EQUINE CLICKER TRAINING.....
using precision and positive reinforcement to teach horses and people
ORCA's Art and Science of Animal Training Conference: 2013
On February 2, 2013 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 5th year the conference has been held and the list of speakers included Susan Schneider, Bob Bailey, Phung Luu, Steve White, Alexandra Kurland, Kay Laurence, Ken Ramirez. There was also a panel discussion.
Last year I posted "ORCA tidbits" on my facebook page and wrote one report on ORCA that included all the talks. This year I wrote a "note" for each talk and posted it on my facebook page. This article is broken into sections, one for each "note" and you can read it from beginning to end or skip around if you prefer. If you have already read the "notes" on my facebook page, this is the same material. You are welcome to read it again, but I didn't want anyone to expect anything new. If you want to read more about ORCA, Mary Hunter also blogs about ORCA on her www.stalecheerios.com website.
This year's talks are listed below. You can skip to a specific one by clicking on the link, or just start reading.
Dr. Schneider’s talk was based on her book “The Science of Consequences” which explores how consequences shape our behavior. For most of us who are already involved in clicker training, this may seem obvious, but she has done extensive research to show just how widespread this was in all areas of our lives and shared a lot of interesting examples that stretched my mind a bit in thinking about reinforcers and how behavior can be shaped. I believe that most of her examples were taken directly from her book, so if this sounds interesting, you might want to put it on your reading list. I should point out that while the book is titled “The Science of Consequences,” she focuses almost entirely on reinforcement.
She started by talking about different kinds of reinforcers and besides the obvious ones (food and other primary reinforcers, access to desired objects, removal of aversives), she included variety and control. Later in the talk, she included attention as a reinforcer. A lot of people talk about the value of giving the animal a “choice” and I think this is another way to say animals are reinforced by the ability to control things about their environment.
She used the example of deer mice that prefer darkness but will still turn a light on and off if they are allowed to control it. It seems to me that this might be a case where they were reinforced by control and variety as turning the light on and off added variety to their environment. She also talked about enrichment programs at zoos and how animals prefer toys that are unpredictable (which provides variety). A lot of enrichment programs also provide hidden food so the animal can choose when to go look for it.
When you are studying animal behavior and looking at consequences, there are a lot of different factors that can come into play. She mentioned a few specific areas that are of interest to animal trainers. Here are some of the ones she discussed:
Delays in reinforcement can have a strong effect on the value of the reinforcement. If the reinforcement is delayed by too much, it loses its value. But, you can build “emotional self control” which will carry over from one area to another so the animal can learn to make choices even if the better choice does not provide immediate gratification. She had quite a few examples of animals learning to choose the “more-later” choice in which they chose to delay reinforcement to get a bigger reinforcement later.
Consequences can be strong enough to create one-trial learning and lead to superstitious behaviors where a behavior is maintained because of some initial reinforcement, even if it has not been reinforced in subsequent sessions. This can be problematic for trainers because these behaviors are very hard to eliminate.
Operant learning affects brain development and scientists can actually measure changes in the brain after an animal has learned behaviors through consequence based learning. They can also measure chemical changes. There has been research showing that animals will work if the reinforcement is stimulation of the pleasure centers in the brain. There has also been research into the effect of the environment on an animal’s genes and how this affects its offspring. She cited a study where they looked at mice that were stressed and how this made their offspring less able to learn through operant learning.
She shared an interesting study about animals learning through observation. One group of pigeons was taught to peck through a paper cover to get to the seed inside a cup. Then they had other groups of pigeons observe the first group to see if they would learn the same behavior. One group watched the pigeons peck through the paper, but did not get to see the pigeons eating the seed. Another group saw the pigeons eating the seed, but did not see them pecking. A third group saw the pigeons pecking through the paper and eating the seed. Of the three groups, only the third group learned how to do the behavior. The pigeons had to see both the pecking and the reinforcement in order to learn how to do the behavior themselves.
Her talk left me with a better understanding of some of the complexities of using reinforcement to train behavior, an appreciation for all the things that animals have to learn in their daily lives, and a bit of amazement for how well animals can learn, even when they are being taught behaviors that are not natural for them. As Bob Bailey likes to say “We do not have to teach animals how to learn. They are learning machines.”
Bob’s talk was jam-packed with information about how to become a better trainer. He used examples from his own work to illustrate the importance of various points and it’s always fun to see all the different kinds of things that he has trained over the course of his career. In the interest of clarity, I am just going to present the content of his talk as a list of key points, but the talk itself was a nice mix of stories about training projects and what he learned from them.
