Tuesday, May 02, 2017

The selfish gene vs the lucky allele

The Selfish Gene was published forty-one years ago (1976) and last year there was a bit of a celebration. I think we can all appreciate the impact that the book had at the time but I'm not sure it's as profound and lasting as most people believe ["The Selfish Gene" turns 40] [The "selfish gene" is not a good metaphor to describe evolution] [Die, selfish gene, die!].

The main criticisms fall into two categories: (1) the primary unit of selection is the individual organism, not the gene, and (2) the book placed too much emphasis on adaptation (Darwinism). I think modern evolutionary theory is based on 21st century population genetics and that view puts a great deal of emphasis on the power of random genetic drift. The evolution of a population involves the survival of individuals within the population and that, in turn, depends on the variation that exists in the population. Thus, evolution is characterized by changes in the frequencies of alleles in a population.

Some alleles confer a beneficial effect on the individuals that carry them but whether those alleles become fixed will depend on a number of features (i.e. population size). Most of them will be lost before they ever become fixed unless the beneficial effect (selection coefficient) is very high. Thus, even beneficial alleles are often more "lucky" than "selfish." In addition, a large number of alleles are effectively neutral and changes in their frequencies are more due to luck than anything else.

No metaphor will accurately describe all of evolution but I would argue that in communicating to the general public (and other scientists) the term "lucky allele" is just as appropriate as "selfish gene." I would also argue that Gould's hierarchical view of evolution as a process occurring at a number of levels (gene, organism, population, species) is better than Dawkins' view where all the emphasis is evolution at the gene level.

It's interesting to see how Richard Dawkins has responded to all the criticism over the years. He published a new edition in 1989 where he stated (p. x) that, "In the dozen years since The Selfish Gene was published its central message has become textbook orthodoxy."

I don't think that's correct. I have pointed out on many occasions that the book is rarely mentioned in modern textbooks and the central messages—evolution at the gene level and adaptationism—are not really textbook orthodoxy.

In the 1989 edition, Dawkins makes sure that readers understand his view of evolution.
The selfish gene theory is Darwin's theory, expressed in a way that Darwin did not choose but whose aptness, I should like to think, he would instantly have recognized and delighted in. It is in fact a logical outgrowth of orthodox neo-Darwinism, but expressed as a novel image. Rather than focus on the individual organism, it takes a gene's-eye view of nature. It is a different way of seeing, not a different theory.
I don't think this way of seeing (gene's-eye view) has become textbook orthodoxy. I think the main emphasis in evolutionary biology textbooks is on the individual as the unit of selection. I also don't think that Darwin's Theory of Natural Selection is seen today as the only way to explain evolution—yet that's the view you get from reading The Selfish Gene. I think Dawkins is correct to say that his book is based on orthodox neo-Darwinism but by the 1980s evolutionary theory had moved beyond that view (Gould and Lewontin, 1979).

