Tuesday, 21 October 2003

Darwin’s dangerous idea

Listening to:

Scarlatti, keyboard sonatas.

Just read:

Daniel Dennett, Darwin’s dangerous idea.

I first read this book a few years ago, coming to it because I’d enjoyed Dennett’s & Hofstadter’s The Mind’s I. Darwin’s dangerous idea has less to say about consciousness and artificial intelligence, but covers a lot of other very interesting ground. Its fundamental theme is the importance of Darwin’s idea, and how it gives rise to a great many interesting ramifications. Indeed, Dennett claims that Darwin’s was the best idea anyone has ever had, more important than the ideas of Newton or Einstein. For example, one reason this idea is so important is because it explains how complexity can arise from simple beginnings without needing to appeal to external miracles, aliens or mystical forces. Dennett lumps all of these “explanations” together under the moniker sky-hooks, to contrast with the mundane cranes that natural selection uses.

Dennett begins with a general discussion of natural selection and evolution as an explanation for complexity. He also describes earlier attempts at other explanations, including an interesting argument by Hume. He also explains some of the many pitfalls that can trap the unwary when thinking about evolution. For example, Lamarckianism is right out, but if you have creatures with even a rudimentary ability to adjust their behaviour (“learn”) then you get something that looks very much like it, thanks to something called the Baldwin Effect. Another example he discusses at length is species and speciation.

Species are not a hard and fast concept: just about any formal characterisation one attempts (necessary or sufficient conditions) will have exceptions. Moreover, speciation (two species emerging from just one ancestor species) is something that can only be detected in retrospect. We look back and say, Ah, that’s the point when those two populations have irrevocably split: there’s no subsequent interaction between them. It’s only with knowledge about what’s to happen in the future that we can identify speciation events.

Dennett is similarly good about Mitochondrial Eve: every woman has one mother, and sometimes a mother has multiple daughters. Therefore, moving back in time, from generation to generation, there are fewer and fewer ancestral mothers (i.e., mothers with descendents now living). Therefore, there must come a point in time when there is just one such mother. She is our Eve, and there is nothing special about her whatsoever. (Our Y-chromosome Adam is likely to have been more recent, and thus a descendent of Eve’s, because men have more sons than women have daughters.)

Dennett includes a couple of chapters about how Darwin’s idea can be invoked to provide non-sky-hook explanations for the origins of life, and even the universe itself. Then he moves onto showing how a variety of scientific ideas and thinkers have come unstuck when dealing with evolution. He singles out Stephen Jay Gould, Noam Chomsky, Jerry Fodor, John Searle and Roger Penrose. He seems pretty convincing on how these thinkers seek to deny Darwin’s unescapable conclusion: complexity (life-forms, consciousness, culture) can all be explained by a simple-minded algorithmic process. On the way he provides some neat thought experiments. One I particularly enjoyed was about building a robot to protect one’s cryogenically frozen body for four hundred years so that it might be revived in time for the arrival of the aliens. What strategies should you adopt in the design of such a robot?

This is a very good book. Occasionally the going is a bit tough, but the ideas within it are both fascinating and very important. So, get out there and read it today!

To review next:

Granta 82: life’s like that.