Wednesday 31 October 2012

Positive training - Ain't Just Cuddles and Cookies


As the debate between pro-Cesar and anti-Cesar simmers along, reading a lot of the comments from the Pro camp it strikes me that there are a LOT of people out there who truly have NO idea what positive training is about, how it works, what you actually do.

The overriding feeling seems to be that positive training is just being nice to your dog, give it sweeties and fuss and ignore the things you don't like in the hope they will go away.

Well, it ain't.


First of all, whilst positive reinforcement (giving cookies or other rewards for doing the thing you wanted) is a huge part, the basis of, positive training - it is NOT the whole story.

There are, for any situation, any breed, any owner, a range of tools in the positive training toolbox.

The first of these is... duh duh duuuuuuuuh...

Safe management - prevent the problem occurring, prevent the problem getting worse.

Bit boring really, doesn't involve anything massively exciting at all.

As the old saying goes, 'practice makes perfect' - the more a dog performs a behavior that he finds rewarding in some way, that you do not like the better he will get at it.

So there's no magical training wand to wave here - prevent, avoid, manage.

If your dog is aggressive over his food bowl, the very FIRST step you take is to remove his need to display aggression. Yep, thats right - feed him in a room on his own, where no one needs to enter whilst he is eating, no one needs to walk past him.

Immediately, the dog is safe,  the problem behavior is not being repeated twice a day so he is not getting 'better' at it.

If your dog is barking in the back garden - first step, take him out there on a long line rather than putting him out there alone. Now you are in charge, and ready to take action.

If your dog has discovered that being put out in the back yard means he gets to play 'I'm not coming back in when you call me, I'm gonna run around and play chase with you' then again, long line, supervised trips outside only.

If your dog is barking like a mad-dog when someone knocks on the door and hes trying to rush out and do who  knows what to the person at the door - put a sign on the door asking visitors to give you time to answer then put the dog behind a gate or a closed door or put the dog on a leash before you answer the door.

Immediately, you are back in control of the situation, the dog can come to no harm, no people can come to any harm.

One of the huge and frequently missed benefits of doing this is that it gives you AND your dog time to chill out, reduce your stress levels, relax a bit.

Never underestimate the importance of being calm and un-stressed when working with your dog, and never ever underestimate just how badly stress affects your dogs ability to learn!

So, I am managing the problem, but that isn't the fix...


No, that is just your 'step one'. The next step is to sit down, shut up, and think. You have a massive massive brain, the potential of which we actually don't yet understand, but we DO know you can figure out really complex problems and process a lot of data with it.

Get processing that data!

Why is our food aggressive dog guarding his dinner from you? The most likely reason is that he fears he will lose it - food is one of the most important resources to any animal. Fear of losing your food is a pretty big deal.
Your dog has NO idea that you buy food each week or each month, or that the chances of there not being enough to eat on any given day are infinitesimally small. He is a dog - in his world, there is ALWAYS a chance there won't be enough food. That's how he is, and you will not change that.

What you can change are his feelings about you approaching his food bowl - and lets be clear here, whilst all dogs for safety do need to know its ok to give someone something they have, the time to practice that lesson is NOT with his dinner bowl that you gave him, and definitely NOT with a dog already fearful of losing his food.

So hes fearful, you need to make him feel safer - how can you do that? Maybe you can move his food bowl to a quieter place - if his bowl is in a place lots of people walk past frequently, that might be to him the equivalent of you trying to enjoy your dinner on the hard shoulder of the M6!

Think back about what you have taught him in the past - have you, or anyone else, ever taken his food from  him? If you have, have you ever told him off for growling or staring in a funny, stiff way when you did it?
If you have done those things, and many people have due to really, stupid advice, then you have built a food aggressive dog - and you can just as easily un-build that, IF you enagage your brain.

So, we have an idea as to why the behaviour is occurring, the reward is obvious - you back off when he does this - now you need to figure out a way of demonstrating to him that he doesn't have to do this, because he is wrong, you won't take his food.

It ought to go without sayign that from here on in, you WON'T take his food from him - hell, I'll stab  you with a fork if you keep taking MY food from me, and most of us are actually the same. Most humans are food aggressive too!

