Interesting article

CorvusCorax said:
At risk of repeating myself, do people believe that no dog, of whatever breed, genetic predisposition, personality type, size, ever needs compulsion?
Click to expand...

I've just googled the word 'compulsion', to be certain of a reply. 'Compelling' a dog to do as we wish will have a purpose at times, but not at others. We don't and can't compel a dog to move away from us and behave in a willing manner. Our 'Purpose Bred' dogs fall into several categories, those 'categories' are various, and this is what I think;

We have those dogs over which we have, and maintain minimal control; English Pointers, Tracker dogs, Hounds (including the Coursing Hounds), and most certainly Sled dogs, amongst others. Complete compliance from such dogs would be counterproductive.

We have those dogs over which we need a 'degree' of control; Sheepdogs, Gundogs and those used for Protection work. Also, and interestingly, there would be those dogs which are used in serious agility competitions. TOTAL control isn't necessary, though the animal must understand that they are part of a team, and that they're not 'out there' to simply please themselves. Such animals, whilst allowed to express themselves can only do so whilst considering their companion.

We also have those dogs which have the sole purpose of being our companions. Compelling them to 'comply' with our wishes is generally, unless we take on the role of a demanding bully, a matter of being done by negotiation. We give-and-take to a greater extent than the two previous 'types'.

Another point; Compelling, or Aversion are differing labels but they will be much the same thing when applied from different directions. For those who would claim that the state of a fully trained dog can be achieved, by the simple expedient of concentrating on what the dog does which is 'Right', whilst ignoring the dogs transgressions, can only have ever had dogs which have the missing ability to reason and think and arrive at conclusions. I have never owned nor known such a dog.

As a flaw in the above points, I'm none too sure where Guide Dogs fit in, but that's another topic!

Alec.
 
Why are we mixing up aversion and compulsion? Compulsion has nothing to do with conditioning or the positive only vs the positive and aversive debate.

Aversives in conditioning are negative reinforcement (an aversive stimulus is removed when a desired behaviour occurs thus making the behaviour more likely to be repeated) and positive punishment ( an aversive consequence follows undesirable behaviour, thus making the behaviour less likely to occur).

Compelling a dog to do something is simply directly making them do it, it has nothing to do with conditioning.
 
CC if a dog is confused about right and wrong behaviour it suggests that the marker has been misapplied, shouldn't have anything to do with whether the trainer used rewards or aversives to encourage/discourage behaviours.
 
Maybe Booboos could talk us through an actual training problem and demonstrate how her science is applied in practice? Anyone suggest a suitable problem? Perhaps the hare chasing scenario I mentioned above? If the science theory works in practice, that should be a piece of cake to do, step by step.
 
Why do you call it my theory? I don't have a problem with mild aversives. Indeed withholding a reward is an aversive if we are going to be strict about definitions so any conditioning trainer will be using aversives in this way. All the clicker trainers I know do use other aversives as well like sudden noises, water and compressed air.

What is the hare chasing scenario? A dog chasing hares? I imagine there are a few different solutions. You could strengthen commands that stop the behaviour like the recall command, so you'd strengthen it by using a long line, the recall game, etc. Or you could condition a behaviour that is incompatible with chasing hares, e.g. I know a JRT who was taught to walk backwards when he saw a cat so that he would no longer chase them.
 
Booboos said:
Why do you call it my theory? I don't have a problem with mild aversives. Indeed withholding a reward is an aversive if we are going to be strict about definitions so any conditioning trainer will be using aversives in this way. All the clicker trainers I know do use other aversives as well like sudden noises, water and compressed air.

What is the hare chasing scenario? A dog chasing hares? I imagine there are a few different solutions. You could strengthen commands that stop the behaviour like the recall command, so you'd strengthen it by using a long line, the recall game, etc. Or you could condition a behaviour that is incompatible with chasing hares, e.g. I know a JRT who was taught to walk backwards when he saw a cat so that he would no longer chase them.
Click to expand...

So, can you please talk us through this? As I assume you've done it, that shouldn't be a problem.

