Sunday, 31 August 2014

How kin selection pushed group selection into the scientific fringes

Group selection lost out to kin selection in the 1960s and 1970s as an explanation for cooperative behaviour. Group selection went on to lead an underground existence - in which a few remaining proponents sought out ways to find things which group selection explained which kin selection did not.

We now know that modern versions of these theories make the same predictions. So there isn't really anything that group selection can explain that kin selection cannot. However this didn't stop the group selection proponents from trying to find these types of phenomena. Indeed, if they had succeeded, fame and glory would have awaited them.

This quest took group selection to the borderlands of science. Kin selection was uncontroversially used to explain obvious adaptations - such as breast feeding, parental care and nepotism. Group selection picked more controversial targets - senescence, the maintenance of sexual recombination and cultural evolution. The theme of these topics seems to have been avoiding rapid refutation of: "kin selection explains that without invoking group selection".

This quest now seems to be mostly over. Most of the group selection enthusiasts have publicly given up their original quest - and have accepted the equivalence between kin selection and group selection.

However the topics group selection is invoked to explain still seem to be skewed away from those that kin selection is used to explain. As a result, group selection is generally offered as an explanation in areas where the theory doesn't function well. This seems like a hangover which group selection is still recovering from.

The "multi-level" terminology

This article will claim that "multi-level selection" and "group selection" should not be understood as being synonyms.

It does appear that "multi-level selection" is terminology that was invented to give group selection a face lift - and allow it to escape from an unglamorous past.

However, group selection is not more or less "multi-level" than kin selection is. The concepts of "relatedness" and "kin selection" can be applied at multiple levels too. One can meaningfully discuss the relatednes between cells in an ant, the relatedness between an ant and its sisters and the relatedness between entire colonies of ants. This has been understood since the 1970s and it can be seen as a result of the generality of the Price equation.

These days, group selection enthusiasts often prefer to use the "multi-level selection" terminology. This article is my explanation of why I rarely follow them - and why I think it is a confusing gambit. Kin selection is "multi-level" too. Using the term "multi-level selection" fails to distinguish between kin selection and group selection. It is true that kin selection and group selection are equivalent - in the sense that modern versions of these theories make the same predictions. However that doesn't mean that proponents use the same methodologies. These are still topics that are worth distinguishing between. The best way of doing that is, I think to use the terms "kin selection" and "group selection".