Bad cat

From Bill Powers (931021.0830 MDT)]

Oded Maler (931020) --

How can I trigger the right reorganization [of the misbehaving
cat], with a minimal number of used sheets?

You can't trigger reorganization directly, and since
reorganization is (we suppose) random, you can't trigger the
"right" one, either. Reorganization is the ability of an organism
to vary its own organization at random until it attains control
of a variable that matters to it. The only way to get a specific
behavior from the cat is to arrange for a variable that matters
to it to depend on its producing that behavior. If doing so
doesn't create any other conditions that constitute errors for
the cat, the cat will reorganize until it finds the required
means of control -- that is, the behavior that is effective. If
you're cleverer than the cat, that will be the behavior you want
to see.

B. F. Skinner, the famous animal trainer, discovered some basic
techniques for doing this, although his interpretations of what
was going on were quite different from those generated by PCT.
The basic process was to establish an environmental
"contingency" that made an input which the animal wanted to
maintain or avoid depend on a specific behavior that the
experimenter wanted to see or not see. A contingency can be
established by an automatic apparatus or by a person who adheres
faithfully to a rule (watching for the desired behavior and
acting according to the rule, reliably, when the behavior
appears). The rest of the process is up to the animal's ability
to discover how to produce one of the behaviors that will restore
its control.

Skinner found that contingencies that encourage NOT producing a
given behavior are effective, but the new behavior that results
is unpredictable. If you don't care what the new behavior is,
aversive contingencies, if not too severe, work fine. The animal
reorganizes away from that behavior in a random direction. If you
arrange for a certain behavior to produce punishing results
(experiences for which the organism has an intrinsic reference
level of zero), the exact behavior that produces those results
will be abandoned. However, the behavior that takes its place
could be just as undesirable to the animal trainer. If you set up
an electrical grid on the bed that delivers a shock when pissed
upon, the cat will quickly reorganize to stop shocking itself and
start pissing somewhere else, perhaps on a different bed or on
the sofa. The likelihood that it will start pissing in the
sandbox is very low. The problem with using aversive
contingencies is that reorganization will stop with the first new
organization of behavior that succeeds in eliminating that
particular disliked outcome, and there is no predicting what the
new behavior will be. Reorganization is random.

Your goal is not to stop the cat from pissing altogether (that
would conflict with the cat's intrinsic needs), but to alter the
place in which it pisses. Skinner discovered that this requires,
with dim-witted animal subjects, taking very small steps indeed.
He called the process "shaping."

The basic principle of so-called "operant conditioning" is to
make something the organism wants to perceive dependent on
producing a behavior that the experimenter wants to perceive. You
could, for example, make the availability of food (when the cat
is hungry) depend on the cat's pissing in the sandbox rather than
on the bed. But if the sandbox is in the basement and the food is
given in the kitchen, it is highly unlikely that the cat's
reorganizations (automatically produced by continued hunger
effects) will result in its rushing to the sandbox in the
basement and then to the food in the kitchen instead of pissing
on the bed. That is too large a reorganization; too many details
of the control systems would have to change randomly in the right
direction, all at once.

What has to be done, Skinner discovered, is to arrange for a
_small_ change in behavioral organization to restore control of
the variable that the organism wants to maintain in a certain
condition. If only a small change in the behavior is required,
the new behavior that will restore control will be in the
neighborhood of the old behavior and small random changes in
organization will quickly find it. Then the contingency can be
altered again by a small amount, and the behavior will change
again by a small amount, and so forth. In this way the organism
can be led through large changes in behavioral organization, but
without requiring drastic reorganizations at any stage and
without any cruelty (if that constitutes an error for the

You could, for example, move the sandbox onto the bed, and
withhold food until the cat pisses a little closer to the
sandbox. Cats like to eat in small bouts, so it shouldn't take
too long to get the cat to piss (or at least stand) in the
sandbox on the bed. Then you start moving the sandbox until it's
where you want it to be permanently, moving as small a distance
at a time as necessary to let the cat continue feeding itself

Of course this takes a lot of time and attention; you have to be
serious about where you want the cat to piss and you have to be
consistent about the rule you apply. If this process is too slow
for you, and you feel very strongly about the cat's pissing on
the bed, put the cat to death. That will solve your problem for
sure. I mention this to remind you of the other considerations of
cat-ownership that may come into play.

