[Martin Taylor 2003.01.11.1517]
From [Marc Abrams (2004.01.11.1436)]
> [Martin Taylor 2004.01.11.1211]
EVERY individual process in a control loop is an S-R process. The
perceptual input function is an S-R process. The comparator is an S-R
process. The output function is an S-R process. It's only when the
loop is completed that we have control. Controlled perception happens
only through the looped interconnection of a bunch of different S-R
processes.There's no contradiction between controlling perception and the fact
that the perceptual signal is constructed by an S-R process.I don't think this is quite accurate. I think you mean that each function is
connected by an s->r process and it is the collection of all the functions
in total that provide the loop.
No, I mean that for each function of the canonical ECU, there is an
input whose value uniquely determines an output. The input may, of
course, be distributed over time and space (i.e. the function may
well contain memory--in fact it almost has to).
That is true, but I'm not talking about
that. I am suggesting that feedback exists _WITHIN_ the input function and
that the production of perceptions requires a _number_ of different
interrelated control processes that ultimately provide us with our
perceptions
That's not inconsistent with the above. What would be inconsistent
would be if the feedback process extended to a path outside the
function under consideration, in the way the environmental feedback
path extends outside the 'skin' of the ECU.
However, I suspect you mean something different from either what you
actually said or what I just said. I suspect you mean that you think
there is some signal from outside that is distinct from the "input
signal" in the way that the reference signal is distinct from the
input signal of an ECU. That's a modellable proposition. The modified
ECU would then have three, not two independent inputs: (1) the
"standard" input vector, (2) the "standard" reference vector, and (3)
the signal that influences the input function directly.
It's a possibility, and a contribution would be to create a model
along the lines of Rick's spreadsheet model to show how it could
work. By the way, it's not the first suggestion along those lines. I
forget who, but I think it was Tom Bourbon who added a signal that
influenced the output function of ECUs in a network.
If you want my uneducated guess, I would guess that a refined model
would incorporate all four input paths to an elaborated ECU (not so
"E", unless you want to call it an "Elaborated Control Unit" :-).
However, I know of no data at the moment that require these changes,
in the sense of being better fit withthem than without them by a
degree that justifies the extra complication.
You of course bring up a very interesting point. _If_ what you were saying
is true. At what scale would feedback become relevant? Because what you are
saying is that you can view _ANY_ part of a process, and decide it's an s->r
system and you would be right, but it would also indicate an incomplete view
of the entire system, which would also be correct.
I don't think I am saying that. What I am saying is that if you
define some part of any circuit so that it has only one input and
only one output, which is a function of that input, it is an S-R
process. However, if the output of the process influences its own
input, then it is not an S-R process.
If the arbitrarily defined part of the circuit has more than one
input, then you might call it a "partial" S-R process if the output
influences neither input. I call it "partial", following the
nomenclature of partial differential equations. It is S-R for either
input, so long as the other is held constant (for at least as long as
the unit's internal memory for the state of that input).
Which view is more
important? And at what scale? Since there are obviously embedded loops
within embedded loops within still other loops. And looking at any 'chain'
of variables _without_ completing a loop would be considered an 'open loop'
or s->r process. I guess it depends on what your purpose is in the
investigation.
Yes, that seems to make sense (though I reserve the right to change
my mind without notice 
Martin