[From Rick Marken (960827.1000)]
Martin Taylor (960827 10:50) --
Rick keeps citing PCT emulations as matching human behaviour within 5%
or better. That can happen, but it doesn't always happen.
So far, the only time it hasn't happened is when poor experimental
procedures were used. But this is irrelevant to the question of the
relative empirical merits of MCT vs PCT. MCT is a much more complex model
than PCT and I know of no instance where MCT has been shown to account for
human behavior better than PCT. At this point, MCT has to be considered a
non-contender.
The question is: Is there _any_ MCT-based control system that does not have
a behaviourally equivalent HPCT system, and vice-versa?
No. The question is whether one can create a situation where the behaviuor of
the MCT model clearly differs from that of the PCT model. I bet that my
variable feedback function situation is, indeed, one where the behavior
of the MCT and PCT model will differ substantially and, thus, make clearly
different predictions about how people will behave.
Me:
But my little feedback function experiment shows that a control system's
structure and parameters tell you _precious little_ about the environment in
which it is designed to function .
Martin:
Your position here is very like your position in the "information about
the disturbing influence" debate: "very little" == "none".
As Eli Wallach said in "The Magnificent Seven": "Generosity! That was my
first mistake"
Ok. A control system's perceptual signal contains no (zero, nada, none)
information about disturbances to the controlled variable. A control system's
structure and parameters tell you nothing (zero, nada, none) about the
environment in which it is designed to function. Does that help?
Bruce Gregory (960827.1135 EDT)
in the United States the right wing wants to return to the
Enlightenment -- Newt would settle for 1928. In Spain, my
friend said, the right wing wants to return to the Inquisition.)
I see things quite differnetly. If the US right were interested in going back
to the Enlightenment I'd be a right wing zealot myself. In fact, the right is
precisely interested in going back to the Inquisition -- with that Torquemada
Gingrich leading the charge against "liberals", "gays" and "athiests". The US
right (like the US left in the 1900s -- remember William Jennings Bryant)) is
now a religious movement and there's nothing enlightening about that!
Bruce Abbott (960827.1055 EST) --
In the future I will use "model" to mean "analog" and "structure and
parameters tuned as required by environmental characteristics" instead of
"implicit model," if that will make you happy.
It will, indeed. But now you can go ahead and use "implicit model" because
I will understand that you mean "structure and parameters tuned as required
by environmental characteristics" -- that is, you are referring to something
about the control system that is _not a model_ (not an analog) of the
environment. Just as I thought. Why did you protest?
Then why do you insist on suggesting that I am trying to link "implicit
model" to the environmental feedback function, period?
I didn't insist that. Sorry if you thought so.
A properly-tuned system would have to take account of the normal effects of
disturbance on the CEV, not so?
Absolutely correct! But, of course, it doesn't take account of them by
modeling them (developing an analog of them); it takes account of the by
tuning it's parameters so that the result is good control.
I'll take someone capable of thinking _rationally_ about the issues over
either one.
But what if you don't like the results of that rational thought. Rationality
is just one level of the hierarchy; there are many people on the left and
right who are quite capable of rational thought. What seems to distinguish
left from right is the references for higher level principles and system
concepts that evaluate the results of this rational process. The drug issue
is a good example of this. A rational analysis of the "war on drugs" shows
that strict enforcement of drug laws has little or no effect on drug usage
but a hugh effect on drug related crime (it increases it). Yet some people
(right and left, I'm afraid) want to continue the drug was because, even
though they don't like crime, they also don't like the idea of "sending the
wrong message"; that is, they don't want to perceive the "wrong" principle.
Politics is not a rationality problem; it's a problem of figuring out how to
organize ourselves to control, as best as possible, all the different kinds
of variables we want to control.
Still waiting for you to describe how all those "wonderful models" of
reorganization of which you spoke work -- you know, the ones that make any
"knowledge" of the environment unnecessary. (:->
You know how it works. The reorganization system is a control system that
controls it's perception of the level of error in another control system
(target system). The reorganization system uses the E. coli approach to
control; it randomly selects a change in a parameter of the target system and
makes no more changes as long as the perceived error in the target
system decreases or remains low; the reorganization system makes another
random change in the target system's parameter if the perceived error
in that system increases or remains large.
The reorganization system has no "knowledge" of the target system's
environment; the reorganization systen has no idea why it's parameter changes
are having an influence on the target system's performance. The
reorganization system only knows how well the target system is controlling
(because it perceives the magnitude of the target system's error signal).
That's all the reorganization system "knows" - - the amount of error in the
target system.
Best
Rick