(Cultural) Multilevel Selection

April 24, 2018 5:55 PM

For the Good of the Species - "Whether group selection is an important evolutionary force, or not, is a highly controversial question in evolutionary science. A substantial proportion of evolutionary scientists still think that it is not." (via)
posted by kliuless (18 comments total)

Depends on the niche? Seems like a prisoner's dilemma writ large. I can't picture how cooperation as a concept could exist without multilevel selection.
posted by BrotherCaine at 6:22 PM on April 24, 2018 [1 favorite]

Recently read both The Social Conquest of Earth and The Righteous Mind which complemented each other really well. I was delighted that TRM was a response to The Selfish Gene in many ways. One of the most basic arguments that I like is that ants remade the world when they became social. It was such a powerful tool that other insects just didn’t have. So too, humans are remaking the world. To me, turning all alutrism into selfishness with logic is twisted. There is real emotional power in being part of a team.
posted by macrael at 7:12 PM on April 24, 2018 [8 favorites]

I find this post a little confusing, because the first link and some of the others are about group selection but then there are some about genetic differences between and within human populations that don't seem clearly connected. Maybe it's explained in the twitter discussions, but I found them incomprehensible, as twitter discussions tend to be. Anyway...

The problem I have with group/multi-level selection is that I've never been able to understand exactly what it adds to the gene-centred view; the more plausible a group selection explanation for evolution is, the easier it seems to be to re-explain it in terms of gene-centred evolution.

Take the ostracod example: it's not likely that the extinct species suddenly reached a certain threshold of size that was too large and then all died, because natural variation would have left some smaller ones around and the species would eventually reach an equilibrium below the threshold. What's much more likely is that the environment changed in some way - say, the introduction of a new predator from the area where the ostracods stayed small - that made being large a liability, and the species couldn't adapt fast enough by shrinking to avoid extinction.

In that example, individuals from the surviving species were certainly more fit for the environment with the predator in it, but you don't need to resort to a mysterious group fitness factor to explain their survival. Maybe the predator had been in their environment for much longer. Or a sudden increase in the size of male ostracods could actually be a response to circumstances that are likely to be associated with oncoming extinction: if the resources available to female ostracods become much less available, maybe the sexual selection for larger males is intensified.

If sudden environmental changes that wiped out larger ostracods were frequent enough, perhaps you'd see selection for traits that tended to counteract the pressure from sexual selection for larger ostracods. But that doesn't require group selection either.

I can't picture how cooperation as a concept could exist without multilevel selection.

One of the interesting things about The Selfish Gene is that only part of it is about how selfish genes are. Most of it is an explanation of how cooperation can occur not just despite genetic selfishness, but because of it. And there's been plenty more work done in the more than 40 years since it came out.
posted by A Thousand Baited Hooks at 8:48 PM on April 24, 2018 [7 favorites]

I know there's a debate about (biological) group level selection and I've never quite understood it. I briefly thought this would be the moment someone explains the reason it's important to me on a day I had time to listen but alas, no. It inspires debate but to me it mostly seems like different types of accounting.

Certainly the ostracods are being selected at the individual level, right? In aggregate the species can't survive but that's because too many individuals are being outcompeted. OTOH the phenotype for bees that gets selected is obviously group behavior.

The one hypothetical I can think of is were the difference would be important is showing group selection where one group of less-greedy animals outcompeted all its neighbors because they as a group avoided a tragedy of the commons (as opposed to a balance between gluttons and ascetes within each group.)

(Side note: Animal Weapons is a fun book about sexual selection in cases like elk and beetles leading to ostentatious oversized weapons only good for ritual combat. Did you know that male deer get seasonal osteoporosis as they need to plunder the calcium in their bones to grow a manly set of antlers?

The human analogies in the book are a boring gimmick that probably helped get it published. But happily they are a super minor part.)
posted by mark k at 9:28 PM on April 24, 2018

So I never went beyond a B.S. in biology, and genetics was a long time ago, but the things I've read have basically said the debate about group selection yes-or-no is basically a "debate" among the public intellectual crowd, who spend far more time on tour flogging their books and articles than they spend in the lab; and that to proper evo scientists the whole thing is a sideshow of labels and egos.

