[From Bill Powers (950802.1400 MDT)]
Bruce Abbott (950802.1130 EST) --
We're of a single mind here. I think I can agree that in the Motheral
data, your explanation of the difference between the curves at different
deprivations makes good sense. The higher system is simply operating at
different amounts of error, which it can't correct because of the
external (experimenter-run) stronger control loop.
In going over the equations some more, I've realized that my first
conclusion was probably too restrictive. The pressing rate isn't
_necessarily_ constant; it just seems to be approximately constant in
this particular case where we fit a straight line either to behavior vs
reinforcement, or time per reinforcement vs. ratio. There are enough
departures from a straight line to suggest a droop on the right in the
actual data, and it is certainly possible to vary the collection time in
the equations and generate very different curves of pressing rate versus
reinforcement rate. This would also result in a different curve fitting
the behavior rate vs reinforcement rate data. I think what we need to do
is vary the collection time to find the best fit of the curves to the
data. Your plots of m versus time per reinforcement are not very
sensitive to changes in collection time -- when you plot pressing rate
against reinforcement rate (using your deduced collection times), you
get much larger changes in the pressing rate than are evident in your
interval calculations. I'll give this a try. Actually, Cyclic2.pas is
set up to show the goodness of fit; maybe all I have to do is change
which variable is being fit to which. Or maybe it's already showing the
right variables. We'll see.
Your scenario concerning variations in two behaviors is sensible. What
still concerns me is the lack of control at the lower levels. How could
behavior vary with a change in higher-level error signal, yet not change
with a change in received pellets? It still seems likely to me that if
the value of the rewards were raised, there would come a point where we
would get a lower pressing rate with the same reference level for food
input.
Before we move to publication, however, I want to take a careful
look at other results from studies in which ratio schedule
parameters were manipulated to confirm that the analysis is not
limited to one type of schedule manipulation (e.g., cyclic-ratio)
or the particular geometry of one operant chamber. And I would
like to further develop our understanding of what is going on in
these experiments, as we still seem to have a number of questions
for which our answers are still speculative. In the end, I want a
generative model that does what the real rat does.
I concur.
···
-----------------------------------------------------------------------
Joel Judd (950802.1425 CST)--
Your're on a roel, Joel. Great poem in the last post, and a nice
suggestion in the present one. How about writing a letter to TIME saying
just what you said? If you make it sound confident enough, maybe TIME or
someone else will wonder what PCT is.
Just be sure that sticking to your principles doesn't lose you your job.
I have enough on my conscience.
-----------------------------------------------------------------------
Best to all,
Bill P.