Clinic themes- Alex K -Part 2

The Poisoned Cue Reference
To help you follow this next section I need to include a bit more background material. In many of my clinics I show two video clips from the "Poisoned Cue" DVD. I won't be defining all the terms I'll be using in this next section. Instead I'll refer you instead to the "Poisoned Cue" DVD. If you aren't familiar with the concepts I'm talking about or haven't seen the video clips, it is well worth watching.

One shows the "Ven" condition. The other shows the same dog working under the "punir" cue. The question Jesús Rosales-Ruiz and his graduate student, Nicole Murray were asking was this: can you shape with a poisoned cue? Does it work the same as any other cue. To answer this question Nicole taught her dog, an eight year old miniature poodle, the same behavior "come" - but in two very different ways. Separate cues were attached to each training method.

The cue "ven" was attached to the teaching process we would all be familar with and very comfortable using. The dog was shaped through successive approximations to come. When it oriented to its handler, she clicked and gave it a treat.

In the other condition, if the dog did not respond promptly to the come cue, in this case the word "punir", it was dragged over to the handler, where it was then given a click and a treat. Dragging doesn't sound so bad, but it was clearly aversive to this little dog.

So in the "ven" condition, the training was done through splitting. In the "punir" condition the training was done through lumping. In both instances the dog learned to come promptly and reliably.

Once the dog was reliably responding to both cues, they wanted to see if they could use the cues to shape other behaviors. Nicole marked her office floor off in grids. She then stood in the middle of the room and waited until the dog walked over a particular square. Now she could easily have captured this behavior with a click and a treat. If she had clicked every time the dog stepped in the square, she would have seen the frequency of this behavior go up. Easy.

But that's not what Nicole did. Instead she cued the dog to come using one of her two cues. She wanted to see if there was a difference in the dog's behavior.

In the "ven" condition, the dog walked over the square, heard the word "ven" and promptly went back to Nicole where it received a click and a treat. In the trial I show at clinics, the dog then immediately returns to the designated square. Its a beautiful example of a clean loop. There's no wandering aimlessly around the room. There's no uncertainty. There is a bit of tail-wagging happy dance, but other than that there is no time wasted. The dog gets his treat and returns to the designated square. Why? because he knows that's how he can get his handler to say the word "ven". This is a behavior he wants to trigger in his human. "Ven" is a predictor of good things to come. Get your human to say the magic word and you can get her to produce goodies. It was a very clear, very simple demonstration of the use of cues to shape behavior.

The "punir" cue tells a different story. In the "punir" condition you see a depressed dog that wanders around the room avoiding the square that turns on the "punir" cue. It's a hard few minutes to watch, but its an important video clip. The "punir" look is a familiar one. I see it in horses as well as dogs. In fact I see it in people. In the "punir" condition there are no clean loops. The dog does wander into the designated square and he does receive a click and a treat for coming, but he doesn't return promptly to the square. Instead he hangs back around his handler's feet. He wanders around the room before finally passing over the square. This isolated clip may look like a clicker training session, but the lack of a clean loop reveals the problems in the training history. Something is clearly wrong with this picture. This isn't the happy, exuberant look we want to see.

What these two clips tell us is we can indeed shape with cues. It reminds us that the effect of cues works in two directions. They tell the animal which behavior is likely to be reinforced next. And they reinforce the preceding behavior. "Ven" told the dog that heading over to its handler was the "hot" behavior of the moment. "Ven" was a reliable predictor of good things. Orient to your handler, and she gives you treats. The cue "ven" acted like a green light, telling the dog when this behavior would work. It also reinforced the preceding behavior because the dog was learning which behaviors could reliably turn on the "ven" cue.

But stir some negative consequences into the pot, distort the process by lumping so there is ambiguity and uncertainty about the outcome, and the results change. Now the dog's behavior becomes more erratic. The ability to shape behavior with a poisoned cue breaks down.

This is important information. If we want to get the most value from our cues; if we want efficiency in training; if we want happy, "ven" animals, we need to be careful how we shape behavior. I wanted to explore this concept more with my own horses, and I wanted to see how I could make more deliberate use of cues as reinforcers using a well-constructed loop. My goal was to create loops in which every behavior in them functioned as a "ven" reinforcer. Robin has been my principle guinea pig for this work.

Mats as Reinforcers
In the loops I created for Robin, initially I decided that I would click and reinforce him only when he was on the mat. If he did something particularly noteworthy off the mat, we would return promptly to the mat where I would then reinforce him with a click and a treat. Once he was on the mat, I deepened it's reinforcement history by clicking him for easy behaviors such as posing which he could do on the mat.

