About Rationally Speaking

Rationally Speaking is a blog maintained by Prof. Massimo Pigliucci, a philosopher at the City University of New York. The blog reflects the Enlightenment figure Marquis de Condorcet's idea of what a public intellectual (yes, we know, that's such a bad word) ought to be: someone who devotes himself to "the tracking down of prejudices in the hiding places where priests, the schools, the government, and all long-established institutions had gathered and protected them." You're welcome. Please notice that the contents of this blog can be reprinted under the standard Creative Commons license.

Sunday, July 26, 2009

Memes, selfish genes and Darwinian paranoia

I’m reviewing a book by philosopher of science Peter Godfrey-Smith entitled “Darwinian Populations and Natural Selection.” (This is not the book review, forthcoming.) Godfrey-Smith makes an excellent argument at some point in the book (chapter 7, on the gene’s eye view) that genes are not at all the sort of things Richard Dawkins and some other biologists think they are. For instance, contrary to the standard view, genes are not “unities of heredity” (and therefore do not last as “individuals”) for the simple reason that crossing-overs (the molecular processes that shuffle bits and pieces of genetic material, the real reason for sex) do not respect gene’s boundaries, but rather cut genes into pieces and shuffle them. Indeed, as Godfrey-Smith points out, for this and other reasons, sophisticated theoretical biologists are abandoning talk of “genes” altogether, referring instead to the more diffuse concept of “genetic material.” As PGS puts it, this is “a stuff, not a discrete unit.”

The interested reader will have to read PGS’s book or wait for my review (forthcoming in Notre Dame Philosophical Reviews) to learn more about the issue of the nature of genes. But what struck me toward the end of that chapter is Godfrey-Smith’s unusual (and, I think, rather compelling) argument that talk of selfish genes (and memes) is one example of a broader “agent-positing” discourse that is shared by, of all people, some evolutionary biologists (though by all means not all, yours truly being one of many exceptions) and theologians!

Here is how PGS himself has characterizes the phenomenon: “Two explanatory schemata can be distinguished within the general agent-positing category ... The first is a paternalist schema. Here we posit a large, benevolent agent, who intends that all is ultimately for the best. This category includes various gods, the Hegelian ‘World Spirit’ in philosophy, and stronger forms of the ‘Gaia’ hypothesis according to which the whole earth is a living system. The second schema is a paranoid one. Now we posit a hidden collection of agents pursuing agendas that cross-cut or oppose our interests. Examples include demonic possessions narratives, the sub-personal creatures of Freud’s psychology (superego, ego, id), and selfish genes and memes.”

I must say that I am rarely struck by a novel enough idea that my first reaction is “wow.” This is one of those instances. There is something profoundly intellectually satisfactory in suddenly seeing disparate phenomena like Augustine’s god and Dawkins’ memes as different aspects of an all-too human tendency to project agency where there is none. Not to mention, of course, the admittedly wicked pleasure I’m getting from imagining Dawkins cringing at the comparison.

Godfrey-Smith refers his readers to another author, Richard Francis, who talked specifically about “Darwinian paranoia” in his 2004 “Why Men Won’t Ask for Directions: The Seductions of Sociobiology,” and points out that there are other ways of thinking about natural phenomena that do not require what Daniel Dennett called the intentional stance. Quoting Dennett: “Here is how it works: first you decide to treat the object whose behavior is to be predicted as a rational agent; then you figure out what beliefs that agent ought to have, given its place in the world and its purpose. Then you figure out what desires it ought to have, on the same considerations, and finally you predict that this rational agent will act to further its goals in the light of its beliefs. A little practical reasoning from the chosen set of beliefs and desires will in most instances yield a decision about what the agent ought to do; that is what you predict the agent will do.”

The intentional stance works well when it comes to predicting what people might do (and in fact is at the basis of Dennett’s work on free will and consciousness), but it is treacherous when applied to things that do not have (conscious) agency. Again Godfrey-Smith on how to pursue research in biology while bypassing Darwinian paranoia altogether: “This is the kind of investigation where someone asks: suppose a population was like this, and such-and-such a mutation appeared, what would happen to it? Thinking this way does not require the idea that genes are ‘ultimate beneficiaries’ of anything.” And, I might add, would save us from a lot of unnecessary misunderstandings and acrimony generated by people who use agent-centered language much too easily and way too loosely.


