Tuesday, July 17, 2007

Is Evolution the Drunken Man as Behe claims?

I was listening to the latest episode of 'The Things That Matter Most' with Dr. Michael Behe, the man that put forth the idea of Irreducible Complexity (IC), as guest. Wow. Some of the worst arguments for Intelligent Design (ID) I've heard to date. It's no wonder that his colleagues at Lehigh University have disowned him.

I've blogged on the many problems of the IC concept before and, even though it is impossible to believe that Behe has never heard of the complete and utter refutations by others with far more expertise than I (little-known people such as Richard Dawkins, Ken Miller, PZ Myers, Jerry Coyne,....), he continues to espouse this idea. Even his mousetrap has been found to be reducible. I'm sorry, Dr. Behe, but this makes you a dishonest broker. 'Irreducible complexity is this fancy phrase that all it means is that you've got a machine like a mousetrap... that has a number of different parts all of which are required [for it] to work.' The mousetrap is a red herring because I have yet to see a mousetrap replicate. Since the primary mechanism by which Natural Selection is removed, the analogy is false one.

Behe arbitrarily declares that structures serving other functions can't be co-opted to be brought together for a new function. 'Even if you are hoping to use thos
e things [parts] for something else, like a doorstop, a doorstop has nothing to do with catching mice.' What's he done here? He's concocted a straw man (hardly original) of Evo. He presumes that catching mice (function) is something not only desirable, but a necessary end point. But function is something we ascribe to a system - the genes themselves do not care what or even whether they provide a function, only that they impart survival value.

So, what we should always be looking for is whether a genetic change imparts a survival advantage, not whether the change maintains function. The latter is simply irrelevant. DNA does not care about any one of its gene's functions, or whether there even is a function for it, only the effect of the gene on survival probability. To illustrate this using the mousetrap example, let's say that a cell contains the parts that are required for the mousetrap and the catching a mouse results in an increase in the chances of survival. If bringing some of the parts together confers some survival advantage (or is benign in terms of survival) then the changes will tend to be passed on to the next generation. In other words, bringing two parts together can increase the chances of survival more than the parts being separate. Then, in a future generation, another part gets added, conferring even more survival value, and so on. Thus, structures better and better suited to survival are produced. Here's the kicker: the function of the early versions of the mousetrap need not even have involved catching mice at all so long as some net survival advantage was gained by combining parts! When looked at in these terms, there is no basis on which to deny the base of the mousetrap to have originally served as a doorstop!

At some point, the functions of the structure changed from whatever they were to by catching mice. This is immediately apparent to anyone familiar with Natural Selection and Behe's poster child for ID, the bacterium flagellum. When Behe artificially rules out functional change the flagellum seems 'irreducible', but he provides no good reason that we should accept omitting such evolutionary paths to this structure. He is simply invoking an artificial restriction that he claims Evolution can not violate (yet it clearly can and does) simply because it makes IC look good and for no other reason that anyone can fathom.

Does the Type III secretory system, which comprises part of the flagellum motor, lose its function when combined with another part on its way to powering the flagellum? Quite probably. Does this matter? Not at all if a net advantage is gained.

For all his philosophizing he has yet to perform a single experiment to confirm irreducible complexity. I can think of a number of experiments off the top of my head to test the validity of IC. One is simply looking at the parts of a complex structure such as the bacterium flagellum and seeing if the parts have analogous structures in related organisms. That the individual proteins of the flagellum are completely conserved and have wholly different functions in related bacteria does not bode well for IC at all and clearly points to an evolutionary process. Why would you, if you wanted to design a mousetrap as opposed to evolve it, use a blender? I think if you wanted to make a mousetrap you would do what is already done - create the individual parts specific to the task.

