House breaking and Magic Hands
By Alexandra Kurland
theclickercenter.com
This first part of this post refers to Panda, the miniature horse I trained to be a guide for the blind. For more on Panda go to the Panda Project section of this web site. The story I am sharing is a follow up on another post which described how a horse had used a behavior her owner had taught her to communicate back to her.
Great post Margaret. I have to add my own example of horses using behaviors we have taught them to communicate with us. This is a Panda story.
Panda is house broken. She has two cues: "Get busy" for urination, and "Park time" to defecate. Sorry for the topic for those of you who are reading this over breakfast, but Panda has provided us with a fascinating example of her ability to communicate with us using behaviors we have taught her.
I saw this behavior most clearly when Ann and I were traveling together on the toll roads here in the Northeast. We can't just pull over when it's convenient. We have to wait until we get to the rest stops to give Panda a relieving break. That means we aren't always on her schedule. It's like traveling with a toddler. When the next rest stop is another forty miles on, we don't always want to push our luck, or her comfort level. So we'll sometimes stop a little early, before Panda really needs a break.
Ann will take her out. Panda has an amazing bladder. She can always oblige on the cue to "get busy", but dropping a pile is another matter. It can sometimes take her five or ten minutes before she is successful. But she clearly knows what she is supposed to do. She stays on task, going through her pre-relieving cycle of behaviors. She turns in small circles, sniffs the ground, looks back at her flank, lifts her tail.
Especially in the winter when it's cold waiting for her, there's always the temptation to say that she really doesn't need to go, and to put her back in the car before she's dropped a pile. But then there's the worry that now that we've started the process she'll need to go before we can get to the next rest stop. So we wait, and she keeps working away at it, and eventually she produces a pile.
(For those of you who haven't been around minis of Panda's size, you mustn't picture that we now pull a full size muck tub out of the trunk of my car to clean up after her. Ann uses the same size plastic bag to clean up Panda's pile that used for her dogs.)
This relieving process illustrates beautifully the difference between a cue and a command. We can't order Panda to produce a pile. When she doesn't produce a pile right away, we can't say it's because she's being stubborn, or she's testing us, or she's showing disrespect. All we can do is give her the opportunity and reinforce her for trying. The rest is up to Panda, and it is quite remarkable to watch her diligently working away at it until she can oblige us with a pile.
But there are times when she really doesn't need to go. Panda has developed a wonderful way of signaling to us that she doesn't need her pit stop. She presses up against Ann's leg.
Pressing against the handler's leg was a behavior I taught her early on in her training. It's part of her version of "the grown ups are talking, please don't interrupt". When I was standing in line at the post office or chatting with neighbors, I would reinforce Panda for pressing her body against my leg. Panda now makes it her responsibility to keep herself attached to Ann. Ann doesn't have to worry about Panda swinging her rear end out into someone else's space. She can feel her pressing her body actively against her leg so she knows exactly where she is.
It's a behavior we taught in one context which Panda has transferred to another. When she really doesn't need to go, she swings over and presses up against Ann. The first couple of times she did this, we tried to send her back to focus on relieving, but she very determinedly swung back around to press against Ann. And when we put her back in the car, she was fine until the next rest stop. So now we believe her when she tells us that she really doesn't need to go.
This using a behavior we have taught in one context to signal us in another is something to keep track of. When you see instances of it with your horses, you should definitely share them with a post to the list.
"Magic Hands"
When I used to show Panda's training videos to groups, I would say that, of course, with the big horses we don't teach them to press in against us. We teach them to move over out of our space. That is true - at first, but I have started to teach the press-in-against behavior as well. It started with Robin. I taught him Panda's lock-in-to-my-body heel position, and then evolved that into "Magic Hands". Michelene just sent in a wonderful report on the success she's been having with Troubadour using the "Magic Hands" exercise.
Basically it's a targeting lesson. You're asking your horse to target, usually his shoulder, to your hand. Depending upon what your horse already knows, there are a number of ways you can initiate this game. With most horses you simply rest your hand on your horse's shoulder. You roll your hand so the heel of it presses against his skin, but beyond that, there's no making him go. When he takes a step, click and treat. At first you are on a very high rate of reinforcement. Every time your horse steps forward, click and treat. In between trials, you either take your hand off his shoulder, or you can scratch him while he eats. This later has the added benefit of pairing scratching with a primary reinforcer so you are increasing the value of scratching as a conditioned reinforcer.
At first you are following your horse, but gradually you change the game and have him follow you. If he starts walking too fast so you have to rush to keep up with him, you'll stop your feet instead. You may even pivot on your heels and take a step back to draw him to you. Your horse will begin to make the association. He gets clicked and treated when his shoulder rests under your hand, and it is his responsibility to adjust his pace and direction to stay attached to you. That's a great concept for a riding horse to understand. It is his responsibility to keep his body attached to yours.
It's a great exercise. In the beginning with a horse that is a non-foot mover it can be as exciting as watching grass grow, but the grass that's growing turns into great things. The liberty ground driving that's at the end of the "Why Would You Leave Me" DVD is essentially a "Magic Hands" exercise.
I call it "Magic Hands" because it does look like magic when its done well. There are no adversives to fade. If your horse leaves, you don't drive him back to you. You can invite him into your space, but you don't make him wrong for leaving. He returns because he chooses to, not because he has to. This gives the exercise a very different appearance from the standard locking on that evolves out of round pen training.
I'll be doing a DVD on it at some point. There are some handling details that are worth highlighting with the visual images. One of those details is what to do if your horse bends his neck prematurely as he steps forward. Many horses will do this. They'll take a step, but then they'll wrap themselves around you, negating their forward movement. Instead of passively allowing the head turning, you can reach forward with your free hand. If you are on the left side of your horse, your right hand will be on his shoulder. Your left hand will reach forward and gently press on his jowl to encourage his head to move forward.
Some horses are initially unsettled by the movement of your hand to their face. So you'll click and treat, click and treat as needed to get your horse comfortable with this gesture. Now you can ask your horse to walk forward with your hand on his shoulder, and make adjustments as needed in the orientation of his head. If he overflexes, you can encourage him to straighten out with a soft gesture towards his jowl.
You're going to look as though you are working your horse in hand - except there is nothing attached to his head. For a horse that worries about the lead or rein, this is a great step to insert. Now when you put him back on a lead, he'll be much more comfortable. Instead of your hand pressing against his cheek, you'll be able to slide along the rope. He'll accept the approach of your hand towards his face. I've used the "Magic Hands" exercise with horses who have become protective of their heads. It's a great transition step that melts away their worry.
Alexandra Kurland
theclickercenter.com
The "Yes Answer" Game
By Alexandra Kurland
Copyright 2007
Yvonne wrote
March 9, 2007
"I have been working with Don Carlos, my wound-up youngster, on shaping on a point of contact. I have the impression that this really worries him a lot. Instead of getting better, he becomes more and more grabby, worried and unsettled, despite the fact that I took him back to kindergarten (just standing quietly for touching the side of his mouth and other areas of his face/neck/withers). I try to keep my engergy level very low and relaxed, but he just doesn't seem to settle down. And his formerly nice, polite food-taking manners are deteriorating in the process as well."
Too Much Too Fast
When horses who are generally polite about taking food become grabby, this is a good indication that you are asking for too much too fast. And when you see an overall deterioration in the horse's behavior, that definitely means that you are asking for something that he doesn't understand. It may mean that the preparation leading to the particular lesson you are working on is not yet in place, there are missing steps, or something about the way you are asking needs to be modified. This is where a video camera can be enormously helpful. You will often spot details in timing that you weren't aware of when you were working with your horse.
It can also be enormously helpful to get a human training partner who is willing to play the part of the horse. They can give you verbal feedback about what the lead rope feels like from the horse's point of view. Little things can have an enormous impact. You said you are trying to keep your energy low. Perhaps to a sensitive horse like Don Carlos this means you feel absent or disconnected, the very opposite effect from the one you are trying to create. Getting feedback from a human can be wonderfully useful. Maybe there's a stiffness in your shoulder that feels heavy to him, or the slide up the lead isn't as smooth as you think it is, or you are using muscle instead of bone rotations so it feels to him as though you are shouting. All of these details matter. And even more, they matter under saddle. You aren't just doing ground work to teach your horse better balance and emotional control. You are doing ground work to refine your riding skills without having to put wear and tear on your horse.
