Debunking the male-female brain mosaic

There is no such thing as a male brain, or a female brain. Instead, our brains are all really a mosaic of male and female parts – we all have an “intersex” brain. This is the claim made by psychologist Daphna Joel and colleagues, based mostly on a 2015 neuroimaging study in humans, but also some previous work in rodents. This idea – especially the catchy phrasing – has caught the public imagination and it has been widely covered in the media. Indeed, a recent editorial in Scientific American, entitled “The New Science of Sex and Gender”, cited this study as support for the view that “To varying extents, many of us are biological hybrids on a male-female continuum”. But what do the data actually show? I will argue below that the interpretation of a male-female mosaic is conceptually mistaken and based on a straw-man argument.

 

I’ve discussed these findings and their interpretation before, as an illustration of how the same data can be used to support diametrically opposite viewpoints. Here I want to consider a related point – how framing a research question from a certain perspective colours the conclusions that you draw from a study. In this case, the framing is wrong from the outset.

This description of the study and its findings is taken from my previous blogpost:

Daphna Joel and colleagues analysed MRI scans of 169 females and 112 males, and segmented them into 116 regions using a standard brain atlas. By analysing how much warping was required to map each brain onto a reference template, it was possible to compare the relative grey matter volume of all these regions across the two sexes. From this group comparison the 10 regions showing the largest sex differences were chosen for subsequent analyses.

So far, so good: the primary finding is that there are statistically significant group differences between males and females in grey matter volume across many brain regions. That’s nothing new – a recent meta-analysis of 167 studies confirms consistent group sex differences in many brain areas between men and women.

The authors went on, however, to ask what could have been a more interesting question: across those 10 regions, how “male” or “female” were the structures of individual brains? This is where the subjectivity comes in – there are many ways to analyse these data and the authors chose arguably the most simplistic and extreme one, which enabled them to draw the conclusion that male and female brains are not categorically different. 

They report that: “35% percent of brains showed substantial variability, and only 6% of brains were internally consistent”. Importantly they chose to classify only those subjects showing extreme male or female values for all 10 regions as “internally consistent”.

So, why did the authors choose those criteria, scoring only the extreme values of these parameters? In a 2017 circumspective, co-written with Margaret McCarthy, Joel states:

 In recent years I have been attempting to answer whether sex differences in the brain “add up‟ to create two types of brains, “male” and “female”. For this to be true the effects of sex should be dimorphic, that is, result in the formation of distinct “male” and “female” types, and internally consistent, that is, that all elements of a single brain are either “male” or “female”.

Are these two criteria valid? For “male” brains and “female” brains to be a thing, must the differences between them really be dimorphic? That is, not just a difference in degree, but in kind, like the difference between male and female genitalia, which come in two clearly different forms. Well, no. There is no reason to set that as the criterion.

An analogy with faces makes this clear. Male and female faces differ on a wide range of parameters – size of the jaw, prominence of the ridge over the eyebrows, fullness of the lips, size of the bridge of the nose, and others (think Jason Momoa versus Natalie Portman). For each of these parameters, there is not a male form and a female form – there is a distribution, which is shifted one way in males and the other way in females. None of these markers by itself provides the means to accurately classify faces as either male or female. But taking all of them together certainly does. We are all very good at telling whether a face is male or female and computer programs can also very successfully perform this classification. So, “male” faces and “female” faces are clearly real things, even though the differences in specific parameters between them are not dimorphic.

(Figure credit: from https://clinicalgate.com/aesthetic-contouring-of-the-craniofacial-skeleton/)

The first criterion thus seems to be far too strict – brains may differ on many individual parameters, between males and females, due to a shift in an underlying distribution, rather than coming in two discrete forms. And the second criterion thereby also falls – if these distributions overlap, then you cannot score any individual region as being either “male” or “female” and the idea of assessing internal consistency on such a basis makes no sense.

Joel clearly acknowledges this fact, in a 2011 review:

There are only a few brain characteristics for which the term sexually dimorphic, which literally means having two forms, is appropriate, that is, for which there is minimal or no overlap between the form of this characteristic in males and females (e.g., the size of the sexually dimorphic nucleus of the preoptic area, which is three to eight times larger in male rats compared to female rats, Swaab, 1995). For most documented sex differences in the brain, however, and in particular in regions involved in behavior, emotion, and cognition, there is a considerable overlap between the distributions of the two sexes.

That’s all very accurate, and you could say it is Joel’s main point, and an important and completely valid one. But Joel goes on, immediately afterwards, to claim that:

It follows, that individuals may have the “female” or the “male” form for any of these non-dimorphic characteristics.

What? No! It totally doesn’t follow at all. The previous sentences clearly establish that “female” or “male” forms DO NOT EXIST.

Joel continues: The question then becomes whether the brain of a given individual is homogenous or heterogeneous with respect to the “male/female” type of its brain characteristics.

This is the straw man (or straw person) argument. If most sex differences are not really dimorphic, then the idea of scoring individual regions as either male or female rests on a conceptual fallacy. It is thus no surprise that hardly any individuals meet this measure of internal consistency. The way the experiment is framed practically guarantees it.

The main problem with this framing is that it ignores the fact that most of the variation in the measured parameters has nothing to do with sex. This is the same situation as with height. If there were no males or females, humans would still show a normal, bell-shaped distribution in height. The sex difference acts on top of that, making males a little taller and females a little shorter, on average, thus creating two, overlapping distributions.

