Error: Most of human history is a recounting of group selection in humans. Every time one group of people displaces another group by virtue of superior technology or social organization, that’s group selection.
Having a belief in, or at least openness to, group selection, is one of my rationality tests.
In related news, this weeks Science has the clearest demonstration of group selection that I’ve seen: The ability to self-pollinate in plants gives individuals a great reproductive advantage; but also increases the likelihood of the entire species going extinct. The presence of a feature (self-pollination) that provides an advantage to the individual, provides a disadvantage to the species, that causes species-level selection.
Most of human history is a recounting of group selection in humans. Every time one group of people displaces another group by virtue of superior technology or social organization, that’s group selection.
That is one definition of “group selection”. However, there is another definition—according to which “group selection” must refer to a different theory from “individual selection”—a theory that makes different predictions. For that you would need to show that the genetic traits that led to technological mastery benefited groups in a way that was systematically different from the way that they benefited the individuals that composed those groups.
I think it suffices to show that selection can operate at the level of the group. Even if all of the traits involved provide some advantage to individuals, if they also provide an advantage to the group, then group-level selection needs to be considered.
It is more interesting if you can show that a trait that does not confer an advantage to an individual, has an effect on group selection. But it is an unreasonable bias to demand that group selection requires traits that do not provide any advantage to an individual, and yet at the same time not insist that the theory of individual selection requires traits that do not provide any advantage to the group.
I should clarify—“group selection” connotes what Tim is describing: Selection for altruistic traits in individuals, by selection at the group level. That’s because, historically, group selection has been invoked only to explain things that individual selection can’t.
However, this has led to people excluding selection at the group level from models and simulations, because “group selection bad”.
This is something of a quibble, but you really shouldn’t think of species-level selection as a kind of group selection. In both group and individual selection, it is the species that evolves. But in species-level selection, the species does not evolve. It is selected—it either lives or dies.
Another key difference—the usual argument against group selection is that it is ineffective since individual selection is a stronger force. That is, individual selection is said to push harder and change the species more than does group selection. But comparing species-level selection and individual selection, it makes no sense to say that one is more powerful than the other. They are playing different games.
This is something of a quibble, but you really shouldn’t think of species-level selection as a kind of group selection. In both group and individual selection, it is the species that evolves. But in species-level selection, the species does not evolve. It is selected—it either lives or dies.
Sorry, but I think this is completely wrong. Species-level selection isn’t “like” group selection. It is group selection. In group selection, groups are selected for or against. That is the mechanism for group selection. That is the mechanism for group initially described by Darwin in chapter 4 of Descent of Man, and defended by Edward Wilson. It just happens not to be the straw-man depiction used by some opponents of group selection. They chose to ignore selection at the group level because it is easier to rebut group selection if you first assume that it doesn’t happen.
This is something of a quibble, but you really shouldn’t think of species-level selection as a kind of group selection. In both group and individual selection, it is the species that evolves. But in species-level selection, the species does not evolve. It is selected—it either lives or dies.
Just because we are dealing with one individual, that doesn’t mean it doesn’t evolve. Check with the definitions of the term “evolution”—they (mostly) refer to genetic change over time. You could argue that they also (mostly) talk about a “population”—and one individual doesn’t qualify as a “population”—but if you think through that objection, it too is essentially wrong.
Perplexed: In both group and individual selection, it is the species that evolves. But in species-level selection, the species does not evolve. It is selected—it either lives or dies.
Tim: Just because we are dealing with one individual, that doesn’t mean it doesn’t evolve.
Uh, I’m pretty sure I just stated that an individual—the species—does evolve. It evolves by way of organism-level or group-level selection. It just doesn’t evolve by going extinct or not.
As for whether one individual qualifies as a population, I’ve thought about that and completely failed to imagine a population of one individual evolving by way of the standard mechanisms of evolutionary population genetics. That kind of population cannot evolve by differential birth, death, or migration. (I suppose it can change by mutation).
The thing you have forgotten in trying to extrapolate the meaning of ‘population’ in this way is that the essential feature of a biological evolving population is that its size is not fixed and its membership changes in time, whereas a population of exactly one entity by definition does not change its membership count in time.
