The Logical Necessity of Group Selection

It seems antithetical to the notion of science.

I totally agree. The costs for these journals is just ridiculous, especially considering most of them are published online too. My university gives access to most academic journals to students, if you want me to e-mail you a copy.

Given the nature of this blog, would it constitute fair use to post the full text?

I wouldn't. You can quote bits of it, but definitely not the full thing.

Michael, thanks for the pointer.

It's unfortunate that the journals are all making us pay to access the text of scientific papers. It seems antithetical to the notion of science. Do you know of anyplace to publicly access the full text for free? Given the nature of this blog, would it constitute fair use to post the full text?

The latest issue of New Scientist has a feature arguing for group selection (multilevel) selection by E.O. Wilson. You can find it here:

http://www.newscientist.com/channel/being-human/mg19626281.500-evolution-survival-of-the-selfless.html

If you consider a more interesting fitness function than ...

But that would be a form of replication. You began one of your arguments with the statement, "Even if we restrict NS to speak only of replicative agents...", which seemed to imply that NS doesn't require replication all the time. That's all I was referring to.

Ears do compete against other ears, and in the ancestral environment, someone who had working ears was much better at survival than someone who was deaf.

Compete for what? Competition is a struggle for resources. Which resource are the ears competing for? They're competing for space for their genes in the gene pool. So, again, it boils down to the genes. Looking at it from the ear's point of view doesn't change the fact that it's the genes which are competing.

This statement importantly misses the point that genetic code does not exist in a vacuum

I didn't mean to suggest that genes can simply replicate themselves without assistance; sorry about that!

Genes do not contain all the information.

Elaborate a little bit more on what you mean by information.

Yes, the ear is “just” a manifestation of the genes, but it is also very true that the genes are “just” a manifestation of the ears (and other aspects of the organism).

I think you're discussing two separate issues here. Organism A's ears are affected by its genes, but not the other way around; once the organism has developed, its body plan stays the same. The genes of A's offspring are affected by A's ears, though, because A's ears affect A's reproductive fitness. But A's ears are derived from his genes. If A's genes affect A's ears, and A's ears affect A's reproductive fitness, which affects A's offspring, then logically A's genes affect A's offspring. The ears are the intermediary step. They're an instrument the genes use to find selective advantages in the environment.

I think we're in agreement, here. It's just that you want to say it one way, and I want to say it another.

You can call, say, deleting all of the programs on your computer that start with the letter A, natural selection. But as biologists use the term, it isn’t.

If you consider a more interesting fitness function than simply "doesn't start with A", and you replenish the deleted programs with new ones that share more in common (statistically speaking) with the surviving programs than they do with the deleted ones, then yes, you actually have natural selection in the biological sense. It's not just a metaphor, I'm claiming that it's the same process.

All a sexual multicellular organism needs to reproduce is its own genetic code and a partner’s genetic code.

This statement importantly misses the point that genetic code does not exist in a vacuum, rather it has been evolved over many generations to exist in a finely structured environment. Trivially speaking, genetic code does not reproduce without all of the environmental scaffolding of the cell in which it resides. Nor can it reproduce if you simply take that cell and place it in a vat of water (for instance). Human genetic code can only reproduce indefinitely if it exists in a population of other humans who sometimes mate. Back in the primordial soup, presumably there were autocatalytic sets of molecules (precursors to DNA/RNA) that under the right environment worked (as autocatalysis does) to recreate themselves (and possibly one or more copies). But again, "right environment" is crucial, and to speak of reproduction without speaking of the environment makes little sense.

My ear doesn’t produce other ears, my genes do. My ear doesn’t compete against other ears, my genes compete against other genes. The genes contain all of the information. The ear is just a manifestation of the genes.

Ah, but your ear does produce other ears. If you consider all of the functional parts of your body together, they are indispensable to your genes being able to "reproduce themselves". There are many parts of your genes which could piecemeal be damaged and they will still reproduce an ear with the capability of reproducing itself. Ears do compete against other ears, and in the ancestral environment, someone who had working ears was much better at survival than someone who was deaf. Genes do not contain all the information. In fact the environment of the genes -- not only the microenvironment of the nucleus and cell, or the broader environment of the organism body, but the entire environment, including the developing organism and its own greater environment through its lifespan -- contains orders of magnitude more information than genes ever could. Yes, the ear is "just" a manifestation of the genes, but it is also very true that the genes are "just" a manifestation of the ears (and other aspects of the organism).

