DOG TRAINING METHODS
Introduction
There are very few ideas that people agree on: the Maple Leafs will never win the Stanley Cup, cold beer is better than warm, and Tim Hortons will always be Canadian seem to be some of the few concepts that are. When it relates to animals, there are even fewer topics that enjoy a consensus. Everything from dogs vs. cats to the ethics of euthanasia is debated. However, one topic tends to bring about more debate than others. The ‘right’ way to train a dog has never been agreed on, instead fueling considerable debate, with both sides claiming that science supports their stance. If both sides claim to be backed by science, which one is correct? The purpose of this paper is to analyze several different training methods in terms of effectiveness, impact on the dog, and the general opinion of society.
The four main methods that will be evaluated are positive reinforcement, clicker training, balanced training and dominance theory. These methods are actually umbrella terms, encompassing many different techniques, and as many tactics as there are trainers. To avoid confusion, the general approach and ideas will be taken into consideration without the individual nuances of each trainer. As one of the newer models, the first method, positive reinforcement, is based on learning theory (Fukuzawa & Hayashi 2012). Notably, many also consider clicker training to be part of positive reinforcement as it follows many of the same ideas. However, based on the definitions of this paper, clicker training will be categorized as its own method. This is due to the learning theory quadrants that are utilized in the training process. While positive reinforcement actually works in two quadrants, clicker training only works in one (Fukuzawa & Hayashi 2012). The third method, balanced training, is also built on the concept of learning theory, though the approach is vastly different. As with many older training and discipline methods, even outside the realm of dog training, balanced training makes use of both reinforcement and punishment, though some techniques have a larger focus on punishment than reinforcement (Schalke, Salgirli, Bӧhm, & Hackbarth, 2009). The final method, dominance training, is the only method that does not follow the model of learning theory. Instead, it follows the dominance theory of wolf pack structure developed by Rudolph Schenkel in the 1940s (John, 2010). This method depends on resource control and respect or fear for the leader of the pack. Though balanced training and positive reinforcement both follow learning theory, they could not be more different. With this mind, it is critical to assess the methods by means of more than which theories they are based on, but also how each idea is executed.
First Method : Positive Reinforcement
Positive reinforcement training is not simply one type of training, but an umbrella term
encompassing many methods. The key factors that identify which methods qualify as positive
reinforcement are the quadrants of learning theory that the training takes place in. These
quadrants are labeled as either positive or negative and reinforcement or punishment (Fukuzawa
& Hayashi, 2012). The term ‘positive’ does not necessarily mean ‘good’, but the addition of a
stimulus. This could be food, an electric shock, praise, or anything added to the dog’s
environment. Like ‘positive’, the term ‘negative’ does not necessarily mean something aversive,
but a stimulus is removed from the dog’s environment.
These two qualifiers are paired with ‘reinforcement’ and ‘punishment’ to make up the four quadrants. Reinforcement is simply describing an action that increases the likelihood of the behaviour reoccurring, while punishment does the opposite; decreasing the likelihood of the behaviour. (Deldaalle & Gaunet, 2013) The four quadrants of learning theory are positive reinforcement (adding something to increase the behaviour), positive punishment (adding something to decrease the behaviour), negative reinforcement (subtracting something to increase the behaviour), and negative punishment (subtracting something to decrease the behaviour). Positive reinforcement training actually does not only stick to one quadrant, but operates in two quadrants- positive reinforcement and negative punishment (Fukuzawa & Hayashi, 2012). As negative punishment is the removal of the stimulus to decrease the chance of the behaviour reoccurring, it is considered a passive or non-invasive punishment (Fukuzawa & Hayashi, 2012).
This is often in the form of the trainer stress signals when taught two basic behaviours (Deldalle & Gaunet, 2014). Overall, dogs trained using positive reinforcement methods are more engaged with their owner, show fewer stress signals, and learn more effectively than when punitive techniques are used (Deldalle & Gaunet, 2014; Herron, Schofer, & Reisner, 2009; Hiby, Rooney, & Bradshaw, 2004). Although the vast majority of positive reinforcement training is centered on learning theory, a new method is in development that focuses on the concept of social learning.
