Category: Puppy Stuff, Common Problems ¤
  Author: Dani Weinberg ¤
   Title: Two Reward System ¤

Last night, I tried John Fisher's Two-Reward training system to retrain 
Ruby's hyper-reactivity to unknown dogs. (You may remember that she's 
my almost-8-month-old German Shepherd puppy, clicker trained from 7 
weeks.) Her motto is: When in doubt, bark. I've done a lot of 
socializing to people but less to dogs (because of lack of good 
opportunities at the right time). When she sees an unknown dog at a 
distance, she barks and strains at the leash. As we approach the dog, 
she either continues her barking display in the dog's face (for a few 
minutes) or simply stops and starts doing normal greeting-behavior 
things (sniffing, circling, etc.). She seems to do better (i.e., move 
to the normal greeting behavior) when there's a group of unknown dogs, 
as opposed to a single one. I don't see anything but unsureness in this 
behavior. I believe this is a developmental thing, and I simply want to 
make sure that we straighten it out soon so that it doesn't become 
fear-aggression later on.

I had enlisted the help of 3 of my students (who are also fellow 
members of our NM Clicker Group) and set up a video camera, so that I'd 
have a record of this. We worked in my training building, with Ruby and 
me at the far end (about 30 feet from the door). I was sitting in a 
chair, and Ruby was on a short leash, attached to a wall hook. Tommy 
was the observer, sitting off to one side, and Mary was our coach, 
making sure that our timing was right.

Katy and her dog Zak (a calm, trained, dog-friendly dog) entered the 
room. Ruby saw them, paused for an instant, and then started barking 
and straining at the end of the leash for about 15 seconds (times are 
approximate), with her tail waving high. The moment she started 
barking, I left her and went to stand at the opposite side of the room 
(about 15 feet away) with my back to her. When she stopped barking, 
Katy and Zak immediately left the room and shut the door behind them 
(First Reward). At that same moment, I returned to Ruby and started 
feeding her some garlic chicken, little piece by little piece, 
stretching out the jackpot for a few seconds (Second Reward).

We repeated this process for a total of 6 times. Here's what happened 
after the first time I've just described.

2nd Repetition:
When Katy and Zak entered, Ruby threw me a quick glance, and then 
barked for about 4 seconds. (We did the same "2-reward" routine, with 
me leaving when the barking started, Katy and Zak leaving when Ruby 
stopped barking and me returning to jackpot her.)

3rd Repetition:
When Katy and Zak entered, Ruby barked for about 3 seconds. (Same 
leaving and returning routine as before.) This time, too, she 
positioned herself right next to my chair, sitting in Heel position (on 
her own, not on my cue!) - in contrast with the first 2 times when she 
had remained a couple of feet away from me. I thought that was a 
significant shift, along with the drastically reduced amount of barking 
time.

4th Repetition:
This time, Ruby did not bark at all but simply watched Katy and Zak for 
about 3 seconds and then looked at me. (Obviously, I did not leave but, 
instead, jackpotted her from my chair for looking at me.)

5th Repetition:
Same as 4th rep - no barking - except for 1 very inhibited bark and 
tiny movement upward out of her Sit, both of which (the bark and the 
movement) she held back on her own. I immediately jackpotted that 
behavior.

6th Repetition:
She watched Katy and Zak quietly for about 5 seconds and then looked at 
me. Jackpot, of course! And we quit!

I think we were all amazed and elated at how well this process had 
worked. It was quick, and it was so clear and easy to watch and respond 
appropriately to Ruby's progress. Before we started, I had tried to 
plot out an approximation ladder of raised criteria (quiet, relaxed 
body, looking at me), realizing that I really had very little idea of 
what would actually happen. So we just went with the flow - for 
instance, reinforcing her for inhibiting her bark and movement during 
the 5th repetition.

Afterward, we spent a few minutes debriefing the event, and that, too, 
was a learning experience. Mary asked me, for instance, what were my 
expectations for Ruby's behavior around unknown dogs. That helped me 
clarify and articulate what I had in mind - basically, that she'd show 
only mild interest in them and strong interest in me. After all, her 
career in obedience and herding will require her to be around unknown 
dogs. Mary was interested in this training because of its possible 
applications to her 9-month-old Pyr Rosie.

