[From Rick Marken (940603.0830)]
Richard Thurman (940602.1300) --
No offense to anyone on CSG-L but I see a whole lot of talk but very little
results. So . . . I can generate very little interest.
I think this is a very good point. It jogs many thoughts. The main one being
"what can one do"? What results would generate interest? It's hard to tell.
I've been working on PCT for 12+ years, I have a book full of results and
files full of unpublished or uncompleted studies. Tom B. and Bill P. have
also produced an enormous body of results -- an amazing amount considering
the fact that they had to do all this work on the side, had no graduate
students to help, etc. The only other "bodies" of PCT results I know of come
from people like Carver and Scheier -- who have produced nothing useful that
I know of. I think this kind of pseudo PCT research is probably less valuable
than no research at all. So what do we do? PCT is in a bit of a bind. It's
difficult to get a PCT research program going (supported by grants and grad
students) because PCT is rejected by the establishment -- the people who give
the grants. So the only people who are doing PCT research are the ones doing
it in their back rooms (like me), and we can't do much on our own -- or have
it noticed by colleagues.
The only hope, as I see it, is to increase the number of people who are doing
PCT research -- real, QUALITY PCT research -- especially the number doing it
in high profile positions in research institutes and universities. That is
why discussion in CSG-L (even though it is all talk) is so attractive to me.
This is the only forum where PCT is discussed in rigorous detail. It is the
only "self-correcting" medium for disseminating PCT ideas (Carver and
Scheier, for example, have decided to go open loop -- never attending CSG
meetings where the very theory they are promulgating is discussed -- so they
just keep diverging more and more from what PCT is about). My hope is the
one or two people listening in (like you?) will start to do some of the basic
experiments and, then, start doing variants and, then, hopefully, produce
some "interesting" research results.
I supose an alternative is for me to go off, stop babbling on CSG-L and just
start doing more research. But for me, it's a question of allocation of
resources. I only have so much time that I can steal away for PCT work. What
is the most effective way to spend that time in order to achieve the goal of
promulgating an interest in and an understanding of PCT ? For twelve years I
spent that time developing studies and demos and publishing them if possible;
the results of this effort, in terms of getting people interested in PCT,
were negligible. I'm still doing research but I would now like to try to
spend my spare time writing in a fairly popular style about PCT -- that's
what I am currently investing my time in. Whether or not it will work, I
don't know. But I also think it is useful to spend my time discussing PCT on
the net. We have apparently gained some new and talented PCT afficionados
via this medium (like you and Martin Taylor and his crew); I think it's
important to keep clarifying the basic PCT points on the net -- so that we
can keep progressing towards an understanding of the control of perception
and not diverge off on a tangent, like that taken by Carver and Scheier, that
leads us back to the dead end of cause-effect modesl of living systems.
So I, personally, think all the talk is worth it -- but then, I think I've
paid (and am still paying) a lot of my PCT research dues so I feel pretty
guiltless about spending my time now "policing" CSG-L.
I'm toying with the idea of putting together a seminar/workshop for
a VR conference in December. I figure using about 4 or 5 hours to
go over your Byte articles. That's about all they (no, I) could handle.
I'll call it "Building Better 'Bots: How To Create Autonomous Virtual
Agents"
Anyone on CSG-L interested in working this angle?
Count me IN for sure. If the program (on advanced HCI technology) that I am
currently supporting holds up there may be a way to incorporate such a
project into my "real" work.
Martin Taylor (940603 09:30) --
Don't confuse the uncertainty used by the analyst with the uncertainty
currently under discussion when we look into Paul Revere's head.
I thought of a "control of uncertainty" study. A signal (S) light appears
(say every second) on the computer screnn and then another light (A) appears
below it and to the right of left with some probability. Eg.
S
A1 A2
So after each S, either A1 or A2 appears. The uncertainty is determined by
the probaility of A1, p(A1), which is 1-p(A2). So if p(A1)=1.0 then there is
complete certainty that A1 appears after S; intermediate values of p(A1)
result in intermediate values of uncertainty about whether A1 or A2 will
appear after each S. Max uncertainty is when p(A1) = .5. The subject controls
the uncertainty by moving the mouse (m) which changes the value of p(A1). A
distrubance (d) is also applied to p(A1) so p(A1) = d + m (with the
constraint that p(A1) be in the range from 0 to 1.0).
Would this qualify as a control if "uncertainty" experiment? If so, then we
could actually do this and see 1) if people can control the uncertainty (ie.
keep p(A1) at some reference level) and 2) what aspect of the display they
are actually controlling (I can think of a couple of possible alternative
perceptual functions that might be tried). Whaddaya think? How about sicking
a research associate on this?
Best
Rick