Social learning is newly applied to behaviour in domestic animals, previously having only being researched in wild animals and children. Animal behaviourists in Hungary, however, have applied these theories to develop a training method called modeling. Modeling is based on the concept that animals learn by watching another animal perform a behaviour. For example, a young female cheetah learns to hunt by watching her mother and copying her actions. This is also how children learn to behave, by copying those around them. The Hungarian behaviourists have taken this concept to the next level; they have tested to see if the theory still applies across species.
To teach a behaviour via modeling, the handler first performs the action, such as ringing a bell or stepping up on a stool, and then encourages the dog to try once given the cue. According to a study performed by Fugazza and Miklosi (2015), modeling allows dogs to learn faster, and retain the behaviours longer than other methods. Not only is modeling effective for teaching cues, but it also helps the dog generalize the behaviour. Generalization is the dog’s ability to perform the behaviour in a variety of different contexts. When asked to perform the new behaviour in a different context, 78% of the dogs taught using modeling were able to perform on cue. This is significantly higher than the dogs taught using another form of positive reinforcement training, with only 9% being able to perform. (Fugazza & Miklosi, 2015)
There have also been many attempts to see if dogs will learn from each other. While social learning originally occurred within the same species, trainers did not know what would happen if dogs were to try and learn from each other when a handler was involved. Scandurra, Mongillo, Marinellli, Aria, and D’Aniello (2015) performed a study on 50 Labrador Retrievers to determine whether watching another dog performing a new behaviour would make them easier to train. The benefits of social learning were again proven when 62.5% of the dogs that watched were able to perform the behaviours, compared to 23.5% of the control group. The most intriguing find, however, was not the fact that dogs were able to learn from others, but that they had to learn this trait to begin with. The dogs that were older were better able to model the behaviour of the demonstration dog (Scandurra et al., 2015), indicating that social learning is not inherently known, but a developed characteristic. This is solidified by the findings of Fugazza and Miklosi (2015), which observed a relatively steep learning curve when teaching the dogs to model human behaviour.
Second Method : Clicker Training and Shaping
Another method was developed along the lines of positive reinforcement training, though it differs in respect to which quadrants of learning theory it utilizes. Clicker training and shaping access only one quadrant, positive reinforcement (Smith & Davis, 2007). Despite the lack of negative punishment, clicker training and shaping are often labeled in the positive reinforcement training camp because the methods differ significantly from ‘balanced’ training, which will be discussed later. Clicker training was originally not developed for use on domestic animals, but for training marine animals to perform tricks (Smith & Davis, 2008).
When training dolphins, whales, and sea lions for shows, traditional training methods were extremely difficult due to the distance the animals had to work at. Timing is crucial in training as animals associate a reward or correction with the behaviour that immediately preceded it (Yamamoto, Kikusui, & Ohta, 2009). This made rewarding from the sidelines for behaviours preformed in the middle of the pool impossible, and it was dangerous and illogical for trainers to be in the water while training the animals to properly time rewards, so an alternative method was developed. The goal of developing clicker training was to find a way to allow trainers to reward safely from the sidelines for a behaviour performed at a distance in the middle of the pool. To do this, a marker was developed. A marker is a word or noise that allows the animal to know which behaviour is being rewarded (Smith & Davis, 2008). Initially, trainers would use a word such as ‘yes’, but during shows their words became drowned out over the cheering of the crowds. To counteract this, they started to use a clicker or whistle, spurring the development of clicker training. When used on dogs, clicker training is extremely effective. Studies have shown that clicker training increases a behaviour’s resistance to extinction, meaning the dog forgets it more slowly after the rewards have stopped. Compared to positive reinforcement, behaviours trained using clicker training are more resistant to extinction (Smith & Davis, 2008). This means that when rewards are completely eliminated, the dog continues to perform the behaviour for a longer period of time.