I know you understand this. The question Don Carlos is posing is worth exploring. Here are some things to consider: Is this the right lesson at the wrong time? That can happen. For example, you could take a horse through the pre-why would you leave me lesson. Its a very simple lesson. You walk from cone to cone to cone, stopping at each cone for a click and a treat. You would think every horse would be able to do this lesson with ease, but not so. There are some who aren't ready even for the simplicity of this lesson. They need other steps before this one. But here's the scenario: the handler tries the lesson, the horse fusses and doesn't seem to be making any improvement, so the handler moves away from the lesson to something else. And then later she tries it again. This time all the good things that the lesson produces pop out. Her horse is a "pre-why would you leave me" superstar and progresses magnificently through the lesson.
So one possibility for Don Carlos is he simply isn't ready for the lesson you are working on, and you need to identify the preparation he needs before working on the lesson you are currently encountering difficulties with.
The "Shaping on a point of Contact" DVD is a complex lesson teaching many levels of refinement. At its core it is teaching the handler how to use the release of the lead to communicate. As the handler becomes skilled at this, very detailed balance shifts become possible.
The “Yes Answer” Game
So how do you explore this with a sensitive horse? You follow a simple rule of shaping: you find a place in the training where you can ask for something and get a consistent "yes" answer back.
You may have to go way back to very simple things to identify the "yes" answer where Don Carlos can experience consistent success. Here's where the short sessions of micro-shaping come in handy. (Refer back to a post I wrote after the Clicker Expo on micro shaping. I would have sent it to the list early in Feb.)
Your first step is to find something you can ask for that gives Don Carlos a 90% or higher success rate. How do you measure your success rate? Video tape your session. When you play back the video, record all the individual behaviors you see. What does he offer?
Suppose you see the following during a sixty second session:
He keeps his head still. Click and treat.
He grabs at the lead.
He looks away.
He sucks on his tongue.
He drops his head.
He keeps his head still. Click and treat.
He stamps his foot.
He bumps you with his nose.
He grabs at the lead.
He drops his head.
Ten behaviors, but only two clicks, so your success rate is a low 20%.
Once you've done this a few times where you really keep track, you'll be able to measure pretty accurately your success rate without having to go through the process of recording the session and counting behaviors, but its a very useful exercise to go through at some point in your training career.
Success Rates
So let's say, you are sitting at that low 20% rate given what you are currently doing. That's way too low. We want 90% or better. Suppose you've been sliding down the lead all the way to the snap, and that's creating the fussiness in Don Carlos. What could you ask for instead? Maybe it is as basic as asking the question: can I stand next to you? Essentially that is the "grown-ups are talking" question. Can Don Carlos keep his head still while you stand quietly beside him? Yes? Click and treat. You should be able to keep him on a high enough rate of reinforcement that he doesn't have time to offer any other behavior than the desired one of allowing you to stand next to him. Basic. Absolutely. But if that's where you need to go to get a consistent "yes" answer, that's where you go.
I refer to this as the "yes" game. You want to set this lesson up so it is so simple, so easy that everything Don Carlos does gets a "yes" answer. He feels completely successful. There are absolutely no "nos" either in the form of a correction, or even in a lack of a click. This is a lesson he is going to succeed in and there aren't going to be long gaps where he has to puzzle his way through to the next right answer. This is an easy game to win. It's like the opening questions on a quiz show. They are so simple everyone gets them right. Hard comes later.
When you've had a short cycle of high success, give him a break. That doesn't mean put him away. It means go do something else with him. Targeting is a great lesson to balance out another lesson that has some emotional intensity attached to it.
Oliver
We used this with Oliver at the last WA clinic. Oliver is the horse featured in the Shaping DVD. Keri has progressed from the stationary mat work to the beginnings of three-flip-three. Oliver is at an age where he can get puffed up and excited especially when he is away from home. Add in working on the precision of lateral work and some fussing at his lead popped out. To diffuse his emotional response we used a strategy that balanced the more challenging exercise of the lateral flexions with something that was familiar and easy. We set up a circle of cones. We reinforced Oliver for touching the cones. We built up that behavior with a high rate of reinforcement so he was eager to touch the cones. Then Keri used the rope handling skill of the rotation of her fully extended arm to ask Oliver to leave the cone.
The arm rotation allowed Keri to be very clear, but not forceful. When Oliver lifted his head, he earned a click and treat. Keri then slid down the lead and asked for the first step into a lateral flexion. Click and treat. She repeated this a time or two, then let Oliver move forward to the next cone in the circle.
By asking for something hard, and then balancing it with something familiar and easy, Oliver stayed very emotionally settled. He stopped fussing at the lead and gave Keri some beautiful steps of a lateral flexion. He was also learning to leave things he wanted. Leaving the cones was setting him up for leaving things such as buckets of grain or tempting spring grass. And he was having a riding lesson, learning how to follow the feel of the rope. "Everything is everything else" certainly applied to this lesson.
Keeping the Balance Between Hard and Easy
So back to Don Carlos. You need to find a step that gives you a 90% or better success rate. Work on it for just a few clicks. For novice clicker trainers I suggest they measure out a set number of treats, or do it by time so they aren't asking for too much. You'll probably be able to gauge this without needing these set references. Click/treat, click/treat just a few times. Get your success. Then shift to something fun and easy for Don Carlos. For example, let him touch a target. The opportunity to touch the target will begin to reinforce the harder work of accepting you in his space. And since emotional effects travel backwards, the all-is-well emotions associated with the easy lesson will begin to trickle over into the harder lesson.
When you've had consistent success over four or five trials in a row, increase slightly the level of difficulty. Don Carlos will tell you how chunked down you need to be and how fast you can move on.
As you begin to progress to the lead and to picking up a connection, and especially when you go to the precision work illustrated in the Shaping DVD, if his fussiness returns, you may want to consider if there is a physical issue going on. He may have some old injury that he is protecting.
We saw this with another of the horses at the WA clinic. Debra has to be very tactful when she works her young horse June Bug on the right side. June Bug is the beautiful paint pony in the riding book who is shown having one of her very early rides. Shortly after those photos were taken June Bug was over tranquilized during a dental exam and crashed head first to the ground. She ended up injuring her spine, and she was in considerable pain for a long time. She is recovering well, but she is still protective of her right side.
In this case we can trace the behavior back to a known injury. But with so many horses that information is a missing link. It is always worth considering if there is a physical reason behind any unwanted behavior you see. Is your horse trying to communicate important information? We need to be respectful of that communication and recognize the behavior for what it is - not disobedience or a lack of respect, but our horse's honest way of telling us that something is wrong.
Magic Hands
Jord-Ann wrote about her success using the "Magic Hands" exercise with one of her stallions. It's a great exercise for sorting through these emotional issues. Yes, you can view it as a simple targeting exercise, but it is targeting with the horse maintaining a physical connection with the handler. Once this has been accepted, it becomes much easier to add the lead rope back in. The horse doesn't feel as overwhelmed by all the information - some of it conflicting - that is presented by both your close proximity and the lead. He can listen to the your requests down the lead without feeling as though you are shouting down a megaphone.
I don't know how much, if any of this applies to Don Carlos, but hopefully it will give you a few directions to explore as you sort this through. The questions he is posing through his fussiness are certainly worth pursuing. I want horses that are comfortable staying with me at liberty, that can work out at the end of a long lead, and that know how to keep slack in a lead. But I also want horses that accept me in close for the work in hand and will allow me to make a direct connection to their head and spine.
Connections
At the WA clinic we worked one of the other horses through flexions at the halt. In the morning "t'ai chi" exercises we looked at the human equivalent of the lesson we were presenting to this mare. I first had people check their range of motion by lifting one knee up to see how high they could lift it.
Then we worked on the equivalent release of the spine in humans that we were asking the mare to give us. When people found this release, they discovered they could lift their knee up significantly higher. Ah ha! Think about how that translates into the gaits of a horse. And indeed that's exactly what we saw in the mare. She was already a gorgeous mover, but she gained even more lift as she went through these exercises.