The blood clotting cascade, another of Behe's examples, can also be tested as to whether it fits the IC model. In this case, we do not have a single structure composed of several proteins, but a series of biochemical reactions. Ideally, one would like to be able to remove one or more enzymes catalyzing reactions in the cascade. Can this be done? Absolutely. Genetically engineering mice to inactiveate genes is routine these days. Dr. Behe has never even suggested such experiments, let alone performed them. Fortunately, less lazy researchers have. Knocking out one gene or even several encoding enzymes in the cascade does not stop the blood of mice from clotting. That knockout mice blood doesn't clot as well is irrelevant, since Behe predicted there should be no clotting ability at all. Of course, all that has to be done is look at the blood clotting cascade in whales and dolphins to see that they have what Behe would have considered an incomplete cascade already.

Indeed, all examples held up as for IC are similarly fatally flawed. Behe claims that 'blind searches do not lead to complex systems.' Really? Again, this is a blanket assumption for which he has no evidential basis for making, and Dr. Behe has never made an attempt at experimentally verifying his ideas. For purely random processes this statement is true, but Natural Selection is anything but random. For examples of such complexity arising from blind searches just google 'genetic algorithms'. A word to the wise: If you are going to put forward a radical idea such as this, you'd better have something to back it up with and that something had better be more than half-baked philosophizing.

Interestingly, Behe says 'I think the most compelling evidence for common decent is when you see features in organisms that seem not to have any particular function but look like genetic accidents and if that's the case then a kind of a parsimonious or a reasonable explanation is that these organisms are both descended from some previous organism. This genetic accident happened in the previous organism and both lines descending from that earlier one inherited the change.'

Wait a minute, Dr. Behe - You accept common decent, but deny macroevolution? What the....? Is he trying to tell us that bacteria containing the flagellum, which are closely related to bacteria having the Type III secretory system from which it evolved, both have a common ancestor, yet the flagellum needed to be added to the 'design'? For this to work, the molecular genetic evidence pointing to close relationship to these two types of bacteria must be wrong. Not bloody likely. He's trying to have his cake and eat it, too, by picking data supporting his position and ignoring all other data.

The question must be asked: why does Dr. Behe accept some parts of Natural Selection while discarding others? The answer is simple: he has a major problem explaining the existence of broken genes. I once heard PZ Myers say what he'd like most explained by supporters of ID, why a 'designer' would equip humans and other primates with a broken enzyme catalyzing ascorbic acid synthesis. This is exactly why Behe brings a limited version of Natural Selection into his world. He realizes that ID, as originally formulated, has no answer to this. But if you accept one part you must accept, if you are to remain intellectually honest, Natural Selection in its entirety. Note to Dr. Behe: the Theory of Natural Selection is irreducibly complex.

I've already blogged about the example contained in his new book, Edge of Evolution, where Behe claims that the development of malarial resistance to
chloroquine through evolutionary processes is improbable. Behe again arbitrarily says that two mutations are required for any level of chloroquine resistance without explaining why no resistance is conferred by a single mutation in the right place except that it suits his argument. 'But suppose in order to be effective it needed not just one but it needed two and with just one that it didn't help or might even hurt the organism. So it had to get a change not only in the left side of its DNA but somewhere in the middle of its DNA as well. Then it turns out that that is a whole lot more difficult than just getting one [mutation].' Sorry, but as many have pointed out, mutation is not the bottleneck in Evolution. The arbritrary assumption that multiple mutations must take place at the same time is simply baseless. And what exactly is he saying? That a designer is making malarial parasites more resistant to human medical interventions in order to make them better killing machines? I don't think he's thought this through...

But my biggest objection is his analogy of the drunken man to Evolution, which falls flat on its face. (I know, I know - bad pun. Bad!) Behe describes his analogy thus: 'Suppose there were this drunk, this dizzy fellow, who had a blindfold on, and you want to get from some place in the city to the top of some tower in the downtown and suppose he's in the suburb. Well if he had to follow a rule that whenever possible you feel the ground going up you have to take a step up and you can never go back down because going back down means becoming less fit in a Darwinian sense, then if this fellow walks along he will climb up onto roofs of cars or onto porches of houses. Once he's up there he'll get stuck and he's not going to find his way to the tower downtown.' Here he's plagiarizing Dawkins' Climbing Mount Improbable example, the peaks of Dawkins' version correspond to the car roofs and porches in Behe's. Such a view gives the false impression that point B is some kind of objective. In Natural Selection there is no desired end point, just what is working at this point.

