Join 3,572 readers in helping fund MetaFilter (Hide)


The False Allure of Group Selection
June 25, 2012 12:57 PM   Subscribe

Steven Pinker on the "False Allure of Group Selection", with comments by Daniel Dannett, Stewart Brand, and others. Richard Dawkins's take on group selection. Jerry Coyne's take.
posted by AceRock (55 comments total) 29 users marked this as a favorite

 
Eliezer Yudkowsky's "Friendly AI" deals with group selection as well. I've been seeing it come up a lot recently.
posted by Algebra at 1:06 PM on June 25, 2012 [1 favorite]


How could you possibly bring this up and not mention EO Wilson's (terribly misguided) publications that sparked this whole thing?
posted by Buckt at 1:14 PM on June 25, 2012 [1 favorite]


This is core of the argument, from both Pinker and Dawkins:
Now, no one "owns" the concept of natural selection, nor can anyone police the use of the term. But its explanatory power, it seems to me, is so distinctive and important that it should not be diluted by metaphorical, poetic, fuzzy, or allusive extensions that only serve to obscure how profound the genuine version of the mechanism really is.
It's getting a bit boring to argue about, but yes it is an immensely reductionist and myopic way of looking at the world. Genes don't literally replicate themselves either, once they're understood as part of a much larger system which mediates their use. Using natural selection and evolution as a lens through which to view the way many/all systems change over time is actually quite useful, the gene-absolutists be damned
posted by crayz at 1:15 PM on June 25, 2012 [3 favorites]


Wilson's NYTimes article supporting multilevel selection.
posted by Jpfed at 1:21 PM on June 25, 2012


It's a little strange to see EO Wilson, who I've often felt is too reductionist in his explanations of human behaviors, being critiqued for apparently not being reductionist enough in this approach.
posted by Panjandrum at 1:36 PM on June 25, 2012 [2 favorites]


Makes sense to me. What was that experiment where two groups of kids were dropped in an abandoned campsite and they almost instantly developed inter-group rivalry what escalated almost to the point of violence? There was an FPP about it.

We seem to instinctively form and associate with groups of that sort, even where there is little distinction. "I'm a Mac" VS "I'm a PC" comes to mind. Why are there so many arguments about something so meaningless?
posted by Ad hominem at 1:42 PM on June 25, 2012


Ad Hominem: that would be the Robber's Cave study, which I believe has less to say about grouping than it does about the manipulation of groups.
posted by rebent at 1:48 PM on June 25, 2012 [2 favorites]


I've been following this whole back-and-forth for a few weeks now and all I can say is that while Wilson might be half wrong about this, he certainly seems half right too. I'm also looking forward to hearing the perspectives of much younger scientists with less entrenched positions.
posted by kuujjuarapik at 1:51 PM on June 25, 2012 [1 favorite]


Why are there so many arguments about something so meaningless?

Maybe because it's not so meaningless. No one's saying that group dynamics don't exist, or that organisms don't benefit from being in groups. What Pinker is saying, and I agree, is that no meaningful selection seems to occur at the group level that can't be explained within the context of plain old ordinary individual selection. If that's not sexy enough for mavericky science folks, the burden is on them to prove their case.
posted by Fritz Langwedge at 1:51 PM on June 25, 2012 [6 favorites]


I like the inversion of altruism in this argument. Altruism is not something individuals have evolved to express on their own part. Instead it is an act individuals have evolved to coerce others into performing.

In other words, I don't help you because of the altruism that lives inside of me; I help you because I've been fooled by manipulations that are external to me. A loss for my genes, a victory for others'.

It's a thought worthy of consideration, regardless of whether or not it is ultimately, fundamentally true.
posted by jsturgill at 1:51 PM on June 25, 2012 [3 favorites]


When all you have is a hammer, when you've invested your life savings in a hammer factory, you point and laugh at the guy with the screwdriver. "Why, that's the dumbest looking hammer I've ever seen!" you say, giving a fist-bump to your mates who all have matching hammers and are co-investors. Pfft, as if it could produce the force necessary to hammer in those weird squiggly nails.
posted by fleacircus at 1:51 PM on June 25, 2012 [5 favorites]


Fritz Langwedge: "no meaningful selection seems to occur at the group level that can't be explained within the context of plain old ordinary individual selection."

this is true of a depressing number of subjects. In my experience, too many people tack on higher-level processes when things can be explained by more elegant, basic principles. Happens all the time in my field, but "basic" means "simple" which bothers people who think they are special and complicated.
posted by rebent at 1:57 PM on June 25, 2012 [6 favorites]


when you've invested your life savings in a hammer factory, you point and laugh at the guy with the screwdriver.

Yeah. If they'd only listened to Lysenko.

