Evolution for Everyone

We have reached the end of the T&R series. In a truth and reconciliation process, truth is required for reconciliation. There must be a consensus on what happened, even if all wrongs cannot be righted. I have had my say on what happened during the group selection controversy. Anyone who wishes to challenge my account is welcome to do so. This period in the history of evolutionary thought deserves the same kind of scholarship that is lavished upon Darwin and his contemporaries. The more scholars the merrier. Much of what I have reported in the T&R series is drawn from my book with Elliott Sober, Unto Others, which was published in 1998 and has largely withstood the test of time. I’d like to think that Samir Okasha, author of the highly respected Evolution and the Levels of Selection (2006), agrees with my account. If not, I hope he will speak up.

Once a consensus is reached on what happened, scientific inquiry can proceed in a more unified fashion than before. I end this series with a summary of what a fully reconciled field of sociobiology will look like. For a more detailed account, please consult my 2007 article co-authored with E.O. Wilson titled “Rethinking the Theoretical Foundation of Sociobiology”.

1) The original problem is the fundamental problem of social life. Darwin put his finger on the key problem with the evolution of social adaptations. Traits that are “for the good of the group” are not necessarily locally advantageous. If they evolve, it is not because individuals bearing the trait survive and reproduce better than individuals bearing alternative traits in their immediate vicinity. This deserves to be recognized as the core problem of sociobiology.

2) There is only one solution to the original problem. If a trait is locally disadvantageous wherever it occurs, there is only one way for it to evolve in the total population–by being advantageous at a larger scale. Groups of individuals displaying social adaptations must survive and reproduce better than other groups, to counterbalance the disadvantage of the same adaptations within groups. All evolutionary theories of social behavior embody this logic. The groups and fitness differentials within and among groups are there for anyone to see, once one knows what to look for.

3) Selection within and among groups are factual matters, not matters of perspective. Once the groups relevant to the evolution of a particular trait are identified (something that all models of social behavior must do), selection within and among groups can be straightforwardly measured–in a model, experiment, or nature. The question of whether a given trait evolves on the strength of within-group selection, between-group selection, or a combination or both has an answer upon which everyone can agree.

4) The categorical rejection of group selection in the 1960’s was an error, pure and simple. It is simply not the case that lower-level selection invariably trumps higher level selection. Period. End of discussion. Textbooks and other accounts that imply otherwise need be revised. The erroneous rejection of group selection was especially tragic because it made the only solution to the original problem appear wrong, motivating a search for other solutions that in retrospect turned out to be the only solution in disguise. Why would anyone want to perpetuate this confusion once it is seen clearly?

5) If it walks like a duck and quacks like a duck, maybe it’s a duck. Avoiding a stigmatized term makes sense as a short-sighted strategy for avoiding controversy, getting your article published, etc., but it is no way for a scientific discipline to conduct itself. If scientists aren’t going to keep careful track of what was said and meant during the history of a given subject, who will? Failing to mention group selection when discussing issues that have always been central to the group selection controversy is poor scholarship and should be grounds for rejection in a peer review journal.

6) Derived issues associated with the group selection controversy should not be confused with the original problem. From Darwin to Dawkins, group selection has been centered on the original problem, as I have shown in considerable detail. During the last few decades, however, a number of other issues have arisen. The first thing we need to know about these derived issues is that they do not bear upon the original problem. Secondly, we need to evaluate them on their own merits. A short list of derived issues includes the following.

