Mutations and Information
From SkepticWiki
Contents |
[edit] Introduction
It is sometime claimed by creationists that
(a) Mutations always decrease, or cannot increase, information.
(b) Evolution requires an increase of information.
Hence, they conclude, evolution is impossible. The idea, like so many creationist arguments, seems to have originated with the astronomer and panspermist Fred Hoyle.
There are a number of fairly obvious problems with this.
[edit] Problem 1: Vacuity of the claim
The most obvious problem with this argument is that the creationists who use it never say how information is to be measured in this context.
If they mean any of the usual measures of information, then they are obviously wrong. Any mutation that lengthens the DNA will increase the number of "bits" in it; and nearly any mutation that lengthens the DNA will also increase the Kolmogorov complexity; and so forth. Indeed, as we shall show below, the claim that mutations cannot increase information will be false for any sensible measure of information.
It is therefore quite a clever tactic for creationists never to say how they are quantifying information. But the fact that they do not do so makes their claim utterly vacuous. They might as well be saying: "Mutations can't increase shrdlu, and evolution requires an increase in shrdlu". Unless they are prepared to go on and explain how to measure the quantity of "shrdlu" in a sequence of DNA, such claims are not true or even meaningful.
In the rest of this article, we shall use the term information™ to refer to whatever mysterious undefined quantity creationists are talking about, as distinct from the well-defined concepts used by information theorists.
[edit] Problem 2: A mathematical argument
However, even without creationists providing their top secret definition of information™, we can see that there is something wrong with their claim. Because it is clear that any sensible definition of the information™ in a DNA sequence must obey two conditions:
- Condition 1: The "null string" consisting of no DNA whatsoever contains no information™.
- Condition 2: Two identical gene sequences must contain the same amount of information™.
Now, consider the fact that every mutation has its opposite. An insertion can be undone by a deletion at the same location; a single nucleotide substitution can be undone by another single nucleotide substitution at the same location; the splitting of a chromosome can be undone by the fusion of the two parts, and so forth.
If one mutation decreases information™, then the opposite mutation must increase it; otherwise, it would be possible for a DNA sequence to undergo two successive opposite mutations, leaving it back exactly how it started, and to have undergone a net decrease of information™ compared to the original, with which it is identical, in violation of condition 2. It follows that if some mutations decrease information™, then others must increase it.
But suppose (you might suggest) suppose that all mutations neither decrease nor increase information™: suppose that they all conserve information™. This would get round the argument in the previous paragraph, but it has problems of its own. For if it was true that no mutation decreased information™, then this would be true, in particular, of all deletions. This would mean that we could delete as much of a DNA sequence as we like without destroying any information™. In which case we could delete the whole thing without destroying any information™! But in that case, the original DNA sequence we started with contains the same amount of information™ as the "null string", which, by condition 1, contains no information™.
So if we have a definition of information™ under which information™ is conserved by all mutations, then would follow from this definition that no DNA sequence would contain any information™; and so the process of evolution would not require any increase of information™, and evolutionary biologists would be under no onus to explain how the information™ got into the DNA --- because there wouldn't be any information™ in the DNA.
Summary of the theorem: For any definition of information™, such that the null string contains no information™ and such that two identical DNA sequences contain the same amount of information™, then either there is no information™ in any DNA sequence, or there are mutations that increase information™.
[edit] Problem 3: Mutations per se place no constraints on evolution
It is clear that a sufficient number of changes to any DNA sequence will change it into any other DNA sequence. To change, for example AAAGTCGTA into CCATAGTGAA requires two insertions, one deletion and two nucleotide substitutions --- or there are many other (indeed infinitely many other) ways to get from one to the other. Evidently, then, a sufficient number of changes to the genes of a fish, for example, will produce the genes of a frog; a sufficient number of mutations will get from the genes of a dinosaur to the genes of a modern bird; just as (by analogy) a sufficient number of changes to the text of Oliver Twist would turn it into Moby Dick.
Whether or not creationists would admit that getting from a fish to a frog would constitute an increase in information™ we cannot say, and, as they cannot define information™, nor can they; but it is clearly what biologists mean by evolution, and the nature of mutations per se does not prevent you from getting from any organism to any other organism.
This fact is related to the fact mentioned in the previous section, that every mutation has its opposite. Imagine if this was not true: if, for example, mutations always decreased the length of a DNA sequence and never increased it, or if substitution could turn purines into pyrimidines but not vice versa. In that case, we could indeed look at two DNA sequences and say, in the words of the old joke, that "you can't get there from here". But there are no such constraints. Mutations per se can achieve any evolutionary change. (Consult our main article on Mutation for details of the various types of mutations.)
So if we want to know whether amphibians can evolve from fish, and whether this actually happened, we need to ask not whether there is a set of mutations that gets from one to the other (for the answer to any such question is always "yes") but rather whether there is a set of mutations getting us from one to the other which, occuring piecemeal, cumulatively, passes through a set of intermediate forms each of which is favored or at least tolerated by Natural Selection. If so, then it is possible; to see if it has actually happened we should also wish to know if we can find some of these intermediate forms in the fossil record (to which the answer is "yes" --- see our article on Intermediate Forms Between Classes) and whether our reconstruction of events is borne out by molecular phylogeny.
In short, anything at all can evolve into anything else by mutation per se; the big question about any such transition is whether it can be brought about by mutation constrained by natural selection, as the theory of evolution claims, and whether the evidence is in accordance with the claim that this transition actually took place. The creationist mantra about mutations and information™ is an attempt to leapfrog these real and relevant questions, which require thought and study rather than the repetition of meaningless undefined words.