Despite these findings, there was little evidence that clicker training facilitated the actual training process, with many studies actually finding no correlation between the use of a clicker and any difference in training (Smith & Davis, 2008; McCall & Burgin, 2002; Williams et al., 2004). Although clicker training does not make training more effective by acting as a marker, it did help to improve the handler’s timing. A study by Yamamoto, Kikusui, and Ohta (2009) revealed just how important timing is on the training process. The response to the command decreased drastically when the handler was delayed only one to two seconds. Not only did dogs take longer to respond to commands, but they also required more commands to finally perform the behaviour. A clicker helps to decrease any delay the handler may have, inadvertently affecting the training (Yamamoto, Kikusui, & Ohta, 2009). In this way, using a clicker in training can have a positive impact on training through impacting the timing, without having a direct impact on the training itself. Another method, shaping, is often used in conjunction with clicker training. Shaping involves building the behaviour in small increments and ‘shaping’ it into the desired final behaviour (Smith & Davis, 2008).
For example, shaping a dog to learn how to spin would require the dog to first just turn its head, then turn its head and take a step, and finally building up to require the dog to make a full rotation. This is often combined with clicker training as the marker is valuable for the dog to know exactly which behaviour is being rewarded. During these training sessions, dogs have the tendency to offer several behaviours in an attempt to figure out how to get the reward, so the ability to point out the exact behaviour is crucial (Yamamoto, Kikusui, & Ohta, 2009).
Both shaping and clicker training operate according to the quadrants of learning theory, but shaping goes one step further to include operant conditioning. This type of conditioning was developed by Skinner, who taught pigeons novel tasks by asking for repetitions of a behaviour. These repetitions started off as a basic behaviour that built in complexity and precision until the desired task was achieved (Blackwell, Bodnariu, Tyson, Bradshaw & Casey, 2010) By requiring problem solving skills, shaping has shown to decrease stress in kennel environments. This is a type of proactive coping style that some dogs learn to help handle the strain of a kennel (Blackwell et al., 2010). Stress still does have a negative impact on a dog’s ability to focus and learn, but dogs that underwent classical conditioning, or shaping, were more active in their kennels and had a lower stress level measured through cortisol (Blackwell et al., 2010). Many other studies have shown similar results to those performed by Yamamoto, Kikusui, and Ohta, (2009), Blackwell et al. (2010), and Smith and Davis (2008). This includes research performed by at the University of Copenhagen used beagles to assess the impact of shaping and frequency of training (Demant, Ladewig, Balsby, and Dabelsteen, 2011; Meyer & Ladewig, 2007).
In addition, Marschark and Baenninger (2002) found that training herding dogs requires the use of operant conditioning and shaping, with other methods causing unsubstantial results. Though this may be true, the type of shaping used by Marschark and Baenninger (2002) includes negative punishment, so it fits more into the category of positive reinforcement than clicker training within the definitions of this paper.
Third Method : Balanced Training
Most commonly, the negative stimulus is social pressure, such as intense eye contact or taking up the dog’s space. Positive punishment is the quadrant that is most strongly contested, though what is determined to be a positive punishment is very dog specific. Tools such as e-collars, prong collars, and choke chains, as well as techniques such as ear pinching, leash corrections, shouting, and physical manipulation.
The general term for these tools and methods is ‘aversive’, which includes any tool or method used outside the domain of positive reinforcement. Even with the clear definition, it is difficult to define what tools are aversive, as each dog reacts differently to each tool (Haug, Beaver, & Longnecker, 2002; Ogburn, Crouse, Martin, & Houpt, 1998). Notably, though negative reinforcement is less common it is still used. Deldalle and Gaunet (2013) explored how dogs responded to either positive or negative reinforcement. They found that dogs trained with negative reinforcement showed significantly more signs of stress with 65% showing at least one symptom, while only 6% of those trained using positive showed any sign. Unfortunately, this style of training also has a negative impact on the relationship between the dog and the owner. The same study revealed that in addition to stress signals, dogs trained by means of negative reinforcement were notably less likely to gaze toward their owner, which was the measure used to determine engagement in the relationship (Deldalle & Gaunet, 2013). Much like positive reinforcement training, balanced training also has smaller sections that are technically within the same guiding principles, but follow much stricter criteria.
