EQUINE CLICKER TRAINING.....
using precision and positive reinforcement to teach horses and people
ORCA's Art and Science of Animal Training Conference: 2012
On March 10, 2012 I attended the ORCA conference at the University of North Texas. ORCA (Organization for Reinforcement Contingencies for Animals) is a student organization at the University of North Texas and the conference is organized and hosted by Dr. Jesus Rosales-Ruiz and the students. It is a one day conference that brings in speakers from all over the world to talk on the subject of teaching and training. This is the 4th year the conference has been held and the list of speakers included Joe Layng, Bob Bailey, Phung Luu, Steve and Jen White, Alexandra Kurland, Kay Laurence, Ken Ramirez, Steve Aibel, and Mike Pool . There was also a panel discussion.
While I've been aware of the ORCA conference for a while, I had never attended it before since it is quite a distance to travel (Pennsylvania to Texas). This year I was invited to spend a few extra days there so I decided to go. It was well worth the trip. The conference was full of useful information, both for training my own animals and for thinking about how to improve my teaching. This is not a "clicker conference" although most of the speakers use positive reinforcement, so there was less emphasis on the details of clicker training and more emphasis on training and teaching strategies for anyone using positive reinforcement.
I think the easiest way to share the information is to pick out some highlights from each talk and then at the end of each section, I will try to share what I got out of it and how I think it will change how I train and teach. If you want more information on any of the speaker, the ORCA website has biographies of all the presenters. I have included some details here for people who might not be familiar to you.
Joe Layng: How We Talk About and Teach What We Do
Dr. Layng has most recently been working on developing educational programs for children (reading). He has an extensive background in science and education and his talk was about how we (as teachers) need to be more aware of areas that can help add clarity and improve progress in teaching.
He started his talk by showing that even though the scientific community has words that they use all the time in research and discussions, if you start to look carefully, you will find that even accepted terms are not used in exactly the same way by all scientists. He used a few examples including "contingency," "FI 30 (fixed interval schedule of 30 seconds)," "chain," and "sequence."
He started with the word "contingency" as an example of a word that is widely used but not clearly defined. Considering it is "the bedrock upon which their field is built," this is rather surprising. He talked a bit about how they use the term and what it seems to mean, so his point was not that definitions don't exist, but that if you talk to different people you don't get the same answers. This makes it hard to communicate clearly. The use of the term "FI 30" was another example of what seems like it should be a clearly defined term, but there is some variation in when to set the timer to start the FI 30 (at the beginning or end of food delivery) which can lead to different results.
He also used "chains" vs. "sequences" as an example, not of the same term with multiple meanings, but of two terms that end up used interchangeably when they are not the same. In a chain, the second response cannot occur unless the first response occurs. This is because it is the result of the first that sets up the conditions for the second (and so on, for longer chains.) Fetching is a good example of a chain because you have to go to the object before you pick it up, and you have to pick it up before you can return with it. Playing the piano and singing are not a chain because regardless of how you break it down into steps, one step is not dependent upon the completion of the previous step. The song might be odd if you sang or played parts out of order, but you could still do it.
This brought him to his next point, which I thought was really important for people who teach, especially if they use a lot of written directions. Perhaps it applies equally to verbal directions, but verbal directions often have some visual component. He had a chart that showed the directions for making Plaster of Paris on the left. There were steps, each one explaining what should be done. But there was no information on how to evaluate what you had done.
On the right he wrote out what you should see after completing each step. If you only had the information on the left, you would have to take your best guess at interpreting the directions for each step and go on to the next one, hoping you were correct. But if you had the information on both the left and right of the chart, you could do step 1, then check to see if you got the results you wanted before proceeding to step 2. This made it clearer than just being told what to do (the steps) because it allowed you to evaluate your progress. These kinds of details matter and they are the details that are often omitted when the person providing the instructions already knows the process. You have to write instructions from the point of view of a complete novice.
His next example expanded upon the idea that teaching needs to provide many different kinds of information. He talked about teaching multiple discrimination and concepts. When you teach concepts, it is just as important to teach "non-examples" as it is to teach "examples." If you want to teach someone to identify a chair, you can show them various pictures of chairs and identify characteristics that are common to all chairs. But unless they also learn to identify what is similar to, but not a chair, then they cannot master identifying chairs. They need to know both what a chair "must have" and what a chair "can have" so that they can identify what makes a chair a chair and what are acceptable variations. By using both non-examples and examples, you can help someone who is having a hard time with a concept because if they think non-chairs are chairs, you show them more non-examples, and if they think chairs are not chairs, then you show them more examples.