So going to the mat became a predictor of good things happening. Just as the dog in the "ven" condition wanted to find behaviors that would trigger the "ven" cue, I was hoping that Robin would want to find behaviors that would trigger the "go to your mat" cue. My goal was to build a complex loop made up of many behaviors but with the mat as the only place where the click and treat sequence was used.

In many ways this is nothing new. People have been building chains using cues as reinforcers for decades. But I wanted to explore this using a new language set to see what emerged out of it. Often times when you change how you describe something, new perspectives are revealed and our overall understanding of how training works advances.

My question was could I shape behavior using the return to the mat as a reinforcer? Could I build my loops in such a way that I would be satisfied that Robin was indeed recognizing that the exact moment when I offered him the return-to-the-mat cue coincided with the moment when I would normally have clicked him? And could I then use this carefully built understanding of the return-to-the-mat cue to shape or refine behavior? The answer was yes, of course, absolutely.

Please note: I was reserving my click and treat in this sequence for a very specific purpose. I am not recommending that you stop clicking and treating your horse except when he is on a mat. A well built loop may have multiple clicks and treats embedded in it, so if your horse needs you to click and treat off the mat - by all means do so. However, if you and your horse are ready for it, it is certainly worth exploring the procedure I'm outlining here. I've gained all sorts of benefits from it. The overall loop was built out of many component behaviors, all of which were built with many clicks and treats. You need to judge when you are ready to apply this lesson to your own training.

Cone Circle Patterns
Building the mat loop was the work I explored over my 2008-2009 winter break.

This past winter I went on to expand the mat work and the cone circles. To build the mat-as-reinforcer concept I had used a standard "why would you leave me circle" with the mat at the center.

In the 2009-2010 winter training cycle I moved the mat out to the rim of the circle in place of one of the cones. So now Robin could stand on the mat, leave the mat, go around the circle only to find himself back on the mat - click and treat!

It was a deceptively simple pattern. I can hear people saying, this is what you've been working on all winter! This is your clinic theme! Going around a circle! How is that new work?

When I'm working on a complex concept, I always like to use simple behaviors. Putting complex on top of complex just makes things, well - too complex. Looking at an idea using simple behaviors and simple patterns helps to reveal the larger, more over-arching concept that I want to explore.

When I moved the mat out onto the rim of the circle, I was confronted with a task for concept training. I wanted Robin to walk around the outside of the cones while I walked on the inside. Most of the time when I have used cones, we've both been on the outside of them, so this was a rule change for him. I needed Robin to understand that people belong on the inside, horses on the outside.

The dictionary that's in my computer defines a concept as:
an abstract idea or general notion; an idea or mental picture of a group or class of objects formed by combining all their aspects.

Having Robin go around one side of a cone while I stepped around the other side was a behavior. Having Robin understand that he was to stay on the outside of all cones, that patterns of cones have an inside and an outside, I would consider a concept.

If my goal had simply been to get Robin around a circle of cones, it would have been an easy enough task, something we would have achieved in very little training time, but I was after this deeper layer of understanding so I set about very systematically building this understanding. Again my foundation was all the previous work I had done with him, especially in our initial loopy training exploration of mats and cones.

The cones I had for him were the short button cones that look like little inflated frisbees. They are too easy to step on and over. To learn about inside and outside I needed something taller. From the internet I got eight 15 inch coloured cones. I love having the variety of colours. Having something other than orange makes things so much more interesting.

For starters I set the cones out in a very small circle. Now when I say small, I mean small. I set the cones out so they touched one another. You couldn't go through them without knocking them down. My circle was only a couple of feet in diameter but Robin could manage it nicely.

I've presented this tight circle exercise at most of the spring clinics. Some horses can manage it, many others struggle. They go wide and can't hold the tight arc around the cones. This goes back to component behaviors. Robin had several component behaviors which helped him manage this exercise. At one stage in our training I'd used half a dozen or so mats all of different sizes and shapes. Some of his mats were smallish, narrow rectangles. The game was to step onto the mat along the wide side, step straight off it, and then to return to it via a very tight circle onto the short side. Robin got quite good at this. Whichever side he stepped off of, he could bend back around and step onto the next side of the mat, even on the very smallest of our narrow mats. So walking around a tight circle of cones was easy work for him.

That doesn't mean he didn't try to take a few shortcuts back to the mat, but he quickly learned that knocking cones over meant we had to keep going past the mat to try again. He became very diligent about staying to the outside of the cones.

So the next step was to expand the cone circle out in slow stages, confirming at each step in the process that Robin would keep to the pattern of "horses to the outside, people to the inside."

