Group selection bleg

What are the best arguments for and against explanations relying on the concept of group selection?  I would like to read more in this area, your suggestions are welcome, please leave them in the comments.  Thanks!


The concept of "social sorting" was created by Michael Philips, author of the blog Pro-Commerce.
It is described in his book Gods of Commerce, 1997.

Kin selection?

Without group selection there would not be multicellular organisms.

There are multicellular organisms, so group selection exists.


uh, what?


Not so fast. In what must count as the last word on the topic. Pinker explains: "If a person has innate traits that encourage him to contribute to the group's welfare and as a result contribute to his own welfare, group selection is unnecessary; individual selection in the context of group living is adequate." The same goes for cells (or genes) in multicellular organisms. "

"Some mathematical models of 'group selection' are really just individual selection in the context of groups. The modeler arbitrarily stipulates that the dividend in fitness that accrues to the individual from the fate of the group does not count as 'individual fitness...'"

"It's only when humans display traits that are disadvantageous to themselves while benefiting their group that group selection might have something to add. And this brings us to the familiar problem which led most evolutionary biologists to reject the idea of group selection in the 1960s. Except in the theoretically possible but empirically unlikely circumstance in which groups bud off new groups faster than their members have babies, any genetic tendency to risk life and limb that results in a net decrease in individual inclusive fitness will be relentlessly selected against."

Ian Maitland

While preposterously annointing Pinker's essay "the last word on the topic" you put extra words in his mouth! Pinker did not write "The same goes for cells (or genes) in multicellular organisms.” So much for "the last word," eh?

Anyway, what you apparently fail to comprehend is that humans (whether in groups or not) are reproductive units. Most of the individual cells in a multicellular organism are not, yet somehow their ancestors evolved to give up individual reproduction.

Most of them are subsumed under the 'levels of selection' arguments, which can range from single genes up to higher order taxa.

Yeah the use of the term 'group selection' is misleading

This is now called Multi Level Selection, see David Sloan Wilson's "Does Altrusim Exist."
Also good introductions on my "diablog" with Wilson
Here's the 1st & 2nd

Note also the prior view of selfish genes collaborating with genome-mates to compete under vehicle-level selection is being challenged by microbiome findings.
Your genome does not contain all the genes you need ("symbiosis isn't rare, it's the rule").

The concept of "the mood association fallacy" will come in handy.

There's essentially only one argument against, and it is fatal: the group is not the unit of selection. End of story.

Explain that to the Helots.

It is perfectly possible to have group selection effects, i.e., selection in favor of traits that harm the individual but benefit the group, without the group being the unit of selection. Indeed, in all convincing group selection explanations the gene (or custom, or meme) is the unit of selection. I would give an example, but unfortunately this box is too small.

You can click and drag the corner of the box to expand it. Also you can keep typing even if you get to the bottom of the box and it will make a scrollbar.

I would give an example, but unfortunately this box is too small.

Yeah, right. "I have discovered a truly marvelous proof of this, which this margin is too narrow to contain."

Exactly. At least scribble something in the margin.

"margin"al revolution,

it isn't? Why not? Sorting requires calculation, calculation takes time and resources.

Take a female xyzasaur. Her mate selection can be males with a huge fin......which 90% of the time makes for better genes or she can evolve a larger brain to figure out with more precision who the best mates are. Opting to select by group can be easier.

Starts and ends with Steven Pinker

This is the best place to start, but be sure to read the replies.

Oops... linked to this below.

I was going to suggest this as well.

I am no expert in the field, but this might be a good place to start:

I think (I'm an interested amateur) the problem is that it's usually unstable.

Imagine a world where competition happens between tribes, and individual humans' properties have less to do with their number of offspring than the tribe's fitness. That's the sort of environment where you can imagine group selection working.

And let's assume further than one of the tribes in this world somehow has every member of the tribe carrying a gene that confers an advantage on the tribe, at the expense of the individual. Now, this tribe has a significant advantage, and it may very well prosper as a result. Over time, maybe this tribe will split as it becomes too big, and its descendants will take over the world with their advantages.

So every individual in the tribe carries this tribe gene, which makes them sacrifice themselves (their own fitness) for that of the tribe. What happens if there are a few members of the tribe who don't carry that gene? They have a fitness advantage within the tribe over everyone else in the tribe. They can act more selfishly than the others, have an extra kid or two survive to adulthood, and still benefit from the more-effective tribe because of the other members' tribal gene. Over time, there is selection within the tribe for people to lose the tribal gene in favor of a gene that helps them get ahead personally. (Another way of thinking about this: Everyone playing the tribal gene strategy is not a Nash equilibrium.)

Now since we're talking genes instead of human choices, maybe this just won't come up--everyone is programmed with the tribe gene. (And we could be talking about "tribes" of termites or something that probably can't do much independent thinking.) But this requires almost no gene flow into the tribe. Every time the tribe defeats a foreign tribe, they must kill every child no matter how innocent and harmless, and every young woman no matter how beautiful. They mustn't allow a single outsider to join their tribe, by adoption or marriage or cuckoldry. They mustn't take slaves unless they sterilize them or somehow make absolutely sure there will never be a mixing of the slaves back into the tribe. Because outsiders have the normal selfish gene--they will not carry the programming to sacrifice for the tribe, and so they will have higher fitness within the tribe than everyone else.

+1. Good explanation.

Good explanation -- of course in the case of termites and bees all individuals are genetically identical, so the unit of selection can be the 'group', in a sense which means nothing for the idea of 'group selection'.

Incorrect. Worker bees are all female and are 75% related to one another due to the haplodiploid system of sex determination in Hymenoptera. Termite workers are of both sexes and share 50% of their genes between individuals.

Turchin's "Ultrasociety" for group selection.

Thanks for the correction

+1 this is the right explanation.

Nobody's disputing that groups are "selected", i.e that the reproductive rate of groups may differ, as multiple commenters above seem to believe. What was once believed, but is now long since throughly debunked, is that individuals with inferior reproductive rate may be selected if they increase the reproductive rate of the group. That situation is unstable as explained above.

This is very good.

This is good but like many toy evolution models it is too static. "Fitness Advantage" is multi dimensional and reacts differently in different environments. So imagine a 'group' fitness gene that gives advantages against rare but catastrophic environmental factors. Something like "having a tight family makes it more likely your whole family will survive the once every 100 year drought". Not having the gene means you can outcompete during the non drought years, but that your lineage gets devastated during the drought year. In order for the group advantage genes to survive you need the environmental hazard to be frequent enough to keep the non-group gene people from totally taking over. Infrequent wars might work also.

Are there actual genes that function like this? I have no idea. But there is nothing about the math or logic of the situation that makes it impossible.

If you are "fit" it just means you reproduce more than others - it's 0-dimensional.

However, conclusions about fitness are obviously dependent on timescale, from our perspective dinosaurs weren't all that fit after all.

I certainly wouldn't say it's impossible, just that for group selection to have a big impact, it seems like the situation must somehow prevent selection within the groups from overwhelming the selection between groups. (And again, I'm an amateur.)

It is argued that group selection cannot exist because of the free-rider problem, within a group, the free-rider will always have greater fitness than the individual who sacrifices himself for the good of the group. Even if a group started with very altruistic members, it would be pushed in the direction of selfishness. But humans can solve the free-rider problem, punishing selfish individuals. This is individual selection, but it is this individual selection that allows group selection to operate. Groups of people will vary in the average level of altruism within them, the more altruistic groups(as in within-group altrusim) will have more success than the less altruistic ones. This is group selection.

