B. F. Skinner elaborated the principles of learning many years ago. Simply put, rewarded behaviors increase in frequency, and punished behaviors decrease in frequency. Rewards (also called "reinforcement") and punishments are designated as "positive" or "negative" depending on whether something is added or subtracted. For example, treats are positive reinforcement, and letting a working horse stop and rest is negative reinforcement (subtracting the requirement to work). Historically, horses have mostly been trained with negative reinforcement.
The natural human reaction to problem behaviors is to punish them, usually by physical punishment such as hitting the horse (positive punishment). Negative punishment is more often used with kids than with horses (for example, taking away the car keys). However, people tend to do a bad job of punishment, so they often fail to succeed in stopping problem behaviors.
It's actually fairly difficult to perform punishment properly. Ideally, the punishment must be administered immediately after the problem behavior, every time the behavior occurs, and at exactly the right intensity. Unwanted side effects of punishment include making the animal fearful and/or aggressive. Punishment also creates a sort of vacuum --- it tells the animal what not to do, but it doesn't tell the animal what it should do instead of the problem behavior. Negative punishment is preferable to positive punishment, but no punishment is best of all.
So, if you shouldn't punish problem behaviors, how do you stop them? Think again about the principles of learning. Most problem behaviors are learned, which means they occur because they've been rewarded. The best way to stop them from continuing to occur is to stop them from being rewarded (a process called "extinction"). Be aware that, at the beginning of extinction, the problem behavior will actually increase, as the animal first tries harder to get the reward. Only when the animal finally realizes that it will no longer be rewarded will the behavior actually stop.
Many people, at this point, are probably thinking I'm crazy. They are convinced that they are punishing their horses' problem behaviors, not rewarding them. However, it is the HORSE that determines what is a reward or a punishment, not the handler. Many times, what we think is rewarding or punishing may not be seen that way by the horse.
The example I'll be using during this blog post is pawing, which is a problem behavior in many horses. So, let me start with using it to explain rewards vs. punishments. One reason why horses paw is because they're lonely and/or bored. Someone ties a horse up, then walks away and leaves it. The horse begins pawing, and the person yells at it. The person thinks that yelling at the horse is punishment. The horse sees it as a reward. It's gotten a reaction to its pawing, so (for at least a moment), it's no longer lonely or bored. When the reaction goes away, the horse paws again and gets another reaction. So, the pawing actually increases in frequency because it's being rewarded, not punished.
The first thing to do, then, in trying to stop a problem behavior is to identify what's rewarding it. The bare bones of a behavior analysis is identifying the behavior (something observable, like pawing) and what immediately precedes it (the antecedents) and follows it (the consequences). The antecedents (tying the horse up and leaving it alone) are usually what causes the behavior, and the consequences (yelling at it) are usually what rewards (or punishes) it. Identifying the antecedents and consequences should help you better understand why the behavior occurs. Tying the horse up and leaving it alone causes loneliness, boredom, and frustration. Frustration is what causes the horse to paw. Yelling at it does not decrease the frequency of pawing, so it is not seen by the horse as punishment. If the horse is lonely and bored, it may see being yelled at as rewarding, so the first step in decreasing the pawing is to stop yelling at the horse (remove the reward).
However, in looking at the antecedents, you've also identified the problem that tying the horse up and leaving it alone causes loneliness, boredom, and frustration. Pawing helps to alleviate the boredom and frustration, so pawing is what is called "self-rewarding" behavior. In other words, the pawing is still being rewarded even if you stop yelling at the horse. How do you address that snag?
As I noted above, one shortcoming of punishment is that it tells the animal what not to do, but it doesn't tell the animal what it should do instead of the problem behavior. The best way to get rid of a problem behavior is to choose an alternate behavior and reward that one instead of the unwanted behavior. If the problem behavior is self-rewarding, then the reward for the alternate behavior must be better than the reward for the problem behavior. The best alternate behavior is one that the animal can't perform at the same time as the problem behavior. (For example, a good alternate behavior for a dog that jumps on people would be sitting. The dog would be ignored if it ran up and jumped on someone but rewarded if it ran up and sat down in front of the person.)
An obvious alternate behavior for the pawing horse would be for it to stand quietly until the handler returns. Therefore, the plan would be to ignore the horse when it's pawing and reward it when it stands quietly. To succeed, you need to be able to reward the horse immediately when it does what you want (stands quietly).
Rewards come in two flavors. A primary reinforcer is something that is a natural reward for the animal, like a treat. A secondary reinforcer is something that is rewarding because it signals that a natural reward is coming. "Good boy!" is a secondary reinforcer if it is always followed by a treat (or some other natural reward). If it is only sometimes followed by a treat, it may or may not function as a secondary reinforcer. Therefore, if you are likely to say "good boy!" sometimes and NOT follow it with a treat, you might want to pick something else (like a clicker sound) to be your secondary reinforcer. The way to make something a secondary reinforcer is just to have a session or two where you make the sound and follow it up with the treat. When the animal starts looking for the treat as soon as it hears the sound, you know that the sound has become a secondary reinforcer.
To be able to reward a horse immediately for standing quietly while tied alone, you need a secondary reinforcer. Lets say youre using "good boy!" So, you tie the horse and walk off, and if it stands quietly, you say "good boy!", return and give it a treat. Ideally, it will stand quietly for at least a short time, so you can avoid the pawing altogether by rewarding the horse soon after you've left it, for standing quietly for a short period of time. Then, you just gradually increase the length of time it must stand quietly before getting rewarded. However, if it starts pawing immediately after you leave, you simply wait for a pause in the pawing, then say "good boy!" and reward it. The timing of the "good boy!" is critical. It must be clear to the horse that the reward is for standing still, NOT for pawing. Once the horse learns it will be rewarded for standing quietly, it will look forward to the reward, which will help decrease its loneliness, boredom, and frustration. At that point, the pawing should cease to be a problem.
Punishment tells an animal what NOT to do, which is exclusively negative and leaves the animal in a vacuum, wondering what it SHOULD do. Asking "how do I stop a behavior?" is equally negative, and it leads naturally to thinking about punishment. Rather than thinking about what you DON'T want, think about what you DO want, then reward the horse for doing what you want it to do. Thinking positively is always better than thinking negatively.
In other words, the answer to the question "how do I stop a problem behavior?" is to ask two better questions: "what is rewarding this behavior?" (so you can remove the reward) and "what behavior should I reward instead of rewarding this behavior?" You and your horse will both be happier if you focus on rewards instead of punishments. Try it!