Why Beat The Dog?
Dogs act like dogs. However, rather than teaching them how to appropriately act like dogs when living with people, many owners try to eliminate the dog's natural tendencies altogether by punishing the poor dog every time he acts like a dog. However, since the dog's behaviors are both innate and habitual, the punishments are often ineffective. Unpleasant? Often! But ineffective nonetheless.
Punishments generally fail to work when they are delayed or are non-instructive. And when punishments don’t work, rather than assessing the overall training program, far too many people become frustrated and respond with a knee-jerk increase in the frequency, duration and severity of “punishment”. We are going downhill quickly. Even though we can prove that in most cases the dog simply doesn’t understand what’s going on and he hasn’t got a clue why he is being “punished”, the rejoinder is always the same — “But he knows it’s wrong!” “He’s doing it on purpose!”
Well here’s the deal. When punishments are aversive (and they don’t have to be) and if they are correctly administered, they work within just one or two trials. Insufficient nastiness is not the most common reason for punishments not to work. Heaven forbid… it’s pretty awful what some people do to dogs, horses and humans under the guise of training. Rather, punishments fail to work because their administration did not satisfy the criteria listed below.
By definition, a punishment simply has to be effective. There is actually nothing in the learning theory literature that states that punishments have to be unpleasant. For most people though, the words “punishment” and “aversive” are synonymous. This is simply not true. Punishment need not be aversive. Punishment need not, and therefore, should not, be painful or scary.
I praise and reward my dogs lots and I reprimand and punish them less. I only use my voice when punishing (with instructive reprimands) and I seldom raise my voice when doing so. It works. And please remember, adequately training and motivating your dog how you would like him to act is by far the best way to minimize the use of even non-aversive punishments for undesirable behavior. If you have taught him what you want him to do and you have motivated him to want to do it your way, then by and large, punishment is unnecessary.
Punishment or Abuse?
Behavior is modified by its immediate outcome. Behaviors will tend to decrease in frequency if always followed by punishment. Often however, undesirable behavior is unaffected by punishment because inappropriate administration renders punishment ineffective. And sometimes, undesirable behavior (and hence the ensuing punishment) actually increases in frequency, usually because the so-called “treatment” (punishment) is the underlying cause of the behavior problem, as with poor heeling, slow recalls, licking, pawing, jumping-up, growling and biting.
To effectively inhibit a specific behavior, punishment must meet a number of criteria. The first criterion is tautological; a punishment must be punishing, i.e., punishment must reduce the immediately preceding behavior. Consequently, repeated and escalated punishment is the best evidence that punishment is not working. Obviously, if punishment were effective, the dog would no longer misbehave and therefore, would no longer be punished. Now, if punishment is not effective, by definition it cannot be termed a punishment. So what is it? Well, depending on the severity, it’s either harassment or abuse.
In order to be effective, punishment should be immediate, of short duration, not overbearing, preceded by a warning and above all, instructive.
The biggest problem is faulty timing. Punishment long after the fact simply doesn’t work. A punishment must follow within one or two seconds to effectively inhibit the immediately preceding behavior. Ideally, both crime and punishment should occur in less than a second. If delayed punishments were effective, obviously the problem would no longer exist and there would be no need to continue punishing the dog.
The instructive aspects of a punishment are the onset and the offset. Commencing punishment informs the dog that he is making a mistake and terminating punishment informs the dog that he is back on track. Thus, short and sharp punishments are most effective because they instructively pinpoint the unwanted behavior, wherein the dog starts to misbehave, he is punished and he stops misbehaving, all in less than second. Continued misbehavior, however, often requires continued punishment. A protracted punishment actually comprises two separate psychological processes — positive punishment and negative reinforcement. These terms are often confused and used synonymously whereas in fact, they are quite the opposite of each other. With negative reinforcement, the immediate termination (negation) of punishment reinforces the preceding desired response, i.e., ceasing misbehaving, and/or starting to behave appropriately (depending on which way you look at it). Negative reinforcement is an extremely potent training tool BUT please bear in mind, negative reinforcement usually involves protracted punishment, and with lengthy punishment, you are walking on thin ice vis a vis the relationship and dog’s temperament. You had better be absolutely certain that you have exquisite timing and that the nature of the punishment is not overbearing.
Whereas horse whisperers have used negative reinforcement to great effect when teaching horses to “join up”, the procedure is not without stress. Torture is the most extreme example of negative reinforcement
Owners and trainers have great difficulty estimating the optimum severity of a punishment in any given training scenario, especially when using aversive punishments. Trainers constantly walk a knife-edge; either they are too lenient and have to up the ante, or they are too severe and they damage the dog’s confidence. Beware of winning the battle and losing the war — by using too severe a punishment to resolve a simple behavior or training problem, only to destroy the dog's temperament in the process. In fact, remember, a trainer must first build up a dog's confidence before even thinking of punishing him at all. If you punish a dog that doesn't yet trust you, the damage to the relationship will likely be severe and longlasting.