That's what we're after. Beautiful, sound horses, and gaits that feel like heaven to ride - which is why these questions our horses pose are worth pursuing. Sometimes it takes some creative thinking to find the answers, but when horses ask questions, it is always worth listening to them. You never know where the process of finding answers will take you, but when you follow the principles of good training, they allows lead to good things.
Alexandra Kurland
theclickercenter.com
Whale Stories
By Alexandra Kurland
Copyright 2007
I've been absent from the list for a bit. I won't try to catch up with all the posts that have been written, except to say a general thank you to everyone who kept the conversation going.
I'm just back from the Cleveland Clicker Expo, and my head feels stuffed to the rafters. You'd think after six or seven Expos, some of the lectures would start to sound stale, but not so. It doesn't matter how many times I hear the lectures on clicker basics, there is always a new twist, a new angle that emerges.
The "ah has" from the Expo are too many to list, but I will share some highlights. On Monday after the Expo we had a Faculty meeting to discuss plans for next year. The meeting is something we all look forward to because Ken Ramirez generally brings video clips of his current training projects. Ken is the Director of Training at the Shedd Aquarium in Chicago so he's always got something fun he's working on. This year's video highlights included some work he's been doing with beluga whales at an Aquarium in Japan.
Bubble Rings
Belugas apparently blow air rings the way some smokers blow smoke rings. The behavior is not well understood and there is no clear predictor of when it will occur. You could watch for days and never see the bubble rings, but by chance a Japanese film crew happened to be on hand at one point when the whales were blowing bubbles. They got some beautiful photos of the rings. The zoo started using the images in their advertising, but then of course the public wanted to see the whales blowing rings. So the zoo needed to get the behavior on cue.
The whales were in tanks with glass side walls so people could watch them swimming under water. The trainers wanted to be able to cue the whales to blow air rings at the audience. The problem was they didn't know how to get the behavior in the first place. They couldn't rely on capturing the behavior since it occurred so irregularly. So getting bubble rings on cue was the training problem the zoo asked Ken to help them solve.
Ken's solution illustrated everything I love about clicker training - the weaving together of a solid understanding of the science of clicker training with good training technique and creativity. He started with a behavior the whales had solidly on cue - spitting water. They would come up to the edge of the tank and on cue spit water at the trainers.
So Ken started by asking the whales to spit water while they were under water. The whales readily transferred the behavior. The movement of their lips for spitting was similar to that of the air blowing behavior, but spitting water under water didn't create the same charming visual effect. It was however a good beginning. The next challenge was getting them to spit air.
When the whales blew air rings on their own, they would first swim up to the surface and get a mouthful of air. So the next task was to get the whales to hold air in their mouths. This was the truly ingenious part. Ken used his own respirator. He cued the whales to open their mouths and then inserted his respirator so they got a big gulp of air. The whales seemed perfectly comfortable with this and held the air obligingly in their mouths.
So the next step was to give the "spit water" cue and see what happened. Ken gave the cue, and the whales responded by blowing an air ring. It wasn't a very good air ring, but the whale was definitely spitting out his air bubble.
Timing and Microshaping
To get a better formed ring, the trainers had to get better in their timing. The trainers Ken was working with tended to click too late. They were clicking when they could see the finished air ring, but that's not the information that was most important to the whales. They really needed to be bridged just as they were forming the bubble ring. So, Ken started looking at shaping micro movements. He said he got the idea for the solution in part from presentations he'd seen at the Expo. Kay Lawrence's discussions about shaping micro-movements in dogs got him thinking about doing something similar with the whales.
To learn what the trainers needed to click, Ken had them study video of the whales blowing bubble rings. He slowed the video way down so they could see frame by frame what the whales were doing. Ken had all the trainers watch the video over and over again until they could recognize the changes and begin to see and respond to them in real time. When he first showed us video of the whales blowing bubble rings, it was hard to see what they were doing, but slowed down it became very clear. You could see the whale form a round circle with its lips, then pull them in and at the same time the shape of its mellon (it's forehead) distorted. It creased in and a micro second later, the whale produced the bubble ring. It floated through the water, a beautifully formed ring of bubbles, as if the whale were blowing kisses at the camera.
At this point Ken could reliably cue the whales to blow bubble rings, but he was still supplying the air. The next step was to have the whales get their own air. As a preparation for that he had the aquarium's trainers include bubble blowing in the whales daily training sessions. When he returned a couple of weeks later, the whales had become practiced bubble ring blowers. So Ken did a session with them where he gave the cue to blow a ring, but did not first give the whales any air. The first whale spit water, but Ken did not bridge and reinforce him. Instead he cued him again. The whale hesitated, then went up to the surface, got his own mouthful of air, came down and very deliberately blew it straight at Ken. Success.
Ken showed some enchanting video of the whales engaging with the spectators. They blew their bubble rings straight at the people so the rings floated towards them before bursting against the glass. The rings looked like fairy rings floating in the water. The whales began adding their own flourishes to the behavior, including blowing bubble rings out their air holes! They would also blow a ring and then do a back flip so they could "catch" the ring with their tail. It was pure delight watching them.
The Horse Connection
So why am I sharing this whale story on a horse list? Apart from the fact that it was just so enchanting, it was a wonderful illustration both of the creativity that is so much a part of clicker training, and of the value of micro-shaping. The analysis of the whale's movement using slow motion video is something we should all be taking note of.
Once you get past the basic lessons where it is easy to see what your horse is doing, seeing detail matters. As clicker trainers we like to think of ourselves as positive trainers. We click and reinforce the behavior we want. We understand that the opposite of positive reinforcement is no reinforcement. It is not correction. But are we really giving our horse a truly positive training experience just because we are using positive reinforcement?
Ken gave a presentation on advanced training concepts. He prefaced his talk by defining what he meant by a beginner and an advanced trainer. A beginner clicker trainer reinforces desired behavior and ignores unwanted behavior. It's that simple. If a novice trainer needs a more advanced tool, Ken said he shouldn't be working with that particular animal. He was speaking as the director of training of a major Aquarium where he routinely has seventy trainers on his staff. Part of his job involves teaching new trainers how to work with what are potentially very dangerous animals.
We are also working with what are potentially very dangerous animals even though we are working with a domestic species. Just as they do with zoo animals, I recommend that horse owners begin with protected contact. This gives a novice clicker trainer the breathing room she needs to reinforce her horse for touching targets, putting his ears forward, backing up, etc., while at the same time remaining non-reactive to unwanted behavior. Learning how not to react to things you don't like can be a hard skill to learn, especially if your horse is presenting dangerous, unacceptable behavior. Protective contact helps. It lets you get your "sea legs" while you are learning the skills needed to be an effective clicker trainer.
That's all well and good, but its the second part of Ken's statement that can get a lot trickier. Its easy to say that a novice handler shouldn't be working in a situation that requires advanced tools. The reality is horse owners are frequently working with horses that are not well matched to their experience level. That means the foundation work is all the more important. Protective contact is a wonderful thing. While you are developing your motor skills, perfecting your timing, and learning the basic concepts that go along with clicker training, it makes sense to keep things simple. That means focusing on the foundation lessons. The more solid you become with the basics, the less frustration you and your horse will encounter later.
Positive Trainers
We are clicker trainers. We want to be positive with our horses, but we need to separate out intent from tools. If you are a lumper, if you are raising your criteria too fast, if you are inconsistent in your timing, if you do not reinforce after every click, you can easily end up with a very frustrated learner. You need to look at the effect on your learner, not just the tools you are using, to decide if the training experience is a positive one.
Jesus Rosales-Ruiz showed a wonderful example of this. One of his students was looking at the difference in response between following every click with reinforcement versus not reinforcing after every click. In the video clips he showed where the dog was being treated after every click, the dog responded promptly to the cues. He was animated. His tail was wagging.
When the same dog was clicked but treated after only fifty percent of the clicks, his body language changed dramatically. The tail stopped wagging. The handler had to repeat cues, and in one very dramatic clip, the dog gave the handler a "time out". He left the training area completely.
Jesus said that the dogs in the study belonged to a friend of the student. Normally when she went over to visit, they greeted her enthusiastically, but after she did the experiment where she clicked but did not treat, the dogs left the room when she came into the house. The experiment didn't just suppress behavior during the trial. It had a global impact on the relationship.
