[Click-l] RE: Methods of teaching behaviors

[Barry McDonald]

---"Shirle Rogers" <dhafss at bellsouth.net> wrote:

> there are several ways to teach the various exercises
> that dogs normally learn (heel, come, sit, down, stay, etc.).
> I wonder if everyone who wants to would be willing to
> briefly describe what method they are using on these
> exercises...

Hi, Shirle.
I'll start you out with some theory to toss about rather than an example of
how I would train a specific behavior. You mention Peggy Tillman. In
"Clicking With Your Dog," she identifies three distinct methods of
obtaining behavior: The Magnet Method, The Capture Method, and the Shaping
Method. To be clear, her Magnet Method is what older trainers would call
Luring or Lure Training, integrated into CT. As far as I know, "Magnet
Method" is her own term. The other two methods are commonly used CT terms.
Capture means simply waiting until the dog does what we want, and then C&T
it. (A dog is bound to Sit sooner or later!) And of course we all
understand Shaping--taking an approximation of a behavior or a tiny piece
of behavior and slowly developing it to resemble the complete behavior we
are looking for though differential reinforcement.

In "Don't Shoot the Dog," Karen Pryor describes Shaping, but does not use
the term Capture, which seems to have come later. However, I believe to
Karen at that time, Shaping and Capturing were two aspects of the same
thing. Usually, a spontaneously emitted behavior needs to be shaped, and we
cannot shape at all unless the dog emits a behavior! In Ch. 2, Karen
addresses Shaping, and has a subsection titles "Shaping Shortcuts:
Targeting, Mimicry, and Modeling." The first method is one that some of us
use formally--target sticks, stationary targets, hand targeting,
etc.--while others use it less formally and perhaps without knowing it, as
when we may slap our leg to suggest a Heel. In a sense, Targeting and
Luring (or to Tillman, The Magnet Method) are really the same method. We
establish Targeting skills in our dogs by transferring the power of the
Primary Reinforcer--the food lure--to an object. So I would suggest that
Targeting is simply the use of a secondary reinforcer to lure. Pryor
suggest that Targeting is an important tool, yet interestingly, throughout
her book, she has little to recommend about Luring. Throughout her career,
she has regarded Luring as one of the "lower" arts of CT, which if used
should be faded ASAP. Karen's slant on this was undoubtedly formed during
her early years as a trainer of marine mammals. Although a whale or
porpoise can be taught to make contact with an obvious target, it is not
usually practical to get a whale to follow a treat you are holding between
your fingers.
Mimicry is dismissed by Pryor as a tool that does not work well with dogs.
She suggests that when we do think we see a dog "imitating" another dog or
a human gesture, it is more likely a case of the animal responding in a
similar way to a similar stimulus--not true mimicry. I don't know about
other trainers, but I have yet to demonstrate a behavior for even the
brightest dog and have it copy my movements. As a matter of fact, unless
trained specifically to understand the gesture, dogs don't have the
foggiest idea what we are doing when we point! How many times have you
pointed at an object to have your dog stare earnestly at your finger?

Finally, Pryor discusses Modeling--which doesn't have the precise meaning
the word has in normal language, By Modeling, she means physically
manipulating the dog--manually or otherwise placing it in position or
moving its body to simulate the desired behavior. When a conventional
trainer applies pressure behind a dog's rear legs to make them buckle to
cause the dog to Sit, that's Modeling.  To quote Pryor: " I am always a
little dubious about modeling as a training device... Until the subject is
doing the behavior or at least trying to do the behavior without being held
or pushed or modeled, I am not sure much learning takes place" (p. 61). One
of the practical limitations of modeling is that applying physical pressure
on a dog can trigger what is called the "opposition reflex"--if you push
the dog's rear down, it will resist and push its rear up! Clicker Trainers
might best appreciate the negative connotations of Modeling by realizing
that conventional snap & pull leash training is a great example of Modeling
combined with punishment.

So Pryor clearly favors Shaping (and what would become known as the Capture
Method) over all other methods she mentions except for Targeting.