The last item he talked about was the use of immediate feedback. Feedback can be either confirmatory or instructional. Confirmatory feedback is information that tells you what you are doing right. To be effective it needs to immediately follow the behavior and can be used to increase, maintain, or capture a variation on the behavior. Instructional feedback is information that tells you what to do differently next time so it needs to immediately precede the next effort.
If someone tells you what to change at the end of a session and you don't get to practice again until the next session, that feedback is not going to have much impact. They would be more effective if they gave you confirmatory feedback at the end of the session and then started the next session with instructional feedback so it preceded the first effort. He described an experiment where the instructor wrote instructional feedback on a paper and gave it each student. One group read it after the session and the other group read it before the next session. The group that read it before the next session showed improvement compared to the other group. I don't remember all the details but I think there was a significant difference.
It was nice for me to have someone talk about what kind of information we need to present (and when to present it) when teaching. It has been a real challenge for me to learn how much information to share so that my students know enough to get started but are not overwhelmed. I like the idea of using non-examples and making sure people understand the result they want to see as they follow instructions. It was also a bit gratifying to see that terminology can be problematic in every field, and that it always pays to clearly define what you do so that someone else can follow your instructions.
Bob Bailey: Looking for advanced dog training? Try "simple dog training" for a change - it really works
Bob Bailey has been training animals using operant conditioning for a LONG time. If you don't know who he is, I suggest you look him up on the internet. His early work was with training dolphins for the Navy and then he ran Animal Behavior Enterprises (ABE) with Keller Breland and Marion Bailey for many years. He is known for his famous "chicken workshops" where he teaches the principles of operant conditioning using chickens.
The theme of his whole talk was that dog training (animal training) should be simple. Simple is NOT easy. At least not for the trainer. It should be simple and straightforward for the trainee. The trainer's job is to analyze the problem until they come up with a plan that is simple. He shared the definition of simple which includes the phrases "not complex, composed of one part" and the definition of easy which was "requiring no reflection."
He moved pretty quickly though some basics which I am just going to list here.
What is training? Making it worthwhile for the organism to change.
The animal ALWAYS has a choice.
The only behavior you can change is your own.
Ask yourself, what do I want? What do I have? Train the behavioral pathway from have to want.
Behavior is anything that an organism does.
It is HARD to make things simple.
Animals are learning machines It is what they are designed to do.
So how do you get to simple animal training? It takes time and effort to develop mechanical and observational skills. You can improve your own training by using video. Bob said multiple times over the weekend that video is the most valuable tool we can use to evaluate our own training. You can also get feedback from trusted sources.
When you train, it's important to really SEE what the animal is doing and not interpret it. He had some video clips from chicken camp where he was asking trainers on day 1 to tell him what the chicken was doing. Most of the responses were "the chicken is trying to," "the chicken wants to," or "the chicken is thinking about..." By the end of camp when he asked the same question, the answer was a description of what the chicken was actually doing.
The training environment is important too. If your training environment is too noisy or distracting, your animal will not be able to hear your signals. You want to start out in a quiet environment so you can start with quiet signals and the animal learns to listen for your signals. Then when you add distractions, the animal has already learned to listen for quiet signals. You want to train the animal from the very beginning to be aware of what you are doing so you never get into a situation where you feel you have to "shout." Along the same lines, when you start a training session you should have clearly defined criteria (write it down) and a clearly defined cue (if you're at that stage).
He also talked a little bit about the law of economics in animal behavior. Animals are always going to work to conserve energy. If you don't have a well thought out training plan and you end up reinforcing a lot of unwanted responses, then you are not going to build up reinforcement history quickly enough for the right responses. This wastes a lot of training time. This is why it's important to collect data so that you can define objectives for each trial and see if you are efficiently working toward the behavior you want.
And finally he asked the question, Do you have a training philosophy? What guides you when you don't know what to do next?
I only presented the bare bones of Bob's presentation here. He had a lot of good information and video clips to show how important it is to clearly define your training plan and explained how training can be broken down into simple steps. A few years back when I was clicker training Rosie for ridden work under saddle, I realized that I had made things too complicated, so I stopped what I was working on and essentially started over again. This time I sat down and really thought about how to make sure she was fluent in the basics before I asked for anything more complicated. The talk today was a good reminder that I need to keep that in mind as there are some new behaviors we are training that probably need to be broken down into smaller steps.
