Our analyses indicate that selection in relation to sex has been an important influence on primate brain architecture. The results showed that sexual selection on males has acted positively on the relative sizes of the medulla, mesencephalon and diencephalon, but negatively on the pons and telencephalon. In the case of social selection, the average group size of females was positively correlated with the relative size of the telencephalon. As has already been shown elsewhere using the same dataset , this latter effect is apparently mediated through a positive correlation between female group size and the relative size of the neocortex. This indicates that female sociality is responsible for the evolutionary change in relative neocortex size that has taken place in haplorhine primates. These results also suggest that social demands on females and competitive demands on males require skills mainly handled by different brain components.
Our analyses fail to support the hypothesis that sexual selection on males has selected for enhanced cognitive abilities, based in particular on the negative correlation between sexual dimorphism and relative telencephalon volume, and the lack of a significant association between sexual dimorphism and neocortex volume. Further indications that selection on physical combat skills are more important is that the mesencephalon, diencephalon (containing the hypothalamus) and amygdala, all involved in governing aggressive behaviors, are positively correlated with the degree of sexual selection, whereas the septum, which has a role in facilitating aggression control, is instead negatively correlated with the degree of sexual selection. Moreover, male group size is positively correlated with the relative volume of the diencephalon (but see Additional file 5) and negatively correlated with relative septum size, further strengthening the conclusion that aggression is an evolutionarily important component of male-male interactions.
The main structures of the brain that were positively correlated with the degree of male intra-sexual competition (the medulla oblongata, mesencephalon and diencephalon) all contain important motor centers. The same is true for the cerebellum and the telencephalon, but the former is not significantly correlated with size dimorphism and the latter even exhibits a negative correlation. Given the distributed nature of the motor centers in the brain, analyses of the main brain structures provides only indirect support for the hypothesis that sexual selection acts on motor centers; more detailed data on specific brain structures are therefore needed.
The medulla oblongata and the diencephalon are important for autonomic nervous system activity. Significant correlations involving these structures suggest that larger body size resulting from sexual selection has placed demands on brain structures involved in handling a larger body. Thus, while not providing direct support, our results are in line with two hypotheses that deserve further scrutiny: that sexual selection selects for (i) brain structures important for success in male-male conflict through better body control, and (ii) better handling of a larger body through increased importance of structures involved in autonomic tasks.
The effects of social selection are more clear-cut in that they suggest that different strategies used by males and females have left marks on primate brain architecture, with female social group size positively and male group size negatively correlated with telencephalon size. Thus, demands of male and female sociality differ fundamentally in skills governed by the telencephalon, and within this region, especially the neocortex. The neocortex is important for cognitive skills involved in navigating complex webs of social relationships [44,45]. As stated in the Background, there are several reasons to expect that such social abilities concern the reproductive success of females more than males. Our results also indicate that the relationship between neocortex volume and male group size actually may be negative, further highlighting the specifically female aspect of handling social relationships in primates. More detailed analyses on the relative sizes of structures within the neocortex would be expected to reveal that it is these areas that process social information that are positively correlated with female group size.
An important question concerns the epigenetic mechanisms by which species differences in brain architecture arise during ontogeny, in particular the relative roles of early, genetically guided ontogenetic processes versus later processes influenced substantially by environmental input and/or hormones. For example, experimental evidence indicates that perturbations of sensory inputs cause neural reorganization [46,47]. However, regional differentiation occurs early in ontogeny, prior to and/or independently of neural innervation from the periphery [48,49], and mutations that influence architectonics prior to innervation by the periphery have been discovered . The role of neurogenesis later in mammalian ontogeny (e.g. in adults) appears to be relatively minor in primates, and restricted to the hippocampus and olfactory bulbs [51-53]. Thus, adult neurogenesis is highly unlikely to explain species differences in overall brain architecture. Finally, many species differences in brain structure size fall outside the range of intra-species variation [18,54]. Thus, while earlier, predominantly prenatal processes are undoubtedly crucial, species differences, such as those analyzed here, may be somewhat moderated by later ontogenetic effects.
From our study of unsexed specimens, we cannot determine if social and sexual selection have produced sexual dimorphisms in brain size, or whether selection acting on one sex has had correlated effects on the other sex, thus producing larger brains for that species. The results presented here therefore emphasize the value of obtaining sex-specific data on brain structures. Several authors have proposed or tested hypotheses for sexual dimorphism in brain structures [10,11,20,55,56]. Several additional hypotheses can be tested. For example, in more social species, females should be expected to have relatively larger neocortices than males. Contrary to our results, however, the fact that reported incidences of innovation behavior are higher in males than in females across species  could be taken to indicate that males have greater cognitive abilities. It can possibly be argued that female primates use their neocortices mainly to navigate social webs, whereas male neocortices are free to perform other tasks. On the other hand, as indicated by our results, in more sexually selected species, a suite of other brain structures should be larger in males than in females. To investigate such sexual dimorphisms, however, sex-specific measures of brain components are necessary.
We have no reason to expect that the patterns presented here apply only to anthropoid primates; similar patterns should be detectable in other mammal clades, provided of course that sufficient variation in sociality and/or intra-sexual competition exists. It is even possible that these factors have driven brain evolution in mammals and thus may explain differences in brain architecture among different mammal orders. This especially concerns the relatively large neocortex-a hallmark of primates and a trait indicated by this study to be a consequence of the high degree of female sociality in primates-but should also involve other structures.