George Williams argued that natural selection results in three categories of features: adaptations, by-products and noise (Williams Reference Williams1966). The same classifications hold for sex differences, and the debate between evolutionary psychologists and gender role theorists can be understood as a debate about which of two categories sex differences in aggression falls into. Because Williams was classifying features and not differences in features, some slight translation is necessary.

Evolutionary psychologists tend to think that sex differences in aggression are adaptive differences; that is, they are sex differences resulting from adaptations designed differently in men and women by natural selection in response to differing ancestral selection pressures. Non-controversial examples include sex differences in body size and maturation rates. Gender role theorists tend to think of sex differences in aggression as learned byproducts, that is, as sex differences that result from learning mechanisms (which were designed by natural selection, of course) that are the same in men and women but create differences because of differential input. Non-controversial examples of this would be sex differences in car seat settings and fear of prostate cancer. Finally, there are arbitrary differences: these are sex differences that result from accidents of history and are not designed by natural selection but also do not stem from learning mechanisms in ways that lead to sensible outcomes. Non-controversial examples would be sex differences in styles of dress, the spelling of names (e.g., Aaron vs. Erin), and culturally agreed upon color symbols (e.g., pink for girls, blue for boys).

The standard of proof for adaptive differences is parallel to those for adaptations. One has to show evidence of complex functional design, geared toward solving an adaptive problem that acted differently on the sexes. Archer has done this for sex differences in intrasexual aggression. He lays out a complex of features, all of which would result from a differential selection pressure (namely, sexual selection). They include, for males: greater variance in reproductive success, greater size and strength, longer maturation times, higher mortality and male-biased conception ratio. I would add that physical differences in size and strength are also supplemented by sex differences in basal metabolic rates (Garn & Clark Reference Garn and Clark1953), heart size, heat dissipation, hemoglobin, muscle-to-fat ratio, and bone density (Lassek & Gaulin, in press). Across all those variables the sex difference is in the direction of males being designed for physical aggression. Additionally, boys prefer rough-and-tumble play, a type of activity that is understood by evolutionary biologists to be practice for future combat (Symons Reference Symons1978). This last point is particularly important because of the overwhelming evidence that sex differences in rough-and-tumble play are not caused by societal expectation. Girls born with congenital adrenal hyperplasia (CAH) as well as progestin-induced hermaphrodism (PIH) are typically raised as girls, and genetically are girls, but experience some heightened organizing effects of androgens during development. As a result, they engage in more boy-like play patterns, including rough-and-tumble play (Daly & Wilson Reference Daly and Wilson1983). All of these coordinated features, each in the predicted direction, provide powerful evidence that natural selection designed males and females differently when it comes to aggressive tendencies.

The evidence required to put sexual differences in aggression in the category of learned byproducts is as follows: (1) identify the adaptation (or adaptations) that aggressive differences result from, and (2) demonstrate why, as a byproduct of their design, those adaptations would lead to sex differences. Gender role theorists have taken steps in these directions, specifying that evolved physical differences in strength and size coupled with a cost-benefit analysis mechanism (and more traditional socialization mechanisms) will produce differences in aggressive tendencies (Wood & Eagly Reference Wood and Eagly2002). As Archer points out, however, the data are stacked against this theory. He mentions that aggression's developmental trajectories are inconsistent with socialization theory, and the role of testosterone and operational sex ratio are difficult for gender theorists to explain. The data from CAH and PIH girls also contradicts the idea that gender roles lead to differential aggression, as the girls generally maintain a female identity even while increasing their aggressive play. Finally, from a theoretical point of view, one has to ask why natural selection would have selected genes that created sex differences in body size and strength if males and females were aggressing at equal rates.

With regard to intersexual aggression, it is important to keep in mind that similarities between men and women on broad measures of aggression can hide sex differences in particulars. For example, men and women have similar rates of spousal homicide that are motivated by sexual jealousy, but it was male jealousy that resulted in both the killing of husbands, usually in defense, and wives, usually out of jealous anger (Daly & Wilson Reference Daly and Wilson1988). As Archer says, both evolutionary and social role perspectives predict that spousal aggression should vary with the relative power of the individuals. The question remains, however: Which perspective will be most useful for predicting and explaining currently unknown features of mate-directed aggression? By studying selection pressures and how they work, evolutionary biologists have been able to account not only for the origin of biological complexity, but the origin of sexual reproduction itself (Tooby Reference Tooby1982), the origin of sexes (Parker et al. Reference Parker, Baker and Smith1972), the existence of maturation and senescence (Williams Reference Williams1957), the origin of aggression (Maynard Smith & Price Reference Maynard Smith and Price1973), its major causes (Huntingford & Turner Reference Huntingford and Turner1987), the existence and function of aggressive displays (Alcock Reference Alcock1998), the magnitude and constituent features of sexual dimorphism (Daly & Wilson Reference Daly and Wilson1983), and sexual differences in aggressive tendencies, homicide (Daly & Wilson Reference Daly and Wilson1988), and rough-and-tumble play (Symons Reference Symons1978). Finally there is a massive amount of primatological data showing how natural selection has designed males and females of other species in ways consistent with the differing adaptive problems they faced ancestrally (Smuts et al. Reference Smuts, Cheney, Seyfarth, Wrangham and Struhsaker1987). With all that in mind, it seems likely that natural selection played some important role in the differential design of male and female minds, particularly in domains defined by differential selection pressures such as aggression.