The Gorilla in the Room
Kathy Sdao shared with us some video clips produced by researchers who study visual cognition. I almost hate to describe one of the clips she showed. Once you know what to look for, the clip doesn't have the same impact. But it is so relevant to this topic I have to describe it. The clip shows a group of people passing basketballs back and forth. Half the group were wearing white jerseys. The other half were wearing black. Kathy instructed us to count the number of times the ball left the hands of someone wearing a white jersey.
Everyone watched and counted. At the end of the twenty second clip she asked for numbers. Fifteen, sixteen, nineteen. One very observant person called out fifteen straight passes, plus four that bounced. That's showing off - or would be except for the one thing that very focused and observant individual didn't see - and that was the person in the gorilla suit who mid way through the clip walked across the basketball court. No one in the room who hadn't heard about the clip before hand saw the gorilla in the room.
That's totally normal. Our brains filter out an enormous amount. We see what we are focused on, what we are directed to see, what we want to see.
Think about all the ways this is significant for our horses, our dogs, our family members, our work.
A trainer focuses our attention on one aspect of the training, and we fail to see the problems emerging in different areas.
Our horse fails to see the cues we are giving just as the people watching that video failed to see the gorilla in the room. We think the animal is being deliberately disobedient, when really it is just a problem in perception.
We're training one criterion, and we get so focused on that that we fail to see some other elements that are falling apart and need our attention.
I'm sure you can find many other ways in which this video clip relates to your horse. For me one of the most important is this: we want to think of ourselves as all positive trainers so we miss the signs of stress when we are lumping our criteria and making the lesson too hard for our learner.
Micro-Shaping and Positive Training
I wasn't able to get to Kay Lawrence's presentation on micro-shaping at this Expo, but I saw it last time and I had many conversations with her through the course of this weekend. I wish everyone could see the video she shows in her talk of the dog she is shaping to put his foot on a stool. In the first clip she deliberately lumps criteria. She clicks the dog for putting his foot on the stool, but she does not click for micro movements. Over a sixty second trial the rate of reinforcement in the dog is very low. He offers a few paw touches, but he is not very focused on the task and shows signs of stress.
In the next clip Kay micro-shapes. She clicks for any lift of the front paw. It doesn't have to be oriented to the stool. Even a slight lift of the paw earns a click and a treat. Over sixty seconds the rate of reinforcement goes way up as does the dog's focus and enthusiasm.
So how do you learn to see the tiny shifts in balance that present clickable moments? Ken's presentation on the beluga whale showed us one way. You video tape your animal and slow down the motion until you can see all the tiny weight shifts that create the overall movement. You learn to spot the movement that occurs just before the clickable moment so you are ready to click.
Timing Exercises
Okay, so you can see the weight shifts, but that doesn't mean you'll be able to click them in real time. How do you sharpen your mechanical skills so you are clicking on time? Theresa McKeon in her presentations on Tagteaching gave people some great exercises for sharpening their timing. She had people click when she opened her hand. She started out slowly, predictably so most people were successful, then she got faster and trickier. Sometimes she held up all five fingers. Sometimes only one or two fingers. Sometimes she started to open her clenched fist, but didn't. The number of missed clicks increased with the difficulty level. It's a great game to play with a friend.
So here's the question: how good is your timing?
Can you have a friend bounce a ball and click each time the ball hits the ground? Other timing games she suggested were watching television commercials and clicking each time there is a scene change.
Positve Trainers
I hadn't intended when I started this post to give you an Expo Highlights report. I really just wanted to share Ken's work with the belugas, but everything truly is everything else. And it ties in so well with something I've been thinking about a lot lately which is this phrase "all positive trainer".
When people describe themselves as clicker trainers, they often say they are all positive. But what does this really mean? And is it possible? If you are shaping and your animal is getting frustrated because you are lumping too many criteria together so the rate of reinforcement is low, are you being all positive? You are clicking and reinforcing correct responses, but your animal is not having a positive experience.
When you add in some clues in the form of negative reinforcement, are you no longer being all positive even though your animal is now more successful and is clearly happier?
I want my horses to have a positive learning experience. To that end I mark behavior with a click and a treat. But to judge if my horse is having a positive experience I must look not just at what quadrant of operant conditioning I am using, but at the experience my horse is having. I can be using what I would consider to be all positive methods, but if I am lumping criteria, or neglecting to reinforce after I click, my horse may be having anything but a pleasant experience.
When you are defining yourself as a positive trainer you must look at a broader picture than just your immediate training plan. What is your intent, your underlying belief system about training and the animals you work with? What tools, concepts and strategies do you use in your training? What tools, strategies and concepts do you avoid? What is the effect of your training on the learner? How good are your mechanical skills? Are you experienced enough to be working with this animal on this lesson - or should you be working on something simpler?
If you are clicking and treating but your horse is showing the kind of diminished response that Jesus saw in his study on treatless clicks, does that mean that clicker training doesn't work or that you need some other tool? Does that mean its okay to correct your horse for unwanted behavior? No. It may simply mean you need to go back a few steps and work on something easier for both you and your horse. There is always a smaller, simpler, easier step you can go back to.
Clicker trainers focus on what they want, and they remain non-reactive to unwanted behavior. But this doesn't mean they don't see the "gorilla in the room". They see the unwanted behavior, but they manage their horse and the environment so that their attention remains focused on marking and reinforcing desired behavior. If you are new to clicker training I would urge you to be on the lookout for the "gorillas in the room". That means paying attention to what your horse is telling you about his comfort level. If he is getting frustrated, take a step back and evaluate your training. Are you asking for too much too fast? Are you inadvertently punishing behavior because your timing is off and you are raising your criteria too fast?
Helix Fairweather gave a great presentation on keeping records. Finding some way of charting progress, whether it is a journal, or the type of record keeping she suggests, can be a huge assist to your training.
Throughout the weekend presenters stressed the importance of a good foundation. Clicker training is a powerful tool. Those beluga whales blowing bubble rings are proof of that statement. Ken's presentation on advanced training concepts showed that clicker training includes a rich storehouse of useful problem solving techniques, but at the core of them is developing good basic skills. Whether it is a horse, a dog, or even a beluga whale, learning good foundation skills is the key to becoming a good solid clicker trainer. The better you are with the basics, the more truly all positive the experience will be for your horse. Add creativity and experience to the mix, and over time that's how great clicker trainers emerge.
Alexandra Kurland
theclickercenter.com
Three-Flip-Three DVD
By Alexandra Kurland
Copyright 2007
I just finished another DVD. This one is titled: "Three-Flip-Three: Understanding Lateral Flexions". There have been so many questions on three-flip-three, I know this has been a long awaited DVD.
Here's an overview of the material covered on the DVD:
The DVD begins with a short review of the "Why would you leave me?" lesson. That lesson, described in detail in the "Why would you leave me?" DVD, uses cones set out on a circle to teach your horse to stay with you and walk on a loose lead. The handler learns to recognize when her horse is connected to her, and when he is leaning in on her, or drifting away from her. As the horse connects more and more to his handler and keeps himself on the circle, the beginnings of lateral flexions evolve.
The "Three-Flip-Three" DVD shows you how to turn this lesson into a counted pattern that gives you the ability to separate from the circle without losing the good balance you have created. Like a gymnast on a balance beam, your horse learns how to keep himself aligned even as the patterns you are asking for become more complex.
The "Three-Flip-Three" DVD eavesdrops in on a lesson where the handler is learning how to move from the "Why would you leave me?" circle exercise to the counted pattern of "Three-Flip-Three". Her questions -the places where she tripped over her own feet, or couldn't figure out how the pieces of the puzzle fit together - are very much the questions many people will encounter when they first explore this lesson. You'll see not the finished, polished product, but how that product evolves, shaping layer by shaping layer.
The DVD also includes some great "t'ai chi walk" analysis of three-flip-three as the handler sorts out the pattern in her own body so she can understand what she wants her horse to do.
The final section of the DVD looks at practical applications for lateral flexions. Lateral flexions are the key to good balance. They help develop great gaits, they maintain soundness, and they build emotional stability. They are also a great tool for such everyday tasks such as helping horses to get past scary objects, or to ride in close to another horse, or to maneuver around obstacles out on the trail.
This DVD begins an exploration of lateral flexions, what they are, how to teach them to your horse, how to adjust them when your horse falls out of balance or overflexes, and how to use them for safety and performance.
