FW: [Fis] The Molecule as Text

FW: [Fis] The Molecule as Text

From: Kevin Kirby <[email protected]>
Date: Mon 14 Nov 2005 - 11:05:10 CET

From: Kevin Kirby
Sent: Sat 11/12/2005 3:35 PM
To: fis@listas.unizar.es
Subject: RE: [Fis] The Molecule as Text

Dear colleagues,

The weekend has allowed me to catch up on this already multi-branching
discussion. Let me take one item that has appeared in two postings: "logic".

Pedro calls our attention to the special embodiment of biomolecular agency,
say across different domains of an enzyme, or across the genome,
analogizing it to the parallel decomposition of logical clauses that can be
"explored and solved in independent steps." On a different topic, Ted
remarks that the logics involved in natural computation are confined to
first-order and probabilistic logics, commenting that this is an
unnecessary limitation.

Well, I would say that logic is both less and more than this. The analogy
of a protein or sequence of DNA to a logical expression in some kind of
clausal form (say a conjunction of disjunctions) is a very lossy analogy.
It is all about using decomposition to help determine a truth value. But
perhaps we could rescue the analogy to logic (even to first-order logic) by
pointing out that things are more relevant when we do not have fully
clausal form. That is, consider the case of complex logical expressions in
which the scope of quantifiers extends across not-too-short ranges, and in
which identical variables reappear at different loci, introducing couplings
across the expression (like pronouns strewn across a paragraph in English).
I am not a biochemist, but the tension between decomposability and coupling
is what these two situations (molecular and logical) have in
common. Actually, to take it further, to escape the centrality of "truth
value" in logical expressions perhaps one should make the analogy to large
Lisp-like programs, the foundation for early 1990s work in Genetic Programming.

Must natural computation rely on first-order and/or probabilistic logic?
That would indeed, as Ted says, be a limitation. But I see the term
"computer science" at least as open as "logic." Well, maybe even more
open. The key insight in 20th century logic was the clean separation of
syntax and semantics, of proof theory and model theory. This is so
ingrained, and so attractive. However, I can detect in these FIS sessions a
sense that this clean separation is out of place when we consider life,
and, a fortiori, I would say, natural computation. Indeed. (I hope to post
something further on this soon.) I think the information-perspective
examination of molecular bionetworks takes computer science to exactly the
same place as linguistics when it began (in some circles) to move beyond
fully formalist syntactic theorizing (e.g. that there were purely syntactic
language universals) to cognitive and construction grammar. In the latter,
syntax is not separable, and the primitives of grammar are not atomic.
Categories like nouns and verbs have little integrity; the notion of a
"construction" is primitive. (See e.g. W. Croft, Radical Construction
Grammar, Oxford 2001). This has allowed (some) linguists to find more
useful ways to talk about the "embodiment" of language. An utterance, like
a protein, is a structured object that goes out into the real world and
plays a role, and has a very delicate, but very principled,
structure-function relation.

This is a sideways approach to Ted's molecules-as-text idea, I suppose.
Texts are structured, amenable to interpretation --but not freely so-- and
are sensitive to milieu.

So, I need now to think about (and attend to these discussions on) how a
formalism, say category theory, can deal with the embodiment of biological
matter in the world….

Many thanks for these stimuli…

-- Kevin
Received on Tue Nov 15 08:51:16 2005


This archive was generated by hypermail 2.1.8 on Tue 15 Nov 2005 - 08:51:16 CET