There are two significant factors in predicting training success. They are reinforcement “density” and duration. Together these create a total reinforcement history for a behavior. Trainers want to maximize reinforcement density (high rate of reinforcement) but still make sure that they are only reinforcing behaviors that meet criteria. If you reinforce for less than you want, that doesn’t count! You can maximize your training density by working at short distances without distractions. He suggests starting in as sterile environment as you can and then adding distractions and complexity. “A weak behavior at 30 meters will not be a strong behavior at 300 meters.”
On a practical note, he mentioned that balloons are an easy way to add distractions. You can start by tethering them low to the ground and then slowly increase the difficulty by giving them more freedom, letting the drift around, and even adding fans to move them in unpredictable ways.
Getting in enough duration is just a matter of doing enough trials until the behavior is fluent. He says to expect it to take hundreds of trials to get a behavior fluent. This is why you need to keep records and take data. In every talk I have ever heard Bob give, this is a recurring theme. You have to keep records in order to be able to evaluate your progress and adjust as needed. Take Data! Take Data! Take Data!
So how do you set yourself up to be able to control reinforcement density and duration? You make a training plan. Bob divides training into 3 steps which are Think, Plan, and Do. In the “Think” phase you get to be creative and come up with possible ways to train the behavior you want. In the “Plan” phase you pick a particular strategy and come up with a detailed training plan including details about the set-up, criteria shifts, and so on. In the “Do” phase you go out and put your plan into action. He emphasized that you don’t mix them up. You should not be doing or planning when you are thinking. You should not be doing or thinking when you are planning, etc… And most importantly, you should not be thinking or planning when you are doing!
This is not a one-way or linear process. You can go back to thinking and planning on any time. Planning is when you can correct errors in your procedure, clean up details, and adjust for unexpected behavior on the part of your trainee. If you have trouble during the planning phase, think of creating a training plan that takes you from what you have to what you want. Planning is also when you need to try to simplify where possible. It is easy to make complicated plan. It takes work to make a simple plan.
Time is the trainer’s most valuable resource. At ABE, they had to learn to train efficiently and effectively in order to stay in business. He learned the value of defining behavior clearly and objectively. There are different ways to collect data. Being able to count behavior is one way to evaluate how the training is going. This is another place in his talk where he mentioned being precise about what you reinforce. It’s not enough to be feeding the animal. You must be marking correct behavior. Correct behavior means behavior that meets the criteria for the current stage of training. This includes behavior that never met any criteria as well as behavior that previously met criteria but does not meet the latest criteria. Feeding for behavior that does not meet criteria is going to create problems because you will have early behavior competing with late behavior. (Early refers to behavior that was trained early on in the plan vs. more finished behavior trained later in the plan).
Time is saved or lost based on what happens during the planning phase. This is where you get to make your plan and test it out. Doing rehearsals is part of planning and you should do them until you can do it without the animal, first in your mind and then on location with all the appropriate equipment. During planning you can run through all the possible scenarios and make sure you have them covered. But….he does say here that you need to be careful of over-analysis. I can’t quite remember how he stated this, but I think the general gist of it was to plan for as much as you can, but not allow that to clutter up your training plan. Simplify it instead.
If this all sounds like you can never do any training by just going out and having fun, or training in a more informal way, Bob did add a few comments here about this. He pointed out that sometimes the plan is to go out and see what the animal can do. That is a plan, just as much as a detailed series of steps is a plan. If you are working with a new animal or species with which you have no prior experience, you may need to just go out and do some training sessions to get a feel for what the animal can do and what seem to be the most effective ways to shape behavior.
Here is his list of what a good trainer should be able to do:
2. Apply a few rules
3. Develop mechanical skills
4. Adjust to the circumstances
5. Know what you want
6. Make it worthwhile for everyone involved (this includes the trainer).
Along with this, here’s his list of what good training includes:
1. Precise timing
2. Observing behaviors
3. Quick decisions
4. Fast trials
5. More trials
6. More reinforcements
7. More expectations
I am going to add a few other pieces that he mentioned elsewhere in his talk. Good trainers identify the skill sets the animal needs for any given behavior. They learn the history of their craft. If you learn the history of your craft, it will explain a lot of the myths about why behavior has to be trained in certain ways. Good trainers are also willing to change. Over the course of the weekend, I heard again and again that the best way to improve our training is to change ourselves. To effectively teach or train, you must be willing to change your own behavior.
