« Blog navel gazing | Gene Expression Front Page | The end of the racial democracy »
April 06, 2003

Human(e) Sciences: Group Selection-the prologue

Well, I said I'd get back to this this weekend, and I've finished Unto Others again, but Darwin's Cathedral remains un-reread. The latter is short, I'll get to it today at some point. As far as the first book goes, I read the 2/3 of it that deals with evolutionary biology pretty closely, but I snoozed through the theoretical psychology & philosophy-of-the-mind parts, though I awoke periodically when they presented data from the laboratory. This happened the first time around, so I think I've satisfied the condition of reproducibility when I submit the hypothesis that most psychology & philosophy-of-the-mind bores me to explain the finding of my semi-sentient mental state when exposed to said stimuli.

Anyhow, if you read the reviews of Unto Others, you'll see they're pretty good, and they tell you to do more than just skim it. Though the content is only 340 pages, I will say it is a 'big book' in the density of argument and supporting data. Some have complained about the rhetorical format the authors use, this gets out of control in my opinion in the second section as the advocacy started to annoy me (almost certainly written by Sober, who is a philosopher by training, and wordy-to-boot from what I could figure out). I found myself examining the equations that were highlighted in special boxes to indicate that less technically oriented readers didn't have to bother understanding them because the text elaborated and explained in loquacious detail what a few lines of math could encapsulate [1]. But there was a reason they presented so much padding, repeated themselves ad nauseam and worked their case as if they were attorneys in front of a slack-jawed jury-Wilson & Sober believe that they are facing a paradigm block. Frustration is a recurring theme in Wilson's tone as he makes the case that group selection has been validated empirically and that W.D. Hamilton himself weighed in on their side in the mid-1970s only to be ignored.

I came into it as an individual selectionist I suppose, but I was open-minded and could have been persuaded by some pretty math and citations from the literature of predictions that matched what was in the lab or the field. But the authors clearly believe that this hasn't sufficed for the past generation to convince the doctrinaire individual selectionists, especially reductionists like Dawkins. In short, Wilson and Sober are presenting in Unto Others a multi-faceted thesis and intellectual exercise, ranging from population genetics & ecology, yes, but venturing into philosophy of science and history and ethics, and the latter can not be reformulated in a few compact equations and appeals to a somewhat different audience than those who would be drawn to the first chapter of the book. I would not be surprised if people interested in psychology were scared by the math in the first few chapters and never made it to the back, because the tone & style shifts radically between part I & II. They are attempting to force what Chris Mooney would term a shift of frames-not just elaborating at length their academic papers. You got to give them props for the attempt, though the execution can be a bit laboring for those of us who have no professional grudge against group selection.

As the comments on my previous post titled 'Groupies' implied, a lot of the debate seems to be around semantics. Wilson confirms this because he comes back over & over to the point that individual selectionists simply redefine clear group selection as individual selection [2]. I haven't read JM Smith's critiques of Wilson yet, so I'm not going to wade too deep into the debate, but let me present the general thrust of the ideas in Unto Others.

Wilson cites the averaging fallacy many times in this book (and brings it up in Darwin's Cathedral too). Instead of looking at the fitness of individuals within the context of their groups and evaluating the groups themselves, biologists just average the fitness of individuals over all the groups. What they mean is this, you have two phenotypes in a species, and a population of animals form discrete breeding pools, then you lump all the individuals from the from the two phenotypes into two pools and then average their fitness as if they were in one whole population and ignore the superstructure of their group organization. Here is a website with more detail on this issue and the general points of Wilson & Sober's thesis.

Groups matter to Wilson, but as everyone knows, "group selection" can't work if they are always isolated. Selfishness takes over because in within-group competition, free-riders always win. If you have groups A & B, with A being filled with half selfish and half unselfish phenotypes, and B being all selfish, it is plausible that the unselfish individuals in A will contribute to greater group success and B will be marginalized-but in the long run, the selfish individuals will still overwhelm the unselfish ones in group A. For unselfishness to not be transitory and group selection to work in the simplest models, there has to be a period of relative isolation of groups followed by dispersion and reassociation-in such a fashion, though within groups the unselfish individuals are less fit than the selfish ones, the reshuffling acts to prevents selfishness from outcompeting unselfishness within the various groups.

