New Yorker article on the evolution of altruism and the kin-selection flap

March 8, 2012 • 5:26 am

UPDATE:  Dr. Philip Ward, a colleague who is an ant systematist and evolutionary biologist, has offered his critique of the Lehrer article, which I’ve added as a comment below the following: 


In several previous posts (e.g., here and here), I’ve described the recent dust-up about whether kin selection (selection of “altruisitic” traits based on their effect on increasing the reproductive fitness of relatives) was an important cause of social evolution in nature, particularly in the evolution of “eusociality” (social insects with a nonreproductive caste and a reproductive queen) and of “altruism” (animals who appear to injure their own evolutionary prospects by helping non-relatives).

The controversy was inspired by a paper published in 2010 in Nature by Martin Nowak et al. (reference at bottom, other authors are Corina Tarnita and the eminent biologist E. O. Wilson), maintaining that kin selection was not a good way to analyze evolution in nature, that it was not a form of natural selection (WRONG), and that the evolution of eusociality was better explained by “group selection” (differential reproduction of groups) than by selection among relatives.

To most biologists, this controversy appears to have been settled: Nowak et al. were wrong. About a hundred and fifty biologists wrote five separate critiques that were published in Nature, along with a lame response by Nowak et al.  As far as I know (and I may have missed something), no paper has been published in support of Nowak et al., and only a handful of biologists (I think David Sloan Wilson was one of them) even supported it with public statements.  Referring to Nowak et al.’s lame reply to their critics, science writer Carl Zimmer said this:

Nowak et al respond to all the criticism and don’t budge in their own stand. They claim that their critics have misinterpreted their own argument. And they claim that sex allocation does not require inclusive fitness. Oddly, though, they never explain why it doesn’t, despite the thousands of papers that have been published on inclusive fitness and sex allocation. They don’t even cite a paper that explains why.

Now Jonah Lehrer has written a big piece in the New Yorker about this controversy (and about the personalities involved), “Kin and Kind.” (It’s behind a paywall, but I’ve scanned it and will send interested readers that scan if they email me.)  Lehrer is a young science writer about whom I have mixed feelings. He’s a good journalist, and his piece is absorbing, but it suffers from several problems.  (You may remember Lehrer as the author of another overblown New Yorker piece about the inherent untrustworthiness of science, “The Decline Effect.”)

What is my take on the piece? As I said, it’s absorbing but flawed.  It will draw you in and teach you a bit about the controversy, but it ultimately suffers from Lehrer’s “he said/she said” noncommittal stance about the article, and from his unwillingness to make any judgment about the scientific issues at hand. It’s not that he’s incapable of that, I think, since he was trained in neuroscience and has written two books on that subject, but he’s more interested in controversy than scientific truth. Such is The New Yorker, which really needs to dig up some better science writers (can I suggest John Crewdson?).

Here are the problems, starting with the trivial ones:

