Cat scientists; Cultural background

[From Bill Powers (951122.1200 MST)]

Martin Taylor 951122 13:15 --

     Back to Jack-the-cat-in-the-box scientist. Jack has discovered
     that if he walks around the back of the box and then comes forward,
     lying down facing the clock, the door opens. There is a
     correlation of 0.78 between starting the walk with the left foot
     rather than the right and the door opening (i.e. it doesn't open so
     reliably if he starts with the right foot, and he has to go start
     the walk again in order to make the door open).

This whole post is a neat example of empirical research. Just keep
trying things that will get the correlation to go higher, and eventually
it will be as high as it's going to get. What you end up knowing is that
if you do behaviors A, B, ... N, the result you want will happen, most
likely. And that's all you know. If someone moved the stick three inches
in the right direction, you'd have to start over. Also, you end up going
through a lot of motions that are irrelevant and unnecessary, but you
don't dare leave anything out because you haven't the vaguest idea of
why you have to do these things. Once you have those standard
experimental conditions set up, and are getting results, you can't
afford to change any of them; you know what's sufficient, but you don't
know what's necessary. This is about how I would expect a cat to
proceed.

Jack-the-cat doesn't actually end up with a model-type theory. He ends
up with a theory in the other sense: repeating what has usually worked
before. That kind of theory is truly a "summary of observations."

But suppose Jack-the-human got a little more systematic about this. When
the door falls open, Jack doesn't just rush out to get the goodies. He
stops and wonders "What was I doing just before the door fell open? And
what happened as a result?" He checks out each part of the action to see
how much it can be varied without losing the door-opening. He tries
starting from different places, eliminating variations that have no
effect on the result. He looks at his relationships to other objects, at
sequences of events; he reasons logically. Finally, even though he is
spending more time studying the situation than actually escaping, he
boils his knowledge down to the essential element: the stick must move.
As soon as he understands that, he can open the box every time from any
starting position and using any handy way of moving the stick. His
actions have ceased to be relevant; all that matters is that a certain
consequence of acting be brought about, in any way that will do it.

Later, out of curiosity, Jack may wonder how moving the stick makes the
door fall open. He starts to guess: maybe there's a string attached to
the stick that runs over the top of the box to the latch and pulls on
it. Examining the latch, he sees no string, but he does see a rod coming
out of the latch and running to the floor, where it disappears through a
hole. Obviously, he reasons, the rod is part of a linkage that runs
under the floor to a place just below the stick when it is centered. He
looks at the stick, and sure enough there's a bit of rod sticking up
through a hole, pressing up against the end of the stick. Moving the
stick lets the end of the rod spring up a little, whereupon the door
opens. By holding the rod down with one finger and then moving the
stick, Jack verifies that it't not the movement of the stick, but its
effect on the rod that opens the door.

AHA. Jack walks over to a place halfway between the stick and the door,
jumps into the air, and comes down with both feet on the floor as hard
as he can, joggling the rod which he imagines to run under the floor.
The door falls open! Now Jack not only has a pretty good mental model of
the link between the stick and the door, but he has tested this model by
using a new operation he had never thought of using before, disturbing
part of the model he has never actually seen, and getting the predicted
result. He hasn't proven that the model is right, but he's starting to
pile up the evidence.

However, when Jack wants to get out of the box, he reaches out from
wherever he is and pushes the stick aside. That's much the easiest way.
All the rest was just to satisfy his desire to know how the thing works.

Back to Jack-the-cat:

     It works, every time, so he knows the theory is correct and
     complete. Or does he?

No, he doesn't. He doesn't have a theory at all. What he has is
_experience_. He knows, from past experience, what to do. He has no idea
at all of why he has to do that.

···

-----------------------------------------------------------------------
Chris Cherpas (951122.0935 PT) --

     I propose for your consideration that Bill P. would not have
     formulated PCT if he had grown up in a culture where there was no
     "behaviorist revolution."

You underestimate my age and misunderstand my background. I was born in
1926, meaning that I was already in school by the time Skinner started
his work. I was fascinated by science, but in my world that meant
physics, not psychology. I didn't learn anything about psychology until
my sophomore year of college, in 1947, and I thought that what they had
to teach about human nature was crap (a Spence-Hull department). My
major was in physics, mathematics, and astronomy. I learned about
control systems from the U.S. Navy, 1944-46, and through building
control systems as a medical physicist in the 1950s. The basic ideas of
PCT were developed from 1953 to 1960, without any association with
psychology at all, except through the late Bob MacFarland who
participated in the project to some extent. My main partner was Bob
Clark, another physicist.

Actually, it would have made little difference to me whether the
behaviorist revolution occurred or not. I didn't think much of the
alternatives, either.

     The first order of business for scientific psychology was to give
     up the ghost, the god/homunculus inside.

That is where the baby went down the drain with the bathwater. In trying
to get free of mystical and religious concepts, scientific psychologists
decided to ignore the problem of how the system works, and thought they
could reach understanding strictly by manipulating externally visible
phenomena. Well over half of this century went by before some behavioral
scientists finally wised up and realized that they HAD to make models of
internal processes, and to start wondering about subjective and mental
phenomena that the old guys had dismissed as imaginary -- for example,
imagination.

     Evolution, as developed by Darwin, didn't require a model of
     something you can't see (except history -- not unlike that "history
     of reinforcement" we're always hearing about).

How about "genes?"
-----------------------------------------------------------------------
Best to all,

Bill P.