I'm a defender of group selection, but I haven't heard some of Robin's arguments for it before, and want to issue some caveats.
> The fact that more species today descended from fragmented habits like rivers, coral reefs, and rainforests, where habitats were smaller, suggests that group selection of species has actually mattered more for DNA than individual selection within species.
Claims about "most species" have serious definitional problems, such as that there are an astounding number of species of beetles, and there are families which have a very small number of species, such as elephants (3), and others with many species, all similar (finches).
That said, Wilson and MacArthur's island theory of biogeography, which they proposed in the 1950s and is now well-validated, shows that the number of species found on an island is a power law of the island's area, with an exponent of about 0.29. This implies that you get exponentially more species produced from a given land area by subdividing it into smaller and smaller islands. And that predicts that fragmented habitats produce many more species per unit area than less-fragmented habitats.
The usual explanation for this power-law relationship does not involve group selection, but I was never entirely happy with that explanation. My intuition says that smaller islands have more speciation events because they have smaller populations, which means population size is more stochastic, which makes group selection events more frequent. Not just because extinctions are more common; also because the brief bursts of co-evolutionary cascades that punctuate punctuated equilibria are more-common since the whole system is more noisy. I just explained this to Gemini, and told it how to model the system using Stuart Kauffmann's NK dynamics and the binomial and Poisson distributions, and it says that approach also produces the same power law (though I haven't yet checked its math, and haven't got the data to check whether the constant exponent is the same). If /that/ is the "true" explanation, then your statement is basically correct.
The other proviso I want to add is that "mattered more for DNA than individual selection" depends on how you measure "mattered more". About 16 years ago i gathered a lot of data on where each HMMER and Pfam pattern (patterns which match evolutionarily related families of genes) first occurred in the evolutionary tree of life, and I found that something like 3/4 of all genes (for which there are HMMER or Pfam patterns; this is a bit problematic because somebody who studied only fungi developed a whole lot of Pfam templates)--anyway, most gene families evolved in bacteria. If we measure what "mattered more for DNA" by counting matching base pairs or amino acid motifs of the coding regions of genes, the answer is going to be that bacterial evolution contributed more to your and my DNA than did the entire evolution of Chordata.
We might well find group selection in bacteria--we definitely find it in one slime mold (not a bacteria, but single-celled), and the evolution of biofilms, in which some bacteria sacrifice themselves to protect the group, sure smells of group selection--but bacterial evolution is very different from eukaryotic evolution, because there is so much lateral gene transfer across species. Bacteria will literally fuck anything that moves.
Australian & American species largely got outcompeted by Old World species once the latter arrived. Similarly, Out-of-Africa humans outcompeted Neandertals, because we'd evolved from much larger effective population sizes, and thus selection was stronger relative to drift, and once agriculture was invented larger populations accelerated natural selection even more. Island species like dodos died out.
"most cultural evolution scholars see group selection as a key force"
I wonder if this is so.
Here, in a single doc, is last year's Paris School alternative to dual-inheritance theory published in Evolution & Human Behavior, followed by commentaries (some of which include dual-inheritance advocates) and authors' response to them:
Also, if memory serves, somewhere near the end of this recent podcast, Robert Boyd seems to hint at some disagreement with Henrich over the extent or importance of cultural group selection:
Dan Sperber once told me that he tried to persuade Dennett that memes are (best understood as) attractors. I guess trying to persuade Dennett of this was never going to work, but I think it’s right.
Maybe Dennett will turn out to be right, but I couldn't help wondering about the relationship between his embrace of memetics and his (to me) overblown panic over "the fragile control systems in our brains" [1] being manipulated by AI [2] -- especially since only a few years earlier, he seemed to be in general agreement with Sperber during this [3] talk on the interactionist theory of reason. (Funny, it was his and Dawkins' supercilious approach to the psychology of religious belief during the heyday of New Atheism that first made me skeptical of memetics.)
What does this mean? In what sense is group selection even a thing with a rate and if so why isn't the rate just the rate at which you cull groups?
If you *could* measure a rate wouldn't it be the rate at which selection pushes features to universality so doesn't high group selection guarantee the exact situation you are worried about with a single shared culture? (maybe we have many groups but if they all agree on X won't they just together act to ensure X is preserved)?
If you just mean selection at the level of a group of organisms of course group selection happens -- multicellular animals are literally groups.
