Training A Reliable Recall

[Erny]

IMO the dog would only anticipate a correction if the correction was first delivered in the absence of good timing (ie the dog was unable to "pair" the correction with the "behaviour" and therefore was uncertain what action is/was required to avoid it).

Many people get the impression that "avoidance training" achieves a dog that then performs with varying degrees of cowardly body behaviour because it is 'scared/concerned' of receiving a correction. Timed well, the dog learns very quickly what action/s to avoid and what actions achieve good consequences. This is in line with the way dogs learn. IE That every action has a consequence.

The 'teaching phase', which should not include corrections, builds knowledge in the dog - teaching it the required/good behaviours - exhibition of which will be to its advantage. With this knowledge behind it, the 'training phase' then leaves opportunity for the dog to discover what the consequences of (a) exhibiting the desired behaviour and (b) exhibiting the undesired behaviour. This sequence makes it easy for the dog to make choices of behaviour which result in favourable consequences. The process (applied properly and appropriately) does not take very long at all and results in a dog who is confident that his/her actions are to its advantage - hence it happily responds in the manner we desire.

What this form of training also achieves, however, is a dog who learns to rely on direction from the handler. This hesitation/dependancy is is not desired for a working dog where it is necessary for the dog to make decisions independently of the handler.

[Working_Setters]

IMO the dog would only anticipate a correction if the correction was first delivered in the absence of good timing (ie the dog was unable to "pair" the correction with the "behaviour" and therefore was uncertain what action is/was required to avoid it).

If a dog has learnt a behaviour (stop to whistle) and has learnt that when they obey there will be praise and when they don't there will be punishment, then I think the dog anticipates prasie everytime it obeys and anticipates punishment everytime it doesn't obey. With regard to the timing, that's fine if the dog is on lead or on e-collar, but if dog is 50 meters away and no e-collar, timing is going to be a problem. That's why gundog trainers take their dog back to "the scene of the crime", repeat and enforce the command (generally a sit whistle the dog blew off), to help the dog pair the behaviour and the punishment. This can all be done much more efficiently these days with an e-collar, but when you're not using one, then during the time I'm covering that 50 meters my dogs know if praise or punishment is coming.

What this form of training also achieves, however, is a dog who learns to rely on direction from the handler. This hesitation/dependancy is is not desired for a working dog where it is necessary for the dog to make decisions independently of the handler.

I don't agree with the extent of this generalisation. I think most/all working dogs need an "off switch", a command that is completely handler dependent. For my dogs it's the stop whistle, I don't care what else is going on, I don't want my dog to think about it, I just want their butt on the ground when they hear that whistle. But in many other aspects my dogs are independent workers, they have to be as there is nothing a human can teach a dog about scent and scenting. Same with the police dog video (Rudy, I think) that was posted here recently. Dog was sent on perp, 2/3 of the way there the dog is recalled by handler, dog's response needs to be reflexive, it is up to the handler, not the dog, as to who gets bitten. Same with herding, sniffer and guide dogs, all needs to have a command the over-rides everything else, to allow the handler to stop/"turn off" the dog.

[MonElite]

during the time I'm covering that 50 meters my dogs know if praise or punishment is coming.

And even if you were Marion Jones that would mean about 6 seconds later.

[m-j]

Also why on earth would you want a dog anticipating correction all the time - wouldnt it be counterproductive?

What Erny said:))

I was talking about stopping one behaviour, where the consequences of doing that behaviour is a bad thing for the dog ie if my dogs chase sheep the farmers going to dish out a far worse adversive than what ever I can . You bet ya I want that dog to think if I chase those sheep I will get nailed:)) If they don't the consequences could be fatal or they spend the rest of their life on lead, the only dog I have had to do this to thought his throat had been cut as he had spent years walking around here off lead.

And even if you were Marion Jones that would mean about 6 seconds later.