  • There is a mistake, which is no big deal in most magazines but is unforgivable in the New Yorker, which has a scrupulous policy of checking and rechecking every single fact. Referring to the criticisms of Wilson, Lehrer says, “There have been denunciations in the press and signed group letters in prestigious journals; some have hinted that Wilson, who is eighty-two, should retire.” In fact, Wilson retired some years ago.
  • Another trivial thing: Lehrer refers to Wilson’s monograph on the ant genus Pheidole, as an “eight-hundred page textbook”. It’s not a textbook, for it’s not for use in any class. It’s a monograph, and no scientist, much less Wilson, would call it a textbook. (As one colleague told me, “I don’t think I’d like to take that class.”)  The mention of this monograph is notable because, according to Lehrer, Wilson is prouder of it than of his famous books Sociobiology and On Human Nature, since Wilson now decries the kin selection that infused those two books. Had Lehrer done a little more digging her (he didn’t dig enough for the entire piece!), he would have found out that this monograph is not held in high regard by other ant systematists, for it doesn’t use modern methods of taxonomy: in particular, it doesn’t take into account variation within species, a sine qua non for proper analysis of species. My ant-y colleagues are of a piece in this opinion, but haven’t criticized the monograph because of Wilson’s status; one person who has criticized  Wilson’s taxonomic work on ants tangentially, including the monograph, is Dr. Alex Wild, author of the superb ant blog Myrmecos (see here and here).
  • The bigger problem: Lehrer doesn’t really delve deeply enough into the controversy to be able to render an opinion, and he mischaracterizes the fracas, simplistically, as a battle between the mathematicians (on the Nowak et al. side) and the biologists (all the critics):
“The mathematicians insist that their critics don’t understand the math, and the biologists insist that the mathematicians don’t understand the biology.”
This is completely bogus.  Many of the authors on the critiques of the Nowak et al. paper were mathematical biologists with a high degree of skill in theoretical analysis, and certainly with the ability to analyze Nowak et al.’s math (Stuart West is a notable example).  And Wilson, of course, is a biologist (Lehrer does note that Wilson doesn’t understand much of his colleagues’ math.)
  • Lehrer discusses haplodiploidy: the system of reproduction in many eusocial insects, which involves females laying haploid, unfertilized eggs that become males and diploid eggs fertilized by a haploid male who mates with the queen; those eggs become sterile female workers. This system produces the peculiar result that female workers share 3/4 of their genes with their sisters—instead of the regular 1/2—and hence may have more of a genetic interest in becoming sterile and helping their mother produce sisters than having their own offspring (to whom they’re related by only 1/2). It’s been known for a long time that while haplodiploidy is associated with eusociality, the association is imperfect: some non-haplodiploid species are eusocial, like termites and naked mole rats, while other haploidiploid species aren’t eusocial.
Lehrer uses this fact to dismiss the importance of kin selection in the evolution of eusociality.  But he neglects a very important paper that says the opposite: a paper of Hughes et al. in Science in 2008 (reference below). That paper shows unequivocally that eusociality in insects involved kin selection: in all eight cases in the Hymenoptera (ants, bees and wasps) in which eusociality evolved, the ancestral species had queens that mated only once rather than multiply.This association was highly statistically significant.
Under the kin-selection idea, single mating facilitates the evolution of sterility since females won’t be related by 3/4 if there is more than one father. The group-selection argument of Nowak et al., which doesn’t depend on relatedness, predicts that eusociality won’t be associated with whether or not females mate once or more than once in the ancestral lineage.
The Hughes et al. paper is prima facie evidence that relatedness, and kin selection, facilitates the evolution of eusociality.  It clearly shows that Nowak et al. are wrong, but the paper isn’t mentioned by Lehrer.
  • If Lehrer had examined the math in the Nowak et al. paper, he could easily have seen that their model was incapable of judging the effects of kin selection, for it didn’t vary the degree of relatedness to see if that would make the evolution of eusociality easier. This was, I think, pointed out in some of the critiques published in Nature.  Lehrer just takes a “hands-off” approach to the model.  But had he dug a little, or done a little inspection guided by a modeller, he could have seen this. Instead he just throws up his hands and says the issue is unresolved.  It isn’t. Nowak et al.’s model cannot say anything about whether relatedness facilitates the evolution of eusociality.
  • Finally, at the end of the paper, Lehrer implicitly accepts Ed Wilson’s argument that altruism in animals must evolve by group selection rather than kin selection, for kin selection simply isn’t sufficient.  That assertion is completely wrong, and Lehrer should have known that. “Altruism,” at least as “evolved altruism” that is not completely self-sacrificial, can evolve via either kin selection in groups of relatives or via individual selection if individuals can recognize and repay others who help them via “altruisitic” acts.  The latter process is known as “reciprocal altruism,” and it’s probably how “unselfish” helping behavior evolved in our primate ancestors. (See my earlier post on how altruism can evolve by means other than group selection.)

Lehrer’s piece, then, suffers from a concentration on style over substance, and, most important, a failure to either dig deeply enough into the issues surrounding group selection, or a lack of understanding of those issues.

It’s not as if making his article scientifically accurate would have made it boring: after all, Lehrer does mention the work showing no statistical association between haplodiploidy and eusociality. He just fails to cite an equally important paper showing a highly significant association between multiple mating and eusociality, which shows ineluctably that kin selection is important in the evolution of eusociality. And he doesn’t talk about the ways that selection other than group selection can promote the evolution of altruism. I know Lehrer knows these other models, because I told him about them when he inteviewed me for the piece (I’m quoted in the article).

The New Yorker apparently likes Lehrer because he can write well and engagingly. But good science journalism requires more than that: it requires a deep understanding of the issues at hand and a means of conveying the substance of a controversy accurately. It wouldn’t have been hard for Lehrer to do that. But I guess the New Yorker doesn’t have any editors who can vet the science.

Oh, and the magazine needs to improve its fact-checking.


Hughes, W. O. H., B. Oldroyd, M. Beekman and F. W. Ratnieks.  2008.  Ancestral monogamy shows that kin selection is the key to the evolution of eusociality.  Science 320:1213-1216.

Nowak, M. A., C. E. Tarnita and E. O. Wilson.  2010.  The evolution of eusociality.  Nature 466: 1057-1062.

95 thoughts on “New Yorker article on the evolution of altruism and the kin-selection flap

  1. “As far as I know (and I may have missed something), no paper has been published in support of Nowak et al”

    The paper has been cited 52 times according to Web of Knowledge. I haven’t actually checked if any of these citations are supporting ones, but some of them look like they might possible do. If I get time, might do a review.

  2. As someone with no expertise in this area, I don’t understand why this has to be an either-or proposition. Why couldn’t group selection operate alongside kin selection?

