, while others again found no gender differences under the right test conditions. A meta-analysis of social and empathetic ability studies confirmed that gender differences are much larger in self-report studies: studies that test physiological responses or that observe behavioural or facial differences consistently found no gender differences or only small differences in favour of females.

This suggests that the study design itself plays a big part in whether its results show gendered performance differences, and there was one consistent factor that influenced these results: the presence or absence of gender stereotypes and of ‘stereotype threat’. Making participants conscious of their gender—and particularly of the typically poor performance of people of their gender on the skill being tested—had a direct impact on their test scores. For example, in a mental rotation test administered to undergraduate participants at an American liberal arts college, the participants were split into three groups who were each given a different questionnaire to fill out before taking the test: one questionnaire ‘primed’ their awareness of their gender identity by asking things like “List three reasons why one might prefer living a coed floor in a dormitory”; another primed their awareness of their status as students at a private university by asking “List three reasons why one might attend a private liberal arts college”; and the third questionnaire acted as a control by asking geography-based questions like “List three reasons why one might prefer living in the Northeast to other parts of the U.S.”. The results showed that females in the group primed to be aware of their gender identity performed worse on average than their peers in the other two groups, while males in the same group performed better.

Similarly, in another mental rotation study participants were given a set of instructions claiming that success in the test correlates highly with success in other tasks that require good visualisation abilities. For half of the participants, ‘masculine’ examples of these tasks were given such as “in-flight and carrier-based aviation engineering, in-flight fighter weapons and attack/approach tactics, and operations requiring the ability to process and integrate electronic warfare data”. For the other half, ‘feminine’ examples were given such as “clothing and dress design, interior decoration and interior design, and handicrafts requiring the ability to work with and create artistically tasteful patterns”. Males who were given the ‘feminised’ set of instructions performed significantly worse than those given the ‘masculinised’ set of instructions.

This is a real methodological problem for studies of sex-based differences in ability, since as soon as participants become aware of the purpose of the study their performance unconsciously starts to reflect existing stereotypical attitudes about men’s and women’s abilities. One explanation that’s been put forward for this effect is that when a stereotype threat is introduced into a situation—ie when the participants become aware that they’re being assessed on a task that people ‘like them’ are usually not very good at—this increases the pressure on them to perform well, which counter-intuitively affects their ability to concentrate on the task and impairs their performance.

However, it also provides an opportunity to improve their performance on tasks not typically associated with their gender by actively countering cultural stereotypes. For example, when undergraduate students were specifically informed before taking a maths test that its results had found no gender differences in maths ability, female participants performed better than their peers in the control group and even performed better than their male peers who were given the same information. Another meta-analysis of similar studies that manipulated stereotype threats found that the results of women and non-Asian people of colour on maths tests significantly improved when their stereotype threat was removed or reduced.

All of this suggests that under regular test conditions people who are popularly believed to be bad at a specific task will perform badly at that task precisely because of those beliefs. Gendered and racial differences in test performances, therefore, are not necessarily the result of ‘natural’ differences in ability but are in fact strongly influenced by social and cultural factors.

This is also supported by data on what’s commonly called the Greater Male Variability hypothesis: the idea that males display a greater range of ability than females in any given area. This hypothesis is often used as an anti-sexist explanation for male prevalence at the highest end of practically all professional fields: it’s not that men are ‘better’ than women since males will also be more prevalent at the lowest ends too, with women being more likely to be positioned in the middle of the scale. However, this hypothesis has been disproved by cross-cultural data showing that evidence of greater male variability in mathematics and spatial abilities is US-specific, and that across both of these areas (plus verbal abilities, for which US data shows no average gender differences) there are some countries where males display more variability and others where females show more variability.

When we skip right across tests of ‘systemising ability’ which are intended to indirectly tell us about someone’s performance in areas like maths and actually just go directly to giving people maths tests, we find that gender differences are cultural rather than innate. For example, in the US the ratio of boys to girls in the 99th percentile of maths ability is around 3:1, whereas in the UK it’s 1:1. When we drill further down into the data on top-performing maths students in the US, we find that race is also a factor: in the 99th percentile there are actually more Asian-American girls than Asian American boys. On top of this, historical changes show that the reported gender differences in maths ability have been reducing over time: as Fine argues, this shows that “mathematical eminence is not fixed, or hardwired or intrinsic, but is instead responsive to cultural factors that affect the extent to which mathematical talent is identified and nurtured, or passed over, stifled or suppressed in males and females” (184).

