Evolution for Everyone

Richard Dawkins did not invent naïve gene selectionism (see T&R X) but he spread it far and wide with the publication of The Selfish Gene. Let’s follow his logic, beginning on page 6 of the 1989 paperback edition:

This book will show how both individual selfishness and individual altruism are explained by the fundamental law that I am calling gene selfishness. But first I must deal with a particular erroneous explanation for altruism, because it is widely known, and even widely taught in schools. This explanation is based on the misconception that I have already mentioned, that living creatures evolve to do things ‘for the good of the species’ or ‘for the good of the group’.

There’s the caution against naïve group selectionism. Good for you, Richard! Now for the explanation of why it is naïve:

[A] group, such as a species or a population within a species, whose individual members are prepared to sacrifice themselves for the welfare of the group, may be less likely to go extinct than a rival group whose individual members place their own selfish interests first. Therefore the world becomes populated mainly by groups consisting of self-sacrificing individuals. This is the theory of ‘group selection’, long assumed to be true by biologists.

Right! This explains how “for the good of the group” traits might evolve. Now for the explanation of why they might not evolve.

The quick answer of the ‘individual selectionist’ to the argument just put might go something like this. Even in the group of altruists, there will almost certainly be a dissenting minority who refuse to make any sacrifice. If there is just one selfish rebel, prepared to exploit the altruism of the rest, then he, by definition, is more likely than they are to survive and have children. Each of these children will tend to inherit his selfish traits. After several generations of this natural selection, the ‘altruistic group’ will be over-run by selfish individuals, and will be indistinguishable from the selfish group. Even if we grant the improbable chance existence initially of pure altruistic groups without any rebels, it is very difficult to see what is to stop selfish individuals migrating in from neighboring selfish groups, and, by inter-marriage, contaminating the purity of the altruistic groups.

Done. Allow me to make three observations about these passages.

1) Dawkins’ portrayal of within- and between-group selection is utterly standard, and that’s a good thing. Throughout the T&R series, I have stressed the simplicity of the group selection controversy. From Darwin to Dawkins, it’s all about what I have called the original problem (see T&R II).

2) Dawkins implies that between-group selection (described in the second passage) is no match for within-group selection (described in the third passage), but he provides no proof. Models (such as the modified haystack model), experiments, and field studies are required to make this determination, not rhetorical flourishes.

3) The status of genes as replicators — what Dawkins calls the fundamental law of gene selfishness — is utterly beside the point. Genes are the replicators regardless of whether altruism wins or loses in Dawkins’ own scenario (also see T&R X). Using the replicator concept to argue against group selection is arguably the greatest case of comparing apples with oranges in the annals of evolutionary thought!

I do not believe the cynical adage “science progresses — funeral by funeral” but I worry that it might be true for Richard Dawkins on the subject of group selection. In my dreams, I imagine him reading my modified haystack model and saying “Well done, David! I have been wrong all these years. It turns out that a gene coding for altruism can plausibly evolve on the strength of between-group selection, even when it is selectively disadvantageous within groups. I do think it is important to keep in mind, however, that when altruism evolves by group selection, it is still an example of gene selfishness because the gene for altruism is more fit than the gene for selfishness, all things considered.”

Scientists would clap their hands red at such an act of nobility, but it hasn’t happened yet. Instead, Dawkins has behaved like a cowboy fighting off the Indians in an old western movie. When one gun runs out of ammo, he grabs another and another. Here are three guns that have run out of ammo.

The selfish gene gun. As we have seen, this gun didn’t have any bullets to begin with.

The vehicle gun. Recall that if individual organisms don’t qualify as replicators, they must quality as something else to be so manifestly well adapted. That “something” is variously called “interactors”, “targets”, or “vehicles” (Dawkins’ term) of selection. In one of Dawkins’ famous metaphors, genes in individuals are like rowers in a rowing crew. Since they are literally “all in the same boat”, they must “pull together” to win the race. As soon as Dawkins threw away the selfish gene gun, he started to claim that groups fail as vehicles because individuals in groups are not completely bound together in a common fate in the same way as genes in individuals. But this was never a requirement for group selection to occur! The groups in the haystack model and in Dawkins’ own portrayal of group selection quoted above aren’t like individuals in this respect, but they suffice for altruism to evolve despite its selective disadvantage within groups.

