[3GSD4IPO]

Read the article. Links and references are included (along with the original referenced article).

Commentary: Efficacy of Dog Training With and Without Remote Electronic Collars vs. a Focus on Positive Reinforcement

In an experimental analysis of the effectiveness of e-collars, China et al. (2020) concluded that "there is no evidence to indicate that E-collar training is necessary" (p. 1). The paper contributes to a wider body of research on the use of e-collars for dog training, much of which is referenced...

www.frontiersin.org

 

[DaySleepers]

Critique and discussion of research is extremely important to the scientific process, so I appreciate that goal of this commentary.

That being said, after reading both the original paper and the commentary, I find that a large portion of the commentary focuses on saying that China et al. didn't study the effects of aversion training... which is true, because the China et al. study never intended to study aversion training (the commentary does acknowledge this). The study was examining the differences between training methods of 'come' and 'sit'. While I'm not arguing that aversion training should be part of the discussion of e-collar bans (it's the only legal use of e-collar training here in Norway, actually), I'm a little skeptical of using an e-collar technique not explored in the original study as justification for the rest of the commentary. If there's valid issues with the original study, the existence of aversion training with e-collars should be a moot point, IMO, or a side note rather than the introduction and justification.

...if delivery of shock was not noticeable in the dogs' reactions (implied if the data extraction were truly conducted under blind conditions), then the authors' conclusion that "dog training with these devices causes unnecessary suffering" (p. 9) is inappropriate.

They are right in that China et al. did not produce evidence in their study that e-collar training causes harm, and have a point in that the wording used was perhaps more emotionally charged than necessary. However, the passage quoted above fails to mention that China et al. made that claim not on their study alone - the data from which showed no significant advantage to e-collar training over other methods - but from multiple studies that explored potentially harmful effects of e-collar training combined with their study's conclusion that they found no increased efficacy with e-collar training. Full quote from China et al. below:

Given the better target behavior response parameters associated with a reward-focused training programme, and the finding that the use of an E-collar did not create a greater deterrent for disobedience; we conclude that an E-collar is unnecessary for effective recall training. Given the additional potential risks to the animal's well-being associated with use of an E-collar (7, 25, 31, 38, 39), we conclude that dog training with these devices causes unnecessary suffering, due to the increased risk of a dog's well-being is compromised through their use, without good evidence of improved outcomes.

I just found the way they presented the claim without acknowledging the researcher's reference to other studies about e-collar use disingenuous. They possibly had valid ground to discuss China et al. appearing biased in the language they used or that making claims about the ethics of e-collar use is outside the scope of the study, but it weakens their argument when they don't take the full context into account.

But the part about using the industry standard correction (lowest level that the dog responds to) compromising training outcomes, that bothered me. Uncritically citing a study done in the 60s as their evidence that "that punishment is most effective when it is delivered at the maximum acceptable level of intensity (Azrin, 1960)" had me raising an eyebrow. Yes, single-event learning with intense negative experiences is EXTREMELY powerful, but that doesn't make it ideal for most learning situations, nor does it take into account the risks to the animal's mental well-being to use such extreme methods indiscriminately (the 1960 study on pigeons is available here, if anyone's curious: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1403961/pdf/jeabehav00201-0033.pdf). As an e-collar trainer @3GSD4IPO I'd appreciate your input on this one, and whether we need to throw out any studies that use less than maximum intensity with their e-collars because otherwise e-collar training is just ineffective(??). That doesn't seem right to me at all.

I also rolled my eyes hard at the 'frog in boiling water' analogy. If this were a pop sci piece I'd be less annoyed, but a thoroughly debunked myth has no place in a scientific journal even as allegory. I can accept that this is a personal pet peeve and has no bearing on the validity of the arguments, though.

