Thursday, September 3, 2009

Evolution, Purpose, & Bad Arguments

I'm just now getting a chance to write about this post yesterday by Jim Manzi, who's filling in at Andrew Sullivan's, on The Evolution of God and Manzi's defense (or at least argument for plausibility of) the thesis of divine purpose behind evolution.

On the most basic level, he's right, of course. The existence of God can't be disproven any more than Bertrand Russel's Teapot can be disproven. But Manzi claims something else; or rather, implies it rather than making an explicit claim.

Manzi begins with an explanation of genetic algorithms as a way of introducing concepts of genetics, including the idea of using evolutionary functions as a way of optimizing on multiple dimensions at once. And while I'd have a minor quibble or two with his wording, his explanation is correct; he clearly gets it.

However. He observes that genetic algorithms are used to find the best combination of 'genes' (which may be data inputs or weights, settings on controls...almost anything, really) for a given purpose. Yes, that's true, according to some externally defined fitness function, a way of boiling everything down to a definitive way of determining, for any two arbitrary members of the population, that this one is a better fit than that one.

In biological systems, of course, the "fitness function" is whether or not the organism survives long enough to breed and for its offspring to reach breeding age. (If it gives birth to a hundred young, and eats 99 of them, that's less fitness than one that gives birth to 3 offspring and carefully nurtures and protects them to adulthood.)

Manzi claims, correctly, that genetic algorithms can select the "best" combination. But: In biological systems, the real world, the "best" combination is determined by survival. This can include high-level activities, especially in social species--if I'm a good neighbor, my children will be taken in and raised even if I'm eaten by a bear.

Manzi seems to be making the implicit assumption that if there is a purpose to it all, the purpose must be to develop a mind capable of apprehending God.

(Which is terribly anthro-centric... Even if there is a purpose, why should it have anything to do with us? Why shouldn't the purpose of creation be to develop the perfect jellyfish? Using Manzi's own logic, this hypothesis can't be disproven.)

But evolution doesn't select for apprehension of the Divine, it selects for survival. Therefore, unless such apprehension, or its precursors, conferred a survival advantage on our ancestors, evolution wouldn't select for it. Is there any evidence that apprehension of the Divine, or its precursors, increased survival? Not that I'm aware of. Yes, religious belief is widespread, and was so in the ancient world, but that doesn't mean it's genetic. A study mentioned in Newsweek a few weeks back found that religious beliefs plummets in advanced societies with low levels of social dysfunction (poverty, crime, etc). If it was genetic, it wouldn't fade out so quickly. On the other hand, social dysfunction was ripe in the ancient world.

(Yes, this is the 'opiate of the masses' theory. Or, if you prefer, the "give me sense of being in control or at least a way to tell myself that someone's in charge" theory. Something can be universal or nearly so, and not biologically determined, if the environmental conditions that foster it are universal as well.)

So. If there's a purpose to it all, evolution probably isn't selecting for it. If you remove the assumption that if there's a purpose to it all then obviously the purpose of all creation is to result in us, you're left with...not much.

(Contrary to what some seem to think, evolution doesn't result in an "ascent" toward higher species... that sort of categorization is something popularizers put on it, but actual people working in the field don't. It's more like a bush branching out in all directions at once, not a tree climbing higher and higher.)

Also, a curmudgeonly point but one that needs consideration: We've existed as a species only about a million years or so, hardly a blink of an eye in geological time, and much less than many other species that were once dominant. Given how rapidly we seem to be making the planet uninhabitable and that we may yet succeed in wiping ourselves out, I'd say the long-term survival value of intelligence is still something that awaits definitive demonstration.

At any rate, Manzi seems to have some recognition of the difficulty he's in here. He then falls back on some unseemly hand-waving, simply declaring it out of bounds for science, in bounds for philosophy, and therefore he'll believe whatever he pleases:

A scientific theory is a falsifiable rule that relates cause to effect. If you push the chain of causality back far enough, you either find yourself more or less right back where Aristotle was more than 2,000 years ago in stating his view that any conception of any chain of cause-and-effect must ultimately begin with an Uncaused Cause, or just accept the problem of infinite regress. No matter how far science advances, an explanation of ultimate origins seems always to remain a non-scientific question.

Now consider the relationship of the second observation to the problem of final cause. The factory GA, as we saw, had a goal. Evolution in nature is more complicated — but the complications don’t mean that the process is goalless, just that determining this goal would be so incomprehensibly hard that in practice it falls into the realm of philosophy rather than science. Science can not tell us whether or not evolution through natural selection has some final cause or not; if we believe, for some non-scientific reason, that evolution has a goal, then science can not, as of now, tell what that goal might be.
That's right, there's nothing for Science to do but throw up its hands and believe in teapots!

The combination of a constantly changing fitness landscape and an extraordinarily large number of possible genomes means that scientists appropriately proceed as if evolution were goalless, but from a philosophical perspective a goal may remain present in principle.
And having decided that it "may remain present in principle," Manzi assumes it into being. But he runs into other problems as well.