He had a few final thoughts about improving your training. He thinks everyone should videotape at least some of their training. Video is the best way to see what is really happening during a session and you should plan to videotape daily. You don’t have to watch it all, just watch enough to see where you are making errors (timing, food delivery, etc..) or the parts where you are having trouble. In a separate discussion over the weekend, he said he recommends you watch a video clip 3 times. The first time is to get a general idea of what is happening. The second time is to focus on what the animal is doing and the third time is to focus on what you are doing (I’m not sure if it matters what order you do these).
I’m going to add a little note here that he did mention the use of punishment. This came up at the panel discussion too so I will elaborate on it later. People often ask him is he uses punishment in his training or if he used it at ABE. He said he doesn’t tell people they can’t use punishment, but he does require that they submit a plan for when, why, and how they are going to use punishment. He found that forcing a trainer to justify and explain the use of punishment was an effective way to get them to reconsider using it. When forced to write a plan, they always seemed to come up with some other way to get the behavior they wanted.
And he left us with this final message. “Beware of he or she who claims the truth. Follow the Science.”
Phung Luu’s talk was specifically about birds, but even if you don’t work with birds, I hope you will read this report because body language is important in working with any species. In addition, I was surprised to see how many similarities there were between the body language of birds and the body language of horses (and probably other species as well).
He started by defining body language as “nonverbal communication through the use of gestures and movements.” There are different levels of subtlety in body language, which means that within each “behavior,” there are sub-behaviors. As you become better at reading body language, you will become more aware of all the nuances of a behavior and of a particular animal. In any discussion of body language, you also have to include the body language of the trainer and he described how there is a feedback loop between the trainer and the animal. The feedback loop consists of antecedents, behaviors, and consequences.
Why learn about body language? In a training situation, a knowledge of body language can help the trainer to reduce unwanted behaviors. The earlier you can spot a change in the animal’s behavior (particularly those related to stress), the easier it is to redirect the animal or do something to reduce the source of the stress. Understanding body language also makes it easier to increase wanted behaviors. If you can “read” an animal well, you can often predict when a desirable behavior is going to happen and that makes it easier to reinforce those behaviors.
It takes time to learn to read body language. The 10,000 hour rule is often used as a guideline for how long it takes to master a new skill and he said that learning to observe and recognize changes in body language is a skill that has to be learned like any other. It takes time.
He went on to talk specifically about body language in birds, using pictures and video as examples. I don’t know much about birds so it was interesting to me to see that I could still guess what their behavior meant at least part of the time, but not at other times. He showed a little video clip of two parrots fighting and some of their aggressive posturing was confusing (to me) as some of the aggressive posturing reminded me of play posture in other species.
Birds rely heavily on visual cues so he thinks they exhibit a wider range of body language than other species. (do you think that’s true?) When he is observing birds he focuses on three things: eyes, feathers, and posture. Here are some things to consider about each of them:
direction of focus (in general, birds have to move their heads to change their focus so eye movement is also associated with head movement, although subtle eye movement might not be).
Intensity of gaze (soft vs. hard look)
Size of pupil (in some birds). He used parrots as an example and I didn’t quite follow what the eye was doing but apparently the pupil changes with intensity of emotion so a change in the pupil could mean the bird is either happy to see you or hates you. That means you have to look at the whole bird to see what the eye means. If anyone knows more about this and wants to add to what I have written, let me know. I hate to post incomplete information but I did think it was interesting that this could go either way (loves you or hates you).
Tight feathers are a sign of tension and indicate the bird is poised for flight
Relaxed feathers indicate calmness and are found on birds who are comfortable in their surroundings.
Erect feathers indicate tension or aggressiveness.
If a bird is not completely relaxed, it will show erect feathers in some areas and relaxed feathers in others. If you are not sure, the neck is a good place to look as that is one of the first places where erect feathers will show up.
When reading feathers, you should be aware that how the feathers look can also be affected by temperature and health. A sick bird may look relaxed as the feathers will not be held tightly, but that can be misleading.
Posture includes head movement, beak and foot movement, and overall body carriage.
Head movement is a useful indicator. Parrots do a lot of head posturing, tipping it from side to side or dirt flipping if on the ground.
Overall shape or body carriage: A bird that is preparing to do something will drop its head and have a hunched posture. This is often seen in raptors.
Beak or foot movement is often the last line of communication before the animal acts upon its feelings.
Flight -I am going to include flight in this category because he talked about it next. I’m not sure if it really counts as a “posture” but a bird prepares for flight before it goes, so I think it’s reasonable to assume that a trained eye could see postural changes before the bird actually leaves. Flight could be either distance increasing or distance decreasing so you do have to look for other indicators to see if a bird is coming toward you because it is pleased to see you or is acting aggressively.