Oh wait, they did.
posted by Fritz Langwedge at 1:58 PM on June 25, 2012 [3 favorites]


you point and laugh at the guy with the screwdriver

Oh, yeah? Well, let me introduce you to my "screw you driver."
posted by It's Raining Florence Henderson at 2:04 PM on June 25, 2012


*not actually aimed at fleacircus, in case that wasn't obvious
posted by It's Raining Florence Henderson at 2:19 PM on June 25, 2012


When all you have is a hammer, when you've invested your life savings in a hammer factory, you point and laugh at the guy with the screwdriver. "Why, that's the dumbest looking hammer I've ever seen!" you say, giving a fist-bump to your mates who all have matching hammers and are co-investors. Pfft, as if it could produce the force necessary to hammer in those weird squiggly nails.

Love it. Going to totally steal it and save it for when I see a debate where this actually describes what's going on.
posted by AceRock at 2:32 PM on June 25, 2012 [8 favorites]


It has been a decade since I read much evolutionary theory, but back then group selection was viewed as a gross error, and the arguments seemed wholly convincing to me. For anyone who is more in touch with the present level of the debate, is there any respected authority except Wilson who genuinely believes in group selection?
posted by Touchstone at 2:35 PM on June 25, 2012


I love the "simple model needlessly complicated" putdown Pinker has. Just to be clear, though, the conceptual argument is much less important than the empirical one. If his last section is right, there's no EVIDENCE of real genetic imperatives to sacrifice, only reciprocal cooperation (and rule-following punishment) when reputation matters, except in hive species. That seems to settle the matter.
posted by anotherpanacea at 2:39 PM on June 25, 2012 [1 favorite]




It has been a decade since I read much evolutionary theory, but back then group selection was viewed as a gross error, and the arguments seemed wholly convincing to me. For anyone who is more in touch with the present level of the debate, is there any respected authority except Wilson who genuinely believes in group selection?
posted by Touchstone at 2:35 PM on June 25 [+] [!]


I can tell you that Steven Pinker should be approached only with great caution and suspicion.
posted by Stagger Lee at 2:49 PM on June 25, 2012 [2 favorites]




I love the "simple model needlessly complicated" putdown Pinker has.

I'll give the guy one thing, he's really good at getting publicity. Considering the quality of his work there's no good reason at all that I should recognize his name so quickly.
posted by Stagger Lee at 2:50 PM on June 25, 2012


I can think of lots of Harvard professors who did some really great original research at the start and are now mostly in the business as popularizers and public intellectuals. E O Wilson is another example. How much of his last book is based on his own original research? Looks like 0% to me.
posted by anotherpanacea at 2:55 PM on June 25, 2012 [1 favorite]


Touchstone - I really don't think so. There are apparently (I haven't read them) some microbial experiments that show group selection is possible, though. I think it's less dogmatically ruled-out than it used to be (honestly, Dawkins might be a bit behind the times), but there is absolutely no way that Wilson is right. He thinks that kin selection/inclusive fitness does.not.exist., and that group selection explains everything. This can't be true. Group selection may explain some small portion of things, though (although Wilson sure as hell hasn't demonstrated this possibility, either, so don't hold your breath).

My expertise is actually in ants, largely because of the stuff EO Wilson wrote. I'm spending my life doing ant research. He's a great man. And I have a lot of sympathy for where he's coming from, having learned about the ants. Hamilton's rule, the simple model of kin selection, doesn't actually seem to be true very often. There are all of these cases in social insects (multiple queen colonies of fire ants come to mind, termites in general, multiple matings in army ants or honey bees) where the relatedness between individuals within a society can be extremely low. In argentine ants, which don't even form colonies, it's always zero. No relatedness. Yet they're extremely successful, social animals. Shouldn't relatedness, and thus kin selection, be a prerequisite for a society that is maintained by natural selection? Clearly it is not. And it doesn't make sense. In Thomas Khun's The Structure of Scientific Revolutions, he talks about theories that describe all of the data to a point, start to become insufficient, for a period everything is in chaos, and then a new theory comes along to replace it. We're in chaos now with the way that kin selection looks right now in ant societies and I honestly could see a minor revolution take place.

BUT - Wilson's wrong. When Einstein replaced Newton, he made the old physics a special case of the new physics. Such it must be with kin selection. We know it works a lot of the time - it makes valid predictions in many simple social insect societies, in the way that birds or primates deal with each other, and in some damn near irrefutable microbial studies. It definitely, definitely is a real thing. Wilson is saying that it is never accurate. I think he's right that we need something new, but that new thing (perhaps a group selection - kin selection complex) will necessarily have to include the cases where kin selection already works today - Wilson's group selection doesn't predict what kin selection predicts, and it is wrong in all of the places where kin selection is right.