Finally, what all of these studies of ‘intrinsic’ abilities fail to acknowledge is that abilities can be taught. We may not all be able to become world-class mathletes, but we can certainly work on our active listening and empathising skills. We can even improve our mental rotation skills by playing videogames (and this may in itself go some way towards explaining why males perform better on mental rotation tests than females on average). By writing off any observed differences in ability between men and women as essential and innate, we are ignoring the human capacity for learning and growth as well as denying the cultural factors that contribute to those differences.

Boys Will Be Boys and Girls Will Be Girls

I’ve covered a lot of arguments so far that challenge the popular belief that men are built more for systematic thinking while women are built more for empathetic thinking, but what about that second theory put forward by proponents of gender essentialism: that men and women are fundamentally interested in different things and therefore engage in different behaviours. There’s no doubt that men and women—and boys and girls—generally do have different interests, but how ‘natural’ is this situation? Is it biologically determined and innate, as the theory goes, or is it a result of society and culture?

A popular argument for the essential nature of binary sex roles is that we can observe behavioural differences between males and females in the animal kingdom, and even divisions of labour in terms of food gathering and child rearing. Since these differences appear to exist in the absence of socialisation, it supports the theory of ‘natural’ sex differences existing in homo sapiens as well.

Unfortunately, it isn’t possible to extrapolate findings from behavioural studies of other animals to humans since modern humans are qualitatively unlike any other animal species that has ever existed. Humans are the only species to have developed government, economics, healthcare, and complex education and language systems. We (mostly) no longer live in natural shelters or forage or hunt for food. Any practical reasons for sex divisions of labour in the animal kingdom are no longer applicable in most human societies. And there is a hypocrisy in distinguishing humans from other animals when it comes to meat consumption and animal experimentation in laboratories, but not when it comes to supporting arguments for ‘natural’ male and female sex roles.

Even in animal studies, the argument that behaviour is determined by sexual dimorphisms such as differences in hormonal balances is debunked by studies that have observed behavioural differences amongst groups of animals of the same species: for example, observations of Japanese macaque monkeys found that paternal behaviours towards offspring varied greatly amongst different social groups. Primatologists have long argued that male and female sex roles in primate societies are actually influenced by social factors in their specific group, rather than being determined by biological factors such as hormones.

But what can studies of human behaviour tell us? Experiments with babies and very young children have been used to argue that sex differences are innate since they are displayed in children too young to have been affected by social norms, however—as with the aptitude tests discussed earlier—the results of these experiments are ambiguous. For example, one often-quoted study by Simon Baron-Cohen and colleagues found evidence of gender differences in how much interest newborn babies show towards human faces (used as a measure of their innate social and empathetic abilities), while a later study found no gender differences amongst newborns but did find that gender differences had developed when the babies were retested when they were between 3 and 4 months old. This suggests that gendered behaviours are indeed learned, and are in fact picked up by babies from an extremely young age.

Other studies have backed up this evidence that the behaviour of very young babies is influenced by their environment even at ages where we wouldn’t intuitively expect socialisation to have take hold yet: for example, 3-4 month old babies prefer to look at faces of the same gender as their primary care giver (ie. female faces if their mother is their primary care giver and male faces if their father is their primary care giver), and at faces of the same race as themselves (but only if they’re raised in a predominantly racially homogenous environment).

Parents often argue that their sons and daughters behave differently despite them practising gender-neutral parenting, and use this as evidence that gender differences must be ‘biological’ because they haven’t taught them to behave any differently. However, teaching gender-specific behaviours doesn’t have to be as conscious as discouraging sons from taking dance lessons and playing with dolls. As the studies above suggest, even very young babies learn to implicitly prefer whatever they are most exposed to: there doesn’t need to be any overt efforts to socialise babies to have specific gendered preferences since their preferences will be unconsciously influenced by their environment. Other studies have also shown that young children are more influenced by non-verbal adult behaviours such as facial expressions, voice tone and body language rather than what adults actually say: as much as a parent attempts to stay gender-neutral in the way they speak to their children, their implicit gender biases may therefore still unconsciously shine through.

Children also develop their own ideas about gender differences by observing how other people behave. During the run-up to this year’s American presidential election, a quote by US Senator Susan Collins was circulating around the internet:

“Role models are important. My campaign manager in 2008 had an 8-year-old daughter. After seeing me give a speech, she asked him, ‘Daddy, can boys grow up to be senators?'”