Behe twists Dawkins' version to make it appear as if the tower (the highest mountain peak in Dawkins' version) can't be reached by random processes without outside help because it is only one person. The likelihood that one person will reach the tower through a random walk with the rule that you can't take downward steps is indeed extremely low.
But Evolution is an ensemble of organisms each taking paths going in any direction without any thought as to where they will end up, not a single one trying to get from A to B. To make this analogy work better, you need many, many drunks (an experiment which seems to be carried out spontaneously during Stampede Week every year). As in Behe's version, you will have many drunks finding car roofs or porches, but there will be those that do indeed get to the tower. In fact, with enough drunks all apexes in the search space will be occupied. His is a horribly inaccurate analogy and I've done my best here to try to correct it.

Are there limits to what can and can not be done by Natural Selection? Absolutely, but it's quite clear that Dr. Behe has not found the Edge of Evolution yet at all.


J. K. Jones said...

Impartation of “survival value” is a function. “Survival probability” is an outcome of a process. As you say, “… natural selection is not random.” And “the Theory of Natural Selection is irreducibly complex.” Non-random processes require design and intention to result in non-random (organized) outputs. Any manufacturing engineer can give you thousands of examples of that in factories all around the country.

Just stating that “In Natural Selection there is no desired end point, just what is working at this point.” doesn’t make it so. Working for what? What is the outcome? Better survivability?

We have never even begin to discuss the causal mechanism for a mutation yet, either. What causes the mutation? How is the DNA changed?

I can be pesky for a “weakest link.”

Shamelessly Atheist said...

I give you one last chance, but if you repeat yourself or force me to repeat myself, that's it. I simply don't have that much patience and as others have told me, I've already shown you more patience than you deserve.

"Impartation of “survival value” is a function." Survival could be looked at as a function, but needless of any purpose. "Non-random processes require design and intention to result in non-random (organized) outputs." Not true. I'll blog at some point examples of non-random processes in Nature.

"Any manufacturing engineer can give you thousands of examples of that in factories all around the country." And not one of those examples are applicable to self-replicating systems. That's where the design analogy breaks down completely. Any machine which cannot self-replicate must be designed because there simply is no other mechanism available. But with self-replication, evolutionary mechanisms become possible. Making comparisons to man-made objects is just a bad analogy.

"In Natural Selection there is no desired end point, just what is working at this point.” doesn’t make it so." No, it doesn't. Nature makes it so. Without adaptation extinction would eliminate all life on Earth with changing environments. "Working for what? What is the outcome? Better survivability?" Clearly, yes.

But better survivability needs no intelligent direction. If the organism is better able to survive in its environment as a result of a genetic change it tends to be passsed on to the next generation. It's a feedback mechanism.

The most common mechanism for mutation is not, as one gets the impression from 50s horror flicks, radiation damage to DNA. This does happen but is extremely rare.

Far more common are point mutations arising from simple uncorrected copying errors. The fidelity of DNA replication is astounding, about 1 error in 10^8 bases. Of these, about 1% go uncorrected. These numbers are hard numbers and not at all reasonably disputable anymore.

Even with this high fidelity, every single one of us has between 100 and 500 genes that we do not share with either of our parents. This astonishingly high number is a function of the length of our genome and the number of divisions sperm and egg cells undergo. The vast majority end up as point mutations in so-called junk DNA or cause a protein to have an amino acid substitution in a region of a protein which has not effect.

One example of this kind of benign mutation turned into a beneficial change in the 80s with the onset of AIDS. There is a small population with a deletion mutation (CCR5-Δ32) that are immune to certain strains of the disease. This mutation was completely unknown beforehand.