Do traits evolve because they are good for the group in which they occur? Darwin thought so, arguing that this was the right way to think about the barbed stinger of the honeybee and human morality. The idea that selection can favor traits that are good for the group as well as traits that are good for the individual was also part of the biologist's toolkit during the modern synthesis. Then, in the 1960s, everything changed. In the space of a few years, George C. Williams published Adaptation and Natural Selection and W. D. Hamilton developed his ideas on kin selection and inclusive fitness. A few years later, John Maynard Smith and George Price laid the foundations for evolutionary game theory. The idea of group selection was attacked not just as factually mistaken but as an example of fuzzy thinking. It was also attacked for being unnecessary—kin selection and game theory were said to deal adequately with apparently altruistic traits, and without using the tainted “g-word.” This anti–group selection consensus was summarized in Richard Dawkins's popularization, The Selfish Gene.
Though many biologists regarded the critique of group selection as a fundamental step forward, there were dissenters. In 1970, Richard Lewontin wrote a review article in which he described the abstract features of the process of evolution by natural selection. If a “collective” contains “particles” that differ in their abilities to survive and reproduce, and if traits of parent particles are correlated with traits of their offspring, the composition of the collective will change. The organisms in a breeding population are one example of particles in a collective. But there are others—the different genes that exist in a single organism and the different groups that exist in an ensemble of populations. From this perspective, group selection is not a confusion. It is part of the Darwinian framework; it is conceptually coherent, though, of course, arguments for its existence and importance must be developed case by case by examining empirical evidence. The idea that selection takes many forms—that intragenomic conflict and group selection need to be considered as well as individual selection—came to form part of what is now called multilevel selection (MLS) theory.
Samir Okasha's wonderful new book... is a philosophical examination of the conceptual framework that MLS theory deploys.... It is gratifying that his book engages the details of mathematical models and at the same time connects those details with broader philosophical questions.
The other early challenge to the dismissal of group selection came from George Price. One reason that group selection looked like a mushy concept to many biologists was that there wasn't much in the way of mathematical models that could be used to anchor one's thinking. Price changed all that by developing a formalism that partitions the change in frequency that a trait experiences into the change due to individual selection and the change due to group selection. Before Price's innovation, Hamilton wrote that the idea of group selection must be “treated with reserve” because it lacked a mathematical foundation. After Price, Hamilton retracted his earlier rejection of group selection and recognized that his own work on inclusive fitness in fact involved a commitment to group selection.
What has happened since those early days of impassioned rejection and isolated dissent? Different biologists give different answers. Some still think that group-selection theory is the work of the devil. Others are comfortable with MLS theory, thanks in part to David Sloan Wilson's work in the late 1970s in which group selection was represented in terms of his idea of trait groups. (Personal disclosure: I coauthored a book in 1998 with Wilson defending MLS theory.) Current friends of the MLS approach emphasize that mathematical models are needed, and these models need to be tested against competitors. Naïve group selectionism should be avoided, but the same applies to naïve individualism.
Samir Okasha's wonderful new book, Evolution and the Levels of Selection, is a philosophical examination of the conceptual framework that MLS theory deploys. Lewontin's early formalism may give the impression that the idea of selection occurring at different levels of organization is straightforward and that the difference between group and individual selection is transparent. The complexities that have become visible since the 1970s show otherwise. One complication arises in connection with the Price equation. Consider this simple example: There are two groups of zebras, one composed entirely of fast zebras, the other entirely of slow ones. Suppose the fast group is less likely to go extinct. According to the Price equation, in this situation there is group selection and no individual selection, because all the variance in fitness is between groups. But surely it is possible that the groups differ in fitness just because there is individual selection for running fast. Selection at the individual level can create a fortuitous benefit for the group (as George Williams put it). The Price equation is unable to recognize this. Biologists have coped with this problem in different ways—for example, by invoking the statistical techniques of contextual analysis and by employing a methodology called neighborhood analysis. Okasha skillfully analyzes the Price equation's strengths and limitations and these more recent attempts to do better.
Another complication that arose as MLS theory developed was that there really are two types of MLS. In discussions of the evolution of altruism, a group's fitness is usually defined as the number of offspring organisms the group produces. But one can also conceive of group fitness in terms of the number of daughter groups (regardless of size) the group produces. This second type of MLS has been important in discussions of species selection and of major evolutionary transitions. Both concepts raise questions about what heritability at the group level means, and here again Okasha does much to clarify what is at stake.
Older discussions of group selection have usually presupposed the existence of biological hierarchy instead of explaining it. The standard question was framed as follows: Given that a species is made up of an ensemble of populations, how might traits that are good for the group evolve? But where does group structure—the existence of herds and hives—come from in the first place? The same question, one level down, is important as well. Rather than focusing on the question of how traits evolve that help multicellular organisms to survive and reproduce, given that such organisms already exist, one also might ask how multicellularity evolved in the first place. This is the subject of the major evolutionary transitions, an exciting area in recent evolutionary theorizing. The new question isn't one of understanding evolution in a preexisting hierarchy but the evolution of the hierarchy itself. Okasha ably discusses how older questions about group selection connect with this newer set of issues.
The units-of-selection problem is part of evolutionary biology, but it has attracted a great deal of discussion within the philosophy of science over the last 30 years. Some philosophers doubt whether the question has a factual answer at all, suggesting that it is a matter of convention whether we choose to view natural selection in terms of group or individual or genic selection. Others have seen MLS theory as a context in which questions about emergence and reductionism matter to the practice of science. And the suggestion has been floated that looking at evolution just from the point of view of genes misses the causes of evolution at higher levels of organization, which has led to reflection on the role of causal thinking in science generally. Okasha makes interesting and novel contributions under these headings. It is gratifying that his book engages the details of mathematical models and at the same time connects those details with broader philosophical questions. Okasha sees both the trees and the forest.