Delusions of gender

« previous post | next post »

That's the title of Cordelia Fine's new book, due out on August 30.

Some reviews: Katherine Bouton, "Peeling Away Theories on Gender and the Brain", NYT 8/23/2010; Robin McKie, "Male and female ability differences down to socialisation, not genetics", The Observer 8/15/2010; "Q&A: 'Delusions of Gender' author Cordelia Fine", USA Today 8/9/2010; Louise Grey, "New book leads 'backlash' against sexual stereotypes that say men are from Mars and women are from Venus", The Telegraph 8/16/2010.

Delusions of Gender joins Lise Eliot's Pink Brain, Blue Brain (published in May and due out in paperback on Sept. 2) in a backlash of experts against the "sex difference evangelists" — authors like Leonard Sax and Louann Brizendine, picked up in the popular press, and promoted by pundits like David Brooks.

Katherine Bouton in the NYT picks up on a passage in Fine's book that also caught my attention when I read it:

Dr. Baron-Cohen’s lab conducted research on infants who averaged a day and a half old, before any unconscious parental gender priming. Jennifer Connellan, one of Dr. Baron-Cohen’s graduate students, who conducted the study, showed mobiles and then her own face to the infants. The results showed that among the newborns the boys tended to look longer at mobiles, the girls at faces.

Dr. Fine dismantles the study, citing, among other design flaws, the fact that Ms. Connellan knew the sex of some of the babies. Because it was her face they were looking at and she was holding up the mobile, Dr. Fine says, she may have “inadvertently moved the mobile more when she held it up for boys, or looked more directly, or with wider eyes, for the girls.”

The discussion of Connellan et al. in Delusions of Gender takes up six pages (pp. 111-117). Fine raises a serious methodological objection, citing the elaborate precautions taken in some other studies of gender differences in newborn eye gaze to ensure that experimenters are blind to the sex of the subjects, and noting that "No such precautions were taken in Connellan's study".

A different version of this same point came up in recent discussions of events in Marc Hauser's lab (see here, here, here, here) where the coding of monkeys' reactions to various sound patterns was supposed to have been done blind to experimental conditions, but apparently wasn't.  I hasten to add that there is no suggestion whatsoever of fraud in Connellan's study, and a key difference is that Hauser's papers asserted that his experiments were done blind to the key independent variable, while Connellan's paper is clear on the fact that her experiment wasn't. (Or more precisely, that the coding was blind, but the experimental procedure itself wasn't.)  However, the potential seriousness of the methodological point is underlined by the fact that failure to adhere to blind coding procedures has apparently been viewed as a form of scientific misconduct in Hauser's case.

In my original post on the Hauser scandal, I wrote:

[The dependent variable] is subjectively coded, by direct observation or by inspection of video recordings. To be believable, the coding should be "blind" — that is, the coders should not know which experimental category they are coding.  […] Unfortunately, this is by no means the norm in all areas of experimental psychology; in fact I'd say that "blind" coding of behavior is the exception rather than the rule, though I don't have counts to show this.

I turned out to be wrong with respect to general practices in the literature using habituation/discrimination methods on infants — there are normally elaborate precautions to ensure that these experimental procedures are run blind — so that, for example, the caregiver doesn't somehow communicate the experimental condition to the subject sitting on her (or sometimes his) lap — and the video recordings are coded blind, so that the coders are not influenced by what they hope or expect the subjects' reactions will be.

But it remains true that this kind of care is not ubiquitous in experimental psychology, and so it's not shocking to find that in the Connellan baby-gaze experiment, coding was blind, but administration of the experiment was not (Jennifer Connellan, Simon Baron-Cohen, Sally Wheelwright, Anna Batki, and Jag Ahluwalia, "Sex Differences in Human Neonatal Social Perception", Infant Behavior & Development, 23:113-18, 2000 — free version here).

In fact, I didn't even remark on this aspect of the study when I discussed it at tedious length a couple of years ago ("Innate sex differences: science and public opinion", 6/20/2008). Instead, I focused on the fact that the sex differences were actually pretty small relative to within-sex variation — with the largest sex difference by far being the drop-out rate — and the fact that the "mobile", far from being the sort of thing that would predict later interest in mechanical devices (as some have suggested) was in fact a weirdly-scrambled face-like object, suggesting instead the conclusion that "Boy babies are innately somewhat more interested in transdimensional monsters than girl babies are":


  1. Mr Fnortner said,

    August 24, 2010 @ 9:03 am

    Knowing neither person, I am not mounting a defense of either. Yet your statement failure to adhere to blind coding procedures has apparently been viewed as a form of scientific misconduct in Hauser's case, is damning to Bouton by implication, in spite of her forthright disclosure of her methods. Hauser's sin was wanton rejection of blind coding….and coding with the results in mind.

