The Panda Project
A Guide Horse for the Blind
Our First Winter - Continued
Report 3: Dec. 2001 - Feb. 2002
Written by Alexandra Kurland
Cinderella Gets her Gown
The other major event in January was our trip to Massachusetts to fit Panda for her harness. When Ann first contacted David Chabot, he was at first a little nonplussed. No one had ever asked him to make a guide harness for a horse before! But he was willing to try.
David works out of his home. His harness shop is a small room at the back of his garage. With all of his equipment there isn't much room to move, but we managed to fit ourselves and both guides into the crowded space.
Here was a major test of Panda's training. Not only were we in a tight space, but she had people fussing over her. This had been the one major issue she had come with, her dislike of being handled. Back in October she was still making ugly faces when people approached her. Now in January she stood perfectly while David took measurements, and tried on different harnesses. I watched her expression the whole time and was delighted to see how calm and relaxed she remained throughout the entire process.
We tested several different choices by walking Panda up and down the driveway. Ann and I took turns feeling out the different styles of harness. David has designed an ergonomically balanced handle that is a huge improvement on the old straight handles that are on Ann's current harnesses. As soon as I felt it in my hand, there was no question that's what we wanted. That still left questions about how to fasten the handle to the harness body, and how it should fit Panda so it didn't press on her back or rub her shoulders. By the time we were done, it was almost four o'clock.
Which brought us crashing head on into the one major hole in Panda's training, and that's centered around the house breaking. Panda learned very early on in her training to urinate on cue. I could ask her to relieve herself and know we could go into a building for an extended period without a problem. In the approximately four months she'd been with me, she'd had no "accidents" of any kind inside.
That was to a great extent Panda's choice. She definitely had urination on cue, but Panda was very secretive about dropping manure. I almost never saw her doing it. At clinics we'd be with her all day, waiting for an opportunity to catch the behavior, and she'd wait until no one was looking to drop a pile in the back corner of her stall. It made getting the behavior on cue a major challenge.
Up to this point this wasn't really a problem because Panda was very selective about where she would go. Houses, indoor arenas, office buildings, shops, these were not suitable locations in her estimation, but cars, or more specifically, Ann's van was. On these long outings Panda would wait until she was back in the van and then drop a pile. Oops! We would offer her lots of opportunities to relieve herself. She would urinate on cue every time, but she held her manure until she was either in the van or by herself in a stall. Catching the behavior so I could put it on cue was proving to be an unexpected challenge.
The interesting thing is the Burlesons, who trained Cuddles, the first mini to be used as a guide, set up their minivan as a mobile stall. Perhaps they had run into the same tendency and simply accepted the behavior as an easy way to manage their horses when traveling. Neither Ann nor I found this to be an acceptable solution. I was not going to have my car turned into a giant liter box. And furthermore this raised the whole question of mobility issues. You can't use public transportation if your guide animal is going to regard it as an appropriate place to relieve.
Up to this point I had observed the pattern, but I hadn't decided on a strategy to eliminate it. I knew my goal was to have a separate cue for urination and defecation. If I could ask Panda to relieve on cue before letting her into the van, I could break the association and transfer the location of her liter box to a more appropriate area. The stumbling block for shaping this was my schedule and her coyness about dropping manure while she was being watched.
Basically the problem was that much of my day is spent on the computer, and the computer is in the basement. I had been putting off this layer of the housebreaking until Panda had enough experience with stairs to spend the day hanging out by the computer. But my schedule was such that I hadn't gotten her to enough locations with good training stairs. She was great on outside steps. I needed to find some inside stairs which were a little easier than the ones in my house. Structure is everything in a process like this, and I just hadn't found the right set up for her yet.
So she was stuck upstairs while I worked downstairs and the housebreaking was stalled at a mid-way point. The long car trip turned the issue into a priority. The solution was to rearrange my house and climb out of the basement. The result has been a major step forward in Panda's training and I get to see daylight! Amazing. The main computers are still in the basement, but I can do email, and other tasks like writing this report upstairs. That means I can now monitor Panda throughout the day and keep her on a schedule.
Training really is a matter of waiting for each layer to emerge. This step in the housebreaking needed several other pieces to solidify before I could address it directly. The main thing Panda had to learn was to stay in her pen. Remember the principle is: "You can't ask for something and expect to get it on a consistent basis unless you have gone through a teaching process to teach it to the horse."
