05 December 2013

About that PNAS Article: Journalism and Neurosexism

"...to those interested in gender equality there is nothing at all frightening about good science. It is only carelessly done science, or poorly interpreted science, or the neurosexism it feeds, that creates cause for concern."

Cordelia Fine Delusions: How Our Minds, Society, and Neurosexism Create Difference (Norton 2010), 238.

A couple of days ago the Critical Neuroscience Facebook linked to a splashy article in The Guardian entitled "Male and Female Brains Wired Differently, Scans Reveal". The Independent (hat-tip Cordelia Fine) followed up with an article titled "The Hardwired Difference between Male and Female Brains could explain Why Men are 'better at map reading'". Both articles referred to a recently published academic article appearing in the Proceedings of the National Academy of Sciences under the lofty title "Sex differences in the structural connectome of the human brain". (Hardwired has a rather interesting history as a keyword in neuroscience.)

In the original paper, the authors use a large sample population sufficient to "elucidate sex differences in networks reliably" and from there they segment their sample into three distinct age groups: 8-13 (n= 158 females and 156 males); 13-17 (180 females - 131 males), and, 17-22 (183 females and 141 males). The authors write: "these groups correspond roughly to the developmental stages of childhood, adolescence, and young adulthood" (page 2). Thereafter, the study explored the so-called human connectome of men and women (connectome is a new neuro-culture buzzword which appears historically as a rather new iteration of the old holism vs localism argument that has traditionally plagued the short history of neurology). Using "fiber tractography", the authors describe gender differences via a structural connectome, and thereafter they show many fine illustrations of wiring differences between men and women, and they then provide a series of boilerplate assessments about the ways these differences translate into behavioral practices. I'm always rather surprised by the tin-ear of authors of studies such as these - intuition? really?

Now while the journalists have been very excited to discover that men and women are different and that their brains are different (here comes boilerplate neurosexism), a whole number of rather interesting science studies question arise in the context of this particular paper. I don't want to violate copyright, so I'm not going to copy and paste in Figure 1 from the original paper. But it is a fascinating example of a series of ontological, technological, and statistical translations leading to a 'wiring diagram', i.e. an ostensibly metaphoric image standing in for a series of evidently absent but detectible and determinative differences. These differences in wiring are then cast into normative social frames and categories. Cordelia Fine does a pretty good job in this essay of unpacking those normative claims both in the paper and also in the the subsequent journalism, but there is a great STS paper waiting there also for someone to come along and look at the 'house of cards.' I'm guessing, but I think the real STS story awaits in the difference between functional connectomes and structural connectomes (I stretch but does this not conjure the old debates in anthropology between structuralism and functionalism?!).

Don't ignore that point, but let me add an observation about this paper (I'm thinking aloud here). The authors parcellated with the Deskian atlas using "FreeSurfer" to label the nodes of each subjects' brain into 95 regions of interest, "68 cortical" and "27 subcortical" (a source here). From there, the authors conjectured fiber tracks to the other regions of interest - they say probabilistically! It sounds reasonable. But here is the thing: if the brain differences between women and men are as great as the authors claim, then why do the authors start with a baseline assumption that their 95 regions of interest would be the same between the genders. In other words, why should the authors get to start with a uniform set of regions of interest for both genders, when they have claimed that they show that human brains are allegedly profoundly different. Now I could be mistaken here - the paper is very dense. But to me, the authors appear to have imagined a Platonic ideal brain connectome that is uni-sexed. One has to ask then: could there be 95 different regions of interest that show the brains are more alike then we thought - probabilistically that is! If not, then it would seem that their model doesn't really reflect the one with which they began. 

Like you, I can't wait for the headlines that say - "men and women, different, but not because of their brains."      

4 comments:

  1. Another sts story could: look how much work they do and how many actors they need to comprehend to stabilize the sex inscription ;)
    Thank you for your comment on this

    ReplyDelete
  2. Hi Stephen,

    "...if the brain differences between women and men are as great as the authors claim, then why do the authors start with a baseline assumption that their 95 regions of interest would be the same between the genders."

    This is an interesting point. I'm no neuroanatomist - I'm on the MR acquisition side of this equation - but it sounds sufficiently interesting to warrant investigation. Would you be willing to post your question on PubPeer, to complement some other issues that have been raised?

    https://pubpeer.com/publications/3CFCCE950D22E7560E9B07C8B63979

    It can be done anonymously.

    I'll be honest. Given the very public airing this paper has achieved, albeit mostly in the British press so far, I think it should get the most thorough post-publication peer review possible. (FWIW I'm Peer 1.)

    ReplyDelete
  3. Thanks for the suggestion. I feel reluctant to interject my historian's viewpoint into the debate. I do think there is some sort of circular reasoning afoot here, but I say that from a perhaps naive critical neuroscience position. It is also interesting to note that Charles Gross edited this particular article. Gross has made many significant contributions to the history of phrenology - and to me this article conjures many of the writings of the 19th century phrenologists, albeit updated in a spectacular fashion of imaging & statistical analyses. The authors should be asked to read and respond to Kurt Danziger's central insights about the construction of psychometric populations. http://www.princeton.edu/~cggross/ency_neuro_phren.pdf on About that PNAS Article: Journalism and Neurosexism

    ReplyDelete
    Replies
    1. Fair enough! I totally understand. If the authors respond to the PubPeer critique then I'll consider raising the point about defining the regions-of-interest (aka "seeds" for probabilistic tractography). And if that turns out to be the Achilles Heel I'll make sure you get the credit for it! (If not, it's on me.)

      Re. phrenology, that is indeed an interesting coincidence. Thanks for the link, I'll read it for my own enlightenment. Seems it would provide an entertaining re-framing of their paper if the gender difference turns out to have been a methodological mistake after all!

      Delete