What explains the willingness of so many young men to volunteer to fight a war for their country, without even really knowing how the war started or what the "cause" is? I don't think its misdirected kin altruism. It's a combination of cowards getting thrown out of groups and more cowardly, less cooperative groups getting ran over by braver, more cooperative groups, compounded over the generations.

A country is a rather arbitrary group to try to reproduce.

If you're interested in group selection in relation with the evolution of altruism (as many, likely most people are), then this is a good piece:

Stuart West is one of the best evolutionary biologists working on this topic, and he brings a lot of direly missing clarity to the topic.

E.O. Wilson in his "Meaning of Human Existence" argues for it, although it has long been rejected by mainstream evolutionary biologists.

The main problem is cheating: how can a gene that benefits the group spread (increase in frequency), if it harms the individual (if it benefits the individual as well, you don't need group selection)? In general, there is no way to prevent cheating by individuals who benefit from group-promoting behavior of others, while not contributing. That's why it is believed to be a non-starter. Wilson's counter-argumant is that human societies have enforcement mechanisms, so groups that require and enforce altruistic behavior would benefit.

So in other words the argument against it is that societies haven't developed enforcement mechanisms which is laughable on its face.

Human societies can do a lot of things to enable cooperation that aren't so workable among wolves or crows or termites. But even there, evolution takes a long time. Your cultural norms and social enforcement mechanisms have to stick around for tens or hundreds of generations to have much effect. And human cheaters can also apply their minds to the problem of getting away with shirking, taking more than their share, sleeping with other men's' wives, and somehow showing up late for the battle with the rival tribe.

> [Group selection] has long been rejected by mainstream evolutionary biologists.

Group or multi-level selectionists are definitely a minority, but it's not like parapsychology or something. It's a legitimate debate.

"Wilson’s counter-argumant is that human societies have enforcement mechanisms" - yes, but someone has to be the enforcer which is a cost to the enforcer (resources, time, risk, opportunity cost etc.). This doesn't solve the problem, only begs the question how does altruistic enforcement (because the enforcer by virtue of enforcement effort is behaving altruistically) ITSELF appear, evolve, and remain .....

Read this:

I consider Boehm's theory of morality to be the best one available. Its based on social selection. Far superior to kin selection, group selection, and reciprocity.


Also this comment on the Gelman post

"A Cooperative Species: Human Reciprocity and Its Evolution" by Samuel Bowles & Herbert Gintis

Blurb: "In A Cooperative Species, economists Samuel Bowles and Herbert Gintis update their ideas on the evolutionary origins of altruism. Containing new data and analysis, their book is a sustained and detailed argument for how genes and culture have together shaped our ability to cooperate. . . . By presenting clear models that are tied tightly to empirically derived parameters, Bowles and Gintis encourage much-needed debate on the origins of human cooperation."--Peter Richerson, Nature

I found Less Wrong's writeup to be informative:

On the other side there's David Sloan Wilson's *Evolution For Everyone*.

I wonder if you might still see some group selection on traits that didn't harm the individual but helped the tribe somehow. That would at least not be actively selected against at the individual level.

I sort of visualize this as two different forces pushing a mass in different directions, but the individual selection force is many times as powerful as the group selection force. There may be cases where the group selection force still makes a difference, but probably almost never when it's pulling in the opposite direction as the individual selection force.

Imagining something that was mildly helpful at the individual level, but really helpful at the group level doesn't seem that hard.

It is if you really really really desperately deny group selection because you like the anti-religious aspects of Darwinism, but are worried about its implications for inter group differences.

You may want to check the blog of evolutionary geneticist Jerry Coyne (University of Chicago). He has posted several times on it (he's not a supporter):
tag: evolution

This is the best explanatory post from Jerry:

These are all his posts on group selection:

Elliot Sober is one of the leading philosophers of science working on group selection. He has an introductory book that discusses the matter:

There's also the Unto Others book he wrote with David Sloan Wilson:

My understanding of these disputes is not rigorous. But my sense is that David Sloan Wilson is right to prefer the phrase "multilevel selection" to "group selection" -- as the commentator Jag Bhalla indicates above.

A further remark: My sense of it -- tell me if this is wrong, anybody -- is that the theories are about the selection of genes, but that theories of what mechanisms operate to select genes consider different levels of constellations, including the organism, the nuclear family, kin, the band, and so on. Since theories at multiple levels have plausibility, we speak of the composite set of plausible mechanisms as multilevel selection.

A big problem in the debates over group selection is that different sides define the term differently. The classic models of group selection involve a population consisting of groups containing some mixture of altruists and non-altruists. Groups with more non-altruists are more likely to go extinct. It’s very hard to make group selection work in this case. It only succeeds if, on average, once a group gets “infected” with a selfish individual, it then goes extinct before it can send out even a single successful selfish migrant. (This is similar to how a disease won’t go epidemic if an average infected person infects less than one other person.) This is an extremely low level of migration (less than approximately one successful migrant per group lifetime), and pretty much everyone agrees this is not plausible for most species, certainly for almost any vertebrates, including H. sapiens.

On the other hand, many models of social evolution can be analyzed using a multi-level selection framework, where one component of selection is selection-within-groups and another component is selection-between-groups. MLS is an accounting method, NOT a separate theory. For example, kin selection can always be cast in MLS form (via the Price equation), as a balance between selection within families (against kin altruism), and selection between families (in favor of altruism). Without selection between families, kin selection doesn’t work. Some people want to say that any time the between-group component of selection affects a trait, you’ve got group selection. Some people think MLS, even if formally correct, is confusing or misleading regarding actual cause-and-effect, and the alternative inclusive fitness framework is the way to go.

How technical do you want to get? For the basic math. Models of Social Evolution, A Guide for the Perplexed” (Chapter 6) is good.

For a balanced philosophically minded treatment, there’s Evolution and the Levels of Selection, fairly technical.

For a more accessible online back-and-forth, you can find David Wilson (pro-GS) below, and follow some of the links to Coyne and Pinker (anti-GS).

Humans are unusual animals, and group selection, specifically cultural group selection, may be unusually important in our species. On cultural group selection (and some comments pro and con), see

Richerson , P., et. al. (2016). Cultural group selection plays an essential role in explaining human cooperation: a sketch of the evidence. Behavioral and Brain Sciences, 39, doi:10.1017/S0140525X1400106X, e30

Also worth checking out is Chapter 5 (The myth of self-interest and the science of cooperation) in Turchin’s War & Peace & War which relates cultural group selection and socially-enforced group solidarity to historical dynamics.

(ADVERTISEMENT) And here’s me on what happens when you add norms to kin selection theory.

There are separate issues that get muddled up under this label.

1. There's a semantic debate among mathematical evolutionary biologists about what the best fitness accounting system is (e.g., inclusive fitness as promoted by the Oxford crowd, or a pluralistic approach favored by most other mebs). This debate will seem totally stupid to economists. This debate shouldn't be confused with the empirical question of whether intergroup competition has shaped genetic or cultural evolution.

2. The difference between genetic vs. cultural evolution, and between cultural group selection and genetic group selection. Many researchers like Boyd and Richerson have argued against the importance of genetic group selection for humans but FOR the importance of intergroup competition shaping cultural evolution. Much modelling suggests that conditions that normally inhibit the importance of GGS in genetic systems are mitigated in cultural evolutionary systems--because of multiple stable equilibrium (think folk theorem with different populations stuck at different equilibrium)

3. Empirically, a lot of evidence suggests that intergroup competition has shaped cultural evolution (institutions, social norms, religions, etc.)

