Furthering PCT

[From Samuel Saunders(19951206:15:18:52 EST)]

Bill Powers (951206.0530 MST)

I suppose we ought to make a list of the experiments that have been
done. There are about 10 or 12 in the demo disks where the PCT
predictions are tested every time a new person goes through them. Rick
and I have done perhaps 5 or 6 together, and Rick alone has done a dozen
more. Tom Bourbon and his students have produced a considerable number
of multiple-person experiments. In all these cases, model behavior and
real behavior correlate 0.9 and above, usually far above. Dick Robertson
and David Goldstein have done one with self-concept; McPhail, Tucker,
and students have done several with the crowd simulation model. Who have
I left out?

I did not mean to belittle the work that has been done. I am sure that all
of this has been difficult, particularly in the face of a lack of
familiarity and even hostility from reviewers. I read Rick Marken's _Mind
Readings_ in August and found it captivating. The one critique I might
have is that almost all of the papers read like demonstrations of the fact
of control, rather than studies of control and control systems. I am sure
that part of the problem is in publishing in non-PCT journals, where
editorial requirements probably include a great deal of background
presentation. Much of the PCT research is unpublished. I read Rick
Marken's unpublished manuscript "The Hierarchical Behavior of Perception"
about 18 months ago and have been anxious to see a follow up ever since;
there are a lot of good ideas there waiting to be explored.

I expect that PCT needs its own journal. I know that _Closed Loop_ folded;
perhaps an electronic journal could be set up to replace it. I suppose
proposing something like that obligates me to volunteer to help; while I am
not yet skilled enough in PCT to be an editor, I would be willing to review
if it would help.

The most important thing, I am sure, is for people who are interested in
PCT to start doing PCT work (mia culpa, mia culpa, ...) instead of debating
the relative merits of PCT and other approaches. I am convinced that the
only way to win acceptance for PCT is to present a corpus of PCT work.
Objectively, the number of PCT publications to date would be considered low
output for an individual, let alone a field. That there are reasons for
this does not change the fact. As scattered as the present published work
is, it is impossible to tell an interesting story based on PCT results that
doesn't involve substantial unsupported speculation. As long as this is
true, speculative counterarguements will appear viable.

My intent in discussing reinforcement views of a PCT work on control of
feeding was to point out that just as EAB experiments appear to be odd
constructions from a PCT view, PCT experiments appear as odd constructions
from a reinforcement theory view. Until there is a supporting framework on
which to hang the PCT work, the reaction is likely to be "I don't know why
anyone would want to do this, but I think I can explain why his weird
procedure produced those odd results." It is fine to lament the lack of
appreciation of PCT, but the only way to do anything about it is to do PCT.
People need to be left after reading PCT work as I was after reading Rick
Marken's unpublished paper saying "then if you did this, that would happen"
and "I wonder if changing that would ...", not saying "you sure are more
scinetitific than I am" or "you sure convinced me that control exists." I
am afraid the only way to get there is the hard way, one paper at a time.

//----------------------------------------------------------------------------
//Samuel Spence Saunders,Ph.D.