Phung Luu: Using modal action patterns to influence behaviors
Phung Luu works primarily with birds and he shared some of his work where he used Modal Action Patterns (MAPs) to work through some behavioral issues with birds. Modal Action Patterns used to be called Fixed Action Patterns. They have been renamed to reflect the fact that they are not as "fixed" as previously thought. He defined MAPs as genetically linked behaviors that are typically exhibited on a frequent basis and are triggered by releasing stimuli. His interest is in recognizing and taking advantage of MAPs in his training, as well as recognizing when a MAP might be interfering with what you want to train.
MAPs can distract animals from the behaviors we want to train and he referred to the article called "The Misbehavior of Organisms" by Keller and Marion Breland. I recommend reading this paper if you have not already done so. You can find it online at the following link: http://psych.hanover.edu/classes/Learning/papers/Breland%20and%20Breland%201961.pdf. What the Brelands found was that in certain situations an instinctive behavior would take over the training process and create some training challenges. This is a good reason to know about the natural behavior of your animal so you can predict how it will react in certain situations.
He showed some video clips of MAPs to illustrate the variety of MAPs and how they are genetically programmed. Baby birds have a feeding posture they assume when they are stroked by their beaks, baby gulls peck at the red dot on the adult gull's beak, stickleback fish respond to the color red with aggression, even when the red item is not very fishlike. In some dog breeds, there are behaviors that are part of the breed such as pointing in hunting dogs and herding behavior in collies. Even though they are genetically programmed, MAPs can change over time as the animal becomes more experienced. Weaver birds build elaborate nests with no instruction or modeling from their parents so the behavior is not learned, but they do learn to make better nests from year to year.
In addition to recognizing the MAPs that are specific to each animal species, he talked a bit about how some animals are more flexible learners than others. Animals that are very specifically adapted to one environment (specialists) tend to have more MAPs than animals that are generalists who live in a variety of environments and adapt quickly. Ospreys are very specific in how they hunt and eat so they are harder to train to do novel behaviors. Ravens are scavengers and very quick learners because they are constantly learning and changing to take advantage of new food sources. Their behavior is more variable which is beneficial if you are trying to shape behavior.
He shared an example of how he used a MAP in his work with condors. He was working with an aggressive condor and he decided to shape "wing pumping" which is a natural behavior for condors, but was a behavior that this condor was not doing. He wanted to train an alternative behavior that was not compatible with biting, but was part of its natural repertoire. He showed how he shaped the behavior and then asked the condor to wing pump in situations where it had previously been aggressive and this led to a decrease in the incidence of aggressive behavior.
He also talked a bit about how to use MAPs to get behavior you want. The condors like to get baths and will spread their wings if you spray them with a hose. They can use the hose to initiate the behavior and then put it on cue if it's something they want at other times. I think we have all taken advantage of MAPs at times in our training. Horse trainers certainly know that one way to teach a horse to lie down is to hose it off and lead it to a sandy spot where it can roll. But probably we could improve our training by being more aware of them. I am wondering if it's easier to maintain a behavior if it's something the animal is genetically programmed to do. Perhaps when we look for incompatible behaviors that we want to train to decrease an undesired behavior, we should look at the animal's natural repertoire first.
Steve and Jen White: Your Results May Vary: the How and Why of Choosing Tools and Techniques
Steve White is a police officer and dog trainer which means he has a lot of experience with both working and pet dogs, and people of all types too. He talked about how we choose training tools and techniques. In his talk he showed a slide with various training equipment (leashes, collars, choke collars, crates, tie-outs, e-collars, etc..) and pointed out that the most important tool, the person's hands, was missing. Equipment can be used in lots of different ways and while there are some pieces of equipment that many of us would not choose to use at all, it's important to realize that the person using the equipment is an important part of the equation.
With this in mind, he talked a little bit about what it takes to become a good trainer and then a good teacher. As a trainer becomes more experienced, it is easy to forget all the little details you learned along the way. Part of this happens because you learn to adjust in small ways without conscious thought and part of it is because you develop good habits which you forget are things you had to learn. As a trainer, this makes you a better trainer, but as a teacher it is important to remember all these details so that you can communicate effectively with your students. This reminded me of Joe Layng's talk from the morning where he showed how important it was to remember and teach the details that allow you evaluate your progress.