Now, I will agree that a population of entities (say, the population of biological species within a genus) can evolve by selection even though its membership count occasionally fluctuates through having only a single individual. The genus does evolve. But the evolution of the genus as a population of individual species and the evolution of the component species as (possibly structured into groups) populations of
individual organisms are conceptually distinct processes.
But here is not the place to continue this discussion. If you wish, please bring it up on sbe, and Dr. Hoeltzer can join in. I think he is getting probably lonely over there since we left, and the newsgroup is dominated by John, Tom, and Peter.
Perplexed: In both group and individual selection, it is the species that evolves. But in species-level selection, the species does not evolve. It is selected—it either lives or dies.
Tim: Just because we are dealing with one individual, that doesn’t mean it doesn’t evolve.
Uh, I’m pretty sure I just stated that an individual—the species—does evolve. It evolves by way of organism-level or group-level selection. It just doesn’t evolve by going extinct or not.
“Does NOT evolve” was the term you used. However, with your clarification, it now looks as though this was mostly just a misunderstanding.
IMO, it is mostly OK to think of species level selection as a type of group selection—where the “groups” are species. Maybe there are some meanings of “group selection” for which this is bad—but I would say: mostly OK.
Yes, a population of 1 changes by mutation. Self-directed evolution is an example of that.
If you A) insist on a population having more than 1 member and B) define evolution in terms of genetic change in populations, then the conclusion is that one big organism would no longer be “evolving” when it changed—which I think would be a totally absurd conclusion—a sign that you had got into a terminology muddle.
That may not be a big deal for today’s organic evolution—but it makes a big difference for the study of cultural evolution. There, populations with only 1 member are much more common.
But comparing species-level selection and individual selection, it makes no sense to say that one is more powerful than the other. They are playing different games.
They are both attempting to influence the same germ line. They are both attempting to influence the same set of traits. It makes reasonably good sense to look at a trait—and to ask whether it is more for the benefit of the individual or the species.
For example, one such trait might be: a love of swimming. That might be bad for an individual (drowning), but good for the species (island speciation).
My apologies. Most theorists say species selection is a subclass of group selection; but Stephen J. Gould says it is not. See the long explanation here.
But comparing species-level selection and individual selection, it makes no sense to say that one is more powerful than the other. They are playing different games.
That is true if you’re talking about features that groups have and individuals don’t, or traits that aren’t inherited genetically. But all the literature on group selection is about the competition between individual and group (including species) selection, within the same game of selecting genes.
I appreciate the effort you are putting into this, but I fear the terminological and theoretical confusion regarding group selection run far too deep. One enthusiastic person is not going to straighten things out in a forum where evolutionary biology is not the central focus. And now that academian has weighed in, the cause is hopeless. ;)
I agree with you (and Tim) that Eliezer’s opposition to group selection was a bit naive and under-informed. But not completely wrong-headed. Many incorrect arguments in favor of group selection have been made over the years. A lot of them were incorrect because they simply did not work. Others were “epistemologically incorrect” because, though they worked, they could be reinterpreted more “parsimoniously” as individual-level selection.
D.S. Wilson’s “Truth and Reconciliation” blog series strikes me as an example of extremely dishonest labeling. What he is really saying is that if everyone who disagrees with him would just accept his version of the truth, then reconciliation will take place. And his book “Unto Others” strikes me as even more dishonest. He defines “group selection” extremely broadly, provides examples of corner cases in which his “trait group selection” mechanism works, and then (here is the dishonest part) claims that if group selection works even in this extreme case, then it will obviously work in other cases.
Then he proceeds to discuss the case which every non-professional has in mind when he thinks of group selection—human evolution with groups = tribes, group death = tribe extinction, and group birth = split-up of a successful and populous tribe. The trouble is that the math of group selection really doesn’t work in this case.
The only cases I know of where the group selection models do work are (1) Species level selection (Gould/Eldredge), examples like your non-selfing plants; and (2) the examples that Wilson gives in which “groups” are rather short-lived entities which “succeed” by keeping their members alive for a while and then returning them safely to the general population, where the individuals reproduce. A good example of a group that Wilson might use as an example of trait-group selection would be a flock of geese conducting a seasonal migration. Such a group might be selected against if it got seriously lost, or blundered into a tornado, or suffered some other collective catastrophe.
