EQUINE CLICKER TRAINING.....
using precision and positive reinforcement to teach horses and people
Clicker Expo 2010: Lexington, Kentucky
I was fortunate to be able to attend Clicker Expo 2010 in Lexington Kentucky in March of 2010. This was my fifth Clicker Expo and I was happy to see familiar faces and meet some new horse people too. The horse people are definitely in the minority at this event, but we stuck together and I think we all learned a lot. If you have never heard of Clicker Expo, you can read more about it on Karen Pryor's website www.clickertraining.com or read some of my previous articles on what I learned at Clicker Expo. My previous articles are listed in the training section. I think the name "Expo" is misleading because this is not a series of demonstrations or about vendors selling products. It is a three day conference with scheduled lectures and learning labs. In the learning labs, you can bring a dog and get hands-on help from the faculty and fellow clicker trainers or you can be an observer and get to watch some training.
The Expo has a core group of faculty and then there are a few new speakers each year. They run five sessions at a time and it is always hard to choose my schedule as I would often like to be in two places at once. Sometimes I have to choose between a topic of interest or a speaker who I particularly like. There is some repetition from year to year as some faculty members teach the same topics each time, but they often add new material and some review is helpful too. One thing I have learned over the years is that since learning to be a better trainer is an ongoing process, the same information can be interpreted or implemented in different ways depending upon a lot of different factors. Revisiting familiar topics is always a good thing.
With this in mind, I chose a mix of familiar faculty members (Ken Ramirez, Steve White, Jesus Rosales-Ruiz, Kay Laurence, Karen Pryor) and new presenters (Eva Bertilsson and Emilie Johnson Vegh). If you are noting that Alexandra Kurland is not on this list, it is not because I didn't want to see her presentations. We discussed this before the Expo and I opted to skip her presentations since I get most of the same information from attending her clinics. If you are not familiar with Alex's work and do get a chance to got an Expo, I strongly recommend that you attend her sessions.
Since the Expo is intended to be educational and Karen Pryor's goal is to educate and bring along future generations of clicker trainers, the Expo often focuses on a certain theme. One year there were a lot of sessions related to shaping. Another year there were a lot of sessions that covered various aspects of adding cues and stimulus control. I don't know if there was a planned theme for this year, but I felt there was a real shift this year away from the mechanics of clicker training (timing, food delivery, how to shape behavior). There were certainly presentations focusing on these details, but I thought that a lot of the presentations were moving beyond that and looking for training strategies that used clicker training or elements of clicker training in slightly different ways. In other words, trainers were looking at how to optimize the power of the clicker to train faster, cleaner and stronger behaviors by using new strategies that were easier for novice trainers and less likely to cause difficulties since novice trainers are going to make mistakes in what they click. I felt this Expo showed that the power of clicker training comes from knowing many different strategies and what to use when.
One of the advantages of attending multiple Clicker Expos is that since I have seen some of the faculty presentations before, I can see how the same trainers are changing their own training over time. Some of the changes are in how they present information or how they teach certain behaviors or skills. But I am also seeing a change in focus. Now that most trainers are proficient in getting behavior, they are looking at other elements of the training session such as the quality of the behavior, the ease of training for new handlers or difficult dogs, long term training strategies, how to develop the physical skills needed for a task, and evaluating the emotional state or energy level of the animals. I thought it was interesting that a lot of the new discussions or training methods were about addressing "sticky" points that we often encounter in horse training, such as how to build duration, how to maintain behavior long term either on a lower reinforcement rate or by using different kinds of reinforcers and how to work around distractions, or when the animal is not mentally ready to learn.
I tend to think of Clicker Expo as being about educating the attendees, but it is worth noting that everyone there is learning. The faculty learn from each other and they are constantly looking for new ways to train their animals or teach their students. I found that I could recognize pieces that they had gotten from each other and it was fascinating to see how each trainer's work was evolving over time. I thought it was fun to see dogs doing something very similar to WWYLM in Kay Laurence's dog learning labs and Karen Pryor now teaches loose leash walking using cones and a method similar to pre-WWYLM. When I attended the agility lecture with Eva and Emilie, they talked about feeding in position, "where the perfect horse would be." After so many years of trying to sort through dog clicker training and modify it for horses, it was fun to see some information flowing the other way.
