Doing PCT

[From Rick Marken (940618.0830)]

Well, I'm off for two weeks to where "the corn is green" (for
those of you who are not familiar with Bette Davis movie of
that name, I mean Wales). One quick (!?!) comment before I
go:-)

The discussion of PCT research has made me realize, once
again, that it is in the realm of testing the PCT model that the
revolutionary rubber of PCT really meets the inertial road of
conventional psychology. The creature we are studying,
according to PCT, is nothing like the creature that has been
assumed by conventional psychologists of all stripes -- SR,
cognitive, psychoanalytic, etc. Conventional psychologists
assume that we are trying to understand a creature that
responds to stimuli or generates responses and that the
reasonable way to study this creature is to watch it (see what
kind of behavior it generates) or "poke it" (see how it reacts to
stimuli). PCT, on the other hand, assumes that we are trying
to understand a creature that is controlling its own inputs. So
just watching such a creature tells us very little because all we
see are the side effects of actions aimed at keeping perceptual
variables (which we cannot see) in intended states. And
"poking" such a creature tells us little because almost any poke
will produce an observable action since we are very likely to be
disturbing one or several controlled variables; "poking" is only
useful in the context of testing for a controlled variable.

Testing to see the effect of "alerting" stimuli assumes that
people are creatures that respond to stimuli. A controlling
creature cannot be alerted by an external event; external events
can only be disturbances to controlled variables or part of the
feedback function that relates outputs to perceptions. So
experiments on "alerting stimuli" simply make no sense if we
are studying a creature that controls. It is possible to design
studies to show that "alerting" stimuli do not alert; to do this,
we would have to get a good description of how those who
believe in "alerting stimuli" think that they work; what is
"alerting". An experiment to show that "alerting" doesn't
work the way proponents of alerting say it does would be like
the studies Tom, Bill and I have done to show that concepts
like "stimuli", "reinforcement" and "planned output" make
no sense when you are dealing with a creature that controls.
Our studies have not been very convincing, however, because
proponents of these concepts always say that we have only
tested "straw man" versions of what these concepts really refer to.

So I think we (Tom, Bill and I) have concluded that this
strategy -- trying to show how existing psychological concepts
are inappropriate when dealing with a creature that controls --
just doesn't work. I think we have all come to the conclusion
that all we can do now is just forge ahead and do PCT research -
- research aimed at studying creatures that control their own
perception. The down side of this approach is that it pretty
much ignores all the psychological research that has gone
before. This probably seems like hubris or a slap at the
researchers who were our predecessors. But it seems that there
is no alternative. Previous research was aimed at studying a
different kind of creature than the one studied by PCT. There is
no way, apparently, to convince people (through experiment
and modelling) that this is true, so it looks like we'll just have
to go about our business, hoping that one or two people will
follow, and take up the PCT approach to research (rather than
trying to "build onto" what has gone before -- obviously a very
tempting seduction).

Going ahead and doing PCT research requires a revolutionary
change in one's goals and interests. The main change is that
one becomes more curious about what perceptions another
creature might be controlling; and less (indeed, non) curious
about how situations (type of "alerting conditions", how a
problem is "represented to" a person, etc) affect behavior. But,
perhaps, the biggest change is turning one's back on shelves
full of psychological research, much of which has become part
of psychology "lore". It is possible that some of this research
can be salvaged -- especially single subject studies in
psychophysics and operant conditioning -- and used as a basis
for PCT research. But I think it will be much more profitable
(in terms discovering what people actually control) to look
ahead and start doing PCT research, assuming that there is
little of value in the past, rather than looking back and trying
to find some previous research study that might have some
relevance to PCT research.

I takes some courage for a psychologist to actually start "doing
PCT" but I think the potential benefits of "doing PCT", in terms
of actually understanding human nature, far outweighs the
costs, such as loss of career, reputation, and financial security;-)

See you all in a couple weeks.

Rick

[Martin Taylor 940619 20:00]

Rick Marken (940618.0830)

Testing to see the effect of "alerting" stimuli assumes that
people are creatures that respond to stimuli. A controlling
creature cannot be alerted by an external event; external events
can only be disturbances to controlled variables or part of the
feedback function that relates outputs to perceptions. So
experiments on "alerting stimuli" simply make no sense if we
are studying a creature that controls.

So your non-constructive criticicsm of my proposed experiment was
based on purely theological grounds, was it?

According to any conventional reading of PCT, almost all perceptions
are not at any particular moment being controlled. You are asserting
with some force that no rational experiment can be done on such perceptions.
Also that no experiments can make sense if they purport to illuminate
the mechanisms whereby control passes from one perception to another.

I think this theology is as irrational as any other.

You started with repeated requests for a demonstration that there is a
phenomenon to be explained. You proposed an experiment on controlling
many perceptions through few output degrees of freedom, which by its
nature could not show the pheonomenon, and for which you did not have
a model "until I see the data [a free quote, not having the original
handy. MMT]". I introduced a modification that used the kinds of
uncontrolled perceptions that seem on phenomenal and physiological
grounds likely to participate in alerting, noting that I don't have
a specific model, upon which you produced a totally destructive critical
attack, based on the notion that if you don't have a model before you
see the data, it is the worst kind of voyeuristic masturbatory experimentation,
likely to condemn the perpetrator to a hell populated with Carvers and
Schreiers.

You are, to be sure, consistently inconsistent.

I have partly programmed the experiment in HyperCard over this weekend.
Perhaps it can be run before you return from Welsh Wales. Maybe we can
have a phenomenon for you to explain when you devise your model for the
experiment you think is not a step on the road to PCT damnation.

Martin