If you'd like a less vitriolic treatment of these things, I can point you to this PDF, published in a lovely journal all about ants, which frankly discusses some of the difficulties. This came out very briefly before Wilson's Nature paper sparked all of the controversy, mind you, so it's untainted by the politics. Note that these authors accept many of the difficulties that Wilson puts forward, but do not come to the same non sequitur of group selection.
posted by Buckt at 2:56 PM on June 25, 2012 [31 favorites]


There is no such thing as a "group," though it may be a useful abstraction at times.

Group selection sounds less like science and more like sentimentality of a kind that pleads for a naturalistically fallacious argument for us all to just get along with each other.
posted by univac at 2:58 PM on June 25, 2012 [1 favorite]


One point to note about Pinker's stance (which is, as usual, very well argued) is his a priori commitment to the origin of behavior within an individual. He speaks of psychological "facts". Psychology shares his commitment. This edifice is, however, crumbling. He also shares the necessary associated materialism that regards mechanistic explanation as trumping all others. The "facts" of physics don't trump all others, but they can play a blinder when the opposition is psychology.

This will probably appear as a severe case of scientistic myopia in the future. Explanations, all explanations, are bounded by their domain. Mixing notional facts about behavior, psychological predicates, and DNA is bound to result in a conceptual clusterfuck. As it does.
posted by stonepharisee at 3:16 PM on June 25, 2012 [1 favorite]


Pinker: You are all individuals.
Wilson: I'm not.
posted by No Robots at 3:19 PM on June 25, 2012 [4 favorites]


I can tell you that Steven Pinker should be approached only with great caution and suspicion.

Why? I mean, most things should be approached with some caution, but why so much for Pinker?
posted by liquidindian at 3:23 PM on June 25, 2012


Thanks Buckt. That was a really illuminating and helpful reply, as was the pdf link you provided.
posted by Touchstone at 3:39 PM on June 25, 2012


Buckt, I have a question if you have the time. The game theoretic explanations of the evolution of altruistic behaviour among unrelated actors rely to some extent on what might be called self policing. Individuals that "cheat" repeatedly are no longer cooperated with, and therefore lose out in the long run relative to more pro-social actors. (I appreciate this is a gross simplification)

So my question is, do any ant species self police in this way, ie do they show behaviour which differentiates between individual ants on the basis of their previous history of cooperation?
posted by Touchstone at 4:04 PM on June 25, 2012


For everyone taking the "Makes sense/seems logical to me" line on this I think A few cautionary words from Dawkins himself might be in order:

In the nicest possible way and with great respect, could I make two suggestions to would-be commenters, based on past experience when this topic has come up-

Please pause before offering your own common sense view. There are topics in science, of which this is one, where common sense is not a good guide. If it were, professional biologists would not have been arguing about it for five decades. There is a large back literature in which the likelihood is strong that whatever commonsense view you put forward has already been proposed and exhaustively discussed. As an analogy, common sense is notoriously misleading when we try to understand quantum mechanics. If you could do physics by common sense, we wouldn't need physicists. To a lesser extent, something like the same thing applies here.

Please note that Homo sapiens is a very peculiar species and probably not the best testbed for the theories under discussion. That is not to say that the theories under discussion will forever be irrelevant to human affairs. But the argument at hand is sufficiently difficult that it is worth trying to understand it and solve it, at least to begin with, without the additional complications that arise with human culture. It's not a bad idea to think about lions or ants or acorn woodpeckers as your model animal when trying to get to grips with these evolutionary theories.

Richard

posted by Decani at 4:13 PM on June 25, 2012 [6 favorites]


This stuff drives me nuts, since group selection is so obviously at work in both humans and other species. Hell, individual human beings are just groups (communities) of genes that have evolved to function in a mutual complementary fashion since the survival of each gene is dependent on the human individual who carries it. Similarly, human individuals have evolved to function in a mutually cooperative fashion since no individual would or could survive without the group.

The Nowak model in the Nowak/Wilson Nature paper sees kin selection as a special case of a more general theory of multilevel selection. Evolution works on the genetic, individual and the group level, but it's an empirical question how those targets of evolution interact and which is most determinative. All three are operative in human groups, which is why you see the characteristically human combination of opportunistic and cooperative instincts. Kin selection types try to see all of this through the lens of kin selection, which is both unnecessary and reductionist. The kin selection paradigm is a metaphor (I would die for two brothers and four cousins!) and a muddled one at that. The central idea that behaviors can't last unless they produce genetic benefits is true enough, the notion that this is only true for cooperation with family members is what's off.

There is nothing mystical about multi-level selection that includes group selection. Anytime the genes of one member of the group help form the environment that determines the survival of other group members, this is an interaction should be taken into account. There are all kinds of real world examples -- for a source, see this article on chicken breeding. For a more general model, see this piece on indirect genetic effects.