Learning to read body language is just a matter of refining your observation skills. You want to observe the eyes, feathers, and posture. He refers to this as the direct (center-immediate) observation zone. You also want to watch the general (peripheral) observation zone so that you are aware of what is happening around your animal. Watching how the animal responds to everything in its environment can help you see connections between behavior, environment, and action.
Sometimes an animal will present unfamiliar body language. If you are not sure what something means, it can be helpful to identify the antecedents and see how the animal ultimately responds. One useful piece of information is if the behavior is followed by distance increasing or decreasing behavior. This can help determine if the animal is feeling threatened or at ease which can then help you narrow down what the behavior means.
I always find it is interesting to learn more about training different species. When watching the videos, I was clearly reminded that working with exotics is different than working with domesticated animals. There needs to be a whole different level of awareness of what the animal is doing. I try to pay attention when I am working my own horses, and pay more attention when working with someone else’s, but honestly I do take for granted that they like me and are not going to hurt me. As someone who started clicker training because of an aggressive horse, this says a lot, but still …the work with the exotics seems different to me.
At the same time, I did notice that there are a lot of similarities. Raising the feathers (or hair) on the neck is a common sign of aggression in dogs. Head posturing is common as a precursor to aggression in many species (horses, dogs, cats..). Changes in posture are obvious in many animals. Someone in the audience asked about “freezing” when meeting another bird and Phung said this was something he has observed. I think one of the best ways to learn more about the animals you usually train is to go work with something else and then come back to your own with fresh eyes. This talk provided a nice reminder of the types of skills trainers have in common and how to improve them.
Steve started his talk off by asking the audience if they had ever experienced a training plateau. He got a lot of “yes” answers. Then he asked if people thought it was normal to have training plateaus, or had been told it was normal. Again, he got a lot of “yes” answers. From the way he asked the questions, it was pretty obvious that he didn’t go along with the idea that training plateaus were unavoidable or that one should accept them as part of the process. Today he was going to share his view of training plateaus; what they are, how to avoid them, and what to do about them.
What is a training plateau?
In a training context, it is a multi-session stall out in skills acquisition.
In an operational context, it is a protracted stall in broadening of generalization that is of undetermined duration.
Note that in his definition he includes both stall outs while you are still actively working toward a training goal, and stall outs that happen because you just stop learning new skills as might happen with an older animal or one that has finished being “trained.” He expects police dogs to keep learning new skills until ….well his exact words were until either the dog or the handler is at room temperature, but you get the idea.
Later on he also pointed out that a training plateau is a sustained stall out. He does not include momentary dips that are part of the normal training process. So duration is important when deciding if you are really looking at a plateau or not.
There are some differences between plateaus in a training context vs. an operational context. These can affect how you handle plateaus. Some of the factors he mentioned are consequences of failure, strategies that are available, if the results are critical (mission critical results), and opportunities to go back to training and get more information about what the failure is telling you. Failure is information. It teaches you the edges/boundaries of your behavior.
Why do trainers hit training plateaus?
Clip end issues (clip end means a problem on the dog’s side):
Resistance to distraction
These all come from frustration or confusion on the dog’s side and animals can go into avoidance, shying (disconnection of gaze), freezing, or opting out (walking away)
They can also lead to reactivity such as aggression, redirection, and displacement behaviors (animal does something instead of the cued behavior).
Loop end issues (loop end means a problem on the handler’s side):
Avoidance: people drop out, don’t show up for a class, skip exercises
Failure to do work outside class: A big part of this is that they don’t feel successful outside of class because they have incorrect expectations
Low rates of +R
Frustration which often leads to punishment.
He spoke a little bit about how punishment is reinforcing to the handler and people feel it is “justified” because it seems to work. But in the long run, animals develop tolerance for punishment so you had to do more to get the same result.
The key to avoiding punishment is to make people stop and think about it. Punishment often happens as a knee jerk reaction to frustration. He compared punishment to “carpet bombing” where indiscriminate use of punishment just suppresses the animal and makes the handler addicted to punishment. If you flatten everything (the bomb analogy) you might have gotten rid of the problem, but you’ve gotten rid of everything else too.
What to do about training plateaus.
Since many trainers encounter training plateaus, there are lots of solutions out there that are recommended to help get you back on track and make progress. He listed ineffective and effective solutions.
Ineffective solutions: Most of these are “conventional wisdom,” common sense solutions and R+ driven.
Ineffective Reinforcement: The first reaction most people have when behavior stalls out is to look at their reinforcement. They get caught in a non-productive cycle of increasing the value or rate of reinforcement and decreasing expections (to get in more reinforcement) and this risks sustaining the plateau. Usually what he sees is that as the reinforcement rate goes up, the performance rate goes down.
