Primate brain architecture has been shown to correlate with both ecological [1-3] and social [4-9] factors. Social and competitive demands often differ between the sexes, however, which should leave predictable marks in the relative sizes of key brain structures [4,8-11]. The aim of this paper was to investigate the relationship between selection in relation to sex and the evolution of brain architecture in primates, and to draw attention to this largely neglected aspect of mammalian brain evolution.
Selection in relation to sex occurs whenever any genetically influenced morphological trait or behavior increases the fitness of one sex, but not the other. One well-known instance of this process in mammals is male-male competition for sexual access to females, which has resulted in a diverse array of characters important for males in conflict situations [12,13]. In primates, for example, this type of male intra-sexual selection has been shown to result in sexual size dimorphism in traits important for male combat, such as canine teeth [14,15] and body mass [16,17].
We used phylogenetic comparative methods to investigate whether two types of selection in relation to sex-intra-sexual selection on males through male-male physical competition and social selection in females-have had predictable effects on species-typical brain structures in primates. To be successful in male-male physical competition, a primate male in a polygynous species not only needs to be large and have imposing canines, but also to display, control and use his weapons and size effectively. No matter what their physical equipment might be, males without such abilities should self-evidently be less able to succeed in competitive situations. A male can potentially be proficient in the use of his assets through quick, agile and skillful movements, thus gaining an edge on his opponent physically, or he can use intelligence to outwit the competition, thus decreasing the amount of direct physical conflict necessary. The former strategy involves physical movements, while the latter involves cognitive processes intended at avoiding such movements.
From this reasoning, we derived four, not mutually exclusive, hypotheses involving male intra-sexual selection. (i) If physical combat is important, the effects of sexual selection on brain architecture should mainly act on brain regions with roles in motor skills and the coordination between sensory and motor information , whereas (ii) if non-combat social skills are important, we would expect to see effects of sexual selection on brain regions with more general integrative and associative functions, such as the neocortex [4-9]. (iii) It may also be the case that larger male body mass, resulting from male intra-sexual selection, puts increasing demands on the parts of the brain that handle autonomic nervous activity. (iv) Finally, selection due to increased demands from competitive situations would also be expected to influence structures that relate to aggression, threat, fear and aggression control.
We investigated four predictions derived from these hypotheses. These are all subject to the caveat that brain functions are often distributed between many separate structures; selection for a specific function will therefore tend to change several functionally connected brain structures simultaneously [18,19]. (1) We expected sexual selection for better motor skills  to influence components within almost all parts of the brain that function in sensory-motor activities. These range from the motor cortex and other cortical areas in the neocortex that initiate voluntary motor activities, through the mesencephalon and diencephalon, to the cerebellum in which fine adjustments of movements are made, and further through the medulla oblongata and pons. (2) In contrast, if males are more commonly in need of strategic cognitive abilities when competing, this should mainly, if not exclusively, be reflected in the relative size of the telencephalon through an increased neocortex volume . (3) If effects of sexual selection are limited to effects that are caused simply by the need to control a larger body, then this should produce relatively larger brain components involved in autonomic nervous activity, primarily the hypothalamus and the medulla. (4) Structures involved in facilitating aggressive behavior (the central gray region in the mesencephalon, the hypothalamus, and amygdala) are expected to be larger in more sexually selected species; in contrast, structures relating to controlling aggression (such as the septum) should be smaller, although this last expectation is not straightforward as it could as well be argued that more aggressive species also need to better control this aggression.
We also investigated a hypothesis related to the drivers of female reproductive success. In comparison with males, female reproductive success is more closely linked to ecological factors, including the acquisition and defense of resources and protection from predators [21-23]. That causes and consequences of sociality should be considered independently for the two sexes is also indicated by observations that separate dominance hierarchies are often maintained for males and females in primate groups , and that primate social groups simply tend to have more females than males [25,26]. Although males also form alliances in primates [27-29], the conditions favoring these alliances tend to be more restrictive both within and across species [30,31]. Thus, cooperation is likely to be more important for females and also more closely tied to ecological demands than in males. In terms of social selection, we therefore focused on one key prediction (prediction 5): we expected that brain components involved in tasks relating to sociality and cooperation are selected for primarily in females [5,11,32], predicting that these brain structures are larger in species characterized by greater female sociality. Such brain centers are primarily, but not exclusively, found in the telencephalon (cerebrum), and within the telencephalon, in the neocortex in particular [33,34].
In summary, at least two sex-specific selection factors should leave imprints on how the brain is organized in different primate species, producing five predictions. Predictions 1–4 involve sexual selection on males for increased sensory-motor coordination, increased size of the areas relating to autonomic functions, cognitive abilities in the context of strategic planning and social networks, and changes in the structures that relate to aggression and fear. The fifth and final prediction involves social selection on females, related to increased cognitive abilities in the context of social networks.
If social or sexual selection exists, it could have one of two effects on sex-specific brain measurements. On the one hand, it could select for larger brain structures in both sexes; on the other hand, it could lead to sexual dimorphisms. Regarding the first possibility, recent research has shown that in anthropoid primates, the degree of sexual selection on males is correlated with larger canine size and greater body mass not only in males, but also in females [15,17]. Effects on females as a result of selection on males could be due to genetic correlations between the sexes in genes determining the character in question [35-39], but this is usually expected to be a temporary phenomenon [36,40,41]. More probable is that selection on females correlates with intrasexual selection operating in males [40,42], although other mechanisms are also possible .
Alternatively, if a selection pressure results in sexual dimorphisms in brain architecture, this would also result in a higher average value for the trait in a species, simply because the values for one sex would be increased relative to the other sex and would therefore increase the mean for that species. We therefore had strong reasons to expect that selection in relation to sex should be clearly detectable regardless of its specific effects on male and female brains. We emphasize these points because the brain volume data examined in our study are from unsexed primate specimens (Additional files 1 and 2) . The results presented below provide a strong argument for obtaining sex-specific measures for more detailed analyses within and across species.