I would give EO Wilson and Martin Nowak -- a working biologist and a brilliant mathemetician -- a whole lot more creedence than two tired old science popularizers like Dawkins and Pinker. Group selection is obvious enough that Darwin believed in it. My general lack of faith in Pinker was reinforced by this article, which is bizarrely hand-wavey. People use family metaphors to refer to the total strangers that they sacrifice for, hence kin selection causes altruism? Umm, ok, right.
posted by zipadee at 4:36 PM on June 25, 2012 [3 favorites]


There is no such thing as a "group," though it may be a useful abstraction at times.
Is this a deliberate play on Thatcher's "there's no such thing as society"? If not, if you really think that the only scientific view of human nature is one that valorizes (and normalizes) selfish behavior and that anything else is "political" or sentimental twaddle, then the parallels here should give you pause.

For the evolutionary scientists in the house: what's the take on Herbert Gintis's comments on the piece? He seems to argue pretty convincingly for how morality and cooperative behavior are eminently explainable in evolutionary terms.

I'll also confess that I'm having trouble understanding where the battle lines are drawn here. Where does Gintis fit? I'd always seen evo psych and sociobiology as much of a muchness, and so was surprised that Dawkins and Pinker et al would be taking on Wilson with such gusto. Evidently I was missing a great deal of the subtleties in the debate.

And is this related to Joan Roughgarden's championing of social selection over sexual selection? Or is that battle taking place somewhere else entirely? (Is it in fact a battle? Or just one eccentric warrior calling a challenge outside the indifferent walls of mainstream evolutionary science?)
posted by col_pogo at 4:43 PM on June 25, 2012


Touchstone, fun question! Firstly, I'll do that whole disclaimer dance - this debate is something I casually read about after-hours, have little investment in, and thus may not have bothered to a requisite standard of intellectual rigor.

Your game theory statement is solid. I get the theory, I don't know if birds and mammals in nature are known to do sort of thing, seems reasonably possible. In ants, individual recognition isn't such a thing. At best, workers can recognize the caste and probably age (which are typically correlated) of a nestmate - but I would guess that two workers of the same caste and age would be identical. So my answer would be I just don't think that the machinery is there for ants to recognize individual cheaters and punish them for it.

That's ok though, because individual cheaters must have some genetic basis for their cheating. Often times, they produce some detectable by-product that lets other individuals know they are "the cheating sort." Thus, this sort of thing certainly comes up over evolutionary time, and can lead to altruism among unrelated individuals. The stalk-forming slime molds come to mind. If I recall correctly, these little groups amoebae sometimes form fungus-like stalks that allow them to disperse out of inhospitable environments. Unrelated individuals aggregate in order to form these stalks, and those that randomly show up in one place end up giving their lives to turn their bodies into the dead rigid structure of the stalk, while those in another place end up at the top of the stalk and get to disperse and survive. It's a random process. Apparently, there is a gene which causes some amoebae to only ever end up at the top of the stalk and thus cheat the system. These individuals actually "smell" different somehow, and the stalk-formers refuse to cooperate with them. You can imagine this unfolding over evolutionary time in just the way you described - first stalk formers benefit, then cheaters benefit, then stalk formers who can detect the cheaters start refusing to cooperate.

In ants, it's all about reproduction. Workers in a colony, particularly in "higher" ants, are typically sterile while their mother, the queen, does all of the reproduction. Sometimes workers, all of whom are sisters, are capable of laying eggs that hatch into males (never females). I'd be happy to explain the genetics in another post (they are much of why I went into ants), but it should suffice to say this: if a queen mates only once, then the workers shouldn't mind allowing other workers to lay eggs which develop into males. If a queen mates more than once, the workers should prefer for all male ants coming out of the colony to be from the queen and not from other workers. This "worker policing" is well understood and has been demonstrated, as predicted, many times. It goes along with the evidence for kin selection that I referenced earlier. So cheats aren't allowed.

