Tom Bourbon [950120.1008]
FROM CHUCK TUCKER 950116
. . .
These comments from Rick, Bill and Tom also indicate
one of my frequent complaints about the previous
PCT research and a "call" for more attention to such
matters in future research. In particular, there are
so many actions that can act as disturbances in these
studies that we must attempt to come up with a standard
protocal that should be used in such studies (along with
the rules for guiding PCT research).
Chuck, the number of possible *sources* of disturbance in one simple tracking
task is *infinite*. There is no way to control for all of them. The idea
that one can attain that kind of control over "extraneous variables" is one
of the silly aspects of traditional behavioral science. In fact, the idea
that you *need* to achieve it is a fallacy. An amazing thing about control,
the phenomenon, is that, within limits that are often quite broad, an
organism controls *in spite of* sometimes pronounced variability in the
world. (Don't forget, the laboratory is as just as much a part of the "real
world" as is Main Street in your home town.)
I have watched hundreds of people run tracking tasks, sometimes alone,
sometimes in small groups of from two to four. I have watched people
control by using two hands on two joysticks, or one hand on a mouse, or one
hand on a mouse and one on a joystick, or one or two josticks strapped to
their forearms or their lower legs, or a stick attached to their back -- the
list is long. Most of those people were "normal" university students, the
others included people with head or spinal injuries, people who walked past
a poster-demo at a psychological meeting, and PCTers at CSG meetings.
Across those many runs, there has been immense variability in environmental
conditions and in the physical conditions of the participants. I have seen
nothing to suggest I *must* use a standard procedure, or standardized
conditions -- or standardized participants, for that matter. Control is an
extraordinarily "robust" phenomenon, to borrow a term from traditional
behavioral science. Control is what life is all about, and it happens all
the time in the messy, variable world of Boss Reality.
I knew but it was
not clearly stated in Rick's research that he was one
of the "subjects" nor was it always stated that the
research was done by person who are "well practiced"
at these tasks. I don't believe for a minute that Rick
or Bill or Tom or whoever was trying to deceive anyone
it was simply a matter of common understanding for them
that was being done.
What do you mean? When I was the subject, I said so. I always said wheter
my participants had practiced a lot, or a little, or not at all. One point
of my little paper on "the accuracy and reliability of predictions by CST"
was that, with a little practice (and just a smidgen of common sense in the
selection of participants), any one subject is as good as any other in a
demonstration of how well the model predicts performance.
Deception? What are you talking about? I really don't understand.
Do you remember how, ever since we started meeting as CSG, there have been
computers set up at the meetings, running all of our demonstrations and
control tasks? Anyone who wants can always come and run the tasks, but some
people *never* avail themselves of that opportunity. If a few members of CSG
had taken a little time at each meeting to run some of the "boring" tracking
tasks, each of them would have a gut feeling for just how robust -- reliable,
immune to variations in trivial features of the environment -- control
really is.
One again, I suggest that everyone who wants to understand the differences
between research methods in traditional behavioral science, and the ones we
use to study control, should read Phil Runkells wonderful book, _Casting
nets and testing specimens_.
This is the same for distance from
the screen, type of mouse or joystick, type of surface
for mouse, and all the other aspects of the research the
"controlled" when doing the research.
Only if we are concerned about certain kinds of measures, Chuck. Most of
the time, those things are unimportant. Utterly unimportant. Think of what
you just said, only imagine you were talking about the conditions for a
study in which you want to learn if students in your classes can drive
from your campus to a city somewhere in Ohio, without experiencing a
serious accident or an encounter with officers of the law. The most
important thing you would need to know or to "control" as the experimenter
would be whether your people can drive "fairly well." For those who can,
the probability they will succeed is greater that .9999999. Within *very
broad limits* they will succeed no matter the condition of the road, the
position of the driver's seat, the outside temperature, which kind of
material covers the steering wheel, the exterior and interior colors of the
car, the number of other drivers on the road, the make and model of the
car, the pressures in the tires. And so on, through an infinitely long list
of variables, some of which *would* or might act as disturbances to the
participant's, but most of which would not matter one farthing. For those
that do disturb, the participant's will control right through all of them.
Of course, I have
always called for a standard set of instructions be
given to the "subjects" but I have never been successful
at this request.
You certainly have repeated that call, and you certainly haven't been
successful. You aer a persistent kind of guy! 
Except for a specific
project where we are looking for some small or subtle differences in
performance, there is no need for standardized instructions. "Drive from
here to X, Ohio. Don't have a wreck and don't get in trouble with the law."
That ought to do it. We could probably get by with saying, "Drive from here
to X, Ohio," and leave the rest to the students. (I am sure that, to
satisfy the members of the human subjects' committee, all of whom believe in
the infinite list of variables that will "cause" your subjects to go astray,
we will need more complex instructions!)
All in all the exchanges about these
exercises have been helpful to me and I hope that future
research will be better for them.
Good. That's what it is all about.
You have taken the huge first steps, Chuck. Unlike most people who enjoy
debating and speculating about PCT, you are doing some of the grunt work.
You are asking large numbers of people to do some controlling. Already, you
have begun to notice some intersting facts about the phenomenon of control,
facts that are unknown to many of the debaters and speculators. I believe
if you stay with it for while and let some of your people gain even a
little experience with the tasks, you will discover that many of the
"problems" of traditional experimental design and experimental control (by
the experimenter) simply disappear.
Keep on having fun. 
Later,
Tom
Except for a specific