This is the fourth DVD in a series I am producing to accompany "The Click That Teaches: Riding with the Clicker". It is intended to be used in conjunction with those other DVDs and the book.
Alexandra Kurland
theclickercenter.com
More on Three-Flip-Three
By Alexandra Kurland
Copyright 2007
Future DVDs: The Lesson Plan
I was asked in a recent post if the new Three-Flip-Three DVD contains any riding. The answer is this DVD focuses on ground work. It explains what three-flip-three is, shows you how it evolves out of the "why would you leave me?" lesson and then shows you how to manage it on the ground. There isn't time in two hours to include riding as well. I know for some of you who are itching to ride, you're thinking: "Not more ground work! When is she ever going to get to riding!?" Remember riding is just ground work where you get to sit down. If you understand this exercise thoroughly on the ground, transferring it to riding is easy. And if you understand it first as a ground exercise, the riding will make much more sense.
So here's my plan for the next few DVDs just so you know where all of this is heading.
The next one, which I will be starting in the next couple days, is on hip-shoulder-shoulder.
Three-Flip-Three and Hip-Shoulder-Shoulder are cornerstone exercises. Understand those and all sorts of other things fall into place, so I want to be sure that I cover them well. The DVD I just finished will not be the only DVD that I make on 3-Flip-3. There is much more that needs to be said about that exercise, but it's a good beginning. What the new DVD will do for many of you is simply confirm that you are on the right track. You'll watch it and say: "yes - that's what my horse is doing. I'm getting it right." That's important confirmation - especially when you are working by yourself.
After the HSS DVD (all these initials!), I'll move into the riding. The first DVD will be on the mounting block lesson and a simple single-rein-riding lesson - riding around cones. I described that lesson in a post a couple of months back.
The mounting block lesson is another one of those key pieces. I use the mounting block to confirm that the horse is ready to take responsibility for a rider on his back, and that he will soften and yield his hips if needed. I treat it as a safety check lesson. Before a rider puts her bones on a horse's back, I want to know that it's safe to do so.
After the mounting block DVD I'll do one on the mechanics of single-rein riding. This will be without horses. We'll have riders sitting on saddles on saddle racks, and I'll go through the "t'ai chi" mechanics of the pick up of the reins. I've got some great footage of riders exploring the mechanics which illustrate well the process. This lesson will focus just on the rider. With the horse out of the picture, it is easier to see what the rider is doing. We have a second person at the horse's end of the reins so the "rider" gets feedback as changes occur.
With the mechanics of the rein pick-up in place, I'll move to the safety lessons of yielding hips and head lowering. The mechanics for these requests are tied in very much to the "t'ai chi" exercises illustrated on the recent "T'ai Chi Rope Handling Exercises" DVD. Everything is everything else.
The rein handling DVD will lead into riding the "Why Would you leave me?" exercise, followed by three-flip-three, and hip-shoulder-shoulder.
From there I can move on to more complex exercises, but this will give everyone a solid foundation.
It's a lot of hours of DVDs to watch (and a lot of hours of DVDs to make!). I wish it could be done much faster. There are many of you who don't need all the detail these lessons go into, but all you have to do is skim through a couple of days posts on the various lists to know that there are others who will welcome all this detail and then some. This approach to riding is not standard issue. We aren't surrounded by others riding in this way, so we don't have pictures of what it looks like, or an understanding of the steps needed to train in this way. Once you understand the system, it's actually a very simple way to ride, but because most of us were taught very differently, it seems complex. If this were the norm, it would seem as simple as it truly is.
I work on the DVDs in between other commitments so they come out at irregular intervals. That means you may be ready to move on before the DVD that you need is ready. The riding book gives you the overall road map, and certainly the posts on this list are great for filling in gaps. If we start using youtube that may help cover lessons I haven't had a chance to get to yet.
Three-Flip-Three
One other quick comment relating to three-flip-three. If you've been working on the WWYLM lesson, you've got the beginnings of 3-flip-3. I do hope as you watch the DVD you'll be thinking: "I'm getting this. That's what my horse looks like." But there is a caveat here and that's that there is no finished end product for 3-flip-3. I often equate 3-flip-3 to a musician practicing scales, or a ballet dancer taking her daily class at the bar. 3-flip-3 serves a similar purpose in your riding foundation. This is where you check in with your horse to see how he is feeling on any given day. Is he stiff or limber? Is the drill team lined up and marching in sync or is he wiggling all over the place? How much lift and engagement do you have?
You are never saying: "okay this is as good as this can get. I'm done now with this lesson. What's next?" Once you understand how to generate 3-flip-3, it becomes a tool you use to explore, refine, adjust your horse's balance. (There's yet another DVD -they're as endless as this exercise.) 3-flip-3 creates performance- it is not in and of itself the finished performance.
Performance Cues and Balance Shifts
When I first heard Kay Lawrence talk about performance cues, I went "but, but, but . . . the things we work on, don't have finished end products that you can box up and package with a tidy cue." I now understand that Kay and I are much more on the same track with our training, but this is still something to think about. There is a tendency, especially when you first start clicker training, to train certain behaviors in a way that I would regard as trick training. Let me see if I can explain this.
It's normal in the beginning to regard backing up, stepping forward, moving the hips over, etc., as discrete, separate behaviors. In that case you ought to be able to put a cue on each one. You ought to be able to say: "back" and have the horse back, or "step up" and have the horse come forward. And absolutely you can teach this to your horse. My horses have come and back up cues which are attached to requests for discrete behaviors. But that is not the core of their work.
Backing and coming forward are connected movements. They are related weight shifts. I want my horse to become very good at shifting his balance, redirecting his energy in any direction I ask for. So the underlying skill I am teaching is the redirection of energy. That's what I want to bring under stimulus control. Not the results of the redirection of energy - but the redirection itself. So if my horse is backing up and has backed up as far as I want, will he shift gears and come forward on request? Can he rebalance quickly, easily?
That changes the initial approach to stimulus control. If you spend a session clicking and treating your horse for going sideways, it shouldn't come as any surprise that the next day when you work with him, he'll be offering sideways. You can treat this as a discrete behavior, like sitting for a dog, and go about putting the behavior on cue and then establishing stimulus control for that cue. Or you can look at it from the perspective of weight redistribution.
Redirection of Energy
The principles of training say that for every exercise we teach, there is an opposite behavior we must teach to keep things in balance. That brings you to redirection of energy. You don't think about shutting the lateral steps down. You think about redirecting your horse's energy so the drill-team balance shifts him out of the lateral work. You can do this by over-rotating the hind quarters into hip-shoulder-shoulder and redirecting him into backing. Or you can use the front end to redirect him. You can straighten out his head and neck so that you can once again redirect the energy into a rein back. From the rein back you can go forward, sideways, or into a halt.
I'm probably going to get myself tangled in this next analogy but think of quantum mechanics. You can use the theory where energy comes in packages or the one where it travels in waves. Packages would be lumps of behavior. In dogs it's: "sit", "down", "roll over". In horses it's "back up", "step up" "whoa".
I want those goal behaviors in my horse, but I create them not as separate packages of behavior, but as a continuum of weight shifts. The better your horse becomes at responding to the weight shifts you are asking for, the lighter and more like heaven he will feel. As I said earlier this is a process that has no finished end result, just many milestones where you step back in amazement and delight as you admire your beautiful horse.
This has turned into a very long answer to a very simple question that could have been answered with one word. Except that one word would have been misleading. Riding and ground work are intertwined. The more I explore ground work, the better my riding becomes and vice versa. That's because at the core of both is this weight redistribution. Step by step in the DVDs I am sharing with you my current understanding of this process.
Every night when I am in the arena and I look across at Magnat at age twenty-nine looking still so very beautiful, I know it is worth all the time and trouble it takes to learn this work.
Alexandra Kurland
theclickercenter.com
Redirection of Energy and Chains
By Alexandra Kurland
Copyright 2007
> Alexandra Kurland wrote:
"Backing and coming forward are connected movements. They are related weight shifts. ... That's what I want to bring under stimulus control. Not the results of the redirection of energy - but the redirection itself."
>
> Pam writes:
“ Ah, this has really set off a bell in my brain. I think this is describing a point of struggling in my understanding. That step of moving from clicking for discrete behaviors vs weight shifts that happen in that behavior.”
