Sunday 22 July 2018

The non-randomness of mutations

Sometimes people contrast organic and cultural evolution by saying that in cultural evolution, mutations are not random, they can be the result of "intelligent design". IMO, the most correct response to this is is to say: "mutations are not random in organic evolution either". I haven't written about the "random mutations" issue too much because others have been capably handling the issue. One exception was here, where I complained about goalpost moving by proponents of random mutation.

Here I want to concentrate on here is explaining the case for mutations being biased towards being adaptive in the organic realm in simple terms.

The basic idea is that the mutations used in evolutionary models are themselves often the product of selective processes. A simple example of this is mutations in multicellular organisms. The differences in organisms between one generation and the next are what shows up in evolutionary models as having a mutational component. However, the mutations are actually the product of a selective process - namely the repeated production and selection of germ-line cells during development.

What about mutations in single-celled organisms? Even measuring these across a single generation introduces selective bias. Some mutations kill their owners. Those muatations won't be found in immediate descendants because they are selectively filtered out.

Another perspective comes from Perry Marshall, who writes: "Damage is Random. Repair is Not". Gene repair mechanisms are common and selectively bias mutations is adaptive directions.

A possible response to these types of claim is to say that it is mixing up selection and mutation. Mutations are not biased towards being adaptive before selection acts on them. That argument represents a declaration of victory by definitional fiat, but that is not its only problem. In many of the cases described so far, the mutation and selection involved are causally separate and take place at different times. The mutation happens first, and then selection happens later. In such cases, defining mutation as changes in the absence of selection seem viableHowever, there is no rule saying that mutation and selection can be so separated. As an example consider mutations that arise during copying. Imagine that somer sequences are more prone to mutation than others. There, mutation and selection occur at the same time and are entangled. Even stereotypically random sources of mutation - such as cosmic rays are vulnerable to this type of entanglement, for example if DNA sections can be shielded or reinforced by methylation, or in other ways.

Since defining adaptive mutations out of existence fails, what other avenues of retreat are available for the "random mutation" proponents? There are probably some involving redefining "randomness". Proponents already redefined this term to mean "not adaptively biased", maybe they could redefine it further to mean "not adaptive". Most mutations are still deleterious, so that position seems more defensible. It would be an enormous retreat, though.

Another defense against the type of argument given above might consist of insisting on empirical evidence. If biased mutations are so common, then we ought to be able to find them empirically. That's a reasonable request. Looking at the existing experiments that have been performed, I don't see an obvious case that decides the issue. So, this seems like a hole in the case for adaptive bias to me. On the other hand, I think the theoretical case is pretty watertight, and isn't likely to be overturned by experiments.

Another response could be to claim that the effects of adaptive bias are minor, and mutaions being random are a good approximation, and one that simplifies things considerably. I could cope with that. I can think of some circumstance where random mutations are not a very good approximation - such as genetic engineering - but they are also currently fairly rare. Randomish mutations might well be the rule inside cells for moost of evolutionary history.

I've been living with an understanding of adaptive mutations for over a decade, and my testimony is that Darwinism survives reasonably well without them. Darwin didn't believe that mutations were random - he know that variation existed, but "mutation" wasn't really a significant concept back then. The concept of "equivocal mutations" came in with the neo-Darwinian synthesis. Random mutations are still available as a theoretical approximation, should they be required. Mutations necessarily being non-adaptive is a dead concept in my book. Its defenders come across as being defenders of out-dated dogma.

The whole idea of mutations being different in cultural evoilution and organic evolution still has merit. Brains are a different environment from cells, and different types of change can happen in them. There are still bounds and limits to what can hapopen inside a brain, though. It isn't the case that anything goes, and so cultural evoiolution becomes an unfalsifiable theory that is compatible with any observations on the grounds that it could be a macromutation. However, the idea that mutations are directed in cultural evolution, but not in the organic realm is mistaken and has to go.

No comments:

Post a Comment