Thats why you use a conditioned reinforcer the message is fairly clear immediately (as long as your timing is good) even if you take 6 sec to deliver whatever you have promised the dog.

cheers

M-J

[Working_Setters]

Thats why you use a conditioned reinforcer the message is fairly clear immediately (as long as your timing is good) even if you take 6 sec to deliver whatever you have promised the dog.

cheers

M-J

On the subject of conditioned reinforcers (CRs), how long is too long, b/t the CR and the Primary?

Also it was suggested to me the other day that it's OK to use multiple CRs (in this case I was talking to a clicker trainer, so it would be click, click, click) then Primary much latter, say up to 5 min later. It was suggested that the dog was "counting up" the CRs and so long as the Primary (food) was appropriate to the number of CRs (mores CRs = more/better food), then this was OK. Any thoughts? Anybody train this way?

I'm told that there are remote clicker collars, you push a button on a electronic box and the dog's collar makes a clicking noise, anybody seen/used them? I guess to be successful as a training tool, the dog must learn that CR now means Primary latter - which I' OK with, but if you're giving multiple CRs and not giving the primary until 5 min latter, I'd have thought there was a danger the CR would become unconditioned (ie multiple clicks now, reward 5 min later, there is a danger that in the dog's mind click nolonger = reward). Anybody got any experience with this one way or the other?

[Rusky]

Also it was suggested to me the other day that it's OK to use multiple CRs (in this case I was talking to a clicker trainer, so it would be click, click, click) then Primary much latter, say up to 5 min later. It was suggested that the dog was "counting up" the CRs and so long as the Primary (food) was appropriate to the number of CRs (mores CRs = more/better food), then this was OK. Any thoughts? Anybody train this way?

I have been thinking about this since a few weeks ago when I was with a girl who trained her horses grand prix dressage to olympic standard, she uses her mouth to make different clicks for different moves. I was absolutely fascinated as I couldn't detect any change in the click but the horse did. So I have been wondering about different click meaning different reward or command, the horses are all positively trained.

[Cosmolo]

Sorry Rusky- i hope you're not trying to draw a parallel between a horse being 'positively trained' and a dog? Horses are prey animals, dogs are predators. I have worked with some of the top natural horsemanship 'experts' in Australia as well as a top dressage coach/ rider and there are elements of compulsion used in both. They work with the horses instincts but get the animal to move using some kind of physical pressure, however slight it may be.

[Staranais]

Just wanted to add that I only know a very little about natural horsemanship, but the techniques I've seen seemed to be based on a combination of mild positive punishment and negative reinforcement. e.g, the trainer would apply physical or emotional pressure to the horse when it was doing an undesired thing, and remove the pressure plus give praise (conditioned negative reinforcer) when the horse started to give the desired behaviour.

I could be dead wrong since I've only seen a little, but that's how it looked to me.

I'd be really be interested in hearing how natural horsemanship techniques compare to dog training methods, if a natural horsemanship pro feels like commenting.

[Kelpie-i]

Also it was suggested to me the other day that it's OK to use multiple CRs (in this case I was talking to a clicker trainer, so it would be click, click, click) then Primary much latter, say up to 5 min later. It was suggested that the dog was "counting up" the CRs and so long as the Primary (food) was appropriate to the number of CRs (mores CRs = more/better food), then this was OK. Any thoughts? Anybody train this way?

WS, I am currently reading a book on this and it recommends that once you have conditioned the dog to both CR and PR, you should commence lengthening the time between the two. It states that the importance of doing so, so that the dog doesn't expect the reward immediately to enables you to keep working. As long as the PR is always given at the end, the dog will continue to recognise the CR signal as a prelude to the PR. I'm not too sure about the dog "counting up" the number of CRs though???

Same with herding, sniffer and guide dogs, all needs to have a command the over-rides everything else, to allow the handler to stop/"turn off" the dog.

With my boy this command is "stop", when I want him to stop dead in his tracks whilst working the sheep. And because herding is done off lead, you don't have any means of correcting him is he doesn't stop , only a body block technique and send him off again as his reward.

