Antisocial; preferring signal over noise

[From Bill Powers (930409.1530 MDT)]

Bill Cunningham (930409) --

Right now, I'm strongly focused on the problem I'm trying to
solve. The variable I'm controlling for is progress toward
solution of MY problem. Put brutally, my interest in PCT, IT
or any other theory is directly proportional to the extent
which the theory contributes to progress.

Sounds pretty antisocial to me. What was that you were saying
about the vital role of social interactions....? Just pulling
your leg.

I think you pretty well describe the situation on both sides of
the discussion.

It occurs to me that somewhere in the future, when PCT begins to
seriously attack problems of higher levels of organization, we
may well run into situations where the mathematics of uncertainty
will be needed to describe how people manage to conduct their
affairs in the absence of clear perceptions. Even in such a case,
I would still resist introducing metaphors like "information,"
but I can see a place for the techniques.

I awoke from a little nap with still another thought developing
out of the latest insights. The kinds of experiments and models
we use in PCT, 1993, involve large unambiguous signals that vary
smoothly from one state to another -- actually, without any
discernible "states" at all. These variables are related by
regular analytical functions. If there is any noise, it's just a
little ripple down near the zero mark on the meter. What on earth
would be the point of taking these nice clear orderly waveforms
and treating them as if they were random variables? Reducing all
this data to terms of bits and probabilities and contingent
uncertainty would simply be to discard most of the data and do
everything the hard way with less reward.

For my part, I want to pursue PCT and modeling as far as possible
by looking for ways to get noise-free data and noise-free
measures of behavior. I suspect that much of what appears now to
be uncertainty in human behavior appears that way only because of
using the wrong model. Even at the level of motor behavior, we
have already found a high degree of order in behavior that has
seemed, in the past, to contain large random components. I think
we will continue to find this kind of order for a considerable
time to come, if only we don't give up when we don't see the
ordering principles the first time we look. I think that
psychology basically gave up on finding true orderliness in
behavior a long long time ago, and reverted to statistical
generalizations as a substitute. I haven't given up, and don't
plan to.



Bill P.