The extended phenotype gun. When the vehicle gun didn’t work, Dawkins decided to bury the vehicle concept altogether by describing genes as having extended phenotypes. Two examples of extended phenotypes are a bird’s nest and a beaver’s dam. The first is an individual-level adaptation in conventional terms; birds that build better nests raise more offspring than birds in the same group that build worse nests. The second is a group-level adaptation in conventional terms; beavers that build better dams are providing a public good for all of the beavers in the pond at their own expense. The fact that the genes result in alterations of the physical environment in both cases is irrelevant. In short, the concept of extended phenotypes doesn’t address the original problem and certainly doesn’t provide a novel solution.

With the Indians closing in, Dawkins has now started to throw chairs, bite, and kick. Consider his response to my recent article with Edward O. Wilson in American Scientist magazine titled “Evolution ‘For the Good of the Group’”

Genes Still Central: David Sloan Wilson’s lifelong quest to redefine “group selection” in such a way as to sow maximum confusion–and even to confuse the normally wise and sensible Edward O. Wilson into joining him–is of no more scientific interest than semantic double talk ever is. What goes beyond semantics, however, is his statement (it is safe to assume that E.O. Wilson is blameless) that “Both Williams and Dawkins eventually acknowledged their error [that the replicator concept provides an argument against group selection]…I cannot speak for George Williams but, as far as I am concerned, the statement is false: not a semantic confusion; not an exaggeration of a half-truth; not a distortion of a quarter truth; but a total, unmitigated, barefaced lie. Like many scientists, I am delighted to acknowledge occasions when I have changed my mind, but this is not one of them. D.S. Wilson should apologize. E.O. Wilson, being the gentleman that he is, probably will.

Gracious! What a hierarchical guy! Dawkins acts as if he is the No. 2 monkey, kowtowing to the No. 1 monkey (Ed) while dishing it out to the No. 3 monkey (me)! As Ed commented to me after reading Dawkins’ comment, “What does he think–that you slipped me a Mickey?”

If you still have the patience, let me make a few observations about this gem of a tantrum. First, why on earth would Dawkins title his response Genes Still Central ? Isn’t he ever going to get over the fact that selfish genes have no bearing whatsoever on the group selection controversy?

Second, I trust that I have provided ample evidence that the original problem has provided the basis for defining group selection for everyone, including Dawkins and myself. There has been no redefining.

Third, when it comes to semantic confusion, you can’t beat selfish gene theory. Genes are “the fundamental unit of selection” but this has no bearing on the “levels of selection” controversy. A gene might be selfish because it is selectively advantageous within groups, or it might be selfish because it evolves in the total population all things considered. Individuals might be perfected “vehicles” of selection now, but we also need to use the term “vehicle” to explain how such perfection evolved. This is how the simplicity of the original problem has turned into a terminological quagmire.

Dawkins continued his tantrum on his website after Ed and I quoted the passage in The Extended Phenotype where he abandons the replicator argument and takes up the vehicle argument:

The Wilson quotation from The Extended Phenotype is a ludicrous attempt to justify their lying statement that I “eventually” acknowledged an earlier error. For one thing, The Extended Phenotype was published way back in 1982, which makes nonsense of Wilson’s “eventually”. But more important, the point I was making in 1982 (and would make again now) was a general one about the important distinction between replicators and vehicles…I was explaining that those models of group selection that had been proposed were vehicle models not replicator models. I was not for a moment suggesting that I accepted those models as valid. They were (and are) invalid vehicle models, as opposed to invalid replicator models.

There you have it from Dawkins himself. The word “eventually” is appropriate for the six-year period between 1976 and 1982, regardless of how much time has elapsed since then. Poor Richard is still trying to fend off the Indians with the butt end of his replicator gun (“genes still central”) and vehicle gun (“invalid vehicle models”). If only Bill Hamilton was still alive to fight alongside and defend his brilliant ideas!