A human hunting party is another example of a “group” such that the mathematics of group selection works. A human tribe of hunter-gatherers is not, unless it is so reproductively isolated from other tribes so as to qualify as a species. I’m pretty sure that this degree of isolation (less than two cross-tribe matings per generation) has never held over any long period of time in human history.
But group selection for cultural traits is another question. If genes get transferred between tribes, but memes do not, then selection at the level of tribes may well help to determine the course of human cultural/memetic evolution.
Well, I seem to have provided you with a long response, which, unlike your own efforts, does not include any links/citations. Sorry about that. You are under no obligation to trust or believe me on this stuff. I will merely assert that I (and tim_tyler as well) have been a serious amateur enthusiast for evolutionary theory for many years. Clearly, you have been too. I do recommend though, that you take a second look at D.S. Wilson’s work in light of my criticisms. He really is pulling something of a bait-and-switch. See if you agree.
It sounds as though I like multi-level selection a bit more than you do.
Evidence from our own species suggests habitat variation can cause significant morphological differences (despite gene flow) which selection can then act upon.
I also find things like this one interesting:
“Senescence as an adaptation to limit the spread of disease”
It sounds as though I like multi-level selection a bit more than you do.
I think so. I’m not quite so purist as Dawkins, but I am pretty close. But I do realize that it is not really an empirical scientific question. It is really simply a matter of what kind of models you prefer. Most cases in which group selection models work can also be explained just as well by individual-level selection or kin-selection.
Speaking of which:
I also find things like this one interesting:
“Senescence as an adaptation to limit the spread of disease”
Yes. Very interesting. Red Queen strikes again. But since they are already thinking about Bill Hamilton, why don’t they take the further step and realize that the senescent death of an old individual not only reduces the population density for the benefit of the group—the death specifically is beneficial to those individuals in the group who are the most immunologically similar to the deceased.
In other words, this mechanism ain’t Red Queen + Group Selection; it is Red Queen + Kin Selection.
Their model exhibits locality (with limited diffusion—V.N. or 5x5 neighbourhood) as well.
So: a death benefits kin not just through immunological similarity—but also because neighbours are likely to be kin—and death takes an adjacent pathogen load out of circulation.
In related news, this weeks Science has the clearest demonstration of group selection that I’ve seen: The ability to self-pollinate in plants gives individuals a great reproductive advantage; but also increases the likelihood of the entire species going extinct. The presence of a feature (self-pollination) that provides an advantage to the individual, provides a disadvantage to the species, that causes species-level selection.
I realize your views may have changed by now, but isn’t that obviously caused by Red Queen effects? Just like all other sexual reproduction?
Error: Most of human history is a recounting of group selection in humans. Every time one group of people displaces another group by virtue of superior technology or social organization, that’s group selection.
Having a belief in, or at least openness to, group selection, is one of my rationality tests.
In related news, this weeks Science has the clearest demonstration of group selection that I’ve seen: The ability to self-pollinate in plants gives individuals a great reproductive advantage; but also increases the likelihood of the entire species going extinct. The presence of a feature (self-pollination) that provides an advantage to the individual, provides a disadvantage to the species, that causes species-level selection.
That is one definition of “group selection”. However, there is another definition—according to which “group selection” must refer to a different theory from “individual selection”—a theory that makes different predictions. For that you would need to show that the genetic traits that led to technological mastery benefited groups in a way that was systematically different from the way that they benefited the individuals that composed those groups.
I think it suffices to show that selection can operate at the level of the group. Even if all of the traits involved provide some advantage to individuals, if they also provide an advantage to the group, then group-level selection needs to be considered.
It is more interesting if you can show that a trait that does not confer an advantage to an individual, has an effect on group selection. But it is an unreasonable bias to demand that group selection requires traits that do not provide any advantage to an individual, and yet at the same time not insist that the theory of individual selection requires traits that do not provide any advantage to the group.
