Life goes non-linear

October 18, 2010 12:29 PM

The chaos theory of evolution
posted by Artw (33 comments total)

Um... Is this article written by the Jeff Goldblum character from Jurassic Park?
posted by Sys Rq at 12:51 PM on October 18, 2010 [1 favorite]

Oh Jesus...
posted by Artw at 12:53 PM on October 18, 2010

No. It was written by Keith Bennett, a professor of late-Quaternary environmental change at Queen's University Belfast.
posted by pwally at 12:53 PM on October 18, 2010

It's quite possible that if I hadn't blanked that from my brain i would have based a shitty pun for the title around it though.
posted by Artw at 12:54 PM on October 18, 2010 [1 favorite]

I know about the theory of evolution. I know about chaos theory. What the article doesn't seem explain is how chaos theory modifies our current understanding of evolution. I saw the term "non-linear" thrown out there, but anyone who has used a differential equation could probably have said the same thing. Is there someone familiar with Bennett's research who can say what is groundbreaking about his work?
posted by Blazecock Pileon at 12:58 PM on October 18, 2010 [1 favorite]

Interesting stuff. Here's the key passage:

The overall picture is that the main response to major environmental changes is individualistic movement and changes in abundance, rather than extinction or speciation. In other words, the connection between environmental change and evolutionary change is weak, which is not what might have been expected from Darwin's hypothesis.

If environmental changes as substantial as continent-wide glaciations do not force evolutionary change, then what does? It is hard to see how adaptation by natural selection during lesser changes might then accumulate and lead to macroevolution.

I suggest that the true source of macroevolutionary change lies in the non-linear, or chaotic, dynamics of the relationship between genotype and phenotype - the actual organism and all its traits. The relationship is non-linear because phenotype, or set of observable characteristics, is determined by a complex interplay between an organism's genes - tens of thousands of them, all influencing one another's behaviour - and its environment.

Not only is the relationship non-linear, it also changes all the time. Mutations occur continually, without external influence, and can be passed on to the next generation. A change of a single base of an organism's DNA might have no consequence, because that section of DNA still codes for the same amino acid. Alternatively, it might cause a significant change in the offspring's physiology or morphology, or it might even be fatal. In other words, a single small change can have far-reaching and unpredictable effects - the hallmark of a non-linear system.

I'll be interested to see what those who know more about it than I do have to say.
posted by languagehat at 1:01 PM on October 18, 2010 [1 favorite]

He seems to be saying that we shouldn't be so tunnel-focused on natural selection as the main driving force of evolution, which is something many evolutionary biologists would agree with. There are many other factors at work, including genetic drift, which he spends a little time on. He goes off the rails for me towards the end by saying things like "we can have no laws of evolution". Okay, sure, maybe our ability to predict evolutionary change due to climate change is extremely limited. But evolution still has incredible predictive power in many other areas. He also seems to favor the work of Fodor and Piattelli-Palmarini, which has been roundly trounced by a number of biologists.
posted by lholladay at 1:06 PM on October 18, 2010

Fodor and Piattelli-Palmarini get everything wrong.

It does interest me how New Scientist seems to have decided to try to undermine pretty well-understood aspects of evolution. Man, that magazine has gone downhill.
posted by Decani at 1:17 PM on October 18, 2010 [2 favorites]

Decani- the text you linked to seems pretty weak in my opinion- lots of fiery rhetoric about people being dead wrong which after all the qualification and details it turns out only to be a disagreement about emphasis and significance. It's also rather undermined since the guy only read a new scientist summary of Fodor and Piattelli-palmarini and not the original text.
posted by leibniz at 1:51 PM on October 18, 2010

leibniz: here's another take-down from a pair who did read the whole thing.

Fodor and Piattelli-Palmarini are not biologists. Fodor is a leading philosopher of mind and cognitive scientist, best known for his ideas about the modularity of mind and language of thought; Piattelli-Palmarini is a cognitive scientist. They do not have new data, new theory, close acquaintance with the everyday practice of evolutionary investigations, or any interest in supplying alternative explanations of evolutionary phenomena. Instead, they wield philosophical tools to locate a “conceptual fault line” in contemporary Darwinism. Apparently unshaken by withering criticism of Fodor’s earlier writings about evolutionary theory, they write with complete assurance, confident that their limited understanding of biology suffices for their critical purpose. The resulting argument is doubly flawed: it is biologically irrelevant and philosophically confused. We start with the biology.


They do a pretty good number on it after that, in my opinion.

