A Positive Trainer Changed His Mind on E Collars. It Made Me Revisit Mine.

I once had a stand up row with a highly respected dog trainer because I watched him use an e collar on a dog so intensely that the dog was screaming and desperately trying to dig itself into the ground.

This was not somebody who had bought a remote collar online the week before and decided to have a go. He was an experienced working trials trainer who had titled numerous dogs. His reputation was precisely why I had paid to attend the course. From what I could see, the dog did not even understand what was expected of it, and when I challenged what was happening I ended up arguing with virtually everyone else there because they defended him.

I am bringing it up now because Nick Benger has recently spoken publicly about changing his position on e collars after more than twelve years as a positive trainer. I respect Nick for being honest about changing his position, even though I disagree with parts of his reasoning and remain unconvinced that the evidence justifies his decision.

I have encountered some pretty big names in dog training who publicly present themselves one way and privately train very differently. I find that considerably more dishonest than somebody standing up and telling you exactly what they do. And both sides of this argument are capable of telling fibs, not just about what the other side does, but sometimes about what they are doing themselves.

I have more respect for trainers such as Michael Ellis who, in the explanations I have heard him give, openly acknowledge that electrical stimulation can cause pain. Because it fucking can. Respecting that honesty doesn't mean I endorse his training methods. Not every use of an e collar causes the same sensation, and not every trainer uses one in the same way, but pretending the painful end of the tool doesn't exist does nobody any favours.

I probably have a slightly different perspective on this from somebody who has always trained reward based because I am what people tend to call a crossover trainer. Earlier in my career I trained using aversive methods. I learned how e collars were used, attended courses about them and wore them myself because I wanted to understand what I was dealing with.

Nick has also talked about trying an e collar on himself and being surprised by how little he felt. That is worth acknowledging, but so is what he was actually testing. He was experiencing low level stimulation, not the top end of what the collar is capable of delivering. Those are very different things. I have experienced the lower levels myself, and yes, they can feel remarkably mild. Keep turning the level up and the conversation changes considerably.

I also used prong collars occasionally earlier in my career. I do not use prong collars now, and I do not use e collars.

A lot of people first came across me through a shock collar video I made years ago when I was still working on the ground in Marbella. I had been out on three separate behaviour consultations in the same day, and all three pet owners were using e collars on their dogs. All three cases were problematic, and I came home from that day incredibly frustrated. I made the video because of what I had just seen, and I will include it further down this article because it is part of how my own position on this developed.

I haven't seen convincing evidence that e collars are necessary for the overwhelming majority of the problems they're used to address. I think the same goals can usually be achieved through good reward based training, sensible environmental management and better training mechanics without introducing an additional aversive stimulus.

That is very different from saying e collars do not work. They clearly can.

The more useful conversation, for me, starts when we separate three questions that constantly get bundled together: can an e collar change behaviour, is an e collar necessary to achieve that outcome, and if it does provide some additional benefit, is that benefit proportionate to the welfare cost we are asking the dog to pay?

Those are three completely different questions.

First, What Does an E Collar Actually Do?

A lot of reward based or force free trainers talk about e collars as though they are always used through positive punishment. The dog chases a sheep, ignores a recall or performs some other unwanted behaviour, electrical stimulation is added, and if that behaviour subsequently becomes less likely, the stimulation has functioned as positive punishment.

That absolutely is one way e collars are used, particularly in aversion training, but it is not the only way. Experienced e-collar trainers often describe procedures that go beyond simply waiting for an unwanted behaviour and applying a shock. These different procedures are worth understanding accurately, although describing their technical complexity should not be mistaken for endorsing their use.

One common procedure is to establish what trainers call the dog’s “working level”. The collar starts at its lowest setting and the stimulation is increased gradually until the trainer sees some tiny indication that the dog has noticed it. That might be an ear flick, a slight head turn, looking around or another subtle change in behaviour. The idea is not to find the point where the dog yelps or visibly recoils. It is to find the lowest level at which the dog appears to perceive the stimulation. E Collar Technologies describes essentially this procedure in its own guidance, starting at zero and gradually increasing the level until the dog first responds.