Of course, in training a behavior, there is no single "better" or "right"
method. The method used needs to first reflect the temperament, anatomy,
and abilities of the individual dog. If I want to use the Modeling rear leg
sweep to cause a dog to Sit, I may find it effective with a Cocker Spaniel,
but it may be a challenge with a 180 lb. Wolfhound. The Luring method of
teaching Sit by raising a treat over the dog's head--which I believe
Tillman uses as an example of the Magnet Method--is also notoriously
useless with some dogs who, presumably after practicing body contortions
for some years, seem to be able to keep the lure in sight without sitting,
no matter where it goes! The Capture Method is great, but it is unlikely
you will capture a 12 pole weave!

Also, each method is not exclusive. When you lure a Sit by raising a treat
over the dog's head, your Luring, as I mentioned, is a close cousin to
Targeting. (Nose follows lure.) Shaping is used with virtually all of the
methods listed so far to perfect the behavior, yet some dogs such as
crossover dogs are reluctant to offer behaviors, and so Shaping may not be
the best choice for getting them started in CT.

So we end up with a tentative conclusion--which seems reasonable
enough--that there is a time and place for all of these methods--except
perhaps Mimicry, where the general behavioral community tends to agree with
Pryor's early pronouncement that dog's are not good "copy cats." We also
see that various methods can be similar and/or complementary to one
another. It is a rare behavior that we "get" into final form via a single
method.

In Melissa Anderson's "Click for Joy," she divides ways of "getting the
behavior" into Modeling, Luring, Capturing, and Shaping. If we combine the
terms of Tillman, Pryor, and Anderson, we end up with:

==SHAPE: AGREED UPON AS A STANDARD CT METHOD.

==CAPTURE: AGREED UPON AS A STANDARD CT METHOD.

==TARGETING: USED IN CT & RELATED TO LURING.

==LURE OR MAGNET METHOD = OLDER TECHNIQUE.

==MODEL: NOT USUALLY RECOMMENDED IN CT.

==MIMIC: PROBABLY NOT EFFECTIVE WITH DOGS.

The main point I wanted to lead up to is that in CT today, there is a clear
preference for some methods over others. I've intentionally listed the
methods above in order of the preference that Karen Pryor and most other
modern trainers would list them. Again, this doesn't mean that any of them
are forbidden! It does mean, however, that some of them offer more
advantages in the long run than others. Some carry problems and potential
drawbacks. Others foster the most preferred kinds of learning. So *given
the choice in a perfect world,* one would prefer to Shape or Capture a
behavior rather than to Lure it or Model it. In my own opinion and the
comments of other trainers, there is a quick "drop off" on the list after
Luring. While the first four methods listed above are used commonly and can
be used effectively, once we get to Modeling, we are making the transition
between Operant Learning and a rather mindless "Do this!" technique in
which WE move the animal ourselves rather than expect it to move. And
again, for most practical purposes, Mimicry is useless.

There are scientific and behavioral reasons why the methods list themselves
in this order. OnE apparent and important organizing principle is that the
list starts with the methods which require the animal to think on its own,
and omit behaviors most spontaneously and by its own choice. It then moves
downward  toward those methods which require the least thought and choice
on the subject's part. (Mimicry--a poor candidate in the first place for
inviting learning in dogs--is the one method that does not really fall into
this scheme.)

I like Shirle's question and request for descriptions of how particular
behaviors are trained by different people. I hope I've made the question
more interesting and thought-provoking by providing this background. Yet
perhaps the textbook doesn't tell all! I've experienced dogs that were dead
from the neck up when it came to shaping, and I am sure someone will
present us with a wonderful example of how their dog learned a complex
behavior solely through mimicry, so the question remains, "Have all our
dogs read the same book on Operant Conditioning?"

===========================================
Barry McDonald -------------------- DogSense
===========================================
Pinto & Little Bits, Basenji Partners in Crime,
rescues Shadow The Dog, Cheyenne,
and Cherokee--Master of Calming Signals.
===========================================
"A dog cannot be bad, it can only be a dog."
===========================================
"Show me a dog that's been trained and
trained and trained and still does not obey,
and I'll tell you who the slow learner is!"
===========================================

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://taylor.uurnet.net/pipermail/click-l_click-l.com/attachments/20040505/e1512401/attachment-0001.htm