    [(myl) I certainly did not, and do not, intend to suggest that any misconduct whatever was involved in the Connellan et al. experiment. In fact, I explicitly make the point that administering experiments (and even coding behavior) without being blind to the independent variable is well within the norms of (many subareas of) experimental psychology — which is why their paper could be (and should have been) published in a respected journal.

    Whether those norms should be changed is another question. My own opinion is that blind administration and coding is not always worth the trouble — and sometimes it's almost impossible. But it's certainly a control that should be considered, especially when the observed differences are fairly small, or when there's reason to wonder whether a "Clever Hans" sort of effect might be involved.

    The problem with the Connellan et al. experiment, in my opinion, was not its design or implementation but its (over-)interpretation, mostly not by the authors but by others. ]

  2. Julian said,

    August 24, 2010 @ 10:49 am

    Just to add some regional bias to the apparent gender bias, I originally read this with an Australian accent. I thought throughout that "mobiles" referred to "mobile phones" (i.e. what we would call cell phones), and they were trying to show that boys preferred mechanical/electronic gadgets to faces.

    It wasn't until I saw the photographs that I hastily retraced and re-parsed what was written.

  3. Blake Stacey said,

    August 24, 2010 @ 11:21 am

    Yes, that "mobile" totally looks like the squid from the end of Watchmen.

  4. Debbie said,

    August 24, 2010 @ 11:27 am

    Out on a limb here, but focal ability and muscle control develops faster in girls than boys unless I've been misinformed. Is it possible therefore that the mobiles, unless completely still as suggested in the blog, could have been easier for children with weaker muscles to look at. Did it have the same level of contrast as the human face in the study? Was it compairable in terms of size…? (My apologies if the answers to these questions are obvious by the accompanying images as I am blind.)

  5. groki said,

    August 24, 2010 @ 1:40 pm

    the conclusion that "Boy babies are innately somewhat more interested in transdimensional monsters than girl babies are"

    q.v. attendance at science fiction conventions. (or have they become more gender-balanced in recent years?)

  6. army1987 said,

    August 24, 2010 @ 1:56 pm

    @Julian: me too…

  7. Mick O said,

    August 24, 2010 @ 2:03 pm

    "Dr. Baron-Cohen’s lab conducted research on infants" …

    This was all just a prank by Ali G?

  8. Theophylact said,

    August 24, 2010 @ 2:17 pm

    They're cousins.

  9. Kragen Javier Sitaker said,

    August 24, 2010 @ 2:42 pm

    It makes a certain amount of sense to use the weird transdimensional monster; if you used, say, a mobile phone (as I also interpreted "mobile"), the result might be something like "babies look at shiny things in preference to faces" or "babies look at large things in preference to small things" or "babies look at things with sharp angles in preference to things with smooth angles" or "babies look at things with brightly-glowing LEDs".

    However, to reduce the sources of variation further, the JC face also ought to have been a flat photograph, rather than a moving face.

    [(myl) The "mobile" was an entirely sensible control, in that it was designed to have the same distribution of colors and local textures and even local visual features as JC's face. And the you could argue that the whole point of using a real face was that it was, well, a real face. But one of the problems with such experiments is that any particular set of choices (including whose face you use) might plausibly influence the results; and you'd like to be able to draw general conclusions about interest in "faces" vs. "non-faces", not about interest in JC vs. a particular strange oval picture with a ball on a string hanging from the middle of it.]

  10. What about Scientific Bases for Differences in Gender? said,

    August 24, 2010 @ 6:57 pm

    […] Language Log further discusses problems in the experimental set-up and conclusion of a study that claims to […]

  11. stephen said,

    August 24, 2010 @ 7:59 pm

    And the you could argue that the whole point of using a real face was that it was, well, a real face.

    I wonder how the results would have been different if different faces were used–very masculine, androgynous, etc.

    Dolly Parton, Nosferatu and Clint Eastwood might evoke very different reactions.

  12. Jonathan said,

    August 25, 2010 @ 12:29 am

    Julian, I'm Australian too, and I didn't think that the older meaning of "mobile" was any less common here.

