Category: Performance Events ¤
  Author: Karen Pryor ¤
   Title: Teaching the Weave Poles ¤

I've been meaning to comment on weave poles for ages, but Norman's post,
deserving of reinforcement, prompts me to do it now.

Norman Welsh says:  "I have read that weave poles are
perhaps the most difficult part of agility to train or shape , and I
have
been giving this some thought. Seems to me that the weave poles are a
behavior chain , and a behavior chain is shaped backwards.......so you
shape the left to right movement around the last pole first and
click.......once that is shaped you put a pole behind it.......shape
that
movement.....now the dog weaves between two poles for a click upon
completion....and one by one you put poles behind and shape until the
dog
has completed the chain.

 Had a bit of a chance to try this out at a one day intro to agility
with
an out of town club in the fall......no one clicked there. When eveyone
else was busy , I took my fluff and my clicker to the weave
poles..........only had a few minutes to spend there , but she seemed to
catch on very quickly . We did not use a target stick but that might
make
it even easier.  We had tried it the usual way  prior on that day and it
really did not seem to give her any information. I remembered reading in
K.
Pryor's LBtW that behavior chains are shaped backwards ...like her
dolphins
weaving in and out of the hoops."


Exactly right, Norman. Weave poles are a homogeneous behavior chain,
that is, the same behavior repeated over and over. An efficient way to
train it is to isolate the basic unit of behavior, train that, put it ON
CUE, and then set up two units, get the animal to do the next to last
unit, reinforce that with the CUE for the last unit, click and treat.
Then introduce the CUE for the next-to-last unit and let the animal
accomplish both units and C/T. Then add a unit in front of those two,
etc. etc. The CUE is your main conditioned reinforcer between units as
you are building the chain. 

Norman suggests a basic unit of one pole, with the behavior being 
/o\ or go around the pole and come out the right way.

Mary Owens posted a description some time ago of training the weavepoles
with the basic unit being two poles, so the dog goes  o/o\
Same behavior, slightly different setup. 

I had lots of fun at last year's Manchester NH clicker seminar training
a Jack Russell to do the weave poles with three poles, so the behavior
was o/o\o. When the dog had learned to go around the middle pole, I
added a Cue: GO! and a hand signal.  GO! o/o\o

When the dog was waiting for the cue (after three trials; this dog
already understood verbal cues) I set up the course like this:
o/o\GO!o/o\o and the dog zipped through. 

That's as far as I dared take time to go, with 125 people watching, but
the next step would be to put the GO in front of the new unit, then add
another unit, etc.

Someone remarked afterwards, bewildered I think, that it worked because
the dog seemed to LOVE the GO! <g> That is indeed why it worked! Of
COURSE the dog loved the GO! Cues are release signals for known operant
behaviors, and as such they become conditioned reinforcers for any
behavior that is happening when the cue is given. Cues are as "loveable"
as the click, for exactly the same reasons (unlike the obed. command,
which is often actually a threat.)

 methods: I used a target to get the two or three first passes into and
then through the first unit, and the first pass through the second unit.
Gail Fisher also had some wire fencing which I used to confine the path
around the first unit, briefly; it really wasn't necessary. The dog
swung wide at first but then (being a JR) shaved all unnecessary moves
down in order to get to the click spot as fast as possible.

We have this on video. It took 5 minutes 20 seconds, to get two units
with comprehension and speed. 

If you think of the weavepole task as being the behavior of going left
and right, you are trying to train two behaviors at once. If you think
of it as a single unit, bend around the pole, coming out right, it's
much easier to establish. Also, the dog will swiftly begin to bend
tightly around the left poles, just out of natural tendencies to
minimize the trip to the next unit.

If you try to arrive at the WHOLE behavior of weaving by running the dog
through the whole course and gradually squeezing down the width of the
course, using the various contraptions and devices people have designed,
you are actually relying on Pavlovian conditioning. Your hope is that
the dog will gradually become accustomed to doing it a certain way,
through many repetitions, whether it understands the rules or not. This
by definition will take months, and since the dog is not giving you an
informed, operant behavior, any amount of errors can take place. And do.

Train it as an operant, and you can get it done in a few sessions.

By the way, I was moved to use this as a seminar demo because Gail
Fisher had these great indoor moveable dissassembling (if that's a word)
weave poles, so I could put them on the stage and build the pole line
backwards. But I'd never trained the behavior before (attendees may
remember that in my ignorance I trained that JR to bend right to left,
not knowing it had to come out on the right). So I practiced with a
person; and I can highly recommend this, if you want to train yourself
before you train the dog. I found an audience member who had never seen
the weavepoles and didn't know what we were talking about, who was
slender and agile enough to pass through them, and I used a target and
shaped her through the basic unit, and then put the behavior on cue,
then added another unit in front, and in fact got her weaving very
nicely through three units, in about the same time it took to do two
with the dog. Great way to see how this works. Any willing child would
do, I suppose, as a practice animal.

Karen Pryor