2.22.2007

what good is half an eye? plenty!

From the Guardian's blog, an interesting tale about a bionic eye turns into an effective argument against Intelligent Design. (via Evolving Thoughts)

Part of the reason that ID is such a seductive argument is the way that adaptation is often referred to in popular culture. No natural history documentary script would be complete without a description of the polar bear as "perfectly adapted" to its habitat or a eulogy to the "exquisite camouflage" of the arctic fox.

But nature is much more interesting than the Discovery Channel would have us believe. Look closely and much of it wouldn't be winning any design awards. Rabbits, for example, have an inefficient and frankly gross way of digesting their food. Your furry pet has a side branch to its gut that is full of enzymes and bacteria. By munching on half-digested morsels from this side branch that have passed out of its backside the rabbit's stomach and intestines have a second go at extracting nutrients. It works, but from a design point of view it is crazy.

Despite looking pretty impressive, the human eye itself is put together in a way that a fairly unintelligent designer could improve on. The rod and cone cells that gather light and convert it into electrical impulses destined for the brain are wired up "back to front". So light hitting the retina has to pass through a maze of wiring before it reaches the light-gathering rods and cones. And anyone with a bad back could be forgiven for cursing our knuckle-dragging ancestors who gave us a spine that is not well designed for upright support. The point is, though, that all of these bad designs, with their echoes of ancestors long since gone, are good enough.


And as to the old creationist argument that half an eye (or half of any complex structure) would do no good, well, consider that one BUSTED. A rudimentary light-sensing patch with 4x4 resolution is far better than no light-sensing equipment whatsoever; a cochlear implant with 16 channels is better than no hearing at all. It's easy to see how just a single cell with the ability to sense where the light is conferred a huge advantage on the organism that posessed it. So too with the first animal to evolve a two-celled light sensing organ. Two becomes 4, 4 becomes 8. Add some protective extracellular material, wire up some neural networks, and add a couple hundred million years of hardcore selection for the ability to sense one's environment more and more acutely... poof! the vertebrate eye!

"Nothing in biology makes sense except in light of evolution." T. Dobzhansky

3 comments:

Charles said...

Nilsson and Pelger demonstrated via a computer simulation how an eye can form with minor mutations to the shape of a sensor membrane and the refractive indecies of a covering membrane.

One of my favorite argument by IDers is irreducible complexity. It's got this brutal honesity to it, basically the creationist saying "I'm to stupid/uncreative to see how this system developed through a discreet serious of improvements"

I think we should flip the script, and use a similar argument. Look at biological systems, and suggest improvements that would be very unlikely to arrise through an evolutionary process, but would be an undeniable improvement.

For example, what about a two lens eye? It would be an incredible benefit to survival. Imagine, zoom vision! But to develop the second lens you would obliterate the quality of the first eye.

You can think of it as a fitness landscape, one hill with a very steep valley leading up to a much higher hill.

kat said...

ahh, landscapes. the term is getting some serious use in computational biology, especially in protein folding and (as you mentioned) evolutionary fitness. maybe i should go back and take some more multivariable calculus, now that i understand the applications better...

and isn't everything ID saying "i'm too stupid/lazy to figure this out, so it must have been GOD"? this is the fundamental critique of ID - that it isn't science, because it doesn't encourage further investigation. without testable hypotheses and the neverending search for better explanations for phenomena, ID is just laziness.

but we already knew that.

charles said...

Mmm, fitness landscapes. Reminds me of the best ID argument I heard. Dembski applied the No-free-lunch theorem, one that explains that there is no optimal algorithm for benefit maximization across all possible fitness landscapes.

Hill climbing (evolution), random walks, simulated annealing all perform equally well on average.

So, one can conclude that evolution would provide no more benefit than completely random, undirected mutation.

Dembski probably got the idea because one of the reasons for the research was to attack the idea that genetic algorithms where somehow superior than more traditional searches in general.

The problem is biological optimization is not a general problem. The kinds of fitness landscapes are not evenly distributed, and in fact trend towards the ones that hill climbing is superior on. Further indicting Dembski's silliness is he failed to appreciate that in evolution, the cost functions are not static, which was the assumption with the NFL theorem.