Before starting this article, I polled the ClickerSolutions mailing list about the training myths—about both clicker and more traditional training—the members had heard. The responses poured in. It became obvious that misunderstandings, miscommunications, and half-truths abound, creating unnecessary walls between trainers. Let's debunk some of these myths.
"Traditional training doesn't work."
AND
"Clicker training doesn't work."
Of course they work. Thousands of pet, competition, and working dogs out there prove the efficacy of both types of training. As long as a specific method follows the principles of learning for an individual dog, that method will work.
"No one method works for all dogs."
True—at least as written. A method is a recipe, step-by-step instructions for training or modifying a behavior. No one recipe works for all dogs, and this is why so many alternative methods—many of them just slight alterations of their predecessors—have been developed. The history of training can, in fact, be traced trainer to trainer through methods.
Unfortunately, methods are limiting unless the trainer also understands the underlying principles. In the past, trainers taught methods rather than principles, and as a result, individual dogs (and sometimes entire breeds) who didn't respond to the methods taught were incorrectly labeled too "stupid," "stubborn," "dominant," "soft," "driven," "aggressive," or "easily bored" to train.
Clicker training, though it contains recipes, is not a method. It is a technology for changing behavior. The principles of learning are heavily emphasized. For the first time, the ability to evaluate a method for an individual dog, to modify it, or even to create an entirely new recipe is taken out of the indefinable realms of "instinct," "experience," and "talent" and taught and explained to everyone—from beginning pet owners to professional trainers.
Clicker training, applied skillfully, will work for every dog mentally and physically capable of learning, because no dog is immune to the principles of learning.
"Traditional training is cruel and inhumane."
AND
"Traditionally trained dogs are fearful/aggressive/miserable."
AND
"Dogs who behave fearfully must have been traditionally trained and abused."
These myths are frequently spouted—and they are both unfair and incorrect. Are there people who use aversives inhumanely in the name of training? Of course. Are there traditionally trained dogs who fail in traditional programs and become fearful, aggressive, or miserable? Unfortunately, yes. However, many remain happy workers, and a seemingly miserable dog is not necessarily a sign of abusive training.
"Because all training can be explained using operant conditioning terms, all trainers are operant trainers."
Children can make some darn cool chemical reactions with a chemistry set, but that doesn't make them chemists. My husband built a functional table—didn't make him a carpenter. I can follow a recipe in a cookbook and can occasionally even throw together ingredients for my own unique dish—doesn't make me a chef. Being a chef, a carpenter, or a chemist requires an understanding that goes far beyond the level of following step-by-step instructions or doing simple experimentation. The same is true of operant training.
Just because a training method can be explained using operant conditioning terms, that doesn't mean the trainer applying the method is an operant trainer. To be an operant trainer, the trainer must understand those principles inside and out and be able to apply them and modify them in any situation to any individual. "Operant trainers" are not limited to clicker training methods (nor is every clicker trainer an operant trainer), but because clicker training so heavily emphasizes the principles of learning from the earliest stages of learning, more operant trainers are clicker trainers.
"Clicker training isn't effective because no clicker trained dogs have achieved {fill in your favorite elite title}."
Clicker training is not a new technology, but it is relatively new in the dog world. Due to grassroots effort, it has slowly increased in popularity over the past fifteen or twenty years. However, only in the past two or three years have clicker training classes begun to appear with any regularity. Prior to that, trainers were largely on their own, teaching themselves with the help of mailing lists, Web sites, and a limited selection of books and videos. Even now most classes are geared toward the pet owner.
A potential competitor who wants to traditionally train his dog has lots of resources at his disposal because many, many people have been down the road before him. There are many existing methods, many experienced competitors to help, many instructors to teach, and many books and videos to supplement.
People who want to clicker train for competition aren't that lucky. Some sports—like agility and canine freestyle—are dominated by clicker trainers and have accumulated a wealth of resources, but other sports—like field training—have nothing to help the new competitor. The old traditional recipes usually don't translate; the trainer—often a beginner himself—must start from scratch.