    1. Exactly! There is a misconception in the field that group selection can be explained in terms of inclusive fitness (kin selection) which is false. The two processes are fundamentally different. Group selection can accelarate evolution by kin selection. In my humble opinion, this is probably common. In some cases a trait that can’t evolve by kin selection alone can evolve if the individuals live in groups. In some cases a trait can evolve by group selection with no help from kin selection, and in others it can evolve by kin selection without groups. Nowak didn’t really address these points, but on the other hand, most of the negative reactions to that paper are based on misunderstandings.

      1. “There is a misconception in the field that group selection can be explained in terms of inclusive fitness (kin selection) which is false.”

        I thought that the issue was whether altruism and eusociality are better explained by either 1)evolution operating with genes as the real units of selection, i.e. processes such as kin selection, or 2) evolution operating at higher levels with whole groups of organisms as the real units of selection, or “group selection”.

        Furthermore, I thought that it was the opinion of most biologists that evolution does not proceed by group selection.

        Therefore, most biologists would not try to explain group selection by anything, because they don’t believe that it is an accurate description of how evolution operates.

        Maybe you meant to write “There is a misconception in the field that *altruism/social behavior* can be explained in terms of inclusive fitness”??

      2. Nope. For any population genetics model of evolution by natural selection in a population with a hierarchical group structure, the effects of selection can always be partitioned in an inclusive fitness or kin selection (KS) way (direct and indirect effects) but also in a group selection (GS) way (within and between group effects). Hence, they are mathematically equivalent. Sometimes it’s more “natural” to work with the KS formulation, sometimes with the GS formulation (although I’d argue with the KS way more often).

        A good recent review here:

        1. You’re assuming that the Price equation propoerly accounts for group level events, and it doesn’t. This is the big problem. GS=KS in special cases when the “fitness” of a group is the average fitness (or total fitness) of the individuals in the group. In models where groups break apart or disperse or suddenly die, etc., the KS=GS result is false, since the Price equation does not apply without changing the properties that allow the partitioning you were talking about.

          1. Here’s my favorite example where average individual fitness obviously isn’t a good measure of the group fitness. The individuals are cells of various types, and the group is a simple organism. One of the cell types is cancerous, and it grows very fast and eventually kills the organism it is in. The average fitness of the cells in an organism with a tumor is large (due to the tumor), but the organism clearly isn’t fit.

            Is this what you’re looking for?

            By the way, thanks for that review – I just printed it.

          2. Thanks for the example, but I wouldn’t say that the average fitness of the cells is large. If the organism dies before reproducing because of the tumor, then the average fitness of the cells is zero. The point being that we have to be careful in defining fitness.

            But I was looking more for a mathematical proof, a model which is a counter example, of your claim that the Price equation doesn’t properly account for group level events.

          3. You can’t “prove” things like that because one can always reformulate the Price equation to account for a new feature, so it’s a moving target. But just look at the standard Price equation like in Okasha’s book, or Gardner, “The Price equation”,Curr, Biol. 2008. You will see that the summations in the Price equation go from 1 to n where n is the number of groups. So n is fixed. But n is not fixed when there are group-level birhts and deaths as well as individual-level births and deaths. Check out for a model where the number of groups and the numbers of individuals in the groups are all dynamical variables.

            And your point about fitness is spot on. There is no good way to define group fitness in a dynamical model, so it’s best to just write down the dynamical equations without even trying to define fitness. But still, I think my example shows that the price equation can’t work for all models with two levels of organization.

          4. You can’t “prove” things like that because one can always reformulate the Price equation to account for a new feature, so it’s a moving target.

            Yes indeed. That’s why I didn’t immediately buy your claim that the Price equation (PE) can’t account for it. One can reformulate it in terms of the PE.

            You will see that the summations in the Price equation go from 1 to n where n is the number of groups. So n is fixed.

            Sure, but n could also be regarded as the number of suitable places for groups to live, some of which happen to be unoccupied. As you say, the PE can be reformulated to accommodate that. That’s why I said that any model can be framed in a KS and a GS partition of selection.

            Check out for a model where the number of groups and the numbers of individuals in the groups are all dynamical variables.

            Thanks for that. I guess you are the author of that paper. I haven’t read it yet, but the math looks impressive! I will try to understand what you’re saying there.

            Seeing as you acknowledge Doebeli as an adviser, I can see better where you are coming from. Doebeli, like Nowak, both from the physical sciences, has some strange ideas, IMO, about modeling evolution. Both seem quite ignorant about population genetics and how it differs from “evolutionary game theory” as they like to call what they’re doing. But perhaps this is getting off-topic and we should have a private conversation. You can figure out who I am and how to reach me from the info in this thread.

          5. Sorry, I’m not clever enough to identify you. Another clue? Or you can find me from that article I referenced. I’m looking forward to chatting.

          6. where the number of groups and the numbers of individuals in the groups are all dynamical variables.

            Thanks! I just figured out what the problem with the GS approach may be, everything else alike: it sounds like it is throwing degrees of freedom at the problem.

            Mathematically equivalent, empirically inferior.

