Dog training is the application of behavior analysis which uses the environmental events of antecedents and consequences to modify the behavior of a dog, either for it to assist in specific activities or undertake particular tasks, or for it to participate effectively in contemporary domestic life. While training dogs for specific roles dates back to Roman times at least, the training of dogs to be compatible household pets developed with suburbanization in the 1950s.
A dog learns from interactions it has with its environment. This can be through classical conditioning, where it forms an association between two stimuli; non-associative learning, where its behavior is modified through habituation or sensitisation; and operant conditioning, where it forms an association between an antecedent and its consequence.
There are a variety of established methods of animals training, each with its adherents and critics. Some of the better known dog training procedures include the Koehler method, clicker training, motivational training, electronic training, model-rival training, dominance-based training, and relationship-based training. The common characteristics of successful methods are knowing the animal's attributes and personality, accurate timing of reinforcement and/or punishment and consistent communication. The use of punishment is controversial with both the humaneness and effectiveness questioned by many behaviourists.
Maps, Directions, and Place Reviews
Definition
Dog training is teaching a response to cues or commands, or the performance of actions not necessarily natural to the dog, and also raising a dog accommodated to his environment by modifying natural digging, barking and eliminating behaviors. Dog training is defined as the purposeful changing of a dog's behavior.
Dog training can be socialisation to the domestic environment, basic obedience training or training for specialized activities including law enforcement, search and rescue, hunting, working with livestock, assistance to people with disabilities, entertainment, dog sports, detection and protecting people or property.
Guard Dog Training School Video
History
Although research into how dogs learn and into cross-species communication has changed the approach to dog training in recent decades, understanding the role of early trainers and scientists contributes to an appreciation of how particular methods and techniques developed.
Before 1900
In around 127-116 B.C. a Roman farmer, Marcus Varro, recorded advice on raising and training puppies for herding livestock. His writings indicate that not only was dog training for specific tasks well established, but that the value of early training was recognised.
In 1848 W. N. Hutchinson published his book Dog Breaking: The Most Expeditious, Certain and Easy Method, Whether Great Excellence or Only Mediocrity Be Required, With Odds and Ends for Those Who Love the Dog and the Gun. Primarily concerned with training hunting dogs such as pointers and setters, the book advocates a form of reward-based training, commenting on men who have "a strong arm and a hard heart to punish, but no temper and no head to instruct" and suggesting "Be to his virtues ever kind. Be to his faults a little blind." Stephen Hammond, a writer for Forest and Stream magazine, advocated in his 1882 book Practical Training that hunting dogs be praised and rewarded with meat for doing the correct behavior.
War years
Konrad Most began training dogs for police work in Germany, and was appointed principal of the State Breeding and Training Establishment for police dogs in Berlin, where he carried out original research into training dogs for a broad range of service tasks. At the outbreak of war in 1914 he was charged with organising and directing the use of dogs to further the war effort. He headed the Experimental Institute for Armed Forces' Dogs during the Second World War, and afterwards ran the German Dog Farm, a centre for the training of working dogs, including assistance dogs for the blind. He played a leading role in the formation of the German Canine Research Society and Society for Animal Psychology. His 1910 publication, Training Dogs: A Manual, emphasised using instinctive behavior such as the prey drive to train desired behaviors, advocated the use of compulsion and inducements, differentiated between primary and secondary reinforcers, and described shaping behaviors, chaining components of an activity, and the importance of timing rewards and punishments. The book demonstrated an understanding of the principles of operant conditioning almost thirty years before they were formally outlined by B.F. Skinner in The Behavior of Organisms. While publishers of the 2001 reprint warn that some of the "compulsive inducements" such as the switch, the spiked collar and the forced compliance are unnecessarily harsh for today's pet dogs, the basic principles of Most's methods are still used in police and military settings.
Marian Breland Bailey played a major role in developing empirically validated and humane animal training methods and in promoting their widespread implementation. Marian was a graduate student under B.F. Skinner. Her first husband Keller Breland also came to study with Skinner and they collaborated with him, training pigeons to guide bombs. The Brelands saw the commercial possibilities of operant training, founding Animal Behavior Enterprises (ABE). In 1955, they opened the "I.Q. Zoo" as both a training facility and a showcase of trained animals. They were among the first to use trained animals in television commercials, and the first to train dolphins and whales as entertainment, as well as for the navy. Keller died in 1965, and in 1976 Marian married Bob Bailey, who had been director of marine mammal training for the navy. They pioneered the use of the clicker as a conditioned reinforcer for training animals at a distance. ABE went on to train thousands of animals of more than 140 species. Their work had significant public exposure through press coverage of ABE-trained animals, bringing the principles of behavior analysis and operant conditioning to a wide audience.