Two recent target articles in BBS with full commentaries and replies are the places to start

I would avoid the opinions of web-publishing non-experts, who have not contributed to the primary literature and don't understand cultural evolution.

This is a great reply. Dividing genetic and cultural is critical. I see a huge amount of evidence against group selection when we are talking about biology. Human cultures-seems like strong evidence for group selection effects. But I'm open to being wrong about all of this.

I'm far from any level of expertise in this area but I strongly agree that it is important to limit the scope of this question and go straight to the primary sources.

Cynically, I wonder if Tyler poses these questions just as a way of probing his readership. He knows to go to the primary sources.


It is indeed more useful to use the term "multi-level selection," even if "group selection" has a certain attraction. But it has the unfortunate element of drawing forth people who apply this in a simple-minded way to economics and get a knee-jerk individualistic pro-laissez faire reaction that leads to a certain amount of ideological ranting that we have seen some of here that is just misleading and unnecessary.

Also, sometimes genetic and cultural selection coincide, as was discussed in an excellent review essay by Joe Henrich published in 2004 in the Journal of Economic Behavior and Organization (which I edited at the time). As noted by some, the free rider problem is an issue, which gets discussed as the "greenbeard" problem in some of this literature and was discussed as such by Henrich in that now old but still good JEBO article.

Pretty good case against group selection here:

Also includes bibliography and follow-up comments.

In particular, commenters Dennett (anti), Haidt (pro), and Dawkins (anti) are well worth the read, along with Pinker's follow-up.

Richard Dawkins is a really remarkable writer. Crystalline.

In the end, I think Pinker ably demonstrates that, right now, Willem of Occam has no need of "group selection". It doesn't improve explanatory power.

Haidt's discussion of language is interesting, because this is one trait that doesn't really help an individual on their own, but it might help one clan outcompete others.

Tentatively anti group selection.

I also Recall Matt Ridely's The Origin of Virtue as being a good (anti) treatment of the subject.

Language doesn't help an individual? I would suggest that it does, as well as benefit the group.

The first guy to develop lactose tolerance has an advantage over others, but who does the first guy who develops language talk to?

I'm not sure if you guys are joking about language, but obviously that's not how it works.

Rather it's that the one guy who was slightly better at using sounds to convey meaning, reproduced slightly more, etc. etc. etc.

No, I get it. Emergent property. What good is half a wing or eyeball etc etc?

My point (and I think Haidt's) is that language is inherently social and groupish, unlike lactose tolerance.

There are a handful of circumstances where group selection is known or suspected to exist. It's far from clear whether these have any relevance to human behavior.

On group selection: Under artificial conditions, observed in chickens

and possibly in spiders, here's the cite to a critique that claims methodological error there:

IMO the best case studies for genetic basis of human behavior have come up inconclusive, and detecting group selection rather than individual selection is difficult even if phenotype and genetics are unambigous. Here's my choice for the best case: The most plausible recent testable idea about human behavior that I know of was higher intelligence among Ashkenazi hypothesized to be a product of positive selection, testable by looking for links with heritable diseases. The most recent test that I know about can't distinguish this possibility from genetic drift.

I'll repeat some of the above to bucket things into categories. I personally find this a fascinating topic.

1) Pinker's Edge essay mentioned above is good starting point, where he argues mainstream view, saying Hamilton inclusive fitness (cousin's will help each other due to relatedness) plus Robert Trivers reciprocal altruism/tit-for-tat game theory are enough framework to explain altruistic behavior where it exists. And arguing group selection is a confused idea.

2) Richard Dawkin's influence is clear in Pinker's essay of course. Dawkins is very hard core in his disdain for group selection. See his reaction on to same essay.

3) The strongest responses supporting idea (not mainstream, but not dismissed either) are by David Queller who argues group selection is an equivalent language to existing theory. So Price Equation can partition fitness in various ways, normal way of inclusive fitness/reciprocal altruism, or alternate languages. Similar to say, math behind Newton's laws can also be expressed equivalently as Hamiltonian. David Sloan Wilson is most prominent advocate of this view. So see their replies as well.

4) Worth noting the Dawkins influence is so strong, that assuming people may have evolved altruistic tendencies is unpopular. So Jonathan Haidt has been influenced by David Sloan Wilson, and I think for the better. It's quite possible to say David Sloan Wilson wrong, and underlying mechanics are described by Hamilton/Trivers fine, and to argue that humans in fact have had this happen. practice people align saying people are selfish like their genes. Dawkins argues strongly that's overselling his ideas, but anyway.

5) Martin Nowak, Corina Tarnita, EO Wilson paper got bashed pretty heavily. Ultimately I think the flaw was arguing not for equivalence, as in David Sloan Wilson, so group selection compatible and additive idea, but for overturning the past completely. So probably best to just avoid that one. But if curious like this version of commentary on it

6) there is a deep irony of course in the very strong position against group selection, as it's clear some of the key giant evolutionary steps in biology are a result of this. For example Eukarya are a merger of bacteria and archaea. And of course multicellular life is itself a group of cells. And colonies of insects, as well as (more debatable) human behavior. Now this is perfectly compatible with Hamilton/Trivers framework, but best to avoid the oversell Dawkins style advocacy of being overly against arguments of explaining behavior as selfish, even if Dawkins work was incredibly helpful in clearing away the intellectual crud.

7) Joseph Heinrich's book is the best recent one on culture-gene evolution, which ultimately is compatible of course with group selection, depending on your definition. The Secret of Our Success: How Culture Is Driving Human Evolution, Domesticating Our Species, and Making Us Smarter. So not sure of your motiviation for reading up on this topic, but if gene-culture, definitely read Heinrich.

I think maybe the key take away on this topic is that #6. A common mistake is to say selfish genes means selfish behavior. A misreading of Dawkins, at least that's Dawkins formal position, but in practice I think Dawkins is closer to that "misreading" than he claims.

And of course multicellular life is itself a group of cells. And colonies of insects, as well as (more debatable) human behavior. Now this is perfectly compatible with Hamilton/Trivers framework, but best to avoid the oversell Dawkins style advocacy of being overly against arguments of explaining behavior as selfish, even if Dawkins work was incredibly helpful in clearing away the intellectual crud.

This is misleading. It implies that Dawkins couldn't explain the behavior of multicellular organisms or of eusocial insects (ants, bees, etc.). But like you said, these behaviors are perfectly compatible with the Hamilton/Trivers framework, a framework that Dawkins has been promoting for 40 years.

This is simply kin-selection and it does not require group selection, even if group selection is "equivalent". In particular, this is not the same as the Newtonian vs Hamiltonian formulations of classical mechanics (Quantum Theory is different), two formulations that are exactly equivalent. Group selection, in contrast, only explains social behavior and is not appropriate for other traits.

*Evolution and the Levels of Selection* (2009) by Samir Okasha received the (apparently prestigious) Lakatos Award for its contribution to the philosophy of science. It was also praised in scientific periodicals. There is a blurb from the philosopher of biology Elliott Sober on the Amazon page saying that it "engages the details of mathematical models and at the same time connects those details with broader philosophical questions."

I made a go at it a few years ago and remember thinking it was very clear. It seems like the sort of book that would appeal to Tyler Cowen.

