Recently, the dog-training social media world has been once again in dispute over punishment. I hate this. Nobody ever really falls out over using toys as a reward.
The arguments are always the same. Some people claim that punishment works. And it may – with many, many provisos. First off, you need to be a great trainer to punish dogs effectively. You need great timing. Most people’s timing is really sloppy. I watched a video of me training Lidy the other week and I was embarrassed how poor my timing was. I’m surprised she ever learns anything. We’re lucky most dogs are intuitive.
Second, you need to understand that punishment simply suppresses behaviour. It doesn’t change the underlying need to perform the behaviour, it just might stop the dog doing it again in future. The need to do the behaviour doesn’t go away. So if the dog’s pulling or jumping or even biting, then punishing it might stop the behaviour in future. But the dog’s motiviations, emotions and underlying need to do that behaviour are still there. Bear this consequence of punishment in mind because it’s going to be really important in the rest of this post.
Third, punishment can cause frustration and aggression. It causes frustration because your dog doesn’t know what you actually want them to do, just that you don’t want them to do that. It can also cause aggression – partly because of the frustration sometimes and partly because it doesn’t do anything at all for your bond with the dog. Will your dog trust you if you punish them? Hell no. This is one reason some trainers are so fond of shock collars because it at least might not seem as if it’s coming from them. This disintegration of trust is also is going to be a really important factor in the rest of this post. Herron et al. (2009) did a study of the consequences of punishers on aggression. You won’t be shocked to realise that certain types of punisher (including a hard stare) increased aggression by up to 40%.
Fourth, punishment reduces all behaviours of a certain type. Dogs just stop doing stuff. If you’ve ever seen people playing the ‘hot’ and ‘cold’ game using shock, you’ll see the same there: they just stop moving in the end. A bit like me in my A level English Lit class. We got told we were wrong so many times, we just stopped answering or responding. What you get are dogs who are afraid to try to do anything.
Fifth, punishment increases distance between the dog and the person responsible for punishing them. Again, why some trainers like shock because it seems as if they aren’t doing the punishment. This might seem seductive until you realise that the sixth potential consequence of punishment is that dogs can generalise wildly during punishment and are apt to learn to connect all kinds of unconnected events to the punishment – an odour, the time of day – you don’t get to pick.
Of course, you don’t get to know if any of these six things will be a consequence until you’ve done it. Great. Totally unpredictable and you can never know whether or not the dog will learn totally the wrong thing.
These six factors are the logical, rational and scientific reasons I don’t use punishment in my training. Not least because most of the dogs I work with have aggression histories and – with no judgement intended on my lovely clients – they’ve already tried punishment already and it failed them. Sometimes it has worsened things. Sometimes it has even been responsible for causing the problem in the first place. I don’t use choke chains, head halters, prong collars or shock collars. I don’t even largely use flat collars any more either. I don’t use ‘no!’ and I don’t use water sprays, training discs, shake jars, compressed air sprays, air horns, citronella collars, spray collars, invisible perimeter fences, stern looks, standing over dogs, rolling dogs… I’m kind of lucky that I don’t have to because most of these things are things people have already tried them.
That brings me to the ethics of training. I don’t use punishment on dogs for two reasons. The first is that I like to hold myself personally to better standards. I don’t use it with people and I won’t use it with animals. You could tell me positive reinforcement was only 20% as effective as punishment and I’d still use reinforcement in my training. The second is that dogs often let us punish them, which is largely to their detriment. They don’t deserve us to use aversives just because they tolerate it. Cats largely don’t. I’m a cat person at heart. My cats would have moved out of the house and voted with their feet if I’d used punishments with them. Positive reinforcement in life works on all creatures from single celled ones up to human beings, if you like to think of life organised in that way. You can train bees to play hockey and fish to do agility courses using positive reinforcement. But you can’t punish a bee. A bee would just sting you. A wasp would sting you repeatedly. Most animals would avoid us completely if we punished them, or they’d avoid the situation. Go try punish a wolf! I listened to the very great Jean Lessard in discussion about punishment with a trainer who uses aversives. He made the point very clearly that wolves wouldn’t tolerate it.