At that point, depending on how the sensation is conditioned and what happens next, it may begin to function more like a cue or signal. We should not automatically call that negative reinforcement just because electricity is involved. Learning theory is about function, not equipment. If nothing aversive is being removed contingent on behaviour and there is no resulting increase in that behaviour, we have not demonstrated negative reinforcement.

Other e collar systems do deliberately use negative reinforcement. Electrical stimulation is applied, the dog performs the required behaviour and the stimulation stops. If removing that stimulation makes the behaviour more likely in future, it has functioned as negative reinforcement.

There is another important complication here though, because “working level” can sound far more standardised than it really is. E collars are not all built with anything like the same scale. Some modern collars divide their output into 100 levels, while other devices have ten, six or even five levels of stimulation. E Collar Technologies, for example, uses a 0 to 100 scale, while older Tri Tronics equipment included models with only five intensity settings.

Those numbers are not comparable units. Level five on one collar does not mean five per cent of level 100 on another, and I would not claim that the maximum on a five level collar is automatically electrically identical to the maximum on a 100 level collar either. The important point is that a collar with only a handful of settings can have much larger jumps between available intensities than one divided into very fine increments.

That becomes pretty fucking important when somebody says they are only moving the collar up “one level”. One level might be a tiny adjustment on a device with 100 settings. On another piece of equipment, the jump between two neighbouring settings can be vastly coarser.

Research by Lines, van Driel and Cooper examined thirteen different e collar models and found substantial differences in electrical energy, peak voltage, pulse number and pulse duration. They also found large differences in electrical impedance between wet and dry dogs. The study does not tell us exactly what every setting feels like to every individual dog, but it makes it very clear that the number displayed on the remote is not a standardised unit of sensation.

I know this personally because I have worn them. At genuinely low settings you may barely perceive anything at all. Nick has described trying low level stimulation himself and essentially wondering what all the fuss was about, and I completely understand that response. The important bit is that he was experiencing the low end of what the equipment can deliver, not the top of it.

Keep turning the level up and the experience changes considerably. These devices are capable of producing stimulation that becomes extremely unpleasant and, at sufficiently high levels, incredibly fucking painful.

Both things can be true at the same time. An e collar can produce a barely noticeable sensation at one setting and a profoundly aversive one at another. Understanding that range is important, because the interesting question is what happens when the mild end of that range meets a dog that suddenly has something much more important it wants to do.

The Bit That Often Gets Left Out

One explanation of e collar training sounds fairly innocuous. Find a very low working level, introduce the sensation carefully, teach the dog how to respond and the dog learns how to switch it off. That can be an accurate description of part of some training systems, but it is not necessarily the whole system.

There is a broader difference in training philosophy here that I think is worth talking about, even if it is a generalisation. Reward based trainers tend to proof behaviour by increasing difficulty gradually. You add a little more distance, a little more duration, a slightly more tempting distraction, and you try to keep the dog successful enough that reinforcement can continue to drive the behaviour. If the dog repeatedly fails, the usual assumption is that you increased the criteria too quickly and you make the task easier again.

A lot of balanced training systems take a different approach to proofing. Failure is sometimes deliberately engineered because the correction itself is part of the training. The trainer creates or increases a distraction that they know is likely to challenge the dog, the dog breaks the behaviour, and that gives the trainer an opportunity to apply the consequence they want the dog to learn from.

That does not mean every reward based trainer keeps every session perfectly errorless, and it certainly does not mean every balanced trainer spends their day setting dogs up to fail. But it is a genuine difference in how failure can be used within the training process.

And with e collars we do not have to invent a hypothetical example. Published e collar training material from Sirius Dog describes creating a hierarchy of increasingly attractive toy distractions, presenting those distractions while the dog holds a trained behaviour, and increasing the stimulation if the dog repeatedly breaks. The explanation given is that distraction changes the dog’s perception of its normal working level.

