11 See, for example, Mark Liberman’s discussion of Leonard Sax’s use of data on rat vision to draw conclusions about human gender difference and single-sex schooling (http://itre.cis.upenn.edu/~myl/languagelog/archives/003473.html).
12 (Hines, 2004), p. 82.
13 These include different timing, different physiological effects and different hormonal mechanisms. For example, while injecting testosterone into female rats soon after birth disrupts the estrous cycle (the rat version of the menstrual cycle), prenatal testosterone during the equivalent critical period in humans and other primates doesn’t have the same disrupting effect. Also, the role of testicular hormones converted to oestrogens in sexual differentiation may be different in rats and primates.
14 For example, Wallen argues that ‘the dominant rat and mouse models of sexual differentiation seem unlikely to apply to human sexual differentiation.’ (Wallen, 2005), p. 8.
15 See (Wallen, 1996). Referring to frequency of ‘threat’ behaviour such as baring teeth and staring.
16 For high doses of prenatal testosterone treatment, late in gestation. Earlier in gestation, no effect of the same high dose of testosterone on rough play is seen. In both rats and rhesus monkeys, prenatal androgen treatment also affects sexual behaviour, for example, degree of mounting. See (Wallen, 2005).
17 Early blocking using flutamide reduces the masculinisation of the genitalia and results in rough play and mounting intermediate between male and female behaviour. Late blocking reduces penis length, has no effect on rough play (even though in females it is testosterone late in gestation that appears to be important in influencing rough play) and actually increases mounting behaviour, which is opposite of what one would expect (Wallen, 2005).
18 Described in (De Vries, 2004).
19 Quoted in (Kolata, 1995), para. 22. Gorski adds that ‘nothing like it has been shown in humans.’
20 (De Vries, 2004), p. 1064.
21 For example, a book for parents published by the Gurian Institute claims that ‘[w]ithout the testosterone hits received in utero by her male counterparts, her brain continued on the female default path, providing specialized circuitry for communication, emotional memory, and social connection.’ (Gurian Institute, Bering, & Goldberg, 2009), p. 32.
22 For valuable discussions of the problems with the orthodox view of the organisational/activational hypothesis, see (Breedlove, Cooke, & Jordan, 1999; Fausto-Sterling, 2000; Kaplan & Rogers, 2003; Moore, 2002; Rogers, 1999).
23 (Moore, 2002), pp. 65 and 66.
24 (Moore, Dou, & Juraska, 1992).
25 (Moore, 2002), p. 65.
26 (Barnett & Rivers, 2004), p. 200. For criticism of the lack of impact of this important work in the scientific community, see (Kaplan & Rogers, 2003), pp. 53–56.
27 (Geschwind & Behan, 1982).
28 Quoted in (Kolata, 1983), p. 1312.
29 Note, according to the model, extremely high levels of foetal testosterone will have detrimental effects on right-hemisphere development, and thus visuospatial function. As several researchers and commentators have pointed out, there is little in the way of evidence for the model, yet despite this it enjoys tremendous scientific and popular appeal and influence. In particular, Ruth Bleier has made an excellent critique of the model, and her criticisms and data have also been well summarised by Carol Tavris (Bleier, 1986; Tavris, 1992). For further critiques see also (Fausto-Sterling, 1985; Grossi, 2008; Nash & Grossi, 2007; Rogers, 1999). A comprehensive account of the data with regard to the Geschwind-Behan-Galaburda model, as it is more formally known, which proposes a link among foetal testosterone, left-handedness, giftedness and immune-system functioning, concluded that ‘[a]n overall evaluation of the model suggests that it is not well supported by empirical evidence and that in the case of several key theoretical areas, the evidence that does exist is inconsistent with the theory.’ (Bryden, McManus, & Bulman-Fleming, 1994), p. 103.
30 (Bleier, 1986).