Stimulus Control
Bells going off is good. I love the expression you used: "a ballet of weight shifts." Clearly it's time to go out to the barn and ask your horse what he thinks of all this.
Here's another way of looking at the weight shifts. When I first learned about cues and bringing behavior under stimulus control, you taught each behavior as a separate entity. You taught "back" as a single, discrete behavior. You got the horse to back and then you started to put the behavior on cue. The problem with this approach is it raises the question of what to do with off -cue behavior.
If you reinforce your horse for backing, you are going to have your horse offering the behavior in the hope that it will get you to play the clicker game. Your horse doesn't at this stage understand about cues. He just knows that in the previous session backing was a good thing. But now you are working on stimulus control so you aren't going to reinforce off-cue behavior. Your horse hasn't read the manual. He doesn't know how this part of the game is played, so he's trying his hardest to get you to click. He's not just backing a step or two. In his frustration he's backing twenty feet. How can you not reinforce that brilliant effort? But if you do, you've just blown your stimulus control, not to mention the havoc you're wreaking with your reinforcement schedules.
Teaching Behavior in Pairs
A better way, one that reduces frustration for both the learner and the trainer, is to teach behaviors in pairs. You teach over and under, left and right. Not "over", and then in a separate session "under". Not "left" today and "right" tomorrow. Not all behaviors you work on will have natural pairs, but many of the movement exercises do. So you teach "back" and then balance that with "come forward".
To get to the "ballet of weight shifts" another way of saying this is I want fluency. I want to be able to ask for back and have my horse respond without any hesitation. Then I want to be able to ask for forward and again have my horse respond without any hesitation. That means he has to understand the cues for both behaviors, and he must maintain his balance so he can respond promptly to them.
So here's a little game I can play with my horse. Let's say I've just given him a treat. At this point I can ask for either back up or go forward. He doesn't know which, so he has to be prepared for either.
I ask for backing. He gives me a prompt response. Click and treat.
I ask for backing again. He gives me a prompt response. Click and treat.
I ask for come forward. He gives me a prompt response. Click and treat.
I ask for backing. He gives me a prompt response. Click and treat.
I ask for come forward. He gives me a prompt response. Click and treat.
I ask for backing. He gives me a prompt response. Click and treat.
I ask for backing. He gives me a prompt response. I ask for come forward. He gives me a prompt response. Click and treat.
I've just built a small chain. I took two behaviors and strung them together using my cues.
And by the way, how many of you inserted the words click and treat between the second and third sentence in the highlighted response? If you were reading quickly, your brain probably filled in those words. You were so used to them being there, you put them in automatically. Your horse will have a similar response. He's so used to saying "yes" to you, to responding promptly to each cue, that he'll be offering the behavior before it's fully registered that you didn't click him for backing. By the time he's noticed, you'll be clicking and treating the second behavior, so the overall effect is still one of success.
Repeat this whole sequence, occasionally mixing in small chains, until your horse is comfortable with the process and understands that he will often have to perform multiple behaviors all for one click and treat.
Behavior Chains
There are two types of chains. One type, the kind I've described here, has as its links behaviors which have cues attached to them. When you chain these behaviors together, the cues become reinforcers for the behavior that preceded them.
This is such an important part of clicker training because it allows us to build complex performance without having to be constantly clicking and treating. But it is initially not all that intuitively obvious. I think that's because for so many of us cues are not positive reinforcers. The cues we use have ambiguous meaning, so we struggle with the idea that they can be used as reinforcers.
The most common example of an ambiguous, poisoned cue is your own name. Think back to your childhood. When your parents called you by name, did that always mean good things were about to happen? Play back the sound of your parents calling you by name. Just thinking about it may give you a tense feeling in the pit of your stomach. That cue certainly carries with it mixed emotions. Sometimes they were calling you to come for a treat and sometimes for a scolding.
So now you are trying to wrap your mind around clicker training and you're being told that a cue can reinforce the behavior that precedes it. Intellectually that makes sense, but it doesn't feel right which makes this concept a struggle for many people to learn and use well.
But let's go back to the example of asking your horse to back up and come forward. Every time you asked him to back up and he met your criterion, he was reinforced with a click and a treat. You've built your expectations in a methodical, clear way, so he's had enormous success associated with your cue to back. When you give the cue, he knows good things follow.
The Emotional Effect Travels Backwards
In classical conditioning the emotional effect travels backwards. When you start to attach cues to behavior, you are creating classically conditioned pairings. Backing, a difficult behavior for a horse to perform, is beginning to be something he really enjoys doing because it has such a strong link to food. So when you give the cue to back, it is now linked with positive associations. "Back", the cue, produces the behavior which leads to goodies. The good feelings associated with eating wash back over the preceding events.
The same thing is happening with come forward. "Come forward", the cue, leads to the behavior which leads to goodies.
Your horse responds promptly to both behaviors. So when you ask him to back, and then immediately follow it with the cue to come forward, he responds almost automatically to the cue to come forward, and when he does, click and treat.
When he backed up, he created an opportunity to come forward, and coming forward led directly to the click and treat. So coming forward reinforced the backing. Back up and the "green light" will appear that leads to the goodies.
This type of chain is maintained by the cues for each behavior. You can build chains that always run in the same direction, that have a set order. Here is where backchaining is so important. You normally teach the last link in this type of chain first so the animal is moving in the direction of the best known, most highly reinforced behavior.
You can also have chains that can go in any direction, where you can mix up the order of the behaviors in the chain. I could ask my horse to back up and then come forward. Or I could ask my horse to come forward and then back up.
There are also chains where you give one cue to start the sequence and the completion of the first behavior in the chain signals the start of the next behavior. Fetching is a great example of this type of chain. You throw out your fetch toy and then give whatever signal you use for your horse to go get it. Your horse leaves you, goes to the object, picks it up, returns to you, and releases the object to your hand - click and treat. You had to teach each of these steps of the chain, but then you merged them to form one long sequence of connected behaviors. Again backchaining is a useful concept for this type of chain.
As you can see from the examples I've given, chains can be very simple. All you need are a couple of behaviors that have cues attached to them and you can form simple chains. If you start thinking about chains, you may find that you have already taught your horse several very useful chains using the clicker.
What is of particular interest for riding are the chains that are maintained by internal cues and can run in any direction. Can you think of examples with your horse where you have the behaviors occurring in a different order within the chains? Can you chain three or more behaviors together, and again be able to change the order around?
Once you have simple chains, you can begin to chain chains together. That's where things really get fun and you start wowing people with your horse's performance.
Alexandra Kurland
theclickercenter.com
Behavior Chains
By Alexandra Kurland
Copyright 2007
This post was prompted by a series of questions on the list about behavior chains. There was confusion over both the definition of a chain, and how chains were built. This post addresses several of the qustions raised in these other posts.
Building Behavior Step-By-Step: Start Simple
There have been a lot of questions about chains. My general advice: Don't over think behavior chains. You all have many behavior chains already in place. Some have become so subtle that they don't feel like chains at all, as Jord Ann described. Some are more obvious because they are in the early stages of development and the cues are very obvious.
Learning about chains is very much like a child learning how to read. You don't plunk a six year old down with a copy of Ulysses. You give him "See Spot Run". And even before he is reading, he is learning to recognize letters and the sounds they make. Then he learns to put several letters together to sound out a simple word. Eventually he can put several words together to read "See Spot run." Listening to a six year old read is an exercise in patience. It's a long time before listening to a youngster read aloud becomes a real pleasure for anyone other than his parent.
We understand the step by step building of skills for something like reading. Teaching our horses about chains follows a similar route from the very simple to the amazingly complex.
Examples of Chains
Lindsey, you asked for examples of chains. I gave some examples of simple chains in a previous post. Here are a couple that are much more complex:
In Jim Logan's wonderful llama videos "Click and Reward" he shows one of his llamas going through an obstacle course. The llama opens a gate, turns around and shuts it, then walks over a teeter totter, backs through a chute, walks under an archway with streamers hanging down, and rings a bell. Throughout the sequence the trainer is out of sight. I may not be remembering all the obstacles. It's been a while since I watched the video, but it's a lovely example of the type of behavior chain where a single cue is given to initiate the sequence, then the animal moves through the chain without further guidance from the trainer. Each obstacle in the sequence tells the llama what to do next. He's learned that completing the obstacle takes him a step closer to the primary reinforcer at the end of the chain.
