Learning that occurs as a result of physical pain or discomfort is classified by psychologists as positive punishment. The commonplace “choke chain” is a good example of a positive punisher—the discomfort of being choked is intended to reduce the dog’s desire to pull on the leash. Dogs have sensitive necks, and so the neck is an obvious target for inflicting pain. The traditional choke chain, also known as a “check chain” or “slip-collar,” and its hardcore variant the “prong-collar” are designed to inflict momentary pain on the dog’s neck when it pulls on the leash. The dog should then learn, via positive punishment, to avoid the pain by not pulling on the lead. But this kind of training is ultimately ineffective: Most dogs whose owners require them to wear such a collar continue to pull, whenever their motivation to move toward something outweighs the pain. They may also habituate to the discomfort of the collar. In the case of the “citronella spray collar,” which is designed to suppress barking by dispensing an aversive odor whenever the dog vocalizes, a study has revealed that it is effective for a week or so, but dogs then habituate to the odor and, after two or three weeks, bark almost as often as they did before the collar was fitted.16
Slip- and prong-collars can be used only when the dog is on-leash and close to the owner (who could therefore just as easily be training the dog using positive reinforcement). Collars that deliver electric shocks allow the pain to be delivered remotely, by radio control. Either the controller can be held by the trainer and the shock timed to immediately follow the undesired behavior—livestock-chasing is a commonly cited example—or it can take the form of an “invisible fence,” a buried wire encircling an area that the dog is to be confined within, such that a shock is delivered if the dog approaches the “fence.”
There is a long tradition in animal psychology of using mild electric shocks to study the effects of punishment on behavior, and there is no doubt that, under controlled laboratory conditions, they do alter an animal’s tendency to behave in a particular way. Dog training is not, however, carried out under controlled laboratory conditions, and so what the dog learns is often not entirely what was intended. For example, research has shown that German shepherd dogs trained to be guard dogs using shock collars were more frightened of their (“expert”) handlers, even when they were not wearing their collars, than were dogs trained conventionally (using a mixture of reward and punishment).17 It seems very likely that these dogs were associating the shocks with their handlers, as well as with the mistakes the dogs had made that triggered the shocks.
When the shock is not timed properly, the dog’s fear and anxiety may be even worse than this. Owners who believe that dogs can “know what they’ve done wrong” may go on applying the shock even after the dog has stopped performing the undesired behavior. Even under controlled laboratory conditions, unpredictable electric shock makes animals much more likely to react aggressively, so it is a real possibility that owners who use shock-collars without proper technique are setting up their dog for an attack that will condemn it to euthanasia. Cases have been documented in which shocks received from invisible fences apparently caused dogs to launch unprovoked and serious attacks on people.18
The pain from the shock itself has a significant effect on the dog’s welfare, ranging from slight to considerable. This depends on whether the shock is used appropriately or not. Because dogs can habituate to aversive events, for the collar to be effective it’s essential to get the level of shock right the first time—too weak, and the dog may habituate to the pain, requiring the trainer to increase the shock level in an attempt to get the dog to respond. There is thus a temptation to apply the maximum available shock right from the start, even though the pain felt by the dog will vary widely, depending on the thickness of its coat, the electrical resistance of its skin, and whether its coat is wet or dry. At worst, this can result in both immediate discomfort and then anxiety if the dog cannot predict when the next violent pain to its neck is likely to arrive.
Repeated delivery of electric shocks is likely to lead to serious fear and anxiety. It is unarguable that the shocks must be momentarily painful; otherwise, they would not have any effect on behavior. A single shock delivered at exactly the right moment to suppress an undesired behavior may have only a transitory effect on the dog’s welfare and, therefore, may be preferable to other, less effective punishments that have to be delivered repeatedly. However, if the dog receives several shocks and is unable to work out why, its heart rate goes up and its stress hormones increase dramatically; both are indicators that its welfare is being impaired. In the hands of inept operators who do not understand the principles behind their use or, worse still, are using the collar as an outlet for their own anger that the dog is not doing what it’s supposed to, electric shocks may not only upset the dog but undermine the relationship between the dog and its owner.
Physical punishment can also be used in another way, in which the dog learns to do something that enables it to avoid a painful sensation—as opposed to not doing something to avoid pain, as in the examples so far. The technical term for this is negative reinforcement. Negative reinforcement is the principle underlying training methods that teach the dog to perform a behavior in order to eliminate pain that is being inflicted by the trainer. One example is a traditional method of teaching young gundogs to retrieve, in which the dog learns to associate the removal of pain with opening its mouth and picking something up. The trainer presents a suitable object—usually a cloth-covered “dummy”—a few inches in front of the dog’s nose while at the same time pinching the dog’s earflap tightly. The dog cries out in pain, opening its mouth and allowing the trainer to insert the dummy into the dog’s open mouth. At that instant, the aversive stimulus (the pain from the pinch) is abruptly stopped. The dog is supposed to learn that picking up the dummy results in the cessation of pain. (This is an example of negative reinforcement.) In the “force-fetch” method, the dog is additionally beaten with a stick while the ear pinch is being applied; in this case, both are abruptly terminated when the dog performs the desired action.
