Fallout in aversive training: a post for trainers

It’s not often I write a post (or two!) for trainers alone – most of my site is guidance for owners of dogs who they have adopted. Often it’s a place where I can send people who ring me who need something for those minor issues: humping, jumping, house-training and so on… things where there are well-established protocols to change behaviour quickly by changing the environment or tackling a single, isolated behaviour.

There’s a reason I have these posts for minor issues. Kathy Sdao talks of doing triage over the phone. I think a lot of dog trainers and behaviourists must do this. You do intake interviews, sure, and I have a 13-page intake assessment form, but those first phone calls go like this:

“Hi Emma…. I’ve been given your number by so-and-so. My dog’s pulling on the lead. What can I do?”

or:

“Hi Emma… I’ve just picked up a new dog and he’s humping my own dog. I’ve tried X, Y and Z but he’s still doing it and my dog is getting stressed.”

or:

“Hi Emma… I’ve just got a new dog and she’s peeing in the house. Can you help?”

Lots of my phone calls go like that. And that’s fine. These are the people you can help in an hour consultation as long as they implement your advice. Cheap and easy hand-holding with tried-and-trusted techniques.

And I don’t worry about hidden blind spots or things clients aren’t considering.

I find people usually start with the most severe problem in a couple of seconds of the call, and although there will be an enormous amount of back-story to fill in which may take an hour or so, even for these ‘simple’ cases, it’s not something that needs an enormous skill to address in non-aversive ways.

Then you get some that go like this:

“Hi Emma… you’re my last resort. Our bouvier bernois has bitten my husband.”

or:

“Hi Emma… please help. We can’t cope any more. Our labrador is growling at our children.”

or:

“Emma… can you take my dog? Otherwise we’re going to have to have it euthanised. It’s fighting with our other dogs and I don’t know what to do.”

And increasingly, I have calls that go like this:

“I’ve got a dobie/shepherd/rott that we’ve been taking to club, but he’s become really unpredictable at bitework, getting more and more aggressive in general, and the club said we can’t take him any more.”

Those are the ones that currently take up a lot of my time. It’s not that there are more of them. It’s just that I’ve got a bit of a reputation as a Last-Chance Saloon following a dog I worked with. I did an assessment of a German Shepherd who’d been trained for “protection” work and who had become mean and uncooperative, unpredictably growly and the club had banned the dog from going anymore when it had attacked the club owner. These clubs are surprisingly popular in France. Malinois and GSDs are the top two ranking breeds in terms of popularity, and a surprising number of those end up at ‘cynophile’ clubs. Whilst I don’t have any problem with obedience heeling, finding strangers to bark at and tracking scents, putting bitework in there needs to be done with care, and it’s invariably this bit that goes wrong. Trainers in France require no qualification and these clubs are renowned for high levels of confrontational training. I’m making no comment and no judgement about how these dogs are trained and the club these dogs are coming from is pretty easy-going in terms of what’s acceptable training-wise. I don’t even really know exactly how the dog in question had been trained, and to a large degree, it didn’t matter. I immediately asked the owners to do a full blood and x-ray/ultrasound work-up under sedation at the vet hospital in Bordeaux. They did so and it revealed quite significant hip dysplasia as well as problems with his spleen. Medication, surgery and 3500€ later, he was almost a different dog. Sure, he needed a bit of counter-conditioning and a complete change in pastimes – no more long walks and definitely no pretend “protection” work – but his owners really did treasure him. Well, kind of. Didn’t stop them sending him out to stud, even with his hip score, but that’s a different post altogether. Anyway, after that, every few weeks, I’d get a call from owners who’d been kicked out of club. The second is a malinois that I’m still working with. The third was a GSD in such poor physical health that the owners wouldn’t foot the bill and put him to sleep. To be honest, when the vet showed us the x-rays, I was surprised the poor dog was still walking and it was the kindest thing altogether. Subsequently, I’m the girl who puts an “off” on a bite – if it’s possible with the dog at all – and I’ve got a pretty good life programme set up for those with whom it is not possible. That programme of lifetime management does start to rebuild the bonds, re-establish trust and make bites less likely, but I’m not one to say it works until the dog is dead of old age and there hasn’t been a bite between them coming to me and them dying. Sadly, a lot of what I advise is managing the environment to keep the dog – and the humans around it – safe. But if you love your dog and euthanasia is unthinkable, then I can help you create a sanctuary-style life for your dog that will minimise the risks. Most of the dogs I work with privately over the long-term have significant health issues that are contributing to aggression, or significant genetic factors. Just so you know where I’m coming from.