Want another example? How about the dolphins that are trained to find underwater mines. The dolphins leave their trainers, swim down to depths of as much as sixty or seventy feet, find the mine, return to the surface and alert the trainer. They are then given a pack of explosives to carry down to the mine. They attach the pack at just the right height to the mine's antenna. Too high and the mine might not detonate from the explosives. Too low and the dolphin could be at risk. The dolphin attaches the pack, then returns to the surface, jumps up into the boat, which then heads off to a safe distance from the mine. That's very neat training, especially since the dolphins are working in the open ocean often near schools of wild dolphins, and there are live fish available to them, not the dead fish the trainers use for rewards.
These are both examples of chains where the next task in the chain signals the completion of the previous task. But what about dressage tests, obedience trials, agility courses? Here the animal needs additional information. The chain is not performed without input from the trainer. The dog needs to know which obstacle in the agility course to go to next. The dressage horse needs to know how many steps of piaffe are required before it moves on to passage. These are chains where cues from the trainer are needed. The cues tell the animal which of several options to do next, and used well they serve to reinforce the previous behavior so a high level of performance is maintained. Want an example?
Watch Attila Szkukalek and his dog Fly perform "The Gladiator". (Do a search on Youtube to find the link.) See if you can spot all the cues he's using in this beautifully choreographed performance.
In comparison to Attila, most of us are probably still at the "See Spot run." stage, but that's all right. The child begins by sounding out individual letters then pieces them together into words and simple sentences. At some point he has the skill not only to read Ulysses, but to write his own great books. In the same way we begin by teaching our horses individual behaviors. Then we string them together to create simple "See Spot run" chains. We can aspire to the level of performances like that of Attila and Fly. I would love to see something comparable with our horses. Perhaps working together through this list, we can get there.
Alexandra Kurland
theclickercenter.com
Behavior Chain Questions: Definitions, Cues as reinforcers, "Punishing Parents", and More
By Alexandra Kurland
Copyright 2007
This post was prompted by a series of posts in which people got tangled up in the definition of behavior chains.
We've been having an interesting discussion lately. Thank you Katie for posting your "Should you click offered behavior?" article. Lots of good food for thought there. And Tanya, great post today on the rope handling. It's such a good reminder to look at all rope handling techniques - including the "t'ai chi" rope handling - from the horse's point of view. Depending upon how you ask, the horse will either glide back with ease or experience deep frustration and physical discomfort. That's why the underlying belief system is so important. Intent changes so much of the experience. And we want to find techniques that match our intent and the relationships we want to have with our horses. Great post!
We've had other great posts these last couple of weeks, but somehow the discussion on behavior chains got derailed, or at least diverted into a discussion of terminology. I think we need to start over. It seems as though there was some confusion over what constitutes a behavior chain.
I'm going to take my cue from Karen Pryor. At the Clicker Expos whenever there is a question on terminology during the panel discussions, she always turns the microphone over to Jesús Rosales-Ruiz. He's our resident Behavior Analyst on the Expo faculty, so it makes sense to go to him for clarification of the science. So that's what I did in the midst of this discussion.
I sent him a summary of the discussion as it was evolving. We seemed to have two definitions of behavior chains. The article by Karen Pryor, "Behavior Chains and Backchaining" (http://www.clickertraining.com/node/111) was cited in several posts. In that article Karen defined behavior chains as: an event in which units of behavior occur in sequences and are linked together by learned cues. . . Of course the biggest and most important chains in dog training are the performance chains: long sequences of many behaviors, linked, reinforced, and thus maintained by cues, in which the individual units may come in virtually random sequences. Running an agility course is an example."
But some on the list took exception to this definition of a behavior chain. They referred to it instead as a sequence, and they defined a behavior chain as: A string of multiple behaviors elicited by a single cue.
Fetch was cited as an example of a behavior chain.
So I asked Jesús what the definition of a behavior chain was. Were both examples chains? Or was it more correct to refer to the first type as a sequence, and only the second was a behavior chain?
Here is Jesús' answer:
"Both are behavior chains. I was not aware of the distinction between chains and sequences. The field of behavior analysis does not make that distinction. The only argument that I know about chaining is the identification of the cue for the next behavior. In some behavior sequences this is not so clear. In this case people tend to think that the organism's own behavior is the cue. So the argument is about what is the cue for the next behavior but all the sequences are considered chains."
I think we saw this confusion over the identification of cues in the discussion over what is a chain. Where people were getting tripped up was exactly what Jesús was indicating. It can sometimes be very difficult to identify the cues within a chain, especially when they are environmental cues, or cues from the handler which have become faded down or merged with other cues.
Simple Chains
I know we could have a very interesting discussion based on Jesús' response. But that's not the direction I want to head off in. I want to return to the original purpose of this exercise which was to incorporate youtube video clips into our discussions. I'd like to see what people are doing with the many exercises we've discussed on this list. And I want to see if people understand a very important part of clicker training and that's how to use cues to link one behavior with another. Here's a simple example:
Most of you - in fact probably all of you - can get your horse to back up. If you taught this with a clicker, you got a small shift of weight back, click and treat, and then gradually built on that until you could ask your horse to back up several steps, click and treat. Cues emerged out of the shaping process, or you attached a cue for backing once the behavior was occurring consistently.
Most, if not all of you, can also get your horse to walk forward following a target. Again you started by having your horse just orient to the target, click and treat, and then you gradually began to move the target so your horse had to walk forward several steps, click and treat. So you had a sequence of behavior that looked like this:
You presented your horse with a cue to back up.
He backed up multiple steps. Click and treat.
You then presented your target, and you may have added other cues as well to indicate to your horse that he was to follow the target forward.
He stepped forward multiple steps. Click and Treat.
You have two solid behaviors, but the "dance" looks a little choppy because you are stopping to treat your horse after each unit. So the question is can you combine the two behaviors into a longer, uninterrupted flow?
Can you ask your horse to back up and, as he satisfies your criteria for backing, present the target and ask him to walk forward?
Does your horse stall out, look confused, frustrated, or cross because he did not receive his expected click and treat? Or does he follow the target forward? Why does he follow the target forward?
If targeting is a strong behavior with a solid history of reinforcement behind it, before your horse has even registered that there wasn't a click after he backed, he'll already be stepping forward to his target. The presence of the cue triggers the behavior before he has time to think - "Wait a minute! What's going on here? You didn't click me for backing!"
And before he can get to this thought, you're clicking and reinforcing him for following the target. We want cues to serve as internal "yes answers" for the preceding behavior. "You backed up. That was correct, so now here's your next cue. Here's your target." The horse sees the target and instead of feeling grumpy, he moves eagerly forward. Why? Because targeting has a strong history of reinforcement behind it.
Building Chains
As I build my horse's understanding of chains, I need to think about how I am going to use them. If I only want the behaviors to run in one direction, I will always ask for them in a set order. "A" always leads to "B" which always leads to "C". Fetching is a chain which most of us run in only direction.
But I might also want "A" to lead to "B" to then back to "A" again. Or I might want to start with "B" and go to "A" and from there to "C".
Backing and following a target would fall into this type of category. I started this example out by having the horse back up and come forward. But I also might want him to come forward and then back up. I do not want to set the order in stone, so in the early stages of building my horse's global understanding of how chains work, I need to show him lots of different combinations. Sometimes I'll ask him to come forward, and then back up. Sometimes I'll ask him to back up and then come forward. I'll be aware of my patterns. I need to make sure that my mini chain doesn't become back up three steps each and every time, then come forward. It's easy to fall into patterns, especially when you have just a little material to work with. You can easily create a glass ceiling where your horse gets stuck if you ask for one more step than the pattern he's practiced.
What Keeps a Chain Going?
Let's look for a moment at motivators. Why does the horse keep working? Chains do not have to be positively built to work. You can have an animal, or a person for that matter, performing long sequences of behavior in order to avoid unpleasant consequences. "I do this and then this, and then this, so this other awful thing does not happen."