There is no doubt that both positive punishment and negative reinforcement, if performed skillfully, can be highly effective in the very short term—provided that we put any ethical considerations to one side and ignore any long-term damage to the dog-human relationship. Things go wrong, for both dog and owner, when the way that punishment works is not properly understood or, worse, when punishment is applied as an outlet for the owner’s anger, frustration, or embarrassment.
The most serious problem with physical punishment is that it is easily misapplied. As with other kinds of learning, a dog will almost always associate any sudden pain or fear with the event that immediately preceded it. Almost every day, I see owners beating or remonstrating with their dogs for being slow to come back to them. This action usually follows some mildly embarrassing incident that the dogs have been involved in and seems to be, in part, an emotional response from the owners and, in part, a sign to anyone watching that they are dissociating themselves from their dogs’ behavior. What are the dogs learning from such incidents? That coming back is (sometimes) a bad idea. Is this going to make them more, or less, attached to their owners? More, or less, likely to come back when called the next time? Dogs don’t think “I was naughty a minute ago, so I deserve to be punished when I get back to my owner.” If anything, they think “Sometimes my owner makes a fuss of me when I go back, but sometimes I get hit; I don’t get it.”
Force-fetch training. The trainer simultaneously pinches the dog’s ear and beats it with a stick until it opens its mouth above the “dummy.”
In short, punishment is often misused by owners who don’t understand how dogs learn. But even if punishment is applied with the correct timing, it is difficult to predict in advance what the dog will associate the punishment with. Will it be the intended event, or will it be the person administering it or the place in which the punishment happened? One of my colleagues recently found his neighbor’s small terrier wa
ndering in the road. He picked it up and carried it back to the neighbor’s house. As he approached she appeared at the door, repeatedly pressing the control for the electric shock collar that the dog was wearing and shouting “Bad dog, bad dog.” The dog jerked violently every time the shock was applied, growling and snapping in fear. From that day on, the terrier growled whenever anyone it didn’t know came near it—it had learned that unfamiliar people mean inescapable pain.
It’s also difficult to get the severity of the punishment right. Typically, owners will gradually escalate the degree of punishment that they use, until they achieve the desired result. Unfortunately, the dog will at the same time be habituating to the punishment, such that the intensity that eventually works (if at all) is much higher than would have been needed if it had been applied correctly in the first place. Contrast this with the use of positive rewards in dog training: If the owner mistakenly gives too much of the reward each time, the value of the reward will merely be somewhat diminished—for example, too big a food reward will just make the dog less hungry. But if a punishment is more intense than necessary, the dog will suffer unnecessarily—and if it is not sufficiently intense to begin with (due to the owner, understandably, erring on the side of caution), it can lead to escalation and therefore also eventually result in too much being applied.
In fact, a growing body of evidence indicates that, in inexpert hands, physical punishment not only is likely to harm the dog but is ineffective as well. Two separate surveys of dog owners have revealed that dogs trained with punishment tend to be less obedient and more fearful than those trained with reward. In the first of these, conducted in the UK,19 364 owners were asked about the training methods that they used for training seven basic tasks, including housebreaking, coming when called, and giving up an object upon command. Vocal punishment was reported by 66 percent of the respondents, and physical punishment by 12 percent. Rewards were also commonly used: 60 percent used verbal praise and 51 percent used food treats. The owners using rewards reported much greater obedience from their dogs than those using punishment predominantly, whereas those using mainly punishment reported a larger number of behavioral problems such as barking at people and dogs, fearful behavior, and separation disorders. The other survey, conducted in Austria,20 also concluded that frequent use of punishment is associated with high levels of aggression, especially in small dogs.
Not all punishment is physical, however, even in dog training. Training almost inevitably involves subtler forms of punishment that most owners wouldn’t even identify as such. In addition to the physical punishments discussed so far, psychologists have identified a category that they call negative punishment, which involves the removal of some reward that the dog has come to expect will occur under a particular set of circumstances. For example, one way to stop a dog jumping up at people is to always ignore it when it does so. No pain or fear is involved; however, the dog presumably becomes mildly anxious when it finds that its strategy for getting interaction with its owner suddenly no longer works. Some trainers in the UK regard even this emotional shift as unethical, but given that even experimental psychologists cannot always determine whether learning is mainly due to negative punishment (brief anxiety that the anticipated reward won’t arrive) or to positive reinforcement (joy when it does), it is probably difficult to avoid all occurrences of negative punishment during everyday interactions with a dog. Perhaps the best compromise is never to use negative punishment on its own but always to offer the dog a rewarded alternative. In any event, withdrawal of reward, as with all forms of punishment, actually works faster when the dog is given an alternative, positively rewarded strategy. Such a strategy—for example, in cases involving jumping-up behavior—might entail making a fuss over the dog only when it is calm.