A few other comments. Videotaping the session was extremely helpful. It 
all happened so fast (I have only 4 minutes and 40 seconds on the 
video) that none of us could have retained a really accurate picture of 
it. Another suggestion, if you'd like to try this yourself, is that the 
neutral dog's handler not use a clicker. That seemed to be an extra 
distraction to Ruby when Katy was clicking Zak (starting with the 3rd 
repetition, I think).

Also, notice that I added some variations to the method as originally 
described by Bianca. Instead of just returning to Ruby as her "2nd 
reward," I returned and jackpotted. I don't know if that's good or bad, 
but it seemed to work nicely and it felt right. (I'll be interested in 
Bianca's and others' thoughts on this.) I was a little sloppy with my 
reinforcement process and hope to clean it up the next time we do this. 
The only time I used an absolutely clear CR ("Yes") was on the 6th 
repetition. The other times, I used a keep-going "Good." Next time, I 
think I'll give the CR as I'm returning to Ruby, following that with 
the jackpot once I've reached her - making sure those are clear and 
distinct events.

I plan to repeat the process, with a different dog, on Friday (if 
you're willing, Rosemary!) and then again next Monday with yet another 
dog (if you're willing, Tommy!). I think that adjusting this variable - 
the unknown dog - may be more useful right now than changing some of 
the others - like the distance between Ruby and the unknown dog. (Katy 
and Zak entered the room and stayed within a couple of feet of the door 
for each repetition.) Any ideas about this?

I'd be very interested in hearing about other people's work with this 
training method. Many many thanks to my training assistants and to 
Bianca for bringing this method to our attention.

Dani

Dani Weinberg Albuquerque, NM

Last night, we had our 4th Two-Reward training session with Ruby. (I 
posted last week about our first session.) It's working incredibly 
well. We've worked in the same location each time but with a different 
dog each time - including one with whom Ruby has some history (that was 
our 3rd session).

Last night she did not bark AT ALL when the dog entered the room with 
the handler! The first time, Ruby's body alerted at the entrance of the 
unknown dog (though she remained sitting), she stood up and leaned 
forward slightly, and then she sat again and gave me eye contact. The 
2nd repetition looked liked her "Ah-Ha" moment, and she actually did a 
sort of double take: eye contact, then a quick look away, and then a 
very deliberate eye contact again. On the 3rd repetition, it was 
absolutely clear to everyone that Ruby was taking the dog's entrance as 
a cue to give me eye contact. On the 4th repetition, she did a little 
experimenting, instead of giving an instant response. And, on the 5th, 
the response to the cue of the dog's entrance was instantaneous. Once I 
saw she was getting it (after the 2nd rep), I raised criteria just a 
bit and required a slightly longer eye contact before reinforcing.

As in the previous sessions, I was not using the clicker but did use my 
verbal CR "Yes" and jackpot treats. Last night, on one of the 
repetitions, I used only a few pieces of Rollover, our usual training 
treat, to satisfy Shirley Chong's suggestion that variation in 
everything really accelerates learning - in this case, the dog never 
knows if her PR is going to be fantastic or just ho-hum. Ruby didn't 
seem to notice the "reduced" value of the treat!

A few bits of information and hints for those of you who want to try 
this. The whole process takes less than 5 minutes, and it's under 3 
minutes once the dog is beginning to get the idea. Also, it's a good 
idea to have some of the jackpot treats hidden on your person, in case 
the dog does not bark and you don't get to leave her. (I was keeping my 
jackpot treats on a chair across the room, at the spot where I went 
when I left her.)

I feel like we've done about all we can do using this one location and 
the 4 different dogs. I'm wondering what the next step(s) should be and 
would appreciate your thoughts on this. Should I raise criteria *very* 
slowly and just start changing the physical location (outdoor instead 
of indoor, for example)? Can I be holding the leash instead of having 
it attached to a wall ring? Should I raise criteria so that the unknown 
dog and handler come in a little closer (they've been at about 25 feet 
so far)? Should I start taking it "on the road" into a natural setting 
like the park, rather than continue doing controlled setups? And how do 
I eventually make the transition from setup to real life?

Dani Weinberg Albuquerque, NM