That is important because it takes us beyond the reassuring description of a dog responding to a barely perceptible sensation. The trainer is not simply waiting for the real world to produce an unexpectedly difficult distraction. In this procedure, the difficult situation is being deliberately created as part of proofing, and the intensity of the electrical stimulus can then increase when the dog fails.

Proofing itself is not the issue. Every trainer needs behaviours to survive distraction, novelty and competing reinforcement. The question is how we get there and what role failure plays in the process.

If my dog repeatedly fails around a distraction, my instinct is normally to ask whether I have increased the criteria too quickly, whether the reinforcement is good enough, whether the dog actually understands the behaviour in that context and whether I need to break the problem down into smaller steps.

In the type of e collar proofing described above, failure can instead become the moment at which the consequence gets stronger.

That is a pretty significant difference.

So when e collar training is described only as finding a tiny working level and teaching the dog how to switch it off, I think that leaves out an important part of the story. The more interesting question is what happens when the dog encounters, or is deliberately presented with, something it wants more than it wants to respond to the handler, and what the training system asks the person holding the remote to do next.

What Does the Research Actually Tell Us?

This is the section where everybody usually grabs whichever paper agrees with them and starts hitting the other side over the head with it.

The evidence is not clean enough for that.

One of the most frequently quoted pieces of e collar research is the Defra funded work involving 63 dogs with off lead related behavioural problems. Cooper and colleagues published welfare and efficacy findings from that work in 2014. Dogs were divided between e collar trainers using e collars, the same trainers working without them, and a separate group trained by reward focused trainers.

The larger controlled study used lower stimulation settings and warning cues than an earlier preliminary phase. Even under those conditions, the e collar dogs spent more time tense, yawned more and interacted less with the environment than the reward focused group. There were not significant differences between groups in some physiological measures, including urinary corticosteroids, and owners across all three groups reported high rates of improvement. The researchers concluded that they found greater welfare concerns without a consistent efficacy advantage for the e collar group.

In 2020, China, Mills and Cooper went back to video from the same 63 dogs and analysed specific measures around recall and sit. That is important because the 2014 and 2020 papers are not two completely independent experiments. The later paper is a more detailed analysis of the same research programme. The reward focused group performed better on several of the measures examined, and the authors again concluded that there was no evidence that e collars were necessary for the training problems studied.

But that paper has also received legitimate methodological criticism.

Sargisson and McLean pointed out that the reward focused group was trained at a different time and location and by different trainers, allocation was not completely random, dogs were on lead and very close to trainers for much of the sessions, there was no useful baseline for some of the obedience measures, and the change to owners handling the dogs on the final day potentially complicated interpretation. They also challenged aspects of the statistical analysis and argued that the study did not directly recreate the predatory behaviour for which many of the dogs had originally been referred.

Cooper, Mills and China published a response defending their methodology and conclusions, including their statistical choices and the relevance of the wider research programme.

So I would not use that study to say science has definitively proved reward based training is superior to e collars for every recall problem.

It has not.

What it does give us is a controlled comparison in which an e collar approach did not demonstrate the clear efficacy advantage people often claim for it, alongside some evidence of welfare concern.

That is useful evidence.

It is not the final word.

Then We Have a Study Where the E Collar Did Better

Johnson and Wynne published a very different study in 2024 looking specifically at stopping dogs chasing a fast moving lure.

In their experiment, dogs trained using e collars stopped chasing within one or two training sessions. The two food based protocols they compared against did not achieve the same result during training. The e collar dogs also refrained from chasing during the first three test sessions, although four of the six dogs in that condition chased again when tested in a novel arena during the final test.

That is a result I think reward based trainers should be prepared to say out loud.

Under the specific conditions tested, electrical punishment suppressed chasing more effectively than the two food based procedures used in the comparison. However, the reward protocols have been criticised for their design, and the result does not establish that e collars outperform well delivered reward based predation training generally.