Now for the part I'm on board with: China et al. showed no evidence that the original unwanted behavior of these dogs was successfully eliminated long-term, and that absolutely bears more research. It was out of scope for the original study, but it's absolutely data that is important if we want to thoroughly compare and contrast various training methods. I can't speak to the discussion on the study design, statistics, error correction, etc. though - it's been far too long since I've been actively looking at these parts of research for anything I half-remember to be reliable. But I agree that designing good data sampling and correcting for possible errors is an extremely difficult part of these kinds of experiment, where so many factors can't ethically or practically be controlled by the researcher. This is why we should never rely on a single study to make sweeping conclusion, especially with something as complex and hard to quantify as animal behavior and animal-human interaction. I wish they had gone into more detail about these flaws and their impact on the results - this seems like where the focus of their argument should've been to me.

I have no qualms with their conclusion, I just feel like a lot of their points are skewed and exaggerated to make their argument look stronger, but only succeed in weakening the more valid criticisms. That doesn't mean the China et al. article is flawless and above reproach by any means, and I see it as just one piece of a large body of research that needs to be critically examined and weighed to draw any overarching conclusions about the efficacy and ethics of training tools/techniques. In the end, I'm not convinced of why China et al. SHOULDN'T be used as part of determining e-collar legislation in conjunction with other research.

 

[parus]

I'm on my phone so a lengthy text, let alone lengthy texts I'd want to put side by side, aren't a realistic option at the sec, but I will point out that there is a fundamental problem with using examinations of efficacy as a primary method of making recommendations about best practices for systems involving living, feeling things: often the most efficacious methods of evoking desired behaviors are not ethical. This applies to everything from economics to animal training.

 

[ClanWorgen]

often the most efficacious methods of evoking desired behaviors are not ethical.

Ethics vary from person to person. Personally I cannot understand this belief many people have that a dog should never ever experience stress or pressure of any sort. Sometimes it takes some stress and pressure to grow. It's a balancing act between pushing a dog enough to improve while not deminishing their desire. Similar when training a dog in weight pull, if you push them too hard they will burn out, but if you push them to their limit that they can succeed it builds them and their desire.

 

[storyist]

Ethics vary from person to person. Personally I cannot understand this belief many people have that a dog should never ever experience stress or pressure of any sort.

In the situation we're considering, it's not just a matter of dog stress or not. There's the matter of the stress, pain, terror, and loss of life of the livestock that has to be factored in - emotional distress to people who care about those animals - monetary loss to farmers, ranchers, and those who pay vet bills. As someone who raised and loved horses for many years, I have to tell you, if a dog was in the pasture harassing my horses, or God forbid a foal, my concern for the dog would be south of zero.

The other thing I see all the time in these discussions is the assumption of idealistic human behavior that simply doesn't exist in the real world. So let's accept that reward-based training can produce a dog with high prey drive that doesn't act on that instinct. Let's also consider how many humans have the skill, time, desire, patience, etc., to actually do that training? How many careless, negligent, ignorant, arrogant people who let their dogs loose in situations where they can cause havoc will do that training? Aiming for and working toward the ideal is one thing. Refusing to accept reality is another.

I was interested to see in the article that countries that ban electric devices on dogs make an exception when it comes to protecting livestock. Sounds unusually reasonable for lawmakers to me.

 

[parus]

In the situation we're considering, it's not just a matter of dog stress or not. There's the matter of the stress, pain, terror, and loss of life of the livestock that has to be factored in - emotional distress to people who care about those animals - monetary loss to farmers, ranchers, and those who pay vet bills. As someone who raised and loved horses for many years, I have to tell you, if a dog was in the pasture harassing my horses, or God forbid a foal, my concern for the dog would be south of zero.

I grew up on a farm and our family business for decades was a riding stable. My job was the care and training of horses, our own and those of others. No one around there back then had ever heard of the concept of "positive" or "force free" training or had any qualms about being rough with an animal that they felt needed it. Even so, there was no use of shock collars on dogs to protect horses from dogs, because this isn't a real-world application of shock collars. A shock collar simply isn't a practical or foolproof tool for managing dogs' behavior around horses. Dogs that are a threat to horses aren't allowed on the property, full stop. Only trustworthy dogs are loose. Livestock before pets, and horses before other livestock.

