How important is group selection?

May 17, 2010 • 7:34 am

Several people on this website have raised the question of group selection, and asked me what I thought of it.  The idea that selection operates on entire groups rather than individuals, and can lead to the evolution of group-level traits (altruism is supposed to be one of these), has been revived by several people. Among them are E. O. Wilson, but especially David Sloan Wilson, who has defended his notion of group selection in a fifteen-part (!) series on HuffPo, infelicitiously called “Truth and Reconciliation for Group Selection.”

I’ve avoided discussing this topic because it’s not my own area of research and it’s a thicket of contentious claims and counterclaims that often seem more semantic than biological.  And I don’t have the mathematical expertise to appraise all the models.  Nevertheless, the field seems to be converging on a solution (which is not without dissent!).

Fortunately, you can get an excellent summary of the state of the field in a little over an hour, thanks to the London Evolutionary Research Network, a consortium of researchers who hold regular meetings and debates. Last July, they held a very nice debate, “Is natural selection at the group level an important evolutionary force?” which is now available at Vimeo.

There were four speakers, each talking for around 20 minutes, and there’s a 70 minute question-and-comment session at the end.  Here are the links to the separate videos.

Herbert Gintis (professor of economics, Santa Fe Institute, University of Siena, and Central European University)

Mark Pagel (professor of Biology, University of Reading)

Samir Okasha (professor of philosophy of science, University of Bristol)

Stuart West (professor of evolutionary biology, University of Oxford)

question-and-answer and discussion session

If you don’t have time for all of these, by all means watch Stuart West.  I don’t know how he did it, but in 20 minutes he managed to sum up the whole debate, beginning with Darwin, moving through the group selection arguments of Wynne Edwards, and assessing modern D.S.-Wilsonian views of group selection.  It’s a masterful performance.

And, as far as I can judge these things, West’s assessment is correct, and, I think, the one most smart people in that area are beginning to share:

1.  The old idea of selection among groups leading to the evolution of group-level traits works only under very special circumstances.

2. The “new” view of group selection (NGS)—the one espoused by D. S. Wilson et al.—gives results that are either wrong or, when they’re right, essentially equivalent to those derived from the simpler and less confusing inclusive-fitness theory (IFT), pioneered by Price and Hamilton and developed in the 1980s.  I’ll show Stuart’s slides:

3.  NGS has not stimulated a productive research program. Virtually every advance in understanding the effects of group-level dynamics on the evolution of social behavior has come from IFT instead. Here West checks off which of the two theories has better helped us understand various biological phenomena (in the slide below, “GS” is group selection and “IF” is inclusive fitness theory).  It’s a slam dunk for IF.

West concludes that while NGS is not usually wrong, it’s not useful, and, anyway, it’s not really new, since its mathematics were already worked out several decades ago.  While advocates of NGS claim that they’re ignored (and, at worst, persecuted), West implies that the theory is simply irrelevant.  His conclusion:

Watch all the speakers if you can, for the 80-minute debate is a painless way to educate yourself on the issues.  And kudos to the London Evolutionary Research Network for holding this and selecting a great panel of speakers.  Let’s see more of these—I’d like one on species concepts and speciation!

18 thoughts on “How important is group selection?

  1. With natural selection operating only on the frequencies of genes in a population (as per Hamilton et al), everything makes sense: you can measure those proportions amongst the individuals and get a quantitative value for evolutionary change. You can see a gene, count it, and identify dynamic statistical trends from it.

    But in group selection, what do you measure? What does one count? What is it that’s being selected for, whose frequencies are changing from generation to generation?

    There’s nothing that fits the bill, so far as I can see, which makes group selection a bit of a non-starter. And if apparent group selection can be more adequately (and more parsimoniously) explained as the effects of genetic selection, why bother with it at all?

    I think Dawkins represents this position by saying “Gene selection is the only game in town” or something similar.

    1. So it’s actually not a theory? Well, _that_ is bound to be confusing. (o.O)

      Now if evo psych was so easily dismissed, we could have a clean slate of ad hoc outgrowths. Though rumor has it shows up like the IF vs GS slide, nothing tested or maybe even rigorously predicted to show?

  2. I have been lurking around the edges for the last couple of years trying to learn about GS and whether there is anything to it. I look forward to watching all of these.

    I did read most of the D.S. Wilson 15 part series on his web site but was not convinced by his arguments and I disliked his tone and style there and unsubscribed a while back.

  3. who has defended his notion of group selection in a fifteen-part (!) series on HuffPo

    Because that’s where people go for serious science these days.