This one example highlights the importance of having visible female role models for young girls, but it also demonstrates how easily children make assumptions about men’s and women’s roles based on what they see going on around them. If a child sees their mum doing most of the housework, and goes to their friend’s house and sees their mum doing most of the housework too, then it would be natural for them to assume that housework is done by women. Beliefs like this can have a significant impact on the attitudes and identities of both boys and girls, and can affect how they behave and how they expect other people to behave based on their gender.

On top of this, gender differences are taught by the different ways males and females are depicted in children’s media. Analyses of children’s books and TV shows have highlighted their stereotypical depictions of boys’ and girls’ personality traits (boys are active, assertive and goal-driven while girls are passive, polite and supportive) and men’s and women’s professions (women tend to work in the home while men tend to work outside of it). TV ads also indicate to children which toys are meant for boys and which are meant for girls.

With all of these influences on children, is it any surprise that they start to exhibit stereotypically gendered behaviours even if parents haven’t actively encouraged them? Gender differences are one of the most visible and salient differences between children: boys and girls are visually segregated from birth in terms of the gendered clothing available to them (blue with car motifs for boys, pink with flower or butterfly motifs for girls). The first thing people ask about babies is whether they’re a boy or a girl. Children—even very young children—pick up on these differences and start to wonder about what it means for them and what makes them different to children of other genders, and once they’ve figured this out they start to conform to these differences and to police them in other children.

Fine refers to the work of social psychologist Henri Tajfel to explain this behaviour: Tajfel’s research found that people are more likely to privilege members of ‘their own group’ and to discriminate against members of other groups, regardless of how these groups are defined. He also argued that a person’s sense of identity develops in part as a result of their group memberships: inclusion in a social group is therefore important in developing self-esteem and a sense of self. Tajfel arbitrarily separated participants in his studies into different groupings and found that even when there was no practical basis for these divisions people still privileged members of their own group over others. Since gender is one of the most visible methods of dividing up children, they will therefore naturally identify more with their own group and make efforts to ‘fit in’ with other boys or girls.

This introduces a big problem for research where cultural ideas of masculinity and femininity are built into the study design itself. Many studies aiming to identify a biological basis for gender differences have started from the false principle that certain toys, careers or play activities are inherently ‘masculine’ or ‘feminine’, and use evidence of children’s preferences for things associated with their own gender to support arguments that these interests are biologically determined. As we’ve seen, there isn’t anything to suggest that these interests haven’t developed as a result of cultural influences rather than biological influences, and making an assumption that boys are more drawn to ‘masculine’ toys such as cars and construction sets and girls to ‘feminine’ toys such as dolls and pushchairs due to innate biological differences ignores the fact that these toys are objects that were specifically created for boys or girls (at least in our culture). Fine also notes that gender expectations are disrupted when toys are modified to subvert binary gender conventions: for example, boys showed a lot of interest in My Little Pony dolls that had been painted black and given spiked teeth, while girls enthusiastically played with pink and purple guns and war helmets.

There are other examples of cultural ideas about gender becoming naturalised, such as the association of the colour blue with masculinity and pink with femininity. Fine discusses the work of American Studies professor Jo Paoletti, who notes that until the mid 1900s in the US the colour pink was actually associated with boys as it was a ‘stronger’ colour, while ‘delicate’ blue was associated with girls. The idea that pink is for girls is therefore barely more than 50 years old, but anecdotal evidence of girls being drawn to the colour pink is still used to support theories of hardwired biological differences.

Gender stereotypes don’t only influence the behaviour of children: they also influence adult behaviour, such as the types of personality traits we want to project and the type of job we want to do. We saw earlier that inclusion in a social group is important in forming self-esteem and a sense of identity, and that this is particularly important when it comes to gender as it’s such a key part of classifying people in our society. The ability to ‘fit in’ with other people of your gender and to be recognised by others as belonging to your gender is therefore an important identity issue, and when we engage in behaviours that go against what is stereotypically expected of people of our gender we are taking a personal risk with our sense of self and of social belonging.

