Joshua Swamidass, in dispute with staunch ID apologist “Deliberateresult” (Joe Palcsak) over at BioLogos, denies (like a surprising number of TEs) that DNA is anything more than “analogically” a code, or a language, in the wider context of his being uncomfortable with considering people as biological machines:
Honestly it sounds like you are seeing people as machines, even in this quote. Of course we make analogies between living systems and machines, language, technology, etc. That is how humans work. We reason about often this way. You, however, seem to think the there is not analogy, that this is actually describing the reality.
You say, DNA is actually a language, no metaphor, right? The only differences between humans and machines are because we are much more brilliantly engineered. Right?
I’m fine with using the metaphors, but you seem to be taking them very literally. I am not comfortable with that. It leads to very poor thinking, and in the case of calling humans and example of technology, it makes me ethically and theologically squeemish [sic].
I’ll come to “machines” shortly, but my major interest here is to examine the “DNA is not really a code” idea. It seems to me as if amongst TEs this idea has followed a growth-path something like this. Gamow’s RNA tie club in the 1950s worked out how DNA operates, and it was Gamow’s use of maths in a “cryptographic” manner that predicted a triplet code for the 20 amino acids, for the deciphering of which Nirenberg, Khorana and Holley were awarded the Nobel Prize in 1968 “for their interpretation of the genetic code and its function in protein synthesis”. Both the Nobel Committee and Khorana in his Nobel lecture were happy to leave scare-quotes off the word “code.”
Even in the the early days, Hubert Yockey showed the application of the new information theory to DNA, and has always insisted that it is not analogically, but literally, a code. He certainly denied that it is identical to “language”, though its Shannon information content (see below) is mathematically equivalent.
The similarity between the code and computer algorithms has not only been used to model evolution, but has even been analysed in research like this fascinating paper, which compares the E. coli genome to the Linux operating system, saying in its discussion:
We have presented a comparative analysis between the transcriptional regulatory network of E. coli and the call graph of the Linux operating system and explored their similarities and differences in hierarchical structure, modularity of organization, and persistence of nodes. A summary of the comparison can be found in Table 2. The two networks are shaped by different underlying design principles, which are deeply connected to the interplay between the systems and their environments.
There is no allusion to “analogy” in the paper – if there were, it would hardly be any more useful a comparison than a scientific paper on some evolutionary psychology story as compared to The Elephant’s Child… though in some other context that comparison might be worthwhile.
Francis Collins called DNA “the language of God” in establishing BioLogos, and whether or not he meant it analogically (of which there is no hint in the book), that kind of description would trigger alarm bells in the predominantly atheist evolutionary science guild. I suggest that, ignoring half a century of information theory insights, they began to push the historically dubious message that everybody has always really meant the code as a mere analogy. That has become the consensus (ie “good”) science. TEs naturally want to follow good science (witness the general antagonism to Third Way approaches to evolution), so they endorse this conclusion.
But is the reasoning actually valid? And more fundamentally, how can a science of efficient material causes make any legitimate judgement on a question of a code, which is an inherently teleological and informational concept, strongly implying a purposeful codemaker? One must surely employ the methods of fields in which more than material efficient casuation is methodologically permitted. Philosophy and theology are such fields, but also information technology and linguistics: biology cannot claim to trump these using methodological naturalism, for it is to argue in a circle.
But let us, from outside that methodology, look at the ways in which the DNA code is directly comparable to language.
Firstly, the methods of Shannon’s information theory deal very successfully with the information content of DNA, in terms of things like noise, error correction and so on. Yockey in his book deals with this in great detail. On the basis that what walks like a duck and quacks like a duck is a duck, DNA contains Shannon information. Or rather, as far as it is “code-like”, it is Shannon information.
Likewise, like human languages of all sorts (including codes and computer algorithms), DNA exhibits Kolmogorov complexity, which sharply distinguishes it from the ordered complexity of non-living chemical or physical systems. The classic example of the latter is the snowflake, hugely variable but completely prescribable from simple mathematical principles and a few variable parameters. And yet astonishingly one still hears TEs (and skeptics – why do they agree on so much, I wonder?) using the snowflake as an analogy for living things only apparently being designed. Yarus’s work suggests primaeval affinities between some amino acids and RNA. To the extent that this is so, it reflects the ordered complexity of the snowflake, not the organised complexity of life and language.
Thirdly, as Gamow predicted, DNA contains symbolism, in which something (a DNA triplet) stands for another entity not in any way directly related (an amino acid). Once again, the work of Yarus barely addresses this fundamental feature of today’s DNA code.
Now fourthly, language (as I have already said) is innately teleological, and science cannot address meaning as such, either to discern it or dismiss it. To quote Yockey (paradoxically attacking the use of his work by Intelligent Design people):
Information theory treats information and meaning as entirely separate (cite Shannon). Information, as defined in information theory, is measurable, and therefore is a suitable subject for science and mathematics (cite Socrates). However, meaning is not measurable and thus falls in the realms of faith, belief, religion and philosophy.
In other words, he implies methodological naturalism can have no opinion about meaning. He omits, of course, that far-from-religious empirical fields like linguistics cannot function without addressing meaning, and neither, in practice, can those who need to apply information theory in practice.
But “function” is a simpler use of language than human discourse: to say, “Quick march!” or “First pluck your chicken” does not exhaust the capabilities of human language, but does employ its main principles. So does the concept of computer language, in which symbolism is made mathematical rather than alphabetic and function is achieved algorithmically. And biologists would have very little to talk about if they did not speak teleologically of DNA in relation to function: calling such talk mere analogy is specious when there is simply no other way to discuss the subject.