Free animals don’t come back for seconds unless… they really want something … or they forget because the punishment lapses. That’s the thing about punishment. You’re committing yourself to a lifetime of aversives. If you use deer scarers to keep deer off your crops, then expect to find them coming back when the thing runs out of gas. Or, if there’s nothing else to eat and they are desperate. Punishment is a lifetime commitment. You can hope that your dog will ‘forget’ the behaviour you don’t like, or that they’ll ‘remember’ the punishment, but if they happen, those are by-products. Punishment needs you to always correct what you see as an error.
Remember that. It’s important. Every time you see the behaviour, you’ll have to correct it.
If you don’t, you’re doing some complicated stuff, learning-wise. Firstly, you’re putting the behaviour on an intermittent schedule of reinforcement. That sounds like stupid science waffle, I know. What it means is that because the dog sometimes gets what they want, they’ll keep doing it. Behaviour that is sometimes rewarded is more resilient even than behaviour that is always rewarded. An example: one is a man using a whip (I know…) to keep dogs away from a big pile of food he’s put out in a hunting kennels. Some of the dogs keep running in and grabbing a piece. All that whip is doing is temporarily suppressing the dogs’ behaviour of running in and grabbing food. Second, some of them keep getting some food, so they’re going to keep doing it. Punishment is always a 1:1 thing. If it’s not, even if it’s 99% punished and 1% rewarded, there’s a risk you’re actually creating a bigger problem. I see this ALL the time with people who punish dogs for jumping up. All it takes is that one time the dog gets what they need – to say hello – and boom, the behaviour is back again. With knobs on.
A lifetime commitment to punishing error. That’s what you need.
There are science-y waffle conditions, exceptions and complications that you might want to explore if you’re a very geeky dog trainer, but in a nutshell, nobody has any need to really understand those if you work off the principle that punishment is a lifetime commitment to suppressing behaviour and punishing errors.
All this is nothing new. I’ve written about this so many times from scientific perspectives and ethical perspectives.
But where it intersects hugely with my work is in rescue and rehoming.
A dog who has been trained with all the good stuff – great food and great toys – can be switched to what we call a lean schedule of reinforcement. What that means is that you don’t need to keep doing it. Unlike punishment. Thus, once I’ve taught a dog to walk loose-lead with treats, I can phase them out. Hoorah. Sure, I can keep it fresh by occasionally bringing the treats out and freshening up on skills, but I don’t need to keep doing it.
When my clients ask, therefore, “Do I need to keep using treats?”
No. Absolutely not. It would actually be better to thin them out.
That is not the same with punishment. Guardians should ask, “Do I need to keep using punishment?”
The answer is, “Probably. Most likely. Especially for problem behaviours.”
It also means that other people can take over the training. I was working with a really sweet reactive girl the other week and she’s had nothing but good stuff since she was adopted. She was throwing out lovely behaviours left, right and centre. She was happy to sit for a biscuit from me (though I don’t encourage strangers to use food with reactive dogs and I really prefer the guardian to do the treat bit) and once she realised that her behaviours also worked on me, well, her world made sense.
It means that, should I ever need to rehome Heston or Lidy, their new guardians would just need to say ‘sit!’ and they would. If they went to people who believed in praise and petting (traditionalist cheapskates!) rather than food, then that’d be fine. I do need to teach them to respond to others asking, but it doesn’t take much to switch from one human to the next.
This is not the same with punishment. When a dog comes from a background of punishment, then it requires the shelter, the fosterer, the adopter, everyone involved in that dog’s life, to continue using punishment to suppress behaviour – whether they agree with it or not.
Take off the bark collar, and it’s likely barking will come back again.
Take off the shock collar you’ve been using to stop the dog running off on walks, and the behaviour will likely come back again.
Teaching your dog using punishments passes that punishment on down the line. It requires everyone else in that dog’s life to also punish the dog. And because you’ve never dealt with the underlying emotion, especially behind problem behaviours, that behaviour is likely to pop right out again if the new family don’t want to use punishment.
Worse still, if you’ve been using direct methods of punishment like hard stares or alpha rolls, water sprays, compressed air, hard words, choke chains or head halters, then a person who has no history with that dog is then put into the very dubious position of having to punish a dog they don’t know. As you know by now, the risk of aggression in such a case is huge. There’s none of the learning history, none of the bond, none of the trust. And when you punish a dog, you absolutely need to have those things in place for it to work if you want to avoid repercussions that are likely to end in aggression.