          7. In models where groups break apart or disperse or suddenly die, etc., the KS=GS result is false,

            Non-biologist here: wouldn’t that mean the hypothesized group selection ceased to operate? (And in any case, wouldn’t such a failure mean GS is restricted where KS isn’t?)

          8. Don’t individuals reproduce and die? That’s how evolution works! There’s variation in the offspring and the “fitter” ones are more likely to have offspring themselves. It’s the same with groups, but in general group fitness does not follow from the fitness of the individuals it contains. When you try to tie the group-level birth and death processes into individual-level fitness measures (e.g., inclusive fitness) you find that things don’t work.

            Your other comment just above this one doesn’t make any sense! You need the variables that are necessary to define the problem. GS and KS are apples and oranges.

    2. Well then, what about meta-group selection? Meta-meta-group selection? Meta-meta-meta group selection?

      The point is, we should not multiply entities unnecessarily. If one mechanism will explain a phenomena, there is no a priori reason to believe any others are operating.

      1. Hrmmm, that came out a teensy bit stronger than I wanted it to. It may be more fair to say: if one mechanism will explain a phenomenon, there is no a priori reason to believe any others are important.

    3. Group selection is a nebulous concept.
      I thought the fact that group selection is too slow and susceptible to mutant invasion, has pretty much invalidated this theory. There must be an implicit assumption here that groups involved in a group selection scenario would be re-forming approximately every individual generation, and no cross-group mating. But then to reform that fast, and no cross-group mating, means that each group is pretty much a small group of close relatives. And you’re back to kin selection. Still doesn’t solve the problem of mutation of non-altruists from within (which would surely outcompete the self-sacrificing altruists). And if you posit that the altruists would kick out the mutants, then it’s back to recriprocal altruism.

      1. This is Jerry’s point. If groups are static (or almost static) so the individuals within them have time to reach an equilibrium before (say) the group splits into two groups, then group selection is impossible. But there is a large range where then group-level events happen often enough for interesting things to happen.

        Check out a similar dicussion on this site from a few weeks ago. I gave some references to some relevent mathematics. Things are not as simple as you think!

        1. Read half of the pdf you posted above in reply to IdoP. The assumptions look very limiting. For example, that all non-cooperator groups must go extinct. And altruism is not really equivalent to cooperator strategy. A group of altruists may not necessarily fare better than a group of non-altruists. Don’t know if the 2nd half addressed this, but the modelling also did not take into account the effect of kin selection within groups. What is the assumption for the relatedness of the members of the group? If random relatedness then the model has even more limited applicability. If related members usually belong in the same group (subgroups within groups, or expanding circles of relatedness in a group), such as hunter-gatherer tribes likely would be, then the model doesn’t calculate which is the real or stronger driver – the fact that the members belong in the group, or the fact that the members are relatives.

          1. Sorry for the late post – probably nobody’s looking any more. I think you are missing how the model works. The basic “data” of the model are rates that various events occur at – birth rates, death rates, fissioning rates, extinction rates, etc. – and those rates are all (arbitrary) functions of the current system state. The key word there is “arbitrary” – there are no restrictions like the ones you mentioned. Except one. The relatedness between individuals is not a variable in the model – but that’s one of the main points. The model shows that interesting things can happen (e.g. group selection) even if all the individuals behave as if relatedness is irrelevant. Including relatedness in the model might be possible (at the individual level) but it’s awkward. My take on that is that kin selection and multi-level selection are apples and oranges. Time to take this discussion off line. Feel free to contact me.

  3. Another feature of the article I find annoying is its characterization of the controversy as a battle with moral implications: if altruism results from kin selection, it’s nasty, but group selection is pure and wholesome. WTF?

  4. Good points all, although I think I can forgive Lehrer the “mistake” about Wilson’s “retirement”. He may be retired from teaching, but I think the call was more that he retire from publishing silly papers that contravene decades of research, (something he clearly has not yet done.)

  5. I find this to be of great interest! I think I have some limited understanding understanding of how eusociality is maintained (stable strategy???) but could not understand how it evolved and no one seemed to give me very good answers. I could not understand how group selection was an acceptable solution. Thanks for the article.

  6. I am interested in the Grand Old Man effect regarding E.O.Wilson. A great scientist & a great man I am sure, but sometimes does one get left behind on a sandbank while the river runs on?

  7. if kin selection is not a form of natural selection, i’d love to hear a more evidential, more logical argument for why any organism should care for its offspring to begin with.

    It’s strange that E.O. Wilson would so easily dismiss his previous work on kin selection without very good evidence to so.  I wonder if he still supports the paper after all the critiques published in Nature.

    1. when I said “care”, I of course also meant it figuratively. I don’t mean to anthropomorphize insects.

      1. No CJ, I think you meant it literally, and in a good way: “to care”, as in
        “to provide actual care”,
        as opposed to
        “having warm fuzzy feelings that don’t have effects in the external world”.

        Only humans do the latter, apparently.

        1. well yes, i was trying, perhaps too hard, to avoid a word that has two, but closely related meanings in people’s minds. Of course it’s possible to not care about what your taking care of.