  1. Massimo, not as aquestion of how "true" the selfish gene theory might be , but of how useful it might be to predict behavior, do you know any examples where it might be counterproductive to think as if the genes had some sort of intention?

  2. I agree about the ever-present dangers of agent-positing discourse, but I don't see why a gene isn't still a useful unit. But it is best defined not as a "unit of heredity" but instead as a "unit of function" or cistron.

    And indeed, if you do follow the research agenda of "suppose we have a population with this mutation", then the next thing to use is population genetics, which requires some sort of unit of heredity.

  3. Angel,

    yes I find talk of selfish genes unproductive any time one takes into account the reality of complex gene-gene and gene-environment interactions.


    first off, as you know there are functional genes that are not in fact cistrons (think of regulatory sequences). Second, as you say, population genetics requires units of heredity, it is rather agnostic about functions.

  4. I've been using PGS's Theory and Reality as my Phil of Sci text and really enjoying the way his mind works. I look forward to reading your review of this book and, most probably, the book itself soon thereafter. Can you just give a quick clue as to the scope of the book?

  5. Whoops, just looked up cistron and realised that its definition is often different to what I was thinking. What I meant was a set of nucleotide sites in close proximity for which mutations affect the same function, assessable via a complementation test. This naturally includes protein-coding sequences together with their regulatory sites, as well as genes (units of function) that do not code for proteins.

  6. Konrad,

    somehow I knew you were going to post a comment on this one... :)

    The book centers on a new conceptual-graphical framework representing the characteristics of Darwinian populations, and aims at understanding the spectrum of evolutionary phenomena from standard Darwinian cases to more borderline ones (in which PGS puts memes, for instance), with an eye toward helping to resolve issues such as the levels of selection and evolutionary transitions.

    I'm almost done reading it, and will write the review soon (due by the end of August). It is definitely worth checking out, though I really wished the publisher had used somewhat larger fonts. (And that's *not* because I'm getting older! :)

  7. Sorry to be so predictable. It sounds like I'll definitely have to look at that.

  8. I worked with genetics in assisted breeding therefore I know the structure of dna, and how it works when one specimen dna crosses with another one.
    The unit is definitely not the gene, neither is the single gene the one that causes manifestations of mosdt of the things, generally is a net of genes that make something express. Still is not pretty clear what are the crossing rules of dna.
    Thinking about the transmission of memes, you are right it is dangerous if the agent or entity that transmits it is not conscious, but I was thinking that different entities can have different levels of conciousness, for example some might understand structure others effect, etc.

  9. "..and in fact is at the basis of Dennett’s work on free will and consciousness), but it is treacherous when applied to things that do not have (conscious) agency."

    What does Dennett say about free will anyway?

    Listening recently to Del Tackett

    (TRUTH PROJECT) http://www.youtube.com/watch?v=XT1laELMSm4&feature=related
    (on Veritology)

    a few weeks ago talking about those who say that there is no such thing as free will. He says to them:

    "Did you have to say that?"

  10. I think PGS's book is excellent, and in particular his argument that the fundamental "unit" of Darwinian evolution is the population, not the gene or trait. But it's a hard read, especially in the setup. Which is not to say it isn't worth it.

    I also had a "wow" moment or two reading it.

  11. From the description above (I haven't read the book) it sounds to me as if PGS is committing the same error that Mary Midgely famously did, reading Dawkins as attributing literal selfishness to genes. Of course, Dawkins' point is only that genes behave, to some degree, as if they were selfish, not that they actually have a selfish "agenda".

    For instance, contrary to the standard view, genes are not “unities of heredity” (and therefore do not last as “individuals”) for the simple reason that crossing-overs (the molecular processes that shuffle bits and pieces of genetic material, the real reason for sex) do not respect gene’s boundaries, but rather cut genes into pieces and shuffle them. Indeed, as Godfrey-Smith points out, for this and other reasons sophisticated theoretical biologists are abandoning talk of “genes” altogether, referring instead to the more diffuse concept of “genetic material.” As PGS puts it, this is “a stuff, not a discrete unit.”