31 (Gilmore et al., 2007), who found that, contrary to adults and older children, in neonates of both sexes the left hemisphere is larger than the right. See also (Nash & Grossi, 2007), p. 15, for discussion of lack of support for the model in studies of adult brains. This is in contrast to research with rats, which has demonstrated the relatively larger right hemisphere in males and the dependence of this on neonatal testosterone (Diamond, 1991). Note that Diamond’s summary of this work also points to the importance of experiential factors in hemisphere asymmetry. I do not know whether researchers have investigated whether the effect of neonatal testosterone on cerebral lateralisation occurs directly and/or via the different social experiences triggered by higher neonatal testosterone – a possibility suggested by the work of Celia Moore described earlier.
32 As Baron-Cohen puts it, ‘the more you have of this special substance [testosterone, especially early in development], the more your brain is tuned into systems and the less your brain is tuned into emotional relationships.’ (Baron-Cohen, 2003), p. 105. It’s not clear that ‘extreme male’ is a good description of the profile of people with autism. You’ll remember from the first part of the book that empathy can be either cognitive (mind reading) or affective (sympathy). In seminal work, Simon Baron-Cohen showed that people with autism struggle with cognitive empathy, that is, they can’t seem to read other people’s intentions, beliefs and feelings with the intuitive ease that most of us enjoy (Baron-Cohen, 1997). Yet several strands of research now suggest that people with autism don’t lack affective empathy (Blair, 1996; Dziobek et al., 2008; Rogers et al., 2007). This is problematic for Baron-Cohen’s thesis because, as Levy has pointed out in (Levy, 2004), according to Baron-Cohen (see [Baron-Cohen, 2003], p. 120) the typical male profile is the precise opposite. Baron-Cohen suggests that men’s empathy disadvantage is greater for affective, rather than cognitive, empathy, the latter being vital for success in domains of predominantly male achievement. (Think how badly a poor mind reader would get on in business, politics or law.) It’s also worth noting the possibility that high foetal testosterone ‘reduces the threshold at which autistic symptoms manifest’, rather than causing autistic symptoms directly, as suggested by (Skuse, 2009), p. 33.
10. IN ‘THE DARKNESS OF THE WOMB’ (AND THE FIRST FEW HOURS IN THE LIGHT)
1 (Gurian Institute, Bering, & Goldberg, 2009), pp. 18 and 19.
2 Remember Celia Moore’s work, which found that early testosterone affected the mother rat’s behaviour. Foetal testosterone levels might affect, say, the physical appearance of the child in some way that influences how the child is treated (for example, by making the face more masculine). It’s also possible that parents who have children with higher levels of foetal testosterone tend to be different from those who don’t, in some way that affects the environment they provide to their children.
3 With respect to the use of maternal testosterone (mT), one clinical study that measured foetal testosterone directly did find that it correlated with mT (Gitau et al., 2005). However, as noted by van de Beek et al. mT levels are not higher in women carrying boys than in those carrying girls, which suggests ‘that maternal serum androgen levels are not a clear reflection of the actual exposure of the fetus to these hormones.’ (van de Beek, et al., 2004), p. 664. Also, testosterone can only act on the brain if it is free (that is, if the testosterone is not bound to another molecule). One way this can be indirectly assessed is to also measure levels of SHBG (sex hormone binding globulin). The more SHBG, the less free testosterone is likely to be available. The two studies that used maternal serum measured both. One found a correlation between a sex-typed behaviour measure and mT but not SHBG (Hines et al., 2002). The other found a correlation with SHBG but not mT (Udry, 2000). There therefore seems some uncertainty as to which (if either) is the appropriate proxy for foetal testosterone (fT) exposure. For amniotic testosterone (aT) ‘there is no direct evidence to either support or contradict’ the assumption that aT is correlated with the levels of testosterone acting on the foetal brain
(Knickmeyer, Wheelwright et al., 2005), p. 521. (van de Beek et al., 2004) suggest aT as the best index of fT exposure, but they also acknowledge the lack of much understanding of the relationship between levels of testosterone in the amniotic fluid – the main source of which is foetal urine – and in the foetal blood. Van de Beek and colleagues note that ‘there is no hard evidence of a direct relationship between amniotic testosterone and fetal serum testosterone.’ (van de Beek et al., 2009), p. 8. Finally, the use of the digit ratio as a marker of prenatal testosterone exposure is controversial and lacks clear empirical support. For review see (McIntyre, 2006). One researcher has complained that ‘[t]he lightheartedness of using certain biological markers in adulthood as indicators of prenatal androgen exposure is not warranted.’ (Gooren, 2006), p. 599. Because digit ratio seems to be the most controversial index of prenatal androgen exposure within this field of interest, I don’t attempt here to provide anything like a comprehensive account of research findings using this technique.