Konrad Lorenz, an Austrian scientist who is regarded as developing the foundations of ethological research, further popularised animal behaviorism with his books, Man Meets Dog and King Solomon's Ring. Lorenz stated that there were three essential commands to teach a dog: "lie down" (stay where you are), "basket" (go over there) and "heel" (come with me).
In 1935, the American Kennel Club began obedience trials, and in the following years popular magazines raised public awareness of the benefits of having a trained pet dog, and of the recreational possibilities of dog training as a hobby. After WWII, the increasing complexities of suburban living demanded that for a pet dog's own protection and its owner's convenience, the dog should be obedient. William Koehler had served as principal trainer at the War Dog Training Center, in California, and after the war became chief trainer for the Orange Empire Dog Club--at the time, the largest dog club in the United States--instructor for a number of breed clubs, and a dog trainer for the Walt Disney Studios. In 1962 Koehler published The Koehler Method of Dog Training, in which he is highly critical of what he calls "tid-bit training techniques" based in "the prattle of 'dog psychologists'". Amongst the training innovations attributed to Koehler is the use of a long line in conjunction with a complete absence of oral communication as a way of instilling attentiveness prior to any leash training. Koehler insisted that participants in his training classes used "emphatic corrections", including leash jerks and throw chains, explaining that tentative, nagging corrections were cruel in that they caused emotional disturbance to the dog. Vicki Hearne, a disciple of Koehler's, commented on the widespread criticism of his corrections, with the explanation that it was the emotionally loaded language used in the book that led to a number of court cases, and to the book being banned in Arizona for a time. Despite the controversy, his basic method forms the core of many contemporary training systems.
Post WWII
In the 1950s Blanche Saunders was a staunch advocate of pet-dog training, travelling throughout the U.S. to promote obedience classes. In The Complete Book of Dog Obedience, she said, "Dogs learn by associating their act with a pleasing or displeasing result. They must be disciplined when they do wrong, but they must also be rewarded when they do right." Negative reinforcement procedures played a key part in Saunders' method, primarily the jerking of the choke chain. The mantra taught to students was "Command! Jerk! Praise!" She felt that food should not be an ongoing reward, but that it was acceptable to use "a tidbit now and then to overcome a problem." Saunders perhaps began the shift away from military and police training methods, stressing repeatedly the importance of reinforcement for good behaviour in training--a move toward the positive training methods used today.
In 1965, John Paul Scott and John Fuller identified the critical periods for learning and social development in puppies, and published Genetics and the Social Behavior of the Dog, a landmark study of dog behavior.
The 1980 television series Training Dogs the Woodhouse Way made Barbara Woodhouse a household name in the UK, and the first international celebrity dog trainer. Known for her "no bad dogs" philosophy, Woodhouse was highly critical of "bad owners", particularly those she saw as "overly sentimental". She described the "psychoanalyzing of dogs" as "a lot of rubbish". Her no-nonsense style made her a pop-culture icon, with her emphatic "sit" and catch cry of "walkies" becoming part of the popular vernacular.
The Monks of New Skete, who were breeders and trainers of German Shepherds in Cambridge, New York, published How to Be Your Dog's Best Friend: A Training Manual for Dog Owners in 1978 and it became an immediate best seller. Despite advocating a philosophy that "understanding is the key to communication and compassion with your dog," they endorsed confrontational punishments which were later shown to elicit dangerously aggressive responses in many dogs.
In the 1980s veterinarian and animal behaviourist Ian Dunbar discovered that despite evidence on the peak learning periods in animals, few dog trainers worked with puppies before they were six months old. Dunbar founded Sirius Dog Training, the first off-leash training program specifically for puppies, which emphasizes the importance of teaching bite inhibition, sociality, and other basic household manners, to dogs under six months of age. Dunbar has written numerous books, and is known for his international seminar presentations and award-winning videos on puppy and dog behavior and training.