I think both WS and Erny are talking about two completely different types of training. Erny is talking about obedience and WS is talkiing working dogs. There is a huge difference between the two and the rules applied to teaching obedience can be very different to the rules applied to teaching working dogs, whether herding, scenting etc.

[haven]

On the subject of Natural Horsemanship, it mainly involves the use of negative reinforcement. There's not alot of positive punishment (if any) used, but of course it depends on the trainer.

You certainly can use similar training methods on dogs but it is very uncommon (or so I have found) to do so, as dogs generally respond far better to other forms of training. As Cosmolo said, horses are prey animals, not predators, so both respond very differently to pressure and horses are far more sensitive to it than most dogs.

It also depends heavily on each individual dog. For example, with the horses I used to work with, I could have them move forward, backward, left, right, change gait etc from metres away (or more) with the subtlest of cues, a slight change of posture, eye contact etc. that would be invisible to most people. Most dogs of a solid temperament need far more pressure than this to respond in a similar manner.

I feel that my training in Natural Horsemanship really helped me deal with Loki, as he is a very fearful dog and far more predisposed to 'flight' as opposed to 'fight' and he doesn't respond much to positive reinforcement. I think the basic premise behind the training can be used on dogs, but it is far more applicable to fearful or timid dogs than those that are confident and of solid nerve.

I also feel that working with a flight animal, like horses, can help to make someone far more aware of the subtleties of body language, including awareness of their own and the effects this has on the animal (horse, dog etc).

Sorry for the O/T

Edited December 19, 2006 by haven

[Staranais]

On the subject of Natural Horsemanship, it mainly involves the use of negative reinforcement. There's not alot of positive punishment (if any) used, but of course it depends on the trainer.

In case I caused some confusion, I just want to clarify that by "positive punishment" I just meant that I saw a lot of applying physical or emotional "pressure" to a horse - for example applying pressure on a rope halter, or pressure from a hander's body language. The "pressure" was then removed when the horse was compliant (that's the negative reinforcement I think you're talking about). I never saw a natural horsemanship trainer cause a horse pain or fear, that's certainly not what I was referring to by the term punishment. Hope that makes sense?

Thanks for your explanation, Haven!

[Rusky]

Sorry Rusky- i hope you're not trying to draw a parallel between a horse being 'positively trained' and a dog? Horses are prey animals, dogs are predators. I have worked with some of the top natural horsemanship 'experts' in Australia as well as a top dressage coach/ rider and there are elements of compulsion used in both. They work with the horses instincts but get the animal to move using some kind of physical pressure, however slight it may be.

Cosmolo

I was simply pointing out the method of training and as you can see most have said that horses are all trained with some negative pressure. I was simply refuting that from the outset.

My response was to the clicks to explain the different sounds and how they were taught to a horse and that I have been thinking about different sounding clickers as a cue... thats it. I did not intend a debate about horse training... just different clicks.

We have all seen wrecked horses. I know there are shocking trainers out there, shocking riders too.

[Kelpie-i]

shocking riders too

That would be me, who had my first official horse riding lesson last Saturday. Legs are still sore today!!

[Erny]

... It was suggested that the dog was "counting up" the CRs and so long as the Primary (food) was appropriate to the number of CRs (mores CRs = more/better food), then this was OK. Any thoughts? Anybody train this way?

I have been thinking about this since a few weeks ago when I was with a girl who trained her horses grand prix dressage to olympic standard, she uses her mouth to make different clicks for different moves. I was absolutely fascinated as I couldn't detect any change in the click but the horse did. ...

Hi Rusky

I may be reading your post (above) incorrectly so forgive me if that's the case. But the way I read it, I can't help thinking you're thinking on a different plane to that of WorkingSetter's post regarding the multiple 'clicks'.

