Bad neuroscience and gender: reading this will change your brain

Warning: reading this post will change your brain. So will choosing not to read it, though, so you might as well continue and learn something. There’s been yet another neuroscience study claiming to have found important, systematic differences between male and female brains. I can’t critique the original paper, because it’s behind a paywall, so I’m going to have to trust Tom Stafford’s analysis of it and Cordelia Fine’s discussion of the paper’s problems. Even so I can see that the secondary reporting is utterly dire, and the interest in this study hinges on multiple misconceptions about gender, statistics, causality and brains.

The Guardian has the least bad of the popular press reports I’ve seen about this, so it’s the only one I want to encourage anyone to read. In fact, to the Guardian’s credit, it looks like all the badness in their article came from the original paper or the corresponding author’s statement. The claims are that in a huge set of brain scan data, researchers found “stark differences” between the “wiring of male and female brains”, and those differences correspond very neatly to gender stereotypes. Towards the end of the Guardian’s article, they casually mention that the gender differences don’t appear until age 13 or 14, but outside that one sentence the implication seems to be that these are innate differences (the BBC article doesn’t even mention the age effect, which is one of the reasons I’m not linking to it).

So, what’s wrong or missing?

No mention of effect sizes: none of the secondary reporting details how big these differences were, and from the critiques I’ve read it sounds like the original paper either doesn’t mention the effect size or doesn’t give enough information to put them in context. This sounds like a minor technicality, but I’m leading with it because it’s very important, and a lot of published psychology research suffers from this problem. We hear that a difference is “significant”, and in day-to-day English that sounds like a declaration that it matters, but in statistics “significant” is a technical term with a very narrow meaning. It only tells us that there is a low (usually less than 1 in 20) risk that the difference in the data was the result of chance. The thing is that the bigger your sample, the smaller a difference you need for it to be statistically significant, and this study features 949 brain scans.

No mention of within-group differences: this leads into another crucial point: there’s bound to be some difference between any two women, and between any two men. For almost everything that isn’t directly related to reproduction, the differences between two random individuals of the same sex is are much bigger than between the male average and the female average. Here’s an example for height:

Sex differences in height in the USEven for a trait on which we can easily see the obvious difference between men and women, 12% of women have a height more characteristic of men, and a full third of men have a height more characteristic of women. I haven’t looked all that thoroughly, but I bet this is the most gender-differentiated statistic outside of chromosomes and reproductive organs, and there’s still that much overlap. Now here’s an example for something subtler: the ratio between the lengths of two fingers:

finger length ratio by gender

Both charts from which is worth reading in its entirety, unlike the articles I’m responding to here.

Now I’m not interested in finger length ratios, and don’t expect you to be either.  The reason I’m showing you this chart is that it illustrates a difference that is statistically significant, but all but meaningless. If you used this ratio to decide if someone was male or female, you’d be wrong almost as often as if you just tossed a coin. Most psychological differences between sexes are smaller still than this, relative to the same-sex variation. I don’t know if it’s true for this study, but any time information like this is missing my suspicion is that it would have been mentioned if it supported the researchers’ claims. If they never looked at all they need to retake undergraduate statistics.

When we look at behaviour or ability, differences between arbitrarily selected groups (genders, races, nationalities, etc) are almost always much smaller than differences within those groups. Almost all the exceptions are things that are clearly culturally learned, like my Latin@ colleagues speaking better Spanish than me.

No reason to assume these differences are innate. At least the Guardian’s report doesn’t explicitly claim that they are, but others do. The Independent’s version has “hardwired” in the headline. I was going to decline to link to it for that reason, but then I realised that they got that directly from one of the researchers:

“What we’ve identified is that, when looked at in groups, there are connections in the brain that are hardwired differently in men and women. Functional tests have already shown than[sic] when they carry out certain tasks, men and women engage different parts of the brain,” Professor Verma said.

But not only have I seen nothing to imply that they’ve found innate differences, there’s even evidence that these differences were probably learned. That crucial sentence in the Guardian’s report:

Male and female brains showed few differences in connectivity up to the age of 13, but became more differentiated in 14- to 17-year-olds.

I realise that some innate traits are hidden until puberty, but the vast majority of differences that emerge that late are culturally specified. This even goes for things very commonly assumed to be intrinsic to gender. For example, when researchers compared competitiveness between genders in a matriarchal society, they found that the women behave more competitively than the men, at just as extreme a ratio as the men did in a patriarchal society. There can be a strong sex difference in a behaviour, and it can still be proven to have no innate component whatsoever.