Prior to the 1980s, Karen Pryor was a marine-mammal trainer who used Skinner's operant principles to teach dolphins and develop marine-mammal shows. In 1984, she published her book, Don't Shoot the Dog: The New Art of Teaching and Training, an explanation of operant-conditioning procedures written for the general public. In the book Pryor explains why punishment as a way to get people to change often fails, and describes specific positive methods for changing the behaviour of husbands, children and pets. Pryor's dog training materials and seminars showed how operant procedures can be used to provide training based on positive reinforcement of good behavior. Pryor and Gary Wilkes introduced clicker training to dog trainers with a series of seminars in 1992 and 1993. Wilkes used aversives as well as rewards, and the philosophical differences soon ended the partnership.
21st century
The 21st century has seen the proliferation of television programs and accompanying books that feature dog training and rehabilitation, including Joel Silverman's Good Dog U, Dog Whisperer with Cesar Millan, It's Me or the Dog featuring Victoria Stillwell, The Underdog Show, Dogs in the City, and SuperFetch. The Association of Pet Dog Trainers advises that television programs are produced primarily for entertainment, and while all programs will have good and not-so-good points, the viewer should critically evaluate the information before deciding which training tips to adopt.
How dogs learn
Operant conditioning
Operant conditioning (or instrumental conditioning) is a form of learning in which an individual's behavior is modified by its consequences. Two complementary motivations drive instrumental learning: the maximization of positive outcomes and minimization of aversive ones. There are two ways in which behavior is reinforced or strengthened: positive reinforcement occurs when a behavior is strengthened by producing some desirable consequence; negative reinforcement occurs when a behavior is strengthened by avoiding some undesirable consequence. There are two ways in which behavior is decreased or weakened: negative punishment occurs when a behavior is weakened by not producing a reinforcing consequence; and positive punishment occurs when a behavior is weakened by producing a consequence that is a disincentive. In combination, these basic reinforcing and punishing contingencies provide four ways for modifying behavior. Reinforcement increases the relative probability or frequency of the behavior it follows, while punishment decreases the relative probability or frequency of the behaviour it follows.
Typical positive reinforcement events will satisfy some physiological or psychological need, so it can be food, a game, or a demonstration of affection. Different dogs will find different things reinforcing. Negative reinforcement occurs when a dog discovers that a particular response ends the presentation of an aversive stimulus. An aversive is anything that the dog does not like, such as a tight choke chain.
Punishment is operationally defined as an event that lowers the probability of the behavior that it follows. It is not "punishment" in the common sense of the word, and does not mean physical or psychological harm and most certainly does not mean abuse. Punishment simply involves the presentation of an undesired consequence (positive punishment) when the wrong behavior is performed, such as a snap of the leash, or the removal of a desired consequence (negative punishment) when the wrong behavior is performed, such as the owner eating the cheese that would have been the reward. A behavior that has previously been developed will cease if reinforcement stops; this is called extinction. A dog that paws its owner for attention will eventually stop if it no longer receives attention.
Classical conditioning
Classical conditioning (or Pavlovian conditioning) is a form of learning in which one stimulus, the conditioned stimulus, comes to signal the occurrence of a second stimulus, the unconditioned stimulus. Classical conditioning is when a dog learns to associate things in its environment, or discovers some things just go together. A dog may become afraid of rain through an association with thunder and lightning, or it may respond to the owner putting on a particular pair of shoes by fetching its leash. Classical conditioning is used in dog training to help a dog make specific associations with a particular stimulus, particularly in overcoming fear of people and situations.
Non-associative learning
Non-associative learning is a change in a response to a stimulus that does not involve associating the presented stimulus with another stimulus or event such as reward or punishment. Habituation is non-associative learning. An example is where a dog that reacts excitedly to a door bell is subjected to repeated ringing without accompanying visitors, and stops reacting to the meaningless stimuli. It becomes habituated to the noise. On the other side of habituation is sensitization. Some dogs' reactions to the stimuli become stronger instead of them habituating to the repeated stimuli or event. Desensitization is the process of pairing positive experiences with an object, person, or situation that causes fear or anxiety. Consistent exposure to the feared object in conjunction with rewards allows the animal to become less stressed, thereby becoming desensitized in the process. This type of training can be effective for dogs who are fearful of fireworks.
