A rant on group selection.

[Nick Thompson]

A couple of people have asked me to write a succinct clarification of the concept of group selection, and having gone to the effort, I thought I might try it out on this group.   Any thoughts?

So individualistic is our age, and so successful have books like Selfish Gene been, getting people to even entertain the notion of group selection is often difficult.    

One way to open the issue up is to start with a clear distinction between vehicles and replicators.  This argument involves some moves that I am uncomfortable with but many people seem to really like it, so try it.  Genes are the only replicators (unless memes have entered the argument).  So a vehicle is any entity whose fate effects the relative distribution of any gene vis a vis its alternative.  So, the argument now becomes, what levels of organization can serve as vehicles.  

Can genes serve as vehicles? Yes in deed.  Any little chunk of DNA that is moving itself around the genome in such a way as to turn up disproportionately in gametes is not only a gene by definition but a vehicle by definition is well.  The famous T-allele of mice is the classic example.

Can individuals serve as vehicles? Yes, also.  If traits are expressed by individuals, and those traits lead to greater reproduction of traitbearers, and this greater reproduction leads to disproportionate distribution of any allele vis a vis its alternative, then individuals are serving as vehicles and selection is taking place at the individual level.  Notice that the fact that differential production of genes serves to mediate individual selection WAS NOT TAKEN AS EVIDENCE against selection at the individual level.  To do so would have been to confuse the distinction between replicators and vehicles.

Can groups serve as vehicles?  You bet!  if traits are expressed by individuals that lead to the expression of traits by groups, and these groups are more productive of individuals that bear group promoting groups than groups without such individuals, and this greater productivity leads to the disproportionate distribution of any allele vis a vis its alternative, then groups are serving as vehicles and selection is taking place at the group level. NOtice that the fact that differential production of individuals (and of genes) serves to mediate group selection WAS NOT TAKEN AS EVIDENCE against selection at the group level.  To do so would have been to confuse the distinction between replicators and different levels of vehicle.  Notice, too, now that selection at both higher levels can be acting simultaneously on the same gene.  If for instance the group promoting trait carried by individuals places them at a disadvantage in comparison within the group, then the total selection, for or against the trait, will be the resolution of its effects at the two levels of organization.   Notice that the two do not need to oppose one another in order to both be effective, although the situations in which they are oppose one another are more dramatic.  

People who hate group selection often do so because they are hard pressed to think of a gene whose additive effects  could make it all the way through the complexities of individual and group development to produce a trait that nature would favorably select.  That you will recognize as the same quandary that I am having with individual selection.  But despite these quandaries, it is not hard to imagine such traits.  One of my favorites is weakness to rhythm.  I can imagine "entrainabillity to a rhythm" as an individual genetic trait which when shared with other individuals of a group leads to higher levels of group coherence and effectiveness.   Another possibility, is male adolescent aggressiveness.   High levels of testosterone in individuals, coupled with the high levels of moralizing typical of adolescents lead in turn to high levels of moralizing aggression in teens ... gang punishment, etc.  It is this weakness in the male adolescent nervous system that is called upon to defend the group against outsiders.  A third example might well be behavioral generality and intelligence.  Notice that for selection to act on well coordinated groups it has to select on some trait, which when aggregated in a group, produces integration and functional differentiation.  Selection is like a coach going into the NFL draft without knowing what position the draftees play.  Such a coach would well advised to pick intelligent trainable athletes.   Tractability is the only trait, which when aggregated, facilitates group functioning, and so is an ideal candidate for group selection.  

Perceptive readers will have already seen at least one reason that this way of making the argument makes me uncomfortable.  The insistence on genes as replicators seems a bit pig headed when individuals seem to be "carrying" group traits from group to group just a surely as genes are carrying individual traits from parent to offspring.  Thus, a better way to think of inheritance when talking about multilevel selection might be fractally -- inheritance occurs more or less at every level or organization.  This move makes the distinction between replicator and vehicle purely relational .. ie, a replicator is just whatever passes the effects of vehicle selection from one generation of vehicle to the next,at any level of organization.  

Nick


Nicholas S. Thompson
Professor of Psychology and Ethology
Clark University
[hidden email]
http://home.earthlink.net/~nickthompson/
[hidden email]



Nicholas S. Thompson
Professor of Psychology and Ethology
Clark University
[hidden email]
http://home.earthlink.net/~nickthompson/
[hidden email]
-------------- next part --------------
An HTML attachment was scrubbed...
URL: /pipermail/friam_redfish.com/attachments/20050228/38d32423/attachment.htm

[Jochen Fromm-2]

Here is my view of group selection:

Group selection is one of the four basic reasons
for cooperation among selfish agents. Agents
are selfish if they act on behalf of "selfish"
genes:

a) Good genes are selfish/egoistic
(see "The Selfish Gene", Richard Dawkins, Oxford University Press, 1976)
b) Good memes are selfless/altruistic
(see "The Meme Machine", Susan Blackmore, Oxford University Press, 1999)

How can an agent acting on behalf of it's own
selfish genes behave selfless if this behavior
reduces the reproductive success ? The answer
is reciprocity (mutual help or support). There are
four fundamental forms of reciprocity:

1) direct reciprocity (agent-agent)
2) indirect reciprocity (agent-group)
3) kin selection (gene-gene)
4) group selection (gene-meme)

As the name says, group selection is natural selection
acting on groups, rather than individuals. It is one of
the four basic reason for cooperation among selfish
agents through an interplay between memes and genes:
altruistic memes (ideas, rules or social norms) which
characterize a certain group increase the fitness of the
genes in this group, and these genes in turn increase the
fitness of the corresponding memes. Both replicators,
genes and memes, share the group as a survival vehicle
in this case and use each other to survive (or to
mutually increase their reproductive success).