Fortunately, Bill Hamilton wrote plenty while he was still alive, showing that Richard Dawkins fights on alone.

To be continued.

Comments

  1. #1 bob koepp
    November 2, 2009

    Dawkins does, indeed, confuse the “unit of selection” and “level of selection” issues, despite their having been teased apart long ago. I got a chuckle from the “three monkeys” analogy.

  2. #2 piker
    November 2, 2009

    The inferences seem to be that because Dawkins has turned out to be wrong, your theories, which are in your view an improvement on his, must necessarily have turned out to be right – and you have been shown the better philosopher to boot.. But not all that much the better if you haven’t recognized that rather than right, you may only be less wrong.

    Altruism and selfishness are not in themselves causative. Nor are they in themselves strategic. They are the effects of a mix of strategies that have evolved from the basic yin and yang of existence: cooperation and/or competition.

    It’s in this mix of strategies where the evolution takes place, not in what the strategists have cooked up or in who gets to eat it. All you and Dawkins can seem to see and do is stick some place names here and there on the table and argue about the propriety of the settings.

  3. #3 bob koepp
    November 2, 2009

    piker – WTF? Monism and pluralism aren’t just place names.

  4. #4 piker
    November 2, 2009

    Oh, did either of these guys place monism and pluralism on the table? Must have missed that evolutionary spin on the art of the casserole.

  5. #5 Colugo
    November 2, 2009

    1) How are you any less gene-centric than Dawkins? It’s just an argument about level of selection of vehicles (or whatever you want to call them – how about ‘modules’?).

    2) While this is all very interesting, and surely more empirically confirmable than angels dancing on the heads of pins, the social, political, and economic implications of the group selection AKA ‘who gets to feel vindicated’ debate is zero, nada, null.

  6. #6 InfuriatedSciTeacher
    November 2, 2009

    the social, political, and economic implications of the group selection AKA ‘who gets to feel vindicated’ debate is zero, nada, null.

    Actually, the social/political implications of the vindication of group selection theory is that we now have another mechanism to explain the success and failure of various cultural constructs, up to and including the nation-state.
    Agreed, there is no instant economic impact, but then again, the purpose of scientific research isn’t solely to result in economic gain. Quite a large number of useful technologies have stemmed from accidental discovery in another investigation.

  7. #7 abb3w
    November 3, 2009

    InfuriatedSciTeacher: Actually, the social/political implications of the vindication of group selection theory is that we now have another mechanism to explain the success and failure of various cultural constructs, up to and including the nation-state.

    …with the caveat that not all said constructs are fully specified by the genetic information? (And thus, that such success/failure may move out of the scope of what is considered questions of biology, even if there is a variation/replication/selection mechanism of a more general evolutionary nature.)

  8. #8 InfuriatedSciTeacher
    November 3, 2009

    So group interaction mechanisms that aren’t inherited solely through genetics aren’t in the realm of biology? Interesting…

  9. #9 frog
    November 3, 2009

    InfuriatedSciTeacher:

    What, do you have some new evidence that evolution precedes on the decade to millennium rate among human beings? Maybe you have a Nobel coming. If you’re talking about non-genetic systems — well, that’s not really relevant at all here. That’s at bottom cognitive science moving towards psychology and is really mostly sociology/anthropology.

    The “social and political implications” in this are the same as in most science and mathematics — which is you never know where knowledge will lead. Who thought the obscure mathematics of Russel –> Godel –> Turing… would lead to internetz porn?

    Now, to the post. The problem with Dawkins is that I see very little math coming out of him — there’s a lot of talk-talk, but very few equations. Anyone can see that there are some conditions under which group-selection will be the dominant determinant or constraint of evolution.

    The question is, what are those conditions? And the only way to generally determine that is mathematically. We need equations, and then we can look at the data and say maybe that the conditions never arise, the conditions rarely arise, the conditions are common…

    But theory without equations is just philosophical jibber-jabber. The “replicator” (or any word describing a scientific object) is only meaningful as a mapping between terms in an equation and empirical data.