But a big part of the problem with kin selection is that often cheaters get away with it much more than they should. My favorite example (and perhaps, unbiased opinion here, one of the coolest things in science) is phenomenon of polygyne (multiple queen) and monogyne (single queen) colonies existing in the fire ant Solenopsis invicta. There's all kinds of crazy going on in this system (which, despite huge phenotypic differences in the varieties, is known to be controlled by a single gene). One of the things is rampant cheating and exploitation in the multiple queen colonies. You have all of these unrelated queens living together producing all of these unrelated workers. In a single queen colony, most eggs that get laid end up as sterile workers and a few end up as reproductive males and females. In multiple queen colonies, it turns out that some queens lay eggs that ALL become reproductives (taking a huge share of fitness), while others lay eggs that ALL become sterile workers (no fitness at all!). We know how it happens, too, and it's genetic - some queens produce lots of pheromones that suppress the other queens' abilities to lay eggs that develop into reproductives. How does natural selection let this happen!? Clearly, the queens which produce only sterile workers don't make it into the next generation, so how, then, do they survive? Why hasn't an equilibrium been struck long ago? Or, to put it in the terms of your question, why don't the queens who are getting manipulated not to produce reproductives evolve to detect this pheromone and immediately assassinate any nestmate who produces it? Eh. I don't really know. You can wave you hand and say "Superorganism!" "Ecology!" "Manipulation!" "Evolutionary dead end!", but I honestly think that no one really understands it yet. Maybe a deeper analysis of genetics and costs/benefits will show that it actually isn't so bad as it looks. I kinda doubt it though - I think one of those hand-waving explanations will have to be formalized and start making sense.

Anyway. It's interesting. Zipadee, your response came up just as I was about to post this and I do have to go. You may be right, but please know that the Nowak and Wilson paper was unanimously discarded by everyone reputable who read it (there were something like 200 scientists who signed on to refute it point-by-point). A brilliant mathematician does not a biologist make. Also note that Wilson's long-time collaborator (and scientifically just as influential), Bert Holldobler, has gone on the record saying he does not agree with any of this. This thing has been hugely oversimplified, by me and definitely the popularizers like Dawkins, but there are some extremely significant flaws in the Wilson argument which he hasn't offered any data to resolve. That doesn't mean he's wrong, but sure hasn't convinced many scientists that he's right.
posted by Buckt at 4:45 PM on June 25, 2012 [16 favorites]


OK, consider me pre-chastised by the Dawkins comment. But still:

The examples of purported group selection are things like altruism or religion, both of which play well into Pinker's argument that individual selection is all that's in play. But what about diversity? Surely there are tradeoffs where the fastest cannot also be the strongest, or the introverts cannot also be extroverts. Isn't there a case to be made that groups do better with a mix of cautious and reckless, worker and soldier, or male and female? It's harder to see how individual selection would give rise to that kind of group strength.
posted by nixt at 4:45 PM on June 25, 2012


My general lack of faith in Pinker was reinforced by this article, which is bizarrely hand-wavey

I don't see it, can you point out the handwavey?
posted by Rocket Surgeon at 4:54 PM on June 25, 2012


I would give EO Wilson and Martin Nowak -- a working biologist and a brilliant mathemetician -- a whole lot more creedence than two tired old science popularizers like Dawkins and Pinker.

Joke's on you. Dawkins doesn't even like John Fogerty.
posted by adamdschneider at 5:01 PM on June 25, 2012 [2 favorites]


This debate is not about science, it's about whether people are nice or mean. Answer: we are both. As for group selection, has the hypothesis been tested?
posted by univac at 5:37 PM on June 25, 2012


Now, no one "owns" the concept of natural selection, nor can anyone police the use of the term. But its explanatory power, it seems to me, is so distinctive and important that it should not be diluted by metaphorical, poetic, fuzzy, or allusive extensions that only serve to obscure how profound the genuine version of the mechanism really is.
This is an interesting caveat that applies equally well to memetics and evolutionary psychology. Our understanding of the mechanism has profound explanatory power because we have a extremely robust understanding of the constraints imposed on genes by the molecules that replicate them and move them around. Allusive extensions to things like behavior, for which there's little consensus on how to define, strike me as premature.

nixt: Genetic diversity is maintained within a population if the population is in a heterogeneous or changing environment. Complicating this is the fact that most genetic traits are a range of values influenced by multiple genes and their interaction with environmental factors. You're more likely to see shifts in the mean over time than strong selection for extremes of the range.

But the show-stopper for things like altruism and religion (and many other behavioral traits in critters with a culture) is that you can't talk about selection at all without a good quantitative model of the inheritance of the trait. And while we might have good models for defining the behavior of social insects, we don't have a model of behavior that allows us to make meaningful comparisons between human and chimp behavior without running into the Clever Hans problem, much less a good understanding of the underlying genetics.
posted by CBrachyrhynchos at 5:40 PM on June 25, 2012 [2 favorites]


And zipadee, because I'm chatty tonight, let me add two more things.

One, regarding "Darwin believed it," you might be wrong and even if not, it's irrelevant. Darwin didn't know about genes, so the levels of selection had not been clarified and evolution was then much less precise than it is now. I also read a scholarly (non pop-sci) book, The Ant and the Peacock, which argued - among other things - that Darwin saw the issues with group selection in ways that his contemporaries did not. That book's from the "group selection is impossible" era though, so it might be wrong.

