Evolution for Everyone

The rehabilitation of group selection took another step with the publication of a commentary titled “Eight Criticisms Not to Make About Group Selection” in the journal Evolution. The authors are Omar Eldakar, my former PhD student currently at the University of Arizona’s Center for Insect Science, and yours truly. Omar conducts wonderful research on male mating strategies in water striders from a multilevel perspective, which I have previously reported upon. Our commentary was inspired by the struggle that we and all other authors who wish to employ a multilevel perspective must endure when we try to publish our work in peer reviewed journals.

In Omar’s case, it might be an article showing that docile male mating strategies are favored by between-group selection, even though docile males are trounced by more aggressive males within single groups. Some reviewers praise the research as an elegant demonstration of multilevel selection in a novel context. Other reviewers also praise the research but raise objections such as these:

“Please remove all references to group selection… significant or not, group selection is not going on here.”

“It is all very well to look at success within and between groups, but can you really extrapolate this to group selection?”

“What the authors show is not group selection, but rather that … mating success depends on the social environment”

” This is a controversial topic, worthy of a mention, but not worthy of the extended discussion, because it is largely lacking in both formal theory and empirical support to justify it.”

“I am not sure how this behavior can be maintained under a group selection scenario unless there is no dispersal and populations frequently go extinct.”

If you know even a little about multilevel selection, you’ll agree that these objections are sadly misinformed. How is this possible, when the peer review process is supposed to rely upon expert opinion? The answer is that all of the reviewers were indeed expert on mating behavior, but only some were expert on multilevel selection.

The same problem exists for any other subject that can be approached from a multilevel perspective. Reviewers who are knowledgeable about the subject and multilevel selection tend to like the research in all respects, while reviewers who are only knowledgeable about the subject at hand raise objections about group selection that are sadly out of date. The hapless authors must then fight to retain their explicitly multilevel perspective, often with an uncertain editor acting as a referee.

To address this problem, Omar and I list eight criticisms about group selection that can be permanently laid to rest based upon current knowledge. Even the gang of 139 that can’t be wrong about inclusive fitness theory would not raise these objections about group selection, although they might raise other more nuanced objections. Here they are, although you’ll need to read the article for our commentary.

Eight Criticisms Not to Make About Group Selection

Criticism #1: The fact that a trait evolves in the total population is an argument against group selection.

Criticism #2: If a trait increases the absolute fitness of an individual, it does not require group selection to evolve.

Criticism #3: Conceptualizing the group as the social environment of the individual is an argument against group selection.

Criticism #4: Frequency-dependent selection is an argument against group selection.

Criticism #5: The fact that a trait can be measured in individuals means that it evolved by individual-level selection.

Criticism #6: Group selection is theoretically implausible.

Criticism #7: There is little empirical support for group selection.

Criticism #8: Group selection requires limited dispersal among groups.

Omar and I end our commentary by pointing out an irony. Knowing the difficulties that they will encounter during the peer review process, many authors avoid using the G-word even when group selection is being invoked in every way except the name. Their silence perpetuates confusion about group selection. The individual gains publications and grants over the short term, while the field as a whole remains muddled about the fundamentals of social evolution. Selection against group selection in the peer review process is a case of the tragedy of the commons in academic cultural evolution.

Confusion about group selection at the professional level will end when we eliminate selection against group selection accounts in the review process. Our commentary is intended to accomplish this goal and we are pleased that it has been published in an authoritative journal such as Evolution.


  1. #1 Russ Abbott
    April 12, 2011

    In Evolution for Everyone you make the point that virtually everything is both an entity and a component of a larger entity. It seems to me that such a perspective is sufficient to allow one to talk about “group selection” simply by talking about selection of the larger entity of which the gene-bearing entity is a member.

    After all, cells are the real bearers of genes, but we don’t talk about the evolution of cells. We talk about the evolution of biological organisms, which may be composed of cells. In fact, at the level of single-celled organisms, it’s sometimes a stretch to talk about evolution at the individual level when so much of evolution at that level involves gene swapping.

    The evolution of multi-cell organisms is group selection in that what is selected is a combination of genes that work successfully together. It isn’t the individual genes or the individual cells that are generated that evolve. It’s the ensemble that evolves. A similar point can be made about evolution of social groups.

    Of course I know you agree with that. What I don’t see is why that’s such a difficult case to make.

  2. #2 stripey_cat
    April 12, 2011

    Russ, it’s because group selection is Wrong. It’s an emotional value-judgement now in many cases. You can’t reason someone out of a position they weren’t reasoned into.

    On a positive note, if my experience as an arts undergrad is relevant, the up-and-coming generation will have no problem with multilevel selection: it’s the generation that was taught in reaction against early, misinformed forms of a concept that will resist any manifestation of it tooth and nail.

  3. #3 OTE
    April 12, 2011

    @Russ. I’ve always thought that widespread rejection of group selection in the 1960s created an environment in which understanding the logic of group selection was secondary to the knowing the basic notion that group selection is wrong. Therefore the ‘logic’ of group selection has been disconnected from the term ‘group selection’ so no one will argue against your formulation, they will just say that what you’re talking about is not group selection.

  4. #4 Sam C
    April 13, 2011

    Is sneering a group level trait amongst group selection afficionados? You clearly envy Richard Dawkins for his past popularity. Insulting everybody who is not A True Believer might make you happy, but it does not win you friends. What is the evolutionary basis for that behavior.

    Basics: fitness is not a trait. It is not in itself heritable. It is an artificial construct. There is differential replication of heritable characteristics by biological entities, that’s fact and the motor of evolution by both selection and drift. How you pile that together into an academic artifice is, to some extent, arbitrary.