It’s also possible to transfer the negative punishment to a secondary punisher, which is analogous to a secondary reinforcer. In the context of dog training, an arbitrary yet distinctive sound can be used; however, it must be the exact opposite of that produced by a clicker. For this purpose owners can purchase “training discs,”21 but a distinctively uttered word can be just as effective. First, an association is built up between the cue (e.g., dropping the training discs on the ground) and a mildly frustrating event (e.g., temporarily removing the dog’s food mid-meal, or offering and then withholding a treat). The cue itself then becomes a secondary punisher and can be used under other circumstances as well, such as getting the dog to stop barking. As with all punishers, including negative reinforcers, the desired response—such as sitting and not barking—should also be rewarded. (This is an example of positive reinforcement.)
Punishment, in the psychological sense, is an inescapable component of the dog owner’s armory of training methods. Today’s most knowledgeable trainers agree that it is in fact impossible to avoid at least some negative punishment (the withholding of a reward that the dog is anticipating) when training a dog. And few would argue that it’s unethical to make a dog feel slightly uncomfortable by delaying somewhat a reward it was expecting. (Indeed, in real life this may be virtually unavoidable. Each time owners give their dogs a food treat or other reward, they set up expectations that the same reward will appear again when the situation is repeated. If the situation recurs and there is no treat, then the dogs are, technically speaking, being punished.) Many trainers who avoid physical punishment nevertheless use the withholding of reward as a way of modifying behavior—clearly a very good way of getting and holding a dog’s attention,22 once reward-based training has begun.
It is the use of physical punishment that remains controversial. Traditions die hard, and “traditional” training methods based on physical punishment are still widely employed. One recent US survey of clients at a veterinary behavior clinic23 found large numbers of owners using confrontational methods, including “hit or kick dog for undesirable behavior” (43 percent), “physically force the release of an item from a dog’s mouth” (39 percent), “alpha roll” (31 percent), “stare at or stare [dog] down” (30 percent), “dominance down” (29 percent), and “grab dog by jowls and shake” (26 percent). All of these actions had elicited an aggressive response from at least a quarter of the dogs on whom they had been used, indicating that none of them could be regarded as safe.
The reason for the widespread incidence of physical punishment was unclear from the research, since few owners in the survey indicated that these techniques had been recommended by dog trainers (although television was the most frequently reported source for the practice of abruptly jabbing the dog in the neck). However, much of the advice reported to have come from trainers did involve the punitive use of collars and leashes, such as prong-collars and forcing-down, which also cause aggressive responses. The survey found no link between aggression and the use of training methods that did not involve physical punishment. Although the majority of the dogs had been brought to the clinic for problems involving aggression, none of the nonaversive, neutral, and reward-based interventions that their owners had used produced aggressive responses in more than a tiny fraction of the dogs they’d been used on, in stark contrast to the aggression triggered by physical punishments.
Why, therefore, do TV companies seem to prefer to publicize methods based on confrontation and punishment? Perhaps because conflict, and its dramatic resolution, make for compelling entertainment.24 Reward-based methods are slower, if surer, and much less dramatic. If dog-training programs were regarded as mere entertainment, then none of this would matter very much. But if the use of physical punishment and other techniques that supposedly reduce “dominance” are adopted in good faith by dog owners, problematic behaviors can easily be exacerbated. When such techniques don’t work as quickly and effortlessly as the TV version seems to promise, there is a risk that owners will escalate the punishment, in the belief that they have not gotten through to the dog. The result can then be a dog that resorts to aggression because it finds this is the only tactic that gets it noticed. Inept use of reward, by
contrast, is likely to result only in an overweight or overdependent dog—both conditions that, while not exactly desirable, can at least be readily resolved.
Given all the scientific evidence that is piling up against the use of physical punishment in training, the question that has to be asked is why such methods remain so popular with owners. The main issue seems to be that dog training originated as a craft and, hence, has no clear route for integrating scientific understanding of dogs into its methods. Neither dog training nor the treatment of behavioral disorders in dogs is a regulated profession, so keeping up to date with the latest developments or, indeed, getting some kind of formal education in the field is not legally required. Television companies also seem not to value formal qualifications; for example, Cesar Millan and Victoria Stilwell (whose approaches, it must be said, are very different) do not mention any academic qualifications on their websites. Increasingly, however, there are moves on both sides of the Atlantic to introduce self-regulation at all levels. The biennial International Veterinary Behavior meetings and the Journal of Veterinary Behavior (neither, despite their names, restricted to veterinarians) are just two of the ways in which new ideas and research are being exchanged internationally.
Dog Sense Page 15