Add to that the other ‘quicker’ cases: over-aroused big dogs who are nipping and/or pulling, uneducated little dogs who guard beds and resources, dog-dog aggression, dog-human aggression, and most of my work is with canine emotional issues. By that I mean that they are either fearful or over-aroused. Many have learned to be aggressive as a very effective tool in their behaviour repertoire. All, and I mean all, have been subject to aversives – be they ‘mild’ like water sprays and spray collars, or more severe – from chokes (prongs are not really well-known in France, but choke collars are everywhere) to electric collars and physical manipulation, stare-downs, rolls, restraint or hitting/kicking. and these have not worked. That’s why they are with me. If these methods worked, they would not be calling me for help.

So, not only have the aversive methods failed, which is one potential hazard of punishers, but you get to see all the well-known behavioural ‘fallout’. These dogs come to me as the poster dogs for why it’ll be a cold day in hell before I use them. They are all I work with. My conversations go like this…

“So, what methods have you tried already to deal with this behaviour? [Insert aversive of choice] … Okay…. and what happened when you tried that?”

Now I’m pretty non-judgey. Plus generally speaking, my clients have no idea of the consequences of their actions or how their behaviours may have contributed to the situation. I might be cross at the trainer or the TV programme or the book they got the advice from, or the breeder, the neighbour or the friendly neighbourhood know-it-all, but my clients have dogs, not a long-term interest in reading scientific manuals and watching endless DVDs about dog behaviour. In my opinion, their ethics are severely out of whack, but I keep that to myself. I like to think that if someone had offered me that advice, I’d have gone, “yeah…. no thanks!” even if it had been the most effective thing on the planet, but I’ve been bonkers about animals since I was a nipper. Plus, I like my clients’ dogs and I want a chance to work with them. That’s not going to happen if I say, “Are you effing kidding me? You thought that forcing your dog to the floor and holding him there was going to work? What planet were you born on??! How about I try that on you?” etc. etc. etc.

But through the owners’ own admissions, they have tried some pretty hideous and gruesome training techniques – and those methods have failed.

Now I realise that kind of statement can attract the comments of trainers who use aversives who say that the owners must have been doing it wrong. I don’t know enough to know how you should punish animals for emotions, but I do know that there is fallout, even if you do it ‘right’. I don’t, by the way, think there is a right way of doing it, for the following reasons. Whether you are using punishment or escape/avoidance methods, the potential fallout is the same whether you do it ‘right’ or you make a hash of it.

The first risk of aversives in learning is that an emotional risk of anger and aggression, which then interferes with your behaviour modification.

Herron et al. (2009) documented the reactions of dogs to a range of aversives (direct and indirect aversives as Doctor Jim Ha calls them) including yelling, staring, water sprays and also those such as hitting or kicking. Some of these methods are almost a 50:50 for aggression as a consequence. Kick or hit a dog and 41% will respond with aggression. But some methods which you’d consider less intrusive, such as staring, also had a high rate of aggressive response. Growling at a dog is almost as likely to elicit aggression as kicking or hitting. Staring has a 1 in 3 chance of causing aggression.

That is one thing that ‘balanced’ trainers either don’t know, don’t understand or don’t explain to owners.

I picked up three dogs this year on my books who had escalated aggression in consequence of stare-downs. One of the dogs was up for euthanasia as a result of their new response, not as a result of the initial behaviour. That is aggression that is a direct result of an intervention – caused by the programme – not by any other environmental change.

That anger and aggression, by the way, may be to the handler, or to others in the environment. You might not see a rise in anger directed to you personally, but the dog can easily redirect. Redirected bites make up a good number of the bite cases I do… dogs who can’t get to a target and turn to the nearest available biteable thing. Often that is another dog in the household. Sometimes it can be a child. Anger and aggression are not the only emotional fallouts. Aversives run the risk of increased anxiety and fearfulness, so I see a number of dogs who self-harm through tail biting or chasing, or flank/foot licking or chewing.

