This is a follow up on the controversy that followed David Dobbs piece entitled “Die Selfish Gene, Die”. Dobbs published a revised and much improved version, one that merits some praise and a serious discussion. When I first read the original piece, I was furious: to me it looked like a mean (and very well crafted) click-bite, designed to create controversy by spinning misleading pseudoscience against the controversial figure of Dawkins. I was enraged because it used erroneous and incomplete concepts to challenge one of the most powerful ideas in the history of human thought: the Selfish Gene. My rage made me do two things: first I tweeted to Jerry Coyne, asking him to engage in the debate; but I couldn’t wait, so I wrote my own rebuttal as well. Both my actions were far from friendly: my own piece is pretty harsh, and Jerry Coyne stroke even harder (and I knew he would either do nothing or strike hard – his replies: part 1 and part 2) [to this day, I don’t know if my tweet did play a role in convincing Coyne to engage].
Things got heated, Dobbs replied to Coyne, Dawkins himself answered, and it all seemed very confrontational (I was gracefully and predictably ignored by the big shots). Dobbs initially tried to play it nice, and said that he only wanted to shake the general public understanding of the Selfish Gene, he didn’t mean to question its scientific validity. I didn’t believe him, and was ready to move on until Dobbs produced a second version of the piece, one that is much improved in all departments. But he wasn’t happy to just rectify his article, he went on and published an explanation, a link to a PDF that highlights most changes, a link to the original version, and another copy of the revised article enriched with links to sources.
What? That is a lot of work, and the improvements themselves also imply plenty of reading and fact checking. First conclusion: I was wrong in my evaluation of Dobbs’ intentions. He just gave me a lesson on journalistic integrity, leaving a clear and open trace of what he did, why and how, while enriching his content significantly. That’s an example that all science journalists should follow, bravo!
I promised him that I will comment on the actual content, so my thoughts are below.
There are three aspects that need to be looked at separately: the first is the personal side that I have addressed above. The second revolves around “public understanding”, something that is at the core of Dobbs’ intentions. The third is technical and scientific. I will address the latter two below, but I can already say that I agree with Dobbs’ attempt to popularise a more complex evolutionary “story”, but still find plenty of his technical details questionable.
Improving the public’s understanding of Evolution
If I was the original author, dealing with editors that wanted a click-inviting nice an catchy title, I would have accepted to call it something like “Why Everything You Know About Evolution Is Wrong“, or something like that, so to make it clear that the aim is to change the general public knowledge and to catch up with the latest science. So what needs to be changed in this imaginary “public understanding”? In four words, the notion: “one trait, one gene” or it’s extended version “one trait, a known and defined set of genes“. But let me start from the beginning. “Selfish Gene” is a handy shorthand for: “natural selection acts on single replicators” (more on this on the technical section below). This second definition doesn’t work well for the general audience but is firmly at the core of what is generally identified as “modern evolutionary synthesis”. The popular definition is gene-centric, the technical one less so. This mismatch can aid the ubiquitous drive toward over-simplification and build a collective understanding of evolution and genetics that is over-simplified and, arguably, wrong. Dobbs finally managed to produce an article that convincingly makes this case, and rectifies as follows:
- Single genes rarely influence one trait directly, they typically work within vast and spectacularly complex network of genes.
- Regulation of gene expression can be and often is more significant than single mutations in transcribed genes.
- Therefore, thinking that one mutation influences one character and that this character can be selected for or against is over-simplistic.
- These considerations are important for the general public because genomics is entering a commercial era, where our everyday behaviours will be influenced by it: it is important that people understand how complex is the interaction between genotype and phenotype. We can even go as far as saying: one genotype can lead to multiple phenotypes (with caveats that I will omit).
He is right, and I applaud the intention. I have no idea on the effectiveness of his article in transmitting this message as my previous knowledge blinds me, but the intention is laudable.
I will argue that unfortunately the execution is still technically flawed and imprecise, albeit massively improved. For brevity sake, I will not highlight what Dobbs has successfully corrected (he did a good job, so the list would be long) and will not cite every single passage that I find questionable. I will also try to avoid going into obscure technical details and make the following section as accessible as possible, please bare with me if you are interested. What I will do is build an argument that explains why the current scientific challenges to the selfish gene model that Dobbs endorses are in fact misguided and unjustified.
Improving scientific knowledge requires to build on the “selfish gene”, not to get rid of it.
First of all, Dobbs plays the abiguity of the term “gene” in his favour, sometimes making me think that he himself is still confused, for example:
Even as a technical term, the word carries at least a half-dozen meanings, and more are added as science finds new tools for exploring the genome. […]
But the gene’s definition is not just semantically vague. As geneticists explore the genome’s previously uncharted stretches, they’re finding that a lot of the work conventionally attributed to ‘genes’ (in the sense of consistent, reasonably well defined clusters of DNA) appears to be done instead by networks of genes and strange DNA elements that doubly defy the selfish-gene model.
These regulatory networks challenge the selfish-gene model first because they include DNA elements not conventionally defined as genes. More important, some researchers believe these networks challenge the selfish-gene model because they often seem to behave not like selfish entities balancing their separate agendas, in selfish-gene style, but like managerial teams regulating the behaviour of individual genes for the interest of the organism.
The passage above is unhelpful, it muddies the waters instead of clearing them out: when thinking in evolutionary terms, there is one and only one definition of gene, that’s “DNA-based replicator”. If the information that is propagated across generations is stored in the DNA sequence, the information is genetic. If not, we are dealing with something else, and in that case, the first question that one wants to address is: is this new carrier of information a replicator? Does it have the potential to make the information virtually immortal by producing new copies at each generation? If it does, natural selection acts on it as it acts on traditional genes, otherwise other concepts are needed. My point is that (and it was main point also in my first rebuttal) as long as the information is stored in the DNA, whatever happens starts with the “selfish gene”, it may be that there are lots of layers of complexity that act on top of the basic principle (and there are), but they don’t invalidate the foundation, they are actually the result of it.
