Training With No Compulsion?

[MonElite]

ive spoken to some trainers that believe that building a solid bond with dog allows you to use compulsion with any negating effects?? any thoughs?? i love the ideas thought and theroys

I dont agree with the any in the above sentence, but I do agree that solid bond with the dog allows the handler to perform more "unpleasantlies" to the dog that to the one with no bond.

Lets take nail cutting as an example. My dogs dont like it, but they let me do it. I dare say they would want to run away or possibly bite in fear if a stranger wanted to do it to them.

This is not to say that over corrections have no effect on the bond, hence what you said is somewhat true and somewhat not.

IMO of course

Edited January 5, 2010 by MonElite

[Cosmolo]

I've used some compulsion with my dogs and with one dog, its actually allowed our bond and relationship to become alot stronger. Using compulsion doesn't automatically create negative fallout.

[Aidan]

In theory, a Differential Reinforcement procedure should produce a very reliable response to stock.

I believe that 'In theory' it might sound as though it could work, but once a dog achieves self reinforcement and satisfaction from chasing and more importantly, killing stock, the only good place for food treats, balls and other rewards is on the ground as you suddenly drop them to chase after the dog in an attempt to get him back. Sorry Aiden, but unless I actually see it being achieved with a dog who has killed stock, then I just cannot believe it, no matter how fancy the terminology sounds on paper.

Sure, don't get me wrong. If a stock killer came to me and I couldn't find someone with the experience to deal with it I would do it the way we know can work. I certainly wouldn't be putting that person, dog and livestock in the position of being my guinea pig but I wouldn't be surprised if someone came up with a reliable +R approach one day.

[Aidan]

And if the dog has a bad habit of some sort (lets say killing chickens) then IMHO its not possible to fix it without corecting the dog for the wrong doing. Unless of course one can find a better reward then chickens

Roast chickens?

But seriously, it's really not about finding a "better reward" than whatever other reinforcers are out there. Unless we're talking about bribery...

Roast chickens

OK, lets say the dog finds chasing and killing chickens ultimate reward, it has killed several of them, what do you do? (other then of course prevent the chasing and killing from happening)

And lets stick of dogs not some other animals

This is actual stock killing and that's not fair on the owner, the dog and certainly not the chooks so I personally would stick with what has been demonstrated to work quickly in the past, and that is a remote training collar. Preferably getting an experienced person in to do it (not my area of expertise), or at least to consult first if they couldn't be there to do it.

The unique thing about this scenario is that lives are at risk, both the chooks and the dogs.

But this is a discussion, not a real situation as far as I know, so let's indulge ourselves.

First up, the problem is not that killing chooks is so inherently reinforcing that we will never compete with it. That's a furphy. There are dogs who chase wallabies that find that just as reinforcing as a lot of dogs who kill chooks and they can be trained to recall using purely positive methods with far lesser rewards, it's been done and done and done.

There are bigger issues. This needs to be 100%, whenever the dog has access to chooks (Kelpie-i mentioned a dog that would suffer severe lacerations to get to chooks so "whenever the dog has access" could have a very broad definition). It needs to be reliable when unsupervised. Day or night. There are a whole bunch of antecedents that would need to be proofed for. It needs to be done within a time frame, most likely a short time frame. There are probably other issues I haven't thought of that would be encountered if one were to attempt this.

All that aside, I would pick a minimum number of alternative responses. Beginning with protected contact I would teach and proof each response, putting any hunting behaviours on an extinction procedure. Then I would move to leashes, muzzles etc Then off-leash. Then protected contact unsupervised, then unprotected unsupervised.

The alternative responses would be either avoidance behaviours (invisible force field type stuff), or nurturing, protecting behaviours. Although it would be trickier, I think the nurturing, protecting behaviours are a better option because I have personal experience with this and trust my GSD with my cats completely as a result. I believe you can start to effect an emotional response, pack drive, through teaching this sort of initially artificial behaviour.

Someone might come up with a better idea. It would have to be a much better idea to bother... It might work quite well in conjunction with an e-collar ("I've got to look after these electric chickens!")

The other things is we must be careful not to pretend that aversives are somehow magical. The laws of learning still apply. I think people put too much faith in an absence of punished responses, and make assumptions about how it generalises. What you don't see doesn't hurt you, until you see it again!

[MonElite]

But this is a discussion, not a real situation as far as I know, so let's indulge ourselves.

That was an example of a real situation I have recently encountered.