There's another problem with that comparison from my own practical experience. In the most effective reward based predation training I've seen and taught, food has rarely been the most effective reinforcer for dogs with strong predatory motivations. We are often working with toy play, chase games and other carefully controlled activities that are much more closely related to what the dog actually wants to do. That doesn't prove those approaches would have succeeded in this particular experiment, but it does raise an important question about whether the reward based procedures being compared were genuinely appropriate for the dogs and the behaviour being tested.

The dogs receiving shocks also yelped when shocked, which the authors themselves described as presumably pain induced. Their behavioural coding otherwise found few stress related behaviours, although the peer review process raised concerns that the welfare assessment was too limited to support particularly confident conclusions about welfare. The reviewers also questioned whether the food reinforcer had been adequately conditioned or demonstrated to be sufficiently motivating for the dogs.

The more recent DeLeeuw and Williams paper similarly notes that the Johnson and Wynne study is important because it is the only published randomised controlled trial in this area, while also highlighting its small sample, an extra training session received by the e collar group and the lack of validation that food was sufficiently motivating compared with other possible reinforcers such as toys.

Again, my conclusion is not that the study should be dismissed because I do not like the result.

The e collar worked better for the exact task and reward procedures tested.

What it does not establish is that e collars are generally better than good reward based predation training, that they are necessary for predatory dogs, or that the welfare cost is trivial.

Those claims would require different evidence.

Predation Is Where This Gets Genuinely Difficult

Predation is probably the part of this whole conversation where I am least interested in pretending I have an absolute answer. There is evidence that targeted electrical aversion can suppress chasing, and there are situations where predatory behaviour creates very real welfare and safety problems. Dogs kill livestock, chase wildlife and run into roads. In some circumstances the dog itself, the prey animal or both can die. I am not going to pretend those situations are trivial because doing so would make the argument easier.

Where I think the conversation goes off the rails is when people turn that into one of two ridiculous extremes. On one side you get the argument that without an e collar a highly predatory dog is condemned to a miserable life permanently restrained on a lead. On the other, you sometimes get the suggestion that deliberately exposing a dog to one severely unpleasant experience somehow makes its entire life no longer worth living. Most real dogs live somewhere between those two fucking positions.

If I have a dog that is genuinely dangerous around sheep, my first intervention is not to stand next to a field full of sheep debating which quadrant of operant conditioning gives the dog the most freedom. My first thought is probably, “Don’t walk the fucking dog loose next to sheep.”

We use management constantly in dog training. We choose where we walk dogs, use long lines, secure spaces and fences, avoid situations where the consequences of a training failure could be catastrophic, and provide appropriate outlets for predatory behaviour. A dog does not need unrestricted access to every possible environment in order to have a very good life.

But management is not bulletproof either. Leads get dropped, gates get left open, equipment fails and dogs escape. Pretending management removes all risk would be just as dishonest as pretending an e collar removes predation from the dog. The fact that management can fail is a legitimate part of the discussion, particularly where the consequences of failure could be serious.

Restriction can have welfare costs too, and there may be individual dogs and circumstances where the trade off becomes genuinely difficult. That is exactly why I am reluctant to make absolute statements around predation.

But presenting the choice as either fry the dog or imprison it for life is fucking ridiculous. Neither is a serious description of the choices available to most owners, and neither helps us have an intelligent conversation about the genuinely difficult cases where the welfare trade offs are not obvious.

Working Is Not the Same as Being Necessary

This, for me, is probably the most important part of the entire discussion. If somebody shows me a dog recalling beautifully after e collar training, that demonstrates that their training produced a recall. It does not demonstrate that the e collar was necessary.

After I crossed over to reward based training I spent around five years training Mondioring alongside balanced trainers. I was surrounded by people using aversive tools, including some extremely skilled trainers. There are balanced trainers and reward based trainers who can out train me, and I am completely comfortable saying that.