My interpretation of your post, Rusky, is that the horse trainer you refer to uses a different sound 'cluck' to produce a different behaviour each time. This could be compared to us using a different word command for each action required of the dog. After all, a word to a dog is simply a different sound pattern.

My interpretation of WS's post is that with each 'click' the dog is taught to anticipate a higher reward level but also that it would be delivered later.

WS - interesting concept and thought provoking, something my ol' grey matter is not up to negotiating tonight after a full-on day today.

I'll initiate my thought processes though, by reflecting on those owners who use multiple commands. Eg. "sit, sit, sit". After a time, most know that if this continues, the dog learns that he is supposed to sit on the third "sit". After all, the dog is hearing a series of 'sound patterns' and responding according to what it believes it has been taught. I guess it could work in the opposite as well, but I do wonder at the effectiveness of it (that's the part my brain needs to rest before it muses on the topic with any clarity).

Would the 'average' dog (whatever you wish to determine as 'average') be able to reliably differentiate that 1 click means a kibble type reward but that 5 clicks means a fresh chicken meat reward? With careful training to ward against the dog anticipating 1 click = chicken reward, yes, perhaps.

But does this make training easier or more efficient? I mean, "Successive Approximation" works in similar fashion but the dog doesn't have to learn the difference between different sound patterns in between times. Of course, 'time span' is built up gradually, as Kelpie-i mentioned earlier. So, I'm still not 'sold' on the idea of what presently seems to me to be further complicating training for the dog.

Groan ..... head hurts. Think it's time to knock off and return another time. :p

:rolleyes: Hey Kelpie-i! Horse riding lessons? What brought that on? You expanding your R & R activities? This going to occur on a regular basis? You going to become my riding partner?!!! Gosh, it's been so long since I've ridden my boy I think I'd be feeling as sore as you do!!! Hope your butt and legs recover soon. ;)

Edited December 20, 2006 by Erny

[Working_Setters]

WS, I am currently reading a book on this and it recommends that once you have conditioned the dog to both CR and PR, you should commence lengthening the time between the two. It states that the importance of doing so, so that the dog doesn't expect the reward immediately to enables you to keep working. As long as the PR is always given at the end, the dog will continue to recognise the CR signal as a prelude to the PR. I'm not too sure about the dog "counting up" the number of CRs though???

Sounds interesting, which book?

It depends on how I think about it - in my limited understand of clicker training, I always used one click, one reward, and kept them close together for fear of unconditioning the click. ie if you've clicked a couple of times, but no treat, I'd have thought the dog is going to stop believing that a click = treat. OTOH there are times when I'll say "good dog" as a dog proceeds through a complex exercise, I might use "good dog" to mark a number of success points in the one exercise, with the dog needing to complete the exercise to get further reward. In this case there are a number of CRs and only one PR at the end of the exercise.

[Working_Setters]

the way I read it, I can't help thinking you're thinking on a different plane to that of WorkingSetter's post regarding the multiple 'clicks'.

My interpretation of your post, Rusky, is that the horse trainer you refer to uses a different sound 'cluck' to produce a different behaviour each time. This could be compared to us using a different word command for each action required of the dog. After all, a word to a dog is simply a different sound pattern.

My interpretation of WS's post is that with each 'click' the dog is taught to anticipate a higher reward level but also that it would be delivered later.

I think this is spot on, at least it is in regard to my post. I talking about multiple clicks, each one the same sound, signifying a reward (just like clicker training), except that the treat is to be given after the exercise is complete.

Some dogs can "count" or at least some version of counting, retrievers doing triples know that after they have retrieved 2 dummies there is still another one to get. However I agree with Erny, I have trouble believing the "average" dog can learn one click = kibble and five clicks = chicken, so if we put the counting aside for a minute, what do people think about multiple clicks (or other CR) during a complex exercise, each CR marking a success point in the exercise. To communicate to the dog that your are pleased with its effort so far, and that it should keep that effort up to complete the exercise and get the PR.

