Re: [FIS] Re: What is information?

From: John Collier <[email protected]>
Date: Tue 27 Sep 2005 - 12:38:00 CEST

Folks:

I hate to keep harping on this, but ...

I basically agree with Kevin here (note that as with Lazlo's submission, I am trying to find commonalities rather than emphasize more arcane distinctions). I think the discussion is going in a very useful direction, at the proper level of generality to get more towards a general notion of information that is useful in science. Some comments.


At 04:54 PM 2005/09/26, Kevin Kirby wrote:

 
Dear colleagues,
 
The cumulative attraction of this online discussion, and now especially Pedro's posting of 22 September, has hooked me.  If you can forgive my late intrusion, I would like to try out the following.
   
"The non-formal properties of information": here I believe Pedro is bringing out something extremely important. I wish I could explain clearly why I think this is crucial to the notion of natural computation, but I need to work at it more. Here, though, are some very crude comments in that direction.

"Formal" is ambiguous between formal in the mathematical sense and in the sense of form. Both are relevant to a general theory of information, but the latter is the most important. The first is "just mathematics".

Look at the idea of the imposition of form on substance. What does this mean? Start with a piece of the world. Take, say, a ribosome in a living cell. I mean a specific ribosome, this one, here, under this microscope now. Individual. Particular. Not the type "ribosome." This sense of "substance" I want to employ here.  (The clumsy phrasing in this paragraph is me trying to distinguish this from the Aristotelian notion of substance as an object of logical predication.)

I note that substance in this sense, which is not so different from the Aristotelean sense, but is closer to the Lockean sense of individual substance, has something that makes it what it is, often called the individual essence. This essence is subject to formal analysis in the first, mathematical sense, but it is also a form in the informal sense. This can be used to bring the two ideas together. I do this in the opening section of my 'Causation is the transfer of information' (1999) http://www.nu.ac.za/undphil/collier/papers/causinf.pdf, though the terminology is quite different. Then, using algorithmic information theory I connect this notion of form to information theory. Then I reify the information, and apply it to the physical world including biology through the negentropy notion. This gives a way of talking about causal powers of an object (which include its functional powers. So we have a mathematical analysis of (informal) form, that we can use to discuss causal relation, including in networks. It is also connected directly to the concept of work capacity.

This ribosome has a function. Now it is a truism that "function" does not inhere in physical objects, but arises from a network of relationships and, perhaps, an interpretive community. We can, in a suitable frame of mind, look at a thing and ignore its function. What are we looking at? That is, take away function from the ribosome, and what is left?   "Form". (The answer seems irresistible.) Now take away form, then what is left? "Substance" maybe? Is that really useful? At first, that sounds like some residual meaningless metaphysical ooze-- ousia, in fact. Isn't all of science about form anyway?

This seems to me to be correct. It should be noted, though, that it is in virtue of form that something can have a function. My recent paper, described above, suggests how to deal with this. Substance doesn't quite drop out, because I found I needed to justify connectivity, as well as, in other papers, the dynamical basis of objects and properties compatible with the informal (but formalized) approach to form.

Anything on the level below form seems below scientific discourse, which seeks out form. When we look at something as a scientist, we give it form. It may even become a "system", with "states." In computer science, representations of things are the object of study. We might represent an agent, a transaction, or a number, for example. (One thing you tell students in a computer science class is that it is meaningless  to speak of the complexity of testing if a number is a prime number; you must instead speak of the complexity of testing whether a representation of a number is a representation of a prime number. Some representations make it trivial; some make it hard.) 

Agreed. My paper was intended to explain away the illusion that substance and its nature was the study of science. I don't think that as scientists we typically give things form (though we may assign it form for the purposes of study). Instead, I think we try to find the form, and what forms underlie it.

But in _natural_ computation, things are different. Form, representation, system, state-- in some way we must be able to occasionally bracket these out and speak of substance alone. For example, an evolutionary exaptation happens when one part of a biological substance, which served one function, is recruited for another. The semantics is broken, and re-formed. There is some discontinuity in how we can track change in substance by change in formal representation. (I think the philosopher Hilary Putnam hit on one part of this in his arguments against computational functionalism.)  I believe these "non-formal properties of information"  are at the core of the Conrad tradeoff between computability and amenability to evolution.

This is interesting. It is noteworthy that in evo-devo it has been found that morphogenetic units can remain constant whereas genetics and function both change. I note, though, that the description of these units in terms of morphology suggests that form is still of the essence. I think that "The semantics is broken, and re-formed." requires semiotic (or hermeneutic) analysis, but that this can still be understood in terms of the informal forms, and more easily so in terms of a formal representation of this idea.
 
I concur that the Conrad tradeoff depends on the informal aspects of information. Information is richer than the notion of a program provides. Algorithmic information theory, however, is a much richer theory, and can describe more aspects of form, but it still cannot fully capture form on its own.

A separate observation: Pedro writes of  looking at the thinker is his/her "spare time, just when he/she is not acting in that peculiar 'scientific' way."  This evokes Wittgenstein (Philosophical Investigations 132 in particular), where he points out that the confusions about language that vex us do not occur when language is playing its real-life role (a non-scientific, non-philosophical role-- in Pedro's scientist's "spare time"). The confusions only occur in when language is taken off-line, as it were, disconnected from its role in life, set on a rack ready to be examined. [W's famous negative simile is that of language in this mode being like "an engine idling".  I just looked up the German text and there is no "engine" mentioned directly: it's just "wenn die Sprache leerlau"ft." Not being a native German speaker, I don't get the nuances, but the notion of language "running on empty" seems engaging...]

Again I concur, and some recent work of mine is intended to get at this issue without throwing out the baby of formal methods: Pragmatist Pragmatics: The functional context of utterances (2005, with Konrad Talmont-Kaminski) http://www.nu.ac.za/undphil/collier/papers/pragmatist%20pragmatics.pdf We use a Peirceaan approach along with some contrary approaches in analytic philosophy to motivate our view, rather than Wittgenstein, who is notoriously cryptic, and not very useful to refer to when trying to attain clarity. (I should not that I admire Wittgenstein, and got an A from Rogers Albritton on my graduate seminar paper for him.)

The privileging of the real-life notion of information connects back to some remarks by Hans several days ago, on the similarities between trying to define information and trying to define energy. In fact, the art in a beautiful loose definition such as (for example) "energy is the ability to do work" is impressive. It is invariant across different sub-notions of energy (Gibbs free energy, Helmholtz free energy�) and changing understandings of fixed notions of energy (pre-relativistic vs. relativistic, say).  And its invocation of work hooks it in both to another set of ordinary-life intuitions (I'm using "energy" pedaling this bicycle) and to another physical concept (work) which can also admit further mathematical elaboration. By so neatly injecting the concept into the web of other physical concepts, it helps one do physics. (Ok- I am not a physicist.) This is at least as valuable as a formal definition that gives precise necessary and sufficient conditions for something to be called "energy." Accordingly, I am a great appreciator of definitions of information such as "reduction in uncertainty" or "difference that makes a difference." They tie this plural concept into a scientifically useful semantic web. If one wishes to charge into the still-dark territory of "non-formal" information, it is definitions like these that will light the way.

Amen.

John
http://www.nu.ac.za/undphil/collier/index.html -------------------------------------------------------------------- Please find our disclaimer at http://www.ukzn.ac.za/disclaimer -------------------------------------------------------------------- <<<>>> Received on Tue Sep 27 12:40:07 2005