[Date Prev][Date Next][Thread Prev][Thread Next]   [Date Index] [Thread Index] [Author Index

Re: [life] MR as ontological; 3 kinds of life



Hi John,
 
See interposed. I think it's pretty safe to say we have some very deep differences in our interpretation of Rosen. 
 
Maybe Judith can comment if she has a chance?
 
Regards,
Tim
 
-----Original Message-----
From: ROSEN Forum [mailto:***On Behalf Of John Kineman
Sent: Tuesday, August 26, 2003 11:38 AM
To: ***
Subject: Re: [life] MR as ontological; 3 kinds of life

Comments below:

Tim Gwinn wrote:
I would strongly disagree. The behavioral homologies to which he referred
were those that were specifically characteristic of organisms, and which
were specifically not present in non-living systems.

If you want to discuss behavioral homologies among other (broader) classes
of systems, then you are discussing some other set of homologies, not the
ones that Rosen specified.
  
What is your evidence for that?  
 
 
It is not a matter of evidence, it is a matter of the criteria Rosen himself used. The criteria Rosen gives for the behavioral homologies are, as I quoted, "between the behaviors exhibited by organisms, homologies which are absent in non-living systems." [AS p. 3] So, when Rosen talks about living systems, he is talking specifically about organisms.
 
Again, if you want to
 
 Mine is his statements I quoted yesterday mentioning in AS how he showed relevance in many disciplines, both phsyical and biological. Specifically 3.3 (pg. 137) is about encoding of physical systems. 
 
 
Yes, 3.3 is "Encoding of Physical Systems". This section is about the general modelling of natural systems - representing natural systems in formal models.  What does this have to do with the scope of specific behavioral homologies of "living systems"?
 
 
I think you are arguing to limit the theory to biology, but for what purpose I don't know. Many aspects of it have general applicability.
 
 
Limit the theory? Your original remark to which I replied was:
 
> Yes. It seems consistent to begin with the "homology" of behaviors. In
> plain English, these systems behave similarly, and so we look first for
> a common way to identify them (the MR), and then a common explanation
> (possibly an MR!). The homology extends to quantum systems, social
> systems, organisms, ecosystems, and psycho-biological systems.
 
 
As I said above, you are free to discover homologies between all those broader classes of systems. They will not, however, be same homologies to which Rosen refers to in his living/non-living distinction.
 
It sounded like you were attempting to extend a definition of "living" to one that exceeds the one Rosen uses where he equates "living state" with "organisms". One which Judith has, I believe, reiterated several times. Again, in that same quote: "I am persuaded that our recognition of the living state rests on the perception of homologies between the behaviors exhibited by organisms".  There is no mention or insinuation of the application of "living state" other than specifically to organisms.
 
Further, from the last page of Life Itself (p. 280): "But complexity, though I suggest it is the habitat of life, is not life itself. Something else is needed to characterize what is alive from what is complex. Rashesky provided this too, in his idea that biology was relational, and that relation meant (as we stated it) throwing away the physics and keeping the organization.....Organization in its turn inherently involves functions and their interrelations; the abandonment of fractionability, however, means there is no kind of 1 to 1 relationship between such relational functional organizations and the structures which realize them. These are the basic differences between organisms and mechanisms or machines."
 
 

  
The only
thing left out are mechanisms, which unfortunately hold most people's
interest. But then he explains how mechanisms can be identified by
commutation of a  modeling relation, and how they can be obtained the
    


A status of 'mechanism' is not related to the commuting of an MR. This would
only be so if the only set of models that existed were simple (computable)
models.
I need the references on this please. There are many statements to the contrary.  
 
 
I can think of no contrary statements. From LI p. 203: "a natural system N is a mechanism if and only if all of its models are simulable." (where "simulable" = "Turing-computable" = "simple"). 
 
Conversely, in the definition of "complex system" [EL p. 306]: "A system is complex if it has a nonsimulable model". So, the very definition of complex system includes the notion of nonsimulable (complex) models. The "nonsimulable model" is (by virtue of being nonsimulable (not Turing-computable)) a complex system. 
 
The (M,R) model is perhaps the most prevalent nonsimulable model that Rosen uses. From EL p. 269: "In conclusion, any material relaization of the (M,R)-systems must have noncomputable models. Hence, they cannot be mechanisms, in the sense we have been using the term, because the (M,R)-systems are too rich in entailment to be accommodated within the world of mechanism."
 
Another prevalent complex modeling approach he uses is "activation-inhibition" networks, model systems which are interlocking networks of partial differential equations. When these differential equations are inexact, this system is complex. [EL ch. 22, p. 337]
 
It is true that complex system will not have a single "largest model", while a mechanism will. (The latter follows as a corollary of a mechanism having all its models be computable.) So, with a mechanism, there will be some single largest computable model which will commute fully with it. Is this perhaps the distinction to which you were referring?
 
 
 
 My thought is that it is a potential ambiguity that is resolved by being clear whether the MR is with a formal description, which then must be simple, or a natural system, which itself can be complex.   This is one of the stickiest parts of the theory, in my opinion, so not at all surprising there would be two interpretations.
Why does the material world need abstract models? Simply because we can
epistemologically consider them as having anticipatory or "model-based"
behavior need not imply that those models exist ontologically as distinct
entities.
Yes, if we look only at gross scale of classical systems where we don't see any apparent complexity. But these systems are not distinct and clearly are related to cases with few components that behave complexly. Thus, for a general theory, it works to explain the apparent mechanistic behavior by commutation with a model, otherwise there must be multiple theories, one for each kind of system. 
 