Modeling, one division of positive reinforcement training, is quite new, but there is a division of balanced training that is quite a bit older, and is, in face, barely used anymore. Koehler training is often referred to as the “yank and crank” (Frawley, 2015) method due it the focus on punishment. According to Koehler’s book, The Koehler Method of Guard Dog Training, a puppy should be corrected within five days of being home if it is still making noise, even if this requires dragging them out from a hiding place. Unlike many other types of training where corrections are based on noise, they Koehler method recommends the owner “give [the dog] a good shake and sting his upper thighs with a bit of folded belt” (Koehler, 1980, p. 63). The idea of this method is to force the dog to perform the desired behaviour out of physical necessity or fear of the consequences, not desire to please. The methods used in balanced and Koehler training can have some negative side effects on the dogs, especially when attempted by dog owners without the aid of a professional.
A study by Herron, Schofer, and Reisner (2009) discovered that direct confrontation caused a significant amount of aggression, with hitting or kicking the dog (43% of dogs reacted) forcing the dog’s mouth open (38% of dogs reacted), and muzzling the dog (36% of dogs reacted) having the largest negative impact. Other aversive methods that were considered non-confrontational were also tested, but the results are quite similar; growling (40% of dogs reacted) and staring down the dog (30% of dogs reacted) still caused aggressive reactions (Herron et al., 2009). In contrast, non-aversive methods caused very little aggression. Most methods resulted in no aggressive behaviour; the highest rate of aggression was caused by using food to trade which resulted in 6% of dogs reacting (Herron et al., 2009). Many other studies confirm that punishment has a negative impact on training (Rooney & Cowan, 2011; Michael, Korzilius, Felling, & Vossen, 2015), but the methods are still commonly used.
Punishment doesn’t have to come from the hands of the handler; it can be applied through tools as well. This includes a tool commonly applied with both the balanced and Koehler techniques: the e-collar. E-collars are a device that is fitted to the dog’s neck that is able to discharge various levels of electric shock under the control of the handler (Cooper, Cracknell, Hardiman, Wright, & Mills, 2014)Many behaviourists are attempting to determine how e-collars impact the training process and the well-being of the dog, including Cooper et al. (2014). This study found that using an e-collar to teach recall, one of the most commonly taught behaviours with this device, was no more effective than using positive reinforcement training, as defined previously (Cooper et al., 2014). In addition, the dogs trained with the static shock exhibited many stress signals such as yawning and lip licking that were not present in positive reinforcement group. The validity of these results is solidified by the many other studies that produced similar findings (Schilder & Van Der Borg, 2003; Riepl, 2015; Overall, 2007; Blackwell, Bolster, Richhards, Loftus, & Casey, 2012; Steiss, Schaffer, Ahmad, & Voith, 2007; Schalke, Stichnoth, Ott, & Jones-Baade, 2007).
Fourth Method :Dominance Training
According to Greenebaum (2010), dominance theory is incorrectly applied to training. This is because dominance is a relationship formed to control resources, while the behaviours taught with this method, such as recall, sit, down, and other basic obedience, do not impact resource control. An article by Westgarth (2014) agrees that dominance training in dogs is ineffective. The paper goes on to explain that the societal structure that is perceived when assessing a human-dog relationship in this light is likely a construction of the envisioned structure (Westgarth, 2014). This means that it is unlikely that the dominant-subordinate relationship exists between dogs and humans as expected by dominance trainers, but they are simply interpreting behaviours to fit their predetermined model.