As a trainer, it's also important to keep advancing your own skills. He brought up the idea of 10,000 hours which is supposed to be the amount of hours you need to become an expert. In one of her books Karen Pryor talks about 10,000 hour eyes and the idea is that you have to observe 10,000 hours of behavior before you get good at interpreting what you see and can call yourself an "expert." Steve says it takes about 5 years to get to 10,000 hours )working every day) and he makes the distinction that it is 5 years, not 1 year 5 times. I am assuming that means that if you want to become an expert, you have to stretch yourself and try new things so that you continue to improve your skills.
Being a good trainer is not just about getting experience and using equipment wisely and well. Trainers also have to make decisions all the time about what methods to use. There are a wide variety of methods being used in the dog training community so it's important to learn to evaluate and choose what is appropriate for each situation. He used examples from police training to illustrate how police officers make decisions about what to do when they are in a situation that requires their involvement, and how they have to look at a lot of variables before deciding how to act. In police training there is a lot of emphasis on appropriate use of force and while he was not advocating the use of force here, he was saying that just as police officers learn to use the concept of objective reasonableness (you don't have to be perfect, you just have to be reasonable), so dog trainers can learn to choose methods that meet their criteria.
Dog trainers need to have a process for evaluating training methods and he described a 3 pronged test which he uses. It looks at effectiveness, intrusiveness, and social acceptability.
Effectiveness is about whether or not the training works, but it covers more than just getting behaviors. It is affected by client expectations, ease of use (can the client do it?), and durability (does it last so the client can continue the work without you?). With most clients that come to him for problem behaviors, he finds that what they want most is "relief." They just want the dog to stop doing the undesirable behavior.
Part of effectiveness is showing them that that he has a superior product so that they are willing to invest their time and energy into it. His goal as a trainer is to meet their needs and to show them the long term benefits of his training method. This means he has to take time to teach them the details that matter, so they can be successful. This part of his talk reminded me a lot of what Joe Layng said about making sure people understand the steps so they can do it successfully on their own. It's also important to get them to recognize the payoff that comes from making time in the beginning so they build a solid foundation.
Intrusiveness is how much you are managing and controlling the dog's life. Does the method require physical restraint, confinement, and other means of limiting a dog's freedom? Because dogs are often part of daily life in many households, management has become a big part of how we deal with our dogs. Dogs have social, mental, and physical needs and trainers need to make sure that management choices are not so intrusive in the animal's life as to cause mental and physical harm. He gave a an "intrusiveness hierarchy" which is (from least to most): R+, DRO/DRA (differential reinforcement of alternate or other behaviors), -P and -R and extinction, +P.
Social acceptability is a function of your culture, peer pressure, and pressure from superiors. Don't underestimate the importance of social acceptability as this can be a big factor in the long term success of your training. These days there is a lot of emphasis on quick fixes (he mentioned owners wanting relief), but there is also an awareness of the needs of animals. Trainers need to educate their students to look for long term goals and this can be challenging in a society that wants results. But if we can teach people that it's important to look a little farther ahead and aim for solutions that are effective and minimally intrusive, we can start to change how society views animal training and change what falls into the category of socially acceptability.
A lot of what Steve talked about with dog training is also true of horse training. Compared to dog training, horse training is still dominated by people who rely heavily on negative reinforcement and punishment. In many places, hand feeding is forbidden and there is a huge emphasis on gaining your horse's respect, which usually translates to "if I say move, you better move..." I think that as horse trainers we have challenges with showing that positive reinforcement training is as effective as other methods because it sometimes requires more work up front than some other training methods. Even though it gives better results long term, it can be hard to get people to try it for long enough to see how effective it is.
I really liked when he said that the best way to promote positive reinforcement is to create a superior product. If we can show the horse world that clicker training creates dependable, capable, and enjoyable horses for any use, then it will be seen as an effective training method. This will go a long way toward increasing social acceptability because I think there are a lot of people out there who would love to use more positive methods with their horses but have to fight social pressure to do it.
Alexandra Kurland: Give me a Break: How to Give Breaks Without Giving Breaks
I'm not sure Alexandra Kurland needs any introduction here, but if you haven't heard of her, she has been a pioneer in bringing clicker training to the horse world. Her web site is www.theclickercenter.com. Her work focuses on teaching good training skills, body awareness, emotional stability, biomechanics of the horse and rider and more.
Alexandra started right off by saying that she doesn't like taking breaks and her horses don't like taking breaks. But she knows that breaks are good for learning. Breaks allow the horse time to process, and allow the trainer time to evaluate the horse's progress and plan the next training session. Breaks are an important part of the learning process for those reasons, but in addition breaks can function as reinforcers and can be used to mark the transition to a new criteria.