I came to the comments to recommend this book as well. I only ever got halfway through it, but found it very clear and convincing.

It's interesting and somewhat depressing to read these comments, which illustrate how little winnowing of ideas is taking place, even in this intelligent crowd. This person cites Pinker, that person sites West or Coyne. Evidently, anyone who has name recognition and writes something on the internet is an authority (Joe Henrich made the same comment).

You can't make sense of the group selection controversy without understanding the concept of equivalence--theories that invoke the same causal processes and differ only in their perspectives. There is a consensus in the peer-review literature that inclusive fitness theory, which was originally thought to explain the evolution of altruism without invoking group selection, is in fact equivalent and includes all of the elements of group selection within its own structure. W.D. Hamilton made this discovery in the 1970's after discovering the Price equation and modern-day advocates of inclusive fitness theory such as Stuart West and David Queller begin their articles by acknowledging equivalence.

Equivalent theories deserve to coexist to the extent that they provide different insights by virtue of their different perspectives. This requires the ability to understand each framework and to be able to translate between frameworks--like being fluent in two languages. Anyone who doesn't get equivalence is like a boorish person who can only speak one language and insists that other languages are wrong.

"Critics" of group selection who contribute importantly to the peer review literature, such as West and Queller, are very different than critics such as Steve Pinker, who does good work in his area but has never written a peer-review article on group selection and doesn't get the concept of equivalence. No one should read Pinker's essay without also reading the commentaries. I think you will find that the commentaries arguing against Pinker (by Joe and myself among others) are more informed than the commentaries agreeing with Pinker, which often sound like cheerleading (e.g., Jerry Coyne's commentary).

My latest book, Does Altruism Exist? Culture, Genes, and the Welfare of Others (Yale/Templeton 2015) provides a concise (150 pp) overview of multilevel selection and its implications, including a chapter on equivalence. I call it a "post-resolution" account because the group selection controversy is over as far as I am concerned. Over half a century is long enough!

This View of Life ( ) contains numerous articles by myself and others on multilevel selection. All of them are anchored in the peer reviewed literature.

I'll end with two more advanced points. First, two theories can be equivalent in some but not all respects. Inclusive fitness theory and multilevel selection theory are equivalent with respect to the simple core models, but the inclusive fitness theory formulation can't be extended to some cases of multilevel selection such as equilibrium selection, ecosystem selection, and human cultural evolution. See here for more:

Second, it is often said that the received wisdom about group selection is correct for genetic group selection and differs only for cultural group selection. Unfortunately, this statement can be found in the peer review literature. Nevertheless, it is demonstrably false, as even a passing acquaintance with the concept equivalence shows. Chris Boehm's thesis of reverse dominance for human evolution argues that our ancestors evolved mechanisms that suppressed the potential for disruptive selection within groups, so that between-group selection became the dominant evolutionary force. This is a multi-level selection argument. The dawning awareness of microbiomes--only since the turn of the 21st century--makes it clear that what we used to regard as straightforward selection among individuals is also selection among ecosystems composed of thousands of species and billions of individuals.

Multilevel selection is everywhere, once you know how to look for it. It is becoming indispensable for the study of human genetic and cultural evolution, as authors such as Boehm, Henrich, and Turchin stress. It is also indispensable for public policy formulation. See here for an example:


What do you think of Dawkins comment? Groups aren't replicators and groups (other than things like beehives) aren't reproducers.

Human cultural evolution is a kinda analogous thing, but it ain't natural selection.

William Wimsatt on “Reductionist Mythology”

"There is another problem with memes that has deep roots in the concept of a self-replicator, or replicator for short. This problem dates back to Richard Dawkins (1976) and the idea that strings of DNA are self-replicators. This is reductionist mythology. The idea is that there is an informational core that contains all of the information necessary to remake itself and, also, a larger vehicle or interactor, and that the informational core is somehow self-replicating. There is no such beast. Von Neumann’s self-reproducing automata had a machine and a tape that together made another machine + tape, but there were no tapes capable of remaking themselves. Similarly, over 200 enzymes are required for DNA replication. Of course, cells are self-reproducing as they go through a mitotic cell-cycle (well-documented by Moss, 2003), but this is a reproductive developmental cycle, not a copying event (see Grisemer, 200; Wimsatt & Grisemer, 2007). There are modules that have like copies of them made, but they don’t do it by themselves, and the capability to do so is always via a larger system."
Memetics Does Not Provide a Useful Way of Understanding Cultural Evolution A Developmental Perspective William C. Wimsatt see section 6

Interview with E. O. Wilson re "group" selection:

The problem is that even "groups" like beehives aren't really replicators or reproducers. The beehive can be conceived as an extended phenotype of the queen, who is an individual that actually reproduces.


See my comment earlier on your reply to Henrich. Genetic and cultural evolution can coincide. Even Dawkins admitted the possibility of multi-level selection in the second edition of his The Selfish Gene, but he tends to argue that it rarely happens, and in some public debates he reverts to a harder anti-multi-level selection position. But even he knows better.

David, thank you for this. I would be very grateful if you would tell us what you think of what I said above about why you use the term "multilevel selection."

You're basically a socialist and your multilevel theories are fronts for socialism, as anyone who has read your Evonomics blog knows.

It's a very partisan, silly blog.

^this guy read about Lysenko and thought, "This guy was right about almost everything, his only mistake was he tried to judge evolutionary biology by whether it was Marxist enough instead of the right way, whether it is capitalist enough"

It’s interesting and somewhat depressing to read these comments, which illustrate how little winnowing of ideas is taking place, even in this intelligent crowd.

It is somewhat depressing to see a highly educated expert in the field come out and imply that people who do not agree with him are stupid. This is not settled science. If science is ever settled. It is an argument. The fact that your own arguments are not yet 100% accepted is not proof that other people are dumb.

Especially it is probably a mistake to try to define your way out of the argument by asserting all the smart people accept your assumptions.

But thank you for the book plug.

He's not doing any of this stuff. He's pointing out that smart people who read blogs and pop-science books have been misled by the vehemence of e.g. Dawkins and Pinker into thinking that group selection is a hopelessly naive theory with no support among biologists. In fact, the Dawkins presentation of group selection is largely a strawman.

I don't even have an opinion on group selection, honestly. My guess would be it's probably true on some level but not all that important compared to gene- and organism- level selection, because I'm no expert and that seems to be roughly the mainstream expert view. But someone like Dawkins is unjustifiably extreme in his rejection of any role for it.


I know that yo know this, but I do find it unfortunate that Hamilton and Price continue to be credited with developing the mathematical conditions for multilevel selection to operate. He is only barely recognized, and I know you know about him (as does Joe Henrich), but it was the James F. Crow who first laid out the mathematical formalism in 1955 in "General theory of population genetics: Synthesis," Cold Spring Harbor Symposia on Quantitative Biology, 20, 54-59.

Of course multilevel selection exists. It's just a different level of abstraction. Sure enough, one does not absolutely need Newtonian equations to calculate trajectory of a bullet, as one does not need group selection to describe population replacement. But in both of these cases, the fine-grained approach is cumbersome and ultimately unproductive.

( I see a lot of selection bias in the comments. No pun intended. )

As far as I know:

1) In the past 30k and certainly in the past 10K years, the driving force in selection has not been mutation, but 'group domestication'. There are environmental selection pressures ( dairy, wheat, disease resistance) sure. But the primary difference has been, just like domesticated animals, domestication of mankind using the same techniques: maturity.

