by Mark Plonsky, Ph.D.
Copyright © 2014
Also available in Finnish (thanks to F. Brenk)
I am an experimental psychologist (professor at a university) with expertise in animal learning & behavior. I am also a dog trainer. People often ask me, “How do you train a dog?” I prefer a scientific approach, since that is how I was trained. In this article, I will describe six concepts that I find useful in training my own dogs as well as helping other people train theirs.
The criterion of a behavior is a clear definition of it (that people can easily agree on). It makes clear exactly what the dog is expected to do. In scientific terms, the criterion is known as an “Operational Definition” or concept defined by how it is measured.
Let us use “the Recall” as an example. When a person calls their dog, they want it to come to them. But is moving towards you good enough? Is sitting near you? How close? In front or at the side? Since I believe it is good to keep it simple, my criterion for the recall is being able to touch the dog’s collar. In other words, my dog has performed the recall when it comes to me and is close enough so that I am able to touch its collar. It is at that point, that I praise the dog.
Another example of the criterion is using the words “Down” (elbows and butt on floor) versus “Off” (4 on floor) with a dog. Each concept is different. Many folks treat this as one cue/command which makes it more difficult for the dog since what it should do is not being made clear. In summary, having clear and consistent criteria for the behaviors that we want the dog to perform makes the training process easier.
Luring is a technique for getting the dog to do what you want without physical assistance. A worthy example is using a piece of food hidden between the fingers of your hand (kept a couple of inches in front of a dogs nose) to pull the dog like a magnet into the position you want it to assume. The dog must be interested (hungry). Using this technique, the dog will typically sit when the food lure is placed above and slightly back over its head. When the dog’s butt hits the ground (the criterion for sitting), one says the word “good” and then releases the food into the dog's mouth. Various positions, such as “the Down”, “the Sit”, “the Stand”, and more active behaviors such as “Rolling Over” and “Heeling” can also be obtained by using the food as a magnet to guide the dog's nose (and thus, body) into the position you want. In still more advanced forms of luring, the lure is moved away from the handler and dog and is then called a “Target”.
From a scientific perspective, luring would be referred to as “Sign Tracking”, “Classical Conditioning”, or “Pavlovian conditioning”. Yes, this concept has to do with Pavlov’s drooling dogs. It is a type of learning where the animal learns what predicts what.
Rewarding is a technique where a pleasant consequence (something good) occurs when the dog performs a behavior. When the dog’s behavior is immediately followed by something pleasant, it will be more likely to perform that behavior again in the future. After spending some time luring the dog to get it to do what you want, the dog will typically begin to anticipate and perform the behavior without the lure. At this point, one should say the word “good” (or use a clicker) and then reward the dog. Over time, the lure can be faded away and the dog will begin to do what you ask without the treat magnet. It should be promptly rewarded for this. Food makes a good reward initially. Ultimately, play and praise can and should be used as well. In fact, whatever the dog enjoys makes a good reward. In other words, pleasant events should be the consequence of appropriate behavior.
The other side of pleasant events is unpleasant or aversive events. Folks speak of punishment to decrease behaviors and punishment typically employs aversive events. Punishment is thus more controversial than reward. Punishment can lead to undesirable behaviors (like hand shyness and other fear responses). I would argue that punishment is more difficult to use appropriately and effectively than reward and is less forgiving of mistakes in training.
From a scientific perspective, reward and punishment deal with “Operant Conditioning” (also called “Skinnerian Conditioning” or “Trial and Error Learning”). This type of learning is concerned with how the animal gets what it wants or avoids/escapes what it does not want.
Shaping is a technique for teaching an animal a behavior by rewarding successive approximations to it. It involves starting with a facsimile of the behavior desired and gradually incrementing the criterion until the desired behavior is obtained.
A simple example of shaping is “the Stay”. Initially, having the dog hold the Stay for a few seconds is good. Over time, we gradually increase the duration of the stay required (shape the behavior) until achieving several minutes. A more complex example of shaping is having a dog turn on a light switch. Typically, this kind of task would be broken down into the various behaviors that make it up and each would be taught separately. Ultimately these behaviors would be “chained” together to form the whole behavior sequence.
Prompting is a technique which involves providing the minimal amount of physical assistance required to get the dog to perform the desired behavior. The word “minimal” is important, otherwise, the prompt is likely to be aversive (in which case we are then dealing with punishment).
Prompting is similar to training wheels on a bicycle. The goal of prompting is to help the dog perform the behavior so that it can be rewarded for it. For example, sometimes luring a dog into a down with a piece of food placed near its nose and then slowly moving it between its front two legs toward the ground is not enough to get the behavior. The dog will put its elbows on the ground, but not its butt. In this case, gently placing a finger between its shoulder blades will often get the dog to drop its butt. A more complex example would be using a wall or fence to help teach a dog to back up when at your side. In this case, the fence serves as the prompt rather than you having to physically manipulate the dog.
Honoring is a set of skills on the part of the dog that involves self-restraint or self-control. For example, in food honoring, the dog has to resign itself to not getting the food (at least for a brief period) in order to get it. In the initial training of food honoring (the “Leave it”), the dog should quickly be rewarded for any behavior incompatible with grabbing at the food. When I train this, the reward involves releasing the dog (with a word like “OK”), so that it can eat the food it wanted but restrained itself from eating. The release word gives the dog permission.
Typically, one starts out by teaching the dog to honor food and then toys follow. Ultimately, other dogs, people, and other distractions should be honored as well. In more advanced training, the dog may be required to maintain eye contact with handler in order to be released.
Each of these six concepts could be discussed in a lot more detail and there are others that might be worthy of inclusion. However, I think this brief overview provides a good foundation for a gentle-handed training philosophy. I believe training should be fun for both the dog and handler. In this way, training is likely to continue and lead to a strong and enjoyable bond. In addition, the dog will become well behaved.