NS is an emergent process that happens regardless of whether the population under study consists of biological entities, computer programs, memes, whatever.

I think you're painting with a bit of a broad brush. You can call, say, deleting all of the programs on your computer that start with the letter A, natural selection. But as biologists use the term, it isn't. We might just be quibbling over terms here.

Even if we restrict NS to speak only of replicative agents, it seems pretty clear to me that multicellular organisms do replicate.

All a sexual multicellular organism needs to reproduce is its own genetic code and a partner's genetic code. A human being who's missing two ears, two legs, and an arm won't produce offspring missing two ears, two legs, and an arm (unless its a heritable genetic defect). Non-genetic material is there to facilitate the reproduction of genetic material, and that's it. This is why Dawkins describes the body as a vessel for the genes. It has no purpose without the genes that control its expression. Therefore you may refer to it as group selection because it involves multiple interlopers, but it's really still selfish, natural selection.

My ear doesn't produce other ears, my genes do. My ear doesn't compete against other ears, my genes compete against other genes. The genes contain all of the information. The ear is just a manifestation of the genes.

Jon, thanks for your thoughtful comment. I used to take the selfish gene argument, as you seem to, in the narrow interpretation, but I've since come to see the inherent problems with this. NS is simply an abstract model created to explain a set observed data. Like any scientific model, it can never be complete -- after all there is a vast difference between the map and the terrain it represents. Models of complex dynamics always deserve re-examination when new evidence is presented.

We can argue over the exact way to state the preconditions of NS, but I think you will agree that as long as the pre-conditions are met, NS is an emergent process that happens regardless of whether the population under study consists of biological entities, computer programs, memes, whatever. (Remember, Dawkins himself was the one to coin the term "meme" and suggest that under the right conditions they evolve via NS).

Even if we restrict NS to speak only of replicative agents, it seems pretty clear to me that multicellular organisms do replicate. It's not important how they do so, and it doesn't matter that there is something called DNA or that we make a distinction between genotype and phenotype (see my next post). They replicate with heritable variation, and therefore meet the preconditions for NS to occur, at least under the standardly accepted definition. And whether you view the cell population in the body as behaving cooperatively, competitively or otherwise, it's clear that they are a group of individual cells (amongst other things). By definition of NS, they are selected for as a group, hence group selection. Group selection exists in at least once instance.

The point of my post was not to suggest that NS needs to be redefined, but rather if you accept the standard definition (and there are several different ways of stating it, but all lead to the same conclusion) you cannot also logically deny the existence of group selection.

Natural selection requires replication and mutation. When you abstract the level of selection from the gene to the group, what are you really talking about? Groups don't replicate and they don't mutate. Individuals within the group replicate, and their individual genes mutate. So when you talk about group selection, you really always end up boiling it back down to the gene and to the individual. There's no analogy to chromosomal crossing over and recombination from meiotic divisions at the group level.

I think that group selective process can and do occur, but that to call them natural selection is a bit too much. It's easy to be tempted to say, well this group of individuals behaves cooperatively, therefore they can be seen as a single entity. One might, as you have, reference multicellular organisms such as humans, with the cells or organs or something being the individuals. But that's a bit of a mischaracterization. A human being isn't just a colony of cooperative cells. A human being's somatic cells all contain the same DNA. Different body cells are specialized for different tasks not because they have different DNA, but because they express the DNA differently. And there are genes which control which cells should express which parts of the DNA. And there are genes which control those genes, and so on.

So all of these cells are working to replicate the same exact genome. It can be seen as cooperation, but it can also be seen as self-interest. A cell doesn't so much care if it replicates itself now, because every other cell around it has the same stuff to replicate. Thus as long as some cell somewhere in the body is replicating, cells will seem to behave cooperatively, even going so far as to commit suicide for the benefit of other cells.

When you abstract to a real group, where every individual has a different genome, we lose this kind of "for the group" cooperation. There are other kinds of cooperation, but they tend to be self-interested altruism. True, "for the group", non-genetic cooperation seems, and I find it reasonable, to be what Dawkins calls a misfiring. The proportion of people who would sacrifice their lives for a stranger to those who wouldn't is extremely small. And so that seems to indicate that it's a fluke. If we go back to the human body again, we can see that just the opposite is true. The proportion of cells which would commit suicide for the group to those who wouldn't is very very high.

Of course, I'm just trying to paraphrase the selfish gene argument. I'm not a biologist or anything, :P