In that light, from the main page of the first link:

He is the founder of a new transdisciplinary field of Cliodynamics, which uses the tools of complexity science and cultural evolution to study the dynamics of historical empires and modern nation-states.

I just broke out in hives.
posted by traveler_ at 10:18 PM on April 24, 2018 [11 favorites]

I recently bought and re-read (first time was in college, years ago) Defenders of the Truth: The Sociobiology Debate which is, despite the title, actually a really excellent and sympathetic study of the way debates arose within evolutionary theory from the 1970s onwards involving the group selection/gene-level selection, as well as sociobiology. I'd heartily recommend it to anyone interested at the history behind this discussion, and why it touched such a deep nerve in so many scientists at the time (and continues to do so). It also has a lot on the eclectic ideas and career of E.O. Wilson, whose relation to group theories of fitness is a lot more complicated than the author of the first link acknowledges (he didn't just "flip" from one side to the other).
posted by AdamCSnider at 3:25 AM on April 25, 2018 [2 favorites]

My impression is that multilevel selection theory and the more widely accepted kin selection theory are converging on the same results, so long as you're using the more current definition of "kin" for kin selection (i.e. not relatedness-via-having-the-same-ancestors, but relatedness-by-genetic-overlap).

For myself, I find multilevel selection theory an easier way to think about certain processes. At the level of DNA, there's selection in favour of selfish DNA like transposons, but at the level of the cell there's selection against uncontrolled copy-and-pasting of selfish DNA since it causes the cell to die. At the level of the cell, there's selection in favour of cells which multiply uncontrollably, but at the level of the organism we call that cancer and there's selection against it because the organism dies.

In both of those cases, the higher-level organization has multiple safeguards in place to stop the uncontrolled selfish replication of the lower-level organization. Our cells have systems to silence and stop transposons. In adult cancers, multiple mutations are required to overcome our multiple anti-cancer defense systems. (That's something I'd be interested in exploring if life ever gives me the time: What effect does the possibility of multiple redundant anti-selfishness systems, as in the multiple redundant anti-cancer systems in our own cells, have on the mathematics of the evolution of cooperation?)

The paper which introduced me to multilevel selection and is still a pretty good read is D. S. Wilson and E. O. Wilson's 2007 Rethinking the Theoretical Foundation of Sociobiology:

The problem is that for a social group to function as an adaptive unit, its members must do things for each other. Yet, these group‐advantageous behaviors seldom maximize relative fitness within the social group. The solution, according to Darwin, is that natural selection takes place at more than one level of the biological hierarchy. Selfish individuals might out‐compete altruists within groups, but internally altruistic groups out‐compete selfish groups. This is the essential logic of what has become known as multilevel selection theory.

Richard Dawkins is not impressed by all of this, but D. S. Wilson suggests that he might be behind the times:

Both [Dawkins and E.O. Wilson] fail to recognize that the era of “kin selection vs. group selection” has passed. Most of the important questions can be asked within either framework and can be translated between frameworks.

Once you get into asking all of these questions in terms of humans, my own opinion is that we're still mostly in the age of Just So stories and suggestive analogies. (Maybe punishment of criminals is like an anti-cancer defense. Maybe wars are like competitions between organisms. Maybe... or maybe not.)

It's all complicated by the fact that we rely more on our brains and less on our DNA for our learning and decision-making than most (or all) other organisms. The human brain, I would suggest, is the ultimate example of conservative bet hedging. Instead of evolving to maximize our fitness under certain conditions, our DNA has instead gambled on building a brain which can learn to handle a huge number of conditions. Instead of our DNA dictating what our cooperative and competitive responses will be, it has passed much of that responsibility over to our brains. So far, from an evolutionary perspective, it seems to be working.
posted by clawsoon at 6:11 AM on April 25, 2018 [6 favorites]

In addition to what clawsoon said, it seems to this layman that the idea of "gene-only selection" is complicated by recent discoveries showing that "the gene for X" is usually more like "half a dozen different genes plus environmental factors that produce X".
posted by tobascodagama at 6:30 AM on April 25, 2018 [1 favorite]

My own half-baked supposition (which is surely not original to me): Perhaps it's not possible to build a brain as flexible as ours without making it capable of selflessness and self-destructiveness. Perhaps trying to explain cooperation and altruism in terms of "genes for cooperation" and the evolutionary trade-offs between individual and group advantage is beside the point.