But does this make training easier or more efficient? I mean, "Successive Approximation" works in similar fashion but the dog doesn't have to learn the difference between different sound patterns in between times. Of course, 'time span' is built up gradually, as Kelpie-i mentioned earlier. So, I'm still not 'sold' on the idea of what presently seems to me to be further complicating training for the dog.

I agree there would be times when it complicates things, for example dog has done 80% of exercise correctly, you've CRed (clicked) 4 times (each time 20% of the exercise was successfully completed), but now dog competely stuffs up the final 20% of the exercise, you've already clicked, so you have to treat, but now are rewarding a stuff up.

Now I remember why I stick to simple time honoured training methods - I can understand them :rolleyes:

[Kelpie-i]

Hi WS, the book is called "Click to Calm" - I forget the author.

I see value in the time lapse between CR and PR as the handler may not always be in a position to give the treat immediately after the click so the dog does not expect the treat to come immediately - only that he knows it WILL come sooner or later.

Remember that you can also use verbal markers to indicate to the dog that he is either on the wrong or right track and use the clicker only for when he gets it absolutely spot on. Like a clicker, verbal markers need to be conditioned otherwise they are meaningless to the dog.

I am not very comfortable with the multiple click notion and like you, would be concerned that it could cause some "un-conditioning" in the dog as well as confusion.

:rolleyes:

Erny: Hey Kelpie-i! Horse riding lessons? What brought that on? You expanding your R & R activities? This going to occur on a regular basis? You going to become my riding partner?!!! Gosh, it's been so long since I've ridden my boy I think I'd be feeling as sore as you do!!! Hope your butt and legs recover soon.

Hey Erny, the lessons were offered to me by one of my clients who is willing to teach me on a more regular basis. What can I say...I loved it and will pursue it. I had only ever ridden trail ride horses before but this is a completely different story. I would be happy to be your riding partner, but perhaps after a few more riding lessons under my belt Legs and butt are a little better today - thanks for asking :p

[Arya]

Interesting to read about the CR as well as the PR idea. I don't have a lot of experience but have been doing this anyway. I use words rather than clicker. The word 'GOOD' = 'Keep going, you're doing really well and doing the right thing'. My dog knows this. The word 'YES' signals the end of the exercise and is followed after a short bridge period with a reward of ball or food, depending on what we are doing. The two reinforcements seem to work well individually and good in the trial ring as she knows to keep going, it's not the end between exercises. Just my take on this. Interesting topic! I like words as inflection can be put into them as necessary. Can't do this with a clicker, I feel. BUT... I've never trained with a clicker. Am I wrong? Can the clicker be used softly or loudly? I'd love to know more.

[Kelpie-i]

Arya, the reason some people use the clicker is because it emits the same sound each time. It is used to mark the precise behaviour you are wanting from the dog and is especially good for teaching complex skills and/or slight movements like head tilts, targeting etc. The fact that the click is always the same is the key in that the dog learns that the sound of the click is always followed by a treat, therefore continues to offer the behaviour more often.

Whilst verbal markers can work well if you keep your voice tone the same each time, they can unfortunately become misconstrued if your tone changes or you are having a bad day etc. Remember that dogs are masters at reading body language and if you are having a "can't be bothered" day then your words will come out as a reflection of that. The clicker, on the other hand, always has the same neutral sound , therefore there is less chance of the dog becoming confused. You can always use a verbal praise with game etc afterward.

There is a whole lot of information about clicker training on the web. It's not for all people though - you either love it or hate it. I use the clicker occassionally, more so for teaching tricks but have used it to polish up obedience etc. Lately I have been using the clicker on aggressive dogs and it has been working wonders.

Edited December 21, 2006 by Kelpie-i

[Working_Setters]

Hi WS, the book is called "Click to Calm" - I forget the author.

Thanks KI, sounds like an interesting book - How are you liking it?

A friend of mine had good success clicker training her dog not to rush the back fence and bark at the neighbours kids.