Saturday, April 18, 2009

We're Getting Close

The Huff Post now has further articles from Sloan Wilson concerning group selection. The latest essentially asks and answers the following: "What is the result of the modified haystack model? It turns out that altruism can evolve by group selection - even when the altruistic gene is initially rare in the total population. The model that led to the rejection of group selection is favorable for group selection after all."

I made a series of comments that I've decided to post here to remind myself I may be getting somewhere with my own take on the general subject (these had to be done in increments due to word limits imposed per post):

royniles
If in fact there is no such thing or nothing equivalent to either a selfish gene or an altruistic gene, but that these so-called traits rise from a combination of other "genes" that recombine differently in different types of groups and environments to achieve different results, then the whole modeling structure and the math devised to measure and predict its development become useless.
In my view, you have mistaken strategies that emerge in life forms as representing some sort of fixed algorithmic structures, readily heritable with predictable results that affect circumstances more than circumstances effect them. A bit too much like evolutionary psychology fantasma or meme mythologica. 
Posted 01:33 PM on 04/17/2009

royniles
Here's some excerpts (and my comments) from an article that references some work connected with your theories that shows to me some serious problems with assumptions:


"Why it Pays for Cheaters to Punish Other Cheaters
A new theory for why we put up with adulterers, steroid-using athletes and the mafia
ByMarina Krakovsky

It's the altruism paradox: If everyone in a group helps fellow members, everyone is better off yet as more work selflessly for the common good, cheating becomes tempting, because individuals can enjoy more personal gain if they do not chip in. But as freeloaders exploit the do-gooders, everybody's payoff from altruism shrinks.


(My comment: This presumes that there are just these two main alternatives, and even with that, not several mixtures of these operating successfully in groups.)


All kinds of social creatures, from humans down to insects and germs, must cope with this problem; if they do not, cheaters take over and leech the group to death. 


(My comment: Not so. Cheaters have to mimic cooperators and hide their strategies to succeed - if they proliferate, they can't stay hidden and can't receive the cooperation that depends on two-way trust to function.)
Posted 03:09 AM on 04/18/2009

royniles
More excerpts: 
What keeps the selfish punishers themselves from overexploiting the group? (Sloan) Wilson readily acknowledges this limitation of the selfish punishment model. Although selfish punishers allow cooperators to gain a foothold within a group, thus creating a mix of cheaters and cooperators, "there's nothing telling us that that mix is an optimal mix," he explains. The answer to that problem, he says, is competition not between individuals in a group but between groups. That is because whereas selfishness beats altruism within groups , altruistic groups are more likely to survive than selfish groups. So although selfish punishment aids altruism from within a group, the model also bolsters the idea of group selection, a concept that has seen cycles of popularity in evolutionary biology.

(My comment: That's basically silly, since all punishment is selfish. How this becomes part of a mechanism for group selection is not at all clear. Because first such a group has to first cooperate openly with other cheaters, and they will need to trust the members that will make up the dominant force needed to succeed as a competing entity. And rather than cheaters, they will have become rebels and/or predators.)
Posted 03:54 AM on 04/18/2009

David Sloan Wilson - Huffpost Blogger
Thanks to royniles for his interesting comments. Quickly...

1) The Scientific American article that he quotes profiles the work of my own recently minted PhD student, Omar Eldakar, who has developed a concept called selfish punishment that results in a mix of selfish and altruistic individuals on the basis of within-group selection. This mix is not necessarily optimal for the group, however. 

2) More generally, punishment is itself a form of altruism when the costs are born by the punisher and the benefits of social control are shared by everyone in the group. Economists call this a 2nd order public good. Causing others to provide a public good by rewarding and punishing them is itself a public good. 

3) It is a common modeling strategy to begin with the simplest possible assumptions and add complexity as needed. I have shown in numerous publications that more complicated models (e.g., genes at multiple loci with epistatic interactions) can shift the balance between levels of selection either way. 

Please remember the most basic point that I am trying to make in T&R VIII and IX: The sweeping claim that group selection is theoretically implausible, and that the search for plausible models had been exhausted by the 1980's, is ridiculous. I don't think that royniles disagrees with me on this point.
Posted 10:10 AM on 04/18/2009

royniles
No, I don't disagree with you there. I just disagree on the dynamics. There was another segment of the article that read: 
'What is more, altruism sometimes evolves without selfish punishment. In a software simulation, Eldakar and Wilson have found that as the cost of punishing cheaters falls, so do the number of selfish punishers. "When punishment is cheap, lots of people punish," Wilson explains. And among humans, there is no shortage of low-cost ways to keep others in line from outright ostracism to good old-fashioned gossip.'


My comment was to be: Note no mention at all of trust as a factor, and no details as to the nature of punishment, which always involves something that signals the withholding of trust, actively or passively. Even the so-called cheaters will join in with punishment because they are still posing as trustworthy and need to protect that image as well as protect themselves from the less prudent among them.


I greatly appreciate the response from Dr. Wilson, and will cut short further musings here - except to add that in my view computer modeling breaks down when one tries to simulate the sense of trust that is the glue allowing cooperative organisms to function at all successfully.
Posted 01:27 PM on 04/18/2009