As Van Kerkhove (2004) explains, even if alpha theory held true for wolf packs, it would still be unlikely that it would apply to dog training as intended. Dogs are so far removed from wolves that, when left to their own devices, do not form strict packs. Instead, they form loose groups that are easily formed and dissolved in an unstructured manner (Kerkhove, 2004). The lack of structure can be associated with the scavenging habits of feral dogs. Wolves are required to for packs to hunt down large prey, without which, a pack structure is quite unnecessary (Kerkhove, 2004). Unlike wolves, feral dogs are not required to take down large prey, instead scavenging for scraps. If wolves were required to hunt smaller prey, would the pack structure disband? Based on the Ethiopian wolf, wolf packs are solid and resilient no matter the prey. These wolves live only in the grassy highlands of Ethiopia, separated from all large prey (Marino & Sillero-Zubiri, 2013). This leaves the wolves to hunt small rodents, using tactics that are not seen in other wolf species (Marino & Sillero-Zubiri, 2013). In contrast, the social structure of coyotes is flexible in relation to the size of the prey (Bowen, 1980). This demonstrates that feral dog behaviour is more closely related to that of coyotes, and not wolves, further solidifying lack of relationship between wolf pack structure and dog training.
Prevalence in Today’s Society
The effectiveness and ease of use of different training methods is only applicable if the public knows about them. To promote their own theories, trainers often teach classes, write books, host seminars, and build trainer development programs. However, this does not necessarily mean that a method will become well-known. To create a large user base of certain methods, there are famous trainers who are the focus of television shows, such as Pit Bulls and Paroles, Alpha Dogs, Dogs in the City, and SuperFetch. One well-known positive reinforcement trainer is Victoria Stilwell, who helps owners with troublesome dogs on the television show It’s Me or the Dog. Since 2005, the show has been wildly successful, originating as a British show, but quickly expanding to include an American version featuring the same trainer (Positively, n.d.). Stilwell is not the only name in positive reinforcement training, though she is the only one with a television show. Other popular trainers include Pat Miller, Patricia McConnell, and Jean Donaldson. With 11 books, 4 DVDs, and a trainer internship program, Miller is arguably one of the largest positive reinforcement trainers (Peaceable Paws, n.d.). The successes of these trainers shows the vast reach of positive reinforcement trainers, and their impact is growing every day. Within the positive reinforcement camp, modeling is still under development, and as such has not gained any traction with the public.
Similar to positive reinforcement, there is also a leading trainer in clicker training. Though Karen Pryor does not have her own show, she does own and operate a successful school for trainers, as well producing over 10 very successful books and DVDs (Clicker Training, n.d.). Part of Pryor’s success revolves around her involvement with the development of the method in her time working with marine life (Clicker Training, n.d.). The method has become popular enough to support a yearly conference, the Clicker Expo, in three locations across the United States. Though clicker training is quite well-known within the dog training community, there are no major television shows or books that promote the method, keeping it somewhat out of the grasp of the general public.
One of the most controversial television dog trainers is Brad Pattison, personality of the show At the End of My Leash, and more recently Puppy SOS. In one of Pattison’s books, Brad Pattison’s Puppy Book, he describes often violent training techniques to train young puppies behaviours such as sit, manners, and even bathroom training. He openly bashes positive reinforcement training while encouraging methods similar to those developed by Koehler, that even include hitting the puppy to teach it to respect space (Elegans, 2013). Though Pattison is no longer very popular due to the controversy surrounding his methods, balanced training is still extremely popular. This is the preferred method of Leerburg Training, a large facility that hosts several online courses as well as producing several instructional books and DVDs (Frawley, 2015).
Cesar Millan is probably the most well-known dog trainer, earning the title the ‘Dog Whisperer’. Cesar employs alpha training methods on his popular television programs Cesar 911 and Dog Whisperer, as well as in his books and DVDs, and instructional manuals. Millan’s first show, Dog Whisperer with Cesar Millan, was National Geographic’s top-rated original series from 2004-2010, making it the first show of its kind to achieve this type of status (PR Newswire, 2010). Today, Millan remains extremely popular, influencing a large section of the general public despite some controversy that has recently developed. (Cesar 911, n.d.)
Despite how many popular trainers are showcased on television, many people still prefer to learn from books, making it important to assess the training methods described in common training manuals. As this is likely what a large portion of dog owners are basing their methods on, it is important to determine which techniques the most common books promote. Considering only the books that do not openly state their methodologies on the cover, a relatively unbiased opinion can be formed on the popularity of each method. For example, Secrets of a Dog Trainer: Positive Problem Solving for a Well-Behaved Dog by Victoria Schade would be excluded from this analysis as it clearly promotes one method on the cover. The reason for avoiding this is to prevent the inclusion of books that would be picked by owners looking for a specific method. It also helps to prevent the selection of more books in one field, which would create an unintentional bias.
