Thorndike's cats; reorganization

[From Bruce Abbott (951115.1530 EST)]

Bruce Abbott (951110.2030 EST) --

Bill Powers (951110.0900 MST) to Martin Taylor

By the way, I am entirely pleased that you and Martin are having this
discussion. It's one I have tried to get going before, but we just couldn't
seem to get past Thorndike's interpretation and focus on what I think are
important lessions to be learned about reorganization from his basic
observations.

I hope my comments didn't "kill the conversation." I was looking forward to
a reply; not only didn't I get one, but with that post the whole thread
seems to have disappeared. I'd particularly like to get a response to this
assertion:

In fact, in his definition of the "satisfying state of affairs," Thorndike
came close to specifying the Test for the controlled variable:

   By a satisfying state of affairs is meant one which the animal does nothing
   to avoid, often doing such things as attain and preserve it.
      Thorndike (1914, p. 245)

A satisfying state of affairs, then, is some (perceptual) state the animal
will do nothing to avoid (i.e., no error = no action), and does things to
attain and preserve [i.e., error due to momentary disturbance (attain) or
continuing disturbance (preserve) = action]. Being out of the box will work
in this definition just as well as having access to milk; all that is required
is that the indicated relationship be demonstrated (i.e., pass the Test).

So, a "satisfying state of affairs" is a state in which some preception,
previously in error, has been brought near its reference value.

It is worth noting that a perception can HAVE a reference value even though
it is currently not under control. Such a perception would be better
described as "out of control" rather than "uncontrolled." But control is a
continuum, and "out of control" is merely the end of that continuum, near
"poorly controlled." It seems to me that reorganization may be a process
operating on any system for which there is a reference for a perception and
yet control is "not good enough" by some criterion. That would include the
extreme case where the perception is completely "out of control" as well as
the less extreme cases where the perception is being controlled but not
"well enough." In this view, reorganization as a process is going on all
the time, and not just when persistent error appears in an intrinsic
variable. When it is not producing "new solutions," it is fine-tuning the
old ones.

Regards,

Bruce