The three papers of this session share a common disagreement: that genetic, or biological, information is substantively encoded by the DNA alone, as John Maynard Smith has recently defended in response to earlier criticisms of the informational construal of genetics and developmental biology. There are two well-worn routes to conceiving of genetic information: that the use of information terminology is mostly metaphorical and heuristic, and that the theoretical use of information meaningfully references biological phenomena. Michelle Little discusses the desirability of taking the first route. Sahotra Sarkar and Stephen Downes each argue for the second, but in a novel way: they present alternatives to Maynard Smith's conception of where genetic information exists.

Organized by: Michelle Little

Michelle Little, Northwestern University
"Genetic Semantics via Function"

I examine the linkage Maynard Smith has recently defended between genetic, or biological, meaning and naturally-selected function. After discussing general benefits of using informational terminology in genetic discourse, I turn to Maynard Smith's own views on why natural selection confers intentionality to genes. Maynard Smith defends biological intentionality, in part, on the basis of an analogy between "blind" natural selection and similarly blind optimization programs. Since the product of an optimization program can be indistinguishable from what a conscious computer programmer could develop we are entitled to say - with homage to Turing - that the optimization program is as intentional as the conscious computer programmer and that both their products subsequently are meaningful. I question the general suitability of Maynard Smith's analogy beyond its heuristic features, and specifically criticize the capacity of natural selection to capture what - scientifically and intuitively - we think is meaningful in the genome.

Sahotra Sarkar, University of Texas
"The Origins of Biological Information"

Sarkar (1996a, b) argued that the customary use of "information" in biology, particularly genetics is metaphorical rather than theoretical. Similar claims have since been defended by Griffiths and Knight (1998) and others. If this argument is accepted, two alternatives present themselves: (i) one could become a complete physicalist and eschew informational talk altogether; or (ii) one can attempt a new explication of information that will give that term some theoretical underpinning. This paper develops the second alternative. Starting with an account of semantic information outlined by Sarkar (2000), which uses three desiderata: (i) differential specificity; (ii) semioticity; and (iii) reverse differential specificity (see also Sterelny and Griffiths [2000]), an attempt is made to show how biological information can have emerged through evolution. This account uses the signalling games introduced by Lewis (1969) to explain the origin of conventions but has the molecular systems in which information emerges evolve according to replicator dynamics. This procedure, previously explored by Skyrms (2000) shows how semoticity can evolve along with specificity. One then has a theoretically adequate concept of biological information. However, contrary to the claims made by Maynard Smith (2000) and many others, biological information does not reside in the nucleic acid sequences of cells alone. Rather, these sequences are mere parts of a much more complicated informational matrix.

Stephen Downes, University of Utah
"Genetic Control"

I take up Maynard Smith's challenge to investigate the role of the information gene by following control hierarchies down stream in developmental cascades. I begin by introducing some definitions of the gene that attribute differing degrees of control to the gene. Next, I introduce some specific examples of developmental control hierarchies: cases in which alternate post-transcriptional splicings lead to different phenotypic outcomes. I then discuss the issue of whether control mechanisms involved in development that are supplementary to DNA sequences can be productively understood in informational terms. One implication of Maynard Smith's view is that supplementary control mechanisms cannot and should not be understood in informational terms. This is not the only position available. First, even Maynard-Smith's own concept of information can be applied to some post-transcriptional controls. Second, some researchers in developmental and molecular biology already invoke informational concepts of control when accounting for both intra and extra-cellular control mechanisms in development. Investigators could adopt this approach merely because information terms are prevalent and hard to avoid but I introduce an example of this approach that has a different motivation. Finally, I briefly consider some more general implications of the appeal to information in biology.