What is actually meant when group selection is denied is something like: we virtually never see selective pressure for a gene which reduces the chance of it's direct carrier reproducing relative to other (not close family) group members but improves the outcomes of group members relative to non-members.
And that claim is probably true. So we can see selective pressure for genes which make our group more successful but we can't explain things like altruistic sacrifice by invoking the group.
"For example, the network feature that most promotes group selection seems to be “modularity”, roughly how many more ties there are within clusters, compared to between clusters. It also matters how similar are people within clusters, how much overlap there is between interaction and emulation networks, how well prestige tracks adaptiveness, how much conformity pressure there is for a behavior, and how much that behavior effects visible outcomes that people care about."
Re. the first sentence, the main pre-requisite for group selection is a correlation between the death rate or survival rate of group members. I don't know if that's what you mean by "ties". When you say "It also matters how similar are people within clusters", that sounds like kin selection, which is the degenerate case of group selection in which it can produce individual altruism, but can be analyzed as individual selection. The rest of the paragraph seems not to come from the group selection literature, and I think you've come up with it yourself, but not told us why. I have no idea what distinction you're making between interaction and emulation networks. Can you explain that paragraph?
> You may have heard that such *group selection* never happens, but that’s wrong. Not only do most cultural evolution scholars see group selection as a key force, group selection also seems to important in DNA evolution, where species are groups.
I've "heard" that (applied to DNA rather than culture) on this very blog https://www.overcomingbias.com/p/group-selectionhtml I've also heard from Greg Cochran on how the math doesn't work, mutations will cause defectors/free-riders to appear in groups who outcompete others.
> The fact that more species today descended from fragmented habits like rivers, coral reefs, and rainforests, where habitats were smaller, suggests that group selection of species has actually mattered more for DNA than individual selection within species.
Most species, or most individuals of any species, or most biomass? And what is the citation for this?
I'm a defender of group selection, but I haven't heard some of Robin's arguments for it before, and want to issue some caveats.
> The fact that more species today descended from fragmented habits like rivers, coral reefs, and rainforests, where habitats were smaller, suggests that group selection of species has actually mattered more for DNA than individual selection within species.
Claims about "most species" have serious definitional problems, such as that there are an astounding number of species of beetles, and there are families which have a very small number of species, such as elephants (3), and others with many species, all similar (finches).
That said, Wilson and MacArthur's island theory of biogeography, which they proposed in the 1950s and is now well-validated, shows that the number of species found on an island is a power law of the island's area, with an exponent of about 0.29. This implies that you get exponentially more species produced from a given land area by subdividing it into smaller and smaller islands. And that predicts that fragmented habitats produce many more species per unit area than less-fragmented habitats.
The usual explanation for this power-law relationship does not involve group selection, but I was never entirely happy with that explanation. My intuition says that smaller islands have more speciation events because they have smaller populations, which means population size is more stochastic, which makes group selection events more frequent. Not just because extinctions are more common; also because the brief bursts of co-evolutionary cascades that punctuate punctuated equilibria are more-common since the whole system is more noisy. I just explained this to Gemini, and told it how to model the system using Stuart Kauffmann's NK dynamics and the binomial and Poisson distributions, and it says that approach also produces the same power law (though I haven't yet checked its math, and haven't got the data to check whether the constant exponent is the same). If /that/ is the "true" explanation, then your statement is basically correct.
The other proviso I want to add is that "mattered more for DNA than individual selection" depends on how you measure "mattered more". About 16 years ago i gathered a lot of data on where each HMMER and Pfam pattern (patterns which match evolutionarily related families of genes) first occurred in the evolutionary tree of life, and I found that something like 3/4 of all genes (for which there are HMMER or Pfam patterns; this is a bit problematic because somebody who studied only fungi developed a whole lot of Pfam templates)--anyway, most gene families evolved in bacteria. If we measure what "mattered more for DNA" by counting matching base pairs or amino acid motifs of the coding regions of genes, the answer is going to be that bacterial evolution contributed more to your and my DNA than did the entire evolution of Chordata.
We might well find group selection in bacteria--we definitely find it in one slime mold (not a bacteria, but single-celled), and the evolution of biofilms, in which some bacteria sacrifice themselves to protect the group, sure smells of group selection--but bacterial evolution is very different from eukaryotic evolution, because there is so much lateral gene transfer across species. Bacteria will literally fuck anything that moves.
