This site is moving to a new domain: check out

“Some of these stories are closer to my own life than others are, but not one of them is as close as people seem to think.” Alice Murno, from the intro to Moons of Jupiter

"Talent hits a target no one else can hit; genius hits a target no one else can see." Arthur Schopenhauer

“Why does everything you know, and everything you’ve learned, confirm you in what you believed before? Whereas in my case, what I grew up with, and what I thought I believed, is chipped away a little and a little, a fragment then a piece and then a piece more. With every month that passes, the corners are knocked off the certainties of this world: and the next world too. Show me where it says, in the Bible, ‘Purgatory.’ Show me where it says ‘relics, monks, nuns.’ Show me where it says ‘Pope.’” –Thomas Cromwell imagines asking Thomas More—Wolf Hall by Hilary Mantel

My favorite posts to get started: The Self-Righteousness Instinct, Sabbath Says, Encounters, Inc., and What Makes "Wolf Hall" so Great?.

Tuesday, July 6, 2010

First Impressions: Unto Others: the Evolution and Psychology of Unselfish Behavior

I remember a period around the time I graduated from college when I was going back and forth between the works of Stephen Jay Gould and Richard Dawkins, trying to locate their most fundamental differences and deciding who had the better empirical and logical support. I often had the feeling that Gould got the better of the argument, but only because Dawkins was making larger claims. At the same time, Gould’s vision never struck me as thoroughly developed, while Dawkins’s was almost hermetically tight.

In 1998, unbeknownst to me, Elliot Sober and David Sloan Wilson came along with a meta-perspective on both evolutionary paradigms, which they spelled out in Unto Others: the Evolution and Psychology of Unselfish Behavior. The book is a painstaking unpacking of assumptions underlying common arguments regarding the processes of evolution and the nature of human motivation. Though tedious at points, it’s ultimately well worth while because it turns out that many of those assumptions, while not exactly wrong, obscure important details.

Case in point, Dawkins relies on a definition of evolution—change in gene frequencies—that is blind to the processes that account for those changes. Specifically, by only attending to the outcomes of genetic competition, researchers inevitably miss the fact that group selection has occurred (yes, group selection!) To illustrate, Sober and Wilson offer an analogy with a case brought against the University of California at Berkley in the 1970’s. The percentage of women applicants who were accepted there was less than that of men. But when the University did a department-by-department inquiry they found that there was no discrimination. How is this possible? It turned out that women were disproportionately applying to more selective departments. The outcome is an example of what’s called Simpson’s paradox:

"To see how this can happen, imagine that 90 women and 10 men apply to a department with a 30 percent acceptance rate. This department does not discriminate and therefore accepts 27 women and 3 men. Another department, with a 60 percent acceptance rate, receives applications from 10 women and 90 men. This department doesn’t discriminate either and therefore accepts 6 women and 54 men. Considering both departments together, 100 men and 100 women applied, but only 33 women were accepted compared with 57 men" (25).

It’s easy to imagine a scenario in which a trait that was maladaptive in an individual group (like a department) could emerge in a larger population by this same process. (Well, maybe not so easy to imagine, but possible.)

Those insisting on individuals or genes as the only units of selection can still say that once the context is taken into account one or another trait is more adaptive than another and that is the one which will evolve, an argument Sober and Wilson refer to as the averaging fallacy. But this argument does nothing to illuminate the process scientists are trying to understand—it rather obscures an important element of it. Sober and Wilson convincingly argue that instead of choosing sides over which unit ought to garner the most attention, we should be able to adjust our focus depending on the question we’re trying to answer. They go on to make a case for the evolution of altruism resulting from group selection.

It seems more arguments are poked with holes in Unto Others than are presented as well sealed. The effect is an opening up of the playing field, an epistemic position the authors call pluralism, not to be confused with the other type of pluralism they argue for in the realm of evolutionary processes. On the whole, this leveling is immensely important—and must’ve been even more so in 98. But since so much of the book is given over to a dismantling of undeservedly prominent arguments and assumptions, I was a little frustrated coming away. I wish there had been more examples of what theories derived from a multi-level selection paradigm would look like. Maybe I’ll find them in some of Wilson’s later books.


caynazzo said...

You have some jackass spamming you.

Anyway, I think group selection works on a metaphoric level and that's it.

If traits aren't grounded in something concrete such as alleles, how do you study them? Models aren't enough.

I highly suggest you listen to this:

Stuart West provides the best summary of the group selection (GS) and inclusive fitness (IF) controversy.

Dennis said...

Yes, I keep getting comments that are links in Chinese. They probably have a program that posts them on random blogs. I don't want to limit access though so I'll just try to keep up erasing them.

The traits Sober and Wilson are discussing are grounded in something concrete, both alleles and observable social norms. But what they're interested in is the processes that cause them.

I will listen to the debate though.

caynazzo said...

Oops, here's the Stuart West video

There are well-known published examples of traits favoring individual and kin selection, but I'm not aware of any instances of traits selected at the group level. Do you know of any?

Keep falsifiability in mind: a trait that's beneficial to the group but not the individual (stand in for gene) or its kin.

Dennis said...

Sober and Wilson use the example of worker bees supposedly working to help close kin reproduce--and then researchers discover they're not related. I suppose this is a type of parasitism.

Anyway, S and W don't exactly present group selection as an alternative to kin selection. A group has to be, in aggregate, different at the genetic level for it to evolve any traits. So kin selection actually is a form of group selection.

You're challenging me to wrap my mind around this stuff more thoroughly. I'll have to post on it more in the future.

Dennis said...

Just watched the lecture. I get the impression West hasn't thoroughly studied Sober and Wilson's ideas. They have a whole chapter on why you can't rule out MLS with Occam's Razor--which I'll have to review before I can summarize.
Also, West's main point is that inclusive fitness theory has been more fruitful over the last thirty years. He claims there has been no conspiracy against group selection, but the fact is it has been dismissed as a matter of orthodoxy for those same thirty years.
And yes, as of yet researchers aren't as good at modelling MLS, but they haven't been at it as long.

caynazzo said...

I barely understand this stuff myself.

A while back I briefly discussed this with a friend who does phylogenetics. I'll try and get him to weigh in on this.

caynazzo said...

His response:

[...]as far as group selection, I'm sure it happens in some circumstances. The numbers seem to make it very unlikely to be very powerful because effective selection requires either large numbers of individuals (or groups in this case) or be extremely strong. Because the numbers of groups in most cases are not very large, for selection to overcome drift/migration it would have to be fairly strong.

Because of this difficulty in getting adequate numbers (either large enough numbers of groups or high enough coefficients of group selection) I would think almost any other explanation is more likely. That's also not to mention that any migration between groups could serve to erase group selection's power also.

However, if I thought that there were sufficiently large numbers of groups and/or there was some reason that selection should be particularly strong, I might believe it. I have little doubt that it does occur at least in a few cases. Pretty much everything does.

I don't know that much about the arguments within the field, except that my undergrad professors seemed to think group selection was numerically impossible in most cases, but I don't know if they were within/outside of the orthodoxy on that one. I would guess that the more statistically inclined evolutionary biologists would be more against it, and more organismic/behaviorally minded ones would be more for it.