Misapplied cues – re-issuing the cue does not work
Misapplied NRMs (no reward markers)
Misapplied KGSs (keep going signals)
These don’t work because if they are applied when the behavior is failing, it reinforces failing!
Effective Solutions: Effective solutions are principle driven, momentum sustaining, and R+ driven. Included in here is how to avoid training plateaus because getting out of a training plateau and avoiding a training plateau use the same strategies. He used two visuals, turning a flywheel and jacking up a house.
flywheel metaphor: flywheels are hard to get started, but easy to keep going. In training you want to keep the momentum going which means you want to be sensitive to any drop in progress and catch it before it becomes a big thing.
Error-less learning or less-error learning: If you keep the animal successful, you will keep momentum going
Component rotation: Most behaviors have some common types of criteria (accuracy, speed, latency, intensity, generalization, duration, and distraction). When you train you should rotate through these components, improving each one a little bit and then moving on.
This will keep your behaviors in balance, keep it interesting for the dog and handler, and provide a road map for steady progress.
He compared it to jacking up a house one corner at a time. If you jack one corner up too much, the house will become unstable (or break) so you just move each corner a tiny bit and then move on to the next one until you reach the end goal.
At the end he talked a little bit about some other elements that can contribute to continued progress to help avoid hitting plateaus. It is important to train in different environments. You start in a simple environment but then move to different environments, moving from the training environment to transitional environments and then the operating environment. He usually has to work in 30 different environments before he has an operational dog.
Video and working with a partner (or two partners) is another element that can add to your training success. He finds that having feedback from an observer can make a big difference in maintaining forward momentum. He called this getting a “reality check.”
He finished with “Plateaus in a Nutshell”
Level the house as you raise it (work on fluency and generalization)
Keep the flywheel turning
Use reality checks
Prevention is the ultimate “remedy”
And… Keep it fun at both ends of the leash.
I am going to guess that most of my readers know the story of Clever Hans. If you are not familiar with it, I suggest you use a search engine (just ask about “Clever Hans”) to look it up. Alex started her talk with the story of Clever Hans and explained that the lesson many people take away from it is that animals can’t really do math. But the real lesson is that horses are very good at reading body language. Clever Hans was reading subtle signs from the people around him to figure out when to stop counting.
A while back she started playing with modifier cues with her horses and was reinforcing them for touching a colored cone on cue. They got pretty good at it, but at some point she started to wonder if they understood modifier cues or were just really good at reading her. These are both valid ways for them to solve the problem of which cone to touch, but one is more about the science of it. Do they know the cue for each color? And one was more about the art of it. Do they read her body language to find the right answer?
She did a little test using novel objects and verbal cues and concluded that they are just really good at reading her. From a certain point of view, this might seem like they had not learned what she intended (modifier cues) but what she got out of it was a deeper awareness of how just how well horses can read our body language and how important it is for her to be aware of what she was doing.
This brought her back to her experiences at clinics where she meets a lot of different horses and handlers. The question she posed on her slide was “why are some people like squeaky toys?” It sounds like a silly question but it is really getting at the idea that horses can read us and they recognize that some people can be moved or manipulated and others cannot. What are the horses noticing that creates different types of behavior?
She posed a few questions:
-why are some people squeaky toys?
-why do different people get different results from the same directions?
-how is your balance? Are you aware of your own?
-can you use bone rotations, or do you use muscles?
-is your handling undermining or supporting your long term training goals?
-can you repeat what you are doing? Do you know how you do things?
-is your training program designed to build good habits in the handler as it builds new skills?
In clicker training, we often look at individual behaviors, but Alex is looking at the big picture to see how all the behaviors fit together to create a complete training program. It is important that each step leads to and supports the next step. One way to look at this is to take the questions listed above and think of them as creating a training loop. She calls this the Balance Loop.
The Balance Loop:
Awareness – become aware of how your balance is affecting your learner
Balance – explore your own balance, look for ways to improve it
Support – does your handling support and strengthen your training goals?
Adjust – can you adjust as needed to support the learner?
Repeatability – can you do it again?
Looking ahead – do you know how what you are doing now will fit into the final work?
Awareness and Balance:
As an example of how awareness and balance affect our learners, she had two volunteers come up and led them and the audience through some exercises to show how a person is more stable when they are standing in balance. If you are familiar with the tai chi rope handling, she had us stand with our feet shoulder width apart over our bubbling spring and test to see how well grounded we were. You can do this yourself by pressing on your chest with two fingers to see if you feel solid or are easily rocked back into your heels or even can push yourself backwards.