This shouldn’t surprise us, given how children are deluged with gender stereotype reinforcement from birth onwards. Ask any new parent: if they announce the sex before birth they get heavily gendered gifts, and if they don’t they get unending social pressure to do so because it would be just impossible to buy or make things for a newborn without knowing if they need to be pink or blue. Even the pink-vs-blue thing is entirely cultural and arbitrary: it was the other way round in early 20th Century America.

Incidentally, there’s a really good evolutionary reason why most differences between humans are cultural, and why within-group variation is so huge. Most animals have evolved to specialise in a particular ecological niche, and their offspring are born much closer to adult capabilities. Humans’ evolutionary strategy has been to take over the world by being the most adaptable animal, and the price of this is that we give birth to unusually helpless young, with brains exceptionally far from adult capability, and needing to be nurtured and educated for many years longer. We are the species that gets the most of ourselves set in the years after our birth, and that’s both why we’re so smart and why humanity is so wonderfully diverse.

The circuit and road network analogies for the nervous system are terrible and misleading. This isn’t as important as the issues above, and I can’t really blame the journalists when (a) properly describing the nervous system is hard, and (b) neuroscientists talk about ‘wiring’ a lot. But it’s still misleading. A brain is what a circuit would be if it were orders of magnitude too complicated for people to really understand it, each component were itself a chemical machine so complicated that it takes significant computing power to model one with our puny transistors, connections massively outnumbered “components”, it had non-localised signaling by diffusing chemicals through the fluid in which it sits, and most importantly the map of connections were changing all the time. I think these analogies are one of the reasons why we tend to mistake “there is a difference between these brains” for implying at least one of “therefore it’s innate”, “one must be faulty” and/or “the difference must be important”, when in fact none of those follow. For a silly but useful example, look at the systematically different brains of London taxi drivers: definitely not faulty, and almost certainly not innate either.

My headline isn’t melodrama, it’s actually a rather mundane truth: any time you learn anything your brain changes. So every time you reward a girl for behaviour that you’d make fun of a boy for, or vice versa, you’re changing that child’s brain. If anything, the most surprising finding of this study is that it takes until 13 or 14 years of age for these differences to be noticeable in brain scans – really that’s evidence for the relatively primitive state of our tools.

The left-brain vs right-brain thing is complete and utter bullshit. Not much to say on this: just that the differences between brain hemispheres on traits popularly described as being “left-brained” or “right-brained” are, just like psychological differences between sexes, much smaller than the differences between individuals. So there’s really no reason to assume that person A will behave any differently from person B because one has more connections within their hemispheres while the other has more connections between them. Some brain functions are extremely localised, but to specific spots rather than a hemisphere, and some have barely any localisation at all. Neuroscience that goes into those specifics can be very illuminating, but talk about generalities like left-brain vs right-brain is entirely useless.

In summary, ignore this paper, it’s full of shit.  Of course, at this point you may be wondering why I’ve bothered devoting over 1000 words to a worthless science paper and some bad newspaper reporting. I don’t really care about this particular paper – it’ll fade from view fairly quickly. I do care about the misconceptions that it and the media frenzy play on, because they’re widespread, enduring and do real harm. Every “women are intrinsically worse at [numeracy/spatial skills/science/intense focus]” story contributes to the systematic discrimination against them in technical fields, and every “men are intrinsically worse at [communicating/emotional literacy/relationships]” story lowers the bar for acceptance of bad behaviour from men. Even in my bubble, where overt sexism has long been unacceptable, I see both hurting real people every day. So please remember:

  • Almost every psychological difference between sexes (or races) is way smaller than the differences between individuals of the same sex (or race).
  • Most of them are also culturally acquired, not innate.
  • If you try to predict how a person will act based solely on their sex (or race) you’ll be wrong almost as often as if you made these predictions at random.
  • The brain’s not really that much like an electrical circuit or a computer.
  • If a research report doesn’t tell you about the both significance level and effect size it’s basically worthless.
  • Humans, and our brains, are amazingly diverse, adaptable and unpredictable.

Acknowledgements: thank you to Erin Kissane, Coda Hale and Tim Maly for drawing my attention to most of the sources cited, and to Melinda for proofreading.

This entry was posted in Science and tagged , , , . Bookmark the permalink.