John Wilikins has a post on my last couple of entries:
In a couple of posts, Scibling Alex Palazzo of The Daily Transcript has given two quite distinct views of what biology is about: information, and mechanism. In the first he argues that what is needed to build organisms is information, and in the second that biology is about machines, things that do work. I want to say that he is wrong about the first and right about the second, and moreover that they are contradictory ways of looking at the living world.
So why does Wilkins have a problem with a discussion equating information with life?
Let me reprise the argument or claim: information has no causal power. Instead, things that have properties have causal structure and information is a property of our semantics, not the world.
OK call me a postmodernist, but I'm a firm believer that all of our theories and models are simply that, MODELS. Does Information have any causal structure? Do we want to be so picky? Isn't "causation" on some level just semantics? After all how do we deffine causation? But lets step back from the edge. Information and causation are short hand for two concepts. Are these ideas useful? Indeed. Many of our concepts and theories have utility in that they help us to understand and predict how nature works.
Here is what I wrote to Wilkins:
Strictly speaking none of our models of how natural systems operate are the "Truth" but simply tools to help us understand and predict how those systems behave. I partially agree with you in that life is not just information, but I also believe that the ideas of "information" and "information processing" are useful concepts in thinking about how living systems operate. When Craig Venter transplanted the genome of one bacterium into another bacterium, the chimeric organism eventually resembled that which is "specified" by the DNA component and not the cytoplasm - why? Sure you can reduce the system down to the physical laws, but another useful model is thinking about what the DNA itself specifies - a particular digital code of A, T, C and G. This genetic program specifies the production of certain RNAs and proteins. Now of course this can be equally explained through the lens of physics and chemistry but explaining it through the shorthand of information can be much more useful. Does a geneticist need to think about every equilibrium or energy expenditure to understand how traits are passed down? or why Venter's chimera turned out the way it did?
- Log in to post comments
I think you've failed to appreciate something about the Venter experiment. Look, Venter takes a genome from one organism and puts it into another; the phenotype changes, neato. But that on its own does not argue for the primacy of genetic material. Consider: what was the same between the two organisms? Well, they both used DNA as genetic material, that's true. But what was also true was that the mechanisms used to interact with and interpret the DNA were largely the same. No matter how much information is contained in DNA, it is meaningless on its own. It is the molecular machines of the cell that give meaning to that information, without which it might as well be noise. If Venter had transplanted that genome into a bacterium which had native machines that recognized an alternative triplet code or used completely different promoter sequences then the organism would have simply died. It is customary to think of syntax and meaning as properties of the DNA in which they are expressed; in reality, these arise from the machines for which the DNA codes.
I completely agree - however those machines were themselves specified by the information of the preceding genome. In biology there is this chicken and egg problem, and I think that it's too simple to say "chicken" or "egg". Within a certain context each concept can be used to give insight into how the system behaves. We can abstract certain concepts. In the post that Wilkins criticized I basically state that biological systems contain information in their DNA but to understand the phenotype we must reconstruct and understand the network of machines that are specified by that code. That is in effect what you are saying, but backwards.
I'm not convinced that there is necessarily a chicken-and-egg problem here, unless you are in the habit of viewing the machines and the information as fundamentally separate. However, we can point to things like ribozymes that essentially are (or at least can be) self-specifying.
I feel this hints at an attractive approach for talking about biology generally. We can view organisms essentially as self-specifying machine networks, the function of which is to make more self-specifying machine networks. This has the advantage of not attributing primacy to either the information (which is meaningless without the machines) or the machines (which cannot be produced without the information). It also has the attractive property of (correctly) placing the specification within the network (i.e. the interactions between protein and nucleic acid components) rather than in the DNA (which on its own has no more than the potential to specify the network).
In this view the cross-specification achieved in the Venter experiment is the result of fortuitous similarities between network elements rather than the primacy of genetic information. A simplistic view of DNA (or machines) as central might make very incorrect predictions about potential experiments in which genomes were swapped between, say, eukaryotes and prokaryotes, or either of these and archaea.
I think we are close to being on the same page here, in that we are both very interested in the network of interactions. I just don't see any need or benefit of saying one aspect is more fundamental than another.
I think that we are exactly on the same page. Insight will be acheived from analyzing the network of interactions - I originally started these posts to critique the DNA-centric view of biology which I think misses the point. Unfortunately this picture of molecular biology is what is sold to the public. The real work and the real insight is not coming out of genome gazing (although a lot of good stuff is comming out genomic analysis) but from studies on the network of "machines".