As most of you know, Edward O. Wilson is one of the world’s most famous and accomplished biologists. He was the founder of evolutionary psychology (known as “sociobiology” back then), author of two Pulitzer-Prize-winning books, one of the world’s great experts on ants, an ardent advocate for biological conservation, and a great natural historian. His legacy in the field is secure.
So it’s sad to see him, at the end of his career, repeatedly flogging a discredited theory (“group selection”: evolution via the differential propagation and extinction of groups rather than genes or individuals) as the most important process of evolutionary change in humans and other social species. Let me back up: group selection is not “discredited,” exactly; rather, it’s not thought to be an important force in evolution. There’s very little evidence that any trait (in fact, I can’t think of one, including cooperation) has evolved via the differential proliferation of groups.
In contrast, there is a ton of evidence for an alternative explanation for cooperation: kin selection, the selection of genes based on how they affect not just the fitness of the individual, but the fitness of relatives that share its genes. Features like parental behavior, parent-offspring conflict, sibling rivalry, and preferential dispensing of favor to relatives, as well as features like sex ratios in insects—all of these are all easily explained by kin selection. And many aspects of cooperation can easily be explained by individual selection: individuals that live in small groups, especially those in which one can recognize group members, can evolve cooperation as an individual good based on reciprocity: the “I scratch your back, you scratch mine” hypothesis. And, as I’ve discussed before, the cooperative and “altruistic” behavior seen in our own species shows many features suggesting that it evolved via individual or kin selection and not group selection.
I’ve covered this issue many times (e.g., here, here, here, here, and here), so I won’t go over the arguments again. Wilson’s “theory” that group selection is more important than kin selection in the evolution of social behavior (published in Nature with Martin Nowak and Corina Tarnita) was criticized strongly by 156 scientists—including virtually every luminary in social evolution—in five letters to the editor, and sentiment about the importance of group selection has, if anything, decreased since Wilson’s been pushing it.
But Wilson persists, to the detriment of his reputation. In a new piece at the New York Times “Opinionator” site, “The riddle of the human species,” Wilson continues to make the same argument that group (or “multilevel”) selection was a key force in making humans (and social insects) the socially complicated species they are. Since his arguments are virtually identical to those published in a NYT Opinionator piece last June, and in his book The Social Conquest of Earth (see part of my review here), I won’t dissect them in detail. I just want to highlight three points that I think make Wilson’s argument for group selection—and against kin selection—deeply misleading. I wouldn’t spend my time writing time-consuming critiques like this were Wilson not famous, influential, and given a big public forum in the New York Times. Someone has to address his arguments!
Here are Wilson’s errors (quotes indented), and my responses:
1. Wilson: Humans are a “eusocial species”:
. . the known eusocial species arose very late in the history of life. It appears to have occurred not at all during the great Paleozoic diversification of insects, 350 to 250 million years before the present, during which the variety of insects approached that of today. Nor is there as yet any evidence of eusocial species during the Mesozoic Era until the appearance of the earliest termites and ants between 200 and 150 million years ago. Humans at the Homo level appeared only very recently, following tens of millions of years of evolution among the primates.
My response: “Eusociality” as defined by Wilson and every other evolutionist is the condition in which a species has a reproductive and social division of labor: eusocial species have “castes” that do different tasks, with a special reproductive caste (“queens”) that do all the progeny producing, and “worker castes” that are genetically sterile and do the tending of the colony. Such species include Hymenoptera (ants, wasps and bees, though not all species are eusocial), termites, naked mole rats, and some other insects.
But humans don’t have reproductive castes, nor genetically determined worker castes. Wilson is going against biological terminology, lumping humans with ants as “eusocial,” so he can apply his own theories of “altruism” in social insects (i.e., workers “unselfishly” help their mothers produce offspring while refraining themselves from reproducing), to humans. But human cooperation and altruism are very different from the behavior of ants, most notably in our absence of genetic castes and genetically-based sterility associated with helping others reproduce. Human females aren’t sterile, and don’t usually refrain from reproduction just to help other women have babies. My guess is that Wilson lumps humans with insects as “eusocial” because he wants to subsume them both under a Grand Theory of Social Evolution.
2. Wilson: Kin selection doesn’t work, ergo it certainly couldn’t have played a role in the evolution of eusociality and human cooperation.
Still, to recognize the rare coming together of cooperating primates is not enough to account for the full potential of modern humans that brain capacity provides. Evolutionary biologists have searched for the grandmaster of advanced social evolution, the combination of forces and environmental circumstances that bestowed greater longevity and more successful reproduction on the possession of high social intelligence. At present there are two competing theories of the principal force. The first is kin selection: individuals favor collateral kin (relatives other than offspring) making it easier for altruism to evolve among members of the same group. Altruism in turn engenders complex social organization, and, in the one case that involves big mammals, human-level intelligence.
The second, more recently argued theory (full disclosure: I am one of the modern version’s authors), the grandmaster is multilevel selection. This formulation recognizes two levels at which natural selection operates: individual selection based on competition and cooperation among members of the same group, and group selection, which arises from competition and cooperation between groups. Multilevel selection is gaining in favor among evolutionary biologists because of a recent mathematical proof that kin selection can arise only under special conditions that demonstrably do not exist, and the better fit of multilevel selection to all of the two dozen known animal cases of eusocial evolution.