    Your argument is largely a figment of a one-eyed imagination; multiple competing formalisms do not entail multiple competing realities.

  5. #5 Ben Hillman
    April 13, 2011

    Speaking as a civilian and not a scientist, I could never understand what the problem was with group selection altogether. Every culture has a warrior class and groups wipe one another off the face of the planet. You lose, you’re gone. The greatest praise in most societies goes to those who give their lives for their country. If that is not group selection, then I don’t know what is. I also remember seeing the same self-sacrificing behavior in the Errol Morris movie about naked mole rats.

  6. #6 joe
    April 15, 2011

    I totally agree that the above criticisms #1-8 are invalid.

    However, group selection seems to be defined by most biologists as “that what W.D. Hamilton and G.C. Williams rejected.”

    Hamilton and Williams rejected most fervently the idea that the evolutionary purpose of traits, like altruistic behaviour, is the good of the species or group.

    They instead favoured the gene’s eye view according to which altruism maximises inclusive fitness (even if the process involves between-group selection swamping within-group selection).

    None of the above points #1-8 changes anything about this rejection of Hamilton and Williams or the possibility to prefer the gene’s eye view.

    As this rejection of reasoning for the good of the group/species has been defining the term “group selection”, however, a “rehabilitation of group selection” would have to show that altruism evolved for the good of the group.

    As inclusive fitness formulations are mathematically equivalent, however, I do not see how a consensus could possibly be reached. Those against group selection will always be free to insists that the quantity maximised is inclusive fitness not group benefit.

    I think the iteresting question is, under which conditions the inclusive fitness of the lower level units becomes a coprehenisve fitness of the higher level unit.

    My hunch is that the answer can be found in literature dealing with the major evolutionary transitions.

  7. #7 royniles
    April 23, 2011

    How could altruism NOT have evolved for the good of the group. Since organisms are in one way or another the product of the social structure that all organisms have necessarily formed and been formed with and within, is an alternative or non-good purpose for the function conceivable? Unless of course you’re one of those who don’t believe that the functional strategies of biology either have or serve a purpose, or were ever meant to..

  8. #8 joe
    April 25, 2011


    Altruism could have evolved for the good of the individual, their inclusive fitness, or for the good of genes. The simplest counter model to your question is probably the green-beard model. If a gene has two effects, one being that its bearer has a green beard, the other that is behaves altruistically to bearers of green beards, green-beard-altruism can evolve exclusively for the good of gree-bearded individuals or green-beard genes. Cooperation can also evolve (or be learned) in game theoretic settings.

    My understanding until recently was that cases where between-group selection swamped within-group selection were cases of group selection but not of group benefit. But that seems to be an open issue.

  9. #9 royniles
    April 25, 2011

    The problem is that in the end, individuals need their groups as a media for the communication that seems necessary for any adaptations to environment (including themselves in that environment), and thus develop cooperative traits that in the end all individuals in the species will consistently share. Altruism requires groups and groups would seem in their turn to require altruism.
    That individuals in turn found benefit in non-cooperation, and value in competitiveness, may well have been of between group benefit, but also of within group benefit and contribute to the degree of diversity that has to be of strategic benefit in what would ultimately evolve to be a competitive game of life. All group competitions in our games requiring some non selfish within group behaviors – altruism in the end being a form of non-selfishness.
    All traits in the end arising from strategically successful behaviors. Arguably that is.

  10. #10 supratall
    May 1, 2011

    They instead favoured the gene’s eye view according to which altruism maximises inclusive fitness (even if the process involves between-group selection swamping within-group selection).

  11. #11 realscienceexperience
    May 12, 2011

    Hello Evolution fore Everyone readers!

    Here is another great blog for you to enjoy: realscienceexperience.wordpress.com

    The Real Science Experience is a blog written by university students who have come from various backgrounds (and countries) and have followed the Bachelor of Science down different pathways from undergraduates to masters. On this blog we discuss what it is really like to study science at a tertiary level.

  12. #12 Logan
    August 20, 2011

    Doesn’t Dawkins’s conceptualization of extended phenotype in evolutionary game theory situations (with inclusive fitness, of course) explain how selection at the gene level produces (is) selection for group behaviors? The seeming group selection is just selection for a gene whose phenotype is active in game theory situations and so can by cooperating can increase its payout. This looks altruistic, but is not because the genes being helped are actually just copies of the ones helping. Thus, your liver does not need to reproduce, because it IS your sperm. They’re the same genes. Consequently (since individuals are just groups of genes), individuals in a group are not entirely different individuals…they are copies. Thus, when i perform an “altruistic” action, I am actually just helping my own copies to survive…i am being selfish. Precisely the same logic can be applied to apoptosis.

    Do we ever see sustainable group selection for groups of genetically diverse organisms (elephants and lions; flowers and leopards; genetically diverse human populations…)? I would contend that in the entirety of history we haven’t. Even within a country like America, whose citizens have mostly (in theory) the same interests, genetic division is the dominant force shaping social policy and behavior.

    So…I guess…I don’t understand the argument in favor of group selection. Why isn’t selfish gene philosophy + extended phenotype + inclusive fitness considered an uncontroversial explanation? Can someone explain it?

  13. #13 Logan
    August 20, 2011

    Also, isn’t it relevant that altruism is actually not altruistic, but simply the first play of a tit-for-tat algorithm? I mean…you could see it as altruistic…but ontologically it just isn’t. A person (the vast majority of the time) is not going to send you a cake if you shoot his mother…he’s gonna come tat you. We are nice to strangers to the extent that it is convenient for us, but that is simply because in game theory situations (and many situations are) it is far more efficient assume cooperate and lose a round occasionally than assume defect and immediately kill the top payout for each round.