  13. Tom Saylor said,

    August 25, 2010 @ 5:29 am

    Mark Liberman posted:

    (so that, for example, the caregivers doesn't somehow communicate the experimental condition to the subjects sitting on their lap)

    I was struck by the agreement issues in this clause, particularly in light of the “ostriches hide their head in the sand” discussion on LL a couple years back. I imagine the subject-verb disagreement is the unintentional result of an incomplete revision (pluralization) of the clause structure (motivated perhaps by discomfort with awkward “his or her” or singular “their” in the original?), but I wonder about the disagreement between plural “caregivers” and singular “lap.” Unintentional also, or do you prefer “their lap” to “their laps” here?

    [(myl) This one is an editing relic. I started with "the caregiver doesn't somehow communicate the experimental condition to the subject sitting in his or her lap". Then I changed the awkward "his or her" to "their". That didn't seem right either, so I made "caregiver" plural, but neglected to change the verb. Now I've given up and gone to "the caregiver … her (or occasionally his) lap".]

  14. Debbie said,

    August 25, 2010 @ 8:25 am

    @Tom, @MYL …'the experimental condition isn't somehow communicated to the subject sitting on the caregiver's lap?' I prefer the Elmer Fudd approach: I choose another word or phrasing!

  15. Debbie said,

    August 25, 2010 @ 9:27 am

    Speaking of proofreading, I meant to say I prefer the Porky Pig approach. Sorry to all Elmer Fudd fans.

  16. J. W. Brewer said,

    August 25, 2010 @ 1:46 pm

    Given that a lot of genetically-based physiological differences between the sexes are only activated, as it were, at puberty, I'm not sure how much you prove by arguing about whether certain cognitive differences (in the nonessentialist overlapping but distinct bell-curve sense) are or are not observable 36 hours after birth. That you can't get a good study sample of post-puberty humans who've been raised by incredibly egalitarian wolves doesn't prove that any differences which might appear at or after that stage are purely artifacts of socialization. This is the standard scientific problem of trying to build a theory based on something of uncertain relevance that you can actually measure rather than fess up that what you should want to be studying is too difficult to measure.

  17. HW said,

    August 25, 2010 @ 6:09 pm

    @JW Brewer

    As far as I can see, the only thing you can do is to disprove the hypothesis that "male and female brains are exactly the same and all behavioural differences between the sexes are cultural". This is a very silly hypothesis, but some people believe it for political reasons unfathomable to me, so disproving it is a worthwhile endeavour.

    If you ask me, the answer to "do male and female brains differ" is "Sure, why wouldn't they?" Ask yourself:

    a) Is there any biological reason why they need to be the same? (Answer: no, our genes can build different brains for different sexes the same way they build different bodies), and

    b) Is there any evolutionary advantage in them being different? (Answer: yes, for all sorts of reasons it's an evolutionary advantage to tailor your brain to your sex)

    So in the absence of any astounding evidence to the contrary, the idea of innate sexual differences in the human brain and mind seems to be the sensible default position.

  18. Around the Web | Savage Minds said,

    August 25, 2010 @ 10:02 pm

    […] Language Log has an appreciation and links to reviews. […]

  19. MJ said,

    August 26, 2010 @ 2:53 am


    "So in the absence of any astounding evidence to the contrary, the idea of innate sexual differences in the human brain and mind seems to be the sensible default position."

    That's incredibly sloppy argumentation. "I can imagine many good theories that do not contradict my basic premises" does not mean "Therefore my deduction should be the default lacking significant evidence."

    There are good reasons for the brains not to differ, or to be plastic such that any "initial" differences may be changed over time. (One reason for them not to differ is that a common default evolutionary position is that as little change as possible is made between types, or rather, as little change as can effect an advantage, such that there must be a compelling reason *for* difference, not for its absence. A good reason for plasticity would be events leading to skewed sex balances or other shifting circumstances requiring more "male" brains or "female" ones, presuming an innate difference; having to wait a generation for a switch in this balance would be a disadvantage.)

    There probably are systematic biological differences (though their significance, size, and spread is still in considerable question). Maybe there aren't and we're seeing various artifacts of other processes. The fact that you can make up just-so stories, or find parallels, doesn't make your position the default one. (There are indeed some sensible reasons to presume difference, but there are sensible and compelling ones not to. I tend to agree that there is some level of difference, but I doubt its significance to most questions of social relevance. But I wouldn't smugly declare that I happen to hold the sensible default that can only be overturned by "astounding" proof in either case.)