When I first brought her into the house, I couldn't go out of her sight. She'd fuss and fret, and I was always concerned if I went too far away that she'd try to climb out over the top of her pen and hurt herself. I went through a lengthy teaching process to build her confidence and comfort level so she will now stay in her pen while I move about the downstairs. When she first went into her pen, I had her on a very high rate of reinforcement. Now I can basically ignore her while I work. She doesn't need constant monitoring or reinforcement for staying quiet. She's now learned that her pen is a place to nap and rest, and amuse herself. She's content to stay in it now through a good stretch of the day.
That laid the necessary piece for the housebreaking. My next task was to get the behavior of producing manure on cue. For this I established a routine for the day. Panda got breakfast in her stall around eight. Mid-morning we'd go out for a walk, then I'd bring her back to her stall to relieve.
Her cue to urinate is "Get Busy!" That's a solid cue she learned in the first week of her training. The new cue to defecate is "Park Time".
I knew that her pattern was to produce a pile of manure around 10:30. All I had to do was wait, and wait, and wait. Actually, it took her about five minutes. Normally, with clicker training you don't attach a cue until you have the behavior, but in this case I gave her the cue. The tone of voice for both behaviors is similar. Panda interpreted my hovering over her as a request to "Get Busy". As long as she was thinking about relieving, I clicked and reinforced each effort with a pellet of grain. And I waited, until finally, just when the cold was starting to really set in and I was thinking she wasn't going to do anything, she lifted her tail and dropped a pile of manure. Click and jackpot! She got a much treasured peppermint.
This is where you see the real power of clicker training, and the advantage of jackpots. It took only two more experiences for Panda to make the connection. Drop a pile and you get peppermints!
Panda wanted those peppermints. She knew what she needed to do. She just didn't always need to go. It was so funny watching her. She was so like a toddler. She could easily oblige when I asked her to "get busy", but this other! She'd circle and turn, and back herself into position against the wall. Nothing. She'd toss her head, and give me a toddler's "But Mom, I don't need to go!" look.
"Park Time" I'd chime out in a cheery voice. She'd toss her head again, and go through her ritual set-up. Eventually, after about five minutes, she'd manage to drop a pile. There were several things going on here. First, she had to learn how to control and produce at will bowel movements, a skill like any other that takes time to develop. And second, I was hoping we would create a conditioned-response connection between urination and defecation. I wanted the cue for the first to help trigger the second. My goal was to be able to take her out for a potty break, ask her to relieve, and have both functions performed so I would know that it was safe to go back inside for an extended period of time. At the moment, I could ask her to empty her bladder, but I had no real control over manure. That was the hole in the system that needed to be filled.
By evening of the first day committed to this project, Panda was four for four. She was clearly understanding what she had to do to get a peppermint, but it was still hard for her to produce on demand. She was taking about five minutes from the time she urinated to the time she defecated.
That was Friday. In a system that requires consistency, Saturday unfortunately we went for another long road trip. We had to drive back over to MA for a second harness fitting, but this time Bob Viviano went with us, so I could ride in the back and monitor Panda. Unlike the previous weekend, this time we had no accidents. As a matter of fact, we caught one important event. Panda started to go, and we were able to pull over so I could whisk her out of the van just in time.
The message hopefully was clear: there are places where it is appropriate to go, and places where it is not. The van is one of those not appropriate places. Since I now had a working cue, I could go ahead and ask her to relieve outside the van. We were parked on the edge of a country road. She was on a steep bank with tempting winter grass poking through the snow to distract her. It took a few minutes in this unfamiliar setting, but Panda finally obliged. Click, peppermints, carrots, extra grain, bells and whistles!
By Saturday evening we were six for six for the day and our time had dropped to less than a minute.
At that point Panda again had a change in the schedule. On Tuesday I was flying out to California for the Equine Affaire. I was going to be gone for almost a week, so Dolores Arste got the job of Panda-sitting and continuing the house breaking process.
I could not have left her in more capable hands. Dolores has trained and bred Racing Siberian Sled dogs and champion Schipperke dogs. She was a Professional Show Dog Handler of many breeds for 20 years, and she is now an American Kennel Club Show Dog Judge. At one point she had 85 dogs at her facility. With credentials like that, needless to say, she has far more experience with house breaking than I.