  8. if kin selection is not a form of natural selection, i’d love to hear a more evidential, more logical argument for why any organism should care for its offspring to begin with.

    Not to reopen a can of worms, but the concept of kin selection, per se, is not necessary to explain care of one’s direct offspring. That’s ordinary old Darwinian individual-level natural selection with no frills.

    1. I’m interested in how you’d explain care of offspring without resorting using the concepts of inclusive fitness and kin selection.

    2. It can be thought of as kin selection, just as caring for one’s sibling’s offspring is. There are costs and benefits of parental care, the costs being forgoing future reproductive investment. That’s why there is parent-offspring conflict. People just don’t think of parental care as a form of kin selection, but it’s easy to see that it can be thought of that way.

      1. And, of course, you can also think of parental care as group selection… although that isn’t a comfortable idea for most evolutionary biologists.

    3. But that’s simply not true in a sexual species. Offspring are not identical to the individual. They are kin of the individual.

      Parental care is absolutely a case of kin selection. The only way it might differ from other forms of kin selection would center on certainty about relatedness (more so on the part of the mother).

  9. Jerry wrote:

    Many of the authors on the critiques of the Nowak et al. paper were mathematical biologists with a high degree of skill in theoretical analysis, and certainly with the ability to analyze Nowak et al.’s math (Stuart West is a notable example).

    No offense, but I wouldn’t call Stu West a “mathematical biologist”, and I doubt he would call himself that. First and foremost Stu is an excellent empirical and theoretical evolutionary biologist, one who often collaborates with more “mathy” types to produce models.

    PS: like you, I am one of the about a hundred and fifty. Proud to be your co-author. 🙂

      1. Linda, there is such a thing as a mathematical biologist, just as there is such a thing as a theoretical physicist, a number theorist mathematician, and a Bayesian statistician. These adjectives indicate specialist expertise in a sub-discipline and it is not uncommon for these specialists to understand something that their non-specialised colleagues often don’t (in fact, that’s the whole point of having specialist expertise).

  10. This is a bit tangential, but the social biologists are trying to move away from the term “reciprocal altruism”. Altruism is a reduction in direct benefits when considering lifetime reproductive success. In contrast, reciprocity leads to a higher direct benefits than pure selfishness (under the assumption others will retaliate). That reciprocity can outperform selfishness has led many to advocate for re-labeling the term “reciprocity”.

    1. Interesting, and from your concise description of the rationale, such a change would seem to make a lot of sense.

  11. Some wil pretend that group selection has already been observed…

    “Decreasing temporal trends in probabilistic maturation reaction norm (PMRN) midpoints, symptomatic of earlier maturation despite environmentally induced variation in growth, have been observed in many exploited fish stocks. Here, we studied the growth and maturation trends of female and male Icelandic cod (Gadus morhua) by estimating PMRN midpoints for cohorts 1964–1999 and found evidence that a shift towards maturation at smaller sizes and younger ages has occurred independently of changes in growth, condition, and temperature. Weighting the data with regional survey abundance estimates to account for spatial heterogeneity in maturity status and sampling intensity did not qualitatively affect the temporal trends. Length-at-age also decreased through the study period, which, through simulations, could be attributed to the energetic costs of earlier maturity at maturing age groups but not at younger ages. These findings support the hypothesis that such changes in maturation schedules are not caused by environmental factors alone but could also reflect a genetic change, potentially in response to intensive fishing.”

  12. …while other haploidiploid species aren’t eusocial.

    All Hymenoptera are haplodiploid, and the eusocial species are vastly outnumbered by those which aren’t.

    1. Outnumbered in number of species, sure. But also outnumbered in numbers of individuals? I suspect not, but please correct me if I’m wrong!

      1. I think he meant the former. It’s not absolutely clear, but that is the only charitable and sensible intepretation.

  13. It is interesting the case for kin selection often spans without distinction the gamut from aphids to apes, as though the same force operates equally within all species. “Autruism” is a word of human construct and connotes an element of human emotion. As Dr. Coyne points out, “How much of ‘human nature’ comes from evolution, and how much from culture? For many behaviors, like religion and criminality, we simply have no idea. Nor does it matter when it comes to solving most problems.”

    As for the eusociability of my 400,000 or so bees “altruism” means nothing. Likewise, “kin selection” means nothing. As Nowak et al. (2010) point out, “relatedness is better explained as the consequence rather than the cause of eusociality”, or as stated by Gadagkar (2001), “ecological, physiological, and demographic factors can be more important in promoting the evolution of eusociality than the genetic relatedness asymetries”.

    Despite immediate and resounding criticism I am attracted to Nowak’s assertion that “multilevel selection drives changes in the colony life cycle and social structures, often to elaborate extremes.” (No tomatoes, please.) Possibly I misread, but I really don’t see any *focus* on kin selection and group selection being mutually exclusive; rather, I was left with the impression that the authors believe there is too much emphasis on kin selection as explanatory of the evolution of eusociality, that this emphasis is intellectually limiting, and that employment of broader concepts of natural selection (K.I.S.S.) provides a more fertile framework for evaluation of multi-level and competing hypotheses.

