[From Bill Powers (931216.1020 MST)]
Osmo Eerola (931216.0910) --
Reading over your post again, it seems to me that we can clear
some things up by looking more closely at your diagram:
_______ Energy
I I flow
------------------I G2(s) I---->
_______ I _______ I_______I
Comp I I I I I
Ref.---->X--------I G(s) I------I H(s) I---I
I diff I______I S I_______I P I
I I
I----------------------------------I
You say that this system "controls perception," with which I
agree. It is true that if the steady-state loop gain is high
enough and the system is dynamically stable, the system will
maintain P = Ref, very nearly. The error signal "diff" will be
kept small.
The perceptual signal P represents some state of the environment,
here shown as the variable S. If H(s) = 1/s, then the perceptual
signal represents the time integral of S, and it is the time
integral of S that is controlled. If H(s) = s, then the time
derivative of S is perceived and controlled. Exactly what it is
about S that is controlled depends on the form of H(s), which in
PCT we call the perceptual input function.
If the time derivative is controlled, then setting Ref to a
constant value will result in S changing at a constant rate, the
rate depending on Ref. By varying Ref, higher systems can vary
the rate at which S changes (provided S doesn't reach a physical
limit).
This is why we say that the form of the input function determines
what is controlled, while the reference setting determines the
state of the controlled aspect of the environment.
···
----------------------------------------------------------------
Best,
Bill P.