One would think that after philosopher Mary Midgley’s monumental misunderstanding of the thesis of Dawkins’s The Selfish Gene, one corrected definitively by Dawkins himself, the record would be clear. But one would be wrong. Midgley not only thought that Dawkins was claiming that—and approving of—the tendency of evolution to always produce selfish creatures (it doesn’t), but she also confused the metaphor of a gene being selfish (it behaves as if it wishes to displace other genes) with a fragment of DNA actually being selfish, which is ludicrous (see Dawkins’s refutation here). Genes don’t have consciousness or desires.
Sadly, Andrew Brown, who increasingly shows signs of willful ignorance rather than just wooly-headedness, commits the same error in a new Guardian piece: “Evolution will punish the selfish? It’s not as simple as that.” Brown is beefing about a new paper in Nature Commmunications (free online) by Christoph Adami and Arend Hintze (reference below; free download). That paper shows that previous theoretical studies of “ZD” strategies, in which one always interacts selfishly with an opponent, were wrong. The earlier work showed that ZD strategies would successfully invade a population of individuals who were either unselfish or behaved “tit for tat” (“I’ll scratch your back if you’ll scratch mine). That would invariably result in populations of individuals who behaved selfishly. And that, of course, is contradicted by the many animal societies that show some cooperation.
What Adami and Hintze showed in their new paper is that ZD strategies are not “evolutionarily stable”: that is, even if they invade a population, they will ultimately lose unless they can for some reason recognize a priori those other individuals with a “nicer strategy”, And, even if ZD strategies became fixed, creating populations of selfish individuals, such populations would be evolutionarily unstable, subject to reverse evolution in which they became more cooperative.
The value of Adami and Hintze’s paper is that it repatriates the evolution of cooperation as a viable outcome of evolution, at least using game theory. And it’s a nice paper because, unlike many game-theoretic models, it uses a fairly realistic genetic model, in which the difference between selfishness and non-selfishness is based on five genes. (Most game-theory studies of cooperation have used no genetics, implicitly assuming that a behavioral difference is based on a single gene, and one that is dominant.) Sure, the paper doesn’t show that cooperation in nature did evolve this way—for theory cannot do that—but it does what theoretical biology is supposed to do: show what can and cannot happen under certain assumptions. Ergo, the evolution of cooperation in societies is still a viable strategy. (See also the press release from Michigan State University.)
Sadly, but understandably, Andrew Brown doesn’t understand the function of evolutionary theory:
The [Nature report] is actually a report of a result in game theory which overthrew an earlier experiment in game theory suggesting that a completely “ruthless” strategy would succeed in a contest with other ones slightly less ruthless.
None of these experiments are conducted in the wild. They are all computer simulations. This is another reason to be slightly sceptical of all these grand results – it’s possible that when the pretty mathematics are fitted to the ugly world, they will break. But that’s true of pretty much all the kind of science that makes news. There’s nothing special to biology about it.
But of course one has to be skeptical of all mathematical results, since they’re all based on simplifying assumptions. Nevertheless, those assumptions are often useful in helping us understand nature. (I am thinking of sex-ratio theory here, as well as my friend Michael Turelli’s theory on how a parasitic bacteria that causes sterility in mosquitoes could sweep through a population. Turelli’s theory has actually been used to eliminate mosquito-carried dengue fever in parts of Australia!) Population genetics has been very successful as a hand-in-hand coupling of theory and experiment. To give but one example, we now understand why there are so many “self-sterility” alleles in plants: those alleles that prevent a plant from mating with itself.
Theory is also good at getting rid of misconceptions based on intuition. I think it was the Scottish statistician George Udny Yule who, based on his intuition, claimed that a dominant allele (say for brown eyes rather than blue) would sweep through a population simply by virtue of its dominance. But, as three scientists showed with their simple “Hardy-Weinberg-Castle” model (one of those models that Brown is skeptical about), brown/blue allele frequencies will stay the same in a large population if natural selection is not operating (along with a few other assumptions). That is, the math showed that the verbal intuition was simply wrong.