Richard Dawkins has just published an article in The Skeptical Inquirer outlining his current view of evolution and how it relates to The Selfish Gene [The Selfish Gene Revisited]. Let's look at it to see if his views have changed over the years. He begins with a defense of the word "gene."
In some ways I would quite like to find ways to recant the central message of The Selfish Gene. So many exciting things are fast happening in the world of genomics, it would seem almost inevitable—even tantalizing—that a book with the word "gene" in the title would, forty years on, need drastic revision if not outright discarding. This might indeed be so, were it not that "gene" in this book is used in a special sense, tailored to evolution rather than embryology. My definition is the population geneticists’ definition adopted by George C. Williams, one of the acknowledged heroes of the book, now lost to us along with John Maynard Smith and Bill Hamilton: "A gene is defined as any portion of chromosomal material that potentially lasts for enough generations to serve as a unit of natural selection." I pushed it to a somewhat facetious conclusion: "To be strict, this book should be called . . . The slightly selfish big bit of chromosome and the even more selfish little bit of chromosome." As opposed to the embryologist’s concern with how genes affect phenotypes, we have here the neo-Darwinist’s concern with changes in frequencies of entities in populations. Those entities are genes in the Williams sense (Williams later called that sense the “codex”). Genes can be counted and their frequency is the measure of their success.
This is complicated. It's perfectly understandable that Dawkins is using the term "selfish gene" as a metaphor but what I don't understand is why today, forty-one years later, he still avoids using the word "allele." And I can understand the word "selfish" as a metaphor to emphasize the power of natural selection but why does Dawkins continue to emphasize the connection between "gene" and "the unit of natural selection" when he knows full well that many "genes" (alleles) survive by luck and not by selfishness? Surely he knows of dozens of "genes" (alleles) that are invisible to natural selection?
A gene achieves its numerical success in the population by virtue of its (phenotypic) effects on individual bodies. A successful gene is represented in many bodies over a long period of time. It helps those bodies to survive long enough to reproduce in the environment. But the environment means not just the external environment of the body—trees, water, predators, etc.—but also the internal environment, and especially the other genes with which the selfish gene shares a succession of bodies through the population and down the generations. It follows that natural selection favours genes that flourish in the company of other genes in the breeding population. Genes are indeed "selfish" in the sense promoted in this book.
Nobody questions the idea that the "selfish" description applies metaphorically to some "genes" (alleles). But it would be nice to see Richard acknowledge two ideas that really are part of textbook orthodoxy: (1) many alleles have no observable phenotype, and (2) many alleles survive (become fixed) by luck and not by selection. I'm puzzled by the fact he didn't take advantage of forty-one years of criticism to acknowledge that the selfish gene metaphor is not a universal description of evolutionary theory.

The next part of his article is a defense of inclusive fitness. Dawkins is an ethologist and he makes it very clear in the original 1976 edition that the book is about animal behavior (p. vi).
I am an ethologist, and this is a book about animal behavior. My debt to the ethological tradition in which I was trained will be obvious.
Fair enough. The Selfish Gene is a book about evolution as it applies to animal behavior so it's not a surprise that Hamilton's views on inclusive fitness would play a prominent role. The problem it is that a generation of readers have assumed that the book is about evolution in general and not a particular aspect of it that happens to interest Richard Dawkins. Dawkins concedes that his view of evolutionary theory only applies to sexually reproducing organisms, thus ruling out the majority of organisms on the planet, but he doesn't really respond to criticisms of his ethological perspective and how inclusive fitness applies to most changes in allele frequencies in a population. All he does is admit this bias, "A central point of the book is the one developed by my friend the great Bill Hamilton, whose death I still mourn. Animals are expected to look after not only their own children but other genetic relatives."

The last part of his recent essay is about coalescence and the dating of allele divergences. In this section he mentions alleles once but then uses the word "genes" to describe two alleles. I don't know why he does this. He explains how you can date the origin of different alleles using coalescence theory and how it can lead to estimates of population size in the past. Then he says,
To summarize, the gene’s eye view of life, the central theme of this book, illuminates not just the evolution of altruism and selfishness, as expounded in previous editions. It also illuminates the deep past, in ways of which I had no inkling when I first wrote 
The Selfish Gene and which are expounded more fully in relevant passages (largely written by Yan, my co-author) of the second edition (2016) of The Ancestor’s Tale. So powerful is the gene’s eye view, the genome of a single individual is sufficient to make quantitatively detailed inferences about historical demography.
I don't understand coalescent theory even though I have access to textbooks by Dan Graur and Joe Felsenstein. However, my readings suggest that it owes very little to the concept of a selfish gene and to orthodox Darwinism. It seems to me that you need a deep understanding of random genetic drift and the effect of population sizes in order to explain coalescence. Here's how Douglas Futuyma explains it in the second edition (2009) of his textbook Evolution.
Earlier in this chapter, we introduced the principle of genetic drift by showing that because gene lineages within a population become extinct by chance over the course of time, all gene copies in a population today are descended from one gene copy that existed at some time in the past. The genealogical history of genes in populations is the basis of COALESCENT THEORY.
I didn't get such an understanding of drift and effective population sizes from reading The Selfish Gene. Surely this is a place where Dawkins could have mentioned the limitations of his forty-year-old gene's-eye view of evolution and how modern evolutionary theory has advanced beyond it in the 21st century?