  10. #10 abb3w
    November 3, 2009

    InfuriatedSciTeacher: So group interaction mechanisms that aren’t inherited solely through genetics aren’t in the realm of biology? Interesting…

    As “biology” is usually anthropologically delineated? Yes, some are considered outside the scope of study, just as physicists usually leave categorizing the variety of carbon compounds that crawl to someone down the hall. While teratogenic, epigenetic and other biochemical infomation transmission modes are still considered “biology”, information transmitted by (for example) a book is usually not considered “biology”.

    frog: The “replicator” (or any word describing a scientific object) is only meaningful as a mapping between terms in an equation and empirical data.

    …or (to be a hair more general) as a nonterminal symbol in a formal grammar with terminal symbols that are empirical data.

    However, I would agree that Dawkins’ lack of mathematics in his argument is a problem; it makes him seem like an English major— or worse, a creationist saying “but how could you possibly…”.

  11. #11 InfuriatedSciTeacher
    November 3, 2009

    @ frog> did you bother to read in context, or just take the snippet and pile on? Perhaps you should try that first… even if it makes distorting my point a little more difficult. Since my original post was in response to #5 and the “what do we get out of this” crowd, you then go on to make my point for me.
    Even so, care to explain how psych, anthropology, and sociology don’t involve some bio? Or are you going to staunchly defend that the social sciences need not use natural science methods? What you term genetic systems aren’t fully so, in most cases (see abb3w on other biochemical means above, and consider the play of environmental factors in gene expression), nor are any of the fields you describe uninfluenced by the genetics of the people involved. The fact that we call the study of group dynamics, cognition, and behaviour something different when referring to humans doesn’t make it any less rooted in the biological.

  12. #12 InfuriatedSciTeacher
    November 3, 2009

    abb3w> semi-agreed; the sharing of information in books isn’t considered a biological process… however the development of the knowledge in that book and the horizontal transmission of behaviour and information can still be examined using a (modified) evolutionary model derived from biology.
    We pretty much agree from the outset, I was making a point about what is considered biology and where biological implications are relevant. Your caveat is accurate, I dislike where people tend to take the “thus” you added to it. Argument from consequences, and not valid as a result, I know.

  13. #13 David Sloan Wilson
    November 3, 2009

    I’m multi-tasking and only skimmed the last few comments, so excuse me if I missed some points, but human cultural evolution is a multilevel process that can take place without any genetic change. Genetic evolution was required to create the architecture for cultural evolution, but thereafter the mechanisms of inheritance are cultural, not genetic. More generally, what game theorists call the replicator dynamic refers to any process whereby the most successful phenotypes increase in frequency in the population. All of them count as evolutionary processes, regardless of the mechanistic substrate. See Richerson and Boyd’s Not By Genes Alone, which has been cited quite a bit in previous comments.

  14. #14 piker
    November 3, 2009

    Dr. Wilson,
    Wow, now you’re redefining what you, yourself, have called “natural selection” to include human cultural evolution. But then as soon as you do that, you’ve gone so far as to exclude genetic evolution as a mechanism of inheritance for other than what I presume is the genotypical architecture. From then on, or “thereafter,” the evolution at succeeding group levels is supposedly restricted to the natural choosing of the most successful phenotypes.

    What has happened to the conditional strategies you had earlier equated with the genetic variety? Are the strategies still exclusive to the phenotypes as a factor in their degrees of success? Or if not, will you now revise your models accordingly?, Because a certain dichotomy of exclusivity still seems to exist in the allocation of traits to their corresponding phenotypes.

  15. #15 meatyphil
    November 3, 2009

    Hmm… Seems like there’s less of a controversy here than Dr. Wilson would like for there to be. Between-group selection surely happens. I don’t think Dawkins would dispute that. I certainly wouldn’t. It seems like Dr. Wilson is just nit-picking. His “selfish gene” doesn’t explicitly acknowledge this fact, but it certainly doesn’t exclude it. I will agree that the “selfish gene” causes problems for the scientifically illiterate, but I don’t think anybody anywhere other than Dr. Wilson is manufacturing so much controversy over the term.