This stuff took a major hammering from all directions at the time it came out last February. Once again, it's interesting that New Scientist still cites it eight months later as though none of that happened. Something is going on over there. Something bad.
posted by Decani at 2:42 PM on October 18, 2010 [5 favorites]

So, just to be clear here, the offending phrase is More recently, and controversially, cognitive scientists Jerry Fodor of Rutgers University at New Brunswick, New Jersey, and Massimo Piattelli-Palmarini of the University of Arizona in Tucson have pointed out philosophical problems with the adaptationist argument.
posted by Artw at 2:47 PM on October 18, 2010

Block and Kitcher are not biologists, either.
posted by No Robots at 2:52 PM on October 18, 2010

Meh. To me, this looks like yet another argument against a straw-adaptationist. Chaos theory in biology isn't especially new; we know that predator/prey modeling is functionally chaotic. The idea that evolution indirectly selects genes via phenotypes which are the sum of inherited and environmental factors is BIO 201 stuff here, and the notion that environmental change might cause a change in the geographic distribution and abundance of a population isn't exactly news.

But it's New Scientist, a periodical that makes headlines on bombastic reporting that mainstream science could be wrong.
posted by KirkJobSluder at 2:53 PM on October 18, 2010 [3 favorites]

Aside from all the normal, extremely well-deserved New Scientist bashing, I don't know why one should be surprised that the fossil record doesn't show a lot of speciation in direct response to climate events. To my understanding (and I'm not an evolutionary biologist, but I've hung out with them a lot), it is thought that the dominant mode of speciation occurs when a population gets separated by some barrier for a long time and genetic drift (and maybe some selection, but selection isn't needed) makes them get farther apart. Environmental changes could assist this by helping move populations around, but it strikes me that it's far easier for species to go where they can already live than suddenly adapt for a new environment.
posted by Schismatic at 2:58 PM on October 18, 2010

Of course evolution involves chaotic systems, because there is feedback between phenotype and environment. Stochastic processes (which he doesn't mention) are also involved through environmentally caused genetic mutations. He never gets around to explaining how any of this flies in the face of adaptation or how the fossil record through ice ages contradicts it. I suspect he thinks that adaption means optimization and thus the same environment should lead to the same phenotype every time. But that can't be true, because we know diverse species occupy an ecosystem, each with its own set of adaptations. Surely this is obvious.
posted by Mental Wimp at 3:11 PM on October 18, 2010

This is a dreadful, sloppy article. Could be that it's been popsci edited to hell, I guess. Damned if I know why Bennett has chosen Fodor & Piattelli-Palmiari to hang his hat on, too: critiques of pure adaptationist evolution go back far longer and are far more cogent than their particular straw man.

I'm profoundly unimpressed with the laboured points about "non-linearity" (a particular culprit: "This self-similarity also indicates that evolutionary change is a process of continual splitting of the branches of the tree." - as opposed to?). "Unpredictable" is also not the same as having low explanatory power, but he's not distinguishing.

In fact, I think he's missing a lot of information: there are more than a few [pdf - chapter 1] contemporary non-adaptationist mechanisms which generate testable, falsifiable predictions of future states of about exactly the patterns Bennett talks about (and which seem to do a good job of doing so). And there's a large and growing body of evidence that seems to be suggesting repeated and predictable patterns in macroevolution - if you look at large data sets at relevant timescales, which I don't think Bennett does here.

Not good.
posted by cromagnon at 3:15 PM on October 18, 2010 [1 favorite]

All fun and games until you publish a book with the title "Darwin wuz Rong!"
posted by ovvl at 4:06 PM on October 18, 2010

Certain new-agey types are fond of taking chaos-theoretical ideas and claiming they validate whatever bogus theory they're on about. Now that this article has the cover of PopSci, I imagine the Intelligent Design proponents are going to be very happy.
posted by LogicalDash at 4:47 PM on October 18, 2010

MUCH easier to follow if you read it in a fake Norn Iron accent. I like Ian Paisley - YMMV.
posted by sneebler at 5:44 PM on October 18, 2010

I'm guessing the next article will be on how this chaos can be modeled as a neural network.
posted by Xezlec at 6:34 PM on October 18, 2010 [1 favorite]

That Block and Kitcher article very much misses the point. A few philosophers point this out early in the comments (3, 10, and 16 are all good) before the thread descends into the usual internet riffraff.
posted by painquale at 8:10 PM on October 18, 2010

the usual internet riffraff.