Case one therefore is: the inheritable (or transmissible) information is stored in the DNA sequence. If that’s true, the challenge for scientist can be divided in two strands.
First Strand: is about understanding how the additional layers of complexity work, for example how the environment triggers the transformation from grasshopper to locust. This kind of problem may be addressed with the tools of molecular and cellular biology, physiology, genomics and more. The interesting fact is that they are likely to require the collaboration of what used to be distinct fields, in what can only be described as a multidisciplinary effort. In this strand, the “selfish gene” concept may be somewhat redundant, because the question pertains “how does this thing work?” and not “how did this thing come into being?”. It is notable that most vigorous challengers of the selfish gene metaphor are indeed scientists that work in the fields I mention above (except probably molecular biologists), they are not evolutionary biologists, and it shows (to some extent, this is understandable and justifiable, see the Note* below).
Second Strand: understand how the apparently simple and straightforward “selfish gene” mechanism of selection can lead to the emergence of such overwhelmingly complex additional mechanisms. For example: what makes it possible that “[some networks of genes] seem to behave not like selfish entities balancing their separate agendas, in selfish-gene style, but like managerial teams regulating the behaviour of individual genes for the interest of the organism”? To answer this question we need to explain the link between the fact that selection acts on single replicators and the clear-as-the-sun evidence that replicators tend to aggregate and build complex networks. As I’ve said in my first piece, this is a very important question that is not limited to evolutionary biology: it spills over into the humanistic domain, because it is likely that it will inform our understanding of sociality and collaboration between complete organisms, not just between genes (you can expect me to write a lot about this subject). Anyway, what I’m saying here is the exact opposite of what Dobbs suggests: these additional and puzzling (or, if you ask me, exciting) layers of complexity do not challenge the selfish gene view. They challenge our human attitude of trying to simplify things and deny the existence of such mindblowing complexity. Scientifically, they require us to embrace and tame this complexity, but we can’t throw away the knowledge that selection acts on single replicators, simply because we have exactly zero evidence that suggests this may not be the case. On the contrary, we have a massive amount of evidence that suggests that the basic principle (selfish gene) is necessary, and probably also sufficient for the emergence of the complexity that surrounds us.
Dobbs also hints at another family of problems, what I’ll call here the second case. This happens when the inheritable (or transmissible) information is not stored in the DNA sequence, but on some other medium. I’ve got news: this happens all the time. It’s called culture, and the information is stored everywhere, minds, books, road signs, hard drives, Internet… This second case defines a number of strands of scientific inquiry, but does not challenge the selfish gene idea, it actually makes it central (again).
Let’s see a few examples.
Third strand: if information is stored and transmitted on non-genetic medium the first thing to look for is the alternative carrier. Epigenetic mechanisms such as DNA methylation, siRNA and other have been proposed, but to me, their link to DNA is so obvious that it makes the attempt futile: they ultimately fall into the classic DNA-based model (first and second strands). Memes (the single replicators in the cultural domain) are a case of non-DNA-based information transmission (and unsurprisingly Dawkins invented the term in “The Selfish Gene” itself), but we know close to nothing to how our brains store information, making this strand pretty dry for the time being.
Fourth strand: understand the replication mechanism that acts on the alternative substrate, and with it, see if the “selection acts on single replicators” model applies or not. In the case of culture, this is again a tricky business because we replicate information in all sorts of ways, making the idea of treating memes as genes problematic at least. I am unaware of other viable cases: as far as I can tell biology is entirely supported by information stored in DNA and brains, nothing in Dobbs article convinced me to change my mind.
Fifth strand: co-evolution. What happens when selection happens on two different kinds or families of replicators (say, as Dobbs suggests, genes and memes) and these interact via their respective effects? Now this is another very interesting strand, and one we are just starting to explore, but again, it does not require us to put aside the selfish gene metaphor, it requires us to understand the levels of complexity that are generated by natural selection. If we want to pursue this strand, we need the selfish gene concept as our starting point, it is the one mechanism that we do understand and that somehow generates the phenomenon that we are studying (pretty much as for strand 2).
Dobbs revised piece makes one central point: it is time to try to communicate the complexity of evolutionary biology to the general public, the simplistic model “one gene, one trait” should be dismissed along with equally simplistic extensions of it. I agree with Dobbs and support his effort in this direction.
He also suggests a secondary point: that the selfish gene metaphor is slowing down scientific understanding. I can only repeat what I’ve said before: this attempt is erroneous, misleading and unsubstantiated. It does not however, invalidate Dobbs’ central message, and that’s the important one.
*NOTE: I’ve been writing about this virtually in all of my posts. Knowledge is made up by models and models are justified by their use. If one wants to study the transformation between phenotypes while keeping the genotype unchanged, as in different cell types, the transition between grasshopper and locust, different individual specialisation in social insects and so on, the selfish gene model is not very helpful. It is reasonable to expect that alternative levels of abstraction (more at the system level) will allow to create explanatory models that can be handled and understood by human beings, hence progressing the field of interest. Therefore it is not surprising that physiologist (as a broad category) find the selfish gene idea overrated, but their occasional claims that the concept is invalid are just as wrong. It’s a model that only marginally concerns their work. Everyone would appreciate if/when they will be able to produce other models of equal explanatory power, but there is no reason to expect the new models to be alternative to the selfish gene, complementary is a safer bet.