  20. John Ward said,

    August 26, 2010 @ 8:43 am

    "So in the absence of any astounding evidence to the contrary, the idea of innate sexual differences in the human brain and mind seems to be the sensible default position."

    This just goes to show the difference between the layman's assumption and the scientific hypothesis. If all you were interested in was whether male and female brains were different, if there was no middle ground, and you didn't care about proving it, that would be the sensible assumption. But in the end, it doesn't tell you very much, does it?

    Scientifically, you can't almost start with the most intuitive hypothesis. The most powerful hypotheses are the ones that are counterintuitive, that seem like they'd be easy to disprove but aren't.

    Besides, proving equality is equivalent to proving a negative (the absence of difference). You can't do it. You have to assume equality and fail to disprove it.

    So if you're determined to test your assumptions, the best way is to start with the counterintuitive assumption that male and female brains are identical and throw everything you've got at it to disprove it. If the difference is large, this should be fairly easy, after all.

    At least, this is how I remember it from my university classes.

  21. Daniel H said,

    August 26, 2010 @ 2:40 pm

    @J.W. Brewer
    What about perceived differences in the way young children play? The common perception is that boys want to run and jump and fight and play with trucks while girls play house and talk to their dolls, practicing to be social when they grow up. This is something that can be observed long before puberty, and Saxes and Brizendines use these apparent differences as proof of their ideas. If sex differences like these were only activated after puberty, then the fact that small children exhibit them would be more likely to prove that they are cultural artifacts instead of innate genetic predispositions. So there is some use to studies like these. The danger lies in the misapplication or over-interpretation of the results.

    The problem here lies in the fact that so many people assume the exact opposite of "male and female brains are exactly the same and all behavioural differences between the sexes are cultural," namely that no differences are cultural (and that's just a liberal agenda) and all differences are biological. Neither of these can be assumed outright, and both indeed sound silly to me. I would go even further than John Ward above, and say that we can't even assume that males and females are different at all. Our "common sense" can trick us very easily. The first step is to prove that there is actually a statistically significant difference between the behaviors of the sexes. Then, we have to be careful not to exaggerate a small difference and generalize it to an entire population. As myl et. al constantly point out, quite often the variation between sexes is far exceeded by that within sexes.

  22. chris said,

    August 26, 2010 @ 3:41 pm

    So in the absence of any astounding evidence to the contrary, the idea of innate sexual differences in the human brain and mind seems to be the sensible default position.

    But at the same time, for any *particular* observed difference, if there's an observed social cause that can produce that difference, it would violate Occam's Razor to presume that there is *also* a genetically preprogrammed cause of the same difference. At least, until the difference is observed in newborns who haven't yet been socialized.

    So there may well be some innate differences, but it's hard to say what they are. Even the tendency for the majority of males to be primarily sexually attracted to females and vice versa isn't necessarily one — since that's also the culturally approved orientation. (I think that that one is reasonably likely to be partly innate, because the adaptive function of it is so strong and obvious, but I could be wrong. Intuition is an unreliable guide to reality.)

  23. Debbie said,

    August 26, 2010 @ 5:29 pm

    Out of curiosity, where would we find these unsocialized infants? My only thoughts on this might be an orphanage where the ratio of males to females is the same both with the infants and the care providers. In North American society, women typically take maternity leave and provide care for the infant thus providing a social context for female gender roles. Personality develops before age one, the first social smile coming at about six weeks of age. Before birth, a hormonal wash bathes male babies. Those with lesser amounts exhibit more effeminate traits. Would it be reasonable to expect that this wash also impacts the brain? Is there not a difference in left and right brain thinking between the genders. The idea that society itself could influence gender and thinking patterns, assuming a difference, intrigues me. If we needed more of a specific gender a male could produce a higher ratio of XY to XX sperm or vice versa. Females on the other hand, could produce a hormonal wash for females or less of one for males. Just a thought. Since we're just beginning to unravel a new layer of the genetic code, I think we are a long way off from unraveling this one!

  24. Gender Differences, The Brain, and Testosterone « Sapien Games said,

    September 3, 2010 @ 2:39 pm

    […] the book so I’m going to make only a quick statement based upon the reviews I just read:  here, here, and […]

  25. Women’s Equality and Neurosexism | Everyday Biology said,

    September 6, 2010 @ 6:23 pm

    […] about Delusions of Gender, see the interview with Fine in USA Today's Science Fair blog, the review post at Language Log, and the article about Fine's book in the Guardian. Composite Image of Brain […]

RSS feed for comments on this post