[Avery Andrews 930502.0639 Eastern Oz Time]
A quick thought from a cursory look at Rick's and Bill's papers: it
seems to me that to get anywhere you can't just present your own ideas,
but need to actively attack currently popular ones, and this means
Kugler and Turvey, along the lines that I've been making noises
about, but keep getting diverted from the actual work (last weeks
diversion being organizing the acquisition of a 486DX-50 machine -
don't have it yet, but will by Tuesday).
The way I'd currently formulate it, I'd say that current `Action Theory'
makes many valuable contributions (emphasis on continuous functions
and dynamcs), but suffers from fundamental limits of scope that give
it no serious prospect of ever providing insight into the means by
which living systems function, survive and reproduce. This is because
(a) there are no guiding criteria for selecting appropriate quantities
to relate via differential equations (the choice of what aspects
of the system is represented by the variables is arbitrary) (b) No
method for figuring out what's going in a complicated system.
PCT addresses these issues by noting at the outset that what living
systems are is parts of the world that maintain local conditions
suitable for their survival and reproduction. Therefore the quantities
worth measuring are the ones that resist disturbance to a surprising
degree (this should include both static maintenance of an ongoing
condition, and achievement of a goal (a new/transient target
condition)). It seems to be the case that maintenance of certain
conditions (adequate level of nutrients in the blood) depends on
others in a complicated way, so you get a hierarchy of systems,
higher level systems maintaining their goals by means of lower level
ones maintaining theirs. The structure of this system can be
systematically investigated by testing for controlled quantities.
And so on.
And somewhere there has to be an attack on the Action Theory attack
on PCT, showing why they are (a) just wrong in their story about why
2nd order control systems (b) wrong in lumping together control
theory with cognitivism. I'd actually like to say that PCT is actually
a *version* of Action Theory, juiced up with some ideas that remedy
the fundamental aimlessness of straight AT (sorta like Chomsky did to
structuralist linguistics). Not sure how to do that at this point.
Avery.Andrews@anu.edu.au