    I'm not quite the same person I was 10 years ago, or even yesterday. Does that mean it's wrong to treat me as a discrete unit? Of course, there's a question of degree here. Just how much do genes change from generation to generation? (I don't know the answer. I'm just an armchair theoriser.) If they change beyond all recognition, then it probably doesn't make sense to treat them as discrete units.

  12. RichardW, not having read the book I can only make the point on the basis of having read Dawkins but it seems to me completely unneccessary for PGS to be thinking along the lines you suggest. It is clear that Dawkins is taking the intentional stance towards genes to some degree. The way in which people have misinterpreted what 'selfish' means in his work is, I think, to some degree due to him having taken the intentional stance and, thereby, making it sound plausible that genes might have personality traits such as being selfish. I would be shocked to hear that PGS is making that error and doubly shocked if he did and Massimo hadn't mentioned it.

    The problem with the comparison you make between the identity of changing persons and changing genes is that individual people maintain a large amount of autonomy from their surroudings, autonomy that at least in certain respects is missing in the case of genes. The situation with genes is more like those concept albums on which the music has no breaks and the way it is divided upon the CD into tracks is to a certain degree arbitrary. What constitutes a song on that album and wouldn't it be better to just talk about song-stuff?

  13. Richard, Konrad is correct, PGS does most certainly not make the same mistake that Midgley did in fact commit.

    Konrard is also correct in his example of concept albums (nice!). To which I will add that individual organisms are not a unit of heredity, so the problem does not apply to them also for that reason.

  14. This is an interesting discussion. Here are a few thoughts. First, I liked the comparison between genes as units and 'songs' on concept albums. I thought about an analogy with the artificial breaking of very seamless movies into 'scenes' on DVDs, but I like the concept album comparison better (especially because I like concept albums). That is exactly the point I wanted to make at that stage in the argument. Genetic material is real, just as music is real on the album. And you can divide stretches of both of them into units, for various practical purposes. But the bondaries between units are not real features in either case.

    Second, regarding the Midgely discussion: I do think she was mistaken. Agential language is used metaphorically by people like Dawkins, so there is no point in arguing against him by saying that the language cannot be literally applied. But some metaphors have peculiar force; they affect our thinking more than we realize. In a discussion with Dennett on this topic, he suggested that my argument is that an agential metaphor should be seen less as a tool that we can pick up and put down at will (his view) but more like an addictive drug. Good comparison!

  15. G'day Peter, you and Massimo are both welcome to the comparison if you like it. I suspect we're all showing our age by owning up to liking concept albums.

  16. I think Jacques Monod in "Chance and Necessity" brilliantly articulates exactly why agential thinking, in his more colourful terms "animism", is such an addictive drug. It plays to deep human psychological needs for a satisfactory "explanation", needs that are ultimately behind the common but sad fact of human discomfort with science, and that are a constant danger to the scientific method.

  17. Thanks, Peter, for your clarification regarding agency. Since Dawkins is not attributing a literal agenda to genes, I can't see what purpose is served by categorising selfish genes with demons and Freudian sub-personalities, which do have literal agendas. No doubt this would be clearer if I read your book, but I was just responding to Massimo's description, and I can't help feeling that his enthusiam for your schema is largely motivated by the obvious satisfaction he gets from seeing Dawkins grouped with demonism and Freud, a kind of guilt by association.

    I don't suppose your discussion with Dennett on this topic can be found online? I'd be very interested to read it.

    Please excuse my ignorance of genetics, but isn't it the case that most genes code for proteins? Since proteins are discrete molecules, it would seem that protein-coding genes, at least, can be considered discrete units, even if each gene isn't necessarily composed of a continuous stretch of DNA. Of course, if a gene is defined as a continuous stretch of DNA then this wouldn't make sense. But in that case perhaps a less restrictive definition of gene would be useful.

  18. Richard,

    many genes do code for proteins, but many are non-coding, regulatory sequences. At any rate, the issue of the function of a gene is different from whether that gene is a discrete hereditary unit. Peter's point is that the latter can be seriously questioned.

    As for me taking satisfaction in seeing Dawkins's ideas compared to Freud's I have admitted so in my post. That doesn't mean there is no substance to the criticism, though.