When I write about Clicker Expo, I often just present the material that was covered in each session, perhaps with some notes for how to apply it to horses. But this time, I have decided that rather than write a report on each lecture that I attended, I wanted to organize this article by topic so that you can get a feel for both the variations between trainers and also what was in common. The topics are not performance related, but more general clicker training topics. I think that for horse people, this is one of the most useful ways to look at the information we get from Clicker Expo. Sometimes my brain hurts from translating dog behaviors to horse behaviors and then dog sports to horse handling so I think looking at general categories is going to be easier and then you can think about how to apply it to your own situation.
The first step in clicker training is getting something to reinforce. In the past, I felt that most dog trainers preferred to use shaping to get behavior. In shaping, you start with a small piece of the behavior and slowly change it over time by raising criteria until you end up with your final behavior. Shaping is very powerful as you can start with just the tiniest bit of behavior and transform it into something else, but it does have a few drawbacks. One is that it takes some skill and practice to be able to shape behavior well. A trainer who is good at shaping has to have good observational skills, good timing and a sense of when to ask for more. It is easy when shaping to either get stuck at a certain stage or end up shaping a behavior that is not quite the same as what the animal would naturally do. In this expo I was seeing more use of creative training strategies and management instead of free shaping everything.
The difficulty of shaping a natural behavior has been talked about in past Expos. At one Expo, Kay Laurence, a dog trainer from the UK, talked about how she sees dogs shaped to spin and the trainer ends up with a spin, but it does not have the flow of a spontaneous spin because the animal does it in pieces. The dog turns its head but doesn't immediately follow with its feet or there is some other variation where the spin doesn't happen as a smooth and natural movement. The more complicated the behavior, the more difficult it is to shape it to become a harmonious movement. If you have ever shaped a horse in do a behavior with multiple components, you know how important it is balance out the different pieces as concentrating on one for too long can create some very odd behavior. At that time, Kay was promoting using luring as a way to train some behaviors. She found that luring created a more natural behavior.
But luring is somewhat controversial with clicker trainers. At one of the first expos I attended (perhaps the same one where Kay was talking about spins), Kathy Sdao gave a lecture on luring and whether or not luring promoted active learning by the dog. Kathy Sdao started out as a marine mammal trainer but now trains dogs. She had found that dogs that were lured a lot did not offer behavior as readily as dogs that were shaped to do behaviors. The presence of the food right in the dog's face actually prevents it from thinking. A common analogy she uses is to compare how well you learn to go someplace if you drive yourself and follow written directions as compared to if you follow someone. If you drive yourself, you are much more likely to be able to repeat the correct sequence of turns the next time. if you follow someone, your focus is on the car in front of you, and you are actually less likely to remember the actual turns.
This difference is promoted as one of the advantages of clicker training. The argument is that by not using food directly, the animal is more likely to think about what it is doing and is therefore more likely to be able to repeat the desired behavior. Since this is one of the benefits of clicker training, I was a bit surprised to see more direct use of food presented as part of training strategies in several sessions at this Expo. I'm not sure if this is really new or if I just noticed it now, but I do think that now that trainers are comfortable using food, they are looking for ways to get more out of the food, either via novel ways of delivery or using food to get behaviors more quickly.
In some cases, the food delivery is actually used to generate the behavior the trainer wants. In Kay Laurence's lecture on Learning Games, she presented three food games that use the food itself to get the behavior started and then she adds the clicker in later, when she can click and get the exact behavior she wants. Her point is that the clicker is powerful and rather than click all those intermediate steps, why not just use the food to get things started and click something much closer to the final behavior? One way to think of it is that instead of shaping from the beginning, she is using food to get the behavior and then capturing it with the clicker.
She showed us three games that use food. One is called "flicking the sausage" and it teaches the dog emotional control by letting the dog figure out the value of waiting for the food to move instead of trying to take it. The trainer sets a piece of sausage on the floor in front of the dog and when the dog backs off a bit, she flicks the sausage. It's a bit like "doggie zen" but the food moves. Flicking the sausage creates an unpredictable pattern of movement so the dog learns the best way to get it is to sit back and wait. If the dog is too close or spends too much time trying to get the sausage, the flicking is delayed and when the handler does flick the sausage, the dog has to turn to chase it.
Another game uses the idea of a "birds nest" where she gets nice heeling patterns (for freestyle) by having the handler unpredictably dribble food from her hand which is held up high over the dog's head. The dogs learn to focus on the hand and track it. They get occasional reinforcement for nice moments, but it is taught as a flowing exercise, not as a way to get specific leg movements. The other game she showed was how to use food to create a great recall by having the dog chase a piece of food that has been thrown away from the handler. Picture yourself throwing a bowling ball in one direction so that the dog has to run past you to chase it. The dog gets the food and you throw another piece in the opposite direction so that the dog has to run past you again. The dog is practicing running at speed toward the handler and once the dog is doing this consistently, she adds the cue and starts clicking.