This is where the “squeaky toy” analogy comes in. Horses quickly recognize when someone is not standing in balance and will bump into or try to move that person. She had us stand so we were rocked back in our heels and we could feel how that made us easier to move when pushed. Then she had us stand in good balance and had us feel how solid we were. Horses pick up on this. A well grounded person working in balance is going to have an easier time working with horses because they will be less fun (and harder) to push around. As an added benefit, if you practice good structure on the ground, it will help with your riding.
Are you aware of the effect you have on the learner. Are you supporting or affecting their movement?
She had a video clip to show how food delivery can affect the balance of the learner. The horse started out standing square and its balance could be changed by feeding so it rocked back a little or feeding so it rocked forward and had to take a step to catch itself.
Can you duplicate what you just did? This is harder than it seems. Most people have limited awareness of what they are doing. They can go through the motion of doing something without paying attention to all the tiny details that make it possible to repeat the same thing. Being aware of those details makes you more consistent.
Is your training program looking ahead?
Do the early exercises/behaviors you teach set the learner up for future success by creating good habits?
She used the exercise “Grown-ups” as an example. Grown-ups is an exercise to teach the horse emotional control (don't mug the handler), but it also has benefits for the handler as they are practicing standing in good balance. Each time the handler delivers food, they get to practice food delivey and returning to a balanced position. In addition, they are also practicing some of the mechanics of Alex's ridden work by putting their hands together at their waist and then separating their hands to feed. Incorporating chances to practice skills that will be needed for a future part of the training program can make a huge difference in how quickly people move through the steps because each new step is not so much "new" as it is a variation on something they have already been doing.
When working with animals that are very sensitive to changes in the handler’s position and technique, taking time to plan ahead will benefit you and your horse.
Kay often encounters people who compliment her on her “drivey” dog or ask her if she can help them make their dog “more drivey.” When asked to define “more drivey,” they are unable to do it in terms of behaviors because what they are really asking for is to have a dog that appears to be enjoying and motivated to do the work. Or they may answer that they want a dog that has energy or a good work ethic. This talk took a look at the concept of “drive” and how it fits into any dog’s training.
Kay showed us a video clip of a german shepherd heeling and said that what she sees in these dogs that people describe as “drivey” is a dog that moves with strength and balance. There is flow and control to the movements and the dog is clearly eager to perform and understands what it is supposed to do. There is no stress. This is very different than a dog that is overexcited and eager to do things, but a bit all over the place both mentally and physically.
Kay’s premise is that most dogs start out as eager learners, but their natural motivation can be lost along the way. It can be lost for many reasons, including:
-insufficient attention to detail
-behavior you want to teach is under-studied due to poor research or planning
-careless, indecisive clicks
-rushed (trainers goes too fast because they are tempted by a glimpse of the final behavior)
-incomplete (behaviors are not robust or strong)
-poor understanding from the learner (shows up as errors under performance stress)
-over-use of directed learning (learning is not internalized or “self-encoded.”
She then quizzed the audience about behaviors they have taught their dog. Hint – if Kay ever asks you if you have taught your dog to sit, down, bark, etc… Say “no.” We don’t teach dogs to do things they already know how to do, we set it up so they do it (or teach themselves to do it). Dogs naturally retrieve and know more about retrieving than we do, so we are not teaching them to retrieve. We are setting it up so they can learn and practice. When we “teach” a dog something it already knows, what we are really doing is setting them up to offer it, putting it on cue, and reinforcing it so that the behavior becomes stronger.
Kay talked a little bit about how motivation can be viewed on a continuum. If the dog is motivated but has not been taught how to channel that energy, you can see stress barking, hyper excitement, and frenetic behaviors. On the other end, if the dog is not motivated, you will see disengagement, dogs that just “go through the motion,” and are easily distracted.
She had a chart that showed arrows pointing to an end goal and how each arrow (arrows representing forward progress) could either lead you to the goal or end up being diverted away from the goal. Some of the things that can make the arrows deviate from a straight path to the goal are:
Uncertainty: if you don’t know what you are clicking, dogs pick up on it and become uncertain
Over-arousal, frenetic behavior: too much practicing kill skills (tug) can make it take longer to get there as the dog is not in the right frame of mind.
The straight path is breaking it down into more behaviors.
She also had a picture of a giant arrow (written as one does with block letters) that was labeled motivation. She showed how we chip away at natural motivation so that there is little left.
This brought her back to the question of if a “drivey” dog (meaning powerfully motivated) is easier to train?