Now bear in mind - dogs can't count. They are not even all that good at identifying 'more' and 'less' - so whatever you do, he has to be able to understand.

Forcing him to back off his bowl so you can lift it and put more in it, is therefore a really stupid idea, because firstly you confront and challenge him (the very thing he feared in the first place!), and then what you have done that is good, he cannot work out. He has NO idea that you just put 8 more pieces of kibble or a tablespoon more meat into that bowl. All he knows is, it was his, you took it.... the fact you gave it back is completely irrelevant!

So what you DO has to be something your dog can understand.

A dog CAN understand that when someone approaches an EMPTY bowl, and throws a single piece of kibble into it, that there was nothing... and now there is something. 

Ok, so now I am thinking about the whys and wherefores... what's next?

Next - set yourself, and your dog, up to succeed.

Make each lesson, whether you are teaching a brand new behavior or trying to fix an unwanted behavior, SUPER easy to get right.

For some reason humans have a weird attitude that a lesson needs to be taught the hardest way possible for it to work or be valuable. I do not understand WHY this is, because it goes against all the evidence we have.

We already understand that this process works with humans - we do not take five year olds and put them into University and expect them to come out with a PhD!

We don't even take 5 year olds and put them in kindergarten and expect them to learn by ONLY telling them what they did wrong!

We put them in nursery or kindergarten, and we make the tasks simple, and we give them lots of praise along the way for getting things right, and when they don't understand, we demonstrate things in a simple way, over and over until they DO understand. And then we move on and make things just a leeetle bit harder.. and repeat the processes.

That is how we turn five year olds, who like shiny things, lick windows, draw on each others faces and occasionally still poo their pants, into PhD students, neurosurgeons, astrophysicists...

If your dog cannot handle an hours walk around busy streets or to a dog parked filled with crazy running barking dogs - don't take your dog to these places! He clearly can't handle that yet.

Take him for ten minutes walk, round quiet streets or to an empty field - work on the things you can do, improving the bond between you, teaching him to focus on you because you are the source of awesome stuff. If you see ONE dog, reward him like mad and then get the heck out of there.

You will be setting him up to succeed, by recognising what he cannot handle and pitching things at an easier level for him. It isn't cheating, it isn't a cop out its walking before you run, its kindergarten before big school!

Is that it?? It isn't seeming all that difficult to me..

 

Well no, thats not 'it' but you are right, it isn't difficult really. You actually already know most of this stuff you just have to let go of some of the misconceptions about what dogs should do, or should not do, and what they 'need to know'.

For a start your dog has NO need of, and cannot actually comprehend, the human social construct of 'good' and 'bad' or 'right' and 'wrong'.

Pyschologists are pretty convinced that human children cannot do this until quite late, and this is reflected in the laws of various countries and states, where the age range is from 7 to 15..

Your dog is never going to reach the equivalent level of understanding, empathy or morality of even a 3 year old, let alone a 15 year old!

It is a complex issue, because dogs can quite clearly learn that 'No' means 'stop doing that' or 'don't do that in front of HER again', but this is in no way like understanding something is inherently wrong, that it will upset or hurt someone.

Humans actually have huge issues with this too, as is evidenced not just by the huge numbers of humans locked away in prisons, but the real variation from one culture to the next in what is considered 'good' or 'bad' behavior.

So forget teaching your dog that he or his actions are 'bad' or indeed 'good' - teach him that you like and will reward certain actions. Teach him that you will not reward others, using the tools I have already discussed and the ones I'll get onto next.

Learn to read  your dog - listen to him.


Humans have a particular skill for learning languages, so you need to use that.

Really look at your dog, learn about canine body language - train your eye to find the subtle tension in a dogs muzzle or lips, the different set of the ears or tail, whether he is holding his body tense or relaxed.

Watch as many dogs as you can - watch every video with a dog in it and turn the sound OFF so that you are not distracted or mislead by what the people are saying.

Take a note of the context of the behavior, what triggered it, what is going on in the background, what follows the behaviour, how others react to it.