Furthermore, Dolores is a competitive trail rider. She's completed several 100 miles races so she has the energy and endurance to housebreak a horse. She was eagerly awaiting Panda's arrival. She had her strategy all worked out, and was planning not only to solidify the "park time" cue for me, but to extend the duration of her inside stays to overnight housebreaking.
While I was in California, I got daily updates from Dolores. Progress was definitely progressing. Panda was solidly understanding the cue. She was learning to go more promptly, in different locations, for both Dolores and her husband, both on and off leash. Dolores was able to keep her with her in the house until about midnight. Panda apparently is a night owl because at that point she would become very active and need to go out.
The interesting part of this process is that Dolores had to keep shifting her model for housebreaking. This is part of good training. A good trainer is always assessing the strategy she's chosen based on the animal's response. A basic principle of training states that there is always more than one way to shape everything. Good trainers learn to observe the effect of reinforcement, and they adjust their training accordingly. This is why you cannot give rigid recipes for training. You have to begin with principles, and the principles teach us how to observe, and how to be flexible so our animals remain successful and our goals are met.
Dolores began by thinking of Panda as a puppy. A puppy will not by choice soil its bed. You can use this in crate training to condition the puppy very easily to relieve outside. But Panda was actually more like an older dog who has learned to go in its crate. She had learned that her stall was an appropriate place to relieve. As Dolores tried to stretch the in-house duration, Panda also decided that her inside exercise pen was an acceptable area. Oops! She dropped two piles within half an hour of each other. Double oops!
Actually, this was just part of the process of learning what works to earn reinforcement and what doesn't. Dolores, I think very correctly, felt that Panda was understanding the "park time" cue, and was simply doing her best to offer Dolores a well-rewarded behavior. She had the first part of stimulus control. We had the cue eliciting the behavior, but we needed to extinguish the off-cue behavior, at least in the house.
To do this Dolores changed her model again. She shifted from dogs to human infants. She started thinking about what makes a toddler decide to use a toilet instead of soiling a diaper. Positive reinforcement for going at an appropriate time along with the discomfort of a soiled diaper combine to create the desired result, bladder and bowel control.
So, the ever resourceful Dolores went shopping at the local pet store and bought a large size dog diaper. Panda hated it at first, but with a few clicks and treats learned to accept it. Apparently, it took just one instance of dropping some manure into the diaper to convince Panda that this was a most unpleasant thing. Waiting to go outside was much better.
By the time I returned from California Dolores had created a solid cue. Panda was reliably going outside in a timely manner. She had not had any more inside accidents, and she could stay in the house with Dolores until about midnight. The doggy diaper was used only in the evening to insure there were no accidents at a time when Panda was normally at the barn with me.
That's excellent progress for just a week's worth of training. Dolores had her on a schedule where she would take her out every hour and a half, and she was reliably relieving within five minutes. When I brought her home on the Tuesday following the Equine Affaire, I shifted the emphasis from building duration for the overnight housebreaking to building more fluency in the cue. Five minutes is way too long in my schedule to be waiting around for Panda. So the first order of business was getting a faster response time to the cue. We're now down from five minutes to less than a minute and usually she goes within twenty seconds.
I'm also slowly building the time between potty breaks. I started her at about an hour and fifteen minutes. We're now stretching her mid-afternoon break to close to two hours. When we're close to the time she needs to go out, I open her pen and let her wander around the kitchen. When she rings the bell hanging on the back door, I get up and take her out.
At first she went directly to her bell because she's been reinforced for ringing it. I no longer click her for ringing the bell. Instead I open the back door and whisk her out to her stall. She seems to be making the connection between events. She is now spending more time in the kitchen before going to her bell. When I'm certain that she understands that the bell is her signal to me that she needs to go out, I'll hang it in her pen. At that point we'll have an elegant system. She'll be able to tell me when she needs to relieve herself, and I'll be able to ask her for both functions.
This is still very much a work in progress, but it has certainly been an interesting exercise both for me and for Dolores in stimulus control. In March we'll have several long car trips so we'll get to see if we have solved the problem of her using the van as a mobile stall. In the meantime her training has shifted to another level. The added control over the housebreaking has meant that Panda gets to spend even more time with me. She's living much more the life she'll have as a working guide. She goes for walks, she runs errands with me, but mostly she sleeps and eats and sleeps some more while I work. And at night she gets to go play at the barn. It's a good life for a young horse and one that seems to suit her well.
Click Here to read the next report, or return to the Panda Project Home Page for the full index.