20 December 2009

Weasels, clouds and biomorphs, part III

The Blind Watchmaker is a superb book by a masterful science writer. It's not just a book about evolution, or even about how evolution works. It's a book about how evolution explains design, and more specifically how natural selection accounts for design. As I wrote before, I consider chapter 3 to be the most important chapter of the book. The chapter is called "Accumulating small change" and it features two different computer programs that Dawkins uses to teach readers about the effectiveness of selection in evolution. Before we play with the Biomorph program in the next post, allow me to set us up by discussing the importance of the program in Dawkins' argument, and by outlining the logic of the program's design.

First let me try to convince you that chapter 3 really is the heart and soul of the book. The chapter is about gradual, step-by-step evolution resulting from natural selection. And as you might already know, natural selection is what Dawkins considers to be The Big Idea, the idea that answered Paley's seemingly insurmountable challenge. In chapter 2, Dawkins makes this clear. Here's how he starts.

Natural selection is the blind watchmaker, blind because it does not see ahead, does not plan consequences, has no purpose in view. Yet the living results of natural selection overwhelmingly impress us with the appearance of design as if by a master watchmaker, impress us with the illusion of design and planning. The purpose of this book is to resolve this paradox to the satisfaction of the reader....
–The Blind Watchmaker, page 21
Chapter 2 famously focuses on echolocation in bats, and I would buy the book just to read Dawkins' description of the engineering feat that is the little brown bat. (He gleefully recounts the utter incredulity of an audience of biologists when the mere existence of such biological phenomena was first described.) And here's his conclusion.
I hope that the reader is as awestruck as I am, and as William Paley would have been, by these bat stories. My aim has been in one respect identical to Paley's aim. I do not want the reader to underestimate the prodigious works of nature and the problems we face in explaining them. Echolocation in bats, although unknown in Paley's time, would have served his purpose just as well as any of his examples. Paley rammed home his argument by multiplying up his examples. He went right through the body, from head to toe, showing how every part, every last detail, was like the interior of a beautifully fashioned watch. In many ways I should like to do the same, for there are wonderful stories to be told, and I love storytelling. But there is really no need to multiply examples. One or two will do. The hypothesis that can explain bat navigation is a good candidate for explaining anything in the world of life, and if Paley's explanation for any one of his examples was wrong we can't make it right by multiplying up examples. His hypothesis was that living watches were literally designed and built by a master watchmaker. Our modern hypothesis is that the job was done in gradual evolutionary stages by natural selection.
–The Blind Watchmaker, page 37
Chapter 4 builds on chapter 3, and the rest of the book deals with how it might all work. Chapter 3 is Dawkins' attempt to show us the power of cumulative selection, and cumulative selection is The Blind Watchmaker. This is the heart of the matter, and Dawkins' argument (and his world) hinges on the success of this idea.

And so Dawkins tackles the concept of cumulative selection in chapter 3, and as we've already seen, he immediately faces a serious problem: the end result of an evolutionary process is the generation of design, of biological machines that are complex and, more importantly, wildly improbable. In other words, such things "can't just happen." The human mind is prone to a serious error when faced with this challenge. The error is to envision complexity arising spontaneously from chaos, in a single step, and thus to conclude that such things cannot be explained naturally. The error is in bold, and Dawkins addresses it first with the simple and effective Weasel illustration. The illustration is highly effective as a corrective for that error, but it fails as a model of evolution, as I explained in the previous post.

The Biomorph program was Dawkins' more serious attempt at modeling the development of complex structures by cumulative selection. It's important to understand just how central the program really is, and thus why it's so silly to make a big deal out of the Weasel exercise. Chapter 3 is the heart of the book, and the Biomorph program is the soul of chapter 3. The Biomorph program improves on Weasel in two very important ways:

1. It models evolutionary unfolding without a specific goal. The Weasel program "homed in" on a particular goal; the Biomorph program has no such constraint.

2. The entities that evolve in the Biomorph program, called biomorphs, "develop," and their development is controlled by a number of factors ("genes") which change (i.e., mutate) in each generation, so that mutations result in alterations to development and thus to new forms.

The biomorphs are tree-like structures, and they are drawn according to simple rules. (This post is decorated with a few that I made using a nice Java applet.) The rules control the branching of the trees (branch at a certain angle or at a certain point on the existing branch, or branch of a certain length, or whatever). The drawing of a biomorph, then, is a representation of embryonic development. And the rules represent the various processes in development.

It should be fairly easy to see how to model the effects of genes: a gene will influence a rule, by assigning a number to the rule (e.g., branch at a bigger or smaller angle). Reproduction is simple: the biomorphs are redrawn, based on the parent's structure, using the same rules influenced by the same genes. Boring? No: mutation acts to change the numeric value of the genes, randomly changing the value by either +1 or -1. The result is a set of offspring, each differing slightly from the parent by virtue of a single mutation.

