[From Rick Marken (940616.1400)]
Martin Taylor (940616 11:00) --
I imagine that you are correct in general terms in how the model would
work, but I'm not clear in detail about what the model would be, in its
linkages between the higher and lower control systems.
I really have no idea how the model will be set up either; until I see the
data. I also prefer a "one step at a time" approach to research. The study
should be as simple as possible and be based on as much as we already know
about how the control model works. It seems to me that the only thing we
should look at in this first study is how one output df is used to control
several perceptual df. The goal is to get a good model of what seems to me to
be the simplest case of output "time sharing".
Be that as it may, my variant on the experiment is to have two different
conditions, one that should show an alerting effect and one that should not.
.
.
.
That's condition 1--no alerting.
Yes. Your condition 1 is the experiment I suggested. This is already a pretty
complicated study in the sense that it could produce data that might be hard
to model. Hopefully, there will be some regularity in the way the subject
moves the mouse cursor from one cursor to another that will provide hints
about how to design the output time sharing model.
In condition 2, everything is the same, except ... If the cursor for a strip
crosses a distance dn going away from the target, emit a string of n beeps
at the pitch associated with that strip.
The question of the experiment is whether there is any difference between
conditions in the likelihood of large excursions.
This kind of experiment is the perfect companion to masturbatory modelling;
voyeuristic experimentation. Where masturbatory modelling is modelling that
is done without any real data to contrain it; voyeuristic experimentation is
data collected without any model telling what kind of data to expect.
Voyeuristic experimentation is exactly the kind of experimentation that is
done by most conventional research psychology.
It is possible that you will find a statistically significant difference
between conditions 1 and 2, especially if you average over subjects. But you
won't know why you got this result. And you won't know why some of your
subjects do worse in condition 2 than in condition 1, if (as is likely) some
do . In fact, you won't know much more after this experiment than that there
is statistically better performance in condition 2 than in condition 1.
Welcome to conventional psychology!
I anticipate that there will be differences, more so if N (the number of
strips) is large, and less so if dn is so small that there are differently
pitched beeps going off all over the place. The optimum value of dn should
increase as N increases, but I don't intuit a reason for the form of that
relationship.
This is conventional psychology at its purist: qualitative assertions based
on extrapolation rather than modelling. Why is all this anticipated? Is
there a working model that actually controls more poorly as N increases?
I have a suspicion that these predictions are not derived from the behavior
of a working model; it sounds to me like they are generalizations from
previous findings. What if there is no "significant" difference between
conditions 1 and 2? No problem, right? We can always say that we just forgot
to take into account some other "variable".
I wrote a textbook on this kind of research, Martin. I know how the
conventional game is played. We can stop playing it now;-) Now we know that
behavior is the control of perception -- so let's find out what people
control, how they control it and why. Bill P. didn't put that chapter about
"The Test" into B:CP just for the heck of it. Try to think of those
operators, who are getting "alerted" when variables go out of limits, as
perceptual control systems. It's an amazing new perspective 
I don't think that one can say that condition 2 is one in which the subject
controls for a different perception (no beep), because the beep occurs only
in imagination until one of the spatial controlled perceptions gets too
far out of control, and when it has occurred, it will not recur unless
the spatial variable's error is reduced in the interim across the d1
threshold.
What the subject controls in condition 2 can only be established by TESTING
FOR THE CONTROLLED VARIABLE. Remember, THAT'S how we do research on control.
The experiment you have proposed is an SR experiment, basically saying "let's
see if these "alerting" stimuli improve control"; it is just being described
with a lot of PCT verbiage. This is precisely the kind of research that I
seem to recall coming out of the Carver, Scheier et al "control theory" camp.
Does it look familiar, Jeff? This is NOT PCT research. PCT research is not
aimed at determining whether something has an effect on performance. PCT
research is aimed at discovering (all together now):
what people control
how they control it
why they control it.
Your research proposal can be turned into PCT research but we have to learn a
lot more about how people "time share" an output in order to control several
perceptions - - and have a good working model of how this happens -- before
we start playing around with variations.
I have never, so far as I am aware, talked about perceiving a disturbance.
I have always talked about the effect of the disturbance on the perception
of a CEV. And that, one does perceive, to the extent that it is not
countered by the effect of the output of the control system.
I'm sorry to be so picky. I know that you don't talk about perceiving the
disturbance; but you do talk about perceiving the effect of the distrubance
(as you note). But even this is not really true for the control system
that is controlling the disturbed perception. For example, when you are
doing a compensatory tracking task you cannot tell (except by monitoring
another perception -- of the output) the extent to which any change in
the position of the cursor is an effect of the disturbance or an effect of
your own output. So although the effect of the disturbance is all that the
system controlling the cursor could possibly perceive, the system cannot
"know" that the effect of the disturbance (alone) is what it is perceiving at
any instant. Again, this is simply because, at all times, p = d + o. All the
system sees is variance in p; unless it has some OTHER way to know (via
another control system) that o is constant for some period of time, it has no
way of knowing whether the variance in p that is being perceived is an effect
of d or o or both :-).
Best
Rick