Australian & American species largely got outcompeted by Old World species once the latter arrived. Similarly, Out-of-Africa humans outcompeted Neandertals, because we'd evolved from much larger effective population sizes, and thus selection was stronger relative to drift, and once agriculture was invented larger populations accelerated natural selection even more. Island species like dodos died out.
Makes sense.
"most cultural evolution scholars see group selection as a key force"
I wonder if this is so.
Here, in a single doc, is last year's Paris School alternative to dual-inheritance theory published in Evolution & Human Behavior, followed by commentaries (some of which include dual-inheritance advocates) and authors' response to them:
https://www.dropbox.com/scl/fi/n7wka68pli0kqxjw8jhk5/Baumard_Andre_Ecological_approach_to_culture_with_commentaries.pdf?rlkey=aro2x0a41rk3o0womvmg0qd6b&e=3&dl=0
Here's a brief summary:
https://www.hbes.com/reconciling-our-three-traditions-the-ecological-approach-to-culture/
Also, if memory serves, somewhere near the end of this recent podcast, Robert Boyd seems to hint at some disagreement with Henrich over the extent or importance of cultural group selection:
https://epthepod.podbean.com/e/rob-boyd/
Yea I believe he also expressed some doubts about cultural group selection in our podcast with him.
Wonder if Baumard & André would persuade Dawkins to drop memetics?
Dan Sperber once told me that he tried to persuade Dennett that memes are (best understood as) attractors. I guess trying to persuade Dennett of this was never going to work, but I think it’s right.
Maybe Dennett will turn out to be right, but I couldn't help wondering about the relationship between his embrace of memetics and his (to me) overblown panic over "the fragile control systems in our brains" [1] being manipulated by AI [2] -- especially since only a few years earlier, he seemed to be in general agreement with Sperber during this [3] talk on the interactionist theory of reason. (Funny, it was his and Dawkins' supercilious approach to the psychology of religious belief during the heyday of New Atheism that first made me skeptical of memetics.)
[1] https://archive.ph/MqDsh
[2] https://www.youtube.com/watch?v=gh2dgsaNY3A
[3] https://www.youtube.com/watch?v=qXsjWo6K4w0
> feature that most promotes group selection
What does this mean? In what sense is group selection even a thing with a rate and if so why isn't the rate just the rate at which you cull groups?
If you *could* measure a rate wouldn't it be the rate at which selection pushes features to universality so doesn't high group selection guarantee the exact situation you are worried about with a single shared culture? (maybe we have many groups but if they all agree on X won't they just together act to ensure X is preserved)?
If you just mean selection at the level of a group of organisms of course group selection happens -- multicellular animals are literally groups.
What is actually meant when group selection is denied is something like: we virtually never see selective pressure for a gene which reduces the chance of it's direct carrier reproducing relative to other (not close family) group members but improves the outcomes of group members relative to non-members.
And that claim is probably true. So we can see selective pressure for genes which make our group more successful but we can't explain things like altruistic sacrifice by invoking the group.
who owns the pipes between clusters
Here's the part I really don't understand:
"For example, the network feature that most promotes group selection seems to be “modularity”, roughly how many more ties there are within clusters, compared to between clusters. It also matters how similar are people within clusters, how much overlap there is between interaction and emulation networks, how well prestige tracks adaptiveness, how much conformity pressure there is for a behavior, and how much that behavior effects visible outcomes that people care about."
Re. the first sentence, the main pre-requisite for group selection is a correlation between the death rate or survival rate of group members. I don't know if that's what you mean by "ties". When you say "It also matters how similar are people within clusters", that sounds like kin selection, which is the degenerate case of group selection in which it can produce individual altruism, but can be analyzed as individual selection. The rest of the paragraph seems not to come from the group selection literature, and I think you've come up with it yourself, but not told us why. I have no idea what distinction you're making between interaction and emulation networks. Can you explain that paragraph?
> You may have heard that such *group selection* never happens, but that’s wrong. Not only do most cultural evolution scholars see group selection as a key force, group selection also seems to important in DNA evolution, where species are groups.
I've "heard" that (applied to DNA rather than culture) on this very blog https://www.overcomingbias.com/p/group-selectionhtml I've also heard from Greg Cochran on how the math doesn't work, mutations will cause defectors/free-riders to appear in groups who outcompete others.
> The fact that more species today descended from fragmented habits like rivers, coral reefs, and rainforests, where habitats were smaller, suggests that group selection of species has actually mattered more for DNA than individual selection within species.
Most species, or most individuals of any species, or most biomass? And what is the citation for this?