2) the primary change that has caused the major differences betwee the groups has been (a) rate of sexual maturity (b) degree of sexual maturity, (c) sexual dimorphism. Ergo, the primary differences are in endocrine expression. And from the data I've seen it's pretty obvious that the majority of the difference in maturity has been testosterone levels.

3) the secondary major change has been how aggressively some groups domesticated their members (east asia, western europe), or how groups have been unable to domesticate their members (africa and the middle east).

4) of the mutations that do occur, these appear to be relatively minor trade offs that are related to these differences (speed vs endurance).

So as far as I now, evolution by mutation, has been trivial compared to evolution by domestication.
This inverts the multi-level selection argument: most genetic mutiation and drift is 'noise' and domestication has been the primary influence (culture), with the secondary influence being territory.

The genome stores 'options' which we seem to express. I am not sure there is much of a case to be made for terribly meaninful genetic variation.

In my work (which seems to have pleasantly shocked the Africans), as far as I can tell, the major differences between regional groups is how successful they have been at eliminating the underclasses and redistributing reproduction upwards.

Unfortunately it's impolitic. But it is what it is.

Curt Doolittle
The Propertarian Institute
Kiev, Ukraine

I'd like to see what the more numerate evo bio/geneticist thinkers have to say. Like Razib or Greg Cochran.

Cochran is fascinating but does not seem to excel at understanding the limits of his own understanding, so he is only at his best when addressing moderately solvable problems. Whatever the concept of group selection means in the intellectual conceptual world Cochran and Razib inhabit, the actual real world working out of what the current roster of conscious biological organisms owes to group selection-type mechanisms - if anything - is likely to be, unless this world is orders of magnitude more simple than it appears, significantly beyond their ken. Bright and amusing guys, though, and Cochran in particular on his blog has developed a fairly unique MidWestern/vaudevillian shtick where he pretends to be a grumpy high-IQ curmudgeon. Goethe and Smith and Kolmogorov and Agassiz, just to name a few disparate names that are familiar, would clearly disagree with his stubbornly held priors on the subject under discussion, unless I misunderstand his sense of humor.

It might be worth reading Nowak, Tarnita, and Wilson's 2010 paper in Nature in favor of group selection:

and why many evolutionary biologists think that their conclusions are wrong:

To my knowledge, nobody has challenged the mathematical appendix to Nowak Wilson and Tarnika yet.

That's because it's correct.

Maybe it's a historical accident that first William Hamilton developed the inclusive fitness approach, and only subsequently did George Price develop the multi-level selection framework. On Bizarro Earth things happened differently. Price did his work before Hamilton, and MLS became widely accepted as THE correct approach after Richard Snikwad popularized it in his best-selling book "Gene Gangs." The alternative inclusive fitness approach was slower to take off, being pushed especially Sloan David Wontdad.

But James F. Crow was well ahead of both of them. He invented what are now called the "Hamilton-Price equantions" in 1955. I can attest from having discussed this with him that he was not bitter about this, having received plenty of recognition for other work.

I didn't know that. Do you have a reference? It would be nice to do posthumous justice to Crow by citing him on this.

David Sloan Wilson's also been dabbling in economics and started the website Evonomics. Here are some titles of their articles:

Does Wall Street Do “God’s Work”? Or Even Anything Useful?

Orthodox Economics Is Empirically Invalid and Theoretically Flawed. Bring on Complexity Economics

Economists Should Stop Defending Milton Friedman’s Pseudo-science

How Bailouts, Deregulation, and Reaganomics Led to the Rise of Donald Trump

Economists Agree: Democratic Presidents are Better at Making Us Rich. Eight Reasons Why.

Finally, The Clearest Evidence That Shows How Money Shapes America’s Elections

We Can’t Wait for an Invisible Hand. It’s Time to Rewrite the Rules of the Economy

Was the Rise of Neoliberalism the Root Cause of Extreme Inequality?

How ‘Competitiveness’ Became One of the Great Unquestioned Virtues of Modern Culture

Plato Got It Right. Tyranny is a Product of Extreme Inequality and Plutocratic Oligarchy

How America Hates Socialism without Knowing Why

The Milton Friedman Doctrine Is Wrong. Here’s How to Rethink the Corporation.

How Perfect Markets Concentrate Wealth and Strangle Growth and Prosperity

Don’t Be Fooled. Corporate Ecosystems Are Filled with Predators and Parasites

Business School Students Are Taught to Extract Resources and Make a Quick Buck Instead of Creating Value

Complexity Economics Shows Us Why Traditional Economics Always Fails

You Don’t Own That! The Evolution of Property

Samir Okasha’s stuff is very good: and more recently

A big problem in the debates over group selection is that different sides define the term differently. The classic models of group selection involve a population consisting of groups containing some mixture of altruists and non-altruists. Groups with more non-altruists are more likely to go extinct. It’s very hard to make group selection work in this case. It only succeeds if, once a group gets “infected” with a selfish individual, it then goes extinct before it can send out even a single successful selfish migrant. (This is similar to how a disease won’t go epidemic if each infected person infects less than one other person.) This is an extremely low level of migration (less than approximately one successful migrant per group lifetime), and pretty much everyone agrees this is not plausible for most species, certainly for almost any vertebrates, including H. sapiens.

On the other hand, many models of social evolution can be analyzed using a multi-level selection framework, where one component of selection is selection-within-groups and another component is selection-between-groups. MLS is an accounting method, NOT a separate theory. For example, kin selection can always be cast in MLS form, as a balance between selection within families (against kin altruism), and selection between families (in favor of altruism). Without selection between families, kin selection doesn’t work. Some people want to say that any time the between-group component of selection affects a trait, you’ve got group selection. Some people think MLS. even if formally correct, is sometimes or always confusing or misleading regarding actual cause-and-effect, and the alternative inclusive fitness framework is the way to go.

How technical do you want to get? For the basic math. Models of Social Evolution, A Guide for the Perplexed” (Chapter 6) is good.

For a balanced philosophically minded treatment, there’s Evolution and the Levels of Selection, fairly technical.

For a more accessible online back-and-forth, you can find David Wilson (pro-GS) below, and follow some of the links to Coyne and Pinker (anti-GS).

Humans are unusual animals, and group selection, specifically cultural group selection may be unusually important in our species. On cultural group selection (and some comments pro and con), see

Richerson , P., et. al. (2016). Cultural group selection plays an essential role in explaining human cooperation: a sketch of the evidence. Behavioral and Brain Sciences, 39, doi:10.1017/S0140525X1400106X, e30

Also worth checking out is Chapter 5 (The myth of self-interest and the science of cooperation) in Turchin’s War & Peace & War which relates cultural group selection to historical dynamics.

Since the issue is not settled, the best read might, for a non-specialist, be this one (I just read it myself and learned something):

I think game theory offers a concise way to understand/defend group selection.

Dawkins (and others) argued persuasively against group selection in the sense that individuals (or genes) cannot do individually non-optimal things to improve for the group. Deviators would overtake the group and that's the end of the story.

But this only says that genes have to play Nash. But of course there can be multiple Nash equilibria. Now if we define groups as frequently interacting individuals which are not necessarily kin, then different groups can play different Nash equilibria. If some equilibria dominate other equilibria, groups playing the better equilibrium will grow over time.

