In recent decades the idea of group selection has gotten a bad rap. In 1871 Darwin outlined what he considered to be the conditions of group selection in terms of self-sacrificing tribal warriors that do not further their own reproductive fitness through their sacrifice, but acting in such a way provides benefits to the tribes that encourage such behavior over those that do not. And these benefits can be cashed out in terms of tribes being able to better compete for resources against other tribes and hence spread more readily across the landscape…we might call this "social selection". From a complex systems standpoint the idea of social selection seems pretty intuitive; it's just regular natural selection where social groups are the agents of a larger system. So what's the trouble?

Over the years other, individual-based, explanations (such as kin selection) were developed and appeared sufficient for explaining Darwin's group selection observations and ideas. In Dawkins' 1976 book, The Selfish Gene, he distinguished two components of the evolutionary process: the replicator and the vehicle. He argues quite well for the idea that the gene is the unique replicator for biological evolution. Animal bodies are the vehicles against which selection pressure works and a body's phenotype is the testing ground for the genotype that produces it. Many people took this to be the end of the road for group selection, but that conclusion is quite unwarranted.

Animal bodies are systems, and hence define environments in which cells (native and foreign) compete for survival and reproductive success. The result of this interaction is a body that maintains a diverse collection of cell types, has techniques to remove harmful agents, and various other system-level operations. If a body has too many selfish agents (cancer cells, viruses, bacteria, etc.) and insufficient institutions to deal with them (immune system, hormone regulators, etc.) then the systems fails and other bodies with better systems can appropriate its resources. A sequence of DNA only programs what a part of a cell does, so the development of a body is a dynamic interactive process in which selection pressure plays an important role. It seems that even examining a body's phenotype invokes group selection as part of the explanation…if one considers the body as a system instead of as the basic unit.

Now what's the trouble? Finding a selection process working at a one level of a system with the environment defined by its metasystem doesn't mean that the process can't be reduced to differential reproduction at a lower level. Seeing natural selection as a system-level process makes it general enough to be applied to social groups, but it's still individuals that reproduce or not…or individual cells that reproduce or not…or individual genes that get expressed or not…or individual molecules that get produced or not. You can describe a stock market crash in terms of the movements of particles but that wouldn't be very insightful into the economic phenomenon. Clearly the process of natural selection makes conceptual sense at higher levels than the gene and useful models can be built employing group selection to explain various social phenomena. But is this Darwinian selection?

There hasn't been very much work in differentiating types of "survival of the fittest" selection processes. Perhaps there is just one general process that we call selection and it applies to many systems at many levels of abstraction. But perhaps there are different underlying mechanisms that produce varying selection forces and the concept of selection can be refined. So it seems that there is a lot of work to be done in identifying the properties of systems that induce selection pressures and developing principled distinctions among these system types…a kind of taxonomy of selection processes. In a follow-up post I will provide some examples of system properties that seem to result from group-level selection and try to develop at least one quality that differentiates the processes.