Human evolution yields behavioral flexibility

Any evolutionary analysis of human sex differences has to account for humans' behavioral flexibility. This flexibility in response to local circumstances is a characteristic feature of the human species. It reflects their evolution in diverse environments with changeable conditions that impinged in differing ways on survival and reproductive outcomes (Wood & Eagly Reference Wood and Eagly2002; in press). For example, in human history, particularly in the late Pleistocene era, climate appears to have been highly changeable. Also, humans and their ancestors engaged in extensive niche construction, meaning that their activities produced changes in the environments in which they lived. Accommodating to such changes required behavioral flexibility, enabled by an evolved capacity for innovating and sharing of information through social learning, yielding a cumulation of culture (Richerson & Boyd Reference Richerson and Boyd2005). Humans' flexibility is evident in their various novel solutions to the problems of reproduction and survival, including tolerance for a wide range of foods, ecologies, and living arrangements.

This flexibility in behavior is at the heart of our evolutionary analysis. Sex differences in behavior arise flexibly from a biosocial interaction, in particular from sociocultural and ecological forces in interaction with humans' biology as defined by female and male physical attributes and reproductive activities (Wood & Eagly Reference Wood and Eagly2002). Women bear and nurse children, and men possess greater size, speed, and upper-body strength. These attributes serve as constraints on behavior such that certain activities are more efficiently accomplished in certain societies by one sex than by the other. Consequently, the sexes participate in a division of labor. Some behaviors tend to be performed by one sex across most societies and time periods, but even these behaviors can be influenced by particular societal circumstances.

In vivid illustration of the flexibility of sex differences in human aggression, women sometimes, although rarely, engage in organized combat in world societies. A prominent example is the “Amazon Corps” of the Dahomey Kingdom of West Africa in the eighteenth and nineteenth centuries. Because endemic warfare had reduced the supply of male warriors, survival of the kingdom required military service by women. These women were trained for combat and fought with distinction in all-female units (e.g., Alpern Reference Alpern1998). A prominent twentieth century example is: the women soldiers of Eritrea. These women fought successfully alongside men in mixed units in the long-term and eventually victorious struggle of the Eritrean People's Liberation Front to win Eritrea's independence from Ethiopia (Bernal Reference Bernal2000). Eritrean women's military service was apparently ideologically driven by the revolutionary movement's modernist rejection of traditional gender relations. These societies overcame the constraints of female reproduction in different ways – in Dahomey by forbidding women warriors to have sex with men, and in Eritrea by providing community childcare. The potential military disadvantages of lesser female physical prowess were surmounted through training and the provision of weapons. In both societies, the female warriors' psychology proved adequate for full participation in highly aggressive combat activity. Also indicative of flexibility in aggressive behavior is the evidence that women in a small percentage of world societies regularly hunted large game (see Wood & Eagly Reference Wood and Eagly2002).

Division of labor produces sex differences in human behavior

Archer has left out the key link in our theory between the division of labor and gender roles, which explains how these roles and associated behaviors change over time and circumstance. Consistent with the social psychological principle of correspondent inference (Gilbert Reference Gilbert, Gilbert, Fiske and Lindzey1998), people infer the traits of men and women from observations of their behavior. To the extent that people observe men and women engaging in different types of behaviors, they regard them as psychologically different. These observations underlie gender roles, which form a shared knowledge structure specifying what men and women usually do and what they should do in a society.

These gender roles, as well as specific social roles such as daughter, boss, and friend, influence behavior through a trio of interacting proximal causes of sex differences and similarities: Roles influence behavior through chemical signals of hormonal changes in interaction with individuals' personal gender identity and others' stereotypic expectations. Our position that the activating influence of hormones is mainly to facilitate role behavior reflects Archer's excellent meta-analysis (Archer Reference Archer2006b). Specifically, testosterone rose in men anticipating and playing sports and highly competitive games, especially among the contest winners, but did not rise in the absence of social roles or provocations calling for aggressive, dominant behavior. Although women have substantially less testosterone than do men, female athletes also recruit testosterone before a competition (see Wood & Eagly, in press). But Archer's (2006b) meta-analysis did not show the reverse causal relation: Studies that experimentally injected men with testosterone or related synthetic androgens found no systematic rise in anger, aggression, or hostility resulting.

Along with hormonal changes, internalized gender roles produce gender identities that act as trait-like determinants of aggressive behaviors. Others' expectations also foster behavior consistent with gender roles. Unfortunately, Archer misinterpreted one of the best-designed experiments demonstrating that aggression is influenced by others' expectations (Lightdale & Prentice Reference Lightdale and Prentice1994). This study did not confound the directness of aggression with its normative manipulation, as he claimed, and thus demonstrated how sex differences in aggression depend on the salience of social norms. Nonetheless, Archer is correct in recognizing our theory's trio of psychological mechanisms (hormonal activation, gender identity, others' expectations) that can yield differences between men and women in aggressive behaviors as well as differences between individuals within each sex.

Sexual selection provides an inadequate account of human bodily dimorphism

Archer maintains that men's size and strength and other physical attributes were sculpted by sexual selection pressures in which ancestral males who were larger and stronger had better fitness outcomes because they were able to compete with other males for access to many mates. Sexual selection pressures presumably also organized human psychology, making men more aggressive than women.