If ever you feel a punishment was too harsh, immediately back-up and have the dog come and sit a number of times in succession, praising the dog each time he does so. Repeat this procedure until you have rebuilt the dog's confidence and enthusiasm and he happily and willingly approaches. Moreover, each time the dog comes and sits when requested, he is being compliant — in fact, voluntarily compliant. If however, the dog does not want to approach you, on no account approach the dog. Keep trying and do not give up until the dog approaches quickly. Yes, it may take you a long time to repair the damage from a single over-the-top punishment. And next time, lighten up — don't be so heavy-handed when training. The idea is to educate the dog by inhibiting an undesirable behavior and replacing it with a desirable one. We do not want to hurt, or frighten the poor critter out of his wits. Please remember, we love the dog. Rather, it is the undesirable behavior that we are trying to extinguish.
When using any kind of punishment and especially aversive punishments, always keep asking the question, “Did it work?” When punishments don’t work, beware of blithely increasing frequency, duration and severity. Change to Plan B. Seek help.
Dogs should be warned before being punished. If not warned beforehand, the dog cannot avoid the punishment, which would be inhumane. If not warned beforehand, the dog cannot learn the meaning of the warning, which would be inane. Warning the dog beforehand converts a relatively ineffective punishment training routine — wherein the dog is punished after misbehaving, into avoidance training — wherein the dog does not misbehave in order to avoid the punishment.
Usually, the warning is the same word that you initially used as a cue. For example, it would be next to impossible to stop a dog from barking altogether. Instead, punishment training would teach the dog to bark specifically at times when he is not punished, i.e., when the owner is away, or when he is not wearing his anti-bark collar. Instead, first teach the dog to “Woof” and “Shush” on cue (basic four-step lure/reward training) and then motivate the dog to want to comply (woof and shush for tummy rubs, couch privileges, tennis-ball throws and of course for supper). Then when the dog is barking at squirrels, instruct (warn) him to "Shush". If he barks, punish him. After barely half a dozen repetitions the dog will now cease barking whenever you ask him to shush. The dog learns that limited barking is fine most of the time but he must be quiet when told to shush. Training is much easier on the owner (as well as the dog) because now, the owner only has to pay absolute attention to the dog’s behavior each time he instructs/warns the dog to shush. At other times, the owner can relax, and so can the dog — knowing that he will not be surprised by an out-of-the-blue punishment.
Perhaps a human analogy will help put it in perspective. Let’s say we play a silly game and then out of the blue I give you an electric shock, and then another, and then yet another. What are you doing wrong? You haven’t a clue right. But I bet your behavior is a bit inhibited right now. Maybe you’re frozen to the spot and you probably don’t like this game any more than you like me. OK. Not fair. So instead, let’s imagine I first teach you the rules of the game; when I say “Frog” the first person to hop up to me gets a hundred bucks. Obviously, it is highly unlikely that I’ll have to use many more electric shocks. When I say, “Frog” the whole crowd will descend upon me as if competing in the Calaveras Frog Jumping Contest. But let’s say one person was a bit distracted, maybe getting a beer out of the fridge, or chatting up another contestant. He didn’t respond and so I zap him. I bet he hops immediately and I bet, it’s one-trial-learning — the next time I say “Frog” he’ll hop instantaneously. So, I trained all of the people with just one electric shock. Not too shabby. There is simply nothing worse than when a severe “aversive” comes out-of-the-blue — without warning and without the option of avoiding it.
I have used the example of an extremely aversive punishment here to make a point. However, I would never use aversive physical punishments when training a dog (or teaching a person). Aversive physical punishments are simply not necessary to achieve happy and willing compliance. Adequate prior training and motivation and the occasional gentle, yet insistent, instructive reprimands are all that are necessary.
Rather than being a non-specific nasty, in fact rather than being aversive at all, a punishment should communicate two things: What is wrong — i.e., the immediately preceding behavior, and What is right — as indicated by the specific environmental context, or better yet, by the meaning of the reprimand. For example, "Outside!" would be a suitable instructive reprimand for a puppy caught in the act of peeing on the living room carpet. The urgent tone and increased volume temporarily startle the puppy (contracting just about every sphincter in his body) and indicate the present behavior is not on but also, the meaning of the single word informs the confused puppy what he should be doing. Similarly, “Chewtoy!” means stop chewing what you’re chewing and chew your chewtoy instead. “Sit!” means, stop jumping–up, chasing your tail, chasing the cat, dashing out the front door… and sit.
When we train dogs, our efforts may not be tireless and consistent but when we use our voice, our feedback becomes analogue and instructive.