    As to my bees? They won’t assist a drowning mate, be she kith or kin. If a bee from an alien tribe enters a hive, death is her fate. However, to introduce a new queen (resplendent with 16 pairs of new chromosomes … no kin intended) I leave her in a protective cage within the hive until she intoxicates all others with her compelling pheromone; then, as if by magic, she is no longer of a alien genes, she is “kin” (in spirit if not in biology).

    If my hive is overcrowded, the workers build several, larger brood cells and curry the queen for the favor of an egg into each … which, unbeknownst to her, seals her doom within the hive. The workers will feed these chosen few a royal jelly that will literally alter their ontogeny. When the first new queen emerges, she seeks out and kills with a vengeance, but with discrimination and only those queens yet un-emerged. As to the old queen? She flees the hive to avoid a similar fate. But must she do this alone? No . . . . she takes nearly one-half the hive workers with her, and with them as much honey as they can carry, to seek out a new home. As trusted members of the hive, they leave, taking a valuable and life sustaining commodity with them, without disagreeable repercussion, although their departure and the removal of honey works to the serious detriment of the remaining members. One week later, if any of the departing bees tries again to enter the old hive, a sure death awaits her.

    I could go on for hours. The mechanisms at play here are in fact multi-dimensional and varied. Not all appear to favor kin or the group. I for one won’t hang my eusociality hat upon kin selection alone, which is the primary point I take away from Nowak et al.

    1. Wonderful post!

      I could go on for hours. The mechanisms at play here are in fact multi-dimensional and varied. Not all appear to favor kin or the group. I for one won’t hang my eusociality hat upon kin selection alone, which is the primary point I take away from Nowak et al.

      Good for you, but other readers saw some additional and very explicit points, such as kin selection being a lousy explanatory framework across the board. I think that point has been soundly refuted by the published responses to Nowak et al.

    2. It’s probably safe to say the succesful introduction of an alien queen to a hive never happens in the wild. So there would be no selection for resistance to intoxication from a non-related queen.

      1. DV: Agreed as to the wild thing, but does it logically follow that “there would be no selection for resistance to intoxication from a non-related queen”? We apiarists play God all the time. For example, natural selection (via man’s “un-natural” hand) gives my bees accelerated resistance to V. destructor. So too, when I introduce a new queen to replace a dead or ailing monarch, I am affecting (if not effecting) natural selection. Without my hand, an entire colony of upwards of 40,000 dies. If the conlony is not amenable to a new phermone, it dies. When a new, native-born queen enters the realm, a new phermone structure is introduced. Thus, by either wild or man-induced means, natural selection would seem to favor the attraction, eh?

        As an aside, it occurs to me one might argue that allowing an old queen to flee with half the tribe and half the honey is an example of eusociality? If you agree, would selection in this case be of the kin or group kind? The debate is healthly; the prisms are varied; the devil is in the definitions.

        1. I mean that it would be expensive to develop a discriminating sense for pheromones. This would only happen if there was selection pressure, but since there is none in the wild, then you’re starting out with bees who already naturally don’t discriminate pheromones for relatedness. Natural selection evolved the attraction for sure, but not a discriminating attraction.

          This is like the imprinting in ducks. There’s no competition in the wild for which adult a chick would imprint on, so ducks are non-discriminating – they would imprint on something that’s not even a duck.

          1. Yes. It’s important to not get carried away with post hoc rationalizations that could just be epiphenomenal. Many things can happen, but what most likely did happen? And in what order?

          2. I think you meant to address that to J James. His original post was indeed very fascinating.

          3. You’re right!! Thanks for the correction! Your input has been enlightening as well.

        2. In my models I call this sort of thing “fissioning” – when a group breaks into pieces, each piece becoming a new group. So I would unquestionably call it a group-level event. My opinion (and apparently yours too) is that one can have group selection and kin selection occurring at the same time.

          If I could better understand everything you said about the population dynamics of your bees, I would see if I could model it with my dynamical equations!

  14. My colleague and friend Dr. Philip Ward, a professor of entomology at the University of California at Davis and a systematist specializing in ant evolution, has given permission for me to post his analysis of Lehrer’s piece, which he sent as an email. It’s below the line.

    Unfortunately, Lehrer does not present a convincing case for the Nowak et al. line of argument–perhaps because it can’t be done. As a result of his discussions with Wilson he seems to conflate the haplodiploidy hypothesis with inclusive fitness, and he apparently thinks that the lack of support for the former makes the latter no longer tenable. He mischaracterizes the natural history of nesting Hymenoptera to argue (incorrectly, in my view) that relatedness is a consequence rather than a cause of eusociality. This is almost certainly not the case: under one of the most plausible scenarios for the evolution of eusociality in insects —the so-called subsocial route—there is extended parental care and increased mother-offspring contact, leading to a situation where the mother is still alive when the offspring emerge as adults. This in turn sets thes tage for the *possibility* of helping behavior arising. The close relatedness of these clusters of offspring comes first, and it surely facilitates the evolution of altruism. Even in the more problematic scenario in which several same-generation females come together and show reproductive division of labor (the semisocial route), this will require between-group genetic variance, i.e., some degree of relatedness among the cooperating females.

