[From Bill Powers (950115.1815 MST)]
Bruce Abbott (950115.1630 EST) --
But speaking of autism, here in the department we have an associate
faculty member and full-time clinician with whom I was conversing a
few days ago about (what else?) PCT. I mentioned that it would be
interesting to conduct a study on autistic children to determine
how their control systems differ from those of more normal
children, and suggested that the differences most likely involve
fairly high levels in the heirarchy. It turns out that the guy
works with autistic children in his practice and that there are
apparently defects in the low-level control systems as well. He
became very interested in what I had to say and asked about whether
our library has a copy of B:CP. I lent him my copy.
There may be a very nice opportunity here for some research, which
would apply the Test to determine what perceptions young autistic
children control as compared to non-autistic kids of the same age,
and how well they are able to do it.
As Tom Bourbon has discovered, even people with terribly discoordinated
motor behavior can in fact do tracking tasks in a recognizeable way --
you should probably get your friend in touch with Tom directly, to
discuss his observations and how he would go about setting up an
experiment.
IF you can get autistic people to do the tracking task at all (1 cursor,
of course), I would be electrified. My impression (uninformed) is that
it's hard to get them to pay attention to any task. If you could get
them to do a tracking task, or any kind of instrumented control task,
you would know what they are perceiving and you would know what their
intention is. These tracking experiments are awfully simple, but when
you think about them, they provide a NONVERBAL window into at least one
little aspect of a person's internal experiences, a window that with
patience and persistence could probably be opened wider. If I were to
watch an autistic person doing a control task, I would feel that I was
sharing an experience with that person, possibly the first kind of
contact I would ever have had past that baffling exterior.
I hope we can all follow through on this. It may be another frustrating
dead end, but considering the plight of autistic people, it's worth a
good try. One can easily understand how the facilitated-communication
people let their hearts lead their heads. Of course we would have to be
extra-careful about that.
···
-----------------------------
On THREECV1: Is there any particular reason why the participant is
giventhe opportunity to see and then accept or reject each set of
stored disturbances?
I have meant to comment on that at least three times and forgot it each
time. The only reason for displaying the disturbances is to let the
experimenter reject those that are uninteresting. You don't have to do
this selecting. In similar experiments in the past, we have actually
calculated correlations and made the program pick sets of disturbances
that correlate less than some criterion amount with each other, like
0.2. Once in a while you can get disturbance patterns that resemble each
other so much that the program's "mind-reading" (TM, Rick Marken)
feature makes a mistake. If the disturbances don't show much correlation
with each other, the automatic application of the Test to select the
controlled variable works pretty much every time. But you can usually
get way with any three randomly-generated disturbances.
Incidentally, I hope you noticed the variable "slow", which sets the
amount of smoothing applied when the disturbance tables are generated. A
value of about 0.001 gives a very easy disturbance, and 0.02 a pretty
hard one to counteract. So you can see how well the model does when the
person is having trouble tracking.
-----------------------------------------------------------------------
Best to all,
Bill P.