  16. #16 Bob O'H
    November 4, 2009

    bob koepp @1:

    Dawkins does, indeed, confuse the “unit of selection” and “level of selection” issues, despite their having been teased apart long ago.

    He did in Selfish Gene: I read somewhere (Defenders of the Truth?) a comment from him saying that he realised this, and it’s why he wrote The Extended Genotype.

    I don’t think DSW understand the approach that the “other side” takes. I don’t just mean Dawkins, but theoretical evolutionary biologists who actually work on this stuff. If he did, he wouldn’t write nonsense like this:

    First, why on earth would Dawkins title his response Genes Still Central ? Isn’t he ever going to get over the fact that selfish genes have no bearing whatsoever on the group selection controversy?

    If one is takes the approach developed by Price, Hamilton etc. then one can analyse multilevel selection by calculating the inclusive fitness of a gene: not just in the individual it is in, but also in others (i.e. one sums over the different vehicles). This is explicitly a selfish gene approach, which leads to the Price equation that Price and Hamilton saw could be used to model group selection. So, for anyone steeped in this approach to evolutionary theory, selfish genes have everything to do with group selection.

    Ironically, I think we need a bit more pluralism here.

  17. #17 David Sloan Wilson
    November 4, 2009

    In response to the last few comments, I am very familiar with the work of folks such as Stuart West, Andy Gardner, Ashleigh Griffin, Alan Grafen, Kevin Foster, Samir Okasha, and so on. All of them agree that when group selection is defined in terms of selection within and among groups (as roughly approximated by the Price equation), then group selection happens all the time. For them, the debate has moved on to another set of issues discussed under terms such as pluralism, equivalence, and whether an example of group selection counts as a group adaptation (Andy Gardner’s most recent paper with Alan Grafen). Here are two quick observations:

    1) The issues that are currently debated among the cognoscenti need to be carefully distinguished from the original problem and its solution, which has been settled among the cognoscenti. As far as the original problem is concerned “we are all group selectionists,” as Andy Gardner once put it.

    2) This is not yet common knowledge among the larger evolutionary community, not to speak of the general public. I’ll wager that 90% of graduate students interpret the current jousting among the experts as proof that the original rejection of group selection is still widely endorsed. Read the textbooks, articles, and books and judge for yourself. I’ll provide proof of my own in T&R XIII.

    3) If the only thing wrong with my T&R series is that everyone already agrees with it, I’m in pretty good shape :)

  18. #18 piker
    November 4, 2009

    “3) If the only thing wrong with my T&R series is that everyone already agrees with it, I’m in pretty good shape :)”

    Not if everyone that endorses your version of group selection is as wrong as you appear to be about the details.

    And certainly numerous posters here, including some you have recognized as colleagues, have gone to great lengths to be specific about the nature of your errors.

  19. #19 abb3w
    November 5, 2009

    InfuriatedSciTeacher: however the development of the knowledge in that book and the horizontal transmission of behaviour and information can still be examined using a (modified) evolutionary model derived from biology

    The model is anthropologically derived from the study of biology, but philosophically the model is merely a particular bit of mathematics. So, while there may be what is mathematically termed group selection in other arenas, that doesn’t make the other arenas “biology”. It just makes biology a useful source of patterns to try looking for in those other arenas.

    InfuriatedSciTeacher: Your caveat is accurate, I dislike where people tend to take the “thus” you added to it.

    I think it’s more that you object to people reasoning that since it’s not the exact same dynamic of variation/replication/selection, that anything learned from biology about such dynamics will be irrelevant elsewhere. That is obviously an error.

    David Sloan Wilson: More generally, what game theorists call the replicator dynamic refers to any process whereby the most successful phenotypes increase in frequency in the population.

    …for which biology serves as example; but not all examples ought be considered “biology”.

    piker: But then as soon as you do that, you’ve gone so far as to exclude genetic evolution as a mechanism of inheritance for other than what I presume is the genotypical architecture. From then on, or “thereafter,” the evolution at succeeding group levels is supposedly restricted to the natural choosing of the most successful phenotypes.

    To argue by analogy to fluid dynamics, the different replicative mechanisms flow with different degrees of inertia and density, but the trait levels of both continue drifting; and sufficiently powerful selective pressures at higher levels may transmit to lower levels.