Try to ditch your human ideas about behaviour, and the things you have been told before, because these may colour your view.

A prime example is the 'guilty' look - this is an appeasement gesture designed to diffuse your mood, which may be a slight tension or annoyance or full on rage at discovering 'something'.

A dog cannot connect performing an act some time ago with your anger or upset now, but he CAN connect the presence of a mess, with your behavior.

He can also read your body language far better than even you can - he can pick up when you are pissed off even if you are smiling and saying nice things!

Humans tend to assume the guilty look means the dog feels guilty, and frequently they get mad at the dog, which produces more squirming wide eyed, rolling on the back type behaviour, which confirms to the human that the dog really IS guilty...

It really isn't - guilt, like knowing right from wrong, requires a degree of empathy and understanding of what is important to you that a dog just cannot possess.

How can a dog know that you spent a thousand pounds on that carpet and that urine almost never comes out? How can he know that the boots that were so comforting to chew, that smelt so strongly of you that he felt better about your absence, were a present from a friend and cost hundreds and cannot be replaced?
How can a dog know that you place empty food containers and left over food in the dustbin because its 'dirty' and you want to throw it out. How does he know that 'trash spread around the house' is not the latest look from Home and Gardens?

So read your dog correctly, and don't expect him to be a human. Expect him to be a dog who thinks like a dog, behaves like a dog.

In turn, he doesn't expect you to swap shaking peoples hands and hugging them for sniffing their butts or weeing on their front gardens!

This still doesn't seem like I am doing all that much... wheres the real dog training?


Ok, here's some real dog training.

For every behavior your dog has that you don't like and wish would stop - think up a behaviour you WOULD like him to do instead.

If he jumps up at people he meets, wouldn't it be better if he sat on the floor to greet people?

If he bolts off like a rocket when you take his lead off, wouldn't it be nicer if he went no further than 50ft and checked in with you in case you have a cue for him to follow every few minutes?

If he gets his feet up on the table when you are eating, wouldn't it be nicer if he lay on his bed quietly chewing on a Kong toy?

When you get to this point you can really say you are getting the 'positive training' ethos - you are thinking about your dog, about what he is saying to you, about why he does the things he does, about how you would rather he behaved.

You are setting him up for success by making the tasks, the lessons you want to teach him, easy, and full of praise and reward.

You actually haven't waved any cookies around at all yet!

So, where ARE the rewards?


Ah ha - well there are lots of rewards - some of them are food. Food is a basic need, and you have access to a lot of extremely valuable food - you are RICH my friend, rich beyond your dogs wildest dreams!

Use food to teach new behaviors - use food the way a parent uses money to teach a kid that doing chores is a good thing.
Use it the way an employer uses it with an employee - theres a salary, and there are bonuses.

When your dog is learning something new, something difficult, then the food may need to be right there, like the £5 for cleaning the car.
When your dog is learning something easy, then the food can be less obvious, maybe its right there, maybe its in your pocket, maybe its in another rom. Maybe its super valuable, maybe its just ok - but its a very real possibility to your dog, so he is going to keep working for the chance to earn it. It's a gamble and we all know, gambling can be very addictive!

When your dog already knows a behaviour, and he can reliably do it no matter what the distraction, what the location, what you are doign as you ask him - then the rewards can be even more varied.

The reward might be being sent off lead to go sniff and run around. It might be a game with a favourite toy. It might be a chunk of cheese just because you would ALWAYS like your dog to be cool about passing another dog.

There is absolutely nothing wrong with using bribery to begin with - its a useful tool - but its important to fade out bribery as soon as you can. Phase it out and replace it with the idea of a salary, perhaps a surprise bonus.

But wait - the rewards are even more varied and complex than this.. the rewards are, for you both, that your dog can go almost anywhere with you. Your dog can go for long walks off the lead, your dog can go visit your relatives, your dog is happy and comfortable with a trip to the vets.

You can also use toys as a reward, you can even get your dog to think that you enable sniffing great scent trails or running fast and then those will be rewards.

But what if I really need to stop my dog doing something straight away?

Well.. positive training is NOT without its consequences, it is not permissive training!