I'd love to have had a pointer that walked backwards every time it saw a hare at a field trial. I was hoping you were going to use the scientific terminology so we could understand what some of your earlier posts were all about.
 
And I assumed you knew what conditioning training was before you said it doesn't work.

OK brief account of conditioning: it is based on a couple of fundamental principles
1. Behaviour that is rewarded tends to be repeated. Rewards are anything that is perceived as such, usually food but could be toys, freedom to run, verbal praise,etc. Note that unwanted behaviour can also be rewarded by mistake, e.g. dog pulls towards the park on the lead, owner walks faster in the direction of the park, for the dog the pulling is rewarded because he gets to the park faster, for the owner it's an arm wrenching experience.

2. Behaviour that is not rewarded tends to be extinguished. By 'not rewarded' I mean omitting the reward from above or associating the behaviour with an aversive, something the dog does not like, e.g. a loud noise, a water spray, an electric collar.

Clarification: what counts as a reward or an absence of reward will vary with each dog and the trainer has to be flexible to find what suits each situation and be careful not to be inadvertently rewarding behaviours he does not want.

3. Conditioning: this is the process by which behaviour is changed by reinforcers, I.e. rewards or the absence of rewards as above. This occurs everywhere and affects animals and humans. E.g. suppose a squirrel makes it onto your outdoor table and finds a bowl of nuts, he is then more likely to search the table the next and may keep searching for weeks and weeks even though you have now hidden the nuts.

The interesting bit is that the change can be brought about by something that has nothing to do with the behaviour, I.e. An association can be made with an entirely different thing. Pavlov noticed this with his laboratory dogs. Dogs salivate when they see food, that is a direct connection between the presence of the food and the reaction, saliva. Pavlov's dogs were salivating when men in white coats walked by the cages because food was always brought in by men in white coats, I.e. The reaction reserved for food, saliva, had now been associated with a completely different thing, white coats. Training takes advantage of this by pairing behaviour with commands through the use/absence of rewards, so that salivating is now paired with white coats.
 
There are four ways to affect behaviour:
1. Positive reinforcement: the reinforcement bit means that you get more of the behaviour, so this is giving a reward to get more of the behaviour. E.g. If a dog sits and you give a piece of chicken the dog is more likely to sit again for another piece of chicken.

2. Negative reinforcement: behaviour is increased when an aversive is removed or avoided. E.g. To get a dog to sit you pull its collar up and push its bottom down, when he sits you stop the collar pressure and take away your hand, I.e. Life gets easier for the dog when it sits (for anyone familiar with dually head collars this is exactly how they work).

3. Positive punishment: this aims to reduce behaviours by following them by an aversive, e.g. The dog barks and receives a shock from an electric collar so it is less likely to bark again.

4. Negative punishment: reducing behaviour by taking away some reward, e.g. The dog barks so the owner does not throw the ball.

Positive training uses 1 and 4. Some positive trainers use mild aversives in 3, e.g.loud noises, water, air. The article is disagreeing with trainers who claim that all training can be achieved through 1 and 4 without ever using 2 and 3.

The role of the clicker: the clicker is just a convenient way of marking behaviour. Marking behaviour is saying this is what I wanted and this is what you are being rewarded for, the reward is to follow. One could just say "good girl" but the click is faster, neutral and easier for many people.
 
Suppose now you decide to use 1 and 4 in your training, how do you get behaviour in the first place so you can reward it? There are a variety of ways:

A. You can lure the dog into the desired behaviour, e.g. hold a piece of chicken level with a dog's nose, move it upwards and as the nose goes up the bum goes down, you have a sit, click and reward it.

B. You can exploit naturally occurring instances of the behaviour. Puppies are scared of the big wide world and are likely to come back to the owner for reassurance, click and treat every time a puppy is near you and you have the beginnings of recall.

C. Exploit the dog's natural responses. If you place your palm next to a dog's nose he is likely to sniff it. Click and treat the moment of contact and you have the beginnings of targeting. Targeting is extremely useful for training other behaviours, e.g. place your hand at your side, ask the dog to target it and walk on...you have heel work.