Hey!
posted by Trochanter at 8:35 PM on October 18, 2010

The Block and Kitcher article led to a response from Fodor and Piattelli-Palmarini, and a further rejoinder to that. Some common ground was established but it's clear that Fodor and Piattelli-Palmarini's original poisition was unsound as presented, and the subsequent definitional retreats are there to be seen.
posted by Decani at 3:29 AM on October 19, 2010

I'm profoundly unimpressed with the laboured points about "non-linearity"

My days of non-linearity are way past, but if I may rant:

Non-linear does not mean "funky and exciting". It has a specific meaning, having to do with the mathematical laws governing the dynamic relationship between measurable quantities. Non-linear is only second to quantum as an adjective signaling 'happy fun science, not the kind of boring science that says we can't have nice things'.
posted by Dr Dracator at 4:59 AM on October 19, 2010

The Fodor and Piattelli-Palmarini argument seems a bit weird to me. From what I can tell, their argument is that because because the problem of distinguishing between adaptive traits and free riders is difficult for biologists, natural selection can't do it either. Therefore, the general theoretical model of evolutionary biology is bunk and must be replaced by "natural history."

The problem is, as Meyers points out, that "free riders" are not exactly new. One form of them actually is part of the evidence that allows us to say that natural selection does appear to be a general theory across the tree of life. Because co-variant neutral mutations appear to be statistically random, we can use them as "genetic clocks" to estimate the degree of evolutionary difference between non-randomly selected phenotypes.

And this ignores the fact that the modern synthesis behind contemporary Darwinism was designed around phenotypic variance and confounding variables. We have a probabilistic and statistical theory for a reason. It's not hard to do the math to see that if the phenotype is influenced by inheritance and there's differential reproduction across the range of the phenotype, the distribution of the phenotype will change over generations.
posted by KirkJobSluder at 7:37 AM on October 19, 2010 [1 favorite]

it's clear that Fodor and Piattelli-Palmarini's original poisition was unsound as presented, and the subsequent definitional retreats are there to be seen.

It was not unsound for any reasons that Block and Kitcher mention in their original article and there were no definitional retreats: despite using language in that response like "they now concede," Block and Kitcher plainly mischaracterized their argument in the first place. All F and P-P did is restate their argument in the book and point out that it had nothing to do with causation, as Block and Kitcher claimed. (It's all especially weird given that I saw Fodor and Kitcher give a public debate months before the review in which Fodor explained his views on peppered moths and the like. It's like Kitcher forgot it ever happened.) And once the misunderstandings have been cleared up, Block and Kitcher agree with F and P-P: there is no "theory" of natural selection. Of course, they go on to say that no one has even been committed to such a strong theory, so there is a little debate over what it is that scientists actually believe. But they concede the whole argument that F and P-P is sound, even if attacking a straw man.

None of this was in the original review, but it was all in the original book. The original review missed the point.
posted by painquale at 9:59 AM on October 19, 2010

painquale: Not having access to the original book, but their restatement is profoundly fishy.

Let's start with their strawman of natural selection starting with... :

There is random variation of phenotypic traits.

This is extremely sloppy. Phenotypic traits follow a statistical distribution that can be modeled using random numbers. That doesn't mean that the phenotypes themselves are random. This distinction is critically important. Because it points directly at the underlying mechanism.

There is some ecological variable that is sensitive to the strength of the correlation of such traits with fitness.

It's not clear what this statement is saying and prima facie seems to be putting the cart before the horse.

There is some mechanism that alters the relative frequency of the trait in the population so that, all else equal, it varies with the strength of the correlation between the trait and fitness.

The problem they have here is that mechanism has been verified both as statistical models and experimentally in enough cases that we can fairly say that the mechanism works. To provide a less sloppy definition of natural selection:

1: Phenotypic variance is the sum of inherited and environmental factors. (P = I + E.)
2: Genetic factors are not evenly distributed across the phenotypic range.
3: There is differential fitness across the phenotypic range.
4: Differential fitness across the range of phenotypes discriminates among genetic factors.
5: The distribution of genetic factors within a population changes over multiple generations.
6: A change in the distribution of genetic factors changes the distribution of the phenotype.

My experience is in microbiology rather than zoology or botany. Microbiologists have the advantage of hundreds to millions of generations in a year. But the mechanism described above has proven to be reasonably general for understanding parasite/host co-evolution, drug resistance, community succession, and how to brew better beer. It's been experimentally verified in animals and plants as well.

But it doesn’t. A way to see that it doesn’t (not by any means the only way) is to consider confounded (linked) phenotypic traits, one but not the other of which is fitness-enhancing. Both traits are then correlated with fitness, so both should count as adaptations according to the formulation of natural selection given above. But only one of them is a cause of selection, so only one of them is an adaptation, and, though both are selected, only one is selected-for. Thus the free-rider problem. Prima facie, free riding is a counterexample to natural selection.