4 I am very grateful to Giordana Grossi for her helpful discussions of the following literature.
5 It’s important that correlations are seen within sex. Otherwise gender socialisation might create psychological differences that then correlate with foetal testosterone for the simple reason that boys have higher foetal testosterone than girls.
6 (Lutchmaya, Baron-Cohen, & Ragatt, 2002). The data from this study are not completely straightforward. For boys and girls together, amniotic testosterone (aT) did indeed correlate negatively and linearly with frequency of eye contact. That is, children with high aT had lower eye contact frequency than children with low aT. However, there was also a quadratic relationship meaning that eye contact frequency decreased with increasing aT in the low aT range (as predicted), but increased with increasing aT in the high aT range. This same pattern appeared when looking at boys separately. In girls only, no relationship at all was seen between aT and eye contact frequency. These data, then, are not consistent with the claim that ‘the higher your levels of pre-natal testosterone, the less eye contact you now make’ (Baron-Cohen, 2003), p. 101. It should also be noted that the methodology of this study was rather odd. Different toys were being presented to the infant during the experimental procedure, which could have differentially distracted some infants more than others. It’s also noteworthy that what was measured was frequency of eye contact (actually, it was not even eye contact, but looking ‘at the face region of the parent’, p. 329) rather than duration of eye contact, although the two were correlated.
7 (Knickmeyer, Baron-Cohen et al., 2005). Multiple regression found that foetal testosterone predicted social relationships score independently of sex. However, within each sex no significant relationships were observed. It’s also worth noting that the difference between boys and girls on this scale was not statistically significant (although there was a trend, with a moderate size of effect) and previous research with the same scale in six-year-olds found no sex differences. So even if amniotic testosterone does indeed correlate with the skills this questionnaire measures, there is not yet convincing evidence that males and females actually differ on them.
8 (Knickmeyer, Baron-Cohen et al., 2006). In this study, four-year-old children watched animations involving shapes. In two of the films, the behaviour of the shapes evokes the perception that they are acting on the basis of mental states. Children were interviewed about what was going on in the film. This involved extensive questioning by an interviewer (see p. 285). There is no mention of this interviewer being blind to experimental hypothesis or amniotic testosterone (aT) status, which seems problematic because an experimenter could unintentionally respond more encouragingly to girls, for example. Use of mental state (expressing character’s beliefs, thoughts, intentions, etc.) and affective state terms (e.g., happy, sad) did not correlate with aT for all children, or within boys or girls. Although girls used significantly more affective state terms than boys, the sexes did not differ in mental state term use. For intentional propositions (e.g., ‘the triangle knew the way’), aT was the only significant predictor in the hierarchical regression analysis. However, within females there was no correlation between aT and use of intentional propositions, but there was a correlation in males. The sex difference in intentional propositions use was at trend level. Boys used more neutral propositions than girls (e.g., ‘There’s a small triangle’). But although aT was the only significant predictor of neutral propositions, aT did not correlate with neutral propositions within boys and girls separately. All in all, the number of negative findings do not make for compelling evidence for the thesis that aT levels are related to the tendency to attribute mental states to animated shapes, and that this tendency reliably differs in males and females.
9 (Chapman et al., 2006) For the Empathy Quotient (EQ)–child version, the only significant predictor in the hierarchical regression analysis was sex. In other words, amniotic testosterone was unrelated to EQ and something other than amniotic testosterone accounts for the effect of sex on score. There was a within-sex negative correlation between amniotic testosterone and EQ score for boys but not girls.
10 For the Reading the Mind in the Eyes test–child version, data confirmed hypotheses. However, as noted in the text, performance did not significantly differ – indeed, the authors report that they have previously failed to find superior performance of girls on this task (Chapman et al., 2006), see p. 140. This, in itself, seems a bit problematic for Baron-Cohen’s thesis. A sex difference in a performance measure would be more convincing than maternal reports of sex differences.