I was interested to see in the article that countries that ban electric devices on dogs make an exception when it comes to protecting livestock. Sounds unusually reasonable for lawmakers to me.

It's other people's loose dogs, or feral dogs, that are a threat to livestock the vast majority of the time, not the dogs of the farmer/rancher or their visitors. So how would legal shock collars help protect their livestock?

I would and have used potentially lethal methods against loose dogs that were placing livestock at risk, so this isn't a bleeding heart thing. People who are pushing for shock collars to be legal for this purpose are either ignorant of the reality of farming, or they're looking for a loophole to make shock collars legal for sale to be put to other uses.

 

[ClanWorgen]

It is unfortunately a slippery slope and banning tools as a whole is pretty far along on said slope. In the states we are battling these laws in every county because it's one step closer to limited pet ownership as a whole. These bans that supposedly improve pets quality of life by eliminating risky dog jobs (hunting, livestock guarding, protection work) also piggy back laws that would prevent dog shows and sports. As with my breed I will always advocate for education rather than legislation because I do believe in freedom rather than putting citizens in padded rooms where they can't make a mistake.

The mistake is believing that supporting Ecollar training is being against reward based training or that Ecollar trainers do not use rewards in their training.

Unfortunately the studies done on positive and negative reinforcement are generally used on the typical pet, so in terms of high drive dogs they don't always address the topic. I can only say that I personally believe based on experience that when a dog is bred a certain way there is sometimes something they want more than anything else. I've seen dogs completely ignore the lightening bolt sent to their Ecollar to save their life because the cat they are after in the midst of traffic is far far more worth it. The catch and kill is the greatest reward of all for some dogs. Ecollars aren't the answer to all problems, they do nothing on my dog in the midst of a scuffle. This is also a breed (APBT) that is willing to rip out its own teeth breaking a fence between it and another animal. As for the argument that if given enough time and effort with no aversive tools this kind of dog could be in the vicinity of other animals and behave calmly, I just haven't seen it so it's hard for me to buy.

 

[parus]

Ethics vary from person to person.

Obviously. For example, in some places dogs are bred for meat. Do you think that should be legal in the US?

 

[parus]

I've seen dogs completely ignore the lightening bolt sent to their Ecollar to save their life because the cat they are after in the midst of traffic is far far more worth it. The catch and kill is the greatest reward of all for some dogs. Ecollars aren't the answer to all problems, they do nothing on my dog in the midst of a scuffle. This is also a breed (APBT) that is willing to rip out its own teeth breaking a fence between it and another animal. As for the argument that if given enough time and effort with no aversive tools this kind of dog could be in the vicinity of other animals and behave calmly, I just haven't seen it so it's hard for me to buy.

You just said that shock collars flatly don't work in these scenarios, so why are you advocating for their use in preference of at least attempting other methods that haven't been actively disproven?

 

[ClanWorgen]

Obviously. For example, in some places dogs are bred for meat. Do you think that should be legal in the US?

My issue has always been with the treatment of meat animals before slaughter, that goes for any animal from chickens to beef to horses to dogs. I am not going to fight cultural differences.

You just said that shock collars flatly don't work in these scenarios, so why are you advocating for their use in preference of at least attempting other methods that haven't been actively disproven?

Who says I haven't? I tried for YEARS with my GSD, but after having him pull a loose husky through our cattle fence breaking it in the process I had to figure something else out. They don't always work, but I have had more success with them. Proper management and enclosures is the only sure way to control animal aggression and prey drive at that level.