No, it takes longer. With a high-drive dog, we need to take extra care of what they learn and keep emotional stability foremost because these dogs overachieve and make errors. Natural motivation exists at some level in all dogs, it needs to be nurtured with forethought and skilled training. My notes are a bit sketchy here but she talked a bit about this idea that a handler needs to do less to achieve high energy results. This is a traditional idea and doesn’t really fit with her experience that getting a high energy (drive) dog is not a result of the handler doing less, but more of the handler getting the fundamentals right and letting the dog’s skill develop so it can rise to the top. You have to examine the smallest detail and get it right.
Some tips on how to nurture your dog’s drive and motivation:
-The learner should always have the right to say “no” and always have a choice to train
-the dog gets to control the pace of learning, can walk away and have another go when it is ready
-the way the behavior is carried out is more important than the behavior
-the dog should believe they are never, never wrong.
-speed develops when the body is functioning with minimum tension
-beware of global descriptions and falling into the trap of trying to train your animal to “be nice,” “behave well,” “be friendly,””have a better work ethic,” “be more drivey,” or “walk on a loose lead.”
-teach component skills and let the animal bring them together
-we can facilitate understanding
-the outcome may be “well-behaved,” “drivey,” or “on a loose leash,” but we should not be trying to teach these outcomes
-let it grow organically in the dog
The emphasis in most training seminars, classes, and other forms of education is on what to do during a training session. In this talk, Ken looked at what happens (or could happen) during all the other hours of the day when you are with (or not with) your animal and the animal is still learning. As horse people, we are often told that you are training your animal every moment you are with it. So this idea is not new to me, but Ken shared his thoughts on a lot of different aspects of what happens during “non-formal” training sessions.
Formal training is the name he uses for dedicated training sessions where you are working on specific behaviors. Informal or “passive” training is what happens the rest of the time. Every interaction we have with an animal has some kind of reinforcing value. And these informal interactions are important as they are a huge part of building a relationship.
In some ways, his point was really very simple. Don’t take your training hat off and look for ways to reinforce behaviors you like during any interaction with your animal. If you have constant awareness of how your animal is interacting with you, or with its environment, you can take advantage of opportunities to reinforce good behavior.
He showed some video of a beluga whale and her calf. They trainers start handling and working with the calf from very early on, but it is not set up as a formal training session for the calf. Instead they are working with the mother, reinforcing her for allowing the calf to be near (as opposed to just letting the calf swim off and explore), and interacting with the calf in small ways throughout the session. They find that the calf will actually start to copy some of the mother’s behaviors if allowed to be right there. This allows them to start to reinforce behaviors they like from very early on.
Part of what makes this work is the trainer having good observational skills and setting it up so that the calf can and wants to participate. At this point he talked a little bit about the “myth” of a trainer with good instinct. He believes that good trainers are a result of good technique and good observational skills.
You can become a good trainer by taking the time to learn the skills you need and spending time with the animals you train. This allows you to develop a relationship with the animal and a good relationship has a lot of reinforcement built into it. He showed some video of a zoo that has a big cat exhibit where people can interact with the big cats. Yes, they were actually running and playing with them. Seeing a person getting into a “pig pile” with 3 or 4 lions was actually somewhat disconcerting but apparently they are able to do this safely and this kind of play is highly reinforcing. It also shows what is possible when you have built trust into a relationship. Trust takes time to develop but it is a powerful tool.
Play in general is a great way to develop your relationship with your animal. He defined play as interacting with your animal with no food, no clicker, and no plan except to have fun. It’s a chance to find out what your animal really likes and interact with them in a more spontaneous way. He did add the little caution that you should be aware of what you are reinforcing as unwanted behaviors can be reinforced through play. So play, but keep wearing your trainer’s hat.
Play and training sessions are ways to provide enrichment to an animal. Another way to provide enrichment is to add things to the animal’s environment that encourage certain activities. Even though we may not think of this as “training,” the animal is still learning and enrichment activities can be considered another way to train when you’re not training. At the Shedd aquarium, they have different ways to provide enrichment, with varying degrees of supervision.
Looking for ways to provide positive interactions with our animals is such an important part of building a relationship that it should not just be limited to training sessions. The scientist in us pushes away the idea of relationships because they are hard to define, but anyone who has seen someone who has a good relationship with an animal knows that it is very powerful. Relationships are about reinforcement history, built in many different situations, and you can’t deny their effectiveness. At the same time… don’t overstate their power. A good relationship is a powerful tool to enhance your training, but it does not replace good training skills.
