[From Bruce Abbott (951207.1810 EST)]
Bill Powers (951206.0655 MST) --
Note that the effect of the pressing system on the eating system, the
switch that you drew, is effected through the appearance or non-
appearance of food in the dish. The eating system reduces the amount of
food left, the pressing system, via the apparatus, increases the amount
left. For schedules that only provide access rather than pellets, the
eating rate determines how much is eaten, and food in dish goes to zero
when access is terminated.
This is a nice model although, as you know, "eating" actually consists of
several subsidiary systems not diagramed. I would like to ask about the
words you have used to label these two systems. In the past I have
experienced some difficulty coming up with appropriate labels for systems
controlling various perceptions (I received a particularly stinging
criticism from Dag on two of them). The problem is that I wanted to talk
about these systems in terms of what they DO, but when I tried to use labels
like the "pressing" system, I was informed rather pointedly that control
systems do not control their outputs, they control their inputs, as if using
such functional labels indicated a confusion on my part about this point.
When I tried to substitute something that would emphasize what perception
was being controlled, it sounded awkward and unsatisfactory.
Now, the labels you have applied to the systems in your diagram seem to
denote behaviors (eating, pressing) that accomplish specific goals (food
eaten, lever depressed) rather than controlled variables. This is O.K.? I
need a little guidance here, because it apparently wasn't O.K. when I used
similar labels, if I understood the criticism correctly. This is the way
Skinner defined operants--not in terms of what movements were made, but in
terms of what those actions accomplished. Minsky used similar labels to
denote his various "agents": e.g., finder, stacker, builder.
Regards,
Bruce