Clicker Training: What’s It All About?
By Gary Wilkes
From a dog’s point of view, obeying a command boils down to two simple problems; “What do you want me to do?” and “What do I get, if I do it?” To put this in human terms, the dog needs information about what to do and a motivation why he should do it.
Before you can teach your dog anything, you have to satisfy the “why” part of the equation. Learning can be trying, tedious and difficult. Without sufficient motivation, the dog is likely to get bored and trot off in search of a wayward cat. When you discover the things that will motivate your dog to work, you can start connecting them to “what” you want the dog to do.
In the early stages of learning a behavior, a dog needs clear information signals that mark which part of the behavior “caused” success or failure. Each successful repetition links a little more information to the correct behavior—like placing another piece in a picture puzzle. If the information is sloppy or ambiguous, it will take more repetitions for the dog to sort out the “Big Picture.” If your information signal is slow or poorly timed, the wrong information is attached to the behavior—like trying to insert a piece from a different puzzle. With an effective information signal, the percentage of correctly applied pieces increases, and the puzzle is solved faster.
Praise as an Information Signal
Historically, the information signal of choice for dog trainers has been verbal praise. While it is natural for humans to use words to identify good behavior, dogs rely more on the timing of the signal than the conceptual meaning behind the words. Since the dog doesn’t really understand English, the signal you choose to mean “good boy” is arbitrary. The results you get, however, are dictated by the dog’s ability to tie a given signal to a behavior.
To get an idea of how this works, call your dog and start a little game of fetch. Pick an arbitrary point about half way across the room, and put a hand towel on the spot. Now, toss the ball so that it will go beyond, but directly over, the towel. As your dog runs toward the ball, he will necessarily pass over the towel. As your dog’s nose passes over the towel, say “good boy”. Don’t worry if your dog doesn’t respond to the words, yet. Just pay close attention to exactly how far the dog travels between the time you start to say “good” and the time you finish “boy”.
Now let’s try it again. This time, instead of using praise, clap your hands together just once to mark the behavior. If you watch closely, you may see the biggest advantage to an instantaneous clap—and the biggest downside to using praise. When you used praise to mark the behavior, the dog traveled a full body length while you said “good boy”. By contrast, the clap marked the precise instant the dog’s nose crossed the towel. When you are using verbal praise, you can only approximate the instant the dog performs the correct part of the behavior.
So, in our rug exercise, praise will allow you to tell the dog that there is a really important patch of rug in your house, about two to three feet in diameter. The same dog, sensing an instantaneous signal, can identify that there was something important about a patch of rug about six inches in diameter. Within a few repetitions, the individual claps could identify a particular place and a particular posture, while the praise would still be working on the location. The point is that a faster information signal can enhance your dog’s speed in seeing the “big picture.”
While it is obvious that an information signal must be quick, its speed is not the only criterion you must consider. Even clicker trainers occasionally forget that no single signal can work in every situation. Rather than being written in stone, the list of acceptable information signals is dependent on the physical abilities of the animal, the training task and environment.
For instance, clickers and praise won’t work if the dog can’t hear them, which makes a whistle a better tool for distance work. Likewise, hand signals are pretty useless if the dog is visually impaired or the behavior requires the dog to be looking away from the handler. Rather than thinking of specific signals, it is more useful to look at the criteria that dictate which signal works best at any given time. Here are four rules that can help you evaluate an “INFO” signal.
I–Immediate: You must be able to present the signal in connection with the correct behavior. It is the timing of the signal that tells the animal WHICH behavior to remember.
N–Noticeable: Whether you use a sight, sound or smell as your information signal, it must be easily recognized across the background sights, sounds and smells of the environment. If you are back lit, hand signals in front of your body will disappear inside your silhouette. If you are training in a room full of other trainers, you will have to modify the way you talk in order for your dog to distinguish your voice apart from all the others.
F–Fast: In order to mark a behavior that may last a mere 1/10th of a second, you need a signal that can be delivered in 1/10th of a second.
O–One of a kind: Unless your Info Signal is unique and consistent, you will force the animal to focus on the signal instead of focusing on the behavior. The idea is to prevent the dog from having to think about anything other than the information about the behavior.
Once you get used to thinking in terms of these criteria, you can start evaluating the signals you use to send information to your dog. For instance, for almost all serious training I use a clicker. It fits the INFO rules better than anything else I have found. That doesn’t mean I always use a clicker. When training at a distance, the clicker isn’t noticeable, so I use a whistle.
If I don’t happen to have a clicker, I simply say “good”. Though the verbal info signal isn’t as fast as a clicker, it’s still immediate and one of a kind. When training deaf animals, I use a fast hand signal as a “clicker”, the wink of a flashlight or the vibration feature of remote collar. Each of these decisions is based on the requirements dictated by the dog and the setting, rather than by a devotion to one signal over another.
The best place to start with clicker training is to realize that a clicker is merely a very precise, but limited construction tool. Clicker training isn’t magic, even if the results sometimes appear to be. Learning to use a clicker doesn’t mean you have to give up your knowledge of “what works” and “what doesn’t”, it merely uses your “dog smarts” in a slightly different way. The essence of clicker training is letting the dog’s behavior dictate your methods, rather than using the wrong tool at the wrong time. If you are willing to take a shot at it, hopefully the results will motivate you to continue using it. ✂