If you look at dog training, many dog trainers work in very short sets, especially if they are training a new behavior. The dog has a chance to come out and work and then goes into its crate or on to its mat to wait for the next short session. With horses I find that the tendency is for people (me included) to take the horse out and work with it for a longer period of time before it goes back to its living quarters. If I go riding, I might have my horse out for 1-2 hours from start to finish. This doesn't mean the horse is actively working the whole time as there are some breaks built into the process, but it does mean that I need to be mindful of when and how I give breaks to use them in the most beneficial way.
So how do you give breaks as a horse trainer? You certainly have the option of returning your horse to its stall or paddock for a longer break, but if you just want to take a short break, she suggests a few ways to do it. One way is to give the horse a mental break by switching to an easier behavior. You can do this informally by mixing up what you work on so that harder tasks are mixed in with easier tasks. Or you can take a break by taking the horse for a little walk or allowing it to stand casually for a few moments. If you want to plan your breaks to mark a change in criteria, you can use the "Microshaping strategy" or Loopy Training to help plan your breaks.
The Microshaping Strategy uses the idea of alternating between two behaviors in a specific way She came up with the strategy when she was teaching microshaping, but you can use it for any behavior. The idea is that you have two behaviors (A + B). You are actively shaping behavior A, but your reinforcement rate is bit lower than you would like, so you insert segments of clicking and reinforcing behavior B into your session. Behavior B is an easy, well known behavior that is used to increase the overall rate of reinforcement of the session. Alex often uses targeting as behavior B. Note that you are not trying to shape or change behavior B.
For example if you were shaping backing and the horse was nicely shifting its weight back, you might offer a target after the click and treat for a good weight shift and then let the horse target a few times. You would then return to backing, but you might start looking for the foot to come off the ground as the next criteria. When you were getting some nice foot lifts, you could offer the target again after you clicked and treated for a better foot lift. You are clicking and treating both behaviors, so the targeting functions as an additional reinforcer for the better effort in behavior A as well as a break and chance to earn reinforcement for known behavior B.
Over time the horse learns to anticipate a change in criteria after the targeting sessions and this is a nice way to provide a little indication that you have now changed criteria. This can reduce frustration if a previously clicked behavior is no longer being clicked. If you want to know more about the Microshaping strategy, it is described on Alex's microshaping DVD.
Alexandra used the concept of Loopy Training to show another way you could decide when to insert a break. She spent some time sharing the basics of Loopy Training. If you have not heard of Loopy Training, she has a DVD on the subject and I have also written an article that explains it in some detail (Loopy Training Article). The idea behind Loopy Training is that good training is a continuous cycle of behavior -> click -> treat -> behavior -> click -> treat instead of just isolated instances of behavior -> click -> treat. If you view a training session as a more continuous (but cyclical) process, it becomes easier to see where to insert a break. You can insert a break when the animal is smoothly repeating the desired behaviors without any extra behaviors or pauses and is getting reinforced on a regular basis. Alex calls this a clean Loop. Taking a break reinforces the clean loop and prepares the animal for new criteria that can be added to the next loop.
In Loopy Training you can use a second behavior as a break (in the same way the Microshaping strategy does) or you have other options such as going for a little walk, putting the horse on a mat and patting or scratching him, allowing him to stand at ease, graze a little bit, or do something else he finds reinforcing. For many years when I let Rosie take a break, it meant giving her a few minutes to watch the squirrels in the hedgerow or whatever caught her eye.
During the break you can assess how things are going and plan for the next session of work. If the break has been well timed, you will see a significant improvement in the animal's behavior. Another advantage to taking breaks is that the horses become more comfortable with pauses in the training. Some horses get agitated when the reinforcement rates change or they think the game might be over. By stopping and restarting multiple times, these horses get more confident that a break in the game is temporary.. They learn to transition easily between different behaviors and different rates of reinforcement and they learn to enjoy periods of rest from mental or physical effort.
I apologize for the shortness of this description of Alexandra's presentation. Since most of my readers are clicker training horses, I feel like I should have more information here. I can only say that Alex used a lot of video material that I had already seen so I didn't take very good notes and only realized later that this makes it hard to write up a good report. I also find it harder to take notes when the material is familiar to me. If you are interested in learning more about how she uses breaks and have specific questions, you can email me and I'll do my best to answer them.
Kay Laurence: Target, Lure or Free-shape? What Difference Does it Make?
Kay Laurence is a dog trainer from the UK who has been a pioneer in using clicker training for dog training. Her website is www.learningaboutdogs.com and she trains dogs of all kinds and for many different types of jobs (agility, freestyle, etc...)