We’ve seen this in meta-analyses of aptitude studies that conclude that women self-report as being more empathetic because women are supposed to be empathetic so this is a trait they would like to project to themselves and to others, but we also see this in people’s attraction to specific ‘masculinised’ and ‘feminised’ careers. I discussed in an earlier post how technology and engineering occupations have been stereotyped as ‘male’ industries and how entering these industries can therefore require women to relinquish their femininity, which introduces a barrier to entry for women that men don’t face. Fine points out that this may explain why some previously male-dominated STEM fields such as psychology, biology, medicine and forensics have become more gender balanced while others haven’t, since it’s easier to reconcile jobs like these—which are perceived as being ‘nurturing professions’—with stereotypically feminine values. However, she also notes that other fields where women have made very little progress could also be perceived as having nurturing and caring aspects in the right light: for example, don’t the physical sciences and engineering also provide lots of opportunities for helping people by creating innovative solutions to problems such as global warming, food shortages or lack of adequate housing? The association of ‘feminine’ STEM fields such as medicine and biology with the human body—which has long been a female domain contrasted with the ‘male’ rational brain—may therefore be more important than the actual impact those fields have on human wellbeing.

Stereotypical attitudes held by our colleagues and peers also have an enormous impact on whether we feel like we ‘belong’ in a particular field. It’s natural for people to want to have a career where their work is valued and recognised, yet women entering male-dominated professions have consistently faced both explicit and implicit sexism that questions their abilities and denies the value of their contributions to their field. Fine provides numerous anecdotes from women entering the professions in the 1960s and 70s, when women were starting to enter the workforce en masse, who found themselves being treated differently by their peers, professors and employers: they were asked dumbed down questions in class, were jokingly asked if they were really a woman when they did well in exams, or were told in job interviews that a company’s senior partners and clients would never agree to working with a woman (and note that this was after employment discrimination on the basis of gender was outlawed). We have already seen that there is on ongoing assumption that men have superior scientific and technical abilities—supported by selective neurological and aptitudinal data—and that even if people don’t hold explicit prejudices against women their behaviour may still be influenced by implicit gender biases: therefore, women in STEM workplaces continue to be at risk of discrimination and singling out on the basis of their gender.

Findings in the Athena Factor report published by the Harvard Business Review show, worryingly, that women in STEM fields may try to mitigate this risk by playing down their femininity and attempting to blend in with their male colleagues, and even by expressing anti-female opinions and denigrating expressions of femininity such as wearing make-up or dresses. As well as threatening women’s sense of identity by requiring them to conform to masculine stereotypes rather than enabling them to behave in whatever way they are most comfortable with, this behaviour also sadly perpetuates the idea that only men can be successful in these fields, and women can therefore only be successful by turning themselves into men.

Historical and cross-cultural studies have highlighted the cultural specificity of ideas of ‘masculine’ and ‘feminine’ professions by showing that these ideas are not constant across time and location. Before women entered the workplace, jobs that we now see as ‘feminine’ such as teaching and secretarial work were performed by men. Computer science in countries like Armenia, Malaysia, Singapore and Thailand has not been male-dominated. In fact, surprisingly, sex segregation tends to be stronger in more economically and industrially developed countries than in developing countries, which goes against the belief that social advancements necessarily lead to more egalitarian working environments. There also appears to be a hetero-normative aspect to this, since research indicates that lesbians may face less stereotype threat than heterosexual females when it comes to entering traditionally masculine professions.

As with aptitude tests that assess performance on gender stereotyped tasks, recognising that gendered behaviours are influenced by stereotypes also provides an opportunity to counter sex segregation by introducing more inclusive narratives. For example, reading children stories featuring anti-stereotypical main characters can influence girls’ choice of toy preferences, and even changing the décor of a computer science classroom to feature less ‘geeky’ objects can improve women’s interest in it as a subject. These are very simple interventions that can help girls and women to feel more of a sense of ownership over stereotypically ‘masculine’ things, however the popular idea that gendered differences in interests are biologically determined risks denying the value of these interventions on girls’ and women’s agency and ability to pursue satisfying and well-paid careers in STEM professions.

Finally, cultural gender roles also give rise to practical differences in living arrangements that influence women’s career decisions. Women still do the majority of housework and child-rearing work in American homes, even if they earn more money than their partner or if their partner doesn’t work at all. Women who give up their careers to raise their children are generally assumed to have made this decision because they are more interested in spending time with their family than in working, but interview evidence in Pamela Stones’s book Opting Out? Why Women Really Quit Careers and Head Home suggests that woman are actually forced into this decision because they don’t get the support they need from their partners around the home to be able to keep working as well as raising kids. Their partners may be very supportive in other ways such as providing for their family financially (which enables the women to give up their careers), but are less likely to offer to give up their own jobs to stay at home so the women can work.