Of course there is another aspect to all this, as cat-owners
know. There is the question of the cat's purpose in pissing on
the bed. Pissing or defecating in outrageously wrong places is,
according to cat-owner's folklore, one of the standard means that
cats use to indicate strong dissatisfaction with something about
the current situation. A normal cat will always piss in the
sandbox if it can't get outdoors. But a cat that is being
violated in some way that matters to it, it seems, will retaliate
by doing something nasty in return. It may be that YOU are being
operantly conditioned by the CAT.

So you must also consider the Test for the Controlled Variable:
if the unwanted behavior is a means of controlling for something
else, perhaps you can find out what the cat wants that it is not
getting, or that it is getting that it does not want. There will
undoubtedly be other indications: cats go around complaining a
lot when things aren't right; it's much like dealing with a baby.
The means of communication (with humans at least) is so crude
that all one can do is keep trying different things until the
complaining stops. In other words, let the cat shape your
behavior, too. If the cat seems to want attention, give the cat
some attention. Don't say "Oh, he just wants some attention." If
giving attention stops the complaining, then something about the
way you give attention provides perceptions that the cat wants or

Perhaps the cat pisses on the bed as a way of telling you that
there is something wrong with your (or someone's) relationship
with the cat. The way to find out is to find out what the cat is
complaining about and fix it. If that stops the pissing on the
bed, then you have eliminated the error that the cat had and you
will have solved the problem. If you establish a congenial
relationship with your cat and the wrongful pissing doesn't stop,
then the pissing isn't instrumental and you can assume that the
cat doesn't care, basically, where it pisses. In that case you
should be able to shape the cat's behavior so it pisses where
it's supposed to.

Finally, don't forget who is boss in your house. In our house,
there is a guest-room with twin beds; if possible, the cat would
sleep on both of them at once. The cat really, really likes those
beds. But so do the cat's hairs, and we have decided that there
shall be no cat hairs on that or any other bed. We can't (or
don't want to bother to) station ourselves permanently in that
bedroom to shape the cat into not wanting to sleep there. So we
use the obvious solution: we keep the door shut.

Periodically, we hear the cat at the closed door, complaining. We
don't yell at the cat for complaining; complaining is a basic
right in this house. We either ignore the cat, or pet it and
soothe it and let it know that at least everything ELSE is all
right. And that seems to be OK with the cat, or OK enough. After
all, into each life a little rain must fall, and cats seem to be
able to accept that without their lives being ruined. If I can
live with the fact that not every scientist in the world thinks
that PCT is wonderful, the cat can live with access to the beds
being physically prevented. And it does. After 13 years, we and
the cat have learned what each of us will tolerate.

None of this requires undue anthropomorphizing. It simply
requires seeing the cat as a living system basically organized as
we are, but with less equipment. One can, to some extent, put
oneself in the cat's place and try to get a feel for what the cat
wants and what its purposes are. Of course you have to remember
that the cat doesn't have much of a vocabulary with which to
express the wants; the system works without much symbol-
manipulation, as far as I can tell. It wants what it wants but it
doesn't necessarily "know" what it wants.

So trying to understand the cat from the cat's point of view is
excellent training for learning to perceive the nonverbal
processes that underly human experience, too. One can come to
appreciate just how much can be handled without elaborate
symbols, even at a rather high level of organization. The same
experience gives hints about what the human symbol-manipulation
capabilities add to the overall ability to control what happens,
particularly in complex or interpersonal situations. Solving a
relational problem with a cat would be SOOO much easier if we
could communicate in words! Communication wouldn't necessarily
guarantee congruence or mutual comprehensibility of goals, but at
least the cat would understand that my intention is for the cat
to stay the hell off the beds, and not to threaten the cat with
death (our cat came to us with considerable distrust of all other
living organisms, and still requires continual reassurance that
we aren't going to have it for dinner or stomp on it. That's hard
to accomplish when you can't discuss principles).

Well, thanks for the opportunity, Oded.



Bill P.