[Martin Taylor 960820 15:00]
Bruce Gregory (960820.1315 EDT)]
(Rick Marken 960820.0750)
to
Hans Blom (960820) --
Ok. Now the questions is: since you have no data that demands an MCT
model of behavior and we have a considerable amount of data that is well
handled by a PCT model of behavior, why do you argue for an MCT model of
human behavior? Is it faith? Respect for the idols of the tribe (of
control engineers)? What?
I suspect that not everyone feels that Occam's razor is as
indispensable as you and I do.
Don't forget that Occam's razor is personally wielded. It's not a weapon
in an absolute Newtonian barbershop. It says that "entities should not
be needlessly multiplied" or some such wording. In more modern language,
if you have a model or a theory that accounts for a lot of your perceptions
pretty well, and someone else draws to your attention other data, whether
you _should_ change your model or theory depends on whether you can do so
more simply by using the model you already have, by taking on board a
whole new model _in addition_, or by taking on a new model _and seeing
whether it fits the old perceptions as well as what you had before_. If
your old model accounts for the new data, you shouldn't change, even if
the other people claim that their new model fits your data as well as their
new data. But then, neither should they change, according to the principles
of Occam's razor.
Occam's razor shaves where the wielder chooses, having no will of its own.
Like everything else in our understanding of the world: "It's all perception."
Hans's claim was that he has seen no situation in which a pure PCT approach
works any better than the MCT approach that he already knows. It costs him
nothing to apply MCT to the data proposed to him, and therefore Occam's
razor cuts out PCT for him, not MCT. He also thinks that MCT accounts for some
data that he cannot account for by PCT, and that makes the Occam's razor
cut even cleaner. For people who do not know MCT, but do know PCT, and who
have seen no evidence that there are data accountable with MCT but
not PCT, Occam's razor cuts the other way.
But all responsible scientists have to shave with Occam's razor, since it
results mathematically in choosing the higher probability theory in almost
all cases. (Remember that probability, like all else, is subjective.
To choose the higher probability theory doesn't mean to choose the theory
that ultimately proves to be right).
The answer to Rick's question quoted above should be that Hans has a
considerable amount of data from his experience as a control engineer
that is well handled by MCT, and none that cannot be handled by MCT. In
that case, why should Hans argue for PCT? Would it be faith? Respect for
the idols of the tribe of PCT enthusasts? What?
The test is to find situations in which the two theories MUST give different
answers, and in which the protagonists of both agree in advance that this
is so. I don't think that has been done, and I'm not sure it can be done.
Martin