During those years I never felt compelled to introduce those tools back into my own training because I was happy with the results I was getting. Now imagine the same five years going very differently. Imagine I went to competition after competition and got absolutely nowhere while the people around me succeeded using aversive tools. It would be incredibly tempting to conclude that the missing ingredient was the aversive tool.

Maybe it was. Or maybe I just needed to become a better trainer.

Those possibilities are not interchangeable.

There is a significant difference between not being able to achieve something with your current skills and demonstrating that it cannot be achieved without using aversives.

I think reward based trainers need to be honest about that as well. If I am working a case and I cannot solve the problem with the skills, strategy and experience I am bringing to it, that tells me I have not solved it. It does not automatically tell me that reward based training cannot solve it.

That is a much more accurate conclusion, even if it is less comfortable for the ego.

And if we use predation as the example, this is probably where my own decision making becomes clearest. If I was dealing with a predation case and I felt I was getting to the edge of my competence, I would be more likely to refer that case to somebody I thought had more skill and depth in that area than I did before deciding the answer was to start introducing aversives.

That matters because there is a huge gap between “I cannot solve this” and “this cannot be solved without an aversive tool”. I do not think we should collapse those into the same statement.

The 2026 DeLeeuw and Williams study is useful here because trainers in their sample described changing approaches in both directions. The most common pathway was away from correction or compulsion towards greater use of positive reinforcement, but a smaller group described starting primarily reward based and later incorporating positive punishment or negative reinforcement. Trainers described education, experience, welfare considerations and perceived effectiveness among the influences on their changing practice. The study does not demonstrate that poor reward based skill causes trainers to adopt aversives, and I do not want to imply that it does.

What it does leave us with is a question I think every trainer should probably ask themselves before deciding that a more aversive tool is necessary: have I actually exhausted the better options, or have I just reached the edge of my own current ability?

The Welfare Calculation Is Bigger Than “Did It Work?”

This is where I think three questions need to stay separate. Did the intervention change behaviour? Was that intervention necessary to achieve the important outcome? And was the additional welfare cost proportionate to the benefit?

A procedure can pass the first question and fail the other two. There are countless things I could do to a dog that would change its behaviour very efficiently, but effectiveness alone has never been enough to establish that something is good training.

Nick has made an argument that, broadly, a brief period of unpleasant training may buy a dog years of increased off lead freedom. I understand the welfare calculation he's proposing, and I think it's worth examining properly rather than dismissing it. But we should also be clear about which end of e collar use we are talking about here. This is not the barely perceptible working level we discussed earlier. If the dog receives a highly aversive electrical stimulus following behaviour such as chasing, and that behaviour subsequently decreases, we are talking about positive punishment.

There's another part of the welfare calculation that I don't think we should ignore, and that's the length of time it takes to change behaviour. If a dog is experiencing significant stress or frustration over an extended period of training, that has a welfare cost too. Simply avoiding aversive equipment doesn't automatically mean the dog is having a wonderful time. I've seen reward based training that has been frustrating and stressful for the dog, and we should be prepared to acknowledge that. Equally, we shouldn't assume reward based training necessarily takes longer, or that a painful electrical correction is therefore the kinder option. If we're going to compare the welfare implications properly, we need to consider the intensity, frequency and duration of the dog's stress, the likelihood of success and the longer term consequences of whichever approach we're discussing.

I wouldn't dismiss the possibility that a short, highly aversive intervention could produce a lasting welfare benefit in an individual case without looking at the evidence. But the burden is on the person proposing it to demonstrate that the benefits outweigh the costs and that a reasonable lower risk alternative isn't available. That means asking how reliably the training generalised, how long the effect lasted, what happened emotionally and behaviourally outside the behaviour we were targeting, whether a lower risk approach could reasonably have produced the same outcome and what management alternatives existed.

Those are not rhetorical questions designed so the answer must always be “don’t use an e collar”. They are questions I think somebody should be able to answer before deliberately subjecting a dog to a highly aversive electrical stimulus in order to change its behaviour.

A Lead Can Be Abused Too

A common response at this point is that any equipment can be abused.

