Thursday, 2 June 2011

Against the extended genotype

One way of modelling human culture is as an extension of human biology. If you do that then one way of modelling cultural information is as part of a human extended genotype. The products of culture would then be modelled as being part of the phenotype of that extended genotype. This type of model is, alas, common in academic studies of cultural evolution. As Mesoudi (2011) puts it:
In a typical cultural evolution model, a population is assumed to be composed of a set of individuals, each of whom posseses a particular set of cultural traits. A set of microevolutionary processes is specified that changes the variation of those traits over time.
An "extended genotype" would make reasonable sense as a model if culture was only transmitted vertically. However, in fact only a few traits are only transmitted vertically. If you introduce "oblique" and "horizontal" transmission the result is more like multiple genotypes than a single genotype - and the "microevolutionary processes" involved can get complicated.

Using this type of model, you can approximately reproduce the same dynamics that are actually exhibited by cultural evolution - if you are prepared to model sufficiently complex micro-evolutionary transmission processes. However, this type of model is philosophically unsatisfactory. As with symbiotic gut bacteria and foodstuffs, it is best to just classify cultural entities as belonging to different species. They have their own lifecycles and inheritance mechanisms and interests. They usually spend some of their lifecycle outside the human body, where they may be destroyed or copied. Modelling them as extensions of the human genotype runs contrary to Occam's razor and makes no sense at all. It leads to byzantine models, which are specific to cultural evolution processes. The correct approach is to use the existing perfectly conventional models of symbiosis. That is the approach taken by memetics.

The extended genotype is sometimes codified in the form of the phenogenotype - as in this 1992 paper.

Alas, "phenogenotype" is very messy terminology, which is best forgotten about. As Herbert Gintis once said:

Durham uses the term 'meme' for a unit of cultural inheritance. I think his defense of this is one of the strongest points in this great book. He shows that culture cannot be identified with phenotype or behavior. It follows that we must drop the term 'geno-phenotype'. In its place we can use the term 'geno-memotype.'
Phenogenotypes were an awful messed-up concept - but in memetics, there is no 'geno-memotype' to replace it. That is pretty-much an unnecessary concept. Instead there is symbiosis.

Evan Louis Sheehan has a nice way of explaining the problem with "extended genotype" models in his book: The Laughing Genes. He says:

Perhaps the ideas that aided in early human survival should be considered as some sorts of extensions to the genes. Then, just as good genes yielded good biological attributes such as strong muscles, good ideas yielded good extensions to biology in the form of such things as clubs and spears. Indeed, this is the way I used to think of cultural ideas, as extensions to the genes that underwent evolutionary development in parallel to the genes. Ideas that provided survival advantage were passed down vertically from generation to generation, and persisted simply because they provided survival advantage, just as some genetically inspired valuable traits, such as keen eyesight or strong muscles might do. I now see this as an incomplete picture of the ways that ideas are able to evolve. I must thank Dawkins once again for the revelation that allowed me to see this.

The most comprehensive treatment of this academic folly is probably Ben Cullen's book: Contagious Ideas: On Evolution, Culture, Archaeology and Cultural Virus Theory. He calls the idea by the term "inclusive phenotype" - since the academic researchers involved bundle cultural and genetic influences into one human phenotype - in what Ben refers to as a "bio-cultural muddle".


No comments:

Post a Comment