The first book, New Dog by Dr. Bruce Fogle, describes a way to use training as a part of a larger comprehensive plan for the long-term health of your dog, and to improve the dog-human relationship. Using advice from trainer Patricia Holden White, New Dog promotes positive, fun training methods that are easy for the whole family to part take in. Included in the book is a description of clicker training for teaching advanced behaviours, though the author also cautions that a trainer should be consulted to ensure proper techniques are used (Fogle, 2008). New Dog clearly promotes positive reinforcement training, extending the idea to include every aspect of the dog’s lifestyle. Alongside positive reinforcement, New Dog also mentions clicker training, though it does not describe the process in detail (Fogle, 2008).
An older manual, The Encyclopedia of Dog Breeds (Cunliffe, 1999) supports both balanced and dominance training. This is shown through the description of how to properly use corrections, or positive punishment, in the training process, and the justification for tools such as choke collars to teach heeling (Cunliffe, 1999). The dominance method is mentioned in both the obedience and social structure sections of the book, though it is not called as such. Dominance training methods are distinguished from balanced training by the mention of scruff shakes, alpha rolls, and the need for a dominant member to take over the pack (Cunliffe, 1999). Similar to the idea of behind the encyclopedia, another common book series that people turn to when attempting to learn something is the For Dummies series. Much like the previous encyclopedia, Dogs for Dummies (Spadafori, 1996) encourages the use of balanced training methods, though it does not mention any aspect of dominance methods. As these two books are quite a bit older, it is not surprising that their methods fall more toward balanced training as that was the most common during that time period. Nevertheless, many of these books are still commonly used, suggesting that the methods they communicate are likely still being adopted by dog owners.
Overall, it appears each training method has more prevalence in certain areas. Positive reinforcement training has more opportunities to teach trainers, showing that the method is still developing and may one day become the most common method. Within positive reinforcement, modeling is still in the early stages of formation, leaving the concepts to be discovered by interested trainers, and not yet impacting the public. Clicker training is another popular method, and the only one having a yearly conference series that focus on solely the one method. Outside of the world of trainers, clicker training is still relatively unknown, being mentioned in only one of the books analyzed. Balanced training is still the leading method in terms of access to literature. The vast majority of the books promoted training using corrections without the use of dominance, which is distinctive of the balanced method. Finally, in terms of famous trainers, dominance training is the most prominent method as Cesar Millan is one of the most well-known trainers.
Conclusion
The purpose of this paper was to analyze four different training methods in terms of effectiveness, impact on the dog’s welfare, and prevalence in modern society. The first method, positive reinforcement, operated in two of the quadrants of learning theory (Fukusawaa & Hayashi, 2012). This method produced the least amount of stress (Herron, Schofer, & Reisner, 2009), and has the largest impact in terms of schools available for trainers. The second method, clicker training, is still a relatively new method. While it did prove to help with timing, there was little evidence that use of a clicker improved the training process to any significant degree (Smith & Davis, 2007). Despite this, clicker training is the only method studied that has a yearly conference series in support of that method alone. Balanced training also follows learning theory, but it operates under all four quadrants, unlike positive reinforcement. Despite research repeatedly publishing that punishment has a negative impact on the welfare of the dog and the effectiveness of training (Rooney & Cowan, 2011; Michael, Korzilius, Felling, & Vossen, 2015), it is still one of the most common methods presented in literature. Similarly, the dominance method, based on the outdated approach developed by Rudolph Schenkel (John, 2010), had very little scientific support. Despite this, it remains of the most popular methods, promoted by television personality Cesar Millan. In conclusion, the most humane and effective method of dog training is positive reinforcement, though this is not reflected in its popularity in modern society. Further research should be conducted on how to improve public understanding of each training method, and why different methods are popular outside of the science.