You, for instance, might control your dog with a choke chain on a walk. But when you pass your dog to a new dog walker who has no history with your dog and you ask them to do the same, well, you’re putting a lot of faith in your dog not turning round when the walker yanks on their neck and saying, “Oh will you f@*k off pulling me!”
None of us want to think that our dogs will ever need to be re-homed, but in reality, it happens. People get ill. People divorce. People marry. People die. People move. People go into homes. People suddenly find themselves having to live with a relative. When that happens, if you’ve passed your dog on to someone else and you’ve got a history of punishment to suppress unwanted behaviour, you’re handing over a time-bomb. And you’re doing it when the dog’s bonds are weakest, when the dog is most vulnerable and when the dog is most stressed.
One example I had recently was a dog of a breed known to be suspicious of strangers. He arrived with a shock collar, which the new guardians put in the bin. They didn’t know why he’d got the collar on, or even what type of collar it was. They certainly hadn’t signed up to use punishments on the dog. The first few times the dog saw strangers, he seemed perfectly normal. Seemed. A couple of weeks in, and the dog gets caught out by a postman coming on the property, and boom, barking and lunging freely as he’s always wanted to but never been able to. The new guardian grabbed the dog and the dog turned on him. Without a conditioned history, we don’t have the permission to grab adult dogs who we don’t know. Two weeks doesn’t give our new dog time to understand our idiosyncrasies.
I’ve heard arguments before from trainers who say positive reinforcement training ends with dogs in shelters. This drives me nuts. They say it ends with dogs being euthanised and being unable to be rehomed. This is a lie. In France’s second-largest shelter, we are embedded in a punishment culture. One local trainer actually uses whips and chains and it shows him doing so on his website. His nickname is Mr Whip. We have positive reinforcement trainers too, but on the whole, France is filled with ‘dresseurs’ who use aversives. Some are mild, others are not. In eight years, we’ve never had a dog surrendered because positive training had failed. Let me say that again. Not one dog ever surrendered because their owner used biscuits to teach them to sit.
We get loads of dogs who’ve had nothing at all. Fine.
And we get loads of dogs who’ve been hit, been shocked, been subjected to choke chains and prong collars. Loads.
For most, they go on to be great dogs. After all, it was only their first owner who was a knob, and other people seem quite nice. Some take months to learn to trust again.
Others are simply so unreliable and have generalised so much about humans being knobs that you’re literally rebuilding them from the inside out. But it’s not like you’re just starting from scratch. You first have to undo the damage that has been done.
It is never dogs who’ll sit for a biscuit who are the problem. They are barely a problem at all. Usually, they tell you that’s what they know, because they try it out on you, hopefully. We don’t get many of those.
When you take off the punishment – when you take off the shock collar – when you remove harsh punishments – that behaviour is apt to come back with friends. It’s also likely to do so with people who have either no idea that they would even need to continue punishments or engage in lengthy retraining. They’re also people who have no history with the dog.
When you have history, you can get away with aversive experiences. I’m not the type to pretend that I don’t say ‘no’ to my dogs from time to time. It’s layered on years of a trusting relationship where they are rewarded for the ‘right’ choices (read: things I’ve decided in my infinite wisdom that are ‘right’) and when sometimes I say no, they stop. I will often cue them to do something they can get rewarded for and we all learn to get along cooperatively.
But when I first knew Lidy, she grabbed a towel out of my hand. I said no. Her look right then was, “F@$k you, bitch, and f@*k your biscuits too.” It set us back for weeks.
Trust is hard won and easily destroyed. Especially with a stressed, vulnerable dog whose primary attachment figures have disappeared and whose world makes no sense either.
I don’t joke when I say that punishment is apt to cause all sorts of untold problems further down the line.
One was the dog whose kennels used a head-halter without telling his guardian. It caused him a neck injury that led to a bite when his guardian returned. He never truly recovered and he was euthanised three years later because his guardian could never rebuild his trust in handling.