 

[DaySleepers]

To clarify my concerns about the commentary's claims that China et al. using e-collar industry standards (finding an individual dog's lowest effective stim level and using that) "compromised" training outcomes:

Their justification of this is citing a study that looked at pigeons in bare 16" cubes with nothing but the target (which previously had only given them food rewards and, at the time of the experimental sessions, also triggered shocks). By trying different punishment intensities and patterns, they determined that the most efficacious punishment was their 'very severe' intensity (15 milliamperes for birds between 450 and 340g), which produced almost immediate cessation of the behavior (pecking the target) and little to no recovery of the behavior once punishment was stopped. That's all well and good. But there's no attempt to examine short-term and long-term behavior changes in these birds unrelated to the experimental setup, nor does it say anything about how 'very severe' electrical punishers may function in real-world training scenarios.

What I'm saying is that in real-world dog training, we cannot target a behavior in a vacuum. We might claim that maximum punishment with the e-collar has the best efficacy because it's most successful at eliminating the target behavior, but efficacy in eliminating the target behavior means little if the technique causes other behavior problems. Trying to compare the China et al. study with dogs in outdoor settings around distractions and interacting with humans with a study on pigeons getting shocked in a bare box and saying that the pigeons prove that the dog trainers aren't using punishment efficiently is questionable logic at best. If Sargisson and Mclean wanted to argue that the trainers China et al. chose were using ineffective methods, despite being recommended by the Electronic Collar Manufacturers Association (ECMA) as experienced representatives of the best practices for e-collar use, the burden is on them to present us with far stronger proof than a six decade old study to justify why their methods were ineffective.

Doesn't have anything to do with 'not wanting dogs to experience stress', just that the argument is poor and willfully ignores why organizations like the ECMA have the standards for e-collar training that they do. The way this is written appears to be arguing that everyone should turn up their e-collars to the maximum safe correction level for every kind of training, including basic obedience like 'sit' which is what China et al. was looking at - since they claim the training in these studies was 'compromised' due to using minimum effective level instead - I can't imagine this is what anyone actually thinks is best and most effective for dogs.

Re: countries making an exception for aversion training with livestock - Norway, despite having banned multiple corrective training tools and being strongly reinforcement training oriented culturally, relies heavily on hunting to manage our large herbivore populations and keep them (and their habitats) healthy and balanced. Part of the ethical hunting guidelines here is the legal requirement for hunting parties have access to a trained and certified blood tracking dog, so that if an animal is only injured during a hunt, they have a better chance of being able to to find it and complete the kill as quickly as possible. E-collar aversion training with livestock is something I've only heard of with hunting dogs like these, who play an important role in ethical hunting practices and need to be off-lead or on a tracking line while working, and who are often working breeds that may be more likely to go after livestock if they come across a field. I suspect this is why the exception was made - Norway takes welfare issues surrounding its wildlife as seriously as domestic animal welfare, so this concession to make tracking dogs safer to work in areas with livestock seems reasonable. Only a few specific trainers have the certification to perform this aversion training, so it's not a free-for-all.

 

[3GSD4IPO]

I am enjoying the intellectualized discussion here. I HOPED that would happen instead of a huge angry match of words!

So, as most responders here know I DO sometimes train with an E Collar and I sometimes train with a prong collar. MOST of my training (and all teaching of new behaviors) is reward based starting with food (the drive is different.. a bit lower and allows for more thinking) and moving up to toys (to build speed and power) and sometimes stepping back to food. I work around mostly high drive dogs or with people who reward drive and use that drive to train. I will also say that I do not universally use e collars and prong collars on every dog I am asked to help with. In every real life situation it is important to READ the DOG. I do NOT RECOMMEND these tools be used by everyone. I also do NOT RECOMMEND BANS of tools (I am going to put up another post about that).

But the part about using the industry standard correction (lowest level that the dog responds to) compromising training outcomes, that bothered me. Uncritically citing a study done in the 60s as their evidence that "that punishment is most effective when it is delivered at the maximum acceptable level of intensity (Azrin, 1960)" had me raising an eyebrow. Yes, single-event learning with intense negative experiences is EXTREMELY powerful, but that doesn't make it ideal for most learning situations, nor does it take into account the risks to the animal's mental well-being to use such extreme methods indiscriminately (the 1960 study on pigeons is available here, if anyone's curious: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1403961/pdf/jeabehav00201-0033.pdf). As an e-collar trainer @3GSD4IPO I'd appreciate your input on this one, and whether we need to throw out any studies that use less than maximum intensity with their e-collars because otherwise e-collar training is just ineffective(??). That doesn't seem right to me at all.