A jump course is a perfect example of this. Many horses truly love to jump and have been schooled kindly and fairly over jumps. But there are also horses that have learned to jump because they have learned they have no option but to jump. If you stop, or swerve to the side, you will be severely punished. Jump, and you may get snatched in the mouth, but the human will not beat you with a stick.
At the end of this kind of chain is safety. That is a huge motivator that draws the horse forward towards the end of a jump course. After the final jump, he gets to rest. He gets to stand with the other horses. His human leaves him alone. And he may even get to go back to the barn. The danger from falling is removed as is the fear of being chased forward from whip and spur. The closer the horse gets to the end of the jump course, the stronger and more eagerly he goes forward. "My horse loves to jump!" Hmm, maybe. Or maybe he is just eager to get to the safety zone at the end of the chain.
“Punishing Parents”
I do not want chains that are built by a "punishing parent". Clean your room or else. You may indeed grow up making your bed every day and putting all your clothes neatly away in your dresser drawer, but there will be an underlying stress to these behaviors, that sense that you are always having to watch your step. If you accidentally drop a sock and forget to pick it up, the "punishing parent" will descend. No matter how many times you have put away all your socks, there is always that looming possibility that you will be caught out.
The "punishing parent" can praise you for being tidy, but twined inseparately around that praise is the displeasure if you make a mistake. Poisoned cues. You can be so habituated to the stress that you are not consciously aware of it. Like the background noise of a city street, after a while you tune it out, but your adrenal glands don't. You may not be consciously aware of the stress until it surfaces as disease, but the stress has been there all along.
Perhaps before things go too far down that ominous road, you go away on vacation, and - separated from all the classically conditioned responses that are built into your normal environment - you become aware of the contrast in how you feel. Awareness creates the possibility for change.
Positively-Focused Environments
I think that is one of the great values of the clicker training clinics. We spend several days together completely immersed in a positively focused environment. People can safely experiment, make mistakes, discover successes. They aren't finding fault or growling at their horses. What they are doing is reinforcing behavior they want, and being non-reactive to unwanted behavior. People work hard at these clinics. We often have very long days, but people leave feeling refreshed, invigorated, as though they have been at a retreat, which in a very real sense they have.
And then, they are back home, back in familiar environments, and the contrast is often striking. Walk back into your work place, your barn, your family, your neighborhood, your larger community that is governed by a "punishing parent", and you now won't be able to tune it out. You'll know the difference, and you'll feel the stress that the underlying threat of punishment creates.
So the questions is do we have to have an underlying threat to maintain behavior? "I go to work to earn a paycheck so the bank doesn't foreclose on my house and take it away from me." Is that what keeps us showing up for work?
For our horses what else can motivate them and keep them working through to the end of a chain? From a clicker training perspective the answer is cues that have been built with a positive reinforcement history. The cues both trigger the next behavior and act as a conditioned reinforcer for the preceding behavior.
Each link in the chain has a strong reinforcement history behind it. Each link in the chain has a cue that triggers it. A chain is only as strong as it's weakest link. If you have a behavior that is fairly new, that is not yet firmly attached to a cue, you want to use it very strategically in a chain. I would, for example, not ask for it immediately following another fairly new behavior. Nor would I use it towards the end of a long chain. If the horse fails to respond to the cue, and the horse stalls out in confusion or frustration, not only that behavior, but all the behaviors that preceded are affected. You have a domino effect where the negative emotion ripples back through the chain, weakening all the connections between the behaviors.
Ripples Back
In classical conditioning the emotional effect travels backwards. That's something else you want to think about as you are considering whether a chain has been positively or punitively built. Notice I am not saying whether a chain has been positively or negatively reinforced. I want to step aside for the moment from the scientific terminology and look instead at the emotional associations the horse has with the individual elements of the chain and the chain as a whole.
If I slide down the lead, and my horse backs up, I can release the lead. That's negative reinforcement. I applied pressure to my horse's halter, a slightly uncomfortable sensation for the horse. I also shifted his balance slightly so there is pressure on his joints. When he took a step back to rebalance, the pressure on his face and his joints was released. In addition, I clicked and treated. As I repeat this, my horse may come to view my hand sliding down the lead as a good thing. It is information he uses to get to his treat faster. If I have been careful in how I use this pressure so it never crosses a threshold level where the horse feels afraid or is in pain, the associations I build can be very pleasant ones.
But suppose my horse doesn't back up promptly. What do I do? Do I wait patiently at a point of contact? Do I offer him other clues, perhaps touching him as well on his shoulder, or shifting my balance more obviously? Do I simplify the request, or change the environment so it is easier for him to learn? The important question is do I keep all these clues and influencing factors under his pain/fear threshold? Or do I escalate to get a faster response? Do I cross the pain/fear threshold? That will alter the associations my horse makes with my initial request.
Touch the lead, back up, get a treat - good deal. The next time the handler touches the lead, the horse will be looking forward to pleasant consequences. The emotional effect washes backwards.
Touch the lead, get whacked hard across the front legs, back up, pressure goes away. This is a very different scenario. The next time the handler touches the lead, the horse may indeed back up promptly, but there will be a little zap of stress associated with the request. "If I don't guess right - will I be hit?"
In the first instance I connect the the back up behavior to another behavior and let the positive emotions ripple backwards. I can ask my horse to come forward, and then slide down the lead to ask the horse to back up. The slide down of the lead has positive associations. In the past it lead to a click and a treat. So the cue to back up can serve to reinforce the preceding behavior, which in this case is walking forward.
Chains Are Not Simple
So that's the basic premise behind behavior chains. However, chains are anything but simple. The motivators, what keeps an animal working through to the end of a chain, can be a complex mixture of negative reinforcers, positively built cues, poisoned cues, and punishers. If internal cues are not maintained well, the individual behaviors in the chain may drop out, merge with other behaviors, become foreshortened. This can be a desired result, or an unintended consequence. Chains can break down entirely. They can collapse like a house of cards, so how do we build strong, unbreakable chains? How do we teach our horses not just individual chains, but a global concept of how chains work? How do we eliminate unwanted behavior from a chain? Where is the beginning and end of a chain? When we pick a particular cue as the starting point and the click and the treat as the end point, are we being arbitrary in that. Is that really the beginning and the end from the horse's point of view? Is there really such a thing? Can we get very off the track and say that all of life is one huge chain. Aargh!!
When something is complex, it often works best to begin by looking at simple models of the thing we want to study. Sometimes we find that the simple model clarifies things and can explain a whole class of much more complex responses. Sometimes, as in the case of much of modern physics, we discover that our original model is too simple and doesn't really work to explain what we are discovering. We may have to redefine our model, or even set it aside entirely as we gather more data.
Chains very quickly become complex, so rather than trying to wrap our minds around complex examples, let's begin by looking at very simple chains. I'd like to go back to the beginning of this discussion on behavior chains and propose that for the moment we table discussions of definitions and instead gather some data. In other words, let's go to people for opinions and horses for answer by testing the following hypothesis:
If we define a behavior chain as: "an event in which units of behavior occur in sequences and are linked together by learned cues. . .", do cues indeed serve to reinforce the preceding behavior? Can you build sequences of behavior using a positively built cue and the behavior it triggers to reinforce the preceding behavior? Will your horse continue working with the same level of enthusiasm he showed for the first behavior? Or does he stall out, freeze up, lose interest, become confused or frustrated when a cue is presented for another behavior rather than a click and a treat?
Let's find out.
Barnwork
I'd like to see some examples of simple chains, two or three behaviors linked together by cues. I am treating this use of cues as a hypothesis. We are told that a cue can act as a conditioned reinforcer. What does that mean in practical terms and is this indeed the case? Is this how chains work? As we proceed with this exploration we may find ourselves questioning some very basic assumptions and redefining some of the terms we've been using. That's fine. In fact that would be a grand use of this list, but before we go there with the discussions, I'd like to see us collect some data.
So its off to the barn with everyone. Can you build a simple chain along the lines of the back up - go forward in the example given in this post? Feel free to use different behaviors, but keep your chain simple at first.
I need to add a P.S. After giving this particular ball a bit of a push forward, I should say that I am entering a pretty intense travel schedule, so my contributions to the list will be sporadic at best. I am sure the discussion that follows will be an interesting one to read on my airplane trips.
Alexandra Kurland
theclickercenter.com