Learned irrelevance is where dogs that are overexposed to a stimulus or cue learn the cue is irrelevant because the exposure has proven to be uneventful. So a dog owner who continually says "Sit, sit" without response or consequence, inadvertently teaches the dog to ignore the cue.
Learned helplessness occurs when a dog ceases to respond in a situation where it has no option to avoid a negative event. For learned helplessness to occur, the event must be both traumatic and outside the dog's control. Family dogs that are exposed to unpredictable or uncontrolled punishment are at risk of developing disturbances associated with the learned helplessness disorder. Punishment which is poorly coordinated with identifiable avoidance cues or response options, such as when punishment takes place long after the event, meet the criteria of inescapable trauma.
Observational learning
Observational learning is the learning that occurs through observing the behavior of others. This form of learning does not need reinforcement to occur; instead, a model animal is required. While the model may not be intentionally trying to instill any particular behavior, many behaviors that are observed are remembered and imitated. The domestic dog is a social species and its social dependency makes it aware of the behavior of others, which contributes to its own behavior and learning abilities. There is, however, ongoing discussion about how much, and how, dogs can learn by interacting with each other and with people.
The term "observational learning" encompasses several closely related concepts: allelomimetic behavior or mimicking where, for example, puppies follow or copy others of their kind; social facilitation where the presence of another dog causes an increase in the intensity of a behavior; and local enhancement which includes pieces of social facilitation, mimicking, and trial-and-error learning, but is different from true observational learning in that the dog actively participates in the behavior in the presence of the other dog and/or other environmental cues. Four necessary conditions for observational learning are: attention, retention, motivation, and production. That is, the dog must pay attention to the dog or person performing the modelled behavior; retain the information gathered about the behavior during the observation; be motivated to reproduce the behavior in a time and place removed from the original; and finally, produce the behavior, or some reasonable facsimile thereof.
Pups between the ages of 9-12 weeks who were permitted to observe their narcotics-detecting mothers at work generally proved more capable at learning the same skills at six months of age than control puppies the same age who were not previously allowed to watch their mothers working. A 2001 study recorded the behaviour of dogs in detour tests, in which a favorite toy or food was placed behind a V-shaped fence. The demonstration of the detour by humans significantly improved the dogs' performance in the trials. The experiments showed that dogs are able to rely on information provided by human action when confronted with a new task. Significantly, they did not copy the exact path of the human demonstrator, but adopted the detour behavior shown by humans to reach their goal. A 1977 experiment by Adler and Adler found that puppies who watched other puppies learn to pull a food cart into their cages by an attached ribbon proved considerably faster at the task when later given the opportunity themselves. At 38 days of age, the demonstrator puppies took an average of 697 seconds to succeed, while the observers succeeded in an average of 9 seconds.
Training methods
Koehler method
Strictly following the model set out in the Koehler Method of Dog Training, some 50 years later, the Koehler method continues to be taught in both class and private training formats. The method is based in the philosophy that a dog acts on its right to choose its actions. Koehler explained that a dog's learned behavior is an act of choice based on its own learning experience. When those choices are influenced by the expectation of reward, the behavior will most likely be repeated, and when those choices are influenced by the anticipation of punishment, they will most likely cease. Once the dog has learned that its choices result in comfort or discomfort it can be taught to make the correct decisions. Action->Memory->Desire encapsulates the learning pattern used by the method; the dog acts, remembers the consequences, and forms the desire to repeat or avoid those consequences. Adherents believe that once the behavior has been correctly taught, it should be performed, thus making any correction, fair, reasonable, and expected. While the model has been used consistently since 1962, some of the punishment procedures described in the book are now not considered necessary, humane, or appropriate by many trainers.
Motivational training
Purely positive or motivational training employs the use of rewards to reinforce good behavior, and ignores all bad behavior. It is based in Thorndike's Law of Effect, which says that actions that produce rewards tend to increase in frequency and actions that do not produce rewards decrease in frequency.
Motivational training has its roots in captive animal training, where compulsion and corrections are both difficult and dangerous, and ignoring bad behavior is not problematic as the animal lives under controlled conditions. As a dog training strategy, purely positive training is feasible, but difficult, as it requires time and patience to control the rewards the dog receives for behavior. Some activities such as jumping up or chasing squirrels are intrinsically rewarding, the activity is its own reward, and with some activities the environment may provide reinforcement such as when the response from dog next door encourages barking.