But in fact, even the “random” elements of evolution that influence the path it takes toward its goal — for example, mutation and crossover — are really pseudo-random. For example, if a specific mutation is caused by radiation hitting a nucleotide, both the radiation and its effect on the nucleotide are governed by normal physical laws.
He's apparently never heard of quantum physics. Those "normal" physical laws include fundamental limits on what's knowable or predictable. We don't call them random, especially an event such as a nucleus ejecting a beta particle, i.e. a bit of radiation-- just because they're too insanely complicated for us to understand--they really are random. This has been taught in any college-level physics course for decades. He seems to admit this a paragraph later, but obviously hasn't thought through what it implies. Yes, statistically we can talk about half-life, which is an average, because we don't need to know which precise nucleus broke down--but when a phosphorus atom inside a DNA strand decays and is suddenly something else, it does matter which precise atom decayed, and which direction the beta particle was heading.

The theory of evolution, then, has not eliminated the problems of ultimate origins and ultimate purpose with respect to the development of organisms; it has ignored them. These problems are defined as non-scientific questions, not because we don’t care about the answers, but because attempting to solve them would impede practical progress. Accepting evolution, therefore, requires neither the denial of a Creator nor the loss of the idea of ultimate purpose. It resolves neither issue for us one way or the other. The field of philosophical speculation that does not contradict any valid scientific findings is much wider open to Wright than Coyne is willing to accept.
Well, only by shifting definitions enough to keep them that way. Of course, the beauty of this particular dog-and-pony show is that the goalposts can always be moved. If we work out, for example, why pi has to have the exact value it does or why our physical laws are the way they are--perhaps no other set of laws would be stable--Manzi can always say "Yeah, but why is that?" and then declare that to be the 'ultimate' question. And when it's answered, he can find another. The problem isn't that "we don't care about the answers," it's that the question is poorly posed.

Might there be a purpose to it all, even if there's no evidence of such purpose? Sure. It can't be disproven. Absence of evidence isn't evidence of absence. But that's no reason to take it any more seriously than hen's teeth or orbital teapots. If someone wishes to believe there's a purpose to the entire universe, and they're it, that's their business, but "you can't prove it isn't!" is also no proof that it is. If you're claiming such a purpose exists, in the absence of any evidence, then you need to produce such evidence, or solid reasons why the lack of evidence doesn't matter. Unfortunately, Manzi has done neither.


Jim said...

Thanks for the very thoughtful comments. a few quick things:

1. I really wasn't trying to argue that purpose is implicit, likely or anything else in evolution, but rather that it doesn't give us any answer to this question one way or another.

2. From the perspective of an organism in actual evolution or in the GA example, there is no understood objective other than survival. Just because the factory control genome doesn't "know" the overal purpose is to maximize output, doesn't mean that isn't the purpose.

3. I didn't want to get into the QM issues, because Coyne's argument was that evolution through natural selection proved there was no divine plan. That's why I was careful to say that evo provides no incremental randomness beyond QM (and, potentially, other scientific findings). Further, while it is a commonplace, even among scientists, that many QM events are non-deterministic, it is not a required impplication of Bell's Theorem (which is typically what is cited as the argument for on-determinism). There is what is termed the 'casual interpreation' of QM.

4. I'm interested to know what the error in my GA description is. I actually pseudocoded the GA for a software product, and used my documentation to make sure I walked through it correctly.

Jim Manzi

KCProgramr said...

Mr Manzi:

Wow, I sometimes forget that anyone other than myself ever reads this....

As to your comments:

1) True. I snarked at length on the "the purpose is to evolve a mind that can apprehend God" meme, and while I've heard it presented with a straight face and I happen to think it's a risible argument, it's not an argument or claim that you made. I was wrong to attribute that belief to you.

That said: as science advances, arguments about purpose keep getting pushed farther and farther back. At one time, the anatomy of the eye was held up as a refutation of atheism; as intermediate stages of the eye's evolution have been worked out, and examples found in other species, there are fewer and fewer such claims made.

As you point out, you can always carry things back to an Uncaused Cause if you push hard enough. But we're veering quite close to orbital-teapot or white-crow territory. Anyone claiming such purpose exists has the burden of producing evidence or at least a good argument for it. As more and more evidence piles up, and a spectacular lack of evidence for purpose becomes more and more evident, the "no purpose" hypothesis becomes stronger.

3) Agreed that QM was off-topic for the post, but you seemed to imply that the only real thing standing in the way of being able to replicate (and predict, and thus deduce a cause for) the whole shebang was a lack of computational power (which I agree would probably take a universe-size computer). I'm not sure that even if we had "all we needed," that it would still be availing. And yes, this drags us firmly up against Bell's Theorem. Which I will readily concede being rusty on.

4) I spend most of my time teaching undergraduates, so I tend to read through the filter of "what could be misinterpreted by a freshman." And (he said, slightly chagrined) I seem to have mostly blipped over the word "roughly" in your every-10,000th gene in the mutation phase. My bad.

I assume that by "roughly every 10,000th gene" you mean that each gene has an independent 1-in-10,000 probability of mutating. No position in the genome is favored or disfavored, and nothing says mutations have to be a certain minimum or maximum distance apart. (At least for purposes of a GA.)

I'd make the quick side observation that "genetic" data can be coded in several different ways, and for some encodings, mutations can take different forms--perhaps nudging a value up or down by some proportion, or for two genes in the same genome to trade places (a transposition mutation). But as the main point of your post wasn't GA's except as a way of drawing a larger point, I wasn't too worried about it. In fact, it's one of the clearer explanations of GA's for the lay reader that I've come across.

Unless, of course, I were to write up one myself. ;)

Thanks for a thought-provoking exchange!