That is true. You can hurt a dog with a lead. You can hurt one with a flat collar. You can hurt a dog with your hands. Reward based trainers can train terribly and cause enormous frustration, confusion and stress.

I do not think the existence of bad reward based training tells us much about the ethical properties of an e collar though.

There is a meaningful difference between taking an ordinary piece of equipment and misusing it in a way that causes pain, and using a device whose purpose is to deliver an electrical stimulus directly to the dog, with an intensity that can be increased to a level where pain or strong aversion is part of how the procedure changes behaviour.

A lead has to be used in a particular way before pain becomes part of its behavioural effect. An e collar delivers electrical stimulation because that is what it is built to do.

At low levels that stimulation may not be painful. At higher levels it plainly can be. In some aversion procedures, the fact that the stimulation is painful or intensely aversive is not an unfortunate side effect. It is part of the mechanism by which the behaviour is suppressed.

There is a broader principle here that we recognise in plenty of other areas of life. Having access to something capable of causing harm does not mean you are going to use it badly. Give me a gun and I am not suddenly going to start shooting people. But if I get into an argument and I do not have a gun, there is precisely zero chance of me shooting somebody in the heat of the moment.

The presence of the tool changes what is possible.

That is not me saying an e collar is equivalent to a firearm. It obviously fucking isn’t. The point is that when somebody has immediate access to a device capable of delivering a painful consequence, frustration can be translated into a stronger consequence very quickly. Without that device in their hand, that particular form of escalation is simply not available to them.

That becomes especially important when we remember who controls the button.

The Human Holding the Remote Matters

This is another part of the conversation that often gets reduced to “experienced hands”.

Experience matters, obviously. A skilled trainer should have better timing, better judgement and a better understanding of what they are trying to achieve than somebody who has watched three YouTube videos and bought a collar online.

But experienced hands are still attached to a human being.

Humans get tired, frustrated, embarrassed, impatient and angry. Trainers are not exempt from that because they have a good CV. In fact, one of the most uncomfortable things about working around experienced trainers for long enough is realising that technical competence does not automatically come with perfect emotional regulation.

I have seen experienced trainers become frustrated and turn the level up.

That does not mean every increase in intensity is emotional or inappropriate. Sometimes the trainer has made a deliberate judgement that the dog has not responded at the current level and changes the intensity as part of the procedure they are using. But there is also a point where increased intensity can stop looking like training and start looking a lot more like retaliation.

The seminar I mentioned at the start of this article is the most extreme example I have personally witnessed. This was an experienced trainer with titled dogs and an excellent reputation, not a novice misusing equipment through ignorance. From where I was standing, the dog appeared confused, the stimulation kept escalating and eventually the dog was screaming and trying to dig itself into the ground.

And this was on a fucking sendaway for a fucking sport.

This was not a life or death situation. Nobody was protecting livestock, stopping a dog running into traffic or dealing with some impossible welfare dilemma. It was a sport exercise.

That experience made the phrase “safe in experienced hands” considerably less reassuring to me.

The issue is not that experienced trainers are uniquely dangerous. It is that expertise does not make somebody immune to frustration, ego or poor judgement in the moment.

And with an e collar, the distance between feeling frustrated and delivering a substantially stronger consequence can be the movement of one thumb.

What About Side Effects We Did Not Intend?

When we talk about whether an e collar worked, we usually mean something fairly narrow. Did the dog stop chasing? Did it recall? Did it stay in the behaviour we asked for?

That tells us whether the target behaviour changed. It does not necessarily tell us what else the dog learned at the same time.

Dogs do not get an explanatory note telling them exactly why an unpleasant stimulus happened. They experience the whole situation. Depending on timing, context and what is most salient to the dog, the aversive event could become associated with the behaviour we intended, but it could also become associated with the sheep, the location, the handler, another dog, a child, a particular cue or something else happening in the environment.

I am not saying that happens every time somebody uses an e collar. It clearly does not. Good timing and good training matter, and plenty of dogs will show exactly the behavioural change the trainer was aiming for.