D. Shaping: brilliant tool for complex behaviours, basically your break down the behaviour into small parts and reward in stages. If you want to teach roll over start by rewarding a down, then a down with the head looking over the shoulder, then with a shoulder collapse, the with the left hind and front legs coming off the ground, then with the dog on her back and finally the full roll over.

E. Delayed gratification: to get more of a behaviour delay the click. For a sit stay, ask for a sit and slightly delay the click, extend the time of delay but always vary the delay. You can work on each component of the sit in turn, I.e. sometimes working on the duration, sometimes working on only square sits, sometimes working on really fast sits.

So how do you teach a dog to run backwards when it sees a cat? First you teach the dog to run backwards. How you do this depends on the dog. I taught one of mine using targeting. She was asked to follow my palm forwards and when this was well rewarded I asked her forwards and a tiny bit backwards. On the first step back I clicked and gave a jackpot (a super reward indicating a particularly desirable behaviour). Then repeated asking for more steps back and the just steps back not forwards. In time I refined the hand signal and added a word command until every time I gave the command she reliably and quickly stepped backwards.

Adding the cat: this has to be done first in a controlled environment where you can control the presence of the cat. The aim is to replace the command for run backwards with the presence of the cat. If you think about it the words 'go back' are as random to a dog as the presence of a cat, it doesn't matter what you pair, it's the conditioning that matters. Pavlov's dogs were conditioned to salivate at white coats, but the white coats were not relevant; had the men been wearing pink coats or hats or whatever the dogs would have been conditioned to the pink coats, the hats or whatever. So you do the same with the cat as with anything else, you do what is necessary to produce the behaviour in the presence of the cat so you can reward the behaviour and strengthen its association with the cat. Here you are going against what the dog would naturally do do the conditioning process will take longer and you have to work on different aspects of it, e.g. once the dog has been conditioned to go backwards in the presence of a,stationary cat, he has to relearn with a moving cat and then a running cat. Repeat often enough and it becomes second nature.
 
Alternatively you can google the three billion sites explaining conditioning, animal conditioning and dog conditioning for more details and examples. Kikopup goes to the trouble of even posting videos of all the training examples.
 
What about dogs who self-reward? To whom even just staring at the cat is much more gratifying than any food that can be offered?
Or a serious livestock or vehicle chaser, who's life is in imminent danger and you don't have time to go through all the steps and repetitions?
 
Booboos said:
Alternatively you can google the three billion sites explaining conditioning, animal conditioning and dog conditioning for more details and examples. Kikopup goes to the trouble of even posting videos of all the training examples.
Click to expand...

Oh dear, I seem to have ruffled someone's feathers! D:

So, I was correct when I used the word "theory"? You haven't actually trained a dog that way? So, maybe it isn't theory after all but hypothesis?
 
JillA said:
Dry Rot, I'd actually like to know, in detailed steps, how you train a fast recall without a reinforcer of some kind?
Click to expand...

I haven't a clue. I think you are confusing me with another poster. I've always managed to effect a fast recall, even when hunting a pack of 14 couple of hounds.
 
Dry Rot said:
Oh dear, I seem to have ruffled someone's feathers! D:

So, I was correct when I used the word "theory"? You haven't actually trained a dog that way? So, maybe it isn't theory after all but hypothesis?
Click to expand...

First of all you are welcome! While it took me a long time to type all this out for you and it was readily availablr elsewhere, your polite acknowledgement of my efforts and your gratitude made it all worth while. I don't know you in RL but you have often intimated how your great age has given you great experience; you should add good manners to your other worthwhile attributes.

Have I trained what? I have done a six month clicker trainers course with Elizabeth Kershaw, I have volunteered for a few years as an assistant trainer atlocal dog training clubs, I have trained 4 dogs successfully and failed miserably with one, I have competed in obedience but lower levels and clicker team challenge competitions and done agility for fun. I have trained a JRT not to chase squirrels by strengthening his recall (high rewards, long line, recall game).
 
CorvusCorax said:
What about dogs who self-reward? To whom even just staring at the cat is much more gratifying than any food that can be offered?
Or a serious livestock or vehicle chaser, who's life is in imminent danger and you don't have time to go through all the steps and repetitions?
Click to expand...