Crucially, then, women’s behaviours and choice of interests—including their career choices and their ability to continue their career after having children—don’t spontaneously arise out of their own internal drives and motivations. They are influenced by our cultural expectations of what women can and should be doing, and the pressures we exert on people to conform to gender roles. Any current study of men’s and women’s behaviours therefore needs to start from a set of first principles: that these behaviours are influenced by wider social and cultural frameworks of gender expectations; and that people have strong personal incentives for complying with these expectations to maintain their sense of belonging and identity. Assumptions that these behaviours arise instead out of innate biologically differences between males and females ignores these first principles, and denies the potential for intervention to improve gender equality for both men and women.

Beyond Male and Female: Intersex and Transgender Identities

A big issue with most of the research on biological sex differences, including the vast majority of the studies discussed in Delusions of Gender, is that it ignores the experiences of people outside of the traditional male/female sex binary, such as people with intersex characteristics or transgender people.

In order for people to fit unproblematically into the male/female sex binary, their chromosomal characteristics and sex hormone characteristics need to match up with a fairly specific outward phenotype (meaning a person’s bodily appearance). A ‘typical’ human female will possess XX sex chromosomes, female gonads (ovaries), a female productive and urinary system and secondary bodily characteristics such as breasts, wide hips, a small build and lack of body hair. They will also have high levels of female sex hormones (estrogens and progestogens). A typical human male will possess XY sex chromosomes, male gonads (testes), a male reproductive and urinary system, and bodily characteristics such as a deep voice, muscular build, body hair and lack of breasts, along with high levels of male sex hormones (androgens).

‘Intersex’ is an umbrella term used to describe a person whose sex chromosomes, hormonal balance, gonads, reproductive / urinary systems or secondary sex characteristics don’t all fall under either the male or the female classification. There are a huge variety of intersex conditions: historically, ‘hermaphrodites’ have been intersex people who possess aspects of both ‘male’ and ‘female’ reproductive systems, occasionally including the possession of both a functional vagina and a functional penis. With other intersex conditions, people may have a chromosomal sex variation outside of the usual XX/XY binary—such as males with XXY chromosomes (Klinefelter syndrome) or females with only one X chromosome (Turner syndrome)—which affect development of their secondary sex characteristics. Intersex people are usually assigned a binary gender classification at birth and then raised as either female or male, often with hormonal or surgical interventions to ‘correct’ aspects of their body that don’t fit into typical male/female classifications, however there has been increasing action from intersex activist groups to prevent irreversible interventions on intersex babies and young children until they are old enough to decide which gender classification(s) best fit their own sense of identity.

A number of studies have been carried out to estimate the number of people with intersex characteristics, but groups such as the Intersex Society of North America and the Organisation Intersex International Australia recommend the estimate of 1.7% of births based on the meta-analysis of medical studies from 1955-1998 carried out by Anne Fausto-Sterling and colleagues. This is a statistically significant number: if accurate, it means there are currently over 125 million intersex people worldwide.

While intersex refers to people whose physical bodies don’t fit into the binary sex system, transgender people are people whose sense of gender identity doesn’t necessarily match their physical sex. Transgender is also an umbrella term, and there are also a huge range of transgender identities: it may refer to people who identify within the binary sex and gender classification system but who identify as the opposite gender to their physical sex, or it may refer to people who identify outside of the male/female binary (such as agender, pangender or genderfluid people). Some transgender people choose to undergo surgical or hormonal treatment to bring their bodies more in line with their sense of gender identity, while others do not.

It’s very difficult to estimate how many people are transgender since, unless somebody chooses to undergo a reassignment process or to legally change their name or sex, there are few official records of transgender statistics. Not all transgender people wish to physically or legally transition, and even if they do there may be barriers preventing them from pursuing this such as fear that they won’t receive support from their family and friends or living in a culture that is intolerant of transgenderism. A survey carried out by the Equality and Human Rights Commission estimates the number of transgender people in the UK at 0.6-0.8%, but these statistics may change as transgenderism becomes more visible.

There is therefore an important question of how ‘sex’ and ‘gender’ are being determined in neurological and psychological studies: are they conducting a genetic analysis of each study participant to determine their sex chromosomal makeup? Are they measuring their sex hormone levels? Are they asking whether their sense of gender identity matches the sex they were assigned at birth, and whether they have any physical intersex characteristics? If not, then their results about ‘males’ and ‘females’ may not be about biological males and females at all.