I am not actually sure how to apply this to horses as we don't use food in the same way and I can't see Rosie chasing after carrots that I throw, but the piece that stood out for me was to look for ways to generate more of the behavior we want before we start clicking. I have certainly played games with Rosie chasing a ball and I may play with some of these to see what kind of behaviors I can generate. The way some people use flags to generate energy by having the horse chase it (NOT run from it) might fall into the same category. I think the other piece to keep in mind here is that Kay's games have a few different goals. They are a way to get more behavior and train people to see behavior before they start clicking. They also allow the animal to work out more of the beginning steps on its own so that the animal is learning from the structure of the game itself and not relying on constant input from the handler.
With horses, we have not been so dependent upon free shaping everything. Yes, we do use free shaping, but many equine clicker trainers still use negative reinforcement, body language and rope handling to generate and build behaviors. If you look at Alex's work, she addresses some of the same issues in that she uses pressure and release on the lead to generate a lot of behavior and to give the horse feedback. I think in the beginning when a new handler is learning the rope handling skills and clicker training all at the same time, clicking and treating for the early stages is useful and keeps the sessions more positive and more productive. But once the handler is more skilled in her mechanics and has more experience in knowing how to keep the training progressing, she can use the release as the main reinforcement and get more of the behavior before she starts clicking and treating.
The other session that had some creative use of food to get behavior was the one on Good Agility Practices by Eva Bertilsson and Emilie Johnson Vegh. They are agility instructors from Sweden and just wrote a new book on Agility. Their focus is on creating dogs that are focused and keen to work. The session I attended was on how to keep your training sessions productive and focused so that the dog never gets a chance to practice being distracted or sloppy. They emphasized making sure the dog is always clear about what it is doing. It is either in a training session which is short and carefully structured, or it is on a mat or in a crate. If you need to move the dog from one place to another, you have a clear procedure for doing this. You can take the dog by its collar or use a tug toy or use a "treat magnet" which is a treat held right on the dog's nose to lure it from one place to another. The idea is that the dog is so fixated on the treat that it is not distracted by other dogs and gets used to moving quickly from one place to another. No training time is lost in moving the dog around and the dog never loses focus. It is either focused on what is being clicked or treated, or it is glued to the treat in front of it.
I talked about this to my friend who teaches agility and she said a big issue with agility dogs is that they get very wound up and in the times between working obstacles or doing a training session, they become distracted or lose focus on the handler. Then the handler wants them to focus on them for the training session and the dog has trouble switching from break time to training time. Eva and Emilie suggest making it very clear when you are working and when you are not. Having simple but specific ways of moving the dogs around makes it easier to keep the dogs from getting distracted and keeps the handlers focused too. If you are moving your dog with a treat magnet, you are not going to stop and talk to your friend. This practice helps manage the dogs and includes a bit of people management too.
Does this apply to horses? I am still thinking on this one. I'm not sure I need to worry about maintaining focus when I am moving my horses around at home. We could certainly use targeting with the nose right on the target if wanted that much control over an animal moving from one place to another. Do we need to? It might be useful in some places, but I would say that most of us don't require (or necessarily) want the same energy level and focus that is desirable in an agility dog. I do think that it is worth looking at if we are clear about setting up our training sessions so that there are times when the horse is very focused on us and that there are other clearly defined break times. I find with horses that we tend to work in longer and less discrete sessions so the horse is on a similar rate of reinforcement over a longer period of time. Last summer I played around with changing the rhythm of my sessions so I had short periods of intense work and then a little break where we did something easier or took a break. I do think this was helpful as it takes a lot of energy to really focus and I got better results from having these shorter and more intense sessions within my longer session than from spacing it out more evenly.
MARKING THE BEHAVIOR
As I mentioned earlier, Kay Laurence is now looking at delaying adding the click until the behavior is more solid. She doesn't give people a clicker until they can clearly identify and see the behavior they want to mark. With her dogs, she is also doing some proprioception work and games where she just feeds without marking the behavior because she doesn't want the dogs actively offering behaviors, she just wants them learning to control their bodies. She showed some videos of training she is doing on exercise balls shaped like eggs. The eggs are flatter on the top and bottom, making them more stable than a regular exercise ball. She will use food to lure the dog up and then uses food to encourage it to move around on the top of the ball so that it gets comfortable standing on an unstable surface. She does have the egg stabilized by wedging it between a wall and the handler's knees so the dog is not dealing with more movement than it can handle. This work on the ball is one of her proprioception games which are designed to develop more body awareness and control and to prepare the dogs for walking on uneven surfaces.