I don't have a problem with aversives. If there is no reward that trumps the behaviour, if there is no way of interrupting the behaviour, no way of avoiding the problem situation and no underlying physical issue then aversives may well be the answer. I have used a sound aversive with great success with one dog and tried sound, water and air aversives with no success with another dog (I wasnrluctant to use shock collar with this latter dog as she is dog aggressive and in think this aversive has been known to make dog fights worse, plus it's notsomething I am trained in using).
 
Booboos, I thought your latter responses were a lot more helpful than some of the scientific jargon that I have found online and I definitely appreciate the time you took to type it out. Bravo :)

As to your last point in post #68, that's the whole tenet of classical conditioning, isn't it? A link between two things that were previously not linked in any way. My dog used to drool when he saw a clicker, which is another reason why I no longer use it, as that wasn't the association I wanted him to make :p

Re compulsion, I see compulsion and aversion as the same thing really, perhaps erroneously on my part, when the dog either MUST do something or MUST NOT.

And if others took the time and showed restraint with an aversive such as an EC as you did, perhaps they would not be such a vilified piece of equipment! And yes, like a lot of toold, they can act as a stimulant as well as an aversive, as you say, if used in the wrong way and without being introduced properly as part of a long term training plan.
 
Last edited:
CorvusCorax said:
Booboos, I thought your latter responses were a lot more helpful than some of the scientific jargon that I have found online and I definitely appreciate the time you took to type it out. Bravo :)

As to your last point in post #68, that's the whole tenet of classical conditioning, isn't it? A link between two things that were previously not linked in any way. My dog used to drool when he saw a clicker, which is another reason why I no longer use it, as that wasn't the association I wanted him to make :p

Re compulsion, I see compulsion and aversion as the same thing really, perhaps erroneously on my part, when the dog either MUST do something or MUST NOT.
Click to expand...

I would see compulsion as being in line with your definition, but aversion as something which is designed to associate an action with something the dog dislikes. So a loud noise, for instance, or a squirt of water that comes when the dog starts eyeballing something to chase, that kind of thing - that's what I would think of as 'an aversive'.

BTW, How Dogs Learn by Burch and Bailey is a really good book for anyone interested in getting a clearer idea of how some of the scientific theories of learning apply to dogs. If anyone is interested.
 
Not really an answer in the spirit of what you were asking of others DR - if you throw down a gauntlet you must expect to have to live by the same tenets. You dismiss everyone's ideas of how animals learn abut are not prepared to tell us how you think they learn. Nuff said really.
 
In fairness, there are a lot of pups around at the moment in the circles I move in and there is very little time spent doing any formal training with them, just conditioning them to believe that the handler is the be all and end all and being around them is the best place to be.
Until you have that focus and attention you can't train much else, not in a solid fashion anyway.
Running up to other people and other dogs is fine, but they don't get played with or fed or fussed or 'rewarded' for leaving. It's tempting to fuss cutesie wootsie puppies when they run up to you but you are not helping, down the line.
Pups of course have to be socialised and learn to be around other people and dogs and be comfortable with them, but not to see them as a big attraction - that is what mum or dad is meant to be.
 
Booboos said:
First of all you are welcome! While it took me a long time to type all this out for you and it was readily availablr elsewhere, your polite acknowledgement of my efforts and your gratitude made it all worth while. I don't know you in RL but you have often intimated how your great age has given you great experience; you should add good manners to your other worthwhile attributes.

Have I trained what? I have done a six month clicker trainers course with Elizabeth Kershaw, I have volunteered for a few years as an assistant trainer atlocal dog training clubs, I have trained 4 dogs successfully and failed miserably with one, I have competed in obedience but lower levels and clicker team challenge competitions and done agility for fun. I have trained a JRT not to chase squirrels by strengthening his recall (high rewards, long line, recall game).
Click to expand...

Oh, I am the first to admit that I have zero social skills. But I am not often deliberately rude. Dog trainers, I've found, tend to be blunt. "Sit" means "Sit". It is not a polite suggestion to have a committee meeting! So I say what I mean even with humans. Sorry if you find that offensive.