This allows for multilevel selection in a way that does not violate individual rationality, and it is therefore robust to the mutation argument put forth by Dawkins.

In one chapter of The Selfish Gene Dawkins admits that multi-level selection can happen. He just says it is not common.

The concept of “the mood association fallacy” will come in handy.

SEP article on biological altrusim
SEP article on fitness
Unto Other by Sober
Robert Trivers' review of Sober

Contrarian take: there is no unit of selection, your theory of fitness should be "mereologically invariant".

Pinker's argument hinges on the same priors that let him dismiss music as "auditory cheesecake," namely that all heritable aspects of phenotype are discrete and, for the most part, additive. This is not really how heritability works overall, it's just how the first measurable unit of heritability worked, and that's what set the tone when bioinformatics first started organizing itself as a field of study (and it's how bioinformaticists would really prefer heritability to work, because that would make the relevant math much simpler). GWAS of autism and schizophrenia, for example, keep producing contradictory data (e.g. "the frequencies of reported risk and protective alleles do not differ between related individuals with or without autism in independent data sets; instead, they reflect ancestral origin." 10.1038/mp.2013.34). Also, transgenerational effects really, genuinely, not-wishful-thinking-here-please-just-look-at-the-goddamn-data, are a thing (e.g. "cumulative multigenerational stress generates sexually dimorphic left-handedness and dominance shift toward the right hemisphere in males" 10.1093/cercor/bhw063).

EO Wilson's 2010 group selection paper was spot-on. He left it vague enough in the less certain parts, and that seems to have bothered a bunch of people. But, ultimately, he suggested that selection for preservation of a costly singular nest resource (for humans, a hearth->village->city progression) could allow very different phenotypes to genetically accommodate one another in nonlinearly additive ways. This is really just an extension of Lynn Margulis' endosymbiosis, but it makes a lot more sense when you realize that cell differentiation and eusocial caste differentiation are both mostly epigenetic.

Going back to Pinker's dismissal of music, phenotypes that are only adaptive at the group scale can't really be transmitted in a discrete manner, but clearly exist. E.g. Lisztomania-type semi-social clusters of humans are group-scale adaptive events, if only because they allow networking around the appreciation of computationally hard things like music (or, well, musician generation is hard, and music generation via musician is easy, but that's the point) and other communicable "great skills" that constitute proof-of-work, e.g. nunchuck skills, bowhunting skills, computer hacking skills, etc. Musical talent is hard to fake and easy to verify, and those are both also hallmarks of costly signals in biology. When you slow down birdsong you end up with Mozart (no, seriously:

"The genetic pathways involved translate not only to other insects but to other species including humans...This is exciting because of the potential it holds for addressing concerns that have to do with diseases like schizophrenia and autism." -- Joel Levine, PNAS Podcast

Anyhoo, hairless rodents are weird, and naked mole rats are eusocial mammals. Hairless primates are weird, and humans are hypersocial mammals. EO Wilson thinks we're eusocial, but maybe something got stuck in the 4-space plumbing with our hips and our brains and we just have a few discrete phenotypes like postmenopause (which is nonreproductive, adaptive, rare among primates, etc.) and various semi-group-adaptive phenotypic vectors like autism and schizophrenia. Or, maybe we're proto-eusocial, and we just have to keep performing caesareans until everyone's head is ginormous. And we'd need queens. Things would get messy.

So group selection could just be selection for systems that only arise in group contexts, and do not have explicit genetic representations that can be picked apart by Ayn Dawkins until they're convincingly individualized. E.g. there are many CNVs that correlate with autism when they're above/below a certain threshold, and schizophrenia when they're below/above a certain threshold. If a protein is produced normally when someone has three copies of a gene, two normal parents can still make a kid with as many as six copies or as few as zero, and that's if you're ignoring mutations like duplications and stuff. Methylation can alter the probability of discrete genotypes in the next generation. Germline methylation can be modified in vivo.

Biology is a bunch of nested membranes punctuated by edge cases. Eusocial organisms are no exception. Ants behave like a viscous fluid when you pile them up. The ones on the outside recognize their position and link up with neighbors to produce a sort of surface tension. Occasionally blobs swallow other blobs and don't digest them. That's almost certainly how we got mitochondria, which seem to have facilitated the oxygen metabolism and genetic bicamerality constraints that produced multicellular life. Similarly, a wacky paper in Medical Hypothesis suggests that bipolar disorder fits a plausible Neanderthal neurotype (10.1016/j.mehy.2011.10.005) and I'm inclined to believe them in spite of their less than prestigious (borderline tabloid) choice of platform because, in this case, BD and depression already have genetic and metabilic overlap, and depression has been subsequently shown to have come at least in part from neanderthal introgression (10.1126/science.aad2149).

Also, according to John Hawks, "[Neanderthals are] a population that we now know...was extraordinarily thin on the ground... At any given time...there were probably never enough neanderthals to fill a good college football stadium in the US." ( Human hypersociality/eusociality may involve genetic accomodation of archaic phenotypes. It happened in ants too: "Abouheif suggests that such evolutionary throwbacks, or ‘atavisms’, are everywhere but often overlooked. They are easier to spot in these ants because the [castes] are so physically distinct. “These things are totally unappreciated, and have been viewed as slips in the developmental system that don’t go anywhere,” he says." (10.1038/nature.2012.9746).

If evolution were purely tree graphs at every scale, maybe the selfish gene people could claim to have the optimal model. But horizontal gene exchange is clearly at least half the picture. Human mate selection relies in part on cues that indicate the extent to which a partner will fulfill the immunological needs of the next generation. That might mean coupling up with a crazy person. It could eve be entirely locally maladaptive for every party directly involved with said crazy person, yet still remain herd-immune-adaptive to have a few crazy couples. Kamikaze is locally maladaptive. Suicide bombing is locally maladaptive. Termites blow themselves up. Etc. The point is that there are co-adaptive traits that can persist in a population, despite not classically conforming to Hamilton's rule. EO Wilson's mathematician coauthor Corina Tarnita tried to explain this, but then the entire community went apeshit and she moved on to other things. Great job, guys.

Mind you, EO hasn't mentioned neanderthal introgression. He formulated his 2010 hypothesis (10.1038/nature09205) in more general terms. But clearly eusociality isn't entirely genetic, and clearly humans have some really weird eusocial traits that we don't share with other extant primates, and since 2009 we've all had to accept the fact that there's a bunch of recently acquired neanderthal genes floating around in different people: "More than six billion people are more than 2% neanderthal. That’s like having 200 million neanderthals walking around [because everyone has a different collection of neanderthal genome fragments]. That is a lot of neanderthal. Neanderthals today, in extinction, are more successful than they ever were when they lived." ( So it doesn't seem like a huge stretch to say that maybe we tied whatever 4-space evolutionary knot the hymenoptera, heterocephalinae, and select alpheidae managed to tie that led them to become so industriously superorganismic.

A bit dated, and written by a nonspecialist (in evolution; he's a law and economics professor). Nevertheless, fwiw, it's an interesting take on the intersection of political and economic preferences, and evolution (including a foray into the group selection debate), offered by a self-professed reformed libertarian. :)

(The subtitle reminds one of books by Ridley and the like. It's actually more academic than those, at least in terms of his writing style.)

The classic argument against group selection: "Adaptation and Natural Selection: A Critique of Some Current Evolutionary Thought" by George C. Williams. Williams was a famous and extraordinarily capable evolutionary biologist. After reading this book, it's hard for me to imagine rational support for group selection.