I'll come back to this with a few equations and graphs to illustrate the points later-suffice to say, I am starting to understand why Wilson had to use so many words to describe what the hell he was talking about. And that is why group selection kind of sucks, Wilson admits that it isn't particularly parsimonious, and it seems eminently vulnerable to sloppy thinking, even if it jives better with some empirical data points (Wilson admits group selection's past faults and bemoans how they have stained the paradigm's reputation). Because of this, it will take a lot of evidence to shift people from skepticism or agnosticism-beauty is not on the side of the group selectionists. Reductionism is seductive and almost religiously appealing to the scientific mind-Wilson makes clear that he has sympathy for those who would speak of a holistic multi-level selection. This converges with some of the ideas of S. J. Gould-and anyone that reads this blog knows that I don't have much sympathy for that man's scientific ideas. That said, I'm not a population geneticist, and I do believe that though Wilson hasn't proven his case, he does add to the contention that individual selectionists are a bit doctrinaire and do not advance their cause by being able to literally explain everything with a just-so story.

More later. And yeah, I'll read the critiques. I know I'm going to get a head-ache over this though, that's one thing I loved about individual or gene level selection, the math ended up to be a bit knarly, but the axioms were clean and easy to grasp. Group selection just knots everything up again.

[1] What that long sentence-irony, no?-basically means is that the authors could have reduced 5 pages here and there to half a page of equations with explanatory captions. But for obvious reasons-this is aimed partially at a popular audience-they did both. If you are math-friendly, you will know when you see an equation to just skim over the text.

[2] One amusing example-if you have group A and group B, with A being more fit than B because A behaves "altruistically" within the group in competition with B, one could say it is group selection, no? Well in the discussion of Myxomatosis, the rabbit-killing virus in Australia in the 1970s, this sort of thing seemed to be happening-virus populations that were less virulent as a whole were spreading. But Wilson recounts with rage that some asserted that the viruses that infected any one rabbit were clones of each other, and so could be thought of as one individual, ergo, it was individual selection (or more properly, gene selection)!

Posted by razib at 03:40 AM

I thought I’d continue on with the discussion from the “groupies” post here.

Troy, you make a good point about the correlation between the genotype and the phenotype. And I was thinking that it might be phenotypes and behaviors that are selected, and only by extension, the genotype. It is possible to code the same or similar phenotypes with different genes – just as you can write two different programs that do the same thing. It’s likely that selection operates at many levels and on many things. So talking of “the” object of selection is, I think, misleading.

And though I sort of understand your intuition about “cultural code,” it’s not quite clear what sort of medium codes cultural information. DNA contains the genetic code, and it could be argued that language, customs, government and other institutions transmit the cultural code. This is probably taken up in “memetics”, but that still has a long way to go to become anything resembling science. Also, interesting idea about information being the (an) object of selection…

David, I think we’re in basic agreement – the differences being mostly a matter of approach. I understand that you’re approaching the idea of group selection from a rigorous, reductionist approach and would like to see the idea developed more coherently. Are you a biologist? I take a more pragmatic approach and view from a common sensical position an idea that seems to be obvious. I’m willing to forfeit some rigor and precision to utilize the idea of group selection. I think this is legit. If a biologist were asked to model a cell from the atom, or even strings, before he can investigate them, he wouldn’t get very far.

Some points about the politics of group selection: anyone thinking that group selection gives support to left wing ideology is seriously deluded. As I’ve heard said, it turns murder into genocide. But there are also objections to the idea of group selection from libertarians, who basically have a post modernist attitude toward society, thinking society is somehow not real. Personally my politics is generically liberal, with a strong emphasis on the importance of the individual, but also recognizing the nebulous concepts of public good, general welfare and the will of the people – all group notions.

Onto your factual points: True, most animals do not live indefinitely in one group, but they do live indefinitely in a group. Sure there is the occasional lone wolf, but most spend there time in one pack or another. This means their behavior is adapted to being a pack animal. And a pack can remain a single entity even as the individuals within the pack come and go, die and are born.

As far as primitive humans practicing exogamy – true. But then again, a group of exogamous tribes can themselves form an endogamous group. Jensen described race as a fuzzy set, which is really the mathematical entity needed to model group selection. I wonder how extensively biologists are using that concept in their modeling (I haven’t yet read the Wilson book to see if it’s in there. If it isn’t, biologists really need to take a trek over to the math department.) Groups can be constructed any which way, from an individual, which from the genes perspective is a group, to kin, to packs, bands, colonies on up to species and beyond to Gaia. All are collections of genes. As the groups expand, the selective force of the group acting on the individual and the genes diminish. Probably something like the inverse square law.

I personally find all this quite thought provoking – you really do run up against intractable philosophical conundrums like the nature of mind, which is why you really can’t afford to skim over those sections – it gives context to the data. I’ll try to read up and comment more, later.

Real life, again…

Posted by: justapolak at April 7, 2003 08:11 PM