    On one point I do agree with Wilson: eusociality is rare because it requires specific preadaptations. The Hymenoptera nicely exemplify this: there are numerous lineages of wasps, containing thousands of species, yet eusocial behavior is confined to a single, relatively recent clade, the aculeate (or stinging) Hymenoptera. Here the ovipositor (a female-limited character) has been modified as a stinging device to paralyze prey and defend against enemies. Nesting behavior and parental care are also widespread in this group, combined with considerable “manual dexterity”, i.e., proficiency in the use of mandibles and legs to build nests and care for young. So if extended parental care leads to overlap between parent and offspring generations, then these traits of aculeate Hymenoptera may predispose offspring to be especially *effective* as helpers. And females are more likely to be able to help than males because they are the ones furnished with nest-building and stinging capabilities. Of course we are still talking about family groups, and relatedness remains important. But the point stands: in some animals (already showing some degree of group living) the offspring may have properties that make them especially potent helpers. There is no sense in attempting to save two drowning siblings if you also cannot swim and the likely outcome is three dead persons.

    Finally, Lehrer does not explain in what way the mathematical model of Nowak and Tarnita is superior to inclusive fitness thinking. OK, so it is mathematically more complex, but is it more biologically realistic? Does it have heuristic value? Can it be used for empirical tests? Of course, I haven’t followed the math either, so this is where I defer to the wisdom of Queller, West and others.

        1. dear Cris: sensu stricto, the fact is that for the nowak et al.”math” result, if you dont understand the math you cant argue against or for, that result. and this is independent of the fact that i can follow the math. sorry if i am not clear enough; long night in these latitudes…. btw, i follow the math….bye for now

    1. Dr. Ward’s comments are thoughtful and provocative, (thank you) but I think he strays a bit here:

      “There is no sense in attempting to save two drowning siblings if you also cannot swim and the likely outcome is three dead persons.”

      I suggest this is not wasp speak and is not kin selection; rather, it is human speak and demonstrates a conscious analysis of risk (“I would lay down my life for two brothers or eight cousins”) which adds another dimension to the discussion. If there were sufficient pressures for wasps to swim, they eventually would, eh?

      As to the math? It is inseparable from the biology; it is what moves the inherently subjective nature of probability into the coherent realm of “scientific truth” accepted by the consensus of a scientific community (not that the consensus is incapable of error). The math is what saves us from the dreaded stain of “post hoc rationalizations”.

  15. I do not receive New Yorker and am commenting in the dark. First off, I must agree with John (#4) with regard to the unending stream of moral drivel journalists and ideologues spray onto the biological playing field.

    With respect to restrictions on the evolution of eusociality, in addition to the facilitating role of haplodiploidy, kin selection depends on an ecology that favors contact between close relatives and on tasks (such as digging tunnels and food sharing) done better through self-sacrifice. The less social ancestors of naked mole rats, termites (roaches that formed colonies in dead logs), and social hymenoptera probably lived in burrows or communal nests where many if not all social interactions necessarily occurred with close relatives born there.

    I would never expect kin selection to be able to generate altruism in pelagic schooling fishes, and the very absence of altruism in all the 100s of such fish species itself supports the kin selection model. On the other hand, fish schools may often be wiped out by predators, and if altruism were ever detected in such animals, group selection will start out as a strong candidate mechanism.

    Group selection is well supported in the evolution of some traits in some organisms. Pathogens, such as viral diseases, may lose virulence (in part) because of group selection. Strains that grow fast and kill their hosts quickly are less likely to be transmitted to new hosts than are strains that cause a lingering death. The spatial ecology of these pathogens has little in common with eusocial metazoans.

    In the 1970s David Sloan Wilson, cited by Jerry above, developed a new model for the evolution of altruism called “trait-group selection.” This mechanism involved neither group nor kin selection (see PNAS, Jan 1975). I do not know how this fits into the story.

  16. Lehrer is a young science writer about whom I have mixed feelings. […]

    … but he’s more interested in controversy than scientific truth.

    Seconded on the first, and the second may explain my own misgivings.

    1. I have had the same feeling about Jonah Lehrer ever since I read his first book, and am glad to have that feeling confirmed by someone as eminent as Jerry Coyne. My sense was that Lehrer doesn’t think rigorously – he doesn’t think like a scientist. And now I understand why: he thinks like a journalist. He’s the same kind of “science writer,” I think, as someone like, say, Malcolm Gladwell. He’s just not in the same class as someone like, say, Matt Ridley (the guy who wrote, e.g., “The Red Queen,” and “The Agile Gene.”) Yay, Matt Ridley!