Gould, S.J., and Lewontin, R.C. (1979) The spandrels of San Marco and the Panglossian paradigm: a critique of the adaptationist programme. Proceedings of the Royal Society of London. Series B. Biological Sciences, 205:581-598. [doi: 10.1098/rspb.1979.0086]


  1. If Dawkins does know he was corrected, or somewhat, a person would be reluctant to see the correction of admit it ot just try to wave it away. We all find thinkers who come up with ideas,get famous by them, and then corrected RESIST the correction.

    Its possible Dawkins does hear the gene not being seen as the selective agent.
    However his textbook claim on the gene would mean he does sincerely see the gene as the selective agent in modern evolutionary biology ideas.

    For evolutionism it must be the reproducing population that is getting evolved.
    No matter what is going on at the gene level. What is being reproduced is what is making a population and that then is what will evolve(including statis).
    After that a reproducing pair can start a new population however its all about populations. Especially to do the great biology changes evolutionism imagines.

    So the selfish gene must be a error even within evolutionism.
    Not only is it not the genes that are the agent, instead instead its successful reproductions, but the genes are not being selfish or tenacious.
    Luck, randomness etc etc, as written in the thread, are nullifying a gene struggling to maintain existence. Successful genes reproducing are nbot all struggling that hard.

    I think Dawkins is seeing the mutation(gene) as the essence of evolution
    while textbook evolutionism sees surviving/reproducing populations as the essence. Thus random mutations , newly, are being seen as important as instantly selection beneficial mutations.
    Population genetics is a successful criticism and correction.

    A YEC creationist however would see the genes as the origin for biology change. Like in human colours instantly changing for a whole migrating population. No selection going on.
    Not selfish but survivalist in a general program of the biology of the entity.

  2. @Larry
    This is the second time that you suggest that inclusive fitness does not exist outside of animals. As stated before: inclusive fitness does occur in all living organisms, including microbes. It will occur whenever there is some spatial pattern that keeps related cells together. If you scroll down a bit from my comment, you'll find that Brian and rich lawler gave some actual examples of features that evolved because they increase inclusive fitness, e.g. quorum sensing in bacteria.

    1. You can try and force microbes into an inclusive fitness analysis, but it doesn't really add anything because the whole difference between "self" and "non-self" is arbitrary in a clonal population. Helping another clonal cell isn't just helping a few genes you carry -- it's helping something indistinguishable from you!

    2. Inclusive fitness makes a trivial contribution to the evolution of life over the past 3.5 billion years.

    3. @Jonathan Badger
      If that were true clonally reproducing and parthenogenic organisms would be the most altruistic beings alive. This is not the case because freeriders and selfish mutants can still spread in a clonal population unless there is some spatial pattern that prevents this.

      Evolution of multicellularity is a consequence of inclusive fitness. Is that trivial?

    4. Dang. I was going to say multicellularity. Good thing I had a quick scan of posts to date.

      The key with multicellularity is the gametic bottleneck. The relationship of diploid genes to the gamete is what stabilises the whole. If the route out of the organism were diploid, there would be no payoff for foregoing somatic replication, despite 100% relatedness. Not to mention the dependence of 'selfish genes' on reduction for their very existence as units, via crossover.

    5. This is a silly debate. The only thing that ultimately matters are mean fitness values. The difference between inclusive fitness models and non-inclusive fitness modes is how you divide up fitness between individuals. If w_i is the fitness of the ith individual in one, then you can write w'_i=w_i+c_i for the fitness of the ith individual in the other, where the sum over all i of c_i is 0.
      There's not a thing in evolution you couldn't adequately address without inclusive fitness. There is also not a thing in evolution where you couldn't provide an inclusive fitness version. Some people might find one more intuitive than the other in some case, so https://www.youtube.com/watch?v=CIYS9EQWkXg

    6. Mean of what? Somatic cells don't really have a fitness if their lineage is ultimately terminated in favour of the same genes in gametes. The status of 'individual' pops up a level, from cell to colony.

