On re-reading my note on ‘Cultural evolution by group selection’, and the comments people have made on it, I realise that some of my terminology was not clear enough - so I hope the following is useful.

‘Heritability’: in referring to the ‘heritability’ of cultural traits I simply meant the tendency of the ‘offspring’ of a group to share some of the traits of the parent group. I did not mean to imply that the basis for this was genetic. The transmission might be entirely through learning and imitation.

‘Societies’: by the unit of ‘society’ or ‘culture’ I had in mind a group of people with substantially the same customs, beliefs, and institutions. Normally they would also speak the same language, interbreed, and live in a continuous geographical area, not separated by major barriers. In modern times, such a group would probably fall under a single political ‘state’ (though some states contain more than one group of this kind). In earlier times, the connection between the cultural units and the political units was looser. Most of the ‘societies’, ‘peoples’, or ‘cultures’ recognised by anthropologists contain several different tribes, clans, etc., which operate as independent political units, and may differ somewhat in their customs. It is difficult to decide what are the appropriate ‘groups’ in considering the possibility of ‘group selection’. But obviously there can be no group selection of cultural traits unless there is a significant cultural difference between the groups, and the strength of selection must depend on the size of the differences.

‘Fitness’: biologists use different definitions of fitness for different purposes. In my note I was thinking primarily of the relative fitness of one cultural trait compared with another, as measured by their relative numbers of ‘offspring’. I assumed, but did not make this clear, that the relative fitness does not vary according to the frequency of the different traits in the population. So, for example, if the trait of polygamous marriage has a fitness advantage compared with the trait of monogamous marriage, then it has much the same relative advantage whether the trait of polygamy is rare or common. This seems a reasonable assumption for most cultural traits, but I should have made it explicit, because it does affect my argument about the likely fitness effect of any particular trait.

‘Mutation’: by the mutation of a cultural trait I simply mean a variation in the trait, which in turn may be inherited by descendent groups. It need not have a genetic basis. Cultural change during the lifespan of a society is very common, so the ‘mutation rate’ for cultural traits, as thus defined, must be high, and is likely to be an important factor in cultural evolution. I have thought about this a bit more since writing my note. Suppose there are two traits, A and B. Assume that there is a constant rate of mutation from A to B and also a constant rate of mutation from B to A. The rates may be the same, or they may be different. In the absence of selection, the proportions of A and B in the population will be in equilibrium when they are inversely as the mutation rates between the two types. E.g., if the mutation rate from A to B is 30%, while the mutation rate from B to A is 20%, they will be in equilibrium when A is 40% of the population and B is 60%. As I mentioned in my note, the mutation rate might well be high enough to swamp the effects of selection in favour of a trait while it is still rare. However, it did not occur to me that in a sense high mutation rates could actually be favourable to selection. Selection of a new variant would initially be a slow process, because the selective advantage applies only to the small proportion of the population who already possess it. But the mutation rate might bring the proportion of that variant in the population up relatively quickly to a significant level. Selection would then ‘take over’, since it would have, so to speak, more material to work on. Selection (in combination with mutation pressure) might therefore be more important than I thought at first.

But don’t get too excited - as I also mentioned in my note, the rate of increase
due to selection would be damped down when the favoured variant becomes
more common. (I am assuming that (a) the ‘population’ has a fixed ceiling, and (b) the two variants have a constant relative fitness. While the favoured variant is rare, it can expand freely, but once it is common, it has to compete against itself as well as its less effective rival, and expands more slowly.) Once the proportion of the favoured variant is above the ‘mutation equilibrium’ level, selection will also be opposed by mutation working in the other direction. At some point a new equilibrium will be reached, where the weakening force of favourable selection is balanced by the amount of net mutation away from the favoured variant. The outcome depends on the actual numerical rates involved, but to crudely generalise, mutation has its greatest effect when the proportion of the favoured variant in the population is either very high or very low, and selection is most important when it is somewhere in between. My gut feeling is still that group selection is likely to be a relatively weak force in cultural evolution, but I wouldn’t bet the house on it.

Finally, I should stress that my estimate of 1,000 years as the average ‘lifespan’ of a ‘society’ was little more than a rough guess, and I hope there will be on-going debate about this. For example, from what little I know about the North American (Native American) peoples, the average lifespan of a society within historical times seems to have been shorter than this, but I don’t know how far this was due to the extraordinary impact of European settlement, horses, firearms, smallpox, etc.

DAVID BURBRIDGE