"John Edser"  wrote in message
news:cc47dc$ghk$1{at}darwin.ediacara.org...
> > > JM:-
> > > No! No! No!  The correct lesson to take from the "new
group selection"
> > > is that groups have a role ONLY IF they are ephemeral and if the
> > > organisms in those groups do most of their breeding outside the
> > > group.
>
> > GH:-
> > This is far too extreme.  The "new group selection"
allows for this
> > possibility, largely as a result of the work of McCaughley and colleagues;
> > but I am quite certain that D.S. Wilson would not agree with you that this
> > defines the bounds of the "new group selection."
>
> JE:-
> "New" and "old" group selection? A rose
> by a different name remains just a rose...

Except in this case, they have come up with something that is truly
new, a tulip, say, but instead of calling it a tulip they decided
to call it a rose - thus creating no end of confusion.

> In your opinion was the oldest
> form of the "new group selection" Hamilton's
> rule? Would you agree that the most important
> thing all these "new" forms of group selection
> have in common is that they cannot be
> tested against nature whereas the "old"
> form of group selection could be?

Two very good questions.  So good that I can't give short answers.

As to what was the earliest "new group selection" model, I don't
know.  However, it is certainly true that Hamilton's 1964 kin
selection model can be understood as a "new group selection" model,
even though it preceeded the "real" origin of "new group
selection"
by a few years.  I think that Price (1970) and D.S. Wilson (1975)
are the theoretical underpinnings of "new group selection".

The question of testability against nature is even more difficult.
"New group selection" means different things to different people.
Some people like to emphasize that the "new group selection" is
"multi-level selection".  If they mean by this that you can have
multiple distinct processes going on simultaneously, then, in
theory, you ought to be able to test against nature whether the
controversial higher-level processes are actually happening.
Actually performing such a test and interpreting the results
may be difficult, but it ought to be possible.

A lot of people call themselves "new group selectionists" or
"multi-level selectionists" even though they believe in the
"old" classical group selection.  They emphasize that the
hypothetical group selection and old fashioned individual-level
natural selection go on at the same time (as if they imagined
that opposition to classical group selection was based on a
misunderstanding of this aspect of the model).  And they
welcome the theoretical support offered by the Price-Wilson
models, because they think that this new theory also
supports the old models.  It doesn't, IMO, but confusion on
this point is fed by the undeniable fact that some of the
same theorists that promote the new theory are also involved
in exploring the loopholes that remain in the incomplete
"disproof" of the classical theory.

However, some people interpret "multi-level selection" to mean
that a single process or strand of a process can be understood
at any of several different levels.  Wilkins, I think, might
react to this by saying that those different levels of selection
don't "really" exist - that there is one particular level that
is the "best" level for viewing that process, and that anyone
who insists on viewing it at a different level is making a mistake.
(I probably exagerate John W's pig-headedness here, but I actually
have some sympathy for this extreme philosophical "monism").
Other people welcome this kind of "pluralism", and I have som
sympathy for that, as well.

To return to "testability", it should be obvious that the
question of whether it is best to view a particular mechanism
as a case of individual level selection or of group level
selection - this question simply is not a testable empirical
question.  So, to the extent that I call myself a "new group
selectionist", I have to plead guilty, John.  My opinion cannot
be tested against nature.

So now to the inevitable follow-up question: Is Hamilton's
model testable against nature?  Well, the key ingredient
of Hamilton's model is that carriers of the allele for
altruism do better than the general population BECAUSE
they receive more altruism than does the general population.
Furthermore, they receive this excess altruism, not because
they somehow CAUSE the altruists to be nice to them, but
simply because their genomes happen to be correlated with
the genomes of the altruists.  We can certainly test whether
the carriers of a particular gene do, in fact, receive more
altruism.  But determining WHY they receive more altruism
is more problematic.  I hope that addresses the question.
---
þ RIMEGate(tm)/RGXPost V1.14 at BBSWORLD * Info{at}bbsworld.com

---
 * RIMEGate(tm)V10.2áÿ* RelayNet(tm) NNTP Gateway * MoonDog BBS
 * RgateImp.MoonDog.BBS at 7/6/04 6:34:25 AM