Comparative research with primates, however, suggests that bodily dimorphism requires a more complex explanation. Despite the bodily metrics that Archer presents, when evaluated in relation to other anthropoid primate species, humans have relatively low male-female dimorphism in both body weight and canines, and presumably in other bodily attributes as well (Plavcan & van Schaik Reference Plavcan and van Schaik1997a, p. 351). Even though across all primate species, greater bodily dimorphism was associated with polygynous mating and male-male competition, dimorphism at the low levels existing in humans “can be found among species with a wide variety of mating systems and competition levels” (Plavcan Reference Plavcan2000, p. 338). In addition, compared with most other primate species, humans have a low operational sex ratio (e.g., Wrangham et al. Reference Wrangham, Jones, Laden, Pilbeam and Conklin-Brittain1999), which also is compatible with low male-male competition. We are puzzled by Archer's failure to acknowledge that, when compared with other primates, the relatively small amount of human size dimorphism does not imply sexual selection through male-male competition.

Also undermining Archer's one-dimensional sexual selection account is evidence that bodily dimorphism was likely influenced by selection on females as well as males. Selection pressures on females are plausible, given that the decreasing size dimorphism as hominids evolved from the earlier Australopithecus to Homo was due to an increase in the size of females relative to males (as Archer notes). For this and other reasons, experts have abandoned Archer's one-sided sexual selection argument in favor of a richer set of possibilities. These newer ideas include not only selection pressures on females but also more varied principles for understanding the cooperative and competitive relations between males and the niche construction of males and females in varied environments (e.g., Plavcan Reference Plavcan2000; Plavcan & van Schaik Reference Plavcan and van Schaik1997a; Reference Plavcan and van Schaik2005).

Now consider the real predictions of the social role/biosocial theory

It is troubling that Archer claims that we believe that sex differences in aggression are “modest” (see his Table 1). Our analysis yields no a priori hypotheses about the size of sex differences but instead anticipates that they pattern in a society according to the trio of proximal causes noted above. In addition, as we explain in the next paragraphs, Archer has freely invented (and then disproved) a social role prediction about developmental changes in aggression sex differences and failed to recognize an obvious social role alternative prediction for greater male than female variability in aggressiveness. He also oversimplifies the social role predictions for cross-cultural comparisons of sex differences.

Archer maintains that our position on developmental changes should be that, “Sex difference should start small and increase with age through childhood, coincident with the cumulative influence of socialization” (see Table 1 of the target article). This naïve prediction assumes that socialization pressures cumulate in some simple way across development, and furthermore, that they overwhelmingly encourage boys' aggressiveness – that they are composed of cheering sections of mothers, fathers, teachers, siblings, and peers, all urging aggressive behavior. On the contrary, boys encounter both discouragement of aggression (e.g., in the classroom) and encouragement (e.g., in sports, self-defense). Prohibitions against physical aggression increase as boys mature, with violence disallowed in the overwhelming majority of employment settings and disparaged in close relationships with friends, family, and romantic partners. Given these complexities, boys learn to express aggression contextually and in patterned ways. We have not made the claim that Archer suggests but instead have remained silent on age trends in aggression because prediction demands close study of the pressures that foster and discourage aggression as boys and girls mature.

With respect to within-sex variability in aggressive behavior, our theory offers an obvious alternative explanation to sexual selection theory. Societies have often provided more opportunities for boys than girls to learn aggression (e.g., in gangs and contact sports). Because some boys and young men avail themselves of these opportunities and others do not, their aggressiveness should be more variable than that of women and girls. Men's greater power and resources in society also bring some men protection from retaliation, and other, subordinate men greater vulnerability to retaliation. Women's more restricted opportunities are consistent with their lesser variability in these behaviors.

Finally, we note that many cross-national comparisons of sex differences, although seemingly an attractive way of testing social role and evolutionary hypotheses, are plagued by ambiguity. Our theory generally predicts a lessening of sex differences with greater gender equality – and this is what Archer (Reference Archer2000a) found with greater male physical aggression to partners in countries marked by lesser gender equality. Yet, gender equality is not the only feature of men's and women's roles that influence sex differences across cultures. Other aspects of roles influence subjective ratings of personality attributes and abilities. In particular, the extent of segregation of men and women can influence the comparison standard that they use to evaluate themselves and others. In traditional cultures in which occupational and other roles tend to be segregated by sex, men and women would judge their own and others' psychological attributes through comparisons with salient others, who would mainly be of the same sex. Thus, a man might rate himself as only moderately aggressive because he is comparing himself with other men, who are generally somewhat aggressive in his society. In contrast, in societies with less segregated roles, a man might compare himself with individuals of both sexes and conclude that he is relatively aggressive. The result of this shifting comparison standard is that sex differences can appear to be smaller in less egalitarian, more hierarchical societies, in which individuals compare themselves to their own sex (Guimond et al. Reference Guimond, Chatard, Branscombe, Brunot, Buunk, Conway, Crisp, Dambrun, Désert, Garcia, Haque, Leyens, Lorenzi-Cioldi, Martinot, Redersdorff and Yzerbyt2007). Therefore, cross-national comparative data based on subjective trait ratings (e.g., Schmitt et al. Reference Schmitt, Realo, Voracek and Allik2008) cannot be taken at face value. Overcoming such confounds requires common-rule (or more objective) measures that disallow standard shifts (Biernat Reference Biernat2003). The reports of behavioral frequencies in the research on intimate partner aggression are an instance of common-rule measures, thus allowing for confirmation of our social role prediction in Archer's (2000a) meta-analysis.

In conclusion, the debate between sexual selection theories of the origins of human sex differences and our social role/biosocial theory will no doubt continue as proponents of each theory hone their views in response to newly emerging empirical evidence. We are encouraged by Archer's extension of standard evolutionary psychology models to include social roles, but we think an adequate theory of sex differences in aggression would take more seriously the flexibility in behavior that follows from a social role analysis.