Epigenetics Cannot Fix the “Too-Slow Mutations” Problem

Recently in Aeon magazine there was an article entitled “Unified Theory of Evolution” by biologist Michael Skinner. The article starts out by pointing some problems in Neo-Darwinism, the idea that natural selection and random mutations explain changes in species or the origin of species. The article says this:

One problem with Darwin’s theory is that, while species do evolve more adaptive traits (called phenotypes by biologists), the rate of random DNA sequence mutation turns out to be too slow to explain many of the changes observed...Genetic mutation rates for complex organisms such as humans are dramatically lower than the frequency of change for a host of traits, from adjustments in metabolism to resistance to disease. The rapid emergence of trait variety is difficult to explain just through classic genetics and neo-Darwinian theory.... And the problems with Darwin’s theory extend out of evolutionary science into other areas of biology and biomedicine. For instance, if genetic inheritance determines our traits, then why do identical twins with the same genes generally have different types of diseases? And why do just a low percentage (often less than 1 per cent) of those with many specific diseases share a common genetic mutation? If the rate of mutation is random and steady, then why have many diseases increased more than 10-fold in frequency in only a couple decades? How is it that hundreds of environmental contaminants can alter disease onset, but not DNA sequences? In evolution and biomedicine, the rates of phenotypic trait divergence is far more rapid than the rate of genetic variation and mutation – but why?

As interesting as these examples are, they are merely the tip of the iceberg if you are talking about cases in which biological functionality arises or appears too quickly to be accounted for by assuming random mutations. The main case of such a thing is the Cambrian Explosion, where we see a sudden explosion of fossils in the fossil record about 550 million years ago, with a large fraction of the existing phyla suddenly appearing. Instead of seeing some slow gradual progression in which we very gradually see more complex things appearing over a span of hundreds of millions of years, we see in the fossil record many dramatic new types of animals suddenly appearing.

The other main case of functionality appearing too quickly to be accounted for by random mutations is the relatively sudden appearance of the human intellect. The human population about 1 million years years ago was very small. This article tells us that 1.2 million years ago there were less than 30,000 in the population. The predicted number of mutations is inversely proportional to the population size, which means the smaller the population, the lower the number of mutations in the population. So when you have a very small population size, the predicted mutation rate is very low. But suddenly humanity about 100,000 or 200,000 years ago seems to have got some dramatic increase in brain power and intellectual functionality. Such a thing is hard to plausibly explain by mutations, given the very low number of mutations that should have occurred in such a small population.

But Skinner tries to suggest there is something that might help fix this “too-slow mutations” problem in Neo-Darwinism. The thing he suggests is epigenetics. But this suggestion is mainly misguided. Epigenetics cannot do the job, because it is merely a kind of “thumbs up or thumbs down” type of system relating to existing functionality, not something for originating new functionality.

Skinner defines epigenetics as “the molecular factors that regulate how DNA functions and what genes are turned on or off, independent of the DNA sequence itself.” One of the things he mentions is DNA methylation, “in which molecular components called methyl groups (made of methane) attach to DNA, turning genes on or off, and regulating the level of gene expression.” Gene expression means whether or not a particular gene is used in the body.

The problem, however, with epigenetics is that it does not consist of detailed instructions or even structural information. Epigenetics is basically just a bunch of “on/off” switches relating to information in DNA.

Here is an analogy. Imagine there is a shelf of library books at a public library. A librarian might use colored stickers to encourage readers to read some books, and avoid other books. So she might put a little “green check” sticker on the spines of some books, and a little “red X” sticker on the spines of other books. The “green check” sticker would recommend a particular book, while the “red X” sticker would recommend that you avoid it.

Perhaps such stickers would have a great effect on which books were taken out by library patrons. Such stickers are similar to what is going on with epigenetics. Just as the “red X” sticker would instruct a reader to avoid a particular book, an epigenetic molecule or molecules may act like a flag telling the body not to use a particular gene.

But these little “green check” and “red X” markers would not explain any sudden burst of information that seemed to appear in too-short a time. For example, suppose there was a big earthquake at 10:00 AM, and then at 11:00 AM there appeared a book on the library shelf telling all about this earthquake, describing every detail of it and its effects. We could not at all explain this “information too fast” paradox by giving any type of explanation involving these little “green check” and “red X” stickers.

Similarly, epigenetics may explain why functionality that appeared too fast is or is not used by a species, but does nothing to explain how that functionality appeared too fast. Epigenetics is making some valuable and interesting additions to our biological knowledge, but it does nothing to solve the problem of biological information appearing way too quickly to be accounted for by assuming random mutations.

Another analogy we can use for epigenetics is what programmers call “commenting out code.” Given some software system such as a smartphone app, it is often easy for a programmer to turn off particular features. You can do what programmers call “commenting out” to turn off particular parts of the software. So the following is a quite plausible conversation between a manager and a programmer:

Manager: Wow, the app looks much different now. Some of the buttons that used to be there are no longer there, and two of the tabs have disappeared. How did you do that so quickly?