I think Kay is really looking carefully at what happens when we click. She has chosen to delay adding the clicker for some games to make sure that the click is marking the exact behavior we want. She is also looking at what happens when we click from other points of view. When we click, the animal goes into food or reinforcement delivery mode and there is a mental change from focusing on the job at hand to taking a break. In her lecture on teaching dogs self control and energy management, she pointed out that when we click an excited dog for doing a behavior that requires some self control, the click itself releases the suppressed energy. And we often accidentally reinforce that by becoming excited ourselves for the animal showing self control. For this reason, we need to be careful about how we use the marker signal (what we are clicking) and what comes after it.
I have seen this with horses where we are teaching the horse that mugging doesn't work. The horse is trying really hard to control itself so that it can earn a click. I click and as soon as I click, the behavior deteriorates. At some level, this is fine, the click does mean the horse can stop working, but it then raises the issue of whether or not you are reinforcing the mugging behavior between the click and treat. I think this is a place where the click is releasing the animal's emotions and we have to ask ourselves if there is something we need to address in our training to either teach the animal to stay calm until the food is delivered or if we need to change something about the training itself (criteria, setup, ...) so that the animal doesn't have to work so hard at self control that the click triggers that kind of release.
Ken Ramirez (head trainer at the Shedd Aquarium) gave a lecture on smart reinforcement and he talked about how important it was to have immediate delivery of food if not using a marker. I think that there is some re-evaluation going on of when you need to use the clicker and when you don't, as well as how to introduce it. Jesus Rosales-Ruiz (professor at University of North Texas) is looking at different ways to do the first clicker training sessions. His work ties in with Alex's Loopy Training and he is looking at finding the smallest loop that teaches the animal what the click means. To do this, he is comparing three ways to introduce clicker training to a new animal. One is to just feed with no marker so the horse learns where to get the food and practices the food delivery. Another is to click and feed with no specific behavior in mind (what they used to call charging the clicker), so you just click and feed every few seconds. The third is to ask the horse to target, click, and feed so the click is immediately associated with doing a behavior. We tend to think of targeting as one behavior, but it really includes responding to the click and eating. Is it easier for the horse if we start with a loop that contains fewer behaviors? Do we need to start right out with the marker signal?
The idea of teaching horses about hand feeding before starting clicker training has been brought up before. There are some horses that don't know how to take food politely from a person's hand and it can make the first clicker sessions more difficult. I have always found that horses can learn good hand feeding skills at the same time as I do the early lessons on clicker training and that using a clicker from the beginning minimizes some of the mugging that can happen if the horse cannot predict when a treat is coming. I have also found that introducing the clicker early makes using food more acceptable to people who are opposed to hand feeding horses. But I don't want to discount what these trainers are doing. I think there is value to recognizing that clicking a lot of early versions of a behavior can create problems. I also think there is value in recognizing when we just want an animal to explore its own body movement without marking anything specific.
But for me, one of the benefits of clicker training is that my horses don't mug me for treats. I have small children and I want the horses to be very polite around food. For this reason, I always click before I hand feed. At this point I am not comfortable with the idea of feeding without clicking so I have to look at what Kay is doing and see if there are some valuable pieces I can pull out of it that still allow me to click and treat. One solution I have found is that I can use food delivery to set up the next repetition or to play with a behavior before I am ready to click it. Once my horse is solid on food delivery and there is no grabbiness about the treat, I can start to experiment with food delivery to see if I can use food delivery to set the horse up for the next repetition or even to allow the horse to practice a behavior that I am not yet ready to click.
For example, I can use food delivery to explore a horse's range of motion with its head and neck to see how it wants to move or where it seems to have difficulty. By changing the location of my hand, I can assess if a horse might have more difficulty stretching in one direction or flexing in a certain way. This ties in with the next section which is about using food, but I wanted to mention it here because I can use food delivery to get behaviors without specifically marking them. I can do the same thing with targeting but if I don't have a target handy, using food delivery can be a convenient way to collect some data in a casual way.