So it goes like this. A parent is selected. The subroutine REPRODUCTION runs, and generates random mutations in each of the genes of the parent (there are 9 genes); the new genes are passed to the subroutine DEVELOPMENT, which draws new biomorphs based on the new genes. The result is a set of 9 offspring, each with a different version of one of the parent's genes. One is selected to be the parent of the next generation. Keep doing this, over and over and over, and you get the program EVOLUTION.

But how does selection work in this program? Recall that the major problem with the Weasel illustration was its goal-directed nature. In the Biomorph program, things are different:
...the selection criterion is not survival, but the ability to appeal to human whim. Not necessarily idle, casual whim, for we can resolve to select consistently for some quality such as 'resemblance to a weeping willow'. In my experience, however, the human selector is more often capricious and opportunistic. This, too, is not unlike certain kinds of natural selection.
The Blind Watchmaker, page 57
Selection, in other words, is done by you, the human who is "playing God."

That's the Biomorph program. Next time: what it demonstrates about evolution.

08 December 2009

Weasels, clouds and biomorphs, part II

Back in September I wrote about the silly preoccupation on the part of various anti-evolutionists with the so-called Weasel program, a simple exercise created more than 20 years ago by Richard Dawkins to illustrate the efficacy of cumulative selection in evolutionary scenarios. My main point was that the Weasel program had one very simple purpose (comparing "single-step selection" – which is purely random – to cumulative selection) and constitutes a trivial fraction of the argument in Dawkins' The Blind Watchmaker.

One might think that Dawkins' basic message – that random one-step flying-together of a Shakespearean phrase or a hemoglobin molecule is impossibly unlikely compared to cumulative selection of intermediate stages – is so elementary that no intelligent person would need to consider it more than once. (Once seems like a lot to me in this case, but never mind.) And yet the error (if that's what it is) is shockingly common. (It forms one pillar of poor Cornelius Hunter's whole enterprise, for example.)

Dawkins understood this problem when he wrote The Blind Watchmaker, so before he unveiled the fascinating program that forms the heart of his case for the power of selection, he took one last stab at making the basic outline clear by going back to clouds. Wait...clouds? (Fans of Hamlet are already sighing blissfully; those who don't get the connection between weasels and clouds should read either Act III, Scene II of Hamlet or Chapter 3 of The Blind Watchmaker.) Yep. Dawkins pointed back at what makes cumulative selection work: the things that are evolving must be able to generate related offspring. And that's what clouds can't do.

Clouds are not capable of entering into cumulative selection. There is no mechanism whereby clouds of particular shapes can spawn daughter clouds resembling themselves. If there were such a mechanism, if a cloud resembling a weasel or a camel could give rise to a lineage of clouds of roughly the same shape, cumulative selection would have the opportunity to get going. Of course, clouds do break up and form 'daughter' clouds sometimes, but this isn't enough for cumulative selection. It is also necessary that the 'progeny' of any given cloud should resemble its 'parent' more than it resemble any old 'parent' in the 'population'... It is further necessary that the chances of a given cloud's surviving and spawning copies should depend on its shape.
The Blind Watchmaker, pages 50-51, italics in the original
Hence the Weasel program.

But I noted last time that Dawkins spent a tiny amount of time and text on the Weasel program, and that he declared it to be "misleading in important ways." The most important, by far, is this: the selection that drove the Weasel program was goal-directed. A better simulation of evolution would be one in which selection is more capricious, more "in the moment." (Survival can be capricious; reproduction happens rather decisively "in the moment.")

Dawkins came up with just such a program, and I mentioned it in the previous post. It's a wonderfully simple simulation of the basic aspects of selection-driven evolution: it includes development, reproduction, genes, and selection, and generates "organisms" with shapes instead of a phrase in all caps. We'll look at that program in the next and final post. But if you want to play with a modern version, you'll find plenty of nice implementations out there. So much more fun than studying...or grading.

07 December 2009


So there hasn't been a post on Quintessence of Dust in three months. Here are some reasons for this.

  1. We're in the midst of a distracting crisis at Calvin College right now. I won't talk about it here, but it's very serious and has already affected my relationship with the college.
  2. I recently completed a major writing project, a book that I coauthored with a colleague at Calvin. Not much else to say about that at this point.
  3. I've coauthored two recent review articles with my colleagues at the Van Andel Institute, and we have significantly expanded the scope of our collaboration, with a major new grant proposal in the works.
During this hiatus, I enjoyed a wonderful attack (and a flatulent followup) by a poorly-equipped ID demagogue, Cornelius Hunter. God bless him: he said I'm "the best of the worst." There's not much there, so I'll ignore his criticism and work instead on completing two interrupted series, one on deep homology & design and one on Richard Dawkins' biomorph program. Then we'll get back to regular forays into the current literature, and we'll focus on "junk DNA" a bit more. Biomorphs are next. See you soon.