You may download a copy of this article for private use, provided you leave this copyright warning attached. However, the distribution of multiple copies, electronically or in hard copy, or the use of this article for any commercial purpose, is a copyright violation.

[For references please consult the original paper]

Sexualities, Evolution & Gender, 2004, 6, 45–53

Keywords: Cognition, sex difference, hormones

The paper by Dr Tone Bleie Evolution, brains and the predicament of sex in human cognition, in Volume 5 Number 3 of this publication, argues against the general acceptance of evolutionary influences on sex differences in cognition, and more specifically, against past sexual division of labour as a source of such differences. Instead, it emphasizes the influence of individual life-history of the sexes as a major determinant. First I must point out that I cannot take credit for the suggestion that our hunting-gathering evolutionary history significantly shaped the behaviour of modern men and women. This is a widely accepted view amongst evolutionary-oriented social scientists, as Dr Bleie is surely aware (e.g., Lee 1968; Washburn and Lancaster 1968). The division of labour among extant simple societies reinforces such a viewpoint (Daly and Wilson 1983). It may be fair to argue that cognitive psychologists have shown a too-ready acceptance of the details of the hunting-gathering schema. Nevertheless, nearly all writers agree that the size of territory roamed by men has been larger than that of women, for hundreds of thousands of years (Lovejoy 1981), and this is still the case in modern times (Ecuyer-Dab and Robert in press; MacDonald and Hewlett 1999). Moreover, it is an inevitable fact that mammalian mothers spend much more time in the care of infants than do mammalian fathers, they are more vulnerable than males, and at least during infants’ perinatal phases are less mobile. These facts alone would predict that females should have abilities skewed towards wayfinding near a home base, while males would require abilities fostering the navigation of distant extrapersonal space. With regard to throwing accuracy, the evidence for the manufacture of missiles in humans is over 200,000 years old, but the probability that unfashioned missiles (stones, clumps of earth) were used well before this is very high. Chimpanzee males are much more likely to throw such unfashioned missiles than are chimpanzee females, though with limited accuracy (Goodall 1986). It is by now a truism that behaviour is determined by  both physiological and environmental factors, acting interdependently. Bleie’s paper appears, however, to take the position, common in the social sciences, that cultural and biological explanations are in competition with one another. It is true that as we have learned more about the biological influences on  cognitive pattern, the past emphasis on sociological determinance of sex differences has necessarily diminished, except perhaps in some feminist writings. However, this need not imply that cultural influences compete with evolutionary forces. A more rational view is that cultural norms arose in conjunction with, and in support of, biological imperatives (Schaller 1997).

Reliability of cognitive sex differences

Another common theme elaborated by Dr Bleie is that sex differences in cognition are small, unreliable, preliminary, or have dramatically diminished in recent decades as the experience of men and women has become more similar. On the contrary, large reliable cognitive sex differences do exist. Favouring males these are: performance on certain spatial tasks (particularly mental rotation), throwing accuracy, and mathematical reasoning tasks; favouring females: verbal memory, and recall of object locations presented in an array. These differences approach approximately a full standard deviation (or an effect size of one) on tests like spatial (mental) rotation, and throwing accuracy; and are well over half a standard deviation on verbal memory tests and object location memory tests. Math reasoning tests show slightly smaller average sex differences, but men are strongly over-represented at the higher end of advanced math aptitude tests (see Kimura 2002). Mental rotation tests are highly correlated with tests of navigation, both simulated and real-life (Galea and Kimura 1993; Moffat, Hampson and Hatzipantelis 1998; Saucier, Green, Leason, MacFadden, Bell and Elias 2002). Bleie’s suggestion that typing or factory assembly-line work of the type done by women depends heavily on mental rotation has no basis in fact. There are other sex-differentiating tests which show smaller and less reliable sex differences, and it may be that some of these have declined in recent decades (Feingold 1992). There are serious problems, however, in comparing heterogeneous unstandardized tests and heterogeneous populations over time, since it is difficult to infer why, if there are any, such changes might occur (Halpern 1989). If we look instead at one spatial test which has been widely used in standardized form on college samples since at least the 70s, the Vandenberg mental rotations test, the size of the sex difference has not changed systematically over the las three decades (Kimura 2002; Masters and Sanders 1993). Similarly, the Scholastic Aptitude test of mathematical reasoning (SAT-Math) appears to have yielded a constant sex difference in USA high school students of about 0.40 sd favouring males, since 1964 (Donlon 1984). The tests showing large advantages for women (verbal memory, and object location memory) have not been extensively studied for as long a period, hence we can’t yet know whether there have been significant changes over time.

Varieties of spatial ability

I must again point out that it is other authors who have discovered and elaborated on the nature of the female advantage on object location memory. It was first demonstrated by Eals and Silverman (1994) through presentation in succession of two object arrays, with some of the objects having exchanged locations in the second array. Subjects were required to indicate which objects had ‘moved’. A later task (McBurney, Gaulin, Devineni and Adams 1997) found a large sex difference when an array of face-down cards had to be turned over in pairs, and whenever two identical cards were turned over, these were removed from the array. Women cleared off their cards more quickly than men, indicating again superior object-location memory. These studies were consistent with other findings discussed in my book, that women more often use unique objects (landmarks) to find their way than do men. Some of these studies have indeed employed real-life situations (Saucier et al. 2002).  The evidence so far indicates that the female advantage on object location memory is restricted to situations where object and locus are processed together. Where location per se is required, men tend to outperform women (Hoesing, Lewine, Yeo and Edgar 1994; Postma, Izendoorn and DeHaan 1998). I have not, by the way, related object location memory to brain asymmetry, as Bleie indicates. There is a compelling parallel to the human data on landmark use, in those rodent species in which males traverse a larger territory than females. Male rats, for example, use geometric cues preferentially in solving a maze problem, whereas female rats prefer landmark cues. These tendencies can be reversed by perinatal hormone treatment (Williams, Barnett and Meck 1990). Obviously, if men excel on one type of ‘spatial’ task (e.g., mental rotation), and women on another (object-location memory), then these two tasks must somehow depend on different brain mechanisms, despite the common label. Dr Bleie suggests that the definition of spatial ability requires more elaboration than I have given in my book (Kimura 1999), but in fact I spend several pages (pp. 52–57) discussing the nature of the various abilities subsumed under the label ‘spatial’. The degree of independence of various spatial abilities can ultimately only be determined empirically.

