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

Re: models - branch fromf GM discussion



Dan,

Interesting comments. I think I know what you mean - that one can reduce
to such models that retain certain forms of uncertainty that we
associate with complexity; and I have had similar thoughts. However I
would also think that models based on statistical uncertainty may not
actually be nature's complexity, but a surrogate that we create to model
complexity by the combination of certainty and uncertainty. That itself
may be an arbitrary dichotomy, i.e., part of the analysis. It is in this
sense that quantum mechanics, while it introduces many of the ideas of
complexity we might appreciate, is also somewhat limiting in its
representation of true complexity. Somehow, I think real complexity
involves the process of how order is incorporated into uncertainty (the
modeler), that is, a complex order rather than an algorithmic one and a
complex unknown, rather than a random one. This seems to get to the
points about simulations vs. complexity as well..

On your second point, I agree there must be some energy cost. Many of
our evolved models are mechanistic, of course, as can be seen in sensory
perception and by how hard it is to get over the habit, so there must be
a cost advantage to many mechanistic models. A cultural example might
be  the gross simplifications leading to judging people on the basis of
stereotypes. It seems most likely to me that the selective advantage for
more sophisticated, and more correct, models of human behavior depends
on the complexity of the social context. Taking this example, the need
to understand other people better than through stereotypes arises as our
global relationships get more complex, making certain mores and
attitudes like what is being revealed in the global press today about
prisoner of war abuse, seem like an obvioius throwback to an earlier,
and simpler, era which most of us would now condemn as at least
ignorant, if not immoral; but which at one time enjoyed a certain amount
of acceptance and success. What we are seeing, I believe, is a culture
that has not adapted itself to the modern era (although one may also be
able to argue that it was never "better" it may have been allowed to
exist as a greater efficience simply because the awareness and
sensitivity that would make it inefficient was not widespread). Clearly
there once was a selective advantage in society to think in simplistic
ways, and given how sharply divided the public still is on most issues,
we don't yet know how much of a complex view of society is going to be
more successful than the simple. Concepts like "cutting through the  BS
or red tape" "cut to the chase" "get to the point" "you're either with
us or against us" (i.e., polarizing choices) etc. all recognize the
value of simple models which still tend to rule society - i.e., some
singular aspect or clear choice becomes the working model of the system
we are manipulating. I suspect where the cost-benfit advantages lie will
always be context dependent.

JJK


Dan Fiscus wrote:


John and John and all,

Some good discussion. Two cents worth from my perspective:

1. Reducing while retaining full or extreme complexity is pretty easy.
There are myriad examples of non-linear models with just three or
even fewer variables that generate completely unpredictable and
non-deterministic output, like Lorenz' chaos equations, 3 body
problems, Robert May's logistic equations with chaos; the chaotic
waterwheel is even a physical embodiment of the Lorenz equations
that retains the complex, unpredictable behavior. Put another way,
the threshold of near infinite complexity is very low - just 3 variables.

2. One aspect of modeling I don't think we have given enough
thought and work to is that any model or modeling approach, in
order to be a viable activity in a living system, has to pay its own
way or even better, provide tangible life and physical benefits. In
other words, any modeling activity has some energetic, material and
other kinds of costs (ecological, biological, relational, risk, time,
etc.),
and these costs or negatives must be more than offset by some
benefits or positives in order for that modeling activity to be
sustainable or adaptive or viable or effective at enhancing survival of
those systems that employ it, over the long haul. I think this criterion
can help show that Rosen modeling in general provides better payback
or "return on investment" or evolutionary value than mechanistic
modeling. I don't have details on how to show this, but think it would
be based on the idea that the world is more complex than simple and
so a complex modeling approach is more suitable and effective as a
basis for survival in such a world than an simple/mechanical modeling
approach. Other angles based on context-dependency as in tuning to
and synergy with a local environment might help, too.

Dan

John Kineman wrote:

Hi John M.,

You are pointing out that we can mean different things by
reduction(ism). Certainly that is true.