There's the other quadrant of learning theory too - negative punishment!

What? You use punishment?

Well, yes. But you need to know the difference.

Positive punishment is all the aversives, such as prong collars, shock collars, hitting a dog, using sounds that have been linked with an aversive, so pulling an example off the top of my head, using a 'tssssst' noise that the dog has been conditioned to associate with a kick or a jab or some other aversive.

Yep, you can condition a dog to flinch as if he has been kicked when he hears a sound, just as you can condition him to prick his ears and salivate when he hears the click of a clicker.

Ah, the magic of classical conditioning!

So, those are the positive punishments - we don't do that. To be effective, the timing has to be perfect, and you still run the risk of 'fall out', ie the dog associating the punishment with something other than what you intended.

Negative punishment is different - it is still a punishment, (defined by something that reduces the frequency of a behaviour), but this time the punishing effect is caused by removing the reward the dog expected.

Sooooo.. if your dog barks when he is outside and every SINGLE time he does this, he is removed to indoors for five seconds, he will learn that barking when outside is not rewarding. He will stop doing it.... but..

I'll come back to the 'but' in a moment.

Negative punishment will not work on its own. You cannot apply it without considering the whole situation. You need a holistic approach, in other words. Consider the breed of dog, consider what his typical day involves, consider the things he finds rewarding and why.

So, that 'but'...

 Yep, there is a but, there always is with punishment. The 'but' here is that the behaviour will ONLY be reduced or stopped IF you address the underlying cause.

Punishment means a behaviour is less likely to occur - but you cannot extinguish a behavior forever using negative punishment.

Back to our dog barking in the back yard - barking is enjoyable on its own, even if nothing else happened, barking can (in the dogs opinion) cause rewarding things to happen (people coming to the property leave, people passing by pass by, small furry animals move quickly...).

So these are his rewards for barking - some of them may be related to the cause of the barking too - if he is anxious about strangers he may bark to protect himself and drive them away. If he is bored he may have accidentally hit upon barking as a fun job to do. If he is a herding dog, he might find the movement he perceives as caused by his actions really exciting, if he is a chasing/sight hunting breed again he might find the movement created really exciting.

This consideration of the problem, rather than the simplistic 'just stop the behavior I do not like' attitude, is actually beginning to provide the answer!


If you find you are applying negative punishment all the time, repeatedly, for a variety of behaviors, you really need to stop, and think.

If those behaviors are getting worse, if there are more and more of them over time, then you have not addressed the underlying problem.

Whilst you are highly unlikely to do any harm to a dog by implementing a time out, by preventing him getting the reward he thought he was getting, the fact that the unwanted behavior is still occurring means that things still are not right, and that may be causing damage.

This is why negative punishment comes last in the toolbox - because to punish anything, you must have the behavior happen before you can do so.

If you are allowing that behavior to happen, then you are missing the opportunities to manage the situation properly, you are not setting your dog up to succeed (in fact you are setting him up to fail), you have not come up with an alternative behavior, you are not demonstrating to him what he should be doing.

So, a time out, whether you remove yourself from the room, or put the dog out of the room, whether you implement it by standing on your dogs lead so he cannot leap at you in a tantrum on a walk, or you bring him in for 5 seconds for barking, is an emergency measure, for a specific problem, that you are addressing in all the OTHER ways as well.

So back to our barking dog again - is he bored? If you don't think so, then review how he spends his day, has he just found that barking is super fun? If so then change what he does during the day, maybe the things you do with him are not the things he would prefer to be doing, and are not as suitable as you might think for him.

Are you preventing the problem from occurring by going out with him - how about blocking his view of things going by, increasing his walks, doing more training with him out there and indoors?

What about pairing the sight of something, or the sound of something on the other side of the fence, wtih a high value reward from you - over time this will teach him that those sights and sounds trigger reward, and he will pause before reacting, giving you time to redirect him to do something else.

So lets have an example of a really extreme behaviour then..


Ok - meet Fred. Fred is a German Shepherd x Alaskan Malamute. He is a BIG guy, he has big teeth, he weighs a lot, he could pull me over.