There's three obvious objections here. The simplest is that they've invented two definitions of adaptation to conflict, and they need to either pick one or the other. The second is identifying the phenotype as the cause of selection rather than the object of selection. The third is that the free-rider problem isn't that big of a deal as I've previously described.
posted by KirkJobSluder at 11:52 AM on October 19, 2010

What they want is a theory of evolution that applies to any creature in any ecology. We doubt that such a theory is possible. We think that how a creature’s trait evolved in its ecology is likely to be very sensitive to which creature it is, and which trait it is, and which ecology it is.

Well, to me this is about like saying that the Jovian moons, the rings of Saturn, supermassive black holes, Trojan asteroids, and type IA supernovas are sensitive to local conditions, that we can't pose General Relativity as an underlying mechanism. Of course the selective pressures of E. coli are different from those of Canis lupus familiaris. The underlying mechanism that differential fitness across a range of phenotypes results in changes in the distribution of genes within a population appears to be general enough to apply to both.
posted by KirkJobSluder at 12:15 PM on October 19, 2010

It seems to me that all of these theories break down at the moment of creation.
posted by Fundraising Ideas at 12:41 PM on October 19, 2010

Really, I really just stepped in to bash on the Block and Kitcher article, not support the Fodor one... I'm ambivalent about Fodor's position. (Whether F and P-P are right or wrong doesn't change the fact that Block and Kitcher misread the heck out of them.) But I'll try to pretend I'm them and address your comments:

There is random variation of phenotypic traits.

This is extremely sloppy.

It sounds like you're objecting to their use of the term "random," but it's incidental to the argument. Just replace "random variation" with "variation" and their argument is unaffected.

a less sloppy definition of natural selection:

1: Phenotypic variance is the sum of inherited and environmental factors. (P = I + E.)
2: Genetic factors are not evenly distributed across the phenotypic range.
3: There is differential fitness across the phenotypic range.
4: Differential fitness across the range of phenotypes discriminates among genetic factors.
5: The distribution of genetic factors within a population changes over multiple generations.
6: A change in the distribution of genetic factors changes the distribution of the phenotype.

I'm not sure they would have a problem with your definition at all. It depends on how you characterize 'fitness'. If 'fit' just means 'likely to survive', then I think they're fine with it. You've given a mechanism in which whole phenotypes are selected. They have no problems with selection of whole phenotypes. They will have a problem if you say that a certain phenotypic trait is fitness-inducing, however. It doesn't look like the mechanism you've given has the resources to distinguish traits that are selected for. This means that we can't talk about the functions of phenotypic traits. We can't say that the heart evolved for pumping blood instead of making a lub-dub noise. It also means that we can't say that there are such things as free riders (if free riders are traits that are not selected for). If you want to disagree with any of this, you need to say something about phenotypic traits in your description of the mechanism.

Note that F and P-P will agree that we have hearts because our ancestors had hearts and the blood-pumping action of hearts in the past helped our ancestors survive. But this is just natural history and not a statement in an autonomous evolutionary theory. It has no lawlike character---nothing about it will generalize to new instances---and that's a necessary feature of sciences.
posted by painquale at 8:33 PM on October 19, 2010

painquale: You've given a mechanism in which whole phenotypes are selected. They will have a problem if you say that a certain phenotypic trait is fitness-inducing, however.

Which is little more than an objection created by semantic hair-splitting as a phenotype is a trait.

It doesn't look like the mechanism you've given has the resources to distinguish traits that are selected for.

#3. Obviously if variations across a phenotypic range create differential fitness, parts of that range are selected for.

This means that we can't talk about the functions of phenotypic traits. We can't say that the heart evolved for pumping blood instead of making a lub-dub noise.

Why not? Across the animal kingdom, the ability to transport oxygen to tissues has a profound impact on an animal's metabolic needs and lifestyle. The fact that some types of hearts make a lub-dub sound is a side-effect But of course, you're just being argumentative because...

Note that F and P-P will agree that we have hearts because our ancestors had hearts and the blood-pumping action of hearts in the past helped our ancestors survive. But this is just natural history and not a statement in an autonomous evolutionary theory. It has no lawlike character---nothing about it will generalize to new instances---and that's a necessary feature of sciences.

Which basically concedes natural selection while denying it because of a claim about local variables. The problem here is that the general mechanism matches both correlational and experimental evidence. It's a robust theory whether we're talking about influenza's ability to evade the immune system, or the yield of industrial soya.
posted by KirkJobSluder at 5:51 AM on October 20, 2010

It seems to me that all of these theories break down at the moment of creation.

I've always wondered: Is there some sort of catalog of awful church sign puns, or do the proprietors come up with them independently?
posted by Sys Rq at 8:06 AM on October 20, 2010

Sys Rq: The minister in my family says yes, along with collections of sermon outlines.
posted by KirkJobSluder at 8:18 AM on October 20, 2010