Evolution and the Levels of Selection [Hardcover]

It is rare that you can find a thoughtful book on the subject of multilevel selection that actually critically reviews all sides of the subject. I have done considerable work in this area. I am pleased that he cites me appropriately and thoughtfully, and that he cites the relevant literature appropriately. That said, it is a technical book. I don't think I would recommend this as your first book on evolutionary theory, and it frankly is quite technical. However, if you want to know what the current controversy about group selection is all about then I strongly recommend this.

A few details: The center piece of this book is the comparison between the Price equation and contextual analysis. The Price equation is a method of partitioning covariances between a trait and relative fitness into within group and between group components. Contextual analysis is a multivariate regression approach in which a partial regression of traits measured on individuals and measured on groups are simultaneously examined. These two approaches follow from different philosophies and lead to different conclusions. Okasha discusses how these two views differ, and provides an excellent rational for choosing between these two approaches, provides important insights into how these two approaches color our view of multilevel selection

I both love and hate this book. I love it because I am particularly interested in the 'Levels of Selection' problem in evolutionary science, and any book on this subject is a good book. But, I also hate it too. Let me begin with the reason why I hate it. Professor Okasha writes in the Introduction, "The book is aimed at evolutionary biologists, philosophers of science, and interested parties from other disciplines. It presumes a basic familiarity with Darwinian evolution, but I try to introduce every topic from scratch. Jargon, whether biological or philosophical, is avoided as much as possible, and explained where it is used. In places the treatment is slightly more technical than is customary in philosophical discussions, but no more so than is necessary to achieve clarity. Inevitably, different chapters will appeal more to some readers than others, depending on the reader's interests. The book is designed to be read as a whole, but there is an element of modularity." As one reviewer pointed out already, this book is technical - very technical. And this is why I hate it. Because this book was published for a general audience, as opposed to being published in a technical journal, I felt that Okasha could have perhaps sprinkled a few metaphors and analogies throughout the text to help explain some of the more difficult passages. It is only because I have "a basic familiarity with Darwinian evolution" that I was able to grind through this book. As an introduction to the `Levels of Selection' problem, I think this book would be entirely beyond the layman's reach. However, with that aside, I did also love this book; what follows are a few reasons why.

For starters, Professor Okasha, does lay out the argument efficiently; "The levels-of-selection is one of the most fundamental in evolutionary biology, for it arises directly from the underlying logic of Darwinism. The problem can be seen as the upshot of three factors, each of which was appreciated to some extent by Darwin himself. The first and most fundamental factor is the abstract nature of the principle of natural selection...Any entities which vary, reproduce differentially as a result, and beget offspring that are similar to them, could in principle be subject to Darwinian evolution. The basic logic of natural selection is the same whatever the `entities' in question are. The second factor is the hierarchical organization that characterizes the biological world. The entities biologists study form a nested hierarchy, lower-level ones properly included within higher-level ones...The third factor concerns not the process of natural selection but its product. Natural selection leads organisms to evolve adaptations - traits that enhance their chance of survival and reproduction." Also, I found the way Professor Okasha applies Price's Equation in the hierarchical setting very interesting and valuable. Furthermore, I found the two sections, 'Particle Fitness and Collective Fitness' (2.2.3) along with 'The Two Types of Multi-Level Selection' (2.2.4), to be enormously beneficial. In fact, these two small sections more than make up for any negatives about the book.

In sum, I believe that for anyone already deeply interested in the 'Levels of Selection' debate then this purchase would be a no-brainer. For others however, there are better places to start: Darwin's Conjecture: The Search for General Principles of Social and Economic Evolution, Evolution--the Extended Synthesis, or The Evolutionary World: How Adaptation Explains Everything from Seashells to Civilization, for example.

Gene's eye view (ch. 5). The early literature on the gene's eye view portrayed it as an empirical thesis: this is the level at which selection really acts. An argument for this is the phenomena of "outlaw" genes, i.e., genes which spread despite being detrimental to the organism's fitness. But it shows only that these are cases of gene selection (which is generally admitted), not that all selection is of this type. Some of the later gene's eye literature (e.g., Dawkins' Extended Phenotype) has toned down the empirical claim and instead portray the theory as a heuristic perspective. For example, altruism (towards kin, who posses largely the same genes) can easily be explained from the gene's eye point of view as a trick devised by the gene which benefits only itself, not its host organism. Strictly speaking, one can tell an equivalent story at the level of organisms (since inclusive fitness can be defined without reference to genes). How are we supposed to tell which is "right" and which is "wrong"? As a heuristic device, however, the gene's eye view clearly has greater intuitive appeal. As for objections to the gene's eye view, a central one is that tracking genes is mere "bookkeeping," ignoring the causality. In fact, such "books" do not even comprise a complete record of evolution. First, they leave out nongenetic inheritance. Second, they track evolution only by smuggling organisms in through the back door in the form of the counting convention that genes are counted once per organism. Clearly the convention is needed (we cannot understand evolution if we cannot tell two identical twins from one fat person, for example), and it also seems to make some sense from the gene's eye point of view to do the counting at the zygotic single-cell stage. By zygotic counting is problematic: there is nothing inherent in the concept of evolution that requires every generation to pass trough a single-cell bottle-neck. Indeed, there are cases (e.g., asexually reproducing plants, insect colonies with queen structure) were other counting conventions make more sense, but which the gene's eye view seems incapable of capturing.

Group selection (ch. 6). One account of group selection is that groups are subject to selection by virtue of having life cycles, offspring, etc. This type of group selection is likely to be drowned out by individual selection since the generations are so much longer. But this does not rule out the possibility of group selection understood more broadly as the claim that the fundamental cause of such traits as altruism is at the group level. In fact, perhaps the most popular way of trying to account for altruism at the individual level, kin selection, can be seen as group selection (in the second sense) in disguise, as follows. Kin selection is merely the special case were the groups are kin groups. People think kin selection avoids group selection because the group can be defined from the point of view of the individual: the rule "be altruistic toward kin" is an individual-level rule. But what determines whether altruism evolves or not is whether altruists interact with other altruists or not (a sole altruist will get exploited and die out, etc.). In other words, what matters is what groups there are: the cause is at the group level. Whether the groups can be defined in terms of individual-level rules or not is incidental. It does nothing to preclude group selection; rather it is a special case of it. The force of this argument is diminished by the following argument that some form of preferential grouping (rather than random grouping) is necessary for group selection to take place: if the grouping is uniformly random with respect to, say, altruism, then altruists and non-altruists face the exact same conditions so any difference in reproductive success must be due to entirely to individual fitness since that is the only difference present. Note that "weak altruism" (which raises the fitness of the individual but raises the fitness of its peers by a greater amount) can easily evolve by such individual selection whereas "strong altruism" (which reduces the fitness of the individual) cannot. This argument assumes that pure individual selection means selection driven entirely by individual differences. The argument falls if one assumes instead that individual selection means selection for the fittest individual within any group (group rather than population because "fittest in population=that which evolves" by definition, so group selection is a priori impossible with this notion of individual selection).