7) My formula is better than yours. Given the same set of biological assumptions about the evolution of a given trait, there is more than one way to calculate what evolves in the total population. Some methods highlight local fitness differentials in addition to the global outcome. Other methods report only the global outcome, for example by averaging the fitness of individuals across groups or the fitness of genes across all contexts. Since all of the methods make the same biological assumptions, the differences between them are empirically empty, as Andy Gardner put it (see T&R XVIII). Nevertheless, one might be preferable to another based on other criteria, such as the compactness of the formula or certain insights that are highlighted by some formulations and obscured by others. For example, Hamilton’s original formulation of inclusive fitness theory obscured the fact that altruism is locally disadvantageous even in family groups, which jumps out of the Price equation. However, the Price equation can misclassify a nonsocial trait that evolves by pure individual-level selection (e.g., type A has a fitness of 1 and type B has a fitness of 0.75, no matter how they are grouped) as an example of group selection when the types are clustered into separate groups. The empirically empty preference of one formulation over another should never be confused with the empirically meaningful issues associated with the original problem.

8) Type one vs. type two group selection. Some traits, such as altruism, can be measured in individuals. Other traits, such as group size, can only be measured in groups. Samir Okasha refers to these as type 1 and type 2 traits, respectively. Whatever the merits of this distinction, it is important to realize that virtually all of the examples of group selection discussed throughout its history have been of the type 1 variety–traits that can be easily measured in individuals but require group selection to evolve because they are locally disadvantageous. A trait need not be a “group-level” trait (type 2) to evolve by group selection. Conversely, a “group-level” trait such as group size can evolve by pure within-group selection, as we saw in T&R XVIII for the endangered bird species discussed by Hanna Kokko.

9) Everything that evolves as a form of individual selfishness. Whenever altruism evolves by group selection, the average altruist is more fit than the average non-altruist in the total population, which can be conceptualized as a form of individual selfishness. The same gambit allows individual-level adaptations to be conceptualized as a form of gene selfishness. The problem with these expanded definitions of selfishness is that they don’t deny what they seem to deny. Gene selfishness is no argument against group selection and neither is the fitness of individuals averaged across groups. Theoretical biologist Alan Grafen is going to great lengths to build a formalism in which natural selection operating at all levels of the biological hierarchy (in multilevel selection terms) can be represented as a form of fitness maximization at the individual level. Strictly speaking, this enterprise is doomed to failure because natural selection operating below the level of the individual, resulting in such things as cancer and meiotic drive, can never be represented as for the good of the individual. Thus, Grafen must assume that these examples are trivial to proceed with his agenda. Even then, however, what is the point of trying to represent natural selection as a maximizing process at a single level of the biological hierarchy, much less the individual level?

10) Group selection vs. group adaptation. Generations of students have been told to avoid “for the good of the group” thinking because it requires group selection. Another way to say “for the good of the group” is “group-level adaptation”. Nevertheless, according to a recent article by Andy Gardner and Alan Grafen, a trait does not count as a group-level adaptation just because it evolves by group selection; it must evolve almost exclusively by group selection. It’s amazing how fast this argument has been taken up as the newest defense of individualism in evolutionary thought. Critics and proponents of group selection alike would have been mystified by it in the 1960’s. For them, the question was whether group selection ever happens. The idea of restricting the concept of group-level adaptation to cases where group selection only happens would never occur to them. It shouldn’t be necessary, but Elliott Sober and I are preparing to spell this out in yet another article as the academic arms race continues.

It’s worth asking why so many derived issues have arisen after the original problem was settled. In the spirit of discussing cultural influences for current science rather than waiting 50 or 100 years, (see T&R VI), I submit that when the rejection of group selection failed, those drawn to individualism felt the need to produce new arguments on its behalf. All of the derived issues buttress the concept of the individual as a privileged level of the biological hierarchy. Individualism is the primary issue at stake and when one argument fails, others are created to take its place. Once we let go of individualism, these arguments seem pointless and contrived.

Adaptations can evolve at all levels of the biological hierarchy, from genes to ecosystems, but only when certain conditions are met. Truth and reconciliation for group selection means regarding this statement in a positive sense and exploring its rich implications.



  1. #1 piker
    November 17, 2009

    From “Rethinking the Theoretical Foundation of Sociobiology”

    “Selfishness beats altruism within groups. Altruistic groups beat selfish groups. Everything else is commentary.”