The Seventh Criterion
The seventh criterion illustrates why punishment-training is much less efficient than reward-training. The dog must know what the owner expects him to do, i.e., if this is wrong, then what is right? Therefore, before punishing a dog for making mistakes, the owner must first teach the dog the appropriate response. Reward-oriented methods require little time because the trainer only has to teach a single response — the correct one. With a punishment-oriented method, however, the trainer must punish the dog every time he does something wrong. Now, without even straining their brains, most puppydogs can think of oodles of ways to err. Punish the puppydog for each mistake and he will come up with oodles more. Remember, a major effect of punishment is to increase the variability of responding. Basically therefore, to reliably train a dog using punishments only, the dog must be punished for making an infinite number of mistakes. This, of course, would take an infinite length of time — an impossibility. Indeed, punishment-training becomes a Myth of Sisyphus — a neverending laborious task. Remember, there is an infinite number of wrong ways, but… THERE IS ONLY ONE RIGHT WAY! Hence reward-training takes only a finite amount of time.
The Eighth Criterion
The eighth criterion illustrates why punishment training is less effective than reward training. In fact, when using punishment-oriented methods, it is virtually impossible to satisfy the eighth criterion. In order for punishment-training to be effective, the dog MUST be punished EACH AND EVERY TIME he misbehaves. The dog only has to get away with misbehaving just ONCE to learn that there are times when he is not punished. Henceforth, the dog will reserve hiss misbehaviour for times when the owner is absent (physically or mentally) and when the dog is off-leash and out of arm’s reach. The owner's control has become punishment-contingent. Owner-absent behavior problems are the direct product of punishments failing to meet the eighth criterion.
On the other hand, using a reward-training method, quicker learning, better retention and improved motivation are all obtained via employing a differential reinforcement schedule, i.e., by NOT rewarding the dog each time it gets it right. Only reward the dog for above average responses, give better wards for better responses and give the very best rewards (jackpots) for the very best responses. Herein lies the difference. Without a doubt, reward-oriented training methods are easier, quicker, more efficient, more effective and certainly, more enjoyable than punishment-training methods.
From Bad to Worse
Once the owner is unable to catch the dog in the act of misbehaving, the effectiveness of punishment-training is reduced to zero. However, human nature being what it is... the fact that the owner cannot punish the dog when he misbehaves, doesn’t mean the owner will not punish the dog at all. On the contrary, the dog is punished when the owner returns (what a wonderful greeting!), or when the dog returns to the owner (what a wonderful reward for a recall!!).
The need for continued, prolonged or severe punishment is an advertisement of the owner's incompetence as a trainer. This tends to irritate most dog owners. The owner's anger is further inflamed by the assumption, that the dog knows he is doing wrong and therefore, must be misbehaving on purpose — an obvious indication that the dog is vindictive and spiteful. Many owners feel let down, frustrated and exasperated and often become openly hostile towards the dog. In frustration, the owner's punishments become more severe.
The Worst is Yet to Come
A well-socialized dog will tolerate an inordinate amount of psychological and physical punishment from his owner. The dog is a social animal and is accustomed to social reprimands. Group harmony is generally maintained by regular unsolicited demonstrations of friendly appeasement by lower-ranking individuals, which tend to offset the need for psychological harassment from higher-ranking animals. Additionally, when a dog transgresses pack rules and oversteps his mark, not only does the blundering underdog readily accept the reprimands but also, he actively makes amends afterwards. The transgressor defers in a stereotyped fashion in an attempt to reestablish peaceful coexistence, so that pack business may continue as usual. If well-socialized towards people, a dog will display similar behavior towards the owner. When reprimanded or punished, the dog immediately defers with a variety of appeasement gestures — whimpering, pawing, licking, jumping-up, rolling over and maybe urinating. In human terms, it is as if the dog is apologizing to his owner for the inconvenience of having to administer punishment. Sadly, the dog's obsequious, apologetic behavior is frequently misinterpreted as signs of guilt, which further enrages some owners, inciting more extreme punishment.
Basically well-socialized dogs are so beatable. Every time they are punished, they come back and “apologize”. And so, they are repeatedly punished. Apart from horses and humans, no other species would tolerate this level of senseless abuse. Some animals would not tolerate a single reprimand. Try giving a pig, a grizzly bear, a parrot or a dolphin a pop on the leash. They would all say the same thing: "Bye bye trainer. Why don’t you train on your own". Some animals might even give the “trainer” the animal equivalent of a pop on the leash before their departure. As would an unsocialized dog. Even our good friend the domestic kitty has this down to a tee. A single social reprimand and kitty Garboes-out, opting for an independent –and-aloof-existence.
If you really want to become an honest and effective dog trainer, work with a variety of species. Training a cat or a chicken to come and heel when called will make anyone a better dog trainer. Moreover, after working with other species, you will quickly be convinced, most dogs are eager to learn, exceedingly tolerant, and easy to train. So, let’s just teach them how we would like them to act and then we won’t need to punish them at all.
To hear more of Dr. Dunbar's insights check out his live appearance schedule, he's likely coming to a town near you!
This article is based on Dr. Dunbar's Behavior column in the January 1990 issue of the American Kennel Gazette. Reprinted with the permission of the author and the American Kennel Club.