 
This does not seem to bear on the question of why these models need to exist ontologically as distinct entities.
 
 

 Protein folding seemed to have been argued by Rosen to be
anticipatory (Life Itself p. 271); however, this strikes me as just a way of
describing the behavior of a system. I personally do not think that Rosen
proposed or suggested that proteins actually have some accompanying
"non-material" (or even material) part that would be the model guiding the
behavior.
  
I am comfortable with considering strange realities at ontological levels, so for me it represents no difficulty at all to imagine a formal domain that is always present. 
  In fact, even traditional theory does this, but it separates it as a Platonic realm where incontrovertible "laws" exist. Admiting to models everywhere, while that may seem radical at first, is far less radical than the Platonic/mechanistic assumption.  
 
 I only says that systems are involved in generating the laws. Some have established rather consistent laws to our perception (mechanical laws), and that is partly because our sensory ability was constructed to capitalize on such consistency. So we see a material world, being a product of its rules. But that is not the only possibility. That makes a great deal of sense to me.  

For example, I do not think that Rosen would propose that organisms have
abstract formal (M,R) models as something that exists (materially or
non-materially) seperately in the world, somehow alongside the organism.
>From my reading he proposes that the models are part of the organism, but have extensions throughout other modeling relations which represent an "infinite mathematical object," i.e., the nested hierarchy of larger and smaller systems.  
 
 
I would disagree wholeheartedly.  I do not see where Rosen proposes that the models are a part of the organism. Instead, I think he indicated repeatedly that models are realized in organisms.  
 
I also recall no point where he intimated any "nested hierarchies" of larger/smaller systems.
 
 
 The most immediate relationship is with the fully embodied model, part of the organism, but information also is shared with larger contexts. This is very suitable for ecology. I read him as proposing to consider models separate from material structures, but both are operative in an organism. Additionally, in an organism, there is the ability to respond to the model for control purposes, thus producing a plethora of anticipatory behavior. That, I think, is all Rosen.  
 
 
Again, I disagree. I do not read Rosen as "proposing to consider models separate from material structures". I can recall no wording that speaks directly to anything like that.
 
 
 
 My own speculation is that the rules governing what makes an organism essentially define an amplification means which capitalizes on anticipatory behevior via evolution and adaptation. It is thus magnifying a property of nature that is apparent in un-structured matter (free particles), but disappears without the organismic way of preserving it.

Instead, the (M,R) model is an epistemological model that is realized or
embodied in the internal organization of the organism itself, and notably,
this organizational aspect is not identical with the organisms structural
organization. 

Yes, exactly my view as well. 
 
 
Hmmm, it seems your view is different. You say models are "separate from material structures".  This is an entirely different claim than Rosen makes (and to which I am making reference above):  "Organization in its turn inherently involves functions and their interrelations; the abandonment of fractionability, however, means there is no kind of 1 to 1 relationship between such relational functional organizations and the structures which realize them. These are the basic differences between organisms and mechanisms or machines." [LI p. 280]
 
Rosen is referring to differing organizational aspects - one structural and one functional - both of which are realized in the one material system. He is not referring to the structural organization as material, and the functional organization as "non-material". I think this deeply misconstrues Rosen.
 
 
 
 
Similarly, my take is that in order to epistemologically
represent these anticipatory systems whose future state determines present
change (such as a protein folding), we must employ formal models in our
modeling relations that have similar characteristics to those natural
systems. These formal models are anticipatory models, which contain the
necessary closed loops of entailment and the necessary descriptive
structure.
Here I think there may be a mis-step. There are many indications that all formal descriptions, taken alone, must be simple. We thus cannot create a complex model of anything, as a purely formal entity. Rosen mentions this problem and says, at best one would have to create an infinite set of models to capture the complex behavior. This is the principle on which I am constructing a modeling/mapping system for ecosystems. 
 
 
As I showed further above, there are complex models.   However, it is true that, in terms of simple models, only an infinite set of such simple models can fully model a complex system.
 
 

I have speculated about complex models as you seem to be doing here. By including human intuition and judgement in, say, a decision support system, we can regain complexity in the model, which now becomes a complete natural system, not just a formal representation. Rosen's references to the power of analogy now become highly relevant, because one can argue that the full set of complex possibilities humans may generate may in some important ways represent the possibilities of nature, simply because we are products of nature and share many similar properties. So, in this way, we can get an anticipatory model for managing ecosystems. In the first way we can get a multitude of simple representations that answer specific questions or represent the range of possibilities, as we fill out the infinite set, presumably establishing some priorities.

 However, just as in the case of the (M,R) model and the organism,
there is also no need for the structure of the anticipatory model to be
realized in the world in any explicit form. Instead, the anticipatory model
is realized or embodied directly in the protein's innate configuration.

  
I would quibble here, saying that no formal representation can actually exist independently (i.e. no Platonic realm). e.g., thoughts in our brain must be in relationship with physical substrate, although not reducible to them. So, somehow, there is a physical correllate of any model, but it may indeed be nearly impossible to localize.
 
 

Thanks,

John