  19. There approximation that a gene is a discrete unit isn't necessarily such a bad one. I'm not sure how many cross overs occur during meiosis, but if there are far more genes than there are crossovers, the odds of any particular gene being cut aren't good, and you can consider it as a somewhat permanent object. In fact, Dawkins knows this, and he knows that sometimes they get broken in reproduction, and he knows that sometimes several genes get carried along through generations without getting broken up. He used a bit of space in The Selfish Gene explaining several different definitions of "gene" that were in use, and he explained that the particular sense that he meant was a segment of DNA that is big enough to get something done, and that actually does something, while being small enough that it's unlikely to be split up during meiosis. Of course, I haven't read PGS's book, so maybe he's saying something new here as far as research people are doing, but it sounds like he's describing the concept no differently than Dawkins did in TSG, which Dawkins said wasn't all that new anyway, just a new metaphor.

    And that leads us to the second point Massimo Pigliucci is startled by: that Dawkins's use of a metaphor of a gene as an agent parallels other uses of metaphors for agency. Now, if you'd bothered to read TGS, you probably would have noticed the parallel yourself, since Dawkins makes it explicit, in order to reinforce that it's only a metaphor. Dawkins was not expressing a secret desire to find agency in the essence of our DNA, rather he was exploiting our innate cognitive bias/intuition toward seeing agency everywhere to explain the mathematical theory of gene propagation.

    I'm sure PGS's book is stunning, but MP didn't really offer any examples as to why it should be stunning.

  20. Peter,

    yes, I've bothered to read TSG, and I wasn't too impressed with it even when it came out. (For one thing, it's a popularization of ideas by others, largely Williams and Hamilton, though Dawkins actually gets the credit for most people.)

    If you are seriously interested in philosophy of science you need to read PGS's book, not just rely on a few comments posted on a blog. (I'm always amazed by how many people take me to task for not going in-depth enough, as if they were reading a scholarly paper, not a blog.)

    As for Dawkins knowing about crossing over, he better. But he clearly underestimates its importance; indeed Dawkins brushes aside other semi-fatal flaws to his idea, like the widespread existence of epistasis (where genes "collaborate" with each other, and thereby cannot be "selfish").

    Finally, of course Dawkins was aware of using a metaphor. But the point is that metaphors have implications and lead people -- including their authors -- to think along certain lines rather than others, which may be misleading, depending on the metaphor.

  21. I don't think Dawkins underestimated the importance of crossover. As I said before, he explained that his use of the word "gene" in that book is not "gene" as a protein coding segment of DNA, but rather as a length of DNA that is long enough to do something (including coding for protein, or regulating coding sections, or whatever else DNA does), and that is short enough to rarely be scissored up during meiosis. It's a fuzzy definition, but that doesn't mean it's not valuable. Have philosophers of science not noticed how scientists use approximations?

    And Dawkins clearly gave credit to Hamilton et al in TSG, explaining that the theory wasn't new, but that he thought the metaphor was new and useful. Now, that might be important to a lay-audience, and even to biologists. It can be easier, more intuitive to think about "selfish genes" than about Hamilton's math.

    Really, I didn't mean to take you to task for not going in depth enough. You were clear that this wasn't the main review. Rather, I think that even in what you meant to be a brief comment on the book, you picked some underwhelming, old ideas to be wowed about. Also, you did say:

    "Not to mention, of course, the admittedly wicked pleasure I’m getting from imagining Dawkins cringing at the comparison."

    Sure, you said that about memes, I guess. But since Dawkins introduced memes as an analogy* for genes, and he only talked about genes having intentions metaphoricaly, I don't think he'd be surprised or cringe at the comparison.

    You also just mentioned,

    "metaphors have implications and lead people -- including their authors -- to think along certain lines rather than others"

    Dawkins would of course agree with this, too. Since he said that's why he thought the "selfish gene" metaphor was valuable and worth writing a book about.

    Really, in my science experience, I've found that some problems are best approached from one metaphor, some are best approached from other metaphors. These various metaphors usually interpretations of different ways to approximate a complicated system. For instance, you could take a Taylor expansion in one parameter and get one approximation or take an integral expansion around another parameter or take a binomial expansion around another parameter, and get different results that are described with different metaphors.

