Apathy About the Levels of Selection

Gene Expression is one of my favorite biology blogs. There seems to always be something interesting going on there. I read Razib's post on apathy about the levels of selection controversy with considerable interest and not at all with apathy.

Philosophers of biology cared a lot in the 1980s about levels of selection. Probably we cared too much. (I think it may not have been possible to call oneself a philosopher of biology at the time without having published a paper on the topic.) Happily, we seem to care less these days, in spite of Samir Okasha's creative efforts to show us why we ought to care. I'd always thought that evolutionary biologists stopped caring, at least for the most part, immediately following the publication of G. C. Williams' Adaptation and Natural Selection in 1966.

An aside: Razib mentions Lisa Lloyd's nice papers on the levels of selection. I especially like her 2001 paper, "Units and Levels of Selection," in Thinking About Evolution (the Lewontin festschrift edited by Singh, Krimbas, Paul, and Beatty). She made that paper available for download because in a review of that volume John Maynard Smith picked out her paper as "the most stimulating essay in the book." I also reviewed the volume and said virtually the same thing. (Lisa didn't mention me. I don't rate like Maynard Smith. Go figure.) My favorite levels of selection paper is Bill Wimsatt's, "Reductionistic Research Heuristics and Their Biases in the Units of Selection Controversy" (in Nickles' Scientific Discovery, Vol. 2, Case Studies, 1980). It's a classic in philosophy of biology and very rewarding if you can survive it.

Razib's post is a reaction to some references made by Coturnix at one of his blogs. I really just want to consider Razib's points mostly in a vacuum, with apologies to Coturnix. Razib makes two main points. The first is that he thinks multi-level selection, the view that selection acts variously on genes, individuals, and groups, isn't generally accepted. I should think he's right about that.

Over and over I hear what you get in Williams: Obviously multi-level selection is theoretically plausible. What's key is whether such phenomena as group selection are important in evolution. I take it that the consensus is that it's not important. The key reason, but not the only one, seems to be that group selection requires that parameters that measure population size, gene flow, selection intensity, and so on, conjointly satisfy a fairly restricted range of values which is just not commonly the case.

Notice that if one accepts that reason, one isn't permitted to say that multi-level selection is a myth. Ultimately, the evidence will decide a given case. And if the evidence demonstrates that the range of relevant parameters meets the conditions necessary for group selection to occur, then there's good reason to think that it's occurring. (If that's good enough  for Williams, and it is, then it ought to be good enough for the rest of us.)

Razib is careful to point out that he doesn't think Richard Dawkins' selfish gene view is the way of thinking about the levels of selection. Thank goodness for that. (It's good to be consistent, after all.) My view, which I may not share with Razib, is that Dawkins has done considerable disservice to our understanding of the levels of selection controversy. There was no reason for Dawkins to, essentially, re-write, with a more popular tone, Williams' book. Williams did a good enough job on his own. And what's worse, Dawkins foisted upon us a slew of tortured metaphors for understanding what's quite straightforwardly understandable by way of theoretical and experimental biology.

The second point Razib takes up concerns the connections, if there are any, between views about the levels of selection and views about politics. I don't really want to go here. But it's not clear to me that you'd get much better than chance results on this connection. Razib attempts to find some potential connections by looking at Fisher, Haldane, Hamilton, Lewontin, and Gould on these issues. Ultimately, Razib thinks looking for these connections is a "fun parlor game." But perhaps a survey looking for the political underpinnings of monists and pluralists about the levels of selection would be interesting. But I think Coturnix is right in his comments on Razib's post when he suggests that the levels of selection and politics aren't necessarily connected. So, we can go out in the field, find the answer, and believe politically whatever we want. Well, thank goodness for that.

At any rate, I read Razib's post with interest, as I said. But what got me chuckling is his addendum, which says, "But, I will say that I am tempted to make some argument about how Robert Skipper's line of reasoning about "genetic draft" is ideologically driven by beliefs as a conservative Democrat :)." It makes me nervous to think (a) that my views about draft are ideologically motivated and (b) that they are driven by beliefs aligned with conservative Democrats. Why? Well, with respect to (a), I'd be arguing entirely for reasons I didn't know I was relying upon. As for (b), I'm way farther on the left than that! (Perhaps that means that good predictions about politics don't arise from analyzing someone's science writing. That would be nice.)

Comments

[coturnix]

Wimsatt's paper is great but, as you say, first you have to survive it. That is why I push people to read Brandon's chapter as it is much clearer for non-philosophers (like me) to understand.

Razib was responding to an ancient post of mine in which I was exploring, jocularly or brutally (depending on your political side, I guess), possible connection between genocentrism and conservatism. He did not like it then and does not like it now, but it was a fun excercise for me at the time.

[Richard R Lawler]

This is an interesting post, particularly the first point by Razib. I think maybe the point he's making should be clarified. He suggests, and you agree with, that multilevel selection "isn't generally accepted." I disagree. Perhaps it isn't generally accepted among philosophers of biology but it seems more than accepted amongst evolutionary biologists. Meiotic drive, fertility selection, kin selection, are all examples of multilevel seleciton in action. Respectively, selection is acting on haplotypes, a mating pair, and interacting groups in these examples--apart from acting at the level of the individual. These are widely occurring (and accepted) phenomena in nature that definitely influence evolutionary dynamics. Whether evolutionary biologists realize that say, fertility selection, is actually a form of multilevel selection is perhaps not always appreciated, but this doesn't detract from the fact that selection is acting on multiple levels. There are many other examples of multilevel selection (e.g., cancer). The apathy that Razib is talking about probably comes from the debate being framed in the "group versus individual" format, complete with lemmings, populations regulation, and the attendant slapping-down of Wynne-Edwards. Now that's a debate whose fitness is zero. I suspect that when you and Razib state that multilevel selection "isn't generally accepted", you are talking about this "group versus individual" paradigm.

[razib]

"group versus individual" paradigm.

to a great extent. i don't care much about labels if the math & data work out.

[John Wilkins]

It's worth noting that Williams in that very same book cites a case of group selection (the famous t-allele) and does not, despite common misunderstandings, deny that group selection can occur, only that if there is a comepting individual selection explanation, it is to be preferred.

Multilevel selection, on the other hand, appears to be a consensus among the scientists, in that if a property is heritable, and subject to selection, it matters not if that property is borne by a gene, or a trait, or an organism, or a deme, or a variety within a population. In fact it has always seemed to me that Dawkins himself, the ogre of reductionism, allows for this in permitting memes to be the subject of selection - they are surely not low level phenomena, whatever they are.

Anything that is capable of reproduction, and which has an interactive bias relevant to reproduction, can be a unit of selection. I doubt this applies to species, myself, but anything lower than that is fair game, and it need not be "reducible" to genetic factors either, since many traits have multiple pathways of development and many genes can cause the same phenotype.

At some point selection becomes sorting, and I suspect that has more to do with the fidelity of reproduction than anything else. But in any case, there is no privileged level of selection.