In order to get a title (in any sport!), you need...
- A trainer who thoroughly understands the training style he is using
- A trainer who thoroughly understands the sport he is participating in and the individual behaviors he needs to train
- A dog who has the talent and physical ability to do the required behaviors at a precise-enough level to consistently win
- The desire to train and compete enough to obtain the title
- The money to train and compete enough to obtain the title
- The time to train and compete enough to obtain the title
- The skill to train to obtain the title
You have to have every single one of those elements. Every one. The reality is it takes years to become good enough to train to the upper levels of any sport, even if you have the resources to help you get there. For those trainers who are pursuing sports where no one has yet forged a path and invented recipes, the road is infinitely tougher and harder.
It will happen. Every year another boundary comes crashing down. All we need is the time to have all of the element fall into place. [Editor's note: Since this has was first published, clicker trained dogs have won all sorts of titles and continue to do so. Clickertraining.com maintains an honor roll where a partial database of honored dogs is kept.]
"Clicker trainers use no punishment."
Incorrect. Clicker trainers use negative punishment, which is the removal of something the dog wants. For example, ""penalty yards" (TM pending, Lana Horton)" is a common method used in teaching loose leash walking. The dog sees something it wants. As long as the dog walks nicely, the trainer lets it walk toward what it wants. However, if the dog pulls, the trainer walks the dog backwards. Walk nicely; get what you want—positive reinforcement. Pull; lose what you want—negative punishment. This method is extremely clear to the dog, because getting or losing what it wants is controlled by the dog's actions.
"Adding an aversive (positive punishment) is more severe, but more effective, than removing a reinforcer (negative punishment)."
Is positive punishment more severe than negative punishment? Is it more effective? What about positive reinforcement? Is it more or less powerful than negative reinforcement? Is it more or less effective?
Every application of reinforcement and punishment, positive and negative, falls on a continuum from mild to extreme. Exactly where the particular application falls on the continuum depends on the individual dog and the specific situation.
Similarly, punishment and reinforcement are defined by their results. So by definition, they work and are effective. Applications that involve positive punishment even if used correctly, aversives may have side-effects. Both take skill to apply correctly, but the potential for negative impact to a training program and the dog is higher if the trainer misapplies an aversive.
"Corrections are vital to ensure a dog knows it must obey."
False. Reliability is a number. Data. Cold data with no relation to "choice" or "control"—or to the method used to get the result. Reliability is obtained through reinforced repetition. True reliability is achieved at fluency, long after the animal is past the point of performing solely because of consequence, positive or negative.
I've been driving cars with manual transmissions for over ten years. In that time, I've had an incredible number of reinforced repetitions (meaning the car did what I wanted it to do) for using the clutch.
I remember learning to use the clutch. It was horrible, because I'm not terribly coordinated. Shifting once the car was moving wasn't too hard, but getting started from a stop was murder. So we did lots of reps. We started in a parking lot—no distractions, few restrictions. When my reliability improved, we raised the criteria and went to a neighborhood street. I was back at square one in that environment. But through practice, I improved. Then we went to more populated streets. Whoa—drop in performance again! But again, I quickly improved. Gradually the streets got busier and harder. We added hills. We added the pressure of cars behind me at a stop light. Man, I had thousands of repetitions before I got relatively smooth at getting that car started.
For a very long time, using the clutch to start the car was a deliberate conscious behavior. I had to think about it every time. Over time that changed, and I don't think about it anymore. I don't have to. I'm fluent in the behavior. Latency is immediate. Reliability is near 100%.
Near. Once in a blue moon, I stall the car. It happens. I ain't perfect, even after more than ten years of repetitions.
And still the behavior is under my control. I can choose not to use the clutch anytime I want to. I can pop the clutch intentionally. Never, never, never will that behavior be out of my control.