Perhaps the real evolutionary trade-off is not between individual and group selection, but between "build a brain flexible enough to conquer the world" and "build a brain that's not capable of acting against the evolutionary self-interest of its genes." Perhaps if our genes tried to force Haldane's logic on our brains, they could only do so by sacrificing the flexibility that makes our brains an evolutionary win.

In other words, maybe evolutionary trade-offs still matter, but not the ones which our studies of E. coli and ants have primed us to look for.
posted by clawsoon at 6:55 AM on April 25, 2018 [2 favorites]

one group of less-greedy animals outcompeted all its neighbors because they as a group avoided a tragedy of the commons

wait, less-greedy animals aren't necessarily able to defend their resources from invaders. I'm lost there.
posted by eustatic at 7:18 AM on April 25, 2018

eustatic: wait, less-greedy animals aren't necessarily able to defend their resources from invaders. I'm lost there.

I believe that this idea is typically framed in terms of the quantity/quality tradeoff. With everyone selfishly consuming as many resources and producing as many offspring as possible, you'll end up with a numerous but malnourished population. Individuals will be small and weak; they'll be easy prey for a group which has underconsumed its resources and is thus made up of a smaller number of strong, healthy individuals.

I've seen it argued that this effect is most often achieved by selfish individuals within territorial or hierarchical species. If they can, they defend more territory (or, for hierarchical species, make a larger demand on the population's resources) than they strictly need to support their own offspring. That way, their offspring grow bigger and stronger, and they also have more insurance against temporary resource failures.
posted by clawsoon at 9:38 AM on April 25, 2018

It seems evident (to me, anyway) that the environment exerts influence on the gene pool of a sexually reproducing species or a subset of the species confined to a relatively isolated niche. New members are the combinatorics of the two parent organisms, so at a minimum the environment operates on that mini-gene-pool. But as the interrelatedness of individual organisms increase, the more their joint behavior serves to mitigate and distort the environmental forces. To the extent these behaviors are mediated by the gene pool of the group, the environment is influencing the composition of that pool in a recursive loop (and note that other organisms of the same species are part of the environment). In addition, joint behavior influences development of the phenotype, as gene expression is influenced by nurture. And ultimately, the environment interacts with the phenotype rather than directly with the genotype. Reproductive advantage through longer life and increased fertility, which depend on better extraction of resources and protection from predation, illness, and weather threats, are certainly partly a function of the pool of genes being expressed.
posted by Mental Wimp at 11:09 AM on April 25, 2018 [2 favorites]

I feel like this problem could be explained much clearer that the discourse the professor uses. Why should it be weird or surprising that evolutionary algorithms discover locally optimal solutions that are globally suboptimal? That's the Elk story in a nutshell: you have a trait/algorithm that gets bona fide evolved, that ultimately is not good for the Elk species relative to other less-greedy traits of the other (competing) prey species. Does it matter if this is called group-level or multi-level or cultural or perhaps I'd personally prefer the term "meta-evolutionary selection" in the sense that evolution is applying its own techniques on itself? The problem is these abstractions are hard to capture and formalize and use, and it seems like the con-side of the debate believes it's easier to treat them as emergent phenomena, or something.
posted by polymodus at 2:04 PM on April 25, 2018

He is the founder of a new transdisciplinary field of Cliodynamics, which uses the tools of complexity science and cultural evolution to study the dynamics of historical empires and modern nation-states.