But the possibility of unintended association is part of the welfare calculation and I think it gets brushed aside too easily when the only outcome being discussed is whether the dog complied.

A dog can stop doing something and we can still have changed more than the behaviour we were targeting.

That is one of the reasons I am wary of using compliance as the entire measure of success.

Even Expert Use and Public Availability Are Different Questions

There is another argument that I think gets muddled far too often.

Somebody might make a case that an e collar can be used competently by a very experienced trainer in a narrow set of circumstances. Even if I accepted that argument completely, it would not automatically follow that the same equipment should be available for anybody to buy online and start experimenting with on their dog.

Those are two separate questions.

If the defence of the tool is that it requires excellent timing, careful conditioning, a proper understanding of learning theory, good judgement around intensity and strong emotional control from the person holding the remote, then that defence creates its own problem.

Most members of the public do not have those skills.

They can still buy the equipment.

They can still turn the level up.

They can still misunderstand what the dog is doing, punish the wrong thing, mistime the stimulation or create an association they never intended.

And as we have already discussed, the equipment itself is not standardised in any meaningful sense across manufacturers. One remote may divide its output across 100 levels while another has only a handful. The number on the screen does not tell an inexperienced owner what the dog is actually experiencing.

So even if somebody convinced me that there was a legitimate place for highly skilled professional use in a narrow category of cases, that would not answer the separate question of unrestricted consumer access.

In fact, if your argument is that the tool is only safe when used by somebody highly skilled, that seems like a pretty strange argument for selling it to absolutely anybody.

What Would Change My Mind?

One question I think is useful in arguments like this is: what evidence would actually make me change my position?

If good comparative research showed that in a clearly defined group of dogs, high quality reward based training and sensible management had repeatedly failed, and that e collar training then produced a substantial, durable improvement in welfare and safety with acceptably low adverse effects, I would have to take that seriously.

I would want to know whether the effect generalised beyond the training setup, whether it lasted, whether the dogs were genuinely better off overall, and whether the comparison had actually tested competent alternatives rather than a weak version of reward based training.

But if the evidence was strong enough, then yes, it should move my position.

I think the reverse question is just as important for people who advocate for e collars.

What evidence would convince you that the tool was not necessary?

If the answer is nothing, because the fact that the dog improved after e collar training is itself taken as proof that the e collar was necessary, then we are no longer making a scientific argument. We are just defending the conclusion we started with.

So Do E Collars Work?

Yes.

That is probably the least interesting question in this entire article.

Electrical stimulation can change behaviour. It can function as positive punishment, it can function as negative reinforcement, and at lower levels it may sometimes be conditioned and used more like a cue or signal. We also have evidence that, under some specific procedures, e collars can suppress behaviour very effectively.

The harder questions are whether they are necessary, what they cost the dog, what else the dog might learn in the process, whether the effect lasts, whether it generalises, whether lower risk alternatives have genuinely been exhausted, and whether sensible management could achieve an acceptable outcome without adding another aversive tool.

Predation is where I am least comfortable pretending there is a simple answer. There are genuine safety problems, genuine welfare trade offs and some cases where management may be restrictive or may fail. I am not going to claim that one unpleasant experience automatically destroys a dog’s welfare, just as I am not going to accept that a dog who cannot safely run loose around livestock has somehow been sentenced to a miserable life.

I also think both sides of this argument need to be more honest.

Reward based trainers should be willing to acknowledge evidence showing that electrical punishment can be effective, including studies where it has outperformed the reward procedures it was compared against. Pretending aversives do not work is not a serious position.

People advocating for e collars need to be equally honest that these devices can deliver stimulation that is intensely aversive and painful, that this can be part of the mechanism by which behaviour is suppressed, and that demonstrating an effect does not demonstrate necessity.

For me, that is where the argument keeps coming back to.

I do not think our job as dog trainers is to maximise obedience at any cost. It is to balance safety, welfare, reliability, freedom, practicality and the needs of the people living with the dog.