To get a truly reliable behavior, there's only one way to do it. Practice with intent. Generalize the behavior. Practice in the conditions in which you need the behavior reliable. Work on latency. Keep records and train until you've achieved the level of reliability you need, whether it's nine of ten or 999 of 1000.
You determine which cues are the most reliable and have immediate responses by training them that way. But never fool yourself into thinking training, no matter how severe the aversive used, overcomes free will.
"The click must be followed by a food treat."
False. The click must be followed by a reinforcer—something the dog is willing to work to obtain. You have a variety of reinforcers available. Some of the most commonly used include:
- Food
- Toys
- Praise, attention
- Opportunity to do something the dog wants to do
- Opportunity to perform a well-known behavior
In a formal training session you want to get as many repetitions as possible. Food is an excellent reinforcer because it can be cut into tiny pieces and eaten quickly. Toys are also a good reinforcer, but playing with the toy takes time, meaning you get fewer repetitions in a session. Praise and attention are wonderful additions to food or toys, but are often not desired enough by the dog to use alone, particularly in distracting situations. The opportunity to do something else is sometimes the most powerful reinforcer you have.
The most important thing to remember is that the dog determines what is and isn't a reinforcer in a certain situation. If the dog doesn't want what you're offering, it's not a reinforcer.
"Clicker training won't always work because food isn't a strong enough reinforcer."
AND
"Instinctive drives and self-rewarding behaviors are so powerful that you must use corrections to ensure reliability."
As mentioned in the last section, food isn't the only reinforcer available. No matter what reinforcer you choose, consider its relative value. One food might be worth more than another food. Your dog's tug toy might be worth more to your dog than food in certain situations. The opportunity to greet another dog may be the best of all! It all depends on your dog and the particular situation.
One of the commonly sited concerns is that there are things the dog wants in the environment that he can't have. Or that the dog has a powerful, instinctive drive to perform a certain behavior, and the trainer can't find a positive reinforcer more powerful than that drive. Guess what—you may not! So that's when clicker trainers move beyond operant conditioning and employ techniques based in classical conditioning, such as desensitization and redirection. Desensitization lowers the strength of the animal's response to particular stimuli. Redirection can be used to actually transfer the focus of the drive from an undesired outlet to a different—even unrelated—desired one.
Imagine… a Border Collie staying on task next to a soccer field filled with screaming, running children… a sporting breed remaining focused in the presence of joggers, squirrels, birds, people throwing balls, and other dogs cavorting… a high-drive Malinois breaking off an attack on a single command without getting that oh-so-desired bite. That's how powerful desensitization and redirection are. I'll give real-life examples and outline specific steps to achieve this kind of control without corrections in a later article in this series.
"Clicker trained dogs will work only when food or the clicker is present."
AND
"You have to use the clicker and food treats forever."
AND
"You have to carry the clicker and treats everywhere you go."
People are terrified of being tied to the clicker and food treats. Fortunately, they need not fear.
The clicker is an event marker, used to identify correct performance during the early learning stage of training a behavior. Once the behavior is fully shaped, on cue, and strong, you don't need its precision anymore. So you can simply replace it with a verbal marker/release word. If you don't have a clicker with you, you can always mark verbally—or even just deliver the reinforcer directly.
Food, as explained in earlier sections, is not the only reinforcer you have available. Yes, you do need to continue reinforcing behavior—at least occasionally—but that reinforcement doesn't have to be with a food treat or even with anything you "give" the dog. Instead the reinforcement might be the opportunity to do something it really likes or, if you're lucky, the behavior itself might become self-reinforcing!
Food can be misused, of course. Some people complain that their dog will do anything as long as food is present. That is an example of how wonderful dogs are at discriminating. If food is visible every time you train—whether used as a lure or in a treat bag around your waist—the dog will quickly come to believe that food is part of the equation. This problem isn't limited to problems fading food. Dogs are frequently trained to perform behaviors only when they are directly in front of their trainer, when their trainer is standing, or when in a specific location. All of these problems can be avoided by making sure that you vary during training everything that isn't tied directly to the behavior.