  17. I used to love The New Yorker, but they lost me when they published a chapter of Darkness in El Dorado that contained numerous elementary biological errors in order to invent an accusation of genocide against a living anthropologist, and decided to publish the piece after these elementary errors were made public. So much for fact-checking.

    The New Yorker still publishes a lot of great pieces, but there’s no way I can forgice it for such a betrayal of science, journalistic ethics, and common human decency.

  18. Alright, here it is….group selection is as real as free will. It starts out as a shared, subtle and selectively neutral epiphenomena that causes a curious behavior among a small group and begins to replicate like a meme. Over time, as the population grows, the meme speeds up selection in the gene pool, selecting for genes that better serve the meme. It becomes pervasive and impossible to shake as the variation to allow for it was too little and too slow.

      1. Free will is not itself a causal agent. Free will is a sensation, in the same way that familiarity, for example, is a sensation.

        The sensation of free will is in effect simply an inference of the brain, (much in the same way as the sensation of familiarity, for example, is an inferenec that something has been seen before, or the sensation of certainty is an inference that a belief is in fact true).

        The sensation of free will is the brain’s inference that the behavior (being experienced as willed) was proximately caused by that brain’s activity — and specifically a kind of activity that is under the actor’s voluntary control. (For example, whistling is in the voluntary control category; hiccuping is not).

        It is evolutionarily adaptive for a brain to be able to distinguish whether an action is or is not proximately caused by its activity versus some other agent’/ and whether the action is the kind that the brain’s “voluntary” circuits control — in the same way that it is adaptive for a brain to be able to distinguish, say, whether an object is moving or stationary, blue or yellow, making a rustling sound or making a crashing sound — or any other sensory discrimination.

        See Daniel Wegner, The Illusion of Conscious Will.

  19. Sean Rice (from Texas Technical University) has an interesting perspective on how group selection might work in “Evolutionary Theory”. I think the math is what get’s people to believe in group selection, the idea makes sense, it just isn’t well supported.

      1. Also, evolution, especially its mechanisms, are essentially statistics. Evolutionary biologists should learn more math.

        1. As a scientifically-literate layman who is aspiring to do a further study in biology, may I ask what other sorts of math are important, apart from statistics?

          Thanks in advance for answers.

  20. Something I recently have wondered:
    A woman’s mitochondrial DNA is shared by her daughter’s children, male & female. However her son’s children do not. Are those children therefore more closely related to her, sharing more of her TOTAL DNA, and is there any evidence that women favour their daughters’ offspring rather than that of their sons?

    1. Yes, in a sense woman’s daughters’ children are “overall” more closely related to her than her sons’ children. However, selection would only favor mitochondrial genes to cause a grandmother to “spoil” her daughter’s kids, not the nuclear genes. I am not aware of any evidence for this.

      However, there are plenty of examples where genetic elements that are passed on exclusively via the mother harm or even kill the mother’s sons!

      Meet Wolbachia: the male-killing gender-bending gonad-chomping bacteria.

  21. This really feels like a “here we go again” moment. Elliot Sober and David Sloan Wilson published a defense of group selection in Brain and Behavioral Sciences several decades ago.

    The community went through the same thing then. The Sober and Wilson argument depended on a blurring together all “pro-social” behavior, whether or not it could be explained by mechanisms other than group selection.

    If I recall, their only data that actually addressed the problem involved animals that live in colonies that are spread out in some ideal way for group selection to play a role. They had a few examples of these, but the data were (I think it was sex ratios) marginal.

    I wouldn’t entirely rule out the possibility that there may be some exceedingly limited circumstances under which group selection might actually be real (though small). But such circumstances are going to be extremely rare and far from stable.

  22. Does group selection occur in humans? Probably. Selection occurs at the level of the gene (e.g. meiotic drive), the clade, and at levels in between. However, there’s no doubt that individual selection is much more important in humans than group selection. But this controversy is not really about something as esoteric as levels of selection.

    It’s about, to quote Lehrer:

    “Can true altruism even exist? Is generosity a sustainable trait? Or are living things inherently selfish, our kindness nothing but a mask?”

    The problem is that rhetoric like “true altruism”, “inherently selfish” and “nothing but a mask” are scientifically meaningless and in any case not resolvable by reference to selection at any level. How is altruism driven by group selection more or less “true” than altruism produced by kin selection?

  23. About the flak against the New Yorker’s fact-checking department:

    One can logically (and at times justifiably) call for the (full) retirement of someone who is demonstrably only partially retired.

    That notwithstanding, given that the “mistake” is not one of Lehrer’s – he is, after all, only relaying what others have written, in order to illustrate the tone of (some of) the denunciations – the word “mistake” is unduly unkind, especially when paired, as it is, with “unforgivable”.

Leave a Comment

Your email address will not be published. Required fields are marked *