Selection for aggression


Jason Collins


January 4, 2011

Masculine appearance in a man is an indicator of their health, which in turn leads to more viable offspring. On this basis, one might assume that women prefer masculine men. However, empirical research into whether women prefer men with more masculine physical features has not shown the strong positive preference we might expect. While masculine appearance is linked to health, that masculine partner may be less interested in a long-term relationship and be unlikely to provide for the child over the long-term. In that case, a woman is likely to weigh up the benefits of a healthy child against the likelihood of provision of care by the father of that child.

As reported in The Economist last month, there has been an interesting exchange in the Proceedings of the Royal Society about what factors may influence this trade-off and lead to variations in preferences for masculinity. In the first paper,DeBruine et al proposed that where health in a country is poor, a woman’s preference for masculinity will be strong as it is important to have healthier offspring. Across about 4,800 women resident in 30 developed countries, they found a significant relationship between health and preference for masculinity that was robust to controls for age and wealth. However, they suggested that other factors such as violence in the society should be researched.

Brooks et al took up this suggestion and used data from the first study to examine whether women are attracted to more masculine men in environments where there are greater benefits to dominance. Using income inequality as an indicator of violence, they found that income inequality may be a better predictor than health of preferences for masculinity. They did add the proviso, however, that health and inequality may be correlated, and that other correlates may mediate the relationship. Brooks et al also tested murder rates as an explanatory variable and found that if they included both health and murder rates in the regression, only the murder rate was significant (although income inequality was a stronger predictor than the murder rate).

DeBruine et al responded that the finding by Brooks et al that male-male competition is responsible for the preference for masculinity was reliant on income being excluded from the murder rate and health regressions. Using new United States data, they also called into question the finding that income inequality was a better predictor than health.

Leaving aside which analysis of the data we should prefer, the relationship suggested by Brooks et al could have some interesting dynamics. If a society is violent, Brooks et al would suggest that more masculine, dominant men would have a fitness advantage. If that were the case, they would come to form a larger part of the population. Does a larger proportion of more masculine men in a population make it inherently more violent? If so, a cycle could emerge where violent societies have a larger proportion of violent men and stay violent. Conversely, peaceful societies would stay peaceful with characteristics linked to greater propensity to violence not being rewarded.

To look at whether that scenario might hold, greater analysis of the strength of the preference variation would be required. Is the fitness advantage large enough that it could have dynamic implications? To analyse this we would need to expand the analysis to countries with greater extremes of violence, with the DeBruine et al dataset restricted to people in developed countries who had identified their ethnicity as white. What is the relative strength of the preference in Iraq or Sudan? Another source of data would be the level of masculinity of men within a population. Is there a higher proportion of masculine men in populations with a long history of violence?