Non-formal interactions are often overlooked, but animals are learning 24 hours a day. Use that knowledge to your advantage. Understanding the laws of learning and applying them consistently and constantly will benefit you and your pet.
ORCA: Panel discussion: This had all the speakers plus Joe Layng who was the keynote speaker last year.
I love panel discussions. It is when you get to hear different trainers answer the same question or comment on each other’s responses and it is a chance to see how much good trainers have in common, as well as how much we are influenced by the specifics of what animals we train and what behaviors we train.
The first question that Dr. Rosales-Ruiz posed was about punishment and how to help people recognize all the different types of punishment that can be occurring. The question really addressed the fact that people think they are not punishing behavior because they are not actively applying an aversive, but there are other ways that punishment (a decrease in behavior) can happen.
It was a great discussion and here are some of the responses:
Ken: Look for punishment in the environment to get people to broaden their understanding of punishment.
Steve: People don’t see that what they think is reinforcing is actually aversive. Learn to look at how behavior is changing and see what is happening. On the video subject, Steve added that it is important to look for the good stuff on the video too. If people see how well they are doing, they will feel more confident and less likely to use punishers.
Bob: Use data (how many people guessed he would say that?). Also, use VIDEO. You can punish behavior by having late timing, poor criteria or being distracted. If you are like that with some behaviors, you are probably like that with all behaviors so you don’t have to take and watch a ton of video, just take enough to get some feedback.
Joe: He shared a story about how pigeons will learn to seek getting shocked if it is the only way to get food. So whether or not something is punishment cannot be measured by whether or not we think it is aversive, we have to look at the behavior. At the same time, the lack of punishment does not mean that you are not applying an aversive. Clearly the pigeons would rather not be shocked but they could be conditioned to accept it as part of what happens when they get food.
Also, look at examples and non-examples so you can observe well. I think what he meant was to learn to read your animal well so that you know what it looks like when the animal is being reinforced vs. being punished and learn to see the gradient between them.
Kay: She added on to Joe’s point about how you can condition an aversive so that the animal seeks it. Her example was a dog who is excited about getting a muzzle on because that is the only time the animal gets to go out. The dog appears to want to wear the muzzle, but you need to be careful about making the assumption that it is not aversive in some way.
Alex: One way to help people feel what is aversive is to do training with two people where one is the trainer and one is the trainee. The trainee can give verbal feedback which the trainer can then use to modify their behavior.
Kay: It’s what the animal perceives as punishing. There are lots of subtle punishers. She used the example of feeding the dog without paying attention to them. The dog is reinforced by food and attention so disengaging cheapens the value of the reinforcement and makes the training mechanical which can be enough to decrease behavior.
Phung: He doesn’t like to talk about punishers and would rather spend time showing people what to do. Everyone wants to defend their beliefs about what they do. Better not to go there.
The second question was on control as a reinforcer. Why does it work and how do you give the animal a choice?
Kay: choice has to be handled carefully. If you have too many choices, it can be difficult, but no choice is not good either.
Joe: too much choice is aversive. You have to teach people to handle choice, otherwise it is overwhelming.
The last question was on how to deal with errors.
Alex: Go have a cup of tea (or coffee or ....) Put your animal away and go rethink your plan.
General comments (I'm not sure who said what) included:
If the animal makes a mistake, laugh and try again. Being “wrong” is not bad or a cause for fear. The animal is not trying to do something wrong.
Joe: We talk about trial and error learning, but as B.F. Skinner says “error” is not a behavior. It is not helpful to think of errors, instead look at what the animal actually did and come up with a plan to replace it with what you want.
Question 1 generated a lot of discussion and most of the panel discussion was spent on it.
Question 2 didn’t seem to generate as much discussion, I think because they all seemed to agree with the first few comments. I was kind of surprised that the idea of choice and control as reinforcers didn’t generate much discussion and that the comments I wrote are all about how too much choice can be difficult. But I think that all of us who clicker train already recognize that choice and control are important so perhaps there was underlying agreement and the only comments were to qualify that so people didn’t think that if a little choice was good, more choice was better.
Question 3 also didn’t generate as much discussion. I think part of this was because they had already discussed punishment which is one way that some trainers handle errors.
The ORCA conference was great. The energy and commitment these speakers have to promoting training using positive reinforcement and educating others is inspiring. I always enjoy the fact that the speakers come from different backgrounds and bring different ideas about training to the mix. If you are considering attending a future conference, or just want to read more about ORCA, you might want to visit their website which is http://orgs.unt.edu/orca/.
Katherine Bartlett, copyright 2013. Please do not share any part of this document without my written permission.