Punishment, Puppies, and Science: Bringing Dog Training to Heel

Three years ago, Valli Fraser-Celin adopted a blonde husky mix puppy, whom she named Husk. Fraser-Celin soon started looking for ways to curb Husk’s “totally wild” behavior, she said, like stealing food from the kitchen counter and barking incessantly at strangers. Based on the advice of a YouTube trainer, Fraser-Celin started using an electronic collar, or e-collar, that delivered a small shock when Husk misbehaved, but said she felt “yucky” about it.

Related

Did That Fox Just Wag Its Tail?

Fraser-Celin rethought her approach after hearing about an animal trainer who taught a grizzly bear to cooperate with medical treatment using only positive reinforcement. If that hulking animal could learn with treats and praise, she thought, why were dog trainers using prong and shock collars? “That was the catalyst into my advocacy,” said Fraser-Celin, who studied African wild dogs for her Ph.D. and now works as a remote community liaison for the Winnipeg Humane Society and advocates independently for positive reinforcement training on Instagram. “I really think that there needs to be regulations that are put into place,” she said, “based on the science and the studies that have shown the best type of training for dogs.”

Fraser-Celin is not alone. Many researchers, trainers, and veterinary and training professional organizations are advocating for greater oversight for dog training, which is largely unregulated worldwide — though they sometimes disagree on the best path of action and choose to focus on the research that reinforces their preferred approach. “Right now, it’s the wild, wild West,” said Anamarie Johnson, a psychology Ph.D. student at Arizona State University with a background in animal behavior and dog training. She recently published a study that analyzed the websites of 100 highly-rated dog trainers across the U.S., which found that most gave no indication whether the trainer had relevant education or certification.

“Anyone can identify as a dog trainer — they can put up a social media page, they can offer services to the public, and there’s no expectations for their education, their continuing education, or their standards of practice,” said Bradley Phifer, the executive director of the Certification Council for Professional Dog Trainers, or CCPDT, an organization promoting science-based training standards. People with little or no education in animal behavior may be advising owners on handling aggression, he added. “There’s a big consumer protection piece here, that if you’re not adequately trained, or you don’t have adequate experience in the industry or in the content, then you shouldn’t be advising people on how to prevent dog bites.”

Some experts and organizations are pushing for greater regulation of the industry. Under an umbrella organization known as the Alliance for Professionalism in Dog Training, two major certification bodies — the CCPDT and the Association of Professional Dog Trainers, or APDT — have jointly proposed model legislation that they hope could be adopted on a state-by-state basis. The legislation would require trainer licensure by a state board, create accountability standards, and require trainers to engage in continued education. Phifer said he’s currently working with legislators in New Jersey, where regulations for dog trainers were first proposed in 2019, and that the joint effort is also making progress in California and Illinois.

But the push for regulation has exposed a schism in the industry over using punishments versus rewards. Under the proposed legislation, certifying bodies would be required to uphold a policy that prioritizes positive reinforcement, though does not entirely rule out punishment — an approach generally backed by research on efficacy and welfare and increasingly popular among training professionals. While researchers and trainers largely agree that punishment-heavy approaches are harmful, they are at odds whether all-out bans on aversive tools are productive, since the approach may work in limited circumstances.

Without clearer rules, the broad gaps in dog training pose “a potentially very large safety risk to the public,” said Johnson, because dog owners are trusting trainers to modify the behavior of animals with “sharp, pointy teeth that live in our house.”

Modern dog training is rooted in the mid-20th-century work of American psychologist B.F. Skinner, who suggested four categories for behavior modification: positive reinforcement, positive punishment, negative reinforcement, and negative punishment. Here, positive and negative don’t necessarily mean good or bad. Positive reinforcement adds something a dog likes to reinforce a behavior, such as a treat or a toy for sitting on cue, while positive punishment adds something aversive, like a tug on a leash, to decrease a behavior. Negative reinforcement removes something the dog dislikes, such as stopping a shock collar when a dog obeys a command, while negative punishment removes something desirable, such as facing away from a dog that is jumping for attention.

Many trainers and animal behavior experts say that aversive methods, which include positive punishment and negative reinforcement, are overused. Two major professional organizations that represent trainers — the APDT and the International Association of Animal Behavior Consultants — now limit the use of tools like e-collars among their members.

In October last year, the American Veterinary Society of Animal Behavior, which includes both veterinarians and behaviorists with doctorate-level education in animal behavior, issued a statement: “There is no evidence that aversive training is necessary for dog training or behavior modification,” referencing 21 studies on the effectiveness of reward-based methods and risks of aversive methods. Alexandra Protopopova, an animal welfare researcher at The University of British Columbia, wrote in an email to Undark that the recent research cited by the statement reflected the “undeniable” risks of aversive techniques, adding: “Ultimately, recent research has also shown that aversive methods do not result in better trained dogs; thereby making traditional aversive dog training methods obsolete.”

The research has raised concerns about dog welfare. In one small study, dogs trained with rewards appeared to be more playful and better at learning a novel behavior than dogs whose owners reported using punishment. In another, dogs reportedly trained with aversive tools were, as the researchers put it, more “pessimistic” than dogs that were not, based on their hesitation in approaching a bowl of food. Some evidence also suggests that use of punishment in training can diminish the bond between a dog owner and their canine.

A 2017 literature review confirmed that, overall, there are welfare risks associated with positive punishment. But the review also noted limitations across the available research. One weakness: Many studies rely on surveys of owners to determine how dogs are treated, making it hard to objectively assess the effects of training methods. Surveyed owners might, for instance, vary in how they define punishment. Those studies are also largely correlational, connecting the self-reported treatment of dogs to their (also self-reported) behavior.