    If they are all good approximations, they'll all converge on the same result as you add more and more terms in the expansion. The "selfish gene" seems to me a good approximation for small gene length and large pools of genes that each gene is interacting with, so that the genes are interacting with a somewhat constant environment of other genes.

    The individual as the unit of selection is also an approximation with its own problems, and so is the reproductively isolated group. You can pretty much circumscribe those problems, and explain when "selfish individual" and "selfish group" type selection effects would be seen. If Darwinian Populations and Natural Selection goes into those effects and when you expect to see them in the natural world, I'm sure it's awesome. I don't think I've seen group selection described with any kind of rigor before, I've mostly just seen (what look to me like valid) criticism of group selection ideas that various people have floated.

    So, really, I just mean to be taking you to task for your strange criticism of Dawkins and his selfish gene metaphor. If you meant to plug the book, you probably didn't need to use it as excuse to take a swipe at TSG.

    *an analogy of something else that natural selection could work on...it doesn't have to be empirically supported or theoretically rigorous for the analogy to be useful for that, you know--maybe you think he takes the idea of religion-as-memes too seriously?

  22. As an example of somebody getting trapped in a metaphor, how about this little item, where Daniel Dennett gets trapped inside the design stance and can't get out.

    This was pointed out by Larry Moran on his Sandwalk blog.

    It seems to me that Dennett gives up way too easily. Function does not necessarily mean design.

  23. Not to turn this into a Dawkins festschrift, but this needs comment. Massimo wrote
    As for Dawkins knowing about crossing over, he better. But he clearly underestimates its importance; indeed Dawkins brushes aside other semi-fatal flaws to his idea, like the widespread existence of epistasis (where genes "collaborate" with each other, and thereby cannot be "selfish").
    Referring specifically to collaboration among genes, Dawkins once said (I can't now find the reference) that he could have titled the book "The Cooperative Gene" with no change in content. Gert Korthof mentions that here:
    The Selfish Gene is "an introduction to evolutionary theory, it explains a number of deeply counter-intuitive results, including how an apparently self-centered process like Darwinian natural selection can account for the evolution of altruism (review: Allen Orr). "As Dawkins explains, the book might as well have been called 'The Cooperative Gene' (Coyne). It would have been better to do just that!

  24. It is interesting that, despite disclaimers to the contrary, every time I write about Dawkins it *does* turn into a festschrift!

    RBH, I know of that comment, but there it seems to me Dawkins is simply being disingenuous. First of all, then why didn't he? (Wouldn't sell, is my guess.) Second, try rewriting the book that way. Not going to work because the metaphor and its implications are too pervasive.

    Peter, nice post, but frankly this is not the place for a point-by-point rebuttal of such long commentaries. I really do suggest that if you are that interested in the matter you read PGS' book, and also (for a critical yet more sympathetic view) Kim Sterelny's "Gould vs. Dawkins: The Survival of the Fittest." I have also written elsewhere on this blog and in Skeptical Inquirer about why both the selfish genes and the memes metaphors are problematic.

  25. Hi Massimo

    I've searched post in this blog trying to understand why you seems to dislike Dawkins so much. If you have some specific writings here or anywhere taht you can point out I would be grateful.
    I like Dawkins in general, but want to hear more qualified critics, that is not religious ones or "atheist but..." ones.

  26. Ene (or anyone who is interested),

    send me an email and I'll email you back my "Dawkins trilogy" from Skeptical Inquirer.

  27. So I'm apparently computer illiterate because I'm interested in your Dawkins trilogy but cannot find your email address anywhere on this blog.

  28. Jordan, yeah, I'm a bit coy about my email address, but you can send your request to:

    massimo at rationallyspeaking dot org

  29. Massimo, having now read your NDPR review I am convinced that I, along with a number of other people, need to read this book. I am also very interested to see what comments Peter gets from the likes of Richerson and Boyd. Next year's European Human Behaviour and Evolution conference is going to be in Wroclaw and I hope there will be some discussion of the book there as it is raising what sounds like a very substantial objection to much of that work.


Note: Only a member of this blog may post a comment.