Secondly, I was thinking more about the popularizers thing and I found it ironic. There are published statements in the literature from many working, influential biologists who do not support Wilson's view. I posted a PDF to one. It's too be expected that those arguments didn't show up on metafilter; they're not written for the public. The fact that Dawkins and Pinker are eloquently recapitulating the consensus to a wider audience just means that they're serving their role in society; you probaby wouldn't even hear of it otherwise.
posted by Buckt at 5:40 PM on June 25, 2012 [2 favorites]


Zipadee, your response came up just as I was about to post this and I do have to go. You may be right, but please know that the Nowak and Wilson paper was unanimously discarded by everyone reputable who read it (there were something like 200 scientists who signed on to refute it point-by-point)

So the 137 scientists who wrote to Nature constitute 'everyone reputable' with anything to say on the topic? Evolutionary bio must be a small discipline. Anyway, as I understand it the conflict between NWT and their critics is not a conflict about the truth or falsehood of a fact, but about the utility or lack of utility of a particular modeling method relative to another modeling method. It is natural enough that scientists get tribal when their modeling methods get questioned; such disputes are notoriously difficult to adjudicate and become very political very fast. My understanding after reading it and the commentary on it was that NWT overstated their case in the rhetorical part of their article (as opposed to the actual model, which stands up up), but they had an interesting case -- and the inclusive fitness types habitually overstate their case in statements like 'group selection does not exist'.

Some to my understanding reputable types sympathetic to NWT who give good discussions of the controversy that I found helpful -- http://www.mendeley.com/research/sociobiology-turmoil-again/#, here, here, and here.
posted by zipadee at 5:47 PM on June 25, 2012 [1 favorite]


Whoops, here is the first Gadagkar link again.
posted by zipadee at 5:48 PM on June 25, 2012


i> The fact that Dawkins and Pinker are eloquently recapitulating the consensus

I have no beef with popularizers. But I don't see either Dawkins or Pinker as eloquently recapitulating a consensus. Pinker gives some kind of personal redefinition of group selection that has nothing to do with what biologists who argue for multi-level selection are talking about. As far as I can tell he seems to be saying that no action which can be hypothesized to benefit an individual in any case can be the product of group selection, and furthermore that when an action does clearly benefit the group at the expense of the individual it's not group selection unless the individual does it automatically, like an ant. I just don't see how that is relevant to the discussion. As for Dawkins, he has been recapitulating Hamilton for several decades now; it was novel in the 70s but not today. The reply by Gintis (who blows apart Pinker on reciprocity) is good. So is the reply by Queller, who makes the point much more simply than I did that the dispute here is over two modeling methods for the same evolutionary process that in mostly get you to the same place, although I think there is somewhat more at stake than he does.

The irony is that I bet everyone in this thread leaves a substantial tip for an anonymous stranger every time they eat out.
posted by zipadee at 6:16 PM on June 25, 2012


zipadee,

I'm sure the model was fine. The question always is whether the model is a good representation of reality - the crap in crap out thing. Other parts of that paper were questionable, as you acknowledge: it's contested whether kin selection is a "special case" of natural selection, or whether the uncanny number of evolutions of eusociality in the haplodiploid aculeate hymenoptera is below the level of significance, as Nowak et al also say. There are actually haplodiploid thrips which also evolved eusociality. Clearly, there's something to it. Given many of Wilson's further statements have concerned these (contested) matters rather than the methodology of his mathematician's model, I think it's safe to say that the model is mostly just a crutch for an argument Wilson already was looking to promote.

I've probably overstated my case, though. I'll put it this way: I haven't read any reasonable defense of that paper or Wilson's subsequent discussions about it in the published literature. I know a lot of ant biologists who think Wilson's been on the wrong track since the late 90s. I'm making an argument from authority or popularity or both and it's entirely possible that I'm wrong. As I said before, this stuff doesn't much interest me and I don't know a ton about it. What I know points to my thesis from above, which was: Wilson's probably wrong, but group selection may exist, and kin selection definitely has problems.

I really should quit at this for the night, though, so I'll have to read your articles and post back later. Thanks for linking them! I'd also like to point out that I completely agreed with your statement that:

Evolution works on the genetic, individual and the group level, but it's an empirical question how those targets of evolution interact and which is most determinative.

Although I'm not so sure of the follow up that it's known to function on all three levels in humans -
I just haven't seen any proof. I haven't looked either.

Anyway, this has been fun, I should go worry about the science things that actually have some bearing on my future.
posted by Buckt at 6:26 PM on June 25, 2012 [1 favorite]


And I just caught your second response. You might be right - both Dawkins and Pinker are doing a bit more than just recapitulating. I'll read the articles you posted soon, scout's honor.
posted by Buckt at 6:28 PM on June 25, 2012


The comments by Herbert Gintis on the piece are among the silliest things I've ever read online:

Yet people do vote, and many do expend time and energy in forming political opinions. This behavior does not conform to the selfish gene model.