The second risk of fallout is a reduction in all levels of behaviour.

Punish a dog or use escape/avoidance methods and find yourself getting less behaviour in general.

I’m reminded of an A-level class with a teacher who would constantly tell us off, so we gave up responding. We stopped offering responses, our homework was minimal but on time, we arrived on time and we left on time. Being late and not handing in homework were also punishable, otherwise you can bet your bottom dollar we’d have stopped offering those as well. We engaged as little as we could.

Now some trainers (and many owners) like dogs that behave less. They do less. A shut-down dog is one step up from an automaton, and that is a dream dog for some people. Not so good if you want to show them, do classes with them or enjoy them. Plus, you find that there is a reduction in behaviours around the primary handler/punisher and a spate of those canine behaviours out of sight. Barking, digging, chewing and roaming are all things the dog does out of sight, as all they have learned is not to do the behaviour around the person dishing out the aversive. Dogs are dogs. That behaviour will more than likely pop out somewhere. And do you want it where you can’t see it (and you can’t intervene or address it) or where you can see it and do something about it?

The third risk of fallout relates to escape and avoidance.

If a dog connects something unpleasant with you, then you’ll notice they want to be around you less. Sometimes that’s inadvertent. Last week was really cold here. I called Heston back and he stood on a frozen nugget of mud that got stuck in his paw. He came back and I picked it out. But the next time I called him, I noticed the latency of the behaviour was less strong. Somehow me calling him and that nugget of mud had become attached, and I noticed hesitance in his recall even though those things were not connected. Avoidance of aversives is well known. It is the principle upon which electric fences work. Businesses use it to create items that make dogs avoid places because of the consequence. This is why someone who uses aversives with a dog may find their dog reluctant to approach. That sucks for recall, even with the best and most well-trained dogs, and I know a good few owners who can’t let their dog off-lead at all, not because the dog goes off exploring the environment, but because the dog leads them a merry dance to get them to come near them at all. Taking dogs to the vet in the shelter is a good example of this. We have to sometimes use quick and dirty muzzling with closed muzzles, and the fallout of this is enormous. The dog wants nothing but to be away from handlers and vets after that.

This leads us to the final risk: generalisation and specificity.

Fear is a great teacher.

Here’s an example. I got in a car accident the other year. A car came towards me and cut right across me so I clipped it. Now I generalise like mad. I’m nervous around cars coming in the other direction. I avoid the place the accident happened. I hate the type of car. I’m on high-alert when I see cars of that colour. I won’t drive into that town on market day. I hold my breath every time I see the spot it happened. I also generalise about things that didn’t happen: I’m nervous around people pulling out of junctions or who look like they won’t stop. My general anxiety when driving is much higher than it was, even in situations that have nothing in common, like driving on motorways.

Animals can do the same and generalise that fear, connecting unconnected things and finding causation in accidental correlations. Our primitive brains handle startle responses inappropriately well and our “mum-mind” or “parent-mind” kicks in, our mind’s airbags, helping us to avoid everything our tiny lizard brains think is vaguely connected with that moment. It puts those airbags on a hair trigger and they keep popping out inappropriately and far, far too often. When something happens again in the same or related circumstances, our mum-minds go into overdrive.

This generalisation is why I don’t want to use products or procedures that build on fear in a dog. Most of the dogs I’m working with are already afraid of one thing or another. Why would I want to put that emotion into a dog when all my work is about taking it out?

You simply can’t predict what dogs will generalise or connect to the situation.

Likewise, you can’t know if they will actually generalise at all. With a reactive dog, I’m very cautious around places in which we’ve run up to unexpected things, but both the dog and I are blithely indifferent in other situations and I don’t exercise the degree of caution that I should. So aversives can cause generalisation and they can cause a lack of generalisation. This is the bugbear of all trainers: what we teach in one place doesn’t always equate learning in another. Why stuff works in a class and seems to have been forgotten in the home.

