Re: Complexity
DON MIKULECKY (MIKULECKY@VCUVAX.BITNET)
Thu, 25 May 1995 10:49:18 -0400
Don Mikulecky, MCV/VCU,Mikulecky@gems.vcu.edu
Some comments on Jeff Prideaux's comments:
Jeff wrote:
> Of course Rosen defines complexity in the following way: (excerpts
> from LIFE ITSELF).
>
> Analytic does not equal synthetic
> Semantic does not equal Syntactic
> Direct product does not equal direct Sum
> external description does not equal internal description
> fabrication does not equal physiology
>
> and
>
> A complex system contains a non-simulable model (something that
> cant be done on a computer)...
>
> A complex system is non-fractionable
>
I don't think he says this or else you said something other than you
meant to. What he DOES say is that complex systems contain functional component
s
which are not fractionable. An active site on an enzyme may be
fractionated, but it looses its identity as a functional component's "home".
> Complex systems contain impredicativities (self-references)... It can't
> be defined without referring to itself...
Does an artificial neural network meet this criteria?
>
> There is no (mechanistic) largest model (that explains everything)
>
> The categories of causation in it cannot be segregated into discrete,
> fixed parts, because fractionability itself fails.
>
> No state set built up synthetically from the states of minimal models.
> This takes us out of dynamical system's theory...
>
> ------------
>
> Of course, now, to understand these definitions, one has to understand
> these concepts (referred to by these terms)...
>
>
> One hint on a possible strategy to find a realization...Rosen implies that
> complex systems may interact with different observables than we measure. Even
> though we may not be able to directly perceive these observables, we may be ab
le
> to know under what circumstances these different observables exist...and then
> put certain appropriate subsystems together so that they can interact using
> these observables... If these subsystems then interact in a way that is
> substantially different than anything our dynamical system's equations can
> predict, then we MAY have a complex realization.
>
> Of course, the reductionists would then just call this a new force and go on
> their merry way...It would then be interesting if a Godel-like theorem (for th
e
> physical world) could be developed that showed that a new complex system
> (interacting with new observables) could be developed for each "new force" add
ed
> to the reductionist repertoire.
The other thing he says is that these are sources of complexity and
emergence, in some cases, because we have been "conditioned" to look at systems
and interact with them in one way. Casti uses this argument almost verbatim
in his 1994 book "Complex-ification" without citing its source [Rosen:
Anticipatory Systems (1985)]. Up to now, Casti has used a lot of Rosen's
stuff almost verbatum, but has referenced him. Is this some influence
of his recent association with the Santa Fe Institute?
Best wishes, Don Mikulecky