Thursday, November 04, 2010

Our Cheatin' Ancestors?

A new study suggests that modern humans are considerably less promiscuous than their forbears. In a story that appeared in the University of Oxford site, the culprit seems to be androgen levels:
Previous studies by the research team have shown that promiscuous species have low index to ring finger ratios while monogamous species have high finger ratios. High levels of androgens, such as testosterone, increase the length of the fourth finger in comparison to the second finger. It is thought that prenatal androgens affect finger length during development in the womb, which in turn is linked to adult behaviour. High levels of prenatal androgens are linked with competitiveness and promiscuity.
I would love to know what the r2 on that analysis is. I am always extremely suspicious of a study that links a single process to a complex social behavior, especially something like this. Later, the article has this:
Dr Susanne Shultz, from the University of Oxford, said: ‘Social behaviours are notoriously difficult to identify in the fossil record. Developing novel approaches, such as finger ratios, can help inform the current debate surrounding the social systems of the earliest human ancestors.
They are notoriously difficult because they often leave few to no clues. This analysis also sidesteps the problems involved in trying to ascertain what other selective forces might have been at work in early hominid development. For example, in Ardipithecus, given that they exhibited incipient bipedality and were, at least in part, arboreal as well, there may have been selective pressures on metacarpal and phalanx length. Also, is finger length one trait dictated by pleiotropy? Or is it polygenic? Is there linkage disequilibrium going on? These sorts of questions could be extrapolated to other hominid species as well and need to be addressed before arriving at the conclusion that is stated in this study.


  1. There is, in fact, quite a sizeable body of literature on the 'index to ring finger ratio'. Apparently, it could also point to sexual orientation. Evolutionary psychologist at my university (Groningen, the Netherlands) love it.
    But I do share your scepticism on the subjet.

  2. What R-squareds are typical in the field? I do controlled experiments in my own discipline, so values in the neighborhood of 0.5 and up are not uncommon (though very low values occur as well). I've even seen some in the 0.8 range. But I understand that field data in the same discipline typically generate nothing higher than about 0.15.

  3. And in anthropological circles, such numbers are also not uncommon. The researcher is very clear to then state "this model explains some of our observations but there is clearly much data that this model does not explain."