There simply aren't enough facepalms in cyberspace. For someone whose stated intent in the first sentence of his comments is to clear up misconceptions, Gintis doesn't demonstrate an understanding of the selfish-gene concept that's anything but a slew of misconceptions.
posted by Fritz Langwedge at 6:38 PM on June 25, 2012 [1 favorite]


Fritz Langwedge: "no meaningful selection seems to occur at the group level that can't be explained within the context of plain old ordinary individual selection."

this is true of a depressing number of subjects. In my experience, too many people tack on higher-level processes when things can be explained by more elegant, basic principles. Happens all the time in my field, but "basic" means "simple" which bothers people who think they are special and complicated.
posted by rebent at 9:57 PM on June 25


Ironically, that sort of reminds me of the Intelligent Design crowd. :-)
posted by Decani at 1:58 AM on June 26, 2012


Look, the two biggest confusions in this argument are

1) that it is about human beings and can be solved with common sense reasoning about the human groups we see around us. It's not. It's about a general biological pattern, repeated at all scales. There is a discussion somewhere in one of Dawkins' early, scientific books which offers a way of understanding these things which torpedoes much of his later rhetoric (btw, another pattern which recurs). But if you read David Sloan Wilson, who is I suppose the most devoted multi-level selection man, it's clear that he started a long way from humans in his reasoning. One of his crucial experiments involves flour mites.

2) Deep confusion about what "a unit of selection" might be. The question was actually cleared up (IMHO) by David Hull, in the Eighties. Natural selection is a two part process, involving replication and interaction. In genetic systems, the two functions are separated. Genes are replicated, but the interactions which determine which will be replicated happen (usually) among their products. Only in the case of "selfish DNA" are the replicators and interactors the same. The Dawkins discussion I mentioned above brings in the idea of a "bottleneck" -- whether all the genes in an organism are dependent on its collective success (the outcome of its interactions with the outside world) to survive. But this is itself of course an example of multi-level selection.

All of the Szathmary/Maynard Smith "major transitions in evolution" can be understood as adding another level of interaction into the mix: none of these becomes *the* unit of selection because there isn't such a thing. They become a level of interaction, which operates as well as all the pre-existing ones.

All this second part of the argument laid out by Novak at the Royal Society a couple of years ago. I have my notes but I think there must be a version of that talk online as well.
posted by alloneword at 3:24 AM on June 26, 2012 [1 favorite]


Ironically, that sort of reminds me of the Intelligent Design crowd.

Well, yeah. My bullshit meter buzzes long and loud when I hear people complain that "the generally accepted theory can't explain such-and-such," when it seems like it explains it just fine. I'm skeptical of dramatic accusations that "everyone's ignoring our evidence" when it could just be that the evidence isn't persuasive. If a "paradigm shift" is necessary before people acknowledge the validity of some idea, the idea probably isn't worthwhile; paradigm shifts occur because of the validity of the ideas, not the other way around.

And I'll admit that having people like Lynn Margulis on board is a strong indication that it's probably not the most sober, objective theory in town.
posted by Fritz Langwedge at 3:51 AM on June 26, 2012


Pinker and Dawkins are both pretty arrogant. Pinker tends to present his opponents' arguments in an exaggerated way designed to magnify the extent of his victory in demolishing them. Dawkins always gives the impression that he has not ever, for a moment, so much as considered the possibility that he might be wrong.*

But I've read this thread and a heap of the linked articles, and... I can't help agreeing with both of them almost completely. I'm not a biologist, but kin selection seems to be a pretty straightforward and logical idea even if sometimes it can be taken too far. The arguments for group selection I've read today, though, seem to consist either of examples of processes that seem likely to be reducible to selection at the kin, individual or gene level anyway, or vague hand-wavy appeals to the inherent goodness of humanity or something.

Gintis' response to Pinker is particular bad; E.O. Wilson's NYT blog article avoids these problems only by being so non-specific it's really hard to tell how and why he disagrees with anyone else.

At least some of the criticisms of kin selectionism seem to be based on the misunderstanding that it disregards all forms of cooperation other than with kin. But this isn't what it's about; rather, it makes the claim that an individual will tend to act so as maximise the replication of the individual's genes even if those genes are in other, related individuals. This doesn't rule out cooperation with unrelated individuals. It just means (to grossly oversimplify) that an individual will tend to cooperate with unrelated individuals only to the extent that it tends to benefit the replication of the first individual's genes.

