Delusions of Gender (22 page)

And there’s still so much inequality to be explained! We need to press on, into the brain itself.

I
n 1915, the illustrious neurologist Dr. Charles L. Dana set out in the
New York Times
his professional opinion vis-à-vis the wisdom of women’s suffrage:

There are some fundamental differences between the bony and the nervous structures of women and men. The brain stem of woman is relatively larger; the brain mantle and basal ganglia are smaller; the upper half of the spinal cord is smaller, the lower half, which controls the pelvis and limbs, is much larger. These are structural differences which underlie definite differences in the two sexes. I do not say that they will prevent a woman from voting, but they will prevent her from ever becoming a man, and they point the way to the fact that woman’s efficiency lies in a special field and not that of political initiative or of judicial authority in a community’s organisation. There may be an answer to this assertion, but no one can deny that the mean weight of the O.T. and C.S. in a man is 42 and in a woman 38, or that there is a significant difference in the pelvic girdle.
¹

The passage of time has not borne out Dr. Dana’s promising idea that the neural circuitry involved in political initiative is located in the upper half of the spinal cord. Without even knowing where in the nervous system the ‘O.T.’ and the ‘C.S.’ are located, I am fairly confident that judicial savvy does not lie in the extra four units of them bequeathed to men. But, at the time, this argument
seemed plausible enough to be published in the
New York Times
. And who knows, perhaps it served to sway, or at least reinforce, opinion on the controversial subject of votes for women.

Today, we can easily recognise the prejudice behind the implications Dana drew from his neurological observations. But even as one hypothesis falls (‘
The connection between the spinal cord and the pelvis? You really think it involved in some important way?
’), another is there to take its place.

As an empirical endeavour, the neuroscience of sex differences began in earnest in the mid-nineteenth century. The findings of Victorian scientists and medical men of the day were ‘a key source of … opposition’ to women’s suffrage and equal access to higher education, notes Yale University historian of science Cynthia Russett.
²
Certainly, as she documents, they improved on the ideas of their predecessors who presented evidence to argue, for example, that women’s intellectual inferiority compared with white men could be seen in the angle of their faces. As asserted by a late-eighteenth-century expert in the measurement of facial verticality, ‘The idea of stupidity is associated, even by the vulgar, with the elongation of the snout, which necessarily lowers the facial line.’

Women did not fare well in such assessments, and were reported to share with the ‘primitive’ and ‘savage’ races an unfortunate lack of facial verticality. It was not long, though, before this crude measure was jettisoned in favour of the more sophisticated cephalic index, namely, the ratio of skull length to skull breadth. The cephalic index was, for a while, thought to be a promising indicator of mental capacity, but was reluctantly abandoned when it became clear that the head shapes of ‘inferior’ social groups, including women, did not segregate neatly from those of ‘superior’ groups. It was later believed, as noted earlier, that women’s intellectual inferiority stemmed from their smaller and lighter brains. And when it became unavoidably evident that one could be slight of brain but substantial of intellect (and vice versa), the hypothesis was reluctantly abandoned, and the brain searched more intimately for the neural correlates of female inferiority.
³

The tape measures and weighing scales of the Victorian brain scientists have been supplanted by powerful neuroimaging technologies, but there is still a lesson to be learned from historical examples such as these. State-of-the-art brain scanners offer us unprecedented information about the structure and working of the brain. But don’t forget that, once, wrapping a tape measure around the head was considered modern and sophisticated, and it’s important not to fall into the same old traps. As we’ll see in later chapters, although certain popular commentators make it seem effortlessly easy, the sheer complexity of the brain makes interpreting and understanding the meaning of any sex differences we find in the brain a very difficult task. But the first, and perhaps surprising, issue in sex differences research is that of knowing which differences are real and which, like the initially promising cephalic index, are flukes or spurious.

In the statistical jargon used in psychology,
p
refers to the probability that the difference you see between two groups (of introverts and extroverts, say, or males and females) could have occurred by chance. As a general rule, psychologists report a difference between two groups as ‘significant’ if the probability that it could have occurred by chance is 1 in 20, or less. The possibility of getting significant results by chance is a problem in any area of research, but it’s particularly acute for sex differences research. Suppose, for example, you’re a neuroscientist interested in what parts of the brain are involved in mind reading. You get fifteen participants into a scanner and ask them to guess the emotion of people in photographs. Since you have both males and females in your group, you run a quick check to ensure that the two groups’ brains respond in the same way. They do. What do you do next? Most likely, you publish your results without mentioning gender at all in your report (except to note the number of male and female participants). What you don’t do is publish your findings with the title ‘No Sex Differences in Neural Circuitry Involved
in Understanding Others’ Minds’. This is perfectly reasonable. After all, you weren’t looking for gender difference and there were only small numbers of each sex in your study. But remember that even if males and females, overall, respond the same way on a task, five percent of studies investigating this question will throw up a ‘significant’ difference between the sexes by chance. As Hines has explained, sex is ‘easily assessed, routinely evaluated, and not always reported. Because it is more interesting to find a difference than to find no difference, the 19 failures to observe a difference between men and women go unreported, whereas the 1 in 20 finding of a difference is likely to be published.’
⁴
This contributes to the so-called file-drawer phenomenon, whereby studies that
do
find sex differences get published, but those that don’t languish unpublished and unseen in a researcher’s file drawer.