Another was a dog who was surrendered muzzled, wearing a shock collar and a prong. Taking these off gave him the freedom to finally express how fearful he was in public. He bit a vet who lifted him into a car. Years of suppressing his feelings about being handled could never be overcome. You don’t get that with dogs who expect a bit of cheese when a stranger handles them.
Finally, a foster who used bark collars. The dog became neck sensitive – a common problem for dogs who’ve had aversives applied to the neck – and putting leads on the dog, handling the dog, moving the dog, even touching the dog turned out to be a nightmare.
I read sometimes of rescues who use aversives. These cases are all reasons why we do not. The fallout for new guardians can be enormous. They make dogs unreliable and unpredictable. From the mildest problems – my girl Flika who ignored her name because it had been used when she was in trouble – Amigo, who trembled when he saw a fly swatter and cowered when you put him on a lead and wouldn’t even go for a pee – Tilly, who was untouchable for months, who wasn’t house-trained and needed rebuilding from the inside-out – to the most severe cases where new guardians have been bitten… punishment causes all kinds of problems that rewards never do. My easiest dogs were Ralf and Tobby, robust old boys who’d happily sit for a biscuit, who never cowered, who never were afraid.
It goes without saying that nobody should ever use aversives on a dog that is not their own – from groomers and house-sitters to kennels and vets.
But I think we should also understand the complications of situations where the original guardian has trained the dog using punishments.
It puts the new guardian in a position where it can be tricky to live with the dog, not least if it obliges them to keep using punishers because it’s all the dog has ever known.
It also means they have to have a hands-off approach until the dog trusts them. No pulling out brambles, no touching the dog by mistake, no collar grabs, no helping the dog into the car, no putting on lampshades after surgery, no lifting dogs onto vet tables. That can be really tough for guardians who’ve always had a hands-on approach of their own and expect to be able to do the same. It is also tough considering how many rescue dogs require veterinary interventions straight away. It’s tough for vet staff in shelters and for unfamiliar vets working with new clients.
So to come back to dog trainers justifying the technical use of punishment… we really need to think of the life of the dog. We might not necessarily be training the dog to be with us all their lives. Heaven only knows what may happen to us. But pass on a dog who has been loved, a dog who knows humans are kind, who trusts humans, who has learned that life’s inescapable bad stuff is invariably coupled up with good stuff.
Don’t pass on a dog who needs a shock collar to walk past other dogs, or who only behaves because they’re scared of you. Punishment engenders fear. That’s all it does. It sets the dog up for a lifetime of that punishment unless you layer in alternatives and you’re an expert trainer who can 100% guarantee there will be no fallout.
And here’s the thing: no trainer can guarantee that. It’s not about whether punishment may work. It’s not about whether it’s ethical or not. Scientists and philosophers can argue the theory between themselves and let us know when they have an answer. It’s about the real-world fallout of punishment for dogs… the real-world consequences. It puts future guardians in jeopardy and it puts the dog at risk.
Consider the dog who has been kneed in the chest if they jump up. It may seem that they’ve ‘learned’ not to jump (usually in the presence of people who knee dogs in the chest) but put them in a new environment, where stress lowers inhibitions, and jumping up is going to come back again. Are new guardians also expected to knee dogs in the chest?!
And consider the dog who has a history of aggression, ‘cured’ by a shock collar. Take the collar off and the aggression is still right there, in a new home, with a new family who may not have even known it was there.
Whenever, then, trainers get clever and punishment and aversives seem like seductive short-cuts, remember the consequences.
For guardians adopting adult dogs, you do not have the privilege of trust. Maybe you used mild aversives with all your dogs before; maybe you are very handsy and feel you should be able to grab a dog. Know that if you do so with a dog you’ve adopted – even months into the relationship – they may not tolerate from you what your previous dogs tolerated.
For guardians raising puppies, know that you are possibly creating a time-bomb if you use punishers and you need to rehome your dog. None of us think that will ever happen to us, but if you rely on chokes, prongs, shock collars, sharp words and forced handling, then you can’t predict that the future guardians will feel the same or even be able to use aversives in the same way your dog’s trust in you gave you licence to do.
It’s not a debate about should we use punishment, how effective it is or whether it’s ethical. It’s knowing that there are real-world consequences that we might not be the ones who have to face up to.