There is always some controversy over different methods used in clicker training and she wanted to compare the same three behaviors (4 feet in a pan, walking along a pipe, and going around a cone) trained in three different ways (luring, targeting, and free-shaping) with three different dogs to see if there were any significant differences in how quickly the dogs learned the behavior and how resistant it was to distractions. This was not intended to be a fully controlled experiment but she tried to keep as much the same as she could.
Before she got into the experiment itself she talked about learning and how there is a learning continuum from self-directed to directed learning. Self-directed learning would be micro-shaping, free-shaping and self-directed learning. Directed learning is where the trainer provides more information to the animal through the use of barriers, molding, pressure and release, luring and so on. Targeting is somewhere in the middle in what she calls guided learning.
When she presents this as a chart, she draws a line with directed learning on one side (the left) and self-directed learning on the other (the right) and this represents the learning continuum. Underneath it she draws a shape like a triangular flag which has the wider side under self- directed learning and comes to a skinny point under directed learning. This represents the mental stamina of the learner. Directed learning requires the least mental stamina because the learner is getting lots of direction from the teacher. That's why it's the skinny end of the flag. Self-directed learning requires the most mental stamina because the learner has to figure it out on their own. That's why it's the fat end of the flag.
I think it's important not to attach value labels to parts of this continuum. It is not that self-directed learning is better or worse than directed learning. You get different things from different ways of learning and what's important is to recognize what your learner needs. One of her points is just to show people that there are costs and benefits to different types of learning. Also, some behaviors are more suited to one kind of learning than another.
Clicker trainers tend to lean toward self-directed learning but she pointed out that this requires more mental stamina. And you also have to consider safety. One would not learn to scuba dive or use a chain saw through self directed learning. On the other hand self-directed learning would be appropriate if you were training a safe behavior that the animal needed to be able to do independently and might require some on the job problem solving.
For her experiment, she used three border collies who had various levels of training and were all keen clicker dogs. Each dog learned one behavior through each method and she had video clips of different stages of the training. Just from watching the videos, I couldn't see any obvious pattern although it did seem like some methods were more suited to some behaviors than others. For the work in the pan, some of the dogs were more careful about their foot placement than others. Free-shaping walking along the pole seemed difficult.
I'm not sure if she took the behaviors to completion, but she did find that there was no obvious different in acquisition of the behavior across the board. All the dogs learned all the behaviors. She did notice that the behaviors were not all exactly the same. So even though all the dogs learned to put their feet in the pan, they each had a slight variation in how they did it. Likewise the dogs followed the pipe and went around the cone, but each with their own variation. For example, the dog who learned to follow the pipe with food luring traveled with more focus down on the pipe/ground than the dog who was free-shaped to follow the pipe. With the behavior of putting 4 feet in the box, the dog who was free-shaped was clearly more aware of where his feet were and was better at getting in and out of the box.
To test for distractions, she had someone sit in a chair next to the area where the dog was working and observed how this affected the dog's behavior. The dogs were all friendly and wanted to engage with the person but they all kept working. She did have to make some adjustments to her training plan, especially for the cone work as the dogs would alter their behavior slightly so that they went toward the distraction and went around the cone.
In the final analysis, she said choosing a training method should be based on evaluating a few different criteria. I have listed them in order here:
First consideration: Mechanics of the behavior. You have to look at what the animal needs to be able to do to perform the behavior. What are the physical and mental skills? You also need to look at how you are going to use or cue that behavior. A behavior that requires self-awareness and independent thinking is going to be more suited to free-shaping. A behavior that is going to rely on a hand cue in the end might be more solid if it is taught with luring as the hand cue can be part of the behavior from the beginning. A behavior that will not use a hand cue should not be taught with luring, but might be better suited for training with a target.
Second consideration: Confidence of the learner. These three different training methods fall in different places along Kay's learning continuum so it's important to consider the mental stamina of the learner. A dog that lacks confidence, or is new to training and hasn't yet learned to offer variations on behavior to get reinforced, is going to do better with more support from the trainer. A dog with a lot of experience in free-shaping or micro-shaping might do better if it is allowed to work independently.
Third consideration: Skill of the teacher. Trainers become more proficient with the training techniques that they use the most. Kay considers herself equally skilled in shaping, luring, and targeting so she has all the options available to her. A trainer who does a lot of luring, but not much shaping, is going to be more skilled at luring than shaping and this should be taken into consideration when choosing a training method. I don't think this means you should only use a method you know well, because that would limit your growth as a trainer. I think what it means is that if you want to get more skilled at using one of the other methods, choose carefully what behaviors and what animals you use so that the animals can still do well even if you make some mistakes.