This is consistent with the gene-centred view of evolution that Dawkins has been on about for decades. It also leaves plenty of scope for cooperation, and seems likely to produce results that are consistent with group selection - for example, a trait that harms the individual who has it but benefits the individual's group might proliferate if the benefit to the group allows each member of the group to outreproduce every non-member. But you don't need group selection to explain this.

So... what am I missing? What gap does group selection fill that individual/kin selection can't?

*Actually, I have a lot of time for Dawkins. But... try to tell me I'm being unfair.
posted by A Thousand Baited Hooks at 6:03 AM on June 26, 2012 [1 favorite]


Does Steven Pinker have an opinion on EVERYTHING? He seems to be the go-to guy for everything from linguistics to natural selection to education these days.
posted by Peach at 6:21 AM on June 26, 2012


Does Steven Pinker have an opinion on EVERYTHING? He seems to be the go-to guy for everything from linguistics to natural selection to education these days.

I don't know about your group, but my group selects for hottie science popularizers with long curly hair and dreamy eyes.
posted by condesita at 6:30 AM on June 26, 2012


Pinker tends to present his opponents' arguments in an exaggerated way designed to magnify the extent of his victory in demolishing them.

I'd say that The Blank Slate goes well beyond "exaggeration."
posted by CBrachyrhynchos at 6:33 AM on June 26, 2012


If theories are referring to religion as "altruism" then we can ignore them without missing anything important. Not Pinker it seems.
posted by Brian B. at 6:49 AM on June 26, 2012 [1 favorite]


I don't know about your group, but my group selects for hottie science popularizers with long curly hair and dreamy eyes.

I can adapt. Really.
posted by Fritz Langwedge at 6:53 AM on June 26, 2012


A Thousand Baited Hooks, I haven't seen anyone explain anything convincingly with group selection that kin selection can't explain. They have tried though, as is evident in Wilson's popular publications or in his introductory chapters to The Superorganism. I might not be smart enough to grasp these arguments, he might be right but not presenting his evidence in a convincing way, or he might just be wrong. The thing to keep in mind is that there are things that need to be explained. I do think Wilson's wrong about group selection, or at least his formulation of it, but that doesn't mean I think he's wrong in seeing a problem. When you're on the front lines looking at the ways ants treat each other, they do a lot of things that a strict interpretation of kin selection would never predict. It's all a bit fuzzier than I'm comfortable with, on both sides.

I don't have (or know of) any rigorous mathematical way to demonstrate this, but I think the best resolution to the problems is by invoking evolutionary constraints. For example, say that ants did evolve into social organisms because of kin selection and the magic 0.75 and all of that - it was down to relatedness and inclusive fitness. But once they were there, many ant societies developed such that the workers became obligately, irrevocably sterile. Once that happened, maybe the colony actually did become a "superorganism," a thing whose function is to produce more colonies, and there simply didn't exist the variation to reverse the sterility of workers. After going down that trail, maybe the superorganism benefits by having hundreds of queens or by having the queen mate multiple times. This goes against the genetic principles that allowed this system to evolve in the first place, but the superorganism still works (clearly: these creatures are taking over the world), so the gene cluster survives. It's not optimal for the workers, but perhaps that's okay... perhaps the workers don't have any say anymore because they've lost their ability to reproduce or to get any of their genetic information from someone other than the queen (who might be under selection pressure to mate multiple times, etc.). If variation arose that allowed the workers to mate again or which allowed the workers to prevent supernumerary queens, then maybe we'd expect them to do it, but from the current standpoint it's just impossible to get there?

I dunno. I've read that argument from a number of ant people and honestly I don't even know whether it falls into kin selection or group selection - the superorganism is clearly a group selectionist's concept. I like to think, though, that ultimately we're talking about the genes the queen bears, some of which are under pressure to make the workers revolt and some of which are under pressure to keep doing things like mating multiply. In that case, the selection pressure for multiple matings just has to be higher than the selection pressure for the workers getting their interests and then this is still resolved in a parsimonious gene/individual oriented way. But maybe the group selectionists claim this argument for themselves, or think they have an argument that could explain it better?

It's a difficult question - much more nuanced than any of these articles let on. Most of the people I know who'd be considered experts on the matter don't care to comment one way or the other. Actually, many (myself included) simply wouldn't care to research or stress over it too much because a lot of this is bordering on unfalsifiability. Hopefully, with genomics and a greater understanding how social animals have evolved, these things might become more clear.
posted by Buckt at 10:15 AM on June 26, 2012 [4 favorites]


Thanks Buckt for your excellent reply.

Some exceptionally good content in this thread, reminds me why I like this site.
posted by Touchstone at 3:46 PM on June 26, 2012


« Older Jay Smooth on why "don't feed the trolls" is not a...   |   Last year, Wired reported that... Newer »


This thread has been archived and is closed to new comments