Proponents of the life-history determinance of sex differences in cognition point out a relation between the past activities of individuals, and their current spatial abilities (e.g., Baenninger and Newcombe 1995). Bleie refers also to an interesting study (Munroe and Munroe 1971) done in east Africa, in which those children who travelled a greater distance from home (typically, male children) were also those who performed best on certain spatial tasks. However, from neither of these findings are we entitled to conclude that the experiential differences determined the cognitive differences. All we have are implied correlations between the two, and one could equally logically argue that it is the better spatial ability that determines the experience. That is, individuals with better native spatial ability are more likely to engage in those life activities that are demanding of such ability. A third plausible explanation is that another factor (perhaps early exposure to androgens) influences both the life activities and the spatial abilities. Disentangling these relationships requires more information than that two things are correlated. Similar commentary applies to Bleie’s reference to the Japanese fondness for paperfolding as an explanation of their higher scores relative to Western samples, on some spatial tasks. Bleie has (to me) a puzzling section on the null hypothesis, in which she appears to take to task empirical researchers for not sufficiently employing this hypothesis in tests of significance. As far as I am aware, any test of the significance of a difference, based on normal distributions, implicitly employs the null hypothesis. It is a given that even when two normally distributed sets of scores differ as between groups, there will be substantial overlap between the groups. One may confidently state that the two groups differ, and it is understood that we are talking of the average difference. Thus to claim, for example, that on a task on which men excel, there are nevertheless women who will score higher than the average man, is to claim the obvious. A more useful statistic now widely employed is effect size, the difference between two groups stated in terms of the variability of the scores, or standard deviations. It seems particularly inappropriate to belabour the use of statistical procedures with reference to experimental studies, yet allow validity for ‘qualitative research’ that generally bypasses any numerical statement of results.

Whatever the ultimate causal agents in the appearance of cognitive sex differences, we know that a potent proximate mechanism is the exposure to sex hormones. This is too large a subject to review here, but readers are referred to Goy and McEwen (1980) and to my book which is the focus of Bleie’s critique (Kimura 1999). It is clear that for most mammalian species, early exposure to androgens or their derivatives (including estrogens) has a lifelong organizing effect on behaviour. This is not limited to reproductive behaviour but includes nearly all sexually dimorphic behaviours so far studied, including problem-solving behaviours. Moreover, variation in hormone levels continues to exert an influence on behaviour throughout life. It is true that both males and females have both androgens and estrogens in their physiological makeup. However, some behaviours (e.g., rough-and-tumble play) appear to depend exclusively on androgens for their appearance, while others require estrogens (albeit converted from androgens). In humans, and perhaps other primates, the role of estrogen in organizing behaviour may be less critical than it is in non-primates (Smith, Boyd, Frank, Takahashi, Cohen, Specker, Williams, Lubahn and Korach 1994; Thornton and Goy 1986). In young adults, the levels of testosterone in men and women overlap very little, in that they differ by about 3 standard deviations (Moffat and Hampson 1996). Evidence outlined in my book, in agreement with other studies, suggests that higher spatial ability is associated with levels of testosterone in the low normal male range. Women who have higher testosterone and thus approach this level do better than women with low testosterone levels, while men with high (but normal) levels of testosterone do worse than men with low normal levels. Since most subjects studied in this way have been Caucasian, we don’t yet know whether this ‘optimal level’ schema will hold across races, particularly since there is some evidence of ethnic differences in hormone levels (Ellis and Nyborg 1992; Ross, Bernstein, Judd, Hanisch, Pike and Henderson 1986; Soma, Takayama, Kiyokawa, Akaeda and Tokoro 1975).

Many reliable anatomical brain differences have been found to differentiate men and women, from basic structures such as the hypothalamus, to differences in systems connecting the two hemispheres, such as the anterior commissure. Some of these have been reviewed in my book. For Dr Bleie to refer to these systems and/or the brain sex differences as ‘literary fictions’ is simply astounding. We don’t yet know the significance of such differences for cognitive function, but to describe them as fiction is to deny the careful investigations of many respected scientists (e.g., Allen and Gorski 1990, 1991; Allen, Hines, Shryne and Gorski 1989; Allen, Richey, Chai and Gorski 1991).  Although we now have a fair understanding of how brain systems work in mediating certain cognitive functions such as memory, language, and the like, we are still far from understanding how individual differences, the variation in such functions from one type of person to another, are mediated by the brain. However, it must follow that if two groups (such as males and females, left-handers and right-handers, or masculine and feminine gender types) differ reliably in some behaviour not simply dependent on physical differences, then their nervous systems must also differ in some way. Where else could the behavioural differences reside? This is just as true of learned as of unlearned behaviours. Brain differences underlying cognitive differences need not, however, be visible in simple structural features. They may take the form of differing organizational mechanisms not apparent by simply viewing the external brain.