We all know that clicker training works by reinforcing desirable behaviors. With horses, we rely on food for most of our reinforcement, at least in the early stages and often for training new behaviors, even with advanced clicker horses. Food is convenient, easy to use and most horses are willing to work for it. With dogs, clicker trainers use food but they also have a lot of options for using play. Even zoo animals are often trained to accept other reinforcers as part of their training program. In a previous Clicker Expo article, I shared how Ken Ramirez trains conditioned secondary reinforcers which are something we can use with horses. A conditioned secondary reinforcer is something neutral that has been conditioned so the animal accepts it as reinforcement. My previous article explains how to condition them, so I am not going to repeat that here. In this Clicker Expo, I attended Ken Ramirez's talk on "Smart Reinforcement" which was a lecture on how to evaluate your use of reinforcers and also what it means to be a positive trainer.
Ken Ramirez is the head trainer at the Shedd Aquarium and he is in charge of teaching many new trainers. He finds that a good understanding of the correct use of reinforcers helps new trainers be more successful. According to Ken, smart reinforcement is effective, produces the desired result, maintains a comfortable animal, and maintains a behavior long term. At the Shedd Aquarium they use both food and conditioned secondary reinforcers in their animal training. With his new trainers, they are only allowed to reinforce behaviors they like and ignore behaviors they don't. If training is not going well, he looks at their use of reinforcement first as problems often occurs because of poor reinforcement. This includes poor choice of reinforcers, low reinforcement rate, and sloppy food delivery.
His simple key strategies for reinforcement are precision marking, timing, clean delivery, placement, stationing (when needed), fairness and balance of reinforcement. Precision marking is identifying the behavior with a marker signal. Timing refers to promptly starting food delivery at the right moment if a marker signal is not used. Clean delivery, placement and stationing (as needed) are all important components of stress free food delivery. Sloppy food delivery at any of these points can frustrate the animal and actually make the training session aversive. If the food delivery is sloppy, it can distract the animal so much that it loses focus and does not repeat what was just reinforced. This leads to a lower rate of reinforcement and stalling in the training.
Fairness and balance of reinforcement refer to looking at the overall training session or what happens over several sessions. Over or under reinforcement can both cause problems. Fairness is important when working animals in groups. He mentioned that if you are working in groups and one animal is asked to do something while the others wait, the others are working too (by waiting patiently) and should be rewarded too. Animals can learn to take turns but this needs to be taught and a good trainer always recognizes when an animal is working hard and deserves reinforcement.
Kay Laurence also talked about reinforcement but from a totally different angle. Kay is a dog trainer from the UK and she trains and teaches many different types of dogs for many different purposes (from competitive freestyle to herding and probably everything in between). She also breeds border collies and gordon setters. I think that the fact that she breeds dogs gives her a different insight into reinforcement because she has the opportunity to observe puppies interacting with each other and with their environment. By watching puppies play, she can see what is naturally reinforcing to them from birth. Why are some games more fun? What do some puppies enjoy doing more?
I attended several of her lectures and one was titled "Learning Games and Play." In it she talks about how puppies learn and how we can use that information to help us develop learning games that we can play with our dogs. This is a totally fascinating topic and made me realize how much of a disadvantage horse owners have. Most of us do not get our horses as foals or get to see them play as foals Even if we do observe our horses in groups, we don't necessarily see the kind of play that happens with young puppies. Horses seem to spend a lot more time eating than dogs and play doesn't seem to be such an important factor in their lives. Still, I think the idea of using play is an important one and even if the horse version of play is different than the dog version, I like the idea of finding out what a horse thinks is fun and using that to make training more interesting.
I think Kay is not the only one looking at play. In recent years, the word "play" has come up in horse training. Horse trainers have come up with "games" to play with your horses and instead of working our horses, we talk about playing with them. But is it really play? Is it the kind of play that horses want to play? can we play like horses? The book "Empowered Horses" talks about how important it is that both parties are willing and eager participants in games. She has a lot of good ideas for how to play with your horse where play is not just another word for training. With horses, we do have to consider safety, but even with dogs, Kay does not necessarily play the same kinds of games with her dogs that they play with each other, or if she does, she modifies them. She does look at dog play behavior to get ideas and to see how they learn. She gave us some examples of food games that are based on a dog's hunting and chase instinct as well as its ability to learn to wait and see what a prey animal is going to do.