We can’t predict a person’s height from their sex, or vice versa, because there are not two distinct forms. What we can say is that any given woman – whatever her actual height – would probably have been a little taller if she were a man (but otherwise genetically unchanged).

I’m going to betray my true nerd background here and use an analogy from Dungeons and Dragons. In this game, each player creates a character – a Fighter or Wizard, or Thief, etc. This process begins by defining values for six different traits: Strength, Intelligence, Wisdom, Dexterity, Constitution and Charisma. Each of these values is determined by rolling three 6-sided dice and adding up the values. Because there are lots of combinations that can give values near the average and few that can give values at the extremes, this creates a normal distribution with a range from 3-18 and a mean value of 10.5, for each trait. So, you’ll have lots of values at 10 or 11, slightly fewer at 9 and 12, and so on, until we get to very few at 3 or 18.

(Figure credit: http://io9.gizmodo.com/the-24-most-embarrassing-dungeons-dragons-character-1524448977)

Now, imagine that after we roll these values we apply a sex modifier. A small difference, up or down, on various traits, depending on whether the character is male or female. (In the game, this is done depending on whether you are an elf, dwarf, hobbit, half-orc, etc.). Let’s say +1 to strength for males and -1 to strength for females. All that would do is shift the mean of the males to 11.5 and the mean of females to 9.5. There would still be a huge overlap. If you define 19 (the only point at the high end with no overlap) as the “male form”, then very few males would show it – the ones who do would have to have started at 18, just by chance.

The same applies in the brain. Say there is some region that is slightly larger in males than in females, such that the distribution in males extends at the extreme end to sizes beyond that seen in any females. If a given male shows such a value, this does not mean that that part of their brain has been extremely masculinised. It is much more likely that they simply started at that end of the distribution to begin with, independent of any sex effects. Being male just shifted it a little further. Likewise, those bits of the brain that are in the middle of the distribution are not necessarily any less masculinised. They probably just started nearer that value.

We have no idea what factors at a cellular level give rise to the primary distribution in size of different brain regions between individual humans and we don’t know which factors give rise to the sex difference – these could be the same or different. The overall size distribution could mainly reflect, say cell number, while the sex difference could reflect growth of nerve fibres (dendrites) and number of synaptic connections. In fact, in animals, different cellular-level parameters are affected by sex in different parts of the brain. How these relate to crude measures of overall volume of different brain regions is largely unknown. More importantly, how the volume of various brain regions relates to differences in their function is almost completely unknown.

So, from that perspective, the imaging results of Joel et al do not support either an anatomical or a functional male-female brain mosaic. Nor do they contradict the idea that there exists a “male” brain and a “female” brain, overall. In fact, the very same data strongly support that idea, when analysed in a multivariate fashion (taking multiple differences into account at the same time).

In my original blog, I stated:

A quick look at panel E of the figure shows that… most of the female brains showed a mostly female pattern (lots of pink) while most of the male brains showed a mostly male pattern (lots of blue – don’t blame me, I didn’t pick the colours!). 

The group differences are clear and highly significant. And even if very few of the males or females are at the extreme end of the distribution for all ten of these regions, the overall pattern suggests that you could build a very good classifier from the volumes of these ten regions taken together, which would be quite successful at predicting whether a given brain scan came from a male or a female. Indeed, this would have been a far more objective test of whether MRI volumetric differences between male and female brains are categorical or dimensional.

This is exactly what three separate groups rapidly did in response to Joel’s paper. Rosenblatt illustrated very clearly how, combining information from even only two variables at the same time can allow you to distinguish between groups (given the right conditions), even when the distributions overlap considerably for each variable alone. (This is just meant to illustrate the general method).

Using that kind of technique, and the data from the paper itself, Del Giudice and colleagues built a classifier that was 69-77% accurate in distinguishing male from female brains. Rosenblatt built one that was about 80% accurate. And Chekroud et al, using a different dataset, built a classifier that was 90-95% accurate in distinguishing male and female brains, when tested on a separate sample.

Rosenblatt summed up the general conclusion:

Given our empirical evidence and the multivariate intuition depicted above, we cannot help but disagree with the concluding statement in the abstract of Joel et al. (1), “. . . human brains do not belong to one of two distinct categories: male brain/female brain” or their statement that “. . .brains do not fall into two classes, one typical of males and the other typical of females.. . .” A simple multivariate analysis using the same data suggests quite the opposite: Brains are indeed typically male or typically female.

Now, what this means is far from clear. First of all, it is important to emphasise that these anatomical differences are slight, compared to the overall underlying variation. Second, we don’t know what the functional consequences of the differences are. None of the observed differences has been linked to any sex difference in a behavioural or psychological trait. Third, the existence of brain differences in adults, by themselves, does not prove that they are all innate. (Though it should be noted that sex differences in brain size and structure are observed even in very young infants).

Fourth, even if all the brain differences were clearly innate, this would not prove that all observed sex differences in behaviour are entirely biological and innate in origin. The most likely scenario is that some behavioural traits do show innate differences between the sexes, on average, but that these are (probably quite strongly) amplified and reinforced by nurture and by cultural expectations.

And finally, (and I can’t believe this has to be said), the existence of male-typical and female-typical forms does not make one better than the other. Nor do group average differences justify sexist expectations of individuals – this is the very definition of prejudice. The fact that there are small, but systematic average differences between the brains of men and women should not in any way undermine the continuing drive for equal treatment, equal opportunity and equal rights. We don’t all have to be intersex to be treated equally.