Studies of people with intersex conditions do raise some interesting questions for research on sexual dimorphism. Fine discusses a number of studies of girls with congenital adrenal hyperplasia (CAH), a condition where they are exposed to high levels of androgen hormones (such as testosterone) in utero. Studies of girls with CAH consistently find that they display a preference for playing with masculine toys and have more interest in masculine career choices than non-CAH control girls. However, other studies have failed to find a correlation between in utero testosterone levels in non-CAH children and gendered play behaviours. Does this mean that girls with CAH have a special biological property that we haven’t investigated yet that influences their engagement in masculine behaviours? Or could they have been socialised differently because of their intersex status? As a result of their exposure to high levels of androgens female foetuses may develop ambiguous or male genitalia: even if they undergo ‘corrective’ surgical and hormonal treatments, could this still affect how their family view them?

Studies of transgender people may also provide useful data on how differences between people of different genders develop, and what role being recognised and treated as a member of a specific gender plays. Are there consistent similarities in neurological patterns and performance on gender-stereotyped tasks in, for example, men who were raised as female vs men who were raised as male? Do neurological patterns begin to change after a person transitions their bodily appearance to a different gender? Acknowledging the flexibility of human gender identities and working this into scientific research may provide entirely new avenues for understanding difference.

Attention to the specificities of human bodies and identities at an individual level is therefore essential in conducting effective research. Bodies within specific sex and gender categories vary enormously and in complicated ways, and research that focusses on determining ‘average’ sex differences ignores the fact that these findings will be inapplicable to the vast majority of people, as well as completely irrelevant to those people living outside of binary sex and gender classifications.

Responsible Research and Biological Determinism

There’s a lot of legitimate reasons for doing research on biological sex differences: as Fine points out, neurological research can help us to better understand why some psychological disorders tend to be more commonly experienced by people of different sexes and genders. Londa Schiebinger also highlights in her book Has Feminism Changed Science? the importance of paying attention to differences in hormone levels and body mass when developing and prescribing pharmaceuticals (an issue that women have particularly suffered from as drugs have historically been tested on males and then prescribed universally). But for this research to be responsible, it needs to be aware of the limitations of its findings and the high potential for negative consequences when data is oversimplified and used to make far-reaching statements about innate sex and gender differences.

There is a very real political agenda behind pseudo-scientific arguments for biological determinism as they legitimate gender stereotypes by making them ‘scientific’ rather than sexist. It’s extremely easy for research findings to be exaggerated or taken out of context and used to perpetuate gender inequalities: after all, if we only pay attention to evidence from a few select neurological studies that suggest men’s brains are ‘naturally’ suited to doing maths then it wouldn’t be in anybody’s best interests to initiate programmes supporting female maths scholars since it draws resources away from supporting the ‘top performers’ and inevitably leads to disappointment for the women who are never able to match up to their male peers. If, however, we utilise some neuro-scepticism and also look at inconclusive and contradictory data from other neurological studies, recognise the limitations and assumptions involved in those studies which do find evidence of male superior ability, and pay attention to findings from sociological and psychological studies that demonstrate how social interventions to encourage budding female mathematicians can demonstrably improve their performance, it becomes much easier to justify these programmes which can make real improvements to women’s lives.

It’s easy for scientific findings to be mis-interpreted when they’re translated from science textbook or journal paper to popular science book or news article, but Fine demonstrates that many of the popular science books espousing biological determinism have actually fabricated data, falsified quotes and exaggerated ‘scientific’ findings of sex differences to support their arguments. She argues that scientists have a responsibility to be aware of how their research is being interpreted and to actively challenge inaccurate interpretations in the media and popular culture, which have the power to do enormous damage. Readers also have a responsibility to exercise scepticism, but with many of the papers cited as ‘supporting’ these arguments locked up behind paywalls it’s often difficult to check how accurately the claims actually fit their sources—and due to the technical nature of many scientific papers it’s a challenge even to assess what its findings actually were. There’s also a wider issue here of publication bias in medical science, which means that studies which find evidence of biological sex differences are more likely to be published than those that don’t. It therefore falls to responsible scientists to prevent the social repercussions that can be caused by the publication of inaccurate summaries of their research.

There is also a responsibility to recognise the variability and flexibility of sex and gender categories, whether this is in terms of physical variations within and between sexes (including intersex bodies) or in terms of gender identities. If sex difference research is to be fair then it needs to be accurate and applicable to individuals, not to a mythical ‘average’ that only exists in statistics papers. It needs to stop talking about ‘male bodies’ and ‘female bodies’ and start talking about bodies possessing different neurological pathways, hormone levels and secondary sex characteristics. It needs to pay attention to specifics, and it needs to stop being hijacked to justify privilege and inequality.