Re: Article: Group selection, a theory whose time has come...again

Bill Morse wrote:

John Harshman wrote:


Bill Morse wrote:


John Harshman wrote:



Bill Morse wrote:



John Harshman wrote:




Bill Morse wrote:




John Harshman wrote:





Bill Morse wrote:





John Harshman wrote:






Perplexed in Peoria wrote:






"John Harshman" <jharshman.diespamdie@xxxxxxxxxxx> wrote in
message news:15L5j.69135$RX.48697@xxxxxxxxxxxxxxxxxxxxxxxxxxxxx






Perplexed in Peoria wrote:







"John Harshman" <jharshman.diespamdie@xxxxxxxxxxx> wrote in
message news:j_H5j.51778$eY.22266@xxxxxxxxxxxxxxxxxxxxxxxxxxxxx







dkomo wrote:

[quoting Sterelny and Griffiths]







"Wilson argued that the trait group is a unit of selection.
Trait groups composed of altruistic animals can outcompete
trait groups of
selfish animals. This can be true even if *within every trait
group* the selfish animals are out-competing the altruistic
individuals."

Can it? How? If the selfish animals outcompete the altruistic
ones, won't altruism disappear in the group?


The trick seems to be that the trait groups don't persist all
that long,
as compared to organism lifetimes. That way you can randomly
reform trait groups (perhaps by fissioning the successful ones),
and simply from the sampling variance, you may get some trait
groups that are more altruistic than the population or
subpopulation from
which you sampled. Also, of course, some that are less
altruistic. But if the parameters are set up so that highly
altruistic groups have a big enough advantage, then the
frequency of altruism may rise in the population as a whole.

The thing is, when most people think of group selection, they
are
thinking of groups that live much longer than individuals.
Things
like human 'tribes'. And getting the parameters to favor the
spread of altruism in a population composed of long-lived,
mostly inbred groups (the picture most people have) is much more
difficult. Almost impossible.


Thus sounds almost sensible to me. How realistic are the model
parameters you need to make it work?


Fairly realistic, IMHO.

IIRC, there is some discussion of a working model in the Sober and
Wilson book, presumably with citations leading to more details. I
seem to recall reading some of the original trait-group papers, as
well. My current understanding of these models and the importance
that group lifetimes be short comes from those sources. Also,
IIRC, those sources contained at least one example in which the
groups persist for a time even shorter than an organism lifetime -
that is, during the life-history of the organism, one phase of its
life (and one important aspect of its survival selection) is spent
in groups.

Sounds like you have to be very careful about the biological
realities here. Trait groups can come very close to being family
groups, and then we're back to kin selection. I'm not sure about
beavers, but in the average cooperative group in birds, only one
female gets to lay any eggs, and often only one male fertilizes
them, and most helpers are offspring from previous years. Further,
the helpers gain other individual benefits from being present in
the territory, among them being a chance to inherit.

So I'm interested, but cautious.

Now if we had a cooperative group of unrelated individuals in which
every member had a shot at reproduction which would be enhanced by
group success, that would be interesting.


The thought just occurred to me. Why wouldn't a monogamous pair be
such a group?


Indeed it would be. So never mind, it wouldn't be interesting. The
evolution of pairs is covered just fine by individual selection.


Arguably, monogamy could not appear by individual selection because it
would
be subject to cheaters. Quoting one J. Harshman, "If the selfish
animals outcompete the altruistic ones, won't altruism disappear in
the group?" Why doesn't this apply to monogamous couples?

Monogamy *is* subject to cheaters. Most monogamous species that have
been studied have a fair proportion of cheating. There are various
mechanisms to punish them too. It's a very interesting and complex
field, as I've said. But I see nothing that group selection is needed
for here.


I am not yet convinced. Yes of course cheating is well known in
monogamous species. But how do the mechanisms arise to control cheating
if the monogamy doesn't predate the cheating?

I don't understand this point. Monogamy could be considered a mechanism
to control promiscuity, so promiscuity must by your logic predate
monogamy.

Per Trivers, monogamy is the expected result of equal parental
investment. Where do you get this "mechanism to control promiscuity"? And
it is clear
that promiscuity is ancestral and monogamy derived, unless you are
maintaining that sponges are monogamous.

I would say rather that the transitions between mating systems, in all
directions, have happened many times, so it doesn't pay much to consider
what the very first mating system was.


Except in each clade in which monogamy evolved, it evolved from
promiscuity. So in each case the mechanism to control cheating had to
evolve before monogamy could evolve.

I'd be interested to know the evidence for your first claim, and the
logical reason for your second.

It would seem to me that there are very few species in which equal
parental investment is required by biology, rather than as a result of
negotiation.

