ANCESTRAL IMAGES?

Dick Pountain/17 May 2010 14:09/Idealog 190

A perennial theme of this column is correspondence between computers and living creatures, and in particular the way DNA resembles or does not resemble a computer program. A couple of random events over the last two weeks popped this topic back into my mind in a way that bears retelling. Darwinism is hardly front-page news nowadays, but it's very rarely off the features page.

Sure enough "What Darwin Got Wrong", a controversial new book by philosophers Jerry Fodor and Massimo Piatelli-Palmarini questions the simple account of natural selection and adaptation offered by most contemporary Darwin defenders. I hasten to add that it offers no support at all to creationists or evolution sceptics, but is a dispute over important details concerning the mechanism of evolution. (There's a magisterial review of it in the May 27 New York Review of Books by Richard Lewontin, who along with his late co-author Steven Jay Gould is no stranger to upsetting Darwin purists). In a nutshell Fodor and P-P demonstrate that the "just-so" stories evolutionary biologists tell about the way particular traits, say plumage colour, are adaptations to an environment are speculative and a small part of the whole truth (most of which will never be known).

Now a rapid change of scene to Scotland. Marion and I are staying with her brother Pip in St Cyrus, a coastal village south of Aberdeen, where I meet a dinner guest, E, who's a genealogist and has just traced Marion and Pip's family tree. This reveals ancestors who lived in St Cyrus 400 years ago, a fact totally lost to the family until now. Pip had left his long-term residence in Edinburgh five years ago and chose St Cyrus because it "felt right". E suggested that this was no coincidence, and we had a friendly disputation about the nature of heritability and whether it could stretch to landscape preferences, which I doubted.

Back in London with genealogy on my mind, I looked out a half-hearted attempt I made a few years ago using the Mormon genealogy website. In the same folder was a PDF from my cousin Joy containing a superb family history she'd recently compiled, which I'd misplaced and hadn't yet read. I told me that my grandmother's family migrated from the Tamar Valley (on the Cornwall/Devon border) in 1877 to Derbyshire where I was raised. Now in the late 1980s close friends of mine moved from London to Gunnislake on the Tamar and I've spent much time there visiting. And there was something about its landscape that I'd found curiously congenial...

I think you can see where this is headed. It's very easy to construct mystical theories about the way the landscape where you're born gets into your "bones" and passed on to your children and children's children. However I'm very much not a mystic, and it's very, very difficult indeed to construct any materialist theory of how any such inheritance could take place.

Once Darwin's version of evolution was supplemented by the discovery of genes, and finally in the 1950s by DNA, theories like those of Lamarck that claim organisms can pass on characteristics acquired during their lifetime go out of the window, and the memory of a landscape is precisely such a characteristic. The "central dogma" of molecular biology holds that information flows only in one direction, from DNA into RNA, from RNA into protein, and from protein into flesh and bone. You don't pass on tattoos to your children because there's no information pathway leading from that bluebird punctured into your arm back into the DNA of your germ cells, and similarly there's no pathway from an image stored in your brain back into DNA.

Here's where we get to the computer/DNA analogy, because this is not true of computer code. Most of a native-code computer program consists of processor instructions that get executed in sequence, but not all of it - there are sections of data mingled in with those instructions, such as strings to be displayed on the screen or look-up tables for selecting pixel colours. There's no one-way restriction on information flow either, so it is possible for a computer program to change its own data, as it does whenever a patch is applied. Configuration programs often used to rewrite literal values in an application's code, and though it may be frowned on by fastidious programmers (who prefer to store data in separate config files and read it at load time) it is possible.   

Is it possible to imagine any mechanism at all via which a landscape could become encoded into human DNA? We're walking the line between science fiction and fact. It would have to invoke epigenetics, the study of histones, alkaline proteins that surround the strands of DNA and act as switches to turn genes on and off. It transpires that the status of such switches, set during an individual's lifetime, can be passed on to offspring, an example being that starved mothers can pass on a metabolic-rate switch to their babies. A visual pattern stored in brain would have to be encoded into histones, these histones inherited and their pattern matched against visual inputs to trigger a gene for an emotional response. Quite far-fetched. Far more likely that Granny mentioned Tamar to me in infancy and I've forgotten...