modeling; applications

[From Bill Powers (950814.0720 MDT)]

Bruce Abbott (950813.1505 EST) --

     Rat 3's reference looks a tad too low, however (noise in the
     data?). What happens if you try, say, 21 g?

Rat 3's data are pretty noisy, and something funny happens near the end:

RAT RATIO WEIGHT INTAKE MEALRATE MEALSIZE (intake/rate)
  3 1280.0 371 14.80 2.20 6.73
  3 2560.0 382 16.90 1.40 12.07
  3 5120.0 351 8.80 1.10 8.00

Somehow, at a ratio of 2560, this rat gets itself together and increases
its intake, which produces a bump to 12.07 in the meal size despite a
drop from 2.2 meals per day to 1.4 meals per day. If you stop the
analysis at ratio 1280, the correlation goes up to 0.98 from 0.93.

As to the reference signal, I just used the intercept from the
correlation program. I'll try a few other values manually, but don't
expect any improvement. Rat 3 may be an example of URS (uncooperative
rat syndrome).

···

-----------------------------------------------------------------------
Rick Marken (950813.1500) --

     Since these data were collected by reinforcemnt theorists, I wonder
     how well the reinforcement model handles these data.

I don't imagine that the issue came up. Reinforcement theory
explanations are seldom of this kind; they're mostly verbal. Bruce will
know.
-----------------------------------------------------------------------
Ed Ford (950813.late evening) --

As I tried to make clear, my questions relate to the way you describe
what is going on in your program, not to what you do. And I wasn't
complaining that your explanation isn't technical enough. Obviously you
don't want to talk in terms of reference signals and error signals and
output functions and all that. I wouldn't do so either in your position.

Here is a bare-bones description of the procedures you talk about:

If a student breaks a rule in class, the teacher asks the student what
the rule is and if the student is willing to obey the rule. If the
student does obey the rule, he or she can stay in class. If not, the
student is sent to another room until he or she does agree to abide by
the rule, submitting a plan which consists of a promise not to break the
rule again. If the student persists in breaking the rules or does not
agree to go to the other room, he or she is sent home or to the parents'
place of work, or failing that is turned over to the police.

Your explanation of how this works is that the students want to be with
their friends, and when this "privilege" is taken away, they will comply
with the rules in order to get the privilege back.

To me, this sounds exactly like the system that was in effect at the
schools I attended 60 years ago. It doesn't sound like a breakthrough
that will create a new atmosphere in schools. It doesn't sound like PCT.

So exactly what is different about your system? Haven't parents and
teachers always told children that the things they enjoy are privileges
that can and will be taken away as punishment for breaking the rules?
Haven't children always been told, "It's up to you -- obey the rules or
lose your privileges?" Isn't this is basically your way of explaining
how your system works?

Of course there are some things you say that do sound more like PCT. You
say, for example, that there's no point in trying to control children's
behavior. You have redefined discipline problems as failure to respect
the rights of others. You've said that it's important to try to get
children to think about the consequences of their behavior instead of
just reacting. You've said that it's important for children to
understand that other people live by rules and expect children to do the
same. You've said that children have a natural understanding of the idea
of rules. You've said that it's important to allow children to choose,
to improve their sense of being in control of themselves by improving
the actuality of their own control. You've changed the meaning of
detention so it becomes a program for learning social skills, taking the
punishment out of it as much as possible.

All of that recognizes the basic principles of PCT. Why, then, must you
explain your system in terms of punishing children by taking away their
privileges when they misbehave? When you say things like that, you
nullify most of the other PCT principles you're teaching. Teachers may
be trying to follow what you say and learn new ideas, but as soon as you
explain the system by saying it depends on taking privileges away, the
teachers can relax and stop trying: here is an idea they're familiar
with, and they don't have to learn all the rest of that complicated
stuff.

The result will be that from the teachers' points of view, you have
simply set up a system that is just like the old one but is more
efficient. Now the teacher is officially permitted to set up a "behave
or you're out of here" system, giving the child one warning and then, on
the second transgression, getting rid of the kid. Heavenly bliss! No
more classroom arguments and back-talk.

Of course that's not the message you want to get across; while you're
there, showing how to interact with the kids, you probably manage to get
a more PCT-ish picture across. But what happens when you leave? LeEdna
Custer remarked at the recent meeting that teachers were going right
back to their old ways, and that she planned to set up re-training
sessions much more frequently. All the teachers are remembering is two
strikes and you're out of my hair.

You and I were raised under the common system of removing privileges as
a way of making us behave. I dare say we both raised our children along
the same lines. I have less of an excuse for not applying PCT than you
did, and our parents had the perfect excuse, but the fact is that we
used a stimulus-response or operant conditioning idea (without calling
it that) and we both had in mind controlling our children's behavior.
That's how everybody raised kids then, and how they themselves were
raised. But we can't develop a new way by saying "What's good enough for
grampa is good enough for me." Or by using our own past methods as the
criterion of success.

It may be a good idea to reflect that you and I have each now lived
about half as many years as the American universal educational system
has existed. The principles used in school are not the way it has
"always been done." The existing methods are not "tried and true" --
they are experimental, and the experiment is obviously failing. PCT may
be able to turn that failure around, but only if we remember not to
bring back into the system the very concepts of human nature that have
been causing the problems wse're trying to fix.
-----------------------------------------------------------------------
Best to all,

Bill P.