Saturday, 22 December 2012

Tim Tyler: Okasha, Evolution and the Levels of Selection


Hi. I'm Tim Tyler, and this is a review of this book:

Evolution and the Levels of Selection by Samir Okasha

I'd previously read Samir's Very Short Introduction to the Philosophy of Science. I thought that was good - and so I had some idea of what this book would be like.

The book contains an interesting and entertaining romp through the territory of group selection. It's what I call a "firehose presentation". In other words, it's a long stream of technical material that doesn't let up. This is a good match for my own preferences in a science book. Samir goes through practically every controversy in the field, and provides insightful opinions and commentary.

The book contains discussions of the Price equation and its significance, causality, emergence, evolutionary transitions, the gene's eye view, species selection, the group selection controversy and kin selection.

I thought the book was interesting and good. However, there were also quite a few parts of it which I disagreed with - or did not like. This is a reflection of the controversial nature of the subject matter.

The book dates from 2006. Throughout most of the history of the field of group selection, many of its advocates considered it to be a super-set of kin selection - often saying things like: relatedness is only one of many ways in which altruists can form groups which are then selected. However in recent years, the quest to find things that group selection explained - and that kin selection did not - seems to have petered out, with many of the most vocal group selection advocates now proclaiming its equivalence to kin selection. Samir's book predates many of these developments - and I suspect anyone writing a book on the subject today would treat the topic rather differently.

The book discusses kin selection only rather briefly. There's a discussion about it in the chapter relating to the group selection controversy, and another one in the chapter about evolutionary transitions. Samir recognises the possibility that kin and group selection might be equivalent, and cites several sources who claim that it is, sometimes approvingly. However, most of this book makes no mention of kin selection.

These days, I think few would approach group selection this way. Kin selection has a rich and successful history, while group selection has spent most of its existence mired in confusion and controversy. Kin selection has been much better studied. So: an obvious approach to many of the topics in this book would be to just use kin selection. However, it is hard to imagine this whole book being written in the language of kin selection. A good number of the issues just seem less important from that perspective. For example, in group selection there's the issue of what counts as a group. This broadly maps onto the issue of what counts as an individual in kin selection - yet this issue seems less controversial. Group selection faces of issue of how to model parly-overlapping groups - since most group selection models feature disjoint groups. Yet the corresponding issue of partly-overlapping families in kin selection seems less contentious. It's hard to escape the impression that the need for this book is partly because group selection is so awkward, difficult to understand and poorly-studied. Since kin selection is much better studied and much more widely used, it seems as though there would be less need for a philosopher to clear up misunderstandings in the field.

Samir offers several digs at the views of Richard Dawkins in the book. He criticises the idea that evolution is based on replicators, offering Hull's comment about them "passing on their structure intact" to claim that the term "replicator" implies high fidelity copying. I think that practically everyone on both sides of this debate agrees that high fidelity copying is not necessary - and it's high fidelity information transfer that matters for cumulative adaptive evolution. No modern users of the term "replicator" in biology use the term in that way - and many of them have objected to this persistent misunderstanding. Of course it's partly Richard Dawkins' fault for assigning an ordinary english word a counter-intuitive technical meaning.

Saimr also criticises the gene's eye view on two grounds. First, he says that it ignores behavioural and environmental inheritance. That isn't true if you adopt an information-theoretic definition of the term "gene" following Williams - since then memes are a type of gene, and the gene's eye view remains valid. Saimr also says that epistasis and "modifier genes" act against the gene's eye view. This is strictly true, but some linearity in the expression of genes is really all that is required to make the gene's eye view useful. Since a linear component in the expression of genes is ubiquitous, this issue seems like a storm in a teacup to me.

Like any complex technical book, there are some mistakes. The most embarrassing one I spotted was where Samir offered an incorrect definition of inclusive fitness - including the "augmenting" but not the "stripping" component - on page 145. Samir's explanations are usually clear - but sometimes the light fades. One such problem comes with the concepts of "MLS1" and "MLS2". Samir introduces these concepts by saying that they represent different focuses of interest on page 56. However on page 59 we hear that "MLS1 and MLS2 are distinct processes" and "whether either occurs in a particular case is a matter of objective fact". At best, this sort of material is very confusing.

Overall, this is a fine book - but I was left wondering if Samir had directed his energies in an appropriate fashion. The book will probably contribute to the modern group selection revival. However that revival seems likely to be accompanied by the usual muddle and confusion that follows group selection around like a black cloud. The problems with group selection at this stage are more sociological than anything else. Yes, groups exhibit reproduction and differential reproductive success, and that affects the course of evolution, but the problem is that practically whenever group selection gets used it results in junk science, or at best, science that is inferior to that which would have been produced by using kin selection. Looking at the mess that group selection has caused in the evolutionary human sciences illustrates this point. Does the world really need more group selection? After reading Samir's book, I was still sceptical. Samir doesn't address sociological questions concerning whether the muddle associated with group selection means that it does more harm than good. Instead, he just wants to clear up the muddle. But in that case: why not use kin selection? It seems much better studied, much less confusing, and has produced much less junk science.

What I think group selection needs most is clearly-articulated reasons to use it in place of kin selection. At the moment, the "why not use kin selection?" question is challenging to answer. Maybe there are reasons - but this book doesn't really provide an answer. It doesn't even ask the question.


No comments:

Post a Comment