    1. Yes, it makes you yearn for those days when people simply cried “wolf!” instead of HuffPoed and puffed their empty ideas.

  4. Good to know. I had to say something about GS in a review of Nicholas Wade’s The Faith Instinct for Free Inquiry because it’s central to his claims. I settled for saying it’s contentious, or perhaps highly contentious – which is amateurese for ‘I think it’s wrong but my thinking that is entirely second-hand so I’d better be cautious.’

    That’s all right though. The point really was that Wade treated GS as unproblematic (very nearly – he did briefly mention disagreement at one point) when it decidedly isn’t.

  5. Thanks for this. As a layperson, it seems like pretty much any question I might be able to formulate about evolution is much more neatly answered by KS than by GS. But there are probably much more interesting questions I’m not even thinking of 🙂

    And — again speaking as a layperson — West’s point about GS being at best a restating of a subset of IF seems strikingly obvious now that I think about it. I mean, the way IF is phrased, any type of selection has to be a subset of it almost by definition… or at least it seems that way to me?

  6. Well, Transformers is still around as well.

    “wrong and just wasting time”

    I think I like that guy.

  7. I am as of now quite in love with Herbert Gintis and want to read his new book which annoyingly isn’t out yet.

    What’s the low-down on him? Do you proper scientist types (ie not like me) think he’s nuts?

  8. An excellent assessment of GS if I ever saw one.

    The one way in which “GS” may have some traction is by analogy with the important of whole, multicellular individual organisms. After all, a multicellular organism is just a group of maximally related single-celled organism. Even though inclusive fitness of an allele is, by definition, all that matters, we recognize that the individual success/failure of individual organisms that contain an allele is often (by far) the largest factor in determining inclusive fitness.

    There are cases in which the success/failure of entire groups of organism, regarded as a “unit”, will turn out to be the largest factor in the success failure of most or all of the alleles in those organisms. In the case of ants, bees, and termites, life/death of the whole colony and the relative success of a colony in producing daughter colonies is clearly more relevant than the sucess/failure of individuals (most of whom, being sterile, are complete “failures” in the sense of individual reproduction).

    It does seem to me, that although inclusive fitness is broadly accepted, that both biologists and lay people continue to speak in terms of selection acting on individuals without controversy despite that fact that inclusive fitness is required to explain many adaptations, especial social adaptations.

    If the GS proponents would just focus on the fact that selection acting on groups is sometimes more important than selection acting on individuals, and if they would define and use the term ‘group selection’ in the same manner that ‘individual selection’ is conventionally used, then all controversy would disapear.

  9. Hmm. I’d have to read West’s arguments to be sure – talks and slides are useless for analysis at this level of detail, frankly – but every argument I’ve read to date that declares Group Selection is mathematically subsumed by Inclusive Fitness seems to commit what D.S. Wilson (& Elliot Sober) rightly describe as “the averaging fallacy” in Unto Others. That is, the assertion that IF explains ALL the phenomena listed on slide 2 and GS explains NONE of them is in fact unsupported by the data and arguments offered, at least for some of those cases: Rather, there are several phenomena for which either or both hypotheses fit the data, and no one on either side of this debate has (yet) conceived of a way to reliably and clearly distinguish between the two. That does not in any way support the contention that GS is reducible to IF, and the further declaration that GS is “unnecessarily complex” is so much hand-waiving intended to obscure the fact that both hypotheses fit the data and that those who dislike GS cannot find a way to rule it out or isolate the relative contributions of IF & GS. Preferring the mathematically and conceptually simpler account to the more complex account is probably a good general operating principle for physics, but I’m not at all sure it can or should be applied in matters biological.

    1. IMO, the “averaging fallacy” complaint is a red herring, or perhaps a red oarfish. For any sort of IF to work there must be ‘structuring’ of the population (either spatial or behavioral) so I don’t see how it can be said to depend on some magical ‘overall fitness’.

      And I don’t think you’ve quite understood West’s claim, that if IF and GS are mathematically identical then there is little reason to prefer the more complex model!

      But don’t just take my word for it:
      (esp. #54, #34, #12 might be useful)

    2. I’m not at all sure it can or should be applied in matters biological.

      Parsimony is applicable to biology since it isn’t a one-way street.

      In physics it is, apparently, a greater likelihood for simplicity of theory. But for all theories applies that it minimizes mistakes, and also later reversals in case it wasn’t the actual model (say, by someone simpler or more predictive to replace it). (There is a paper on that effect, in fact.)

      So it makes sense as a research program to use it greedily, even in cases where a set of pathways (contingencies) is possible.

