Thursday, August 11, 2011

A Re-Post: Studying Sex Differences in Science: A Story

This is the third post in the series, originally from here.

In 1995 a wife-husband research team published the first study appearing to show that men and women process language differently in their brains. Sally and Bennett Shaywitz used functional MRI to study the brains of nineteen women and nineteen men during three different language tasks.

One of the tasks, identifying rhymes, showed gender differences in the relative activation levels of the brain. Lise Eliot in Pink Brain. Blue Brain. writes (pp. 185-6): exhibited strong activation of the lower portion of the left frontal lobe, while women tended to activate the same frontal area but on both sides of the brain. Of the nineteen women, eleven exhibited this bilateral pattern and eight activated just the left hemisphere (like men). So the results of this study seemed to indicate that in processing language, or at least during this particular rhyming task, women were more likely to use both hemispheres while men used exclusively the left hemisphere. As one of the first reports to find a sex difference by using functional MRI, this study got a lot of press. An article in the New York Times Science section promptly declared: "Men and Women Use Brain Differently, Study Discovers," and the findings continue to be highlighted even in recent popular works.
I remember the publication of those popularizations and the mileage they went. Like from here to sun and back again! Take into account the fresh interest in the genome study and the soil was well prepared for something like this to be interpreted as evidence of permanent, stable and innate differences between men and women. In many places it was.

What happened next in this interesting field, you might ask. Two things. On the one hand the research in the field continued. On the other hand, several popularizers harnessed the idea that men and women have totally different brains, that this difference is innate, and that the world should be organized to respect those differences. Part of that organizing was the idea that boys and girls should be educated separately and with different methods. Michael Gurian and Leonard Sax are famous advocates of innate sex differences as the basis for single-sex schooling, and Gurian, in particular, keeps appealing to the fMRI and PET scans to make his point*.

Sadly, what has happened in studies which use those methods has weakened the arguments of guyz like Gurian and Sax. It turns out that the way the brain looks in those scans can change based on how it is used. This means that the kinds of differences that 1995 study found don't necessarily tell us anything at all about the innateness of the observed differences in use.

Even more sadly for Gurian and Sax, later studies failed to replicate Shaywitzes' original finding. Eliot again (pp. 186-7):

Like any good research, the Shaywitzes' study inspired many attempts at replication. By 2008, twenty-six comparable brain-imaging studies were available for Iris Sommer and her colleagues at the University of Utrecht in the Netherlands to synthesize using meta-analysis.

Their overall conclusion: there's no sex difference in language processing. While some studies reported results similar to the Shaywitzes', others did not. Some even found that women processed language more strongly on the left side. When you put all the findings together, it's a wash; there is no significant difference in the way men's and women's right and left hemispheres are activated by language.
All this has to do with the idea that brain lateralization might differ between men and women in language use. This doesn't seem to be the case. But no worries! We still get a lot of popularizations based on exactly that idea, even though it has now been removed from the relevant university-level textbooks.

As Cordelia Fine puts it in Delusions of Gender, after discussing the above study and some additional research (p. 138):
So let us , with healthy skepticism, summarize all of this as clearly as we can. Nonexistent sex differences in language lateralization, mediated by nonexistent sex differences in corpus callosum structure, are widely believed to explain nonexistent sex differences in language skills.

Why does this story matter? Because the popularizations of research such as the Shaywitz study were seeds to the new single-sex education movement and also because of statements like this one (from Fine, p. 139):

For example, a consensus statement titled "The Science of Sex Differences in Science and Mathematics" links female "interhemispheric connectivity" to an advantage in language skills and male within-hemisphere connectivity to superiority in "tasks requiring focal activation of the visual association cortex", that is, visuospatial tasks.
That women don't appear to have any language skills advantage in interhemispheric connectivity is worth remembering.

The morale of this story? Perhaps the fact that it's one of many similar ones, as far as I can see. Almost any study finding sex differences will be given powers it should not have, as the final and eternal explanation of all observed gender differences. Whole edifices will be built on that one finding. When it's ultimately accepted as a false lead, the process begins anew with some other study. All this has real costs, psychological as well as monetary, which could be avoided if popularizers took more care and if researchers themselves played a role in explaining the limited role of such findings.
Cordelia Fine's Delusions of Gender discusses the imprecision of these imaging techniques and the meaning of the "blobs" one sees. They are nowhere near as easy to interpret as one tends to assume.