The overall feeling I got from her talk was that there are a lot of variables that come into play when choosing how to train your animals. It is important to be attentive to the immediate needs of the learner as well as looking at what you hope to get long term from the training. While we tend to focus on training 'behaviors" with the end result as the important part, every time we train a behavior there is an opportunity for both the trainer and the learner to become more skilled at some aspect of training and by choosing wisely we can both expand our abilities without creating frustration.
I thought it was interesting that both Kay and Steve talked about choosing training methods but from very different angles. Part of it is a reflection on their jobs, but I thought they also complemented each other nicely. Unless you are very specialized in what kinds of training jobs you take, most trainers are going to end up doing some training which is about training new behaviors (for dog sports or specialized jobs) and some training which is about dealing with problem behaviors. It is helpful to have advice on different aspects of choosing training methods.
Ken Ramirez: Lessons from the aquatic world
Ken Ramirez is an expert on marine mammal training and is the executive vice-president of animal care and animal training at the Shedd Aquarium. His website is www.kenramirez.com.
Ken's talk was about how concepts that came out of the marine mammal world have been transferred to trainers working with other species. He started by talking about how positive reinforcement was first used in the marine mammal world partly because of necessity but then continued (and really took off) when the trainers realized that there were a lot of benefits to positive reinforcement training. The use of positive reinforcement improved the animals lives in ways that went beyond what they had been able to do before with traditional methods.
In this presentation he chose to talk specifically about the LRS (least reinforcing scenario), which is one tool that marine mammal trainers use and he shared how this could be used with other species. The LRS was originally called the Least Reinforcing Stimulus for anyone who remembers that term. It is now called the Least Reinforcing Scenario because it is not always a specific stimulus. Ken described it as a way of standardizing what you do when you ignore unwanted behavior. Most of the examples he used showed using a LRS when the animal did not respond to a cue correctly, but I think the LRS is also used when the animal offers an unwanted behavior during a training session.
Here is his description of how to use a LRS:
It is a 3-5 second neutral response (the duration will vary depending upon the speed at which you are working, think sloth vs. dog)
It is intended to make an impression on the animal (wrong answer) without creating frustration
It is not a fixed posture so it's not about "freezing." He says that you want to change as little as possible about what you are doing, just do a neutral pause. So if you are looking at the animal when it makes the mistake, you can keep looking at it. You don't have to break eye contact or do anything to indicate that you are having a reaction to the animal except hesitating for a moment. Your intention is to be as non-reactive as you can, so that the only thing the animal observes is the break in the flow of reinforcement.
You are looking for a calm response to the LRS. This doesn't mean you do the LRS until the animal is calm, but that if the animal seems to get upset when you use a LRS, then you need to change how you do it.
And most important... the animal should be immediately asked for an easy behavior right after using the LRS so that reinforcement starts coming again. When he first said this I sort of wondered if you might get some interesting results if the LRS predicted an easy opportunity for reinforcement on the next behavior, but apparently that does not happen. The LRS stands out as a pause in the flow of reinforcement so it clearly marks the "wrong answer" and if the trainer asks for the behavior later in the session, it is usually performed promptly.
The examples Ken showed of him using an LRS were all during sessions where the animal was performing behaviors that were already on cue. I don't think he uses it during a shaping session for new behaviors, but I can't say for sure. He had some video clips of him working with animals at the Aquarium and also with his dog and in each case, there was a very slight hesitation when the animal responded incorrectly to a cued response and then the training moved right back on.
One thing I will mention is that when he is training (as seen in the video clips), he is very precise and clear in his actions so that when he pauses it is very meaningful to the animal. If you are frequently distracted or train in a more casual way so that you and your animal are not totally focused on each other, I think it would be harder to have the LRS be as meaningful.
Ken had invited two speakers to share his time slot. They are both trainers at Sea World and talked about their training program. Steve Aibel and Mike Pool talked about "Balance" and how animals have physical, social and mental needs.
Trainers at Sea World are responsible for giving signals, observing behavior and providing consequences. They are also responsible for the physical, mental and social needs of their animals. He and Mike presented two "terms" that the use to describe what they do. The first term HELPRS is about long term planning to create balance in an animal's life.