      [In fact, I thought bayesian methods in phylogenies used it to great effect?! If so, it is a direct counter-example.]

  10. It looks like many here (and at that “excellent” conference that gerrybald cites) have missed a major controversy going on among students of animal societies. It was initiated by James Hunt who studied multiple origins of sociality in a wasp phylogeny in which all the species had about the same above-background “kin selection potential” and found that sociality evolved only when there were very strong ecological incentives.

    But since ~2005 EO Wilson, Hoelldobler, etc, have hijacked the controversy.

    Some people have indeed started saying that crucial for the existence –and thus for the evolution– of animal societies is not (or not so much) “kin selection” but rather the fact that there are very rewarding ecological niches out there that allow biomachines which adopt group-approaches to foraging and interference competition to be much more effective trophically (i.e., only secondarily “evolutionarily genetically” more “successful” !) than biomachines which adopt “solitary” strategies.

    In other words: people have begun realizing that “altruism” requires also ecological advantages rather than only the promotion, or not, of altruism genes and the rejection of cheater genes.

    The social-ant colony, e.g., is an ecological machine that out-competes at the foraging- and interference-competition level most other organisms in almost any terrestrial ecological settings, i.e., a social-ant colony in the field is not necessarily just an example of an ESS immune to “selfishness” mutations that may undermine the genetic encoding of the sociality.

    EO Wilson indeed has always made a big deal of the fact that ant species monopolize nearly 70% of the insect biomass on earth, but he did not realize the implications of this until the wasp guy rubbed it to him and his coterie.

    This means that evolution by “natural selection of individuals” delivers niche-occupancy strategies that suffice to claim only ~30 of the trophic energy monopolized by the insect Bauplan (assuming termites and other social insects are insignificant biomass-wise).

    The situation among many mammals is the same. Wild-dog packs, e.g., beat the hell out of tigers, and biomass wise the packs must dominate…

    It is time for gratuitous faux-a-prioristic flawed-math arguments to be confronted with ultimate natural-historical facts.

    And it is also time that the too-many cheapo-applied-math peddlers posturing as evolutionary biologists learn that natural selection is not the same as evolution by natural selection, that differential fitness is always caused by differential ecological performance (and never by “genes”), and that evolution by natural selection is just something that “may” happen to genes provided that there is differential ecological performance at some level of biological organization, which however does not “prove” that the differential ecological performance is at the level of genes.

    In other words, Sober’s 1984 [1984 sic!] book “The Nature of Selection” should be required reading for every evolutionary biologist.

    PS. i found this very recent paper below that studied this group-performance vs. “kin-selection” issue within a clade of sponge-living shrimp.


    Kin structure, ecology and the evolution of social organization in shrimp: a comparative analysis

    Author(s): Duffy JE (Duffy, J. Emmett)1, Macdonald KS (Macdonald, Kenneth S.)2

    Source: PROCEEDINGS OF THE ROYAL SOCIETY B-BIOLOGICAL SCIENCES Volume: 277 Issue: 1681 Pages: 575-584 Published: FEB 22 2010

    Times Cited: 2 References: 62 Citation MapCitation Map

    Abstract: Eusocial societies present a Darwinian paradox, yet they have evolved independently in insects, mole-rats, and symbiotic shrimp. Historically, eusociality has been thought to arise as a response to ecological challenges, mediated by kin selection, but the role of kin selection has recently been questioned. Here we use phylogenetically independent contrasts to test the association of eusociality with ecological performance and genetic structure (via life history) among 20 species of sponge-dwelling shrimp (Synalpheus) in Belize. Consistent with hypotheses that cooperative groups enjoy an advantage in challenging habitats, we show that eusocial species are more abundant, occupy more sponges, and have broader host ranges than non-social sister species; and that these patterns are robust to correction for the generally smaller body sizes of eusocial species. In contrast, body size explains less or no variation after accounting for sociality. Despite strong ecological pressures on most sponge-dwellers, however, eusociality arose only in species with non-dispersing larvae, which form family groups subject to kin selection. Thus, superior ability to hold valuable resources may favour eusociality in shrimp but close genetic relatedness is nevertheless key to its origin, as in other eusocial animals.

  11. Hi there,

    I’m the secretary for LERN (London Evolutionary Research Network) which you mention in the post. First of all, thank you for mentioning our society, the debate was our most successful event to date, and second, we would have very much enjoyed having you as one of our speakers in this year’s LERN debate on epigenetics, unfortunately we didn’t hear back from you so we had to go with other people! In case you are interested, here is the event’s information and the line-up:



Leave a Reply