My response: There is so much fail here I don’t know where to start. The first paragraph is basically correct except that Wilson omits “individual selection” along with “kin selection” as an accepted evolutionary process that can promote the evolution of cooperation. As I mentioned, selection on individuals in small groups can allow the evolution of cooperation without any need to invoke the unparsimonious process of differential group survival based on genes.
Wilson’s claim that the “special conditions of kin selection” demonstrably do not exist is an egregious and (I think) willful misstatement. Kin selection can cause evolution whenever the genes in an individual benefit relatives that share copies of that individual’s genes, and can do so whenever the benefit of that behavior to the recipients, devalued by their degree of relatedness to the donor (a figure usually ranging between 0 and 1, but which can be related if an individual helps another less related to it than the average member of the population) is greater than the reproductive cost to the donor. (“Hamilton’s rule”: rb > c.) That is known to obtain in many cases, and explains things like parental care, parent-offspring conflict, sex ratios in insects, and many other features (see the five letters in Nature mentioned above, which list some features of social behavior that clearly evolved by kin rather than group selection).
The mathematical “proof” given by Nowak et al. does not show that group selection is a better explanation than kin selection for social behavior in insects, for their “proof” does not vary the level of kinship, as it must if it could allow that conclusion.
The second egregious and false claim in this paragraph (a paragraph that’s the highlight of the piece) is that “multilevel selection is gaining in favor among evolutionary biologists” because of the Nowak et al. paper. That’s simply not true. The form of multilevel selection adumbrated in that paper is, to my knowledge, embraced by exactly four people: the three authors of the paper and David Sloan Wilson. There is, and has been, no increase in acceptance of group or multilevel selection in the past ten years. The Nowak et al. paper has sunk without a stone, except to incite criticism by other biologists and excitement by an uncomprehending press.
3. Wilson: Eusociality in insects arose not via kin selection, but via the initial construction of a defended nest site.
The history of eusociality raises a question: given the enormous advantage it confers, why was this advanced form of social behavior so rare and long delayed? The answer appears to be the special sequence of preliminary evolutionary changes that must occur before the final step to eusociality can be taken. In all of the eusocial species analyzed to date, the final step before eusociality is the construction of a protected nest, from which foraging trips begin and within which the young are raised to maturity. The original nest builders can be a lone female, a mated pair, or a small and weakly organized group. When this final preliminary step is attained, all that is needed to create a eusocial colony is for the parents and offspring to stay at the nest and cooperate in raising additional generations of young. Such primitive assemblages then divide easily into risk-prone foragers and risk-averse parents and nurses.
My response: Phylogenetic studies show that eusociality in Hymenoptera always originated in species whose females mated only once: this is a statistically significant result. And that alone militates for kin selection as an important factor in eusociality: if a female founds a colony consisting only of full siblings (as is the case when she mated only once), they are more related to each other than if she had mated multiply. In the later case, colonies would consist of half-sisters or even more distant relatives, making kin selection less efficient.
Further, relatedness is high in virtually every species of eusocial insect with the exception of a few highly derived species of ants that have many queens. The connection between relatedness and eusociality is exactly what we expect if kin selection is important in social evolution, and is not expected if Wilson’s nest-based group selection was important. The model of Nowak et al., which starts with the construction of such nests by single females who stay in the nests with their offspring, produces precisely the condition in which relatedness can promote the evolution of sterility and cooperation. They argue that this relatedness is a consequence of their model and not a cause of eusocial evolution, but that’s unconvincing, for they do not vary the level of initial relatedness in their model.
Wilson’s claim, the theme of his newest book, is that humans are both angels and devils: we are both selfish and cooperative species, and this combination of good and bad is what makes our species unique. (That’s not true, of course, because many species show that mixture of behavior. Lions, for instance, cooperate when hunting, but when males take over a pride they immediately kill all the female’s cubs, which are unrelated to them. And that, by the way, is due to kin selection, because those cub-killing males replace the cubs with new cubs containing their own genes, including the genes for killing cubs. Cub-killing could have evolved only by individual selection and not group selection, for while killing another male’s cubs is good for an individual, it’s bad for the group, forcing females to waste reproductive energy.)
Yes, we have both selfish and cooperative behaviors, though most of our “cooperative” behaviors that didn’t arise through culture arose through forms of selection that involve maximizing our reproductive output—individual and kin selection. There is not a scintilla of evidence, in humans or any other species, that group selection has been responsible for the evolution of any adaptation. In contrast, individual and kin selection have productively explained the evolution of “problematic” traits like altruism and cooperation. They have been tested and work.
Why does Wilson keep writing article and article, and book after book, promoting group selection? I’m not a psychologist, so I don’t know the answer. What I do know, though, is that his seeming monomaniacal concentration on a weakly-supported form of evolution can serve only to erode his reputation. His theories have not gained traction in the scientific community. That doesn’t mean that they’re wrong, for, in the end, scientific truth is decided by experiment and observation, not by the numbers of people initially on each side of an issue. But the facts of science already show that Wilson is unlikely to be correct. What is sad is that, as a great natural historian, he doesn’t recognize this.
Wilson’s reputation is secure. It’s sad to see it tarnished by ill-founded arguments for an unsubstantiated evolutionary process.
h/t: Phil Ward, Laurence Hurst