I should clarify—“group selection” connotes what Tim is describing: Selection for altruistic traits in individuals, by selection at the group level. That’s because, historically, group selection has been invoked only to explain things that individual selection can’t.
However, this has led to people excluding selection at the group level from models and simulations, because “group selection bad”.
This is something of a quibble, but you really shouldn’t think of species-level selection as a kind of group selection. In both group and individual selection, it is the species that evolves. But in species-level selection, the species does not evolve. It is selected—it either lives or dies.
Another key difference—the usual argument against group selection is that it is ineffective since individual selection is a stronger force. That is, individual selection is said to push harder and change the species more than does group selection. But comparing species-level selection and individual selection, it makes no sense to say that one is more powerful than the other. They are playing different games.
Sorry, but I think this is completely wrong. Species-level selection isn’t “like” group selection. It is group selection. In group selection, groups are selected for or against. That is the mechanism for group selection. That is the mechanism for group initially described by Darwin in chapter 4 of Descent of Man, and defended by Edward Wilson. It just happens not to be the straw-man depiction used by some opponents of group selection. They chose to ignore selection at the group level because it is easier to rebut group selection if you first assume that it doesn’t happen.
Can you provide a reference for that usage?
Just because we are dealing with one individual, that doesn’t mean it doesn’t evolve. Check with the definitions of the term “evolution”—they (mostly) refer to genetic change over time. You could argue that they also (mostly) talk about a “population”—and one individual doesn’t qualify as a “population”—but if you think through that objection, it too is essentially wrong.
Uh, I’m pretty sure I just stated that an individual—the species—does evolve. It evolves by way of organism-level or group-level selection. It just doesn’t evolve by going extinct or not.
As for whether one individual qualifies as a population, I’ve thought about that and completely failed to imagine a population of one individual evolving by way of the standard mechanisms of evolutionary population genetics. That kind of population cannot evolve by differential birth, death, or migration. (I suppose it can change by mutation).
The thing you have forgotten in trying to extrapolate the meaning of ‘population’ in this way is that the essential feature of a biological evolving population is that its size is not fixed and its membership changes in time, whereas a population of exactly one entity by definition does not change its membership count in time.
Now, I will agree that a population of entities (say, the population of biological species within a genus) can evolve by selection even though its membership count occasionally fluctuates through having only a single individual. The genus does evolve. But the evolution of the genus as a population of individual species and the evolution of the component species as (possibly structured into groups) populations of individual organisms are conceptually distinct processes.
But here is not the place to continue this discussion. If you wish, please bring it up on sbe, and Dr. Hoeltzer can join in. I think he is getting probably lonely over there since we left, and the newsgroup is dominated by John, Tom, and Peter.
“Does NOT evolve” was the term you used. However, with your clarification, it now looks as though this was mostly just a misunderstanding.
IMO, it is mostly OK to think of species level selection as a type of group selection—where the “groups” are species. Maybe there are some meanings of “group selection” for which this is bad—but I would say: mostly OK.
Yes, a population of 1 changes by mutation. Self-directed evolution is an example of that.
If you A) insist on a population having more than 1 member and B) define evolution in terms of genetic change in populations, then the conclusion is that one big organism would no longer be “evolving” when it changed—which I think would be a totally absurd conclusion—a sign that you had got into a terminology muddle.
That may not be a big deal for today’s organic evolution—but it makes a big difference for the study of cultural evolution. There, populations with only 1 member are much more common.
They are both attempting to influence the same germ line. They are both attempting to influence the same set of traits. It makes reasonably good sense to look at a trait—and to ask whether it is more for the benefit of the individual or the species.
For example, one such trait might be: a love of swimming. That might be bad for an individual (drowning), but good for the species (island speciation).
My apologies. Most theorists say species selection is a subclass of group selection; but Stephen J. Gould says it is not. See the long explanation here.
That is true if you’re talking about features that groups have and individuals don’t, or traits that aren’t inherited genetically. But all the literature on group selection is about the competition between individual and group (including species) selection, within the same game of selecting genes.
