“Men and women really do think differently, say scientists,” reported The Times, after a study was released earlier this month on sex differences and the brain. “The much-maligned but longstanding idea that women enjoy discussing their emotions while men are mostly excited by cars may be true after all,” the article continued. Yahoo Sports declared: “There are really mental differences between the sexes (and men are more analytical).” Similar headlines were repeated in news outlets around the globe.
This study, overseen by the Autism Research Centre in Cambridge, headed by Simon Baron-Cohen, claims to be the largest ever analyis of psychological sex differences and autistic traits. It is undoubtedly a hugely impressive study in terms of the size of the groups tested and in the detailed analysis of the resulting enormous datasets – but not for the reasons the headline writers thought.
To understand the study, we need to go back to the spring of 2017, when Channel Four produced a website for a documentary on autism. Users were able to take four tests: the Autism Spectrum Quotient (a test of autistic traits), the Sensory Perception Quotient (a measure of typical or atypical sensory sensitivities) as well as the Empathy Quotient (a measure of awareness of the feelings of others) and the Systemising-Quotient ( a measure of interest in and understanding of rule-based systems of all kinds). A total of 695,166 participants completed the tests. The researchers selected only those who indicated whether they were male or female (losing 22, 887 who ticked “other” or “prefer not to say”).
The measures from the four tests were then subjected to detailed statistical analysis to test predictions from Baron-Cohen’s own theories.
The first, the Empathising-Systemising theory, makes predictions to explain “typical” sex differences: women, on average (and it is important to bear this caveat in mind) will score more highly on measures of empathy, whereas men, on average, will score more highly on measures of understanding rule-based systems.
The second, the Extreme Male Brain theory, suggests that individuals on the autism spectrum will be more likely to score highly on systemising measures. An additional suggestion is that, having equated systemising with autism, individuals working in science-based subjects (science, technology, engineering, mathematics – known as Stem subjects) will have a higher number of autistic traits than those in non-Stem subjects.
The press release on this study summarised the finding thus: in the typical population, women, on average, scored higher than men on empathy, and men, on average, scored higher than women on systemising and autistic traits. These sex differences, however, were reduced in autistic people. On all these measures, autistic people’s scores, on average, were “masculinised”: that is, they had higher scores on systemising and autistic traits and lower scores on empathy, compared to the typical population.
The team also calculated the difference (or “D-score”) between each individual’s score on the systemising and empathy tests. A high D-score means a person’s systemising is higher than their empathy, and a low D-score means their empathy is higher than their systemising.
According to the study, in the typical population, men, on average, had a shift towards a high D-score (more systemising), whereas women, on average, had a shift towards a low d-score (more empathy). Autistic individuals, on average, had a shift towards an even higher D-score than typical males.
Men, on average, had higher autistic trait scores than women. Those working in Stem subjects, on average, had higher systemising and autistic traits scores than those in non-Stem occupations. And conversely, those working in non-Stem occupations, on average, had had higher empathy scores than those working in Stem.
But here is the most interesting finding of the study, easily lost in all the headlines. The report found that differences in empathising and systemising account for 19 times more of the variance in autistic traits than other measures, such as sex. A better predictor of autism is a high d-score, rather than whether you have a penis This should be of major importance, because it highlights the flaws in arguments about sex differences. Indeed, it should be hailed as a key finding in diverting attention away from the notion that biological sex is the primary driver in differences between females and males, typical or atypical. But I have a sinking feeling that this is not the effect that the reporting on this paper will have
The Empathising-Systemising theory is fundamentally a theory about brain types – brain types which can be demonstrated to be more common in one sex or the other, but where biological sex can only, at best, be allocated the role of an influence rather than a primary cause. It is not a theory about women’s brains and men’s brains. And yet I suspect that this is not the message that will be received, even though the study didn’t even take any brain measures.
This is not surprising. In a contentious area of debate such as sex differences (of any kind), language matters. Baron-Cohen’s 2003 book, The Essential Difference, begins with the firm statement: “The female brain is predominantly hard-wired for empathy. The male brain is predominantly hard-wired for understanding and building systems.” Somewhat later on, we get a caveat that you don’t have to be a man to have a male brain, or a woman to have a female brain. Why, you might ask, is it called a male brain or a female brain, then?
This kind of linguistic smoke and mirrors is also evident in the current paper. The authors are at pains to note that “the observed average sex differences likely reflect an interaction of biological and cultural factors” and “the E-S theory does not allow one to make predictions about an individual’s psychological profile based on their biological sex, and to do so would be stereotyping, which is pernicious”. They even note that misunderstanding of the Extreme Male Brain of Autism is “likely based on only reading the name of the theory [my emphasis] but not its actual claims” and urges a “careful reading” of the Empathising–Systemising Theory.
But a sentence such as “..autistic females are masculinised [my emphasis] in both brain structure and function”, or even the use of the term “sex differences” in both the title and title and opening sentence of the paper will tip the scales. The authors’ cautions are buried in the closing stages of the paper – they would have been so much more telling if they had found their way into the abstract or even resulted in giving the paper a different title. “Testing the Empathizing-Systemizing Theory in half a million people” would still have been an impressive headline.
If we could only focus on what brains can do and how they do it, rather than the sex of their owners, we might have better insights into the links between brains and ability. For example, now we know that performance on spatial tasks (such as map-reading) are much more influenced by your videogame experience than your sex, we might start to make some progress in understanding where such skills come from and how we might improve them.
There are also questions that should be asked about the measures they took, irrespective of the fact that they are from half a million (self-selected) individuals. It is noted, but not highlighted, that all these measures are based on self-report (where participants answer directly without the interference of researchers). In other words, they reflect how each of these half a million individuals thought about themselves as they entered their “agree/disagree” answers to statements such as “I am fascinated by dates” or “I would notice if someone added five grains of salt to my cup of water” or “If anyone asked me if I liked their haircut, I would reply truthfully, even if I didn’t like it”. Such self-report measures are prone to the kind of distortions caused by stereotypes – people who know you are measuring empathy are more likely to present themselves in more of an “empathic” light than their unprimed behaviour might indicate. The participants in this study were aged between 16 and 89 years old – plenty of time to have absorbed the gendered messages to which they will have been exposed. Yes, the authors note the importance of “social experience” but this message is rather more muted that the “sex differences” one.
Similarly, rather little is made of the fact that, in these huge cohorts, the differences between the groups are actually very small. “Effect size” is measure of the amount of overlap between scores from different groups, in this case, females and males). If an effect size is small, it means that knowing which group an individual comes from will not allow you to predict how well or badly they might perform on the task in which you are interested. In this study some of the effect sizes are, at best, moderate and often small. Attention should be drawn to the findings, for example, that although males scored higher on the Autism Quotient, the effect size was 0.18, whereas the higher score for females on the Sensory Perception Quotient reflected an effect size of 0.15. So very, very small percentages of the variance could be accounted for by the fact that the respondents were biologically female or male.
In an era where bombardment by stereotypical gendered messages is ever present and where we are still subject to widely publicised outbursts concerning, for example, women’s unsuitability for scientific careers, I am concerned about the message that may be extracted from this paper. If we could only focus on what brains can do and how they do it rather than the sex of their owners we might have better insights into the links between brains and ability. Understanding how brain activity can support empathic behaviour is much more important than knowing whether the brain concerned belongs to a woman or a man.
Gina Rippon is Professor of Cognitive Neuroimaging at Aston University. She is the author of The Gendered Brain, which will be published 28 February 2019 by The Bodley Head.