First of all, we are discussing studies wherein attempts are made to isolate that which is being studied. When isolating for a single thing, attempts are made to eliminate other influences. Of course, that is nearly impossible to do and that is why any study also has to show repeatability (I think).

We have a study here of pigeons. Pigeons are not dogs. It is doubtful they have the retentive learning that dogs have (now that is a guess).

They do not discuss WHAT the maximum acceptable level of intensity is. In my experience this varies largely with a dog and with the situation. An experienced e collar user training things on a training field will want the intensity high enough to get a response and clarity but never so high as to make the dog vocalize. A correction needs to be enough or it is nagging and nagging doesn't work (with kids, spouses, siblings, dogs or horses!!). Too much and/or unfair and you create problems.

I can only offer examples here based on a dog that has been properly introduced to the e collar (a whole 'nother discussion).
In a low drive exercise the e collar level is much lower. On a scale of 1-100 on a double box with one dog it may be a 7 while on another dog it may be a 20.. and on some dogs you don't need it at all. In a high drive exercise it could be a 30 on one dog or a 60 on another dog. Some dogs the e collar isn't even acknowledged when they are in high drive. In a high drive situation with a dog that has had extensive training the e collar is a reminder of "you MUST listen even if you are in high drive." Gradually increasing stim is nagging. Pick a level based on knowing the dog. Use that level. If it is insufficient STOP. Do something else. NEXT training you know.. you have to use a higher level.

NO ONE who is knowledgeable wants to use too much stim especially in low drive situations. In high drive situations it is RARE that too much stim does damage to the future training of the dog.

And, I will add, NO ONE who is serious about training and uses an e collar (or a prong) wants to create detrimental psychological fall out wherein future behavior is damaged beyond repair (or is damaged, period). E collars are not typical equipment for teaching a new, cued behavior.

I am avoiding the statements or "always" or "never" because training is individualized and while the dog wants to see black and white in training, aspects of training dogs are rarely black and white.

I HOPE this answers the question?

As to the comments on aversion to chasing livestock that is called "Trash Breaking" or breaking the dog from chasing an unwanted target. It is a common practice on hounds that work off leash and go after target game. We also have, in my State, a "leashed tracking dog" license to track down wounded game. I do not work with hounds working game for which they were bred to hunt.

 

[DaySleepers]

That's about what I thought re: practical use of e-collars - thank you. It seems to align with the ECMA guidelines for best practice in most ways as well.

Yeah, I'm definitely not arguing that the 1960 pigeon study was bad research, just that the commentary authors are applying it poorly. They're pointing out that these ECMA-recommended trainers are finding the minimum effective correction level for the dogs they're working with. They then assume that this means:

A) This correction level is necessarily 'mild'. Minimum effective level, as you point out 3GSD, doesn't necessarily mean 'mild' depending on the needs of the dog and situation. As these are experienced e-collar trainers specifically recommended by the ECMA as the best representatives of ethical and effective e-collar training practices, it's foolish to assume they're using a corrective level that's so 'mild' as to be ineffective. Because they make this assumption, they also assume the corrections used by the dog trainers are equivalent to the corrections used in the 'mild corrections' section of the pigeon research.

B) That finding the minimum effective stimulation that works for each dog is a process that mirrors the pigeons who were subjected to an increasing punishment slowly over period of several days, hence why they make the 'frog in boiling water' analogy about the dogs getting habituated to the punishment (I might be making some assumptions of my own here because it's not explicitly stated, but I see no other reason to have brought up the 'increasing intensities' point in the commentary if they weren't addressing the process of finding the minimum correction level).

and

C) That success of a correction in dog training is solely based on how many repetitions required to eliminate the unwanted behavior and how likely that behavior is to reoccur without the corrections, eg the same criteria used to measure success in the pigeon experiment. Completely ignoring the possibility of short- and long-term behavioral fallout or taking the dog's mental state into account, and also ignoring that the ECMA trainers were also implementing rewards, making these two learning scenarios not directly comparable.