Inasmuch, then, as science can compare the DNA code to language, the evidence is that it is more than an analogy. On the other hand the differences one can state between language and DNA seem to have more to do with detail than disanalogy. So, to begin with, the information in DNA is always transmitted chemically, whereas in human language it is neurochemical and aural, or visual, or many other modalities. In computer code it is always electronic, but needn’t be – a mechanical abacus or Babbage machine will do the same job, if inefficiently, because information is immaterial.
Likewise one could assemble a protein synthetically from a written script derived from the gene, and it would function in the cell, or synthesize bastardised tRNAs that would happily form proteins with all the wrong amino acids, that failed to function. Chemistry cannot be mistranslated – only codes.
Secondly, speech always in practice carries sophisticated levels of meaning (including metaphors and analogies!) and ultimately entire experiences of the world. Understood as linear code, DNA does not. And yet we should not assume too much from our limited knowledge of DNA, which may be akin to the first understanding of a foreign language, in which we may soon find how “!ktoh” means “dog”, but will need many years of study to see how the dog functions as a totem animal. Only recently we have found the importance of DNA’s 3-dimensional configuration, and its multi-level control functions. Who knows what “meaning” may not be hidden in what remains undiscovered? The point is that what difference we find has not to do with analogy, but divergence of purpose (and weakness of scientific methodology): where they can be compared, language and DNA match at essential points.
To explain this further, consider Joshua Swamidass’s objection to the machine metaphor for the body, an objection which is valid at some levels, but not at others – and notably not at the level of science. I remember being shocked in 1st year school physics when Mr Spencer described a simple lever as a machine. To the eleven-year old me a “machine” had a hundred moving parts and a spring, but physics takes a simpler view.
Now, it is simply a fact (on which I’d stake my medical reputation, if I still had one) that the musculo-skeletal system consists of (virtually) nothing but a system of levers. Palaeontologists can determine the shape of dinosaurs from calculating the muscle-bulk necessary to work a particular limb. These levers are not analogies – they are true machines, in the same sense that physics defines them.
Of course, as any Aristotelian will quickly tell you, they are not artifacts, and in formal terms comprise integral parts of a unified “substance” that is far more than a machine, in all kinds of ways. Nevertheless, when you design an artifical limb, you start by modelling the (natural) machinery it has to replace. And when Joshua speaks as a scientist, rather than an Aristotelian metaphysician, his objection lacks weight. Theologically, many people are also “squeamish” if they think of human beings, created in God’s image, as “animals” – and yet Scripture itself is clear that though we are more than animals, our “animality” is real, not analogy.
Therefore one needs to demonstrate, not simply assert, the differences between DNA and a true code, and saying (as one BioLogos poster did last week) that DNA is not a code because it is “all chemistry” is as crass as saying that language is “nothing but neurochemistry”. Why do so many people find it so hard to see why the discovery of a medium does not preclude a message?
Another aspect of this has arisen only very recently, in a Scientific American article on human language. This describes the increasing abandonment of Noam Chomsky’s longstanding linguistic paradigm (not to exaggerate its influence) that the human brain has a hard-wired “grammar” which underlies the structure of every human language, making apparently diverse tongues actually variations on a rather mechanistic template. The article says:
Instead the new research shows that young children use various types of thinking that may not be specific to language at all—such as the ability to classify the world into categories (people or objects, for instance) and to understand the relations among things. These capabilities, coupled with a unique human ability to grasp what others intend to communicate, allow language to happen.
In the new usage-based approach (which includes ideas from functional linguistics, cognitive linguistics and construction grammar), children are not born with a universal, dedicated tool for learning grammar. Instead they inherit the mental equivalent of a Swiss Army knife: a set of general-purpose tools—such as categorization, the reading of communicative intentions, and analogy making, with which children build grammatical categories and rules from the language they hear around them.
What this appears to mean, in the comparison of “language” with “DNA code”, is that if language is not fixed by a mechanism, but only by its function, it becomes a lot less secure to speak of “analogy” between human language and DNA. One might as well speak of Xhosa or some other rare tribal tongue as only analogically comparable to English, being constructed entirely differently and, perhaps, even using different pathways of the brain. But significant elements of what is common to such disparate human languages is, as far as one can tell, common also to the genetic code.
That is emphasised if one compares DNA to the disanalogy of animal communication. When I did social psychology at Cambridge, the first book I read was by Margaret Mead, though I can’t remember the title. In it, she pointed out the difference between the limited vocalised signals of an animal, which communicate by “calling out” a fixed direct response (such as the alarm call that signifies “run” and uniformly evokes running), and human language, which combines symbolic elements to communicate any number of complex messages.
On this comparison, clearly the chemical affinities noted by Yarus resemble far more the animal than the human model. The fact of such communication in animals has no direct analogy with human language. In fact, it is only when language breaks down under strong stimuli that laughter, screams or wailing occur at the expense of language. It would seem to me that chemical affinities similarly work against the development of a flexible code, rather than forming a foundation for it.
One more thing, regarding the approach of Christians in this matter. Remember that science, operating under methodoloical naturalism, has no authority to comment either way on “meaning”, or lack of it, in DNA even at the relatively simple level of teleological function. But as Christians, if we believe that Christ the Logos is the source of the creation of all things, including living entities, what do we mean by excluding the existence of a true information function within the mechanisms we know to be involved in the evolution and development of life? What basis do we have for such an assertion?
New Scientist article speaks of “a unique human ability to grasp what others intend to communicate”. That certainly applies across the 6,000 languages of mankind, and it extends to a large extent to our relationship with animals too, which is why we can tame or train them (not to mention the more subtle communications of crops or even cakes, according to one TV baker recently, who said her success comes from “listening to her cakes”!). Is it really that outlandish to suppose that God should want us to understand his communication in nature, and might equip us, since we are made in his image, to do so more or less reliably?