I'm not sure I disagree, but I don't understand the statement. There are
species (e.g. penguins) in which the young will not survive if either
parent dies. I am sure there are many others where the young will not
survive as well with one parent but can still survive. Now clearly the
underlying biology significantly affects the family system that is
selected. Why are penguins monogamous and ringed seals (which share a
similar niche) not? Because, as mammals, the initial parental investment of
female ringed seals is so high. Why is male-only parental care much more
common in fish than in other vertebrates? Because the female can lay eggs
and swim away, leaving the male holding the bag.

Isn't that true for most vertebrates other than mammals?

He hasn't fertilized the
eggs yet, so he has more to lose at that point than the female.

Why? Fertilization is hardly a big investment for most males, even fish.

And don't mechanisms to prevent
cheating indicate that the phenomenon being studied is in fact group
selection?

Why? A mechanism to control cheating is presumably advantageous to the
cheatee.

But it is presumably disadvantageous to the cheater, which remember is
the same species. It is only if both potential cheatee and potential
cheater do better if cheating is controlled that the mechanism will be
selected - in
other words if the selection is operating at the level of the group (in
this case a group of two).

That doesn't follow at all. Conflicts between sexes are well known, and
as long as males and females have slightly different genomes (which of
course they generally do), both sexes can be selected separately. They
can even be selected when the sexes aren't genetically different, but it
takes a bit more work.

Yes they _can_ be, but it takes some work - you have to either get the trait
to appear on a sex chromosome, or you have to get its expression controlled
by a sex hormone.

Which appears to be easy to do.

And you have to do all this before you evolve monogamy,
because you have to have a trait that benefits the cheatee more than it
handicaps the cheater in order to prevent a cheater from undermining
selection for monogamy.

I really don't understand this, given that cheating is nearly ubiquitous
in socially monogamous species.

Note that we have wandered here from the arguments given in the
Wilson and Wilson article, but I still find it an interesting question.
Especially since monogamy occurs among many taxa, it would seem that the
prerequisites for group selection are also widespread- which would make
group selection much more feasible in principle.

*If* we allow your assumption that monogamy implies group selection, for
which I see no justification.

No, even if monogamy is simply an example of reciprocal altruism, the
fact that the mechanisms exist to control cheating means that true group
selection is possible.

It would seem to me that in order for there to be true group selection,
we would require that group advantage be contrary to individual
advantage. I see no indication of that here.

You missed my point. Yes, group advantage has to be contrary to individual
advantage in order to qualify as group selection. But even if monogamy is
reciprocal altruism, it requires the ability to recognize another as a
group member, it requires the ability to coordinate behavior with that
other, and it requires the ability to punish cheating.

By the first presumably you mean the ability to recognize a mate. Why
not just say that? Why is punishment of cheating necessary for monogamy?
One can in fact model social monogamy quite well just by individuals
maximizing their own fitness.

Yet these abilities
are exactly the prerequisites for true group selection using
the "classical" drivers of kin selection and reciprocal altruism.

I wouldn't consider either of those to be group selection.

So we
should expect group selection to be possible in any clade in which we see
monogamy.

I really missed your point that time. It seems merely a series of non
sequiturs.

.

Relevant Pages

  • Re: Article: Group selection, a theory whose time has come...again
    ... Trait groups composed of altruistic animals can outcompete ... frequency of altruism may rise in the population as a whole. ... The thing is, when most people think of group selection, they ... Monogamy *is* subject to cheaters. ...
    (talk.origins)
  • Re: Article: Group selection, a theory whose time has come...again
    ... Cheaters will decrease the fitness of other individuals while attempting to enhance their own, so in the end the fitness of the cheater will decrease as well, hence cheaters will be selected against by group selection. ... kin selection and reciprocal altruism models contra group selection that I "don't understand natural selection." ... Groups and Superorganisms" there is a fairly long and detailed discussion of group selection and the beaver example is used in various spots. ... Trait groups are groups of organisms, each of which feels the influence of the others with respect to some trait. ...
    (talk.origins)
  • Re: Article: Group selection, a theory whose time has come...again
    ... Trait groups composed of altruistic animals can outcompete trait ... of altruism may rise in the population as a whole. ... one important aspect of its survival selection) is spent in groups. ... Monogamy *is* subject to cheaters. ...
    (talk.origins)
  • Re: Article: Group selection, a theory whose time has come...again
    ... Trait groups composed of altruistic animals can outcompete trait ... of altruism may rise in the population as a whole. ... one important aspect of its survival selection) is spent in groups. ... Monogamy *is* subject to cheaters. ...
    (talk.origins)
  • Re: Article: Group selection, a theory whose time has come...again
    ... groups composed of altruistic animals can outcompete trait groups ... then the frequency of altruism may rise ... The thing is, when most people think of group selection, they are ... Monogamy *is* subject to cheaters. ...
    (talk.origins)
Loading