Until
next week, 
AI and the Notion of Progress (Part 1): The Extended Phenotype
Dateline: June 14, 1998
Reference: Dawkins, Richard (1982). The Extended Phenotype: The Long Reach of the Gene. New York: Oxford University Press.
I read The Extended Phenotype (which is more technical than Dawkins' lay-oriented books and already 16 years old, but still in print) only recently, partly because I have grown interested and knowledgeable enough in matters pertaining to evolutionary development to want and to be capable of reading a relatively technical book about it, and partly because Dawkins is such a great scientist and writer.
After reading it, I sensed it might shed some light on the evolutionary progress toward the intelligent machine, the Machina sapiens of (perhaps!) my dreams. But I was, and remain, unsure about just what that light is. Meantime, I also read Gould's much more recent Full House, which has nothing particularly to do with phenotypes but everything to do with the notion of evolutionary progress.
The very idea of "progress" has come under fire from several renowned scientists and philosophers, and I've never bought their arguments. Because (it seems to me) it is necessary to accept the idea of progress if one is to accept the idea of Machina sapiens, I plan to discuss, in spite of Gould's counter argument, what I think it tells us about the notion of evolutiionary progress. Since this entails both a lot of thought and a lot of writing, I shall break this article up into three weekly sections. I start this week with a summary of Dawkins' book.
The Extended Phenotype
The book’s key thesis extends upon Dawkins’ earlier theory (expounded in The Selfish Gene, OUP, 1976) that the fundamental unit of biological life is the gene, and that everything in an organism that is "higher" (for want of a better word) than the gene (i.e., the cell, the organ, the whole organism) exists only to serve the gene and help to replicate it. The organism is merely a carrier, a vehicle, for the gene.
The phenotype is the definition or classification of an organism or group of organisms with reference not merely to its genetic characteristics (the genotype) but to physical characteristics or properties produced through the interaction of the genotype with the environment as the organism develops through adulthood. Genes instruct the body of an organism to develop in a certain way (say, to grow to a certain size, to grow a hard shell) to suit the organism's environment. Tallness may be an advantage in some environments but not in others, therefore it is to the gene’s advantage, if it is to survive, to change the shape of its host organism to help its host survive.
Dawkins extends these phenotypic effects of the gene to include the environment itself—the gene can change the form and structure of the environment outside the body of the host organism. The gene's vehicle thus becomes not just the organism, but that organism's environment as well. Examples of extended phenotypic effects include the beaver dam (including how tall the dam is) and the lake that results from the damming of the stream (including how large that lake shall be), and the termite mound (including the size of the mound).
But the theory goes further: extended phenotypic effects apply not just to external inanimate objects such as dams and mounds, but to integral parts of living organisms. For example, there is evidence that the thickness of the shell of some snail species may be an extended phenotypic expression not of the snail’s own genes, but of the genes of a parasitic microbe for which the snail is the parasite’s external environment. But for the parasite—or, rather, but for the gene in the parasite responsible for making the snail grow a thicker shell—the snail's shell would be thinner, and but for a thicker (and therefore more protective) shell, the parasite—and more importantly, its genes—might not get a chance to replicate, since its environment, the snail, would be more vulnerable to encroachment by predators.
Going further still, Dawkins' theory claims that an organism's behavior (and not just its morphology—its form and structure) may be controlled by the genes of another organism. Evidence for this is again most clearly seen in the world of parasites, some of which invoke suicidal behavior in their hosts in order that the host can more easily be found and eaten by its predator, which will usually be an organism in which the parasite can proceed through another stage of its own life cycle.
Which leads to Dawkins' most startling conclusion: that a phenotype extends beyond the organism to embrace other organisms as well. His best example is the cuckoo, whose phenotype extends to embrace the birds that rear its own offspring. The surrogate parents are, not just in effect but in fact, controlled by the cuckoo's genes.
Dawkins affirms that it is not his intent to belittle the organism; rather, he wants us to view the organism in a new light—as a member of a sort of extended family, a network, that can include organisms of different biological kingdoms, all working (unknowingly) to ensure the survival of certain genes. He is careful, however, to disavow much resemblance between his theory and Lovelock's Gaia hypothesis, which claims that the entire earth is just one huge organism. Dawkins disputes Gaia, and shows that his own extended phenotype does not extend very far spatially; perhaps at most a few miles in the case of the beaver's lake.
* * *
So much for the extended phenotype. Next week, we will see what Gould has to say about the notion of progress, and the week after that, I'll put it all together in the context of artificial intelligence.
Until
next week,
NEXT WEEK: AI and the Notion of Progress (Part 2): Full House
Help Wanted: Got questions or comments on this article or on any other AI-related subject under the sun? Post it in the AIBB!