And worse still, we don’t get to pick whether a dog gets really general about what caused the emotion, or whether they fail to generalise at all. I’d like Heston very much to generalise that all people are nothing to be afraid of, and I seem to spend most of my life teaching him that this one is fine, that one is fine… they’re all fine. But he doesn’t generalise the good stuff. He isn’t actually very specific about the bad stuff either. I wish it was ONE set of circumstances that set him off, but I’m buggered if I know which people he’ll like and which he won’t. Like Lidy my shelter project dog at the weekend. Aggressive responses with raised hackles to one woman. Non-aggressive responses and submissive behaviours to another. I could say “it’s a hat thing”, but they were both wearing hats. I could say it’s a height thing, but both were the same height as me. I am absolutely buggered if I can see a pattern in who she’ll have raised hackles with and who she won’t. So when you use fear (P+) as a method of teaching, you don’t get to choose if the dog will a) start generalising wildly about non-connected things that were present at the time, or b) never generalise and only ever ‘obey’ in a very particular set of circumstances. That’s too unreliable an approach for me.

Another side-effect we don’t often talk about and isn’t always part of the ‘canon’ on what constitutes fallout from aversive methods is that of transfer behaviour. Say your dog is seeking attention, so it paws you. You punish it for pawing you. That basic desire hasn’t gone away, even if the behaviour has, and so it may be expressed in other ways. You might then find the dog jumping up. So you punish jumping up. Now the dog can’t get your attention through pawing you or jumping. One day it humps a cushion. You stop what you are doing and give it what it wants. Even if that is a good shouting at, your dog’s basic desire for your attention has been satisfied and your dog has learned a good way to get you behaving in a particular way. Emotions and functions will out. Whatever your dog feels like it needs to do, why the behaviour started in the first place, has not been addressed, and that emotion or need will find a way, somehow. Thus you end up like ‘whack-a-mole’ trying to nip a range of behaviours in the bud one after another.

These are all side-effects, consequences and fall out that trainers should be aware of when using punishment, aversives or coercion. Certainly, they are every single reason I have for not using aversives in my work. Not least that the dogs I work with have already had aversive behavioural modification which has been ineffective and has left the owner with a dog who has one or more of the behaviours outlined above. It hasn’t got rid of the problem and in fact, behaviour has deteriorated with one or more of the undesirable consequences outlined above. Generally speaking, this is the majority of my caseload. Behaviour was significantly unwanted to warrant an intervention. The intervention has caused problems of its own and the original behaviour has increased in intensity, duration or frequency.

Thus, the negative fallout of aversive, punitive or coercive methods is well established – certainly ‘force-free’ groups and forums are liberal in sharing the fallout of non R+ training methods. I believe that these side-effects should be shared with all owners who choose to go down the route of applying pressure, applying an aversive stimulus or using escape/avoidance training with their dog. You’ve got to know what the side-effects might be.

But then, I also believe that it shouldn’t take science to make it patently obvious why something is ethically and morally heinous. Animals are emotional beings. Punish an animal for fear or for hyperarousal and you will see that not only is it ethically unacceptable given many less aversive approaches available for every single situation, but it is also a pointless exercise. Punishing an emotion won’t stop the emotion. It will only increase it or suppress it. Not consequences I want.

But is positive reinforcement the panacea it is claimed to be, and is it entirely without fallout? Many trainers would have you believe that you can’t go wrong with positive reinforcement  – it’s safe in the hands of non-experts – and yet if you ask me, it is as alarming to consider R+ to be without drawbacks as it is to consider P+ or R- to be without drawbacks.

And if we don’t consider those drawbacks, all we are doing is sending our clients with their dogs right to the doorstep of “balanced” trainers, saying “well, I tried that positive reinforcement thing… and that didn’t work!”

If it’s frustrating for P+/R- trainers to use their methods without explaining the fallout, it’s no good entering into R+ training without being mindful of the fallout of that too. If we aren’t as clued up on the side effects of our methods as we should be, it will not only cost us clients who will then walk right over to ‘the dark side’ but it could also have a negative impact on their dogs and their behaviour.

In the next post, I’ll explore the potential fallout or side-effects of positive reinforcement, along with some ways to address those possible issues.