[Martin Taylor 920816 01:30]
(Bill Powers 920815.0930)
The rest of the environment contains so few instances of significant
amplification that they're pretty unusual. It's hard to think of a case in
which the gain of a human controller isn't higher than that of the
environment by a wide margin.
I don't think the point is worth belabouring, but elasticity has much the same
perceptual effect as an interaction over a physical variable with another
controller that has a different reference. At one extreme we have a totally
non-resisting medium, in which (apart from disturbances) the physical variable
is readily induced to provide the desired percept. At the other end of the
scale, we have crystalline elasticity, in which the controller can have only
a trivial effect on its percept, because the world resists strongly any attempt
to deform it. In between, there is a whole range of impedances, both
resistive and reactive. Whatever the real-world impedance might be, it affects
the dynamics of control, and the behaviour of the controller is intimately
bound up with the real-world impedance of the provider of the percept. So it
is difficult to say straight out that the control is totally in the head,
even when there is no identifiable active controller affecting the same
physical complex variable. (Impedance is a term from linear dynamics, but
consider any nonlinear extension of it, in the above).
As to the mirror diagram, it is an epistemological tool. This diagram could
be expanded to show the hypothetical world of physical reality between the
two mirroring hierarchies.
My reference to the mirror diagram was not to the one linking two controllers
in the Layered Protocol structure, but to the one depicting the idiosyncratic
structure of the Universe as seen by any one controller. (Bill, did you ever
get my Paris abstract with these pictures in it?).
Both the mirror diagram and the considerations of loop dynamics as affected
by "real" world impedance argue that there is validity to the view that
the internal and external worlds cannot be completely separated. All we have
to work with is perception, after all.
My formulation puts a physical world
running by physical laws between the organisms which are control systems
running by closed-loop laws. That seems to work passably well.
Yes, mine, too, in the Layered Protocol structure as integrated with PCT.
And before, but not so explicitly.