I've read this book before.
posted by Apocryphon at 1:37 PM on April 26, 2018 [1 favorite]

"To me, turning all altruism into selfishness with logic is twisted. There is real emotional power in being part of a team."

It's been a long time since I read The Selfish Gene, Extended Phenotype, or The Ancestor's Tale, my takeaway from those together was that there was more power to altruism and cooperation than just emotional, but a very real, physical and evolutionary advantage for it.
posted by GoblinHoney at 3:26 PM on April 26, 2018 [1 favorite]

fwiw...
-The Key to Everything[*]
-A Brand-New Version of Our Origin Story
-George Church on bioengineering (Church Speaks)
-Encore Geoffrey West on why some things stop growing and other things don't
-The canine genome (Dog breeds and humans and genetic diversity and stuff; next FPP)
posted by kliuless at 6:54 PM on April 26, 2018

A Thousand Baited Hooks: One of the interesting things about The Selfish Gene is that only part of it is about how selfish genes are. Most of it is an explanation of how cooperation can occur not just despite genetic selfishness, but because of it.

And yet the book comes to this conclusion:

Be warned that if you wish, as I do, to build a society in which individuals cooperate generously and unselfishly towards a common good, you can expect little help from biological nature. Let us try to teach generosity and altruism, because we are born selfish.

Because Dawkins based his conclusions on early math in evolutionary theory which required, in order to be tractable with calculus:

• infinite population size
• random mating
• additive genetic effects, and
• a single gene for cooperation

and never updated his thinking to more current math and simulations which allowed for smaller populations, non-random mating, and multiple genes for cooperation, he was forced to the conclusion that in a diploid population (i.e. like humans) the selfish would always outcompete the altruistic because individual selection would always work faster than group selection.

The arc of the book is first explaining how cooperation can evolve if you've got clonal individuals, then explaining how cooperation will be destroyed if you've got sex and don't have a relatedness-increasing genetic quirk like haplodiploidy to make up for it.
posted by clawsoon at 3:58 AM on April 27, 2018 [2 favorites]

A Thousand Baited Hooks: One of the interesting things about The Selfish Gene is that only part of it is about how selfish genes are. Most of it is an explanation of how cooperation can occur not just despite genetic selfishness, but because of it.

And yet the book comes to this conclusion:

Be warned that if you wish, as I do, to build a society in which individuals cooperate generously and unselfishly towards a common good, you can expect little help from biological nature. Let us try to teach generosity and altruism, because we are born selfish.

There's no contradiction here, because of a subtle distinction: cooperation versus generous and unselfish cooperation towards a common good. Reciprocal altruism and kin-directed altruism are forms of "cooperation" just as much as purely non-reciprocal altruism; I was responding to an observation that cooperation didn't seem possible without multi-level selection, which is a misunderstanding.

and never updated his thinking to more current math and simulations which allowed for smaller populations, non-random mating, and multiple genes for cooperation, he was forced to the conclusion that in a diploid population (i.e. like humans) the selfish would always outcompete the altruistic because individual selection would always work faster than group selection.

The arc of the book is first explaining how cooperation can evolve if you've got clonal individuals, then explaining how cooperation will be destroyed if you've got sex and don't have a relatedness-increasing genetic quirk like haplodiploidy to make up for it.

I don't get the bit about maths and simulations; it's a while since I read the book, but I just skimmed through it and I can't find any suggestion that it's based on the kind of simplistic maths you mention. He certainly doesn't assume random mating or a single gene for cooperation!

But on "the selfish would always outcompete the altruistic" - this is true (so the argument goes) only for a very specific understanding of "selfish" (i.e. genetically selfish, which is not the same as individually selfish) and a very specific understanding of "altruistic" (where "altruism" means actions benefitting unrelated individuals at some cost to oneself, with no possibility of direct or indirect benefit, i.e. non-kin-directed, non-reciprocal altruism). This certainly doesn't rule out cooperation, which can be entirely consistent with genetic selfishness.
posted by A Thousand Baited Hooks at 6:00 AM on April 27, 2018