Not all studies share that limitation, including a government-sponsored study in England that directly compared two training approaches. Researchers at the University of Lincoln recruited two types of trainers: Those recommended by e-collar manufacturers and those who use positive reinforcement. The trainers worked with 93 dogs that had trouble responding to their owners when they were called, instead choosing to chase livestock, run after other dogs, or just simply ignore any plea to come.

At the end of the trial, owners of dogs from both groups were satisfied with the results — more than 90 percent reported to researchers they saw improvement in their dog’s recall. However, the researchers also noted more signs of stress in the e-collar group, including yawning and, in some dogs, yelping. In a second study reviewing videos made during the initial trial, the team found that the dogs trained using positive reinforcement had faster response times.

“Anyone can identify as a dog trainer — they can put up a social media page, they can offer services to the public, and there’s no expectations for their education,” said Bradley Phifer, the executive director of the Certification Council for Professional Dog Trainers. Visual: fotografixx/Getty Images

Some have critiqued the work, however. Rebecca Sargisson, a psychologist at the University of Waikato in New Zealand, published a commentary of the second study with a co-author that called into question its methods and conclusions. The recall tests, she noted, were mostly performed with dogs on a long leash, which didn’t necessarily show how they’d behave off-leash, and the researchers didn’t measure baseline performance. Sargisson also expressed concerns on total bans for e-collars. In New Zealand, for instance, the devices are successfully used to teach hunting dogs to stay away from Kiwis, the endangered, flightless national bird. But, she added, e-collars still shouldn’t be readily available online and in pet stores. 

Jonathan Cooper, a coauthor on the training comparison studies, said that in instances where endangered species are on the line, the critique makes a “fair point” and his research team published a detailed response to the commentary. Some positive reinforcement trainers note additional risks in using shock collars. Kat Camplin, a dog trainer based in Redding, California, said that she’s worked with dogs that had been through rattlesnake training — where dogs are shocked after sniffing a snake — that had also become terrified by things that aren’t snakes, like garden hoses.

Still, some dog trainers aren’t convinced by the growing body of research. Ralf Weber, a trainer based in Southern California who uses both e-collars and play-based rewards, is skeptical of the conclusions positive reinforcement supporters draw from the literature. He said that they are cherry-picking findings to support their position, adding that learning can be stressful for dogs regardless of the method (he pointed to one controversial study where a verbal cue indicating a dog didn’t perform as desired was associated with higher cortisol levels than an e-collar shock). While Weber acknowledges that e-collar misuse is rife — “I can go to YouTube and find hundreds of videos of people who shouldn’t be allowed to have these, abusing dogs left and right” — he argues that they are a valuable tool in certain cases, like stopping a dog from chasing wildlife. 

It’s hard to design a perfect study to test the two approaches, Johnson said. Researchers would need to recruit skilled trainers, ensure the training approaches were applied consistently, and control for differences in dog abilities and personalities — a massive undertaking.

“The key thing that I try to be mindful of with my scientific understanding is that I can recognize that punishment does work,” Johnson said. The issue, she added, is that by the time an owner applies that approach “it’s so convoluted and diluted,” it could ultimately harm the dog.

While researchers and trainers debate what to take away from the welfare studies, advocates continue to push for regulation. Phifer said that if one or two states adopt the Alliance for Professionalism in Dog Training’s model legislation, it will likely be easier for others to follow suit.

Still, even among professional organizations, there’s argument over what’s the best way to protect dogs and consumers. If you were to ask 100 dog trainers, you would “get 100 opinions on where the line should be,” said Benjamin Bennink, a dog trainer and vice chair of the Association of Professional Dog Trainers. But, he added, there should be some form of regulation: “You wouldn’t go to an unlicensed dentist, you wouldn’t even get an unlicensed plumber or electrician.”

Some argue that going straight from no regulation to licensure requirements is too great a leap. Kathrine Christ, executive director of the International Association of Animal Behavior Consultants, said that starting with limited regulations that enhance accountability for trainers would be preferable to licensure requirements. “We’re not necessarily ready to take a step towards promoting licensure or promoting, you know, intrusive types of regulation, before we can tell you that A, it’s worth the money for the taxpayer or B, it’s worth the money and cost for the people and the profession,” she said. As evidence, she shared a Brookings Institution paper that found that occupational licensing in other professions provided little benefit in terms of service quality and safety.

Weber said that the dog training industry is currently “an unmitigated shit-show” and he supports licensing, but worries about how to do it without creating a “bigger problem down the line.” Limits on tools like e-collars and prong collars, he added, have had unintended consequences in some countries. In Germany, for instance, a ban on a type of collar forced law enforcement to pull police dogs, which were trained to respond to the collars, off the streets, while dog sport trainers in Finland have turned to more severe punishment tools such as cattle prods. 

Instead of outright bans, Weber and Johnson both propose a different starting point: basic education for dog trainers. Weber said that Australia may be a good model, where a nationally-recognized certification requires trainers to take courses in behavior science.

As for bridging divides in the training community, Bennink said that organizations can put out all the data and position statements they want, but at the end of the day, trainers simply need to show what they can do. “If I’m literally showing you this dog can now do this,” he said of teaching a new behavior, “that’s going to convince more people than, unfortunately, any amount of scientific data.”

UPDATE: Due to an editing error, Jonathan Cooper was only referred to by his last name in the original version of this piece. His reference has been updated.

Ula Chrobak is a freelance science writer based in Nevada. You can find more of her work at her website: https://www.ulachrobak.com/