I appreciate the effort you are putting into this, but I fear the terminological and theoretical confusion regarding group selection run far too deep. One enthusiastic person is not going to straighten things out in a forum where evolutionary biology is not the central focus. And now that academian has weighed in, the cause is hopeless. ;)
I agree with you (and Tim) that Eliezer’s opposition to group selection was a bit naive and under-informed. But not completely wrong-headed. Many incorrect arguments in favor of group selection have been made over the years. A lot of them were incorrect because they simply did not work. Others were “epistemologically incorrect” because, though they worked, they could be reinterpreted more “parsimoniously” as individual-level selection.
D.S. Wilson’s “Truth and Reconciliation” blog series strikes me as an example of extremely dishonest labeling. What he is really saying is that if everyone who disagrees with him would just accept his version of the truth, then reconciliation will take place. And his book “Unto Others” strikes me as even more dishonest. He defines “group selection” extremely broadly, provides examples of corner cases in which his “trait group selection” mechanism works, and then (here is the dishonest part) claims that if group selection works even in this extreme case, then it will obviously work in other cases.
Then he proceeds to discuss the case which every non-professional has in mind when he thinks of group selection—human evolution with groups = tribes, group death = tribe extinction, and group birth = split-up of a successful and populous tribe. The trouble is that the math of group selection really doesn’t work in this case.
The only cases I know of where the group selection models do work are (1) Species level selection (Gould/Eldredge), examples like your non-selfing plants; and (2) the examples that Wilson gives in which “groups” are rather short-lived entities which “succeed” by keeping their members alive for a while and then returning them safely to the general population, where the individuals reproduce. A good example of a group that Wilson might use as an example of trait-group selection would be a flock of geese conducting a seasonal migration. Such a group might be selected against if it got seriously lost, or blundered into a tornado, or suffered some other collective catastrophe.
A human hunting party is another example of a “group” such that the mathematics of group selection works. A human tribe of hunter-gatherers is not, unless it is so reproductively isolated from other tribes so as to qualify as a species. I’m pretty sure that this degree of isolation (less than two cross-tribe matings per generation) has never held over any long period of time in human history.
But group selection for cultural traits is another question. If genes get transferred between tribes, but memes do not, then selection at the level of tribes may well help to determine the course of human cultural/memetic evolution.
Well, I seem to have provided you with a long response, which, unlike your own efforts, does not include any links/citations. Sorry about that. You are under no obligation to trust or believe me on this stuff. I will merely assert that I (and tim_tyler as well) have been a serious amateur enthusiast for evolutionary theory for many years. Clearly, you have been too. I do recommend though, that you take a second look at D.S. Wilson’s work in light of my criticisms. He really is pulling something of a bait-and-switch. See if you agree.
Thanks for your 2p on D.S. Wilson’s Unto Others.
It sounds as though I like multi-level selection a bit more than you do.
Evidence from our own species suggests habitat variation can cause significant morphological differences (despite gene flow) which selection can then act upon.
I also find things like this one interesting:
“Senescence as an adaptation to limit the spread of disease”
Josh Mitteldorf , John Pepper
http://www.mathforum.org/~josh/Epidemics-JTB.pdf
I think so. I’m not quite so purist as Dawkins, but I am pretty close. But I do realize that it is not really an empirical scientific question. It is really simply a matter of what kind of models you prefer. Most cases in which group selection models work can also be explained just as well by individual-level selection or kin-selection.
Speaking of which:
Yes. Very interesting. Red Queen strikes again. But since they are already thinking about Bill Hamilton, why don’t they take the further step and realize that the senescent death of an old individual not only reduces the population density for the benefit of the group—the death specifically is beneficial to those individuals in the group who are the most immunologically similar to the deceased.
In other words, this mechanism ain’t Red Queen + Group Selection; it is Red Queen + Kin Selection.
Yes: sex and death!
Their model exhibits locality (with limited diffusion—V.N. or 5x5 neighbourhood) as well.
So: a death benefits kin not just through immunological similarity—but also because neighbours are likely to be kin—and death takes an adjacent pathogen load out of circulation.
I realize your views may have changed by now, but isn’t that obviously caused by Red Queen effects? Just like all other sexual reproduction?