Neuroimaging studies of sex differences are certainly not exempt from this problem. It’s important to realise that the patches of colour you see on brain scans don’t actually show brain activity. Although it may seem as though fMRI and PET enable you to see a snapshot of the brain at work (or, as popular writers Allan and Barbara Pease claim, ‘to see your brain operating live on a television screen’),
⁵
this simply isn’t the case. ‘Unfortunately, these pretty pictures hide the sausage factory’, as one neurologist put it.
⁶
fMRI doesn’t measure neuronal activity directly. Instead, it uses a proxy: changes in blood oxygen levels. (PET uses a radioactive tracer isotope, which attaches itself to glucose or water molecules, to indirectly track blood flow.) Busier neurons need more oxygen and (after an initial dip) active brain regions have higher levels of oxygenated blood, because blood flow to that area increases. The oxygen is carried by the haemoglobin in red blood cells, and haemoglobin has slightly different magnetic qualities depending on how much oxygen it’s carrying. This creates a signal in the scanner (which pulses a magnetic field on and off). Neuroscientists then compare the difference in blood flow in brain regions during the task they’re interested in, with blood flow during a control task or rest state. (Ideally, the control task involves everything the
experimental task entails – button pressing, word reading and so on – except for the psychological process you’re particularly interested in.) Researchers test for significant differences in blood flow in various locations of the brain regions during the two tasks, and if tests indicate that it
is
significant, a blob of colour is placed at the appropriate location on the picture of the brain.
⁷

In other words, those coloured spots on the brain represent statistical significance at the end of several stages of complicated analysis – which means there’s plenty of scope for spurious findings of sex differences in neuroimaging research. Many studies use both male and female participants. The researchers may well check for gender differences but, if none are found, make no mention of it in the published report. What’s more, because imaging is so expensive, a small number of participants is the rule rather than the exception, and small neuroimaging studies may be especially unreliable, because nuisance variables (like breathing rate and caffeine intake, or even menstrual cycle in women) can dramatically change the imaging signal without having any effect on behaviour.
⁸

Neuroimaging also brings with it the teething problems of a technology that’s still in its infancy. There are healthy controversies in the neuroscientific community regarding how statistical analysis should best be done. There’s nothing wrong with this in itself, of course. But it is a little disconcerting that neuroimagers are now finding that reported sex differences in brain activation haven’t been put to adequate statistical testing, or can come and go depending on how the analysis is done, or can fail to generalise to a distinct but similar task within a second group of men and women, or that the kind of analyses used to establish sex differences in brain activation can also ‘discover’ brain activation differences between randomly created groups (matched on sex, performance and obvious demographic characteristics).
⁹
For all these reasons, it’s critical not to place too much faith in a single study that shows sex differences but instead to look for a consistent pattern.

The importance of this becomes very clear when we consider the influence of the nonstick theory of Norman Geschwind and
his colleagues who, you’ll recall, suggested that high levels of foetal testosterone in males result in a left hemisphere that is underdeveloped relative to the right. This led to the idea that male brains are more lateralised (or specialised) than female brains, on average. That is, males tend to stick to their shrivelled left hemisphere when grunting monosyllables and use the roomier right hemisphere when processing visuospatial stimuli. By contrast, women’s brains are supposedly less lateralised: during both language and visuospatial tasks, women tend to use both sides of the brain.

Now this is not regarded as an unimportant ‘I say
to-may-to
, you say
to-mah-to
’ sort of difference within the scientific community. A specialised, keep-it-local structure is supposedly what underpins male superiority on certain visuospatial tasks. By contrast, the more collaborative ‘Left? Right? Hey, we’re all in this together’ approach of the female brain supposedly explains their superior verbal skills, because they can more easily integrate information processed in different parts of the brain. The other side of the coin, however, is a more cramped design for spatial processing. Purportedly, this is because there is more competition between verbal and spatial circuits in the female, bilateral brain, which also, supposedly, has a relatively thicker and more bulbous corpus callosum, which is the bundle of neurons that connects the two hemispheres. This superior corpus callosum (especially a part of it called the splenium) supposedly enables faster and more efficient cross-talk between the hemispheres.
¹⁰

There is something a little curious about the relationship between (some, at least, in) the scientific community and the idea of greater male lateralisation. It is a bit like that of the wife who determinedly overlooks the plentiful signs that her husband is shifty, unreliable and worthless, while inflating the significance of occasional dependable behaviour. Even in the 1980s, researchers were pointing out major flaws and yet, as Ruth Bleier noted in 1986, even ‘devastating criticisms by two leaders in the field of cognitive sex differences and lateralization have done nothing to stem the flood of research’.
¹¹

Neuroimaging has provided a new way for researchers to show their loyalty to the hypothesis. Yet as neuroscientist Iris Sommer and her colleagues have shown, despite the new frisson of excitement wrought by the introduction of new technology, the data are as faithless as ever. Sommer and her colleagues reviewed (twice) all functional imaging studies of language lateralisation in a meta-analysis. (A meta-analysis is a statistical technique for putting together all studies that have investigated a particular question, taking into account the size of the study, to get a more accurate overall picture of the empirical situation.) The first meta-analysis (in 2004) put together data from more than 800 participants, and the second, in 2008, included more than 2,000 participants. In both meta-analyses they found ‘no significant sex difference in functional language lateralization’.
¹²
Interestingly, they also found that studies that found sex differences tended to have smaller sample sizes than those that didn’t. As Sommer and colleagues suggest, this may be a sign that the file-drawer phenomenon is at work, with biased reporting of chance findings from smaller studies.