Programmer: It was easy. I just “commented out” some of the code.

Such “commenting out” of features is similar to gene expression modification produced by epigenetics, in which there's a “let's not use this gene” type of thing going on. But the following is a conversation that would never happen.

Manager: Wow, the app looks much different now. I see there's now some buttons that lead you to new pages the app never had before, which do stuff that the app could never do before. How did you do that so fast?

Programmer: It was easy. I just “commented out” some of the code.

The programmer would be lying if he said this, because you cannot produce new functionality by commenting out code. Similarly, some new biological functionality cannot be explained merely by postulating some epigenetic switch that causes some existing gene not to be expressed. That's like commenting out code, which subtracts functionality rather than adding it.

I can give Skinner credit for raising some interesting questions, but he does little to answer them. The problem remains that biological information has appeared way too rapidly for us to plausibly explain it by random mutations.

For every case in which random mutations produce a beneficial effect, there are many cases in which they produce a harmful effect. Long experiments on exposing fruit flies to high levels of mutation-causing radiation have not produced any new species or viable structural benefits, but produce only harm. We have so far zero cases of species that have been proven to have arisen from random mutations, and we also have zero cases of major biological systems or appendages that have been proven to have arisen from random mutations. So why do our scientists keep telling us that 1001 wonderful biological innovations were produced by random mutations?

It's rather like this. Imagine Rob Jones and his family get wonderful surprise gifts on their doorstep every Christmas, left by an anonymous giver. Now suppose there is someone on their street named Mr. Random. Mr. Random behaves like this: (1) if you invite him into your home, he makes random keystrokes on whatever computer document you were writing; (2) if you eat at his house, he'll give you probably-harmful soup made from random stuff he got from random spots in his house and backyard, including his bathroom and garage; (3) if you knock on his door, and ask Mr. Random for a cup of sugar, he'll give you some random white substance, maybe sugar or maybe plaster powder or rat poison. Now imagine how silly it would be if Rob Jones were to look on those fine Christmas gifts on his doorstep, and say to himself: Let me guess who left these – it must have been Mr. Random!

Randomness, Survival of the Fittest, and the Origin of Biological Complexity

According to Neo-Darwinism, the astonishing biological complexity of the natural world arises from a combination of random mutations and natural selection. The idea is that random mutations produce random changes in organisms, and that natural selection (or survival of the fittest) causes helpful random mutations to proliferate. Neo-Darwinists maintain that this can account for the appearance of useful complex features such as wings, eyes, elephant trunks, giraffe necks, and so forth.

Let us try to imagine a situation that might involve a combination of randomness and survival of the fittest. Imagine you are a football coach at a college or university. Every year many students sign up for the football team, hoping to gain the on-campus prestige enjoyed by college football players. This provides you with a great deal of randomness. Some of these applicants will be strong, and some will be weak. Some of the applicants will be in good physical shape, and some will be in poor physical shape. Some will be fast, and some will be slow.

But where does the “survival of the fittest” come into play? That happens with your discretionary roster cuts. Given this random pool of applicants, you will be able to create a “survival of the fittest” effect by cutting from the football roster any aspiring team members who are too weak or slow or who cannot catch or pass or punt the football.

Now let's suppose the randomness of the applicant pool and the roster cuts are the only factors involved. You simply take your starting pool of aspiring football players, subject them to various physical tests, and cut from the roster a certain number of people who fail to perform well on the physical tests. You do nothing else. So you have randomness, and survival of the fittest. Will this result in a winning football team, one that produces the coordinated functionality needed to win football games?

Of course, it will do no such thing. Producing a winning football team requires both design and a huge amount of coordination. The design comes from designing particular football plays that your team will execute. The coordination comes when particular players are assigned particular positions, and these players repeatedly practice smoothly coordinated football plays. Without this design and coordination, your football team will be a mess. When a play starts, players will just wander about, without anyone knowing whether they are supposed to block, throw the ball, catch the ball, or punt.

So clearly for a football coach, randomness plus survival of the fitness does not yield coordinated functionality. But someone may argue that this analogy is inadequate, because it only involves survival of the fittest and does not involve the idea of differential reproduction. Differential reproduction means that those who are more fit to survive will reproduce more frequently, and those who are less fit to survive will reproduce less frequently. Differential reproduction is basically the same as natural selection, and is actually a better term for such a thing (since nature does not actually choose or select anything, natural selection is not a literally accurate term, even if it may be accurate in some figurative sense, to at least some extent).

So let's imagine a better analogy, one that will include both randomness and differential reproduction. Let us imagine a tall skyscraper filled with monkeys who have been trained to type on laptop computers. Let us imagine that each laptop has an email program or full-screen instant messaging program on its screen. Let us imagine that all day long the monkeys are randomly striking the keys, with such activity being encouraged because the more typing the monkeys do, the more food appears. (This could easily be accomplishing by programming the laptops to count the keystrokes and send a message to some feeding apparatus, whenever a hundred keystrokes were detected.) The random typing of the monkeys gives us all the randomness we could ever ask for. 