FOOD AND FOOD DELIVERY
Food is, of course, a convenient means of reinforcement and I have already shared some of the information that was presented on new ways to use food, whether or not to click, and reinforcement in general. In addition to using food to get or reinforce a behavior, food delivery can be used in specific ways to increase the productivity of your training session. I think new trainers underestimate the importance of food delivery. Horse trainers tend to notice it because poor food delivery can lead to problems with horses such as mugging, biting, increased frustration and general grumpiness about the training session. Alexandra Kurland has spent a lot of time mapping out the elements of good food delivery and she emphasizes consistent and safe feeding techniques in her clinics. Her session on Loopy Training points out the importance of monitoring the behaviors that occur before and after food delivery because clean and efficient food delivery is so important.
Kay uses food in her learning games and uses creative delivery of food to generate behaviors and teach animals self management. In the session on training self control. she talked about how important food delivery is for setting the tone and pace of a training session. It is important to recognize and modify your food delivery depending upon how the animal is doing. With animals that are naturally slower, food delivery that is too quick can rattle them and interfere with their ability to process what is happening. She points out that animals that are tense will rush you through food delivery, which makes you tense and then everything goes downhill from there. Our natural tendency is to rush to deliver the food if an animal is emotional, but she points out that if the animal is rushing you, it is better to deliver the food slowly and carefully. By drawing out the feeding process, she can slow the pace of the session and help the animal learn to slow itself down. You can use a slower pace of food delivery to ease the dog into a new and slower training rhythm.
I don't think she means that an animal that wants to work at a slower pace has to stay there forever, she just means we need to recognize the animal's needs and go from there. For an animal that works at a slower pace, you can experiment with using food delivery to add energy to a situation by choosing a reinforcer that generates more energy or delivering it in such a way that the animal becomes more animated. The point is to recognize if you are delivering food in a way that meets the animals needs and then see if you can shift it in the direction you want and then re-evaluate if that is helping or hurting the situation.
There are a lot of elements to food delivery including location, speed and presentation. She talked about using anticipation. For some of her dogs, the anticipation of the treat is what motivates them, so she will make quite a process out of picking just the right treat and presenting it to generate enthusiasm. It is common for dog trainers to vary the reinforcers, with a mix of treats or treats and toys. Kay chooses the reinforcement based on what kind of energy she wants to generate in the animal. If she wants to build energy, she will use a toy that gets the animal a bit excited, or throw the food. If she wants to slow the dog down, she will slowly and carefully deliver food. You can get a lot of variation by having several reinforcers and different methods of food delivery.
A jackpot is another way to vary food delivery and reinforcement value. The subject of jackpots came up in Jesus Rosales-Ruiz's session on clicker research. His student found that jackpots created unpredictable behavior and that even if they did seem to have a positive influence on the next few repetitions, it was not maintained. This contradicts the anecdotal evidence from most trainers which is that jackpots seem to work. I think they are going to do some more research on this, but when I talked to him, he said that his feeling was that the improvement that trainers see after a jackpot is not from the increased amount of food, but because most trainers change their mode of food delivery, presentation, energy level and emotional tone when giving a jackpot. We get excited and make a big deal out of it and his feeling is that the change in our behavior is more important than the quantity of food.
I thought this was interesting because anyone who has done a certain amount of clicker training can see that clicker training is not just about the food. The animals get very involved in the game and in finding the right answer. As I mentioned earlier, Ken Ramirez pointed out that not only is sloppy food delivery inefficient, but it can actually be aversive to the animal. If the food delivery frustrates the animal or causes it to keep continually losing focus, the animal might choose not to play the game at all.
IMPORTANT COMPONENTS OF GOOD TRAINING AND TRAINING STRATEGIES
This is sort of a "catch-all" category. I wanted to share some elements of putting together a good training program that were presented in various sessions, but don't necessarily fit together into a cohesive unit. I haven't talked at all about Steve and Jen White. They are dog trainers who work in a variety of different areas including police work, search and rescue, pet dogs and horses. Steve's presentation was on vexing training issues and he discussed his DIP-IT protocol which is his method for working through a training challenge. DIP-IT stands for Define the problem (get the real baseline), Isolate the problem, Plan your remediation, Implement your plan, and Take another look. I covered DIP-IT in a previous article so you can read more about it in my previous clicker expo articles if you are interested. I think most of us use some version of DIP-IT when we train, but having it written down is helpful if you get stuck. It gives you some structure and can point out where you missed a step.
