Skepticism

[From Rick Marken (931114.1600)]

Martin Taylor (931114 01:20)--

One should test one's
own position more severely than others. If you can't find another way
to do something as simply and effectively as the way you believe is
right, then it may well be right. But without applied scepticism, you
can build wonderful castles in the air. I know you have modelled and
tested for decades, but the sceptical checking should never stop.

The test for controlled variables is the basis of PCT. We don't suggest
a PCT explanation of a phenomenon unless control is explicitly determined
by "the test" or implicitly revealed by resistence to natural
disturbances. PCT is based on skepticism about control -- "the test" is
the embodiment of that skepticism.

The level of skepticism seen in PCT is certainly NOT seen in sciences that
assume behavior is an open loop process. If there are any castles in the
air, they have been built by open loop, feedforward modellers -- not PCTers.
So your suggestion to Bill P. that he be "skeptical" seems a bit gratuitous,
especially since you have not suggested a similar degree of skepticism from
those (like Hans) who are advocating feedforward models. Why not encourage
some skepticism amongst the hugh majority of psychologists who are not
skeptical AT ALL about the open loop nature of behavior? Just a teensey
weensey bit of APPLIED skepticism (testing for controlled -- or uncontrolled
--variables) could have saved the social sciences from the continuing
idiocy that is spelled out in non-skeptical detail in every issue of
Psych Review, JEP, BBS and the Harvard Educational Review (right Gary?).

if there exists a mechanism like the Lang-Ham connection,
which allows a change in reference to generate a signal usefully close
to the right magnitude, a mechanism that is ALWAYS active, then that
mechanism would inevitably allow some useful action when for some
reason feedback was eliminated.

The way science is typically done is to try to explain observations with
as simple a model as possible. Additions to the model are not usually
made until you make observations that CANNOT be handled be the existing
model. The Lang-Ham model explains nothing that cannot already be quite
precisely handled by the existing model. It seems to me that a skeptical
person would want to be at least as skeptical about models as they
are about observations. If someone shouldn't believe that a behavior is
closed loop until it has been shown to be such (via "the test" I presume)
why should one believe that a model has possible merit when there is
NOTHING to suggest that it does? There are NO observations demanding
a Lang- Ham explanation of control. There MIGHT be some day -- but there
are none now. Shouldn't we be skeptical of answers (like Lang Ham) to non-
existent questions?

How, in
any case, would one know whether disturbances were for the moment
absent? (Aha...Is it a presumption that one COULD do this that is
at the core of Rick's insistence that "there is no information about
the disturbance in the perceptual signal?" If he thought I thought
that, it would explain a lot of fruitless past arguments).

No. My insistence has been and always will be based on the
fact that a controlled variable, p, is AT EVERY INSTANT
a simultaneous result of both disturbances AND its own outputs;
in the simplest linear, additive case p = o + d. So the system
controlling p has NO WAY of knowing the extent to which the
value of the perceptual variable AT ANY INSTANT is determined
by the disturbance, d. The only way out of this is to claim
that the word "information" actually means something so esoteric
and weird that it doesn't matter (to me anyway -- certainly not
to the control system itself) whether THAT kind of "information"
about the disturbance exists in the perceptual signal or not.

Best

Rick

[Martin Taylor 931115 10:40]
(Rick Marken 931114.1600)

So your suggestion to Bill P. that he be "skeptical" seems a bit gratuitous,
especially since you have not suggested a similar degree of skepticism from
those (like Hans) who are advocating feedforward models.

There is an intrinsic difference between being sceptical about claims of
"some, occasionally" and claims of "all, always." "Some, occasionally" is
itself a statement of scepticism, which Hans has been making. You, and
to a lesser extent Bill and Tom, seemed to be making the "all, always"
claim, which deserves a rather greater level of scepticism.

The test for controlled variables is the basis of PCT. We don't suggest
a PCT explanation of a phenomenon unless control is explicitly determined
by "the test" or implicitly revealed by resistence to natural
disturbances.

You still don't seem to have realized that testing for whether the variable
is controlled is not a test for whether there is a feedforward component
in the action mechanism. "The Test" determines the fact of control. It
is not an exclusionary device for testing the existence of other factors.

Why not encourage
some skepticism amongst the hugh majority of psychologists who are not
skeptical AT ALL about the open loop nature of behavior?

I do, I do! To the boredom of my colleagues. But not on a mailing list
where the contributors already are basically committed to the idea that
all behaviour is the control of perception. And here, I don't advocate
scepticism about that basic statement. (Mainly because I cannot see any
way it could be false; maybe I should be more sceptical about it, but I
can't find a way).

The Lang-Ham model explains nothing that cannot already be quite
precisely handled by the existing model.

A big claim, indeed. Bill and I have been having a little dialogue with
simulations on this question over the weekend (no answers yet).

ยทยทยท

======================
I see that you still don't understand the basic concept of information.
Apparently I was correct when I said:

How, in
any case, would one know whether disturbances were for the moment
absent? (Aha...Is it a presumption that one COULD do this that is
at the core of Rick's insistence that "there is no information about
the disturbance in the perceptual signal?" If he thought I thought
that, it would explain a lot of fruitless past arguments).

Your response, although it starts with "No," is completely compatible
with my presumption. You see, I have absolutely no quarrel with your:

So the system
controlling p has NO WAY of knowing the extent to which the
value of the perceptual variable AT ANY INSTANT is determined
by the disturbance, d.

but totally disagree with the immediately following:

The only way out of this is to claim
that the word "information" actually means something so esoteric
and weird that it doesn't matter (to me anyway -- certainly not
to the control system itself) whether THAT kind of "information"
about the disturbance exists in the perceptual signal or not.

(I can't, obviously, disagree with "to me, anyway"). But I do disagree
with "esoteric", "weird" and emphatically with "certainly not to
the control system itself." The entire functioning of the control
loop depends on not "whether" but on "how much of"

THAT kind of "information"
about the disturbance exists in the perceptual signal

As I said to Tom last week, it seems fruitless to continue this discussion
here. It gets so wound up in misunderstandings of the basic notions of
probability and information that nothing useful is communicated. I have
been encouraged by a private posting to continue my "Information,
Perception, and Control" paper, and maybe I will do so. Meanwhile, you
do have a copy of the prologue, which should help a bit toward understanding
the probability basis.

Martin