The mat served several functions. It was our place of reinforcement. Leave the mat, return to the mat - click and treat! It created enthusiasm for returning to the mat. The mat helped to form the pattern and it built a willingness to keep going around the circle. That's on the plus side.

The mat also has a negative side. (I'll call it this for the moment, but I need to define more clearly what I mean by negative because actually the the effect is quite positive. Don't worry about the "Alice Through the Looking Glass" effect. It will all get sorted.)

The more Robin understood that the mat was a good place to be, the more of a draw it placed on him. This draw can create problems if you don't manage it well. But it is this draw, this desire to get back to the mat that I was using to explain more of the cone-circle rules. Robin wanted to get back to the mat. Especially as the circle got bigger, it made sense to cut corners, to slip in on the inside of the last cone to get to the mat, but that only delayed getting to the mat.

So the fun of mats is to make the mat a highly desirable place to be, to magnify for the horse the meaning of mats, and then to use their enthusiasm to explain another set of rules. You can get to the mat, but you have to go there with me, keeping slack in the lead. You can go to the mat but you have to get there by going around a set of cones.

The challenge for the handler is to teach these rules without poisoning the mat. The behavior must remain "ven". (See the Poisoned Cue DVD for an explanation of these references.)

It's easy to set out rules, have obedience, but end up with a horse who really doesn't want to play. It's very much like giving a small child a colouring book and then insisting that the child stay within the lines. A creative child may feel smothered by this "rule" and not want to play at all.

If I take a horse up to a mat who is rushing to get to this favorite place, and at the last moment insist that he back up before I let him go forward again, I could very easily end up creating some ambivalence towards the mat.

"If I try to step on it, will my handler release the lead, or will I collide with her hand and get bumped by the lead?"

That is often the question the horses face. If the handler has been lumping the approach to the mat and not teaching the component behaviors that create good mat manners, you can easily end up with a horse that simple tolerates being asked to stand on a mat. I want a horse who loves his mat! (See the Shaping on a Point of Contact DVD for details on teaching good mat work.)

Over the winter Robin gave me a very clear indication that he was eager to play with our cone-circle games. The barn where he lives is attached to the indoor arena. At night while I am cleaning his stall, I often turn him out in the arena. Robin likes to have the arena door open so he can monitor my progress as I do the evening chores. He stands at the threshold of the door waiting for me to finish. He could wander off around the arena looking for scraps of hay left over from other horses. He could walk out the arena door into the barn aisle to raid the hay pile or visit with other horses. He could make a dash for the barn door and have a romp around outside. He does none of those things. Instead he waits for me at the arena door. He wants me to finish my chores, get his mat and his cones, and to come play with him! I do love clicker-trained horses!

Once I was satisfied that Robin understood inside and outside, even when I expanded my cone circle out to a full lunge-circle size dimension, I changed my pattern. I got more cones and set out a figure eight with the mat at the join of the two circles.

Lots of non-clicker trainers use cones. Lots of trainers set the cones out in patterns, including figure eights. What made this more than just a standard exercise was the addition of the mat linking the two halves of the figure eight. Robin understood the pattern. He understood that whatever size the circle expanded to, whatever direction we set off in, his job was to take me around the circle back to the mat.

What emerged was a wonderful tool for working on a variety of training challenges. Suppose you are an insecure rider getting your confidence up by riding a horse who understands figure eights and mats. You know exactly what he is going to do when he leaves the mat. He's going to circle around the cones and return to a halt at the mat. He's not suddenly going to grab the bit and drag you off to the spooky end of the arena to visit with the other horses. He's not going to pick up a trot and dash off to the gate. He's not going to plunge his nose to the ground to eat grass. He's going to quietly, and calmly take you around the circle to return to his mat.

Suppose you're riding a new horse whose gaits you are just learning to ride. His trot is bigger, springier than anything you're accustomed to. It would be grand if you had someone to lunge you while you get used to his bigger movement. You don't want to shut him down while you are learning to ride him. But there's no one around to help you with that, so instead you teach him the cone circle figure eight with the mat in the middle. You leave the mat, ask for a trot, and what is he going to do? He'll pick up a nice trot, trot around the circle and return to a halt on the mat. You get the benefit of a great up transition, you only have to manage a few strides of trot, and then he's bringing himself into a soft down transition to stop at the mat.

Suppose you are teaching a young horse to jump, or you're gaining more confidence as a rider in your jumping ability. You could easily set out ground poles or a small cross rail at the far side of the circle. Your horse will leave the mat, head off around the circle, trot comfortably over the obstacles and return to the mat for a halt, click and treat. Again, he's practicing only the behaviors you want in a nice clean loop.

There are many more uses for, benefits of this pattern, but I'll leave it up to you to discover them.