But you seem to have missed the point entirely. (Incidentally, rather than typing that lot out and, as you say, it is all available on the Internet anyway, wouldn't it have been more sensible to cut and paste?).

What I was hoping for was a point by point analysis of how you tackle a specific problem, like hare chasing, with your scientific explanation of each stage. A hare gets up in front of your dog, your dog chases….Now, what do you do next and what is the scientific term? Personally, I would hang my coat on a fence post next to where I parked the car and go home and have my tea. Ten to one, when I returned an hour later, the dog would be curled up asleep next to my coat. :) I suppose even that will have a scientific term to make it less easy for us mortals to understand.

Just to repeat, The standard of dog training here in the UK is appalling and I do not think trying to explain things with a lot of long winded scientific terms (which even the scientists cannot define with any degree of consistency!) helps the situation. Normal people don't talk like that.
 
DR there is nothing science can do for you when the dog runs off chasing a hare; either the dog is trained well enough not to run off or you retrain later. But your question is a bit like saying "I jumped off a cliff and never flew therefore the science of aeronautics is rubbish".
 
CC attention to the handler is generally important in clicker classes. Beginners classes usually cover attention exercises, e.g. dog maintains attention on handler even when the handler waves food about, and some unwanted behaviours, like pulling on the lead, are ignored until the dog refocuses on the handler (which is then immediately rewarded).

If compulsion means making behaviour happen that tends to be against clicker principles. So let's say you want to reach a sit to a puppy, you lure him into position with food click and treat, you repeat three times and then you do nothing and wait. It is very important to wait for behaviour to be offered a the dog has to figure this out himself. Clever breeds, like collies, can identify the behaviour after one reward. Less clever breeds can take much longer, I've see Afghan hoiunds take weeks to get it. Advanced dogs will know to offer behaviours on a trial and error basis until they find the one you will reward.
 
Dry Rot said:
Just to repeat, The standard of dog training here in the UK is appalling and I do not think trying to explain things with a lot of long winded scientific terms (which even the scientists cannot define with any degree of consistency!) helps the situation. Normal people don't talk like that.
Click to expand...

Nor does any decent dog training instructor. If I'm explaining to my class what an extinction burst is, for example, the words 'extinction' and 'burst' will not cross my lips. INstead, I explain it in terms of a vending machine, so:

Say your dog jumps up, and you don't want it to. One way to try and deal with this is to ignore the dog. (There are other ways, but we'll go with ignoring as the chosen method in this instance, as being most appropriate for this purely hypothetical dog. In a real situation, I would give them a couple of other tools too, but this is still how I would explain what this particular phenomenon is).

If you suddenly start ignoring the dog when before you've given it attention, the problem may actually get worse before it gets better. It's like this - imagine you have fifty pence in your hand, and a vending machine has magically appeared in this field. So you bung your fifty pence in and select a calorific treat of your choice. This has worked every other time you've used a vending machine, so you fully expect to be adding to your waistline shortly.

But oh dear - the machine has taken your money, but is not spitting your sweeties out! What do you do?

Well, you might try pressing the buttons again. Then you might try pressing them really hard a few times. If that doesn't work, you might try knocking or shaking the machine. Maybe you'll even give it a good kick. Then, eventually, you'll give up and wander off.

Well, as far as your dog is concerned - YOU are the vending machine. Jumping up is the fifty pence. And your attention is the lovely treat selected. So if jumping up has always worked before, your dog is likely to give it another go. Maybe if that doesn't work, they'll have a good bark at you. But if you stick to your guns, they will give up. (Of course, they might try something else to get your attention instead, like running away with something they shouldn't have. But they won't be jumping up).

The scientific term for that behaviour is an extinction burst. You don't have to use that to explain the concept - although some people do like to know some of the science behind what's being said - everyone learns differently. But being aware of its existence is worthwhile either way as it means you can reassure owners that what is happening is normal, and not due to their dogs being extra naughty, or them being bad handlers - both of which are real anxieties that people in dog training classes have.

That is a real example, which I have successfully used in a real setting, which was understood by the people listening.
 
You must log in or register to reply here.
Top