Steve also presented some very clear information on how he approaches training. For one thing, he doesn't believe that training to get rid of unwanted behavior is any different than training desired behavior. You use the same steps for both. He also doesn't believe in failure. He likes to say "failure is just information." Instead of just focusing on results, he judges a training session as successful if the dog ends up ahead of where it started OR if you learned some useful information from the session. He says his training relies on 22 basic principles. These are Karen Pryor's 10 laws of shaping, 8 ways to get rid of unwanted behavior and 4 rules for stimulus control. These are all in Don't Shoot the Dog if you don't remember them. It was a good reminder of the importance of the basic principles behind clicker training.
In Ken Ramirez's talk on smart reinforcement he talked he talked about what it means to be a positive reinforcement trainer. This comes up on a fairly regular basis on clicker lists and there seem to be varying definitions of clicker trainers vs. trainers who use clicker trainer vs. all positive trainers vs..... What exactly does it mean if you say you are a positive reinforcement trainer? He shared Susan Friedman's Hierarchy of Effective Procedures which is a diagram shaped like an inverted cone. There are 6 categories, marked by lines and these are (from the top down): medical, nutritional, physical; antecedent arrangements; positive reinforcement; differential reinforcement of alternative behaviors; extinction, negative reinforcement, negative punishment; positive punishment.
The goal of positive trainers is to only move down the cone only if all other possibilities have been exhausted. So if a problem was occurring, one would address the physical, nutritional and medical needs first. If those were ok and the problem persisted, the trainer would look at antecedent arrangements. Antecedent arrangements include such things as management. If your dog chews your shoes, putting your shoes away would be addressing the antecedent to shoe chewing. I think the rest are pretty self explanatory but the idea is that you don't move down the cone toward punishment unless you need to. As horse trainers, we are going to be sensitive to the location of negative reinforcement in this hierarchy, but I think it is helpful to recognize that most animal trainers use negative reinforcement for correction and not for shaping behaviors as we often do.
In Kay Laurence's talk on Learning games, she presented some games and then defined her rules for evaluating games. Games need to be carefully chosen so that they have some learning value and they are low risk for both the trainer and the animal. With her dogs there are different types of games including solitary games with the environment or toys and interactive games with their siblings, peers, adults and other people. Any game requires some skills so you need to decide if you and your dog have the necessary skills to play them safely. Or maybe you can learn the skills through playing the game, but then you have to decide how to structure the game to teach those skills. Some of the things that animals learn through play are physical skills, experimental learning, self management, social interaction and boundaries, competition, and how to release stress.
A game needs to be appropriate for a handler and dog so some games that are games dogs play with their siblings or peers may not be appropriate for a person and a dog. In addition, each game needs to have clear rules, and the handler should have a clear structure or plan for what resources to use, what learning is happening, how to use the game to generate desired behavior and what games are appropriate for individual dogs. Kay's knowledge of what game is appropriate for what dog was obvious in her learning lab. She uses whippits (a stuffed animal on the end of a lunge whip) and tug to teach dogs self control. I watched her use tennis balls to teach a border collie to focus on her job and ignore distractions. The tennis balls were sheep and the dog was rewarded for a good effort by getting to chase the sheep. If the dog didn't focus, a sheep escaped out the back and she lost her opportunity to chase it. She also uses proprioception games to develop self awareness, balance, coordination, build strength, build trust, develop cues and puzzle solving.
One of the advantages of Kay's learning games is that they are set up so the animal learns from the game without being directed by the trainer. The trainer sets up the game, but the dog is not micromanaged within the game. The point of the game is for the dog to learn to manage itself. This could mean learning to manage its own body in games that develop its physical skills or learning to think through problems or figure out what works and what doesn't in other kinds of games. Rather than clicking every little right answer, she allows the dog to keep going until the answer is very clear and the dog has learned it for itself. This works because the games are carefully set up so that dog gets reinforcement within the structure of the game when it gets the right answer.
In some of Kay's games that were about teaching emotional control, she showed how you can take some of the pressure off the dog by training more casually. If the dog's energy level was too high, she would pretend to do something else and wait for the dog to calm itself down instead of expecting the dog to calm itself down while she was staring at it and waiting. This made a lot of sense to me because last summer when I was working some new horses, I learned the value of taking a break. If the horse got a bit wound up about trying too hard or throwing too many behaviors at me, I might quietly wait it out, but for some horses, this was still putting pressure on them. Instead I would just go do something else for a moment, and let them sort themselves out. It might be enough to take a little walk or I could go adjust the jumps, poles or cones or do anything that showed them I was not asking anything of them for a few minutes.