George Mason University's library has both a paper copy and an electronic version. The book was published in 1996.

Williams argues against a straw man version of holistic group selection that is irrelevant for modern mulit-level selection approaches. The latter in effect build up from microfoundations, if you will. The aggregate emerges from the mathematically specified interactions between memes at the lowest levels. Williams fails to deal with that. This is sort of like saying there is no macroeconomics because ulitmately it is individual people who make important economic decisions (although of course there are things like governments, blah blah).

Sorry, In my comment above, I made a significant typographical error. The book was published in 1966, not 1996. As I said, the book represents the classical argument against group selection, but the date is worth noting.

My comment above was intended to apply to evolutionary biology and genetic evolution, not more recent theories of cultural evolution. I think it's unfortunate that social theorists of cultural evolution have not only appropriated the vocabulary of organic evolution, but also used it to assert the primacy of cultural factors in human evolution. This goes beyond the widely accepted view that human genes and human culture have co-evolved, and continue to do so. Sorting all this out in a blog comment is utterly impossible,

For a very brief introduction to recent theories of cultural evolution, I'd recommend the topical entry found in "The Stanford Encyclopedia of Philosophy."

No two things are alike. Even fundamental particles (if there even are such things) cannot share the same space/vector at the same time. Bacteria evolve almost as fast as viruses - which is good since it allows tracing in foodborne illness cases. Indeed differences immediately emerge even among clones, especially if they're in a biofilm. There are really no such things as groups of the same thing; only groups of not-too-different things which we inspect only superficially lest the otherwise readily revealed differences scandalize our project of categorizing everything. It's the temptress of over-generalization that hearkens to you.

To me group selection is a mistaken theory of functionalism. It confuses genealogy with motives for reproduction. A best explanation of functionalism is Ruth Garrett Millikan's book Language, Thought, and Other Biological Categories, see . In Millikan's account, a function is a reason why a behavior or action or product is reproduced, in that it improves the fitness and survival of the carrier of that behavior or action or product. Consider then that, a surviving function or behavior or product starts in one group or area. Its improvement of fitness and survival benefits its original group. Then it spreads out, but not evenly nor equally. Over time, given diffusion, the correlation of reproduction with a given group decreases, but rarely fully disappears. Still, a biological or genealogical explanation is not incompatible with technology transfer. For example, Switzerland and Japan took technologies of other nations (watches, autos) and improved on them in design, reliability and marketing, despite their development in Britain, Germany and the US. Subspecies learn, mimic and improve on other subspecies. "Group selection" can jump. Which suggests "group selection" is a poor explanation.

Is the question about group selection in biological evolution, or group selection in social change and learning?

Depends on who you ask. David Sloan Wilson focuses on media-transferable social stuff ("cultural evolution"). EO Wilson focuses more on heritable stuff that persists even if you're adopted at birth by martians ("biological evolution"). Combining the strategies of the Wilsons is really difficult. They speak dramatically different languages. Group selection really needs a mesoscale theory to bridge the two. But transgenerational epigenetics is still haram because we now have the technology to keep a lot more old people alive and they have careers to protect goddamnit! It's analogous to the split between relativity and QM:
"If you don’t include quantum mechanics, then you can copy information any way you want. And if you don’t include relativity, then there [are] no causality constraints. So this question actually intimately involves both.” --Patrick Hayden (

If you ignore biology, you end up thinking that people can be swapped out arbitrarily like little cogs (Chomsky's view, ironically enough) which really screws with long-term models of group selection (because pressures on ancestors aren't transmitted, even through standard natural selection). If you ignore fuzzy linguistic pattern matching social scientists, you end up atomizing your area of focus down to an impossibly small collection of discrete genes. It's the same fallacy that led Jim Watson to conclude in 1998 that "Judah [Folkman] is going to cure cancer in two years." Paul Davies (who is currently chewing through a big fat hail mary cancer research grant) puts it succinctly:
"What is epigenetics? Well, a lot of newspapers give the misleading impression that DNA is somehow a blueprint for an organism. This word “blueprint” crops up again and again. It is clearly fallacious; you can see it immediately because if you compare our DNA with that of say a mouse, it’s not a lot different. In fact, I’ve got this quote here, “Murine and human genomes qualitatively differ in the structure of only 300 genes” and, obviously, there are more than 300 ways in which a mouse and a human differ. So that’s a sort of trivial example. Another example is if you take say a cortical column in the brain - immensely complex - and the figures are, if you take the brain as a whole, the human brain has about 100 billion neurons and about 100 trillion synapses. So, it would be ludicrous to suppose that the wiring diagram of your brain is somehow captured by the information content of DNA.
All of this type of information comes from somewhere else. So, where else does it come from? A clue comes from the developing embryo. So, all cells in your body contain the same DNA, and yet these cells are different. When the embryo starts out, each of the cells is pluripotent, and can develop in a variety of different ways…and the point is that although they are genetically identical, not all the genes are switched on at the same time."

Full disclosure: I'm squarely in Roger Penrose's cluster of wacky. I strongly suspect that the same sources of experimental frustration in studying the dead world persist when studying the living. But that would be inconvenient, so instead let's all keep shouting past one another until we all get diabetes or antibiotic resistant staph. Actually, D-Wave-type hardware might finally make studying this stuff computationally feasible. That would be convenient.

Correction: Paul Davies finished chewing through his grant. And it was from the US NCI.

My B.

I thought Richard Dawkins had effectively destroyed all group selection arguments on Chapter 1 of the Selfish Gene. As an aside to this discussion, isn't the efficiency criterion in economics (for example, the idea that the common law promotes economic efficiency or wealth maximization) a textbook example of the fallacy of group selection arguments?

He has modified his views since, Enrique. Read the full second edition, although he thinks it is rare.

Thanks for the pointer. My understanding is that there are two problems with group selection: (1) how do we define the "group"?, and (2) how does one protect the group from free riders and defectors and other exploiters?

Henrich calls the second the "greenbeard problem."

Well, this very recent paper may be of use:
Jussi Lehtonen - Multilevel Selection in Kin Selection Language

From the abstract - Few issues have raised more debate among evolutionary biologists than kin selection (KS) versus multilevel selection (MLS). They are formally equivalent, but use different-looking mathematical approaches, and are not causally equivalent: for a given problem KS can be a more suitable causal explanation than MLS, and vice versa. Methods for analyzing a given model from both viewpoints would therefore be valuable. I argue that there is often an easy way to achieve this: MLS can be written using the components of KS. This applies to the very general regression approach as well as to the practical evolutionarily stable strategy (ESS) maximization approach, and can hence be used to analyze many common ESS models from a multilevel perspective. I demonstrate this with example models of gamete competition and limitation.

Let me take a stab....

Suppose there's an individual gene that exhibits in individuals being more inclined for people they grew up around. Easy to see how that confers an evolutionary advantage. Such people will be inclined to be good to whatever tribe they were born in and likewise tribes would be inclined to view such individuals as 'good children'. Since humans are social creatures getting along with society is an advantage over not being able to get along.

Now groups splinter all the time and form with random differences but they will like the 'loyal to my group' trait so that confers an individual advantage.

Now groups value different things at different levels and there are some things that help the group while harming individuals in the group. This brings me to my hypothesis on the 'gay gene'....which I came up with during debates about whether homosexuality could ever be 'natural' or not.