11 Recently, Auyeung et al. (2009) reported correlations between amniotic testosterone and subclinical autistic traits, using two questionnaires. One of the questionnaires, the Autism Spectrum Quotient Child, was separable into subcomponents that included a mind-reading scale and a social skills scale. However, although these subscales both correlated with foetal testosterone, the authors do not present within-sex correlations.
12 (Voracek & Dressler, 2006), for example, found no relationship between digit ratio and either EQ score or Reading the Mind in the Eyes performance, in their large-scale study. As noted earlier, however, I do not attempt here to review the digit-ratio findings.
13 (Auyeung et al., 2006), p. S124.
14 (Levy, 2004), p. 319 citing Einstein quotations from H. L. Dreyfus & S. E. Dreyfus, Mind over Machine (New York: Macmillan, 1988), p. 41.
15 (Baron-Cohen, 2007), p. 161.
16 (Marton, Fensham, & Chaiklin, 1994). Both quotations on p. 467, from Yuan T. Lee and Konrad Lorenz.
17 (Houck, 2009), p. 66.
18 (Auyeung et al., 2006).
19 Baron-Cohen argues that systemising ‘needs an exact eye for detail, since it makes a world of difference if you confuse one input or operation for another.’ (Baron-Cohen, 2003), p. 64. However, it seems to me that one could just as plausibly argue that good empathising requires attention to detail, because otherwise you might, for example, fail to notice the important emotional leak that tells you what the other person is really feeling, or how you might be best able to make him or her feel better. In addition, the benefit of attention to detail would seem to depend on whether the right detail is being attended to. Focus on something irrelevant will not be helpful to understanding a system. And sometimes, as the earlier quotations from the Nobel Prize winners suggest, breakthroughs in understanding require a feel for the bigger picture, beyond the details of the component parts.
20 (van de Beek et al., 2009). There was an unexpected positive correlation between levels of amniotic progesterone (a hormone associated more strongly with females) and playing with boyish toys! The researchers suggest that this may be a spurious effect.
21 Speed of rotation did correlate positively in girls with amniotic testosterone (aT), but boys’ rotation speed seemed to get slower with increasing aT, and they performed no better than the girls (Grimshaw, Sitarenios, & Finegan, 1995). And as Hines points out, it is performance accuracy – which did
not relate to aT – on which a sex difference is normally seen (Hines, 2006a).
22 (Finegan, Niccols, & Sitarenios, 1992). No sex differences in performance were seen.
23 (Auyeung, Baron-Cohen, Ashwin, Knickmeyer, Taylor, & Hackett, 2009), the Block Design Test. No sex difference in performance was seen.
24 (Brosnan, 2006; Puts et al., 2008; Voracek & Dressler, 2006).
25 (Gurian Institute, Bering, & Goldberg, 2009), p. 35.
26 (Connellan et al., 2000).
27 (Sax, 2006), p. 19.
28 (Lawrence, 2006), p. 15.
29 (Baron-Cohen, 2007), p. 169.
30 (Nash & Grossi, 2007).
31 (Nash & Grossi, 2007), p. 9.
32 (Leeb & Rejskind, 2004), pp. 4 and 10, respectively.
33 The article itself states that ‘[c]are was taken not to film any information that might indicate the sex of the baby’ (p. 115), suggesting that such information was available. Additionally, in an interview with Edge magazine, Simon Baron-Cohen notes that sometimes Connellan did learn the sex of the baby because of clues such as congratulation cards (Edge, 2005a).
34 For example (Batki et al., 2000; Farroni et al., 2002). Regarding preference for motion, Philippe Rochat writes that ‘[i]nfants from birth tend to be more attentive to objects that move than to stationary objects. In devising experiments, researchers know that infants are much more engaged by dynamic compared to static displays.’ (Rochat, 2001), p. 107. The study looking at preference for eye gaze (versus eyes closed) in newborns was conducted by the same team as Connellan’s study, and may have used the same populations of newborns. (Connellan’s face was used as the stimulus for both studies.) Interestingly, this study found that newborn boys had no less of a preference for eye gaze than did girls.
Delusions of Gender Page 29