If they had used the 1960s study in conjunction with other research about punishment in dogs, the tandem use of punishment and reward, the effects of intense punishment on mental distress and behaviors, I wouldn't have taken much notice of the one study. But as it stands it reads like their argument is "we can't use this research to inform us about e-collar training's efficacy because these experienced trainers who follow the best practice standards for this tool, as defined by the ECMA, aren't doing it right, here's a decades old paper about pigeons in boxes to prove it". Which in the end feels like "e-collar bans are bad because e-collar trainers aren't using e-collars right". I don't think this is what they were TRYING to say, but their argument is so bad that this is how it comes off to me.

It doesn't help that they, as you pointed out, never define "maximum acceptable level", which is also not defined in the 1960s paper so they didn't get it from there, and so is essentially meaningless. Does this mean the maximum level that doesn't cause physical injury? The maximum level that doesn't cause behavioral fallout? How would one even determine that without running the risk of causing that fallout in the process?

Pigeons are incredibly smart though - I always think of the study where they learned to discriminate between paintings by different painters and generalized that to paintings they'd never seen before (they performed as well as trained human college students), or where they were taught to recognize malignant tissues on radiographs with something like 99% accuracy. They also have incredibly good long-term memory. Obviously they aren't going to be directly comparable to dogs in terms of cognition and behavior in all ways, but I don't think they're a terrible model for behavior or learning research, just - again - that the way the study's been used here is not supporting their argument.

My FiL trained and judged tracking dogs for a while, and they had one dachshund (before I knew them) who'd been trained off livestock with an e-collar (don't know if it has a specific name here), so most of what I know in that regard is through that perspective. He had one dachshund who kept after a badger even after it had bitten through the dog's tongue (this was accidental, dog hadn't been deliberately set on the badger, just was in the wrong place at the wrong time), and one that tried to follow a trail across a fast-flowing, deep river that was something like 30m(?) wide (if I'm remembering the story right - it was an area that'd be dangerous for people to cross, let alone a dachshund), so they can absolutely be tenacious little guys if they're motivated to do something. I can see why they'd need to break a fixation on sheep or other livestock.

 

[storyist]

Dogs that are a threat to horses aren't allowed on the property, full stop. Only trustworthy dogs are loose.

Off topic from the study, but how do you/did you achieve this? The problems I've seen don't come from dogs the livestock owner had control of and could ban from the property but from loose dogs owned by others. I've never seen fencing for horses or cattle that could keep a dog out.

My thought was that for the kind of irresponsible neighbors I've had trouble with, given the incentive of Animal Control, court dates, fines, and lawsuits from neighbors for vet bills, people like that still aren't going to do any real training and aren't going to start effectively confining the dog but would use an e-collar to stop the behavior since negligent as they are, they aren't willing to have the dog seized, euthanize it themselves, etc. However, I agree that the best aversive in these cases is a gun, and here in Colorado, shooting dogs harassing livestock is legal. I don't know how true that is in Eastern states and other countries.

The only aversive training with e-collars I've actually seen is snake training, which as I've said before I was talked into for one of my dogs when I started training tracking, and which I'd never do again. I'd forego the tracking before subjecting one of my dogs to that again, but the fact is my high prey drive Rottie was so affected by it she was spooky about anything strange on the ground for at least a year after. Whether she'd react to a snake now, years later, I don't know as it hasn't happened.

As to the study, from what I've seen in dogs it seems to me the effort to find the least necessary electric shock could simply condition hard dogs to the pain and help them keep going in spite of it, like nagging with ineffective collar corrections. Soft dogs aren't the problem.