Suppose there was a gene that didn't guarantee someone would be gay but caused the chances to be much 10% instead of 0.001%. A pure gay gene that resulted in a gay person 100% of the time would be at a huge disadvantage in terms of evolution (how would it get passed down?). But the gene above could easily get passed down because most people with the gene will not be gay and will reproduce as much as anyone else.

But by itself the 'gay gene' seems to be at a disadvantage. If you are passing it down, 1 out of 10 of your children are unlikely to have kids while your buddy who doesn't have it will see 10 out of 10 his having kids. But suppose a group or tribe just happened to tolerate a bit of the population being gay?

Consider the 'surplus male problem'. More men than women means men fighting each other for wives and getting in trouble. This is the flip side to Pinker's story about the fierce warrior type benefiting the group...yea they benefit the group when the group is under attack but they can be a liability when they come back from battle and start demanding all the weaker men give up their wives and food to them. But a group with more men is at a physical advantage in terms of offense and defense against other groups. Groups that tolerated gays more might find themselves at a bit of an advantage. Throw that together with an 'affinity gene' and you might get a gay gene being selected for *even though* it's a disadvantage for any particular individual who has it.

This is probably not the 'cause' of homosexuality but perhaps I've sketched out something that could square group selection dynamics with the fact that selection happens on the individual level.

"Suppose there was a gene that didn’t guarantee someone would be gay but caused the chances to be much higher….like 10% instead of 0.001%." you might want to email joe

Something that should be noted is that all the libertarians here who are knee jerk rejecting "group selection" because it is "socialist" or whatever, really do not know what they are dealing with. I note that most anarcho-capitalists, while asserting methodological individualism, argue for the ability of people to self-organize and cooperate in small groups. Many are big fans of Elinor Ostrom, who argued for a variety of self-organizing human groups.

So some of these people need to think more carefully about what they are advocating or think (and evolutionary biologists who think that multi-level selection implies that socialism is wonderful, and some do or did, see Lewontin, should be disabused of their misconceptions as well).

You should definitely check out Peter Godfrey-Smith. He has a few interesting articles (and books) in the field of philosophy of biology and particularly natural selection. I have enjoyed his (i believe pragmatist ?) view on the whole issue of levels (ind, group, multi) as well as units (genes,individuals, cultural) of selection.

This article is very useful:
“Individualist and Multi-Level Perspectives on Selection in Structured Populations"

But there are more on levels of selection and population structures, altruism...

and yes, i believe we should apply the idea of “gestalt-switching pluralism” in the social sciences as well when looking at some issues such as relationship between individual interest and social institutions. The view about causation and relevance of processes at both levels is better informed when drawing a parallel to similiar ideas in the mentioned article

I recommend "Evolution--The Extended Synthesis", 2010, edited by Massimo Pigliucci and Gerd B. Müller. This collection is about much more than group selection, but that debate is now very old, and it is good to consider it in light of recent broader evolutions of evolutionary theory. David Sloan Wilson has a very good chapter in this book, "Multilevel Selection and Major Transitions", which covers the basics of group selection theory and the history of the debate. You would probably also enjoy the final chapter by Werner Callebaut.

For a very short (4 page) introduction to these issues, see the 2011 paper by Omar Eldakar and David Sloan Wilson, "Eight criticisms not to make about group selection".

However, looking over the comments I see that David Sloan Wilson has replied directly to your bleg, so why not go to the source!

This new book argues that aging in animals is an artifact of group selection. The idea is that we are programmed to age as a way to free up resources for younger generations and avoid population explosions that can lead to extinction. (bad for the individual, good for the group). The fact that calorie restriction lengthens life is offered as evidence that programmed aging can be mitigated in times of scarcity.

The argument that deviants without the aging gene would overrun the population is addressed by assuming the gene only comes into play after individuals no longer reproduce...

1. One important data point involves not cases where a trait harmful to the individual but helpful to the group is selected for, but to traits that are neutral with respect to fitness.

We know that ancestry informative, fitness neutral traits of members of groups have again and again and again exploded in frequency as a result of the expansion and success of the group in question. These ancestry informative genes have surged at the expense of genes informative of other ancestries, which have generally vanished or become extremely rare.

2. Likewise, we know that there have been many historical instances in which one population with a particular gene pool has completely, or nearly completely replaced another population with a different gene pool completely or nearly completely. For example, this happened in Ireland sometime around 2500 BCE.

Now, replacement does not in and of itself imply fitness based selection. As best we can tell from the case of Ireland, some combination of bad weather and unsustainable farming practices led to the collapse of the agricultural economy and a dramatic decline in population. Then, the weather got better, the soils regained their fertility, and a new people moved into the vacuum and peopled the greatly depopulated land at a time when the selective pressures applied to the first population were gone. It could be the second population survived not because it was more fit genetically, but because it lived someplace with a more abundant ecology and better weather at the time when the first population in Ireland was wiped out, and that the dire circumstances that wiped out the first population in Ireland never recurred (the current population in Ireland genetically is very similar to the people who moved in around 2500 BCE).

And, even if the newcomers collectively were more fit, it could very well be that their fitness was due almost entirely to their culture (e.g. agricultural technologies that were more sustainable and handled harsh conditions better) rather than to their genes.

3. One of the most striking instance of genetic selection in the last 10,000 years or so in humans has been the evolution of lactose tolerance in adults, which went from being almost entirely absent, to being widespread, in a few hundred years.

There is strongly suggestive evidence that this genetic benefit was largely confined to only certain populations and that this genetic advantage helped the group succeed. But, this in and of itself doesn't imply group level selection. Does excluding non-group members from acquiring this fitness enhancing gene through endogamy count as group selection (at least at the cultural level), however, constitute group selection? Because, allowing your group to have a fitness enhancing gene while denying it to another group, allows your group to access resources of the outgroup that lacks the gene through conquest that could not have been achieved by any one individual in the group.

All this is to say that the focus of group selection inquiry may be too strong on the hardest of cases while not paying enough attention to cases where the group and individual are not so diametrically at odds and there is mutual reinforcement between the group and the individual beyond what would be present with panmixia.

Along the same lines, one of the distinct factors of the Basque population is a high frequency of an Rh factor that on average reduces fertility in couples involving an outsider and an insider, relative to those with two insiders. These gene strengthens group solidarity and reinforces endogamy, while decreasing the fitness of individuals in mixed couples and not improving the fitness of individuals in unmixed couples. The effect size isn't huge, but it would seem like a classic case of group selection in the extreme case of adversity between individual and group fitness. - "Lynn Levy brings us the story of George Price, who bopped through the 60s like some kind of scientific Forrest Gump. He worked on the atom bomb, transitors, computer-aided-design, and eventually turned his attention to the problem of altruism. Lynn talks to Oren Harman, author of The Price of Altruism, and George's daughters Annamarie and Kathleen, who help us get to know this complicated genius. In 1967, George left his family, went to London, and wrote a mathematical equation to explain why one creature might sacrifice its own interests for another. Carl Zimmer helps us understand why altruism is such a problem in the first place, and how family might hold the key to understanding apparently selfless acts. The so-called Price Equation changed biology ... and ultimately led Price to spend the rest of his life trying to transcend his own equation."

Discussed in detail in this podcast:

Ekkehart Schlicht's On Custom in the Economy has an interesting discussion of group evolution.

Here are two posts I wrote when grappling with this a couple of years ago:

What is multilevel selection?

Groups, kin and self interest:

Comments for this post are closed