But what about the differential reproduction – how can we get that? We can simply imagine that there is a roving editor walking around the skyscraper, examining the laptop screens. Whenever the editor finds a laptop screen that gives some good prose, the editor presses some keystroke on the laptop that sends out this typed output to quite a few other laptops in the building, via email or instant messaging. So if the output on the laptop screen is a decent sentence, maybe such a sentence gets transmitted to 10 other laptops, and put on the screens of these laptops. If the output on the laptop screen is a decent paragraph, maybe such a sentence gets transmitted to 100 other laptops. So this is differential reproduction, a kind of natural selection in which fortunate random output gets reproduced much more frequently.

Will such a system result in the large scale appearance of coordinated complexity? Under such a system would we expect to eventually see lots of laptop screens on which there were good poems, letters, essays, articles, recipes, or intelligent pieces of computer code? Absolutely not. Despite the combination of randomness and differential reproduction, we should not expect any monkey's laptop to have anything but gibberish on it. Even if such a system were kept running for a billion years, we would not expect for coordinated complexity to appear to any large degree.

We can think of two general reasons why this would be true. The first reason is what we may call the discarding of preliminary implementations. If we are to imagine that our roving editor acts like natural selection in the natural world, we must imagine that the editor would reward only work which had received a certain level of functional quality. So if a monkey's laptop contained a sentence such as “I think what we should do about the gun violence problem is nanae anowe anslweonw assfw,” such a sentence (a preliminary implementation of a functional sentence) would not be rewarded with increased reproduction. In order for the random output to be rewarded with increased reproduction, it would have to have a high degree of ordered, coordinated complexity, a level of functionality very, very unlikely to be achieved by chance.

The second reason why this skyscraper of typing monkeys would not result in large-scale coordinated complexity is that in the very rare cases in which random output produced functionality that was rewarded with increased reproduction, the randomness of the typing monkeys would further degrade that functionality over time. So if a monkey ever typed a good sentence, and that sentence got transmitted to the screens of 10 other monkeys, the subsequent random typing of those monkeys would quickly degrade the coordinated complexity, and tend to gradually turn it into uncoordinated gibberish.

Similar reasons cast great doubt on the claim that in the natural world, some combination of random mutations and natural selection can produce coordinated complexity. The discarding of preliminary implementations is something we should expect to be constantly occurring in a natural world ruled by blind chance. For example, if an eyeless organism started to develop a light-sensitive dimple that could be the start of an eye, we would expect natural selection to throw away such a useless feature – unless by some incredibly improbable coincidence it happened to have almost simultaneously arrived by chance with other changes needed for some earliest version of functional vision (changes such as an optic nerve conveniently stretching from such a feature to the brain, and also quite a few changes in the brain needed to process such visual input). To give another example, if an organism by chance developed a wing stump, we would expect natural selection to discard such a feature, as it would provide no immediate benefit. The idea that natural selection will tend to discard anything that isn't useful comes from Darwin himself.

Imagine you are hired to work in a junkyard, and you are told to build useful things from the odds and ends lying around. But there's a problem: you are closely watched by a dimwitted foreman with no foresight or imagination, a guy you nickname “the Brainless Boss.” You start out by finding an axle and two wheels, and fit them together. You think to yourself: this can be the start of a good wagon or cart. But then your foreman sees what you are doing, and throws it away in the trash bin. “Not useful – make something useful,” the foreman says. You then find some wood, and nail together three pieces of wood to make a U-shape. You think to yourself: this can be the start of a bookshelf. But your foreman arrives, and throws what you have made into the trash bin. “Not useful – make something useful,” the foreman says. You realize that your chance of making anything useful are not good, because the “Brainless Boss” will always be discarding your preliminary implementations. So it would be in the natural world ruled only by chance and natural selection. Lacking any foresight or imagination, natural selection would always be acting like this “Brainless Boss,” discarding preliminary implementations that were not yet useful.

Similarly, in the natural world we would expect that random mutations would vastly more often degrade coordinated functionality than to improve it. The more complex and coordinated a piece of functionality is, the more likely it will be that random changes will degrade it rather than improve it. For example, random changes in computer code will be 100 times more likely to harm the code than to help it.

Part of the appeal of Neo-Darwinism is its simplicity. Neo-Darwinism tries to explain the appearance of biological complexity by giving us the real simple equation that randomness plus survival of the fittest eventually yields astonishingly coordinated wonders of biological complexity. The problem is that this equation is not accurate. Randomness plus survival of the fittest does not yield coordinated functionality. The answer to the origin of biological functionality must be something vastly deeper or more complicated than this simplistic little equation.