    7. Of course somatic cells have a fitness - it's 2p, where p is the probability that the cell will divide before it dies. If the argument is that we are shifting a level here, I agree. But that's explicitly not something kin selection can handle - it deals with fitness at one level.

    8. I am probably using 'kin selection' in a somewhat loose sense, while retaining the idea of gene copies helping other copies not in their direct lineage. I don't really know how strict population genetics deals with this, given the implicit assumption (an many models) of efficient population stirring, which acts in direct opposition to the benefit in real-world multicellularity.

    9. I don't really know how strict population genetics deals with this

      Simple Haldane-Wright-Fisher models of Darwinian selection just don't: they assign fitness values to genotypes and evaluate those to the population mean fitness. It is true that fitness of genotypic classes already incorporates inclusive fitness, but that does not really give a lot of intuition about kin selection and such. I really don't understand Simon's point, so maybe he could elaborate a bit?

    10. I don't understand Dawkins' forcing of his selfish gene idea onto coalescent theory. I've read both books and the link is pretty tenuous. If anything, it's misleading. The Ancestor's Tale traces species back from present to past, as they coalesce into common ancestors. In Dawkins' book, there is a single history (or topology) that one traverses back in time. Coalescent theory shares this retrospective approach, but it is specifically not concerned with any specific topology of alleles as they trace back. In fact, the shape/topology of a given gene genealogy as it traces from a present sample of N to a past sample of a single ancestral allele is mostly unimportant; rather it is the statistical properties of numerous simulated genealogies that is important; one can then extract useful summary statistics from these simulations in order to determine the possible evolutionary scenarios that produced the pattern of variation in the present sample of N alleles.

  3. This comment has been removed by the author.

  4. "The Selfish Gene" certainly is influential inasmuch as people are still talking about it - at least, you are.

    His health issues aside, I think Richard Dawkins would be disappointed if his ideas about science didn't develop over forty years, including new research (done by other people). Then, he can write another book.

    "The Selfish Gene" and "The Lucky Allele" both sound like titles for children, but "The Selfish Gene" may be considered as poor teaching for the very young child because selfishness is rewarded with success. This is not to say that it isn't true. Also, the story involves sexual reproduction prominently.

    "The Lucky Allele" I am imagining as a story about the little allele that could, and did, but it also should acknowledge the other alleles that could, and they tried, but they didn't - because they weren't lucky. But "lucky" isn't a thing, it is just an outcome. Another outcome was equally possible. This may be a useful lesson for a small child in a large public school class. But the lucky allele also has to succeed by its efforts, or by those of its carriers. The bad allele may prosper, by accident, but the way you should wager is on the better allele.

    Luck is a thing in scenarios where a person or a creature has the friendship of leprechauns, or God. For instance, in the bible, the brothers Esau and Jacob. Esau is honest, hardworking, loyal, rather gullible. Jacob is a liar and a thief. God favours Jacob. God apparently likes clever, sneaky people. (Or, Esau is described as a hairy man, which is gross. Or God forms emotional attachments at random.) Although you could also argue that Jacob made his own luck (including leaving home hurriedly after committing so many crimes. Twice.) But God also helped him a lot in the story. In the bible, the nation of Israel is descended from and named after Jacob. I tell the story only to make a point about "luck"; I am fairly confident that "Jacob" is a person who never existed, and so is God.

    I have no idea if this interests you at all.

    Perhaps also a point about what the bible tells Israel to believe in. I think this advice is not good.

    In the time after Jesus (who also may not have existed), God now doesn't like clever people and prefers gullible ones. The bible says that.