Clicker training
Clicker training is a nickname given to a positive reinforcement training system based on operant conditioning. The system uses conditioned reinforcers which are able to be delivered more quickly and more precisely than primary reinforcers such as food. The term 'clicker' comes from a small metal cricket adapted from a child's toy that the trainer uses to precisely mark the desired behavior; however, some trainers use a whistle, a word, or even a light as the conditioned reinforcer. The trainer delivers a primary reinforcer, such as a toy or treat, after the noise or signal.
Electronic training
Electronic training involves the use of an electric shock as an aversive. Common forms are collars which can be triggered remotely, or that are triggered by barking, fencing that delivers a shock when a dog wearing a special collar crosses a buried wire, and mats that can be placed on furniture to deliver a shock. Some aids deliver an aversive such as a spray of citronella when triggered. The use of electric shock aversives for training dogs is the subject of considerable controversy. Supporters claim that the use of electronic devices allows training at a distance and the potential to eliminate self-rewarding behaviour, and point out that properly used, they have less risk of stress and injury than mechanical devices, such as choke chains. Opponents cite the risks of physical and psychological trauma associated with incorrect or abusive use.
In one study laboratory-bred Beagles were divided into three groups. Group A received an electric shock when the dogs touched the prey (a rabbit dummy fixed to a motion device). Group H received a shock when they did not obey a previously trained recall command during hunting. Dogs in group R received the electric shock arbitrarily, i.e. the shock was administered unpredictably and out of context. Group A did not show a significant rise in salivary cortisol levels, while group R and group H did show a significant rise. This led to the conclusion that animals which were able to clearly associate the electric stimulus with their action, i.e. touching the prey, and consequently were able to predict and control the stressor, did not show considerable or persistent stress indicators, while animals that were not able to control the situation to avoid the shock did show significant stress.
In 2004 a study was published that was based on the observation of a variety of breeds trained for protection work using shock collars, which showed that although electronically trained dogs can excel as guard dogs, their behavior toward humans and work circumstances changed, often indicating heightened uncertainty and reactivity.
Lindsay says of this study, "Schilder and Van der Borg (2004) have published a report of disturbing findings regarding the short-term and long- term effects of shock used in the context of working dogs that is destined to become a source of significant controversy ... The absence of reduced drive or behavioral suppression with respect to critical activities associated with shock (e.g., bite work) makes one skeptical about the lasting adverse effects the authors claim to document. Although they offer no substantive evidence of trauma or harm to dogs, they provide loads of speculation, anecdotes, insinuations of gender and educational inadequacies, and derogatory comments regarding the motivation and competence of IPO trainers in its place."
Model-rival training
Based on the principles of social learning, model-rival training uses a model, or a rival for attention, to demonstrate the desired behaviour. The method was used by Irene Pepperberg to train Alex the African Grey Parrot to label a large number of objects. McKinley and Young undertook a pilot study on the applicability of a modified version of the model-rival method to the training of domestic dogs, noting that the dog's origins as a member of large and complex social groups promote observational learning. The model-rival training involved an interaction between the trainer, the dog, and a person acting as a model-rival, that is, a model for desired behaviour and a rival for the trainer's attention. In view of the dog, a dialogue concerning a particular toy commenced between the trainer and the model-rival. The trainer praised or scolded the model-rival depending on whether the model-rival had named the toy correctly. It was found that the performance times for completion of the task were similar for dogs trained with either operant conditioning or the model rival method. In addition, the total training time required for task completion was comparable for both methods.
A Hungarian dog training group called Népszigeti Kutyaiskola use a variation of model-rival training which they describe as the Mirror Method. The mirror method philosophy is that dogs instinctively learn by following the example of others in their social sphere. Core to the program is including the dog in all aspects of the owner's life and positive reinforcement of copying behaviors. Mirror method dog training relies on using a dog's natural instincts and inclinations rather than working against them.
Dominance-based training
The concepts of "pack" and "dominance" in relation to dog training originated in the 1940s and were popularized by the Monks of New Skete in the 1970s. The model is based on a theory that "dogs are wolves" and since wolves live in hierarchical packs where an alpha male rules over everyone else, then humans must dominate dogs in order to modify their behavior. However, recent studies have shown that wolves in the wild actually live in nuclear families where the father and mother are considered the pack leaders, and their offspring's status depends on their birth order which does not involve fighting to attain a higher rank, because the young wolves naturally follow their parents' lead.