    Except when the altruistic groups are beating the selfish groups within the context of an even larger group.

  2. #2 Don Monroe
    November 17, 2009

    Thanks for this series. I now feel much more confident that I can deal fairly with group-selection issues as I encounter them in my science journalism.

  3. #3 Bob O'H
    November 18, 2009

    For example, Hamilton’s original formulation of inclusive fitness theory obscured the fact that altruism is locally disadvantageous even in family groups

    Can you explain what you mean? It sounds like you’re saying Hamilton’s c was hidden away somewhere. I guess the problem is it’s not clear what “locally disadvantageous” means.

    Whenever altruism evolves by group selection, the average altruist is more fit than the average non-altruist in the total population, which can be conceptualized as a form of individual selfishness.

    Did you really mean to write that? You earlier stated the following:

    Isn’t he [Dawkins] ever going to get over the fact that selfish genes have no bearing whatsoever on the group selection controversy?

    and yet now you’re using the same selfishness argument that Dawkins did. You’re also being careless using the term “individual selfishness”: one can easily gets the impression you’re talking about individuals being selfish.

  4. #4 Guy
    November 18, 2009

    DSW wrote “If a trait is locally disadvantageous wherever it occurs, there is only one way for it to evolve in the total population–by being advantageous at a larger scale.”

    I think this is too narrowly described, and your own discussion of the meiotic drive issue reveals this when you wrote “natural selection operating below the level of the individual, resulting in such things as cancer and meiotic drive, can never be represented as for the good of the individual.”

    In a multilevel selection framework, selection at any level can entail local disadvantages at any other level. Such disadvantages can be top-down effects, as indicated in your first statement. They can also be bottom-up effects, as in the maladaptive effects on individuals that can result from meiotic drive.

  5. #5 Alan Kellogg
    November 21, 2009

    Now, is a super organism comprised solely of specimens of a single species, or can it consist of separate species? Such as, say, a community consisting of humans, cats, and dogs?

  6. #6 piker
    November 21, 2009

    Those would be super symbiotic organisms, with superfragilistic behavioral dynamics.

  7. #7 devadatta
    November 22, 2009

    DSW: “Individualism is the primary issue at stake and when one argument fails, others are created to take its place.”

    It doesn’t necessarily have to be a defense of individualism per se, rather a defense of the position the person currently has, as giving it up might be too costly. Presumably, a lot of persons have adapted to what the majority of people have believed in this area, making it itself an interesting case of cultural group adaptation.

  8. #8 InfuriatedSciTeacher
    November 23, 2009

    It’s normally applied to social insects that have few actual reproductive members within the hive itself; the hive or hill, in that instance, is what is considered to be the superorganism. Lewis Thomas discusses that without the technicalities in Lives of a Cell, if you’d like further example. The idea of a human superorganism then would mean just humans, although the contribution of other species in, say, a termite mound might be something that bears further examination. I don’t know much about entomology other than aquatic larvae, so I can’t really comment at length on that.

  9. #9 Guy
    November 24, 2009


    Sure. For example, you could not survive without other species (mostly prokaryotic) that are critical components of the ecosystem that define your functional system as an organism.

  10. #10 piker
    November 24, 2009

    For more on the altruistic prokaryotic, see: Origin of Group Identity, Subjugation of the Individual; Prokaryotic Group Living – Blooms, Slime and Mats
    Publisher Springer US

  11. #11 Heresiarch
    November 25, 2009

    Applying sociobiological thinking to human societies is a perilous undertaking (as you know). But the sociobiology of the New World Order must be exposed!!

  12. #12 BaldApe
    November 29, 2009

    Thanks for a fascinating series. I am eager to read more on this.

