From Bill Powers (950728.0825 MDT)]
Dick Robertson (950726.1833CDT) --
Sorry you couldn't make the meeting this year -- many people asked if
you were coming. Next year? (July 17 - 21).
It looks as though most of the vote is with you and against splitting
the list.
···
-----------------------------------------------------------------------
Erling Jorgensen (950726.2230) --
Your observations of levels of perception are just what I have hoped
people would try to do. They're right out there in the world; all you
have to do is pay attention. Looking is far more important than applying
particular labels. If more people were looking, perhaps we could start
improving on that tired old list that I offered.
----------------------------------------------------------------------
Marc Abrams (072795 1100 EST) --
I understand the points you are making about splitting the list. As you
can see, the sentiment seems to be running against that idea.
I think your suggestion was reasonable -- actually, it comes up from
time to time. However, we've always decided against it in the past, in
part because of our experience at CSG meetings where all sessions are
plenary and everyone listens to everyone else and tries to understand.
Of course applications people don't grasp everything the modelers say,
and vice versa, but they try. In the trying, both segments of the group
learn. The better the applications people understand the basic
principles of PCT, the better they will know what they're doing out in
the field where there is nobody to answer questions. And the better the
theoretical types understand what happens out in the field when
practical people try to apply PCT, the better they will see the snags
and loopholes in what they have developed. Everybody benefits.
I'm not terribly interested in increasing the body count. What matters
the most is _understanding_. People who look in on the list and don't
even try to understand aren't of much use to us. It's really up to the
people on the list who are developing an understanding to carry what
they know into the larger world. There's a strong tendency in our
dillettante anti-science world for people to pick up a little vocabulary
and then rush out to proclaim themselves experts just because they can
say the right words. That is the road to cultism. The only antidote I
know of is to avoid intellectual snobbery, either forward or reverse. We
are here to help each other understand PCT better.
The problem with bringing "other methods" into PCT is that the other
methods are certain to involve traditional concepts of what people are
and how they work. We have had very bad luck with this approach. It's
much better for people who understand PCT as well as some other field to
work out the applications themselves. We don't have a library of books
and papers that others can just look up; most of the development of PCT
is going on right here, on the net. The real leaders of PCT for the
future are the people who are listening and talking -- but mostly
listening -- on CSG-L right now, learning as much as they can so they
can carry on the work when all of us geezers are gone.
One point: the List is not like a conversation. Some people hestitate to
start a new thread because it seems like interrupting the active
writers. But there's no way to interrupt on the net. You can read what
you want to read, and write what you want to write without stepping on
anyone else's discussion. If sociologists, clinical psychologists, or
management consultants want to start a new topic, all they have to do is
start it and see who replies. Nobody is under any compulsion to comment
on posts that aren't interesting to them, and there is no rule here
about starting a new topic. As far as I'm concerned, there's no social
hierarchy here, either. If the Old Man says something foolish, he
invites correction just like anybody else. If you want to be cowed by
your own concept of who deserves to be heard more than somebody else,
that's your own fault (not YOU, Marc, I mean anybody).
----------------------------------------------------------------------
Bruce Abbott (950727.1220 EST) --
We seem to be on the same course. I'm working on a new program to
display the relationships in our new representation of the environment,
and will post it pretty soon.
One thing I've found already. The "constancy" of the deduced within-
ratio pressing rate is EXTREMELY sensitive to the assumed collection
time. I think the method of deducing the collection time by plotting
ratios against time per reinforcement is suspect, particularly because
the curves are NOT constant, but show a definite downward curvature on
the right side where it counts. This curvature increases very rapidly as
assumed collection time is shortened. When we do our own rat
experiments, we are going to have to get direct measurements of
collection time -- that's the only reliable method.
It's good that you have brought this subject up; it has a large effect
on the way the model fits the data. From now on we'll have to include
collection time in all models of this kind of behavior; it can't be
ignored.
A two-level model is definitely appropriate, as you suggest. I think,
however, that for the time being we can fake it: just say that when a
reward occurs there is a delay while it is collected. We could probably
kludge a model ad hoc, but I'd rather wait until we have some
alternative ideas that we could test experimentally. There's no point in
just hammering together an arbitrary model that will reproduce the data
-- that's always possible. What we want is a model that conforms to some
principle that we can then test.
Hey, not a word in there about reinforcement. I'm starting to
sound like a control theorist! (About time, huh?) (;->
I goggle in admiration. But you've been a control theorist longer than
you've sounded like one.
-----------------------------------------------------------------------
Best,
Bill P.