Animal behaviorists assert that using dominance to modify a behavior can suppress the behavior without addressing the underlying cause of the problem. It can exacerbate the problem and increase the dog's fear, anxiety, and aggression. Dogs that are subjected to repeated threats may react with aggression not because they are trying to be dominant, but because they feel threatened and afraid.
Researchers have described several reasons why the dominance model is a poor choice for dog training. First, a relationship based on dominance is established to gain priority access to scarce resources, not to impose particular behaviors on the less dominant animal, so the dominance model is irrelevant for most of the behaviors that people want from their dogs, such as coming when called or walking calmly on a leash. Second dominance-submission relationships, once established, are constantly tested and must be regularly reinforced. Thus people, particularly children and the elderly, may not be able to retain their rank and are at risk of being injured if they attempt to do so. Third, dominant individuals gain priority access to resources, but only while they are present, establishing dominance over a dog does not guarantee its behavior when the dominant individual is distant or absent.
Relationship-based training
Derived from the theories of symbolic interactionism, relationship based training exploits the patterns of communication, interpretation and adjustment between dogs and their trainers. Building on a positive relationship between them, the method sets out to achieve results that benefit both the dog and the trainer, while at the same time enhancing and strengthening their relationship. The basic principles include ensuring that the dog's basic needs have been met before beginning a training session, finding out what motivates the dog and using it to elicit behaviours, interpreting the dog's body language to improve communication between dog and trainer, using positive reinforcement to encourage desired behavior, training incompatible behaviors to replace unwanted behaviours, and controlling the dog's environment to limit the possibility of unwanted behaviours. A relationship-based approach to dog training is not reliant on using particular training aids or treats, the relationship is always there, and the connection between dog and trainer is sufficiently powerful to achieve the training goals.
Factors
Training can take as many forms as there are trainers, however a detailed study of animal trainers found common characteristics of successful methods: thoughtful interpretation of what the animal does prior to training, accurate timing and consistent communication.
Communication
Dogs have become closely associated with humans through domestication and have also become sensitive to human communicative signals. Generally, they have a lot of exposure to human speech, especially during play, and are believed to have a good ability to recognize human speech. Two studies investigated the ability of a single dog that was believed to be exceptional in its understanding of language. Both studies revealed the potential for at least some dogs to develop an understanding of a large number of simple commands on the basis of just the sounds emitted by their owners. However the studies suggested that visual cues from the owner may be important for the understanding of more complex spoken commands.
Understanding
For any of these techniques, consistency of the owner's training/behavior and level of engagement can influence the effectiveness of any technique applied.
Innate characteristics
In considering the natural behaviours of specific breeds of dogs, it is possible to train them to perform specialised, highly useful, tasks. For example, Labrador retrievers are the favoured breed for the detection of explosives. This is because of a combination of factors including their food drive which enables them to keep focused on a task despite noise and other distractions. Most working breeds of dogs are able to be trained to find people with their sense of smell (as opposed to their sense of sight). Cocker Spaniels are able to be trained as part of a termite detection team. Their relatively small size enables them to fit into small spaces, and their light weight allows them to walk on areas of ceiling which would be dangerous to anything heavier. In fact, although unusual, termite detection dogs are much more reliable at detecting termites than humans who rely on a basic system of tapping and listening. Because of their ability to learn signals by sight and for their energetic and athletic natures, German Shepherds are able to be trained for work alongside search and rescue teams and human apprehension teams.
Individualized and/or class training
Individualised training is used with dogs that have an urgent or unique training problem such as fear, hyperactivity, aggression (and other related problems), separation anxiety, biting, excessive barking, insecurity, destructive behaviors, walking difficulties, and inappropriate elimination. This type of training would normally be undertaken where the problem naturally occurs rather than a class situation. Class training can be effective in encouraging socialization and play with a peer group.
Specialized training
Dogs are also trained for special purposes such as CGC Certification; for dog sports, including but not limited to competition obedience, dog agility, herding, tracking, and flyball; and to undertake particular roles such as detection dogs, assistance dogs, hunting dogs, police dogs, SAR (search and rescue dogs) or guard dogs.
Tools
Source of the article : Wikipedia
EmoticonEmoticon