"Clicker training is a snap to learn."
AND
"Clicker training is too difficult for beginners."
Unfortunately and fortunately, neither of these is true. Clicker technique is simple, but it isn't easy. Even if you have a good recipe to follow, it takes a certain amount of skill to clicker train correctly. Fortunately, however, dogs are an extremely forgiving species who work very hard to figure out what their trainer wants. Although "timing" is often touted as the be-all, end-all for clicker training, the reality is, all pet behaviors (and a great number of competition behaviors) can be taught by someone with utterly abysmal timing.
Other people are overwhelmed by the amount of theory in clicker training. They just want to follow a recipe and train their dog. That's okay too! Many people are finding and following recipes with little or no understanding of how they work. They aren't operant trainers, but they're still achieving their goals.
"The clicker is a magic, necessary tool for training."
I'm amazed at the number of people who point the clicker at their dog, click it, and, when nothing happens, complain that clicker training doesn't work. The clicker is not a remote control.
The clicker is a tin noisemaker. When you first use it, it's completely neutral. However, by associating it with food or other reinforcer, it takes on reinforcing qualities. Its power, however, is not as a reinforcer but as an event marker. As an event marker, it is (in skilled hands) a scalpel, capable of shaping incredibly precise behaviors. A verbal marker is, by comparison, a butter knife. What do you need? If the butter knife is adequate, sure, you can train without a clicker. But if you need a scalpel, the clicker—or other similar, species-appropriate marker—is a superior way of getting precise behavior without the risks associated with aversives.
"The clicker cues behavior."
AND
"Clicker trainers go around with a million uncued behaviors."
AND
"Clicker trained dogs constantly throw behaviors."
A cue "names" and elicits a behavior. The clicker marks that behavior when it occurs. For two reasons, clicker trainers don't add a cue until the dog is offering exactly what we want.
- When the pup is learning the behavior, we want him to concentrate on the behavior. At that point, the cue is meaningless to him anyway—just another bit of "noise" to sort through. In the beginning, make learning easier on your dog by minimizing distractions, including meaningless cue words.
- We want the cue to be associated with the final, perfect form of the behavior. If you add the cue in the beginning, you run the risk of having the unfinished version of the behavior crop up when you least want it to—like during the stress of competition—even though you continued to shape a more precise behavior.
First get the behavior you want in the form you want it. Then add the cue as soon as the dog is actively offering the behavior you want. For a simple behavior that could happen the first day!
More complex behaviors may take more time to shape. If the behavior is extremely complex—a behavior chain, for example—you can add cues to the individual parts of the chain, and then add a cue for the entire chain when it's complete. Or, if the behavior is a single but very elaborate behavior, you can use temporary cues as you shape the behavior, replacing them with a permanent cue when you've shaped the final behavior.
"You can't praise your dog when you clicker train."
Of course you can praise! After you click, praise all you want. I do encourage trainers to be quiet before the click so the dog can think about what will earn him that click. Once he has earned it, however, celebrate all you want. Let training improve relationship—love your dog!
Conclusion
I love clicker training, and I'd love to teach you about it, but not by using half-truths or attacks on a perfectly valid way of training. All trainers, no matter what kind of training they practice, have the same goal: to make life better for dogs and their humans. Learning new techniques is about solutions, not about condemning old techniques. Facts. Education. Understanding.
Editor's note: The article above was originally written for The Commentator, the newsletter of the Curly Coated Retriever Club of America.
Great clarification!
Wow, this really helped me clarify some of the finer points of clicker training. The advice in this article is just great, and it really helped provide some good structure and a better understanding of clicker training for a newbie like me!
As I continue to read the articles and blogs on the site, I find myself getting more and more enamored with the entire clicker training concept and my goals for both myself and my new four-legged furry friend taking on more breadth and definition!
Post new comment