The Group Delusion

Jan 9, 2008

by Richard Dawkins

This week’s New Scientist has an article by Daniele Fanelli announcing an apparent change of mind by E O Wilson. This has been picked up by the Daily Telegraph under the headline Scientist renounces insect ‘kin selection’ theory and by the Independent under the headline Evolutionists at war over altruism’s origins . New Scientist asked me to reply, but they gave me a very tight limit of 650 words. I decided that I could fit into this limit only with references to other publications, and I took great care to upload those publications to the web, and asked New Scientist to publish the url:

http://www.simonyi.ox.ac.uk/dawkins/writings/index.shtml

Unfortunately they omitted to do so.

My article, such as it is, follows. Josh has very kindly set up hyperlinks to the other articles.

EDWARD WILSON has given us a characteristically fascinating account of the evolution of social insects (see page 6 and BioScience, vol 58, p 17). But his “group selection” terminology is misleading, and his distinction between “kin selection” and “individual direct selection” is empty. What matters is gene selection.

All we need ask of a purportedly adaptive trait is, “What makes a gene for that trait increase in frequency?” Wilson wrongly implies that explanations should resort to kin selection only when “direct” selection fails. Here he falls for the first of my “12 misunderstandings of kin selection (pdf)“, that is, he thinks it is a special, complex kind of natural selection, which it is not (Zeitschrift für Tierpsychologie, vol 51, p 184).

In the true sense of kin selection, offspring are “kin” just as siblings are. Parental care and sibling care both evolve because copies of genes for caring are present in beneficiaries. Genes promoting feeding of larvae by sterile workers are passed on by those larvae — sisters, nephews, and so on — destined to become reproductives. That’s kin selection, and it maintains sterile worker castes in insect colonies. Wilson could not dispute that.

What he does dispute – perhaps correctly – is that eusociality originated through related females clubbing together because of kinship. It could also originate through unrelated females nesting together. But to call this “group selection” is massively confusing. A better approach is John Maynard Smith’s concept of evolutionarily stable strategies (ESS). “Stable” means that when most individuals follow the strategy, no alternative does better. If “breed cooperatively” were a stable strategy for unrelated females, this would furnish a good preadaptation for the evolution of eusociality.

Jane Brockmann, now at the University of Florida, Gainesville, and I explored this (pdf) with an ESS model (pdf) developed with Alan Grafen of the University of Oxford, using Brockmann’s extensive fieldwork on solitary digger wasps, Sphex ichneumoneus. When following the ‘dig’ strategy, a female digs a burrow, provisions it with prey on which she lays a single egg, seals the burrow and departs. But burrows may be abandoned (for good reasons that can then disappear) and this opens the way for an alternative strategy: ‘enter’ an existing burrow and take it over, saving the time and effort of digging. The disadvantage of entering is that the original owner may not have abandoned the burrow, and you run the risk of a dangerous fight. So the decision whether to enter or dig is a gamble.

With too much entering in the population, not enough new burrows get dug and chances rise that a given burrow will be occupied. Selection would therefore favour digging. With too much digging, many abandoned burrows go begging, and individuals should enter instead. Grafen’s ESS model predicted an equilibrium frequency of digging versus entering with equal benefits to each. Brockmann’s field measurements were rich enough to test this prediction, and it was, with reservations, fulfilled.

Brockmann and I then postulated an ecological “landscape” (pdf) over which the parameters governing Grafen’s model might vary. A change in ecological conditions might move digger wasps from “aggressive space” (where S. ichneumoneus sits) to “tolerant space” (where diggers benefit from being joined by an enterer). From here there is a smooth gradient to “cooperative space”, where both parties benefit from sharing. Our review of the literature uncovered wasp species that apparently take such intermediate positions. From here, the evolutionary journey to full eusociality is easy.

Revealingly, Wilson’s great book Sociobiology allots only four sentences — in the chapter on group selection — to ESS theory. Kin selection is also here, as a form of group selection (Misunderstanding #2 (pdf) )! Evidently Wilson’s weird infatuation with “group selection” goes way back, unfortunate in a biologist who is so justly influential.

My articles referred to can all be found at
http://www.simonyi.ox.ac.uk/dawkins/writings/index.shtml

As a prelude to this rather overblown episode, E O Wilson collaborated with his (unrelated) namesake D S Wilson on a paper on Group Selection, also in New Scientist. I did not write a proper reply (I was unfortunately extremely busy) but I did write a letter, correcting one outrageous lie about me, and New Scientist solicited a reply from the Wilsons in the same issue:

Genes still central
Richard Dawkins writes:

David Sloan Wilson’s lifelong quest to redefine “group selection” in such a way as to sow maximum confusion – and even to confuse the normally wise and sensible Edward O. Wilson into joining him – is of no more scientific interest than semantic doubletalk ever is. What goes beyond semantics, however, is his statement (it is safe to assume that E. O. Wilson is blameless) that “Both Williams and Dawkins eventually acknowledged their error…” (3 November, p 42 ).

I cannot speak for George Williams but, as far as I am concerned, the statement is false: not a semantic confusion; not an exaggeration of a half-truth; not a distortion of a quarter-truth; but a total, unmitigated, barefaced lie. Like many scientists, I am delighted to acknowledge occasions when I have changed my mind, but this is not one of them.

D. S. Wilson should apologise. E. O. Wilson, being the gentleman he is, probably will.



David Sloan Wilson and Edward O. Wilson write:

Our comment about Dawkins specifically relates to the error of using the replicator concept – genes as the “fundamental” unit of selection – as an argument against group selection. Dawkins writes in The Extended Phenotype (p 115): “The point here is that we must be clear about the difference between those two distinct kinds of conceptual units, replicators and vehicles… The majority of models ordinarily called ‘group selection’… are implicitly treating groups as vehicles. The end result of the selection discussed is a change in gene frequencies, for example, an increase of ‘altruistic genes’ at the expense of ‘selfish genes’. It is still genes that are regarded as the replicators which actually survive (or fail to survive) as a consequence of the (vehicle) selection process.”

Another error is to suppose that within-group selection poses an insuperable problem for between-group selection. Dawkins has yet to acknowledge this error and we apologise if our article seemed to imply otherwise. Finally, Dawkins seems to think that one of us has somehow confused the other. We are united in our view about group selection, which we converged upon through separate lines of enquiry.



The Wilson quotation from The Extended Phenotype is a ludicrous attempt to justify their lying statement that I “eventually” acknowledged an earlier error. For one thing, The Extended Phenotype was published way back in 1982, which makes nonsense of Wilson’s “eventually”. But more importantly, the point I was making in 1982 (and would make again now) was a general one about the important distinction between replicators and vehicles. I mentioned group selection only to clarify that distinction. I was explaining that those models of group selection that had been proposed were vehicle models not replicator models. I was not for a moment suggesting that I accepted those models as valid. They were (and are) invalid vehicle models, as opposed to invalid replicator models.

I might well write something more lengthy about this, but I thought I ought to put something up about it now, in view of the attention that it seems to be getting in the newspapers. It is a great pity Bill Hamilton is not alive to defend his brilliant ideas.

Leave a Reply

View our comment policy.