Alexei Sharov wrote:
> Reply to John Kineman:
>
> >Hence, practical "reality," that is, anything we are able to study through
> >the senses, cannot be separated from models.
>
> This fits very well into the concept of pragmatism that I try to
> defend. In my discussion with Don I used the word "reality" as a
> substutute for an "ultimate model" no matter if it exists or not.
Here it probably will help to state our position.....that reality can not be
captured by an infinite number of models....but may be approached satisfactorily
close by one for certain practical purposes.
> For me it was important to show that we can talk about "mechanisms"
> and "non-mechanisms" only in application to models. As a result,
> the notion of reality is no longer interesting, which was the purpose
> of my arguments.
>
> What you call "practical reality" is not something independent from
> our thoughts (what people usually think of reality). Instread,
> "practical reality" is a result of our activity that follows specific
> models that humans have developed in their biological and cultural
> evolution. We consider objects mechanisms if our behavior directed
> to these objects is simple. For example, a slave-owner may treat
> slaves as mechanisms. Thus, being a "mechanism" is not a property
> of an object but a property of our interaction with that object.
Neat! I like that example!
>
>
> >Our reference to something "real" in a practical sense, by this reasoning
> >MUST involve some kind of model, and therefore MUST involve the concept of
> >level or scale, which is inherent in our definition of "things" or
> >"entities."
>
> I agree that the notion of scale is very important and that some
> objects will simply dissolve when viewed at a different scale.
> It seems that scale is not a part of the model but it corresponds
> to the kind of interaction with an object. For each scale we may
> develop a specific model that represents best our interactions at
> that scale.
Yes, that's how we see it too.
>
>
> >Here I consider two options:
> >
> >a) It is only simple because we look at it through a simple model. If we
> >looked at it as a complex thing, then it is complex;
> >
> >(b) It IS functionally simple because we MUST identify "it" before we can
> >contemplate "it," and thus the "it" refers to a specific scale and level of
> >practical reality that in every observation/interaction we or anything else
> >can make, is predictably and repeatably simple. It is not a matter of more
> >precise measurements revealing complexity, because at the level at which
> >this would occur (for an inanimate object), the "it" we are referencing no
> >longer exists. "It" is a macroscopic "it" defined only in terms of our
> >interaction with it at the macroscopic level. At the level where "it" is
> >complex, "it" is no longer "it" and we are no longer obtaining knowledge
> >about "it."
>
> >Indeed, the level at which a car becomes complex is a level
> >where the concept of "object" itself no longer applies.
>
> My impression is that we can talk about a car even at a molecular level.
> Of course it will not function as a car, but it does not matter
> because we look at the complexity of an object and not at its
> specific functions.
>
> >2. Living macroscopic "things" exhibit complexity at the same scale at
> >which the "thing" is defined, hence they retain both definition and
> >complexity.
> >3. Non-living macroscopic "things" exhibit complexity only at a scale where
> >the "thing" is no longer defined (or changes definition). Hence, in these
> >cases, the "thing" does not retain both definition and complexity, and
> >hence "it" is not complex (even though something else at this level is
> >complex).
>
> I had no intention to associate Rosen's
> "complexity" with life or organization. Life requires a semantic
> closure (self-observation), but in our discussion with Don we always
> talked about human models of other things (which is a much simpler
> matter).
>
> >>2. Models of real systems may be mechanistic (attempt to
> >>describe and control every detail) and non-mechanistic (that
> >>have some internal freedom, e.g., neural networks,
> >>genetic algorithms).
> >I would suggest classing these as mechanistic and quasi-mechanistic (i.e.,
> >incorporating concepts of freedom, such as uncertainty, into otherwise
> >mechanical models). I do not understand how a model can be entirely
> >non-mechanical without loosing the ability to define terms or reference
> >something relevant to us. Even theological models employ ideas of cause and
> >effect, which is part of the mechanical program. Do we not have difficulty
> >discussing a non-mechanical reality without using mechanically-based
concepts?
>
> This is an interesting question! I believe that some models may exhibit
> complex behavior in the following sense. Theoretically it is possible to
> follow their dynamics using cause-effect chains. However, it will not help
> us to understand how the model works because in the next run all
> cause-effect chains may change. It will take millions of years to trace
> all possible model trajectories and we still gain nothing. Theleological
> or phenomenological explanations may appear much simpler and useful.
> You can call this quasi-mechanistic models, but for me it looks like
> true complexity.
There is a class of models called "relational models" which do not use dynamics
or
the usual mechanistic tools. Also I have a list of topologically based models
which yield radically new information about the mechanistic world! As our
repertoire of ways of interacting with systems grows, so will our repertoire of
models and vice versa.
>
>
> >Can someone provide a specific example of an entirely "non-mechanistic"
> >model that is neither metaphore nor metaphysics?
>
> All models are metaphors (at least there is no way to prove that
> they are not).
A semantic problem......we call metaphor those incomplete modeling relations
which
decode without an encoding. I apologize for doing this since some seem offended
by
it, but I hate to forsake a well defined language for discussing these issues to
wallow in semnantic ambiguity.
>
>
> >how about (e) consider ALL models (including mathematics and logic) to be
> >approximations of reality, mechanical models being best suited for those
> >things that behave mechanically and quasi-mechanical models being best
> >suited for those things that do not; and recognizing that in any ultimate
> >ontology, all distinctions blur and only the instance of experience itself
> >is left as a means of understanding.
>
> "Approximation" makes sence only if you have a converging series.
> Scientific revolutions do not show much of conversion. I prefer
> to talk about metaphors. This is more safe because it does not
> require convergence.
You can also approximate with a set of distinctly different(not derivable from
each
other) models which may or may not result in a kind of convergence on the
natural
(real) system.
>
>
> I would say that mechanistic models are good for modeling objects
> with which we have simple interactions. It is amazing that very
> complex processes like movement of cars on a highway in rush hour,
> which involves a lot of passion, can be modelled by a couple of
> simple mechanistic equations.
>
> -Alexei
>
> -------------------------------------------------
> Alexei Sharov Research Scientist
> Dept. of Entomology, Virginia Tech, Blacksburg, VA 24061
> Tel. (540) 231-7316; FAX (540) 231-9131; e-mail sharov@vt.edu
> Home page: http://www.gypsymoth.ento.vt.edu/~sharov/alexei.html
respectfully,
Don Mikulecky