Sometimes that means training harder. Sometimes it means changing the environment. Sometimes it means referring to somebody with more skill in a particular area. Sometimes it means accepting that management is the sensible answer rather than trying to engineer perfect control over every behaviour in every environment.

And before adding something more aversive, I think there is one question worth asking very fucking honestly:

Do I actually need to do it this way?

References and Further Reading

Lines, J. A., van Driel, K. & Cooper, J. J. (2013). Characteristics of electronic training collars for dogs. Veterinary Record, 172, 288. DOI: 10.1136/vr.101144. The study measured 13 collar models and found substantial variation in electrical characteristics and impedance under different conditions. PubMed

Cooper, J. J., Cracknell, N., Hardiman, J., Wright, H. & Mills, D. (2014). The Welfare Consequences and Efficacy of Training Pet Dogs with Remote Electronic Training Collars in Comparison to Reward Based Training. PLOS ONE, 9(9), e102722. DOI: 10.1371/journal.pone.0102722. PLOS

China, L., Mills, D. S. & Cooper, J. J. (2020). Efficacy of Dog Training With and Without Remote Electronic Collars vs. a Focus on Positive Reinforcement. Frontiers in Veterinary Science, 7, 508. DOI: 10.3389/fvets.2020.00508. This is the later analysis of the same broader 63 dog research programme, rather than an independent replication. Frontiers

Sargisson, R. J. & McLean, I. G. (2021). Commentary: Efficacy of Dog Training With and Without Remote Electronic Collars vs. a Focus on Positive Reinforcement. Frontiers in Veterinary Science, 8, 629746. DOI: 10.3389/fvets.2021.629746. This is the methodological critique discussed in the article. PubMed

Cooper, J. J., Mills, D. S. & China, L. (2021). Response: Commentary: Remote Electronic Training Aids; Efficacy at Deterring Predatory Behavior in Dogs and Implications for Training and Policy. Frontiers in Veterinary Science, 8, 675005. DOI: 10.3389/fvets.2021.675005. This is the authors’ response to that criticism. Frontiers

Johnson, A. C. & Wynne, C. D. L. (2024). Comparison of the Efficacy and Welfare of Different Training Methods in Stopping Chasing Behavior in Dogs. Animals, 14(18), 2632. DOI: 10.3390/ani14182632. This is the chasing study where the e collar protocol outperformed the two food based protocols under the conditions tested. MDPI

Bastos, A. P. M., Warren, E. & Krupenye, C. (2025). What evidence can validate a dog training method? Learning & Behavior, 53(3), 227 to 228. DOI: 10.3758/s13420-024-00658-9. This peer reviewed commentary questions the suitability of the comparison protocols and the welfare conclusions in Johnson and Wynne. PubMed

Johnson, A. C. & Wynne, C. D. L. (2025). Chasing solutions: A response to Bastos et al. (2024). Learning & Behavior, 53(3), 229 to 231. DOI: 10.3758/s13420-025-00672-5. Including the response keeps the references balanced rather than presenting the critique without the authors’ reply. PubMed

DeLeeuw, J. L. & Williams, T. J. (2026). Professional dog trainers’ perspectives on training methods: ethical and evidentiary insights. Frontiers in Veterinary Science, 13, 1744448. DOI: 10.3389/fvets.2026.1744448. The published article was corrected in April 2026, so we should link to the current corrected version. Frontiers

Sirius Dog, Lou Castle. Proofing the Behaviors. This is not scientific research. It is included as a primary example of an e collar training procedure describing a hierarchy of increasingly difficult distractions and increasing stimulation when the dog repeatedly breaks behaviour.

E Collar Technologies. Pro Educator Owner’s Manual. Manufacturer material describing precision low level stimulation and the equipment’s adjustable stimulation functions. Again, this is product documentation, not welfare research.

Tri Tronics Bark Limiter owner’s manual. Useful only for supporting the factual point that some older equipment offered five intensity levels. It should not be used for any broader scientific claim.

Further listening