    (It’s been a decade since I got my “D-” in fluid mech, but I suspect there would be similar partial differential equations involved.)

  20. #20 piker
    November 5, 2009

    abb3w,
    That fluid dynamics analogy is like one I read where where a biologist was comparing natural selection as a process to that of water running downhill, explaining that “selective environments make choices in precisely the same sense that rapidly moving water makes choices among bits of gravel of different sizes and weights, sorting the gravel bar below a rapids into a roughly ordered sequence according to size and weight.”

    Except that water does not choose in a biological sense – unless you can argue that it chooses to unfailingly obey the forces of nature. Neither does water have a variety of traits which have in common a choice making function. Such function necessarily contributing to their evolution, whether it be cultural or genetic.

  21. #21 abb3w
    November 14, 2009

    piker: Except that water does not choose in a biological sense – unless you can argue that it chooses to unfailingly obey the forces of nature.

    Indeed; and this is an excellent parallel to evolution. At the lowest levels, it’s just an application to the equations of the Second Law of Thermodynamics for pairs of systems connected by mass-energy (and thus, entropy) flow.

    The analogy is less excellent at representing “variety of traits”. To extend the analogy that far, one needs to recognize that the variety of traits correspond to positions in the n-dimensional space for the (nonuniform probability) random walk. Since we usually deal with water flows in only three space dimensions, physically modeling this requires a vast oversimplification; however, the 3D partial differential equations can be expanded to handle N dimensions easily, even if the physical world can’t.

    Like many analogies, it only attempts to try to reduce properties of some mathematics to more familiar context. Don’t mistake limits of the analogy’s model for limits of the underlying mathematical model.

  22. #22 piker
    November 14, 2009

    abb3w,
    Don’t mistake mathematics or its models as determinant of biological stratagems.
    Mathematics is a wonderful and virtually indispensable tool for helping us to think and communicate – but it can’t think for us or communicate without us. It doesn’t draw its own inferences. Analogy can make strategic comparisons that mathematical models cannot.

    You won’t agree I’m sure but then you have your categorical assurances and I have mine.

  23. #23 daedalus2u
    November 15, 2009

    piker, analogies are only crude non-mathematical and non-quantitative models with only some crude qualitative similarities. In effect they are like an architectural model of a building compared to the blueprints and design calculations (which are more like a mathematical model of the building).

    Analogies are useful models for some things, particularly when the mathematics known to be important is intractable. We shouldn’t try and use them where we cannot show that they are useful.

  24. #24 piker
    November 15, 2009

    Blah blah blah. Analogies serve purposes clearly beyond the abilities of those who can’t conceive of the nature or benefits of the purposive to begin with.

  25. #25 piker
    November 15, 2009
  26. #26 abb3w
    November 18, 2009

    piker: Analogy can make strategic comparisons that mathematical models cannot.

    One of the precursor points of Godel’s work was how whatever can be semantically expressed, can be mathematically expressed.

    For example, our current remarks are both being expressed in English, via words written as letters, which are expressed as numbers, and manipulated by computers operating on mathematical principles (as with Church-Turing automata). Language is simply a form of mathematics in disguise.

    (Expressed and compared efficiently, of course, is entirely another matter.)

  27. #27 piker
    November 18, 2009

    Mathematical symbols are an analogical tool but their meaning cannot be defined by the mere fact of their presence in the formulae.

    Other than that, I don’t choose to debate an issue that is really not a matter for debate.

  28. #28 daedalus2u
    November 19, 2009

    abb3w, excellent comment; all analogies can be put into the form of a mathematical model, they are just weak mathematical models. An attempt to map the properties of one object onto another object with a 1 to 1 mapping.

  29. #29 AI
    September 26, 2011

    Concerning Dawkins simplistic argument against group selection. Yes, altruism as he describes it may not be stable by itself, but it can be easily protected by other traits, for example if the group rewards the kin of those who sacrificed themselves with higher status, making them more desirable partners, the combo will be stable, and groups with both adaptations will outcompete groups of selfish individuals.