Brown goes on to commit the Midgley Error: assuming that “selfish” is anything more than a metaphor, for both genes and computer codes:
The interesting question is how we come to describe a fragment of computer code as “ruthless”. I put scare quotes around “ruthless” because the attribution of moral qualities like ruthless, selfish and even altruistic (assuming for the moment that altruism is moral) to computer programs is the essential mechanism by which these stories spread. Yet of course no one who thinks about the matter for a moment supposes that computer programs (or genes) can be selfish, altruistic or ticklish. It’s just that no one would give a damn about them if they weren’t described emotively.
The root of all this anthropomorphism is Richard Dawkins’ first book, The Selfish Gene, which remains a masterclass in science writing. Anyone who could pass an exam after reading it would have a sound understanding of evolutionary biology – and of rhetoric.
That exam has only one question: “In the light of the text of the book, candidates will explain why ‘selfish’ does not mean ‘selfish’ and ‘gene’ does not mean ‘gene’.” The author himself would score about 80% on it (and claim for the rest of his life that the examiners hadn’t read the book). The average headline writer would score 0.
. . . But it is worth pointing out why it matters that genes and programs can’t be selfish in any interesting or important way. That’s not because believing this leads us to misunderstand genes, or to program computers badly. It’s because it leads us to misunderstand selfishness, which is a moral quality displayed in the acts and choices of responsible beings.
Brown hasn’t scored 80% here; he’s scored zero. That’s because he misses a major aspect of the book: that the term “selfish gene” is just a metaphor, Mr. Brown! Gene replication behaves as if the genes are selfish. There is no morality in either computer code and genes (the Midgley Error). The emotive description only confuses those people who can’t think very hard; otherwise it is quite enlightening. It is in fact a very good metaphor. Do you beef, Mr. Brown, when your car doesn’t start, and curse it for being “cranky” or “uncooperative”? Or do you object in general to any metaphor that is anthropomorphic?
And look at this backhanded “compliment” Brown gives Dawkins’s book (which of course has sold millions of copies, far more than anything Brown ever produced):
What makes the Selfish Gene such an interesting and important book is that it contains in itself all the arguments you need to understand why it is absurd to call genes (or computer programs) “selfish” – and then sometimes, and with equal force, ignores them.
I’m not knocking contradiction here. It’s the flaws and the self-contradiction that make the book compelling.
No, what makes the book interesting and important is that it explains in a compelling way how natural selection works, not because its premise is absurd or in any way contradictory. In an afterward, and repeatedly since its publication, Dawkins has explained not only that selfish genes can cause cooperative behavior, but that he could equally well have called his book The Cooperative Gene. DId you miss that, Mr. Brown?
Brown goes on to blather about how people can use computer programs in a humanly selfish way,—for example, bilking Greeks out of their pensions—and of course that’s true. But it’s completely irrelevant to Dawkins’s thesis, and is merely another cheap shot. He concludes with this:
Of course we’re machines subject to physical and chemical laws. But we are such immensely complicated conglomerations of such machinery that we need a new and different set of concepts – things like morality, responsibility, ruthlessness and selfishness – to describe their interplay. To talk as if the same concepts could be applied to genes or program fragments and to human beings is dangerously misleading – even though it’s fun and makes for memorable headlines.
No, it’s not dangerously misleading. Really, Mr. Brown, what “dangers” have resulted from Dawkins’s enlightening metaphor, save the loss of trees expended on corrected the stupid misunderstandings of people like yourself and Mary Midgley? Have you ever cursed at your computer as if it were willfully misbehaving, or treated any piece of machinery as if it was malicious? I thought so.
This is just another of Brown’s pieces reflecting his venomous attitudes towards Dawkins, but it also shows his misunderstanding of both biology and good writing. Yes, I suspect, as many readers have suggested, that the Guardian keeps Brown on simply because his stupidity draws corrective comments and lots of hits, but doesn’t there come a point when this kind of blather is inherently embarrassing to what used to be a good newspaper?
________________
Adami, C. and A. Hintze. 2013. Evolutionary instability of zero-determinant strategies demonstrates that winning is not everything. Nature Communcations online: doi:10.1038/ncomms3193