    Your point #10, particularly this:

    according to a recent article by Andy Gardner and Alan Grafen, a trait does not count as a group-level adaptation just because it evolves by group selection; it must evolve almost exclusively by group selection.

    reminds me of debates about sexual selection where those opposed to the idea that sexual selection might exist defined it in a way that made it logically impossible for it to exist. IOW, the evolution of a trait only counted as “true” sexual selection if it hurt the overall fitness of the organism. Making it easier to find a mate meant that the organism’s fitness was enhanced, so it couldn’t be sexual selection.

  13. #13 sikiş izle
    December 1, 2009

    We may sometimes use other businesses to perform certain services for us, such as maintaining the Site and our mailing lists, processing orders and delivering products, sending postal mail and providing marketing assistance and data analysis. We may provide Personal Information to those businesses when that information is necessary for them to complete a requested transaction or otherwise perform their duties

  14. #14 sikiş
    December 5, 2009

    I spent quite a long time reading all those releases, trying to figure out what exactly all that means. But of course, these were written by experienced PR and legal teams of the two institutions in maddening legalese I don’t understand. So nobody really knows the details. What I could gather

  15. #15 Carmi Turchick
    August 2, 2010

    I admire all of your work on group level selection, but for me it can be reduced to a very simple issue. The Theory of Evolution posits selection as a result of the inevitable, given that Malthus was correct, competition for resources required to survive and reproduce. In group territorial species this competition for resources takes place at the group level, as well as at the individual level. If competition for resources occurs at a specific level then selection must also occur at that level, or the entire Theory of Evolution is wrong and we must start over. While an individual bear may control a territory and the resources in it, I seriously doubt there are examples of individual humans doing the same. Therefore to talk about human evolution in a way consilient with the Theory of Evolution, we must accept that individual humans who had no group and were not subject to any group level selection were selected out, their fitness was zero. This means group level selection must be included, as no individuals were subject to only individual level selection.

    As for the altruism issue, I am curious as to why you did not mention the recent work on selfish being self-limiting, something I have been writing about for some time myself. If we consider a human group to be a superorganism then we can consider the selfish to be infecting parasites. There are then a variety of possible outcomes. Given that transmission for the human selfish parasite depends on the long-term survival of the host, we can expect the selfish to limit their virulence, which also limits their fitness adavantage over altruists. Further, a group of altruists is a fitness limiting resource for selfish and as such they will attempt to avoid sharing it with other selfish, resulting in fitness costs for selfish. This can break the infinite regression issue for punishment, and there is empirical data showing that selfish punish selfish as often as altruists do. Finally, as the altruists and selfish co-evolve selfish must pay the cost of avoiding detection by the altruists, and in many cases this will require acting as an altruist, actual cooperation, when no deception is possible. This further raises the relative fitness of altruism and limits the advantage of selfish.

    I would also like to make a point about altruism requiring group level selection. It is not clear to me that this is so. In this paper (Mesterton-Gibbons, Michael and Dugatkin, Lee Alan. (1992). “Cooperation among unrelated individuals: evolutionary factors.” The Quarterly Review of Biology. Vol. 67, 3, (Sep., 1992) pp.267-281.), the authors argue for “by-product mutualism” They state “There are three categories of cooperation among unrelated individuals: group selected behavior, reciprocal altruism, and by-product mutualism…. The mechanism for by-product mutualism is the common enemy of a sufficiently adverse environment. A prerequisite for this mechanism is the boomerang factor, that is, any uncertainty that increases the probability that a noncooperator will be the victim of its own cheating.” In an environment where any single individual was vulnerable to predation, humans literally needed to have others to watch their backs. Evolution is about odds, not certainties. If you and several others went to the aid, altruistically, of a fellow group member being attacked by a lion, for example, your odds of survival would be high and your net cost would be low. If others went to fight lions and you never did, your odds of survival would drop sharply. Hard to hide that kind of thing. If no one in the group ever went to the aid of others in the group, the fitness of each member of the group would be much lower. So being willing to risk one’s life for others paradoxically in fact lowers one’s risk of losing that life, regardless of group level competition.

New comments have been disabled.