His problem is that on the lead, he freaks out when he sees another dog. All he wants to do is get over there and muller that dog. He lunges, he screams, he snarls and he redirects his frustrations on to his owner if they are not careful.

Every walk is a nightmare, he has pulled his owner over and dragged her down the road, he has gotten to another dog and bitten it badly. He bit his male owner when he gave him a pop with the leash, hanging off his arm and really meaning business.

This guy is dangerous, to his owners and to those around him!

So what do we do?

Manage/Prevent/Avoid - Fitted with a front fixing harness and a Dogmatic headcollar, with two leads, one lightweight to the headcollar and one heavier to the harness, Fred can no longer pull his owner over.

She holds the leash to the harness slightly shorter than the leash to the headcollar, which means IF he lunges, he hits the harness, not the headcollar. The ONLY purpose the headcollar has, is to turn Freds head away and thus break his direct line of sight to another dog, should one appear.

Next, his owners now only walk Fred for 10 minutes at a time - fortunately their street is quiet and they can also drive to a quiet field a few minutes away.

They replace Freds normal 2 hours of walking per day, with one hours worth of 10 minute walks - thats six walks per day, but only 10 minutes each.

On those walks, Freds owners concentrate on Fred walking nicely on a loose leash, and Fred concentrating on them (they have some chicken in their pockets, but Fred generally knows how to walk to heel unless he sees something, so the rewards are randomly given just to keep his interest).

The main thing Freds owners do is keep their eyes peeled for other dogs - the second they spot a person who looks like they have a dog with them, they about turn and then briskly march the other way.

Fred begins to relax, because he is getting enough exercise (his owners do puzzle solving stuff, clicker training and ball games at home), because six x 10 minute training sessions a dya is HARD work. He is no longer having to deal with the sight of other dogs, so his stress levels drop. His bond with his owners increases because of the increased training sessions and games at home, and because there is no longer two hours of stress each day on his walks.

Setting him up to succeed- after a few weeks of almost complete avoidance, Freds owners begin to reward him whenever he sees a dog. They are careful not to allow him to see many dogs, maybe one every other day, though life is never that kind  so sometimes they see more.

If Fred reacts, they remove him from the situation as quickly as possible. Because no one is popping leashs at him or trying to force him to sit or shouting at him to tell him he is bad, Fred no longer redirects his frustration onto them.

Fred begins to pair the sight of other dogs with a reward, his owners are careful NEVER to get him so close that he reacts - they are setting him up to succeed!

Freds owners continue to monitor his stress levels, and how close he can be to another dog. They look at his body language and take things at his pace.

Teach replacement behaviors -  Fred's owners think it might be a good idea to teach Fred to step behind them and stay behind them, meaning their bodies block his view. They teach this at home first, then in the garden, and then begin to use it out on walks, first when there are no dogs present, then when some dogs are in sight.

All the while they keep up the focus on them, using rewards, and they keep Fred below threshold so he never reacts, and they keep up the counter conditioning, reinforcing the idea that other dogs are NOT something to be scared of or to react to.

Is that it?

 Well - yes, and no. Fred's owners need to keep in mind that their expectations, that he can walk past another dog just a few inches away, and not react no matter what that other dog does, is probably not realistic. Dogs do not like being trapped (and on a leash is trapped), dogs do not like being forced to walk head on at one another as we do daily when we leash them and walk along a pavement.

But what IS realistic is that they can over time change Freds instinctive, emotional, gut reaction on sight of another dog. They can regain control, they can reduce Fred's stress and they can help him become the best Fred he can be.

The process involves no pain or fear, but understanding and careful manipulation of Freds environment and management of his behaviour.

In comparison to using positive punishment, Freds owners do not ever have to get into a physical battle with him. They do not have to guess at a level of aversive to use that will result in him being more scared of them than he is of another dog, (and then maintain that level of fear for LIFE), and the whole process is in fact enjoyable, which is what owning a dog SHOULD be.

It is also worth noting, Freds are not that common - and if every owner understood how positive training really works, very few Freds would exist at all, because the problem could be dealt with before it ever got to the level Fred was at.



No comments: