On the Use of "Information Talk" in Biology - Useful or Misleading Analogies

[Likewise very drafty - same caveats apply]

Analogies form a vital tool in exploring a new field of study, since they provide a familiar structure into which we can attempt to arrange the new, unfamiliar phenomena. Faced with a structureless bulk of newly-discovered facts, a researcher may well struggle to find the inherent patterns that allow the data to be carved up into simpler blocks, divided into high-level structures and low-level details, and eventually re-described in a new language that allows for a unified understanding. However, if the investigation is guided by an analogy with some other, better-understood part of the world, the process can be accelerated; if we suspect that system XYZ has some structure analogous to known system ABC, and we find find that part X seems to play a role in XYZ like that of part A in ABC, then what could Y be? Quickly a couple of hypotheses suggest themselves...

Of course, the above is only the loosest of sketches of how analogies can guide research[\footnote{For a more careful look at this subject, see [?],[?],and [Corfield]}]. An important fact to note, and a key fact in all that follows in the "biological information" debate, is that analogies can only give us good hypotheses if they themselves are "good" analogies (in a sense to be developed shortly). Of course, hypotheses that turn out to be disconfirmed by the investigations designed to test them can in the long run be useful, if that disconfirming evidence brings with it greater insights. By "good" hypotheses we mean those that bring deeper understanding, those that are confirmed and form part of a more developed theory of the system in question. Such hypotheses are (more likely to be) produced by correct ideas about the structure of the system.

Thus, to provide helpful guidance in our research, we need analogies which point to genuine correspondences with other systems. Moreover, the correspondences should be deep, rather than superficial, picking out important features of the systems referred to. Analogies are better or worse, (very) roughly speaking, depending upon how many features the related systems share, how significant those features are in their respective systems, and how well the relations between those features match up across the comparison. Put another way, good analogies are those that pick out deep abstract structures that are relevantly instantiated in both systems.[\footnote{Disclaimer again: this is necessarily a very rough sketch, but it will suffice.}]

On the Use of "Information Talk" in Biology - Introduction

[Note: This is clearly very much still a draft version. Very clearly, very drafty]

In biology, especially molecular biology and genetics, much use is made of the analogy between the structures and processes of genetics and those of the world of human artifacts such as computers and radios. Thus it is said that genes "carry information" about proteins and phenotypic traits, DNA is "read", translated, transcribed, and decoded; mutations arise by "copying errors", and are sometimes fixed during the duplication process by "proof-reading" mechanisms; and so on. Note that this is not merely a feature of popularizations of science, translating complex ideas into language more easily understood by the public; on the contrary, the information metaphor was vital to in developing the first understanding of how DNA functions. [\footnote{Quote JMS - no Rosetta stone; studying it just by chemistry would take forever.}]

However, the details of how good this analogy really is have come under considerable scrutiny by philosophers of science, with opinions varying widely as to whether such talk of "biological information carried in genes" is (at one extreme [JMS]) literally correct, and in need of precise fleshing out, or (on the other hand [Griffiths]) both imprecise and misleading.

The general question of what makes an analogy good or bad, useful or misleading,is obviously too large to be tackled within the scope of this essay [\footnote{Find some refs on goodness of analogies in the Phil Lang literature.}]; nor would it be appropriate to attempt to give a full account of how analogies are used and are understood, or how they inform, convey meaning, or provoke new ideas. Fortunately, such a detailed analysis of analogy and metaphor is unneccesary for the present case, and a few general remarks, outlined in the next section and developed elsewhere, will suffice.

Having set out some of this general background, we turn straight to the central disagreement between the two sides. Taking John Maynard Smith and Paul[?] Griffiths as representatives of the opposing camps, we find that the point of opposition is easily described. Maynard Smith thinks that the analogy with information storage and processing is a good one[\footnote{Indeed,good enough that such "information talk" should be taken as literal, not merely metaphorical - but this distinction goes beyond our present scope.}], while Griffiths claims that it is misguided, drawing from a too-limited stock of examples on the biological side, and therefore more misleading than useful.

To explore this disagreement, we will examine some of the ways JMS defines the analogy - the features of genetic systems that are analogous with conventional information processing technology - and defends it against some of the simpler objections of critics. We then consider some further objections by Griffiths, which seem to support a position that he calls "the Parity Principle", namely that 'there is no sense in which genes carry information about phenotypes that is not also exemplified by non-genetic factors'. Thus talk of "genes carrying information" (as distinct from non-genetic factors) implies a priviledged distinction where none exists, and so is misleading.

In the final section of the essay, we consider how a defender of the analogy such as JMS should respond to Griffiths' objections. Can his examples of non-genetic factors which meet JMS's definition of information-cariers be somehow excluded? Can this definition be further refined or re-worked in order to cover all and only the factors JMS requires for the information analogy to hold firm? To answer this question, we need to know what it is about the analogy that makes it worth defending. Why does JMS want to justify talk of "molecular biology as a branch of information theory"? It can only be that he thinks such a view will be useful in developing our understanding of biological systems. Bearing this in mind, we consider how an information analogy (refined to take account of all the relevant data about genetic and non-genetic factors etc.) could be useful in biology, and how it might be possible to establish such an analogy more rigorously.

Some links

Gmail
Bristol Webmail
Google

Atrios
Pharyngula
Billmon
Language Log

Lovely Sophie

--
Because, after all, a sequential collection of written articles doesn't officially count as a blog unless it has a picture of a cat on it. I don't make the rules...