I think one of the big pieces I got from Kay's presentation was that good trainers recognize how to set up training situations so that the behavior you get is the natural one for the dog to do in that circumstance. She is a keen observer of dog behavior and talked about how we make our jobs harder by trying to teach behaviors where they do not naturally occur. This is especially true of behaviors that require our dogs to stay still. She believes that dogs are very good at self control and energy management and she is not a fan of suppressing behavior. Dogs can learn to control their energy from the very beginning if you present them with exercises that encourage self control. The food game "flicking the sausage" teaches dogs self control and focus because the dog learns to wait and watch the food. Hunting or herding dogs will naturally freeze in certain situations when watching other animals and she takes advantage of that.
She talked a lot about energy management too. She likes to build confidence and calm from the start and is very aware of generating more energy or arousal than she can control. You always need to be monitoring the dog to make sure that you are not generating more energy or arousal than either of you can handle. This is very similar to how Alex uses head lowering or other foundation lessons to balance out exercises that generate energy or might make a horse more tense. As performance animals, we want our horses to be able to change their energy levels as needed.
Alexandra Kurland's sessions were on "Loopy Training" and "6 Strategies to get you through training blockages." The 6 strategies are Train by Priority, Keeping things in Balance, Always know where your mat is, 300 peck pigeon, Overcoming fear and the power of cues and the MicroShaping strategy. I didn't attend these sessions so I only have her power point presentation. Perhaps she will jump in and share some details. The 6 strategies cover a lot of training issues and if you have a training problem or are just looking for information on how to set up a training session so that it is a positive experience for both horse and trainer, one of these strategies will probably work for you.
I am going to give VERY brief descriptions here. Those of you who are familiar with Alex's work will recognize that most of this information has been presented on her DVDs. "Training by priority" refers to setting criteria so that that most important one is taught first. Once it is solid, you can then add more criteria to improve the quality of the behavior. "Keeping things in balance" means that for every behavior you teach, there is an opposite behavior that you must teach to balance it out. "Always know where your mat is" is about remembering that goals can interfere with good training. Yes, you should know your end goal (she used the example of putting the horse on the mat), but you should remember that how you get there is as important as getting there. The "300 peck pigeon" is an exercise in building duration and teaching emotional control. "Overcoming fear and the power of cues" looks at how powerful it can be to recognize what is cueing undesirable behavior and how you can change the horse's physical and emotional response by changing the cue or the meaning of the cue. "The MicroShaping" strategy looks at using an easy behavior as reinforcement for learning a harder task so that the rate of reinforcement remains high and the horse gets a chance to have a break from the harder task but keep earning reinforcement.
There were a lot of other tidbits that I picked up in various sessions. Ken Ramirez gave an interesting talk on teaching mimicry and it had a lot of details about how to set up the training sessions to teach mimicry. Eva Bertilsson and Emilie Johnson Vegh shared some training clips that showed dogs learning to make noise. They teach the dogs to knock over towers of pots and pans so that the dogs are comfortable with sudden noise in the agility ring. They also shared a flow chart of a great training session. Included in the planning part of any session is how to set up the session, what resources are needed, and what to do if something goes wrong. Karen Pryor talked about how to deal with the public on issues such as the use of punishment, especially now that television seems to promote trainers who rely on punishment.
Both Ken Ramirez and Steve White talked about the importance of collecting data and keeping notes on training sessions. Perhaps it is because we work in longer sessions or perhaps it is because most of us are training our own horses for fun, but I don't meet too many horse trainers that collect data as they train. I keep a training journal but I don't record precise numbers on anything. However, I am starting to think that there is probably value in being a bit more organized about data collection and I have been playing around with sample data sheets for collecting information on training some basic behaviors. I think that having some data sheets would be useful if you hit a training block, are training a brand new behavior, or were working with team and had to share information. I also think there is value in encouraging new clicker trainers to collect data so that if they have training issues, they have precise information to share with their teachers or anyone who is offering them assistance.
I certainly came home with lots of ideas for things to try and new ways to look at training. It was exciting for me to see how much clicker training is evolving and changing to address different training issues such as developing better teaching methods, looking at the whole animal and its needs, and how to help owners be more successful in training their own dogs. Most of the clicker trainers I know are not professionals, but just looking for better ways to do things. It was nice to see how there are always new ways to improve your training and anyone can access them if they understand the basics and then add a bit of creativity.
If you have questions on something that I have covered here, you can email me for more information (email@example.com). If you want more information on a specific topic or faculty member that I have mentioned here, most of them have web sites and/or written resources available. A list of the faculty members is available at www.clickertraining.com.
Katie Bartlett, 2010 - please do not copy or distribute without my permission