I have included the list here because I think it's a good reminder of how we can use our training to impact the lives of our animals and how training should address each animal's needs. They are:
The second term VRRV is about balance in training sessions and can be written out as:
They emphasized how important it was to make reinforcement variable and not fall into patterns. Mike mentioned a few times how patterns can decrease motivation and how you don't want to let a pattern dictate what you do, that you want the animal to tell you what it needs. I'm still thinking on this one as I'm not entirely clear what he meant. I think perhaps it is as simple as observing the animal to make sure that we are providing the right reinforcement to keep the animal actively engaged in the learning process so that training doesn't become routine and boring. But that's just my guess.
Finally they shared a little bit about how marine mammal training has sparked some innovative programs that teachers and educators are using and how this can change children's lives by promoting positive reinforcement and a sense of working together among children. I know that learning about clicker training has certainly affected how I interact with people so this makes perfect sense to me.
I'll leave you with the 3 ways to provide balance in your life that Steve Aibel shared. He got this from a speech by a basketball coach, but I didn't catch his name.
Laugh (find joy)
Move to tears (be moved and connected)
At the end of the formal presentations, they had a panel discussion where all the presenters were on stage to answer questions submitted by the audience. I don't have the exact words for the question, but this was the general gist of it: If I am training my dog and I am having fun, as shown by my body language and enthusiasm, does this make it more likely that my dog is having fun?
Since the panel was a mixture of animal trainers, educators and scientists, there were pretty varied responses. I didn't write down all the answers but some that I did note were "I don't know, I need more data," "happy trainers provide better reinforcement," "what is the definition of fun?" and "dogs like it when we are happy because then they know they are safe."
It was a bit of a loaded question because as they pointed out, "fun" is not a clearly measurable quantity so evaluating an animal's responses for "fun" is not easy to do. It would be easy to answer the question as "yes" based on anecdotal or personal experiences where the dog seems to get more enthusiastic when we do, but that doesn't mean that other factors could not have been important, and that still doesn't tell us if the dog is having "fun." One member of the panel did suggest that it would be interesting to set up an experiment to see if the attitude of the trainer affected the number of reinforcers delivered or the speed at which new behavior was acquired.
I had sort of mixed feelings about the question itself. This was a conference that was exploring positive reinforcement training and one of the reasons many of us have found and stayed with positive reinforcement is that it is a training method that promotes the creation of thinking and confident learners. Is this the same as creating animals that are having fun? I don't know.
I do think that if I bring up my energy level and enthusiasm, and my animal doesn't respond by mirroring the same emotions to some extent, then I need to look at why there is a disconnect. But on the other hand I need to recognize that all animals are different. Bringing enthusiasm to my training with my dog makes him more energetic, but with my guinea pig it sends her running to her house.
For most of the above report I tried to pretty accurately present the information from each talk, although I did insert a few comments here and there. I also filled in details where my notes were a bit incomplete so I hope that I have not added too much of my own interpretation to the material presented. I think I've gotten spoiled by Clicker Expo where I can get a notebook with the power points and just add my own comments. At any rate I did want to share a few thoughts about what I got out of the conference.
I really enjoyed the conference and thought it was a nice mix of speakers and covered a wide range of related topics. Prior to learning about clicker training I just thought of myself as a horse owner and rider, although I always had an interest in animal behavior. Now each year I learn a little bit more about topics that are related to what I do, and I see just how much more there is to learn. Sometimes this is overwhelming, but other times I find it comforting to know that there is information out there that can help me learn to be a better horseperson, trainer, teacher, and communicator.
The conference didn't have an overall theme, but there were some common threads. I came away with a reminder that it is the teacher's job (whether the student is an animal or a person) to be mindful of the small details that can make a difference in how well we communicate. Good training requires a lot of careful planning and preparation up front so that the actual implementation goes smoothly. This includes knowing your animal, all the details of the behavior you want to train, and having enough tools in your toolbox so that you can choose the best one for the job. I also felt encouraged to see that many of us face the same training problems regardless of what species or behaviors we are training, and that we can support each other on our own individual journeys.
And as a final note, this report only covers the one day conference. I was actually out there for 3 days which included the conference and two days of private talks hosted by Dr. Rosales-Ruiz with the presenters and invited guests. As I have time I will write up my notes from the other two days and share them. Thank you to Alexandra Kurland who kindly invited me so I got to go for all 3 days. And thank you to Dr. Jesus Rosales-Ruiz and the ORCA students for their work and being such good hosts. I enjoyed each and every talk and meeting everyone.
Katherine Bartlett, copyright 2012. Please do not share any part of this document without my written permission.