Rosen, I believe, discussed it in terms of explaining a system in terms
of its components, i.e., reducing everything the system does to its
parts and not attributing any causes to its organization.  Since that
view - that the organization comes from the parts - is essentially the
mechanical view of nature, Newtonian, etc., that's what I meant by
"physical reduction:" referring to the common way the term has been
used/connoted.  An example (from Blackburn Dictionary of Philosophy) is
"reducing biology to chemistry, supposing that no distinctive biological
facts exist, or chemistry to physics, supposing that no distinctive
chemical facts exist." Clearly RR believed that certain distinctive
facts exist in biology that are not represented in physics.

However, the broader and more technical definition is simply explanation
of one thing in terms of something else. Again from Blackburn:
"reductionism (reductivism) A redutionist holds that the facts or
entities apparently needed to make true the statements of some area of
discourse are dispensable in favor of some other facts or entities."
Hence there is a sense where the use of any kind of model is a
reduction, for working purposes.

Those were the two usages I was referring to -- our own personal
definitions may be equally interesting, but the "confusion" I was
referring to is that most people, most scientists anyway, will relate to
reduction in the physical sense, i.e., as "reductionism," as defined by
Blackburn, reducing biology to chemistry and physics, and it is that
sense that RR seemed most opposed to. In the broader sense, however,
where any model is a kind of reduction, even Rosen theory cannot escape
it - it is still a way of thinking about nature, i.e., a surrogate for
nature, but it is not reduction to physics.

For more fun and grins, one can point out another confusion - that
saying RR theory is not a reduction to physics is incompatible with the
statement that RR theory implies "new physics" because in the latter
case we are again saying that what is new would be captured by the
supposedly improved physics. My personal belief  is that even new
physics can't do it without ceasing to be physics, and I think at the
core that was Rosen's belief too, but talking about new physics is a way
of getting the physicists to be a little more humble about the scope of
their discipline and to really be saying that some "new science" is
needed.. The new science in the Rosennean sense is Rosennean biology. If
one wants to play tit-for-tat, calling it "new physics" is reducing that
new part of physics to biology, where the "new physics" presumably comes
from. But unraveling that circularity is unimportant; all I'm saying is
that we need to distinguish between traditional reductionism and
modeling in general, which is a more general form of reduction.

JJK

John M wrote:

Hello, John,
do we have a concensus on reductionism? On another list and some years
back someone overpoured me with ½ dozen theoretical definitions from
diverse philosophers, in which I was not interested. To your remarks:
First: I never mentioned "physical reductionism".
I don't know about any connection of "ism"s to physical measures or
measurablity. Spiritualism comes to mind.
Reduction, however, brings to my spiritualism (just for the pun of it)
a reduction <G> meaning: to reduce a total view into some part of it,
substituting the part for the original (total?). The reduction occurs
by selecting aspects to consider, called: boundaries, WITHIN which we
observe and visualize. Topical is a broad selection.
A second step is "model" formation, we pick (consciously or by
occurrence) a limited cut from a topic, call it a name and consider it
as a "total" for our observation. All of our sciences are topically
identified, - all are reductionistic. That's my stance. \
(Except for (pure) math, which is a 'world of its own', with its
language and logic. As long as it is not "applied" to sciences, when
it runs into (contributes to) difficulties by reductionism.  - the
equational formalism of models.)

>"I would prefer to state it that way - "best available concept >of
nature." - "<
Available to us - at what level? The Flat Earth? It changed.

>We can take it as given that the way we understand nature >is through
models.<
Aren't we working on improving our understanding? Can you 'understand'
the wholistic interconnectedness, an unlimited impredicativity of a
natural system by clean-cut models? Isn't Rosenism a step forward from
the "model-view"?
Yes, models are useful, practical and usable. Conclusions have been
drawn in successful development for technology and the underlying
(reductionist?) science (formalism),  in a very complex (complicated)
fashion.
Models and IMO reductionism:  predicative, T-computable.
Wholism (RR-complexity): impredicative,T-non-computable.

Your take--------

John M


-- © 2004 John J. Kineman all rights reserved