[LINK] Steven Pinker on "The false allure of group selection"

post by hairyfigment · 2012-06-19T20:14:15.842Z · LW · GW · Legacy · 12 comments

This essay at Edge touches on a few possible meanings for the term "group selection." Pinker argues that as a form of memetic theory it has no explanatory power, and that group selection for genes does not fit the evidence. He focuses on humans with some mention of insects that live in hives. So the essay doesn't seem surprising, but it does seem rather Hansonian.

12 comments

Comments sorted by top scores.

comment by David_Gerard · 2012-06-24T17:01:12.386Z · LW(p) · GW(p)

Jerry Coyne has an excellent piece, prompted by Pinker's piece: The demise of group selection. He bludgeons the idea to death in a manner any LW reader would enjoy.

He also notes why group selection is dead amongst evolutionary biologists, but is showing up from vocal proponents: Templeton Foundation money. Sigh.

comment by timtyler · 2012-06-20T00:53:18.472Z · LW(p) · GW(p)

Pinker's article is scientifically mistaken. There's a long-established scientific consensus about group selection models - that says they are equivalent to kin selection models - and represent a different partitioning scheme. Pinker's article isn't part of this consensus - he doesn't understand the topic.

Replies from: Luke_A_Somers
comment by Luke_A_Somers · 2012-06-20T13:33:49.514Z · LW(p) · GW(p)

His definition of group selection may exclude that from 'group selection' and include it within what he approves of - 'gene selection'.

It seems to me he's hacking away at the bone when he should be cutting at the joints - evolution can act on first-order effects (I'm faster, so I escape a predator or catch my prey) where you only need to consider yourself to see the benefit, or second-order effects (I cooperate on the hunt so we can all eat), where you only need to consider yourself and those with the same relevant genes, or third order effects (pea-hens preferring peacocks with big plumes), where you need to consider yourself and those with certain other relevant genes and those without...

But by considering group selection to be this separate thing, he's denying that ultimate role, and thus defines it out of existence.

Replies from: timtyler
comment by timtyler · 2012-06-21T00:47:34.501Z · LW(p) · GW(p)

His definition of group selection may exclude that from 'group selection' and include it within what he approves of - 'gene selection'.

He just says:

In this essay I'll concentrate on the sense of "group selection" as a version of natural selection which acts on groups in the same way that it acts on individual organisms, namely, to maximize their inclusive fitness (alternatively, which acts on groups in the same way it acts on genes, namely to increase the number of copies that appear in the next generation; I will treat these formulations as equivalent).

That is not good enough to rule out the groups of the "new" group selection - which would be an especially foolish thing to do anyway, considering that he specifically says he disagrees with that at the start of the essay:

I am often asked whether I agree with the new group selectionists, and the questioners are always surprised when I say I do not.

Pinker's article is out of touch with the truth on the topic.

Replies from: hairyfigment
comment by hairyfigment · 2012-06-25T18:28:24.618Z · LW(p) · GW(p)

What are you trying to argue? Pinker's article says,

Some mathematical models of "group selection" are really just individual selection in the context of groups.[2] 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."

If you think that "new group selectionists" would agree with this, why call them that? The 'old' kind doesn't seem to have gone away, since you yourself cite articles explicitly arguing against them:

Several recent papers, however, have objected that inclusive fitness theory is unable to deal with strong selection or with non-additive fitness effects, and concluded that the group selection framework is more general, or even that the two are not equivalent after all. Yet, these same problems have already been identified and resolved in the literature.

Another of your cited abstracts gets more specific:

In this paper, we start by addressing Traulsen and Nowak's challenge and demonstrate that all their results can be obtained by an application of kin selection theory. We then extend Traulsen and Nowak's model to life history conditions that have been previously studied. This allows us to highlight the differences and similarities between Traulsen and Nowak's model and typical kin selection models and also to broaden the scope of their results. (emphasis added)

Coyne's anti-group-selection piece quotes an overlapping group of authors as follows:

“No group selection model has ever been constructed where the same result cannot be found with kin selection theory”. “The group selection approach has proved to be less useful than the kin selection approach.” (ditto) “The application of group selection theory has led to much confusion and time wasting.” It is, as the authors say, “easy to misapply, leading to incorrect statements about how natural selection operates,” it is “not distinct from kin selection”, and it “often leads to the confusing redefinition of terms and the use of confusing jargon.”

You also cite people who would at least partly disagree with the bolded part of that. But even they might agree with the last point. They appear to recommend the dual-perspective approach for people who thoroughly understand the individual fitness approach already, and could explain any valid evolutionary argument in those terms.

Replies from: timtyler
comment by timtyler · 2012-06-27T00:24:42.919Z · LW(p) · GW(p)

What are you trying to argue?

My article's title is a good synopsis:

Group selection and kin selection are formally equivalent. I.O.W. they make the same predictions.

Richard Dawkins, E. O. Wilson, Martin Nowak and Steven Pinker are not on board with this. They have yet to join the modern scientific consensus about group selection.

comment by ChristianKl · 2012-06-24T14:36:44.612Z · LW(p) · GW(p)

Some mathematical models of "group selection" are really just individual selection in the context of groups.[2] 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." But the tradeoff between "benefiting the self thanks to benefiting the group" and "benefiting the self at the expense of the rest of the group" is just one of many tradeoffs that go into gene-level selection. Others include reproductive versus somatic effort, mating versus parenting, and present versus future offspring. There's no need to complicate the theory of natural selection with a new "level of selection" in every case.

If an additional parameter in the model helps the model to make better predictions then it's good to use it. The results of the predictions justify the parameter.

Maybe Pinker makes this error because he make a wrong assumption about evolution being deterimistic. In real life there a lot of randomness in evolution. Some gene's who are advantageous simply disappear in genetic drift. Some "bad" genes win just because they are for a bunch of time located next to a really "good" genes.

There are simply a lot of mutations. Life isn't fair. The good mutations only win on average. Adding parameters that track individuals or groups help to produce a more accurate model.

Replies from: David_Gerard
comment by David_Gerard · 2012-06-24T17:02:44.342Z · LW(p) · GW(p)

Jerry Coyne says that the vague idea of group selection at best gives the same result as the selfish-gene model, with more complicated mathematics, not less complicated - it fails Occam.

comment by Thomas · 2012-06-19T20:32:31.188Z · LW(p) · GW(p)

The evolution is just going on at all the levels.

Genes, for example, may loose everything by the actions of their products. We can decide and abandon the DNA, one day. But even smaller actions we already do, inflict the genes.

Just every level is susceptible, groups as well. Pinker is not right this time.

Replies from: David_Gerard
comment by David_Gerard · 2012-06-19T21:05:36.962Z · LW(p) · GW(p)

See, that's the sort of idea that seems like it must be true, and you can make out that it's a coherent position, but it really isn't how evolution actually works. They did in fact go and look.

Replies from: timtyler
comment by timtyler · 2012-06-21T23:48:56.523Z · LW(p) · GW(p)

Have you not heard that group selection models and kin selection models are formally equivalent?