But hey your indulgence in theory was nice, shame with that particular dog all has been tried and failed.

She even ignored highest settings on ecollar.

Mind you 100% reliable recall when on long line.

I guess super smart dog (not mine if anyone is wondering)

[Aidan]

shame with that particular dog all has been tried and failed.

She even ignored highest settings on ecollar.

That rings alarm bells. Was this dog otherwise trainable?

[MonElite]

Yes.

but it has reinforced its high prey drive with self reward of chasing and sucessfully killing several chickens and ducks. So there is no better reward to it than the chook.

And for the purpose of this discussion I asked how would one train a dog like this with no corrections.

Locking up the chickens or the dog is management not training. The dog is being currently managed, meaning it has a very limited freedom.

[Aidan]

And for the purpose of this discussion I asked how would one train a dog like this with no corrections.

Not saying it would present a workable solution (management is probably the only solution in this case), but you would need to start a long way from chickens and ducks in order to reinforce an alternative response. As far away as you need to be, only moving closer when the alternative response is fluent at that distance.

e.g my GSD would get very wound up if another dog were chasing a ball, I had a friend toss a ball for her dog at the other end of a football field. That worked OK, but we got quicker results bringing them in closer and having her toss the ball in the air to herself, then tossing it in the air and letting her dog get it, then just rolling it to her dog and proceeding that way.

[Aidan]

shame with that particular dog all has been tried and failed.

She even ignored highest settings on ecollar.

That rings alarm bells. Was this dog otherwise trainable?

I should explain, I asked this because there are certain conditions that inhibit learning through aversion. OT, I apologise.

[MonElite]

As far away as you need to be, only moving closer when the alternative response is fluent at that distance.

Of course this is what you would need to do and it, again, sounds good in theory. However there are situations where its not possible to do.

The dog chasing chickens example.

It was a xmas day chooks in a pen at the back of the house, dog in the yard. No need to say going nuts to get to the chickens as it has never seen them that close.

In order to have peace and quiet one solution would be to lock the dog up in a crate for the whole day, but this was not a good option.

We let the dog out it run to the choocks, we wacked it on the head with the broom (im sure Ill be told off for that) about 3 times when it attempted to carry on like a fruitloop, then stood between the dog and the chickens for 30 minutes and if it tried to go anywhere near the area we raised the broom it eventually understood that its a no go zone.

It tried few more times during the day but a verbal correction was enough, or the sight of the broom.

To me achieving something satisfactory in a short period of time was the goal and we did. It doesnt mean that the dog is trained not to chase the chickens as Im sure it would if given an opportunity.

I just belive that sometimes a short term result is important, hence there is no time for the pure positive training.

Mind you I still think that dog could not be taught not to kill chickens with pure positve methods.

Edited January 6, 2010 by MonElite

[Kelpie-i]

I'm with Monelite here.

Often these people want the problem fixed yesterday! They don't have the skills, patience and sometimes the inclination to put all the learning theory and weeks/months of conditioning work into action to stop their dog from chasing and killing stock. We need to be realistic when it comes to the average Joe with their hobby farm full of chooks and ducks. They are not dog trainers and often cannot be bothered having to hear/do all the steps involved with purely positive training. It's just simply takes too much effort for them so sometimes a quick fix solution is the answer.

I'm all for using positive methods for nearly all of my training and I always do, but for specific situations where it is difficult to control a dog's instinctive urge, I won't trust any method that does not deliver a consequence, and I don't just mean withdrawal of a treat.

First up, the problem is not that killing chooks is so inherently reinforcing that we will never compete with it

I believe there are variables to this ie, depending on how often the action has been reinforced (self reinforced), depending on whether the dog has successfully achieved full predatory drive sequence and/or whether it's just the thrill of the chase with no killing involved. It would be extremely difficult (albeit not impossible) to compete with a dog which has fulfilled the entire predatory drive sequence.

[corvus]

First up, the problem is not that killing chooks is so inherently reinforcing that we will never compete with it

I believe there are variables to this ie, depending on how often the action has been reinforced (self reinforced), depending on whether the dog has successfully achieved full predatory drive sequence and/or whether it's just the thrill of the chase with no killing involved. It would be extremely difficult (albeit not impossible) to compete with a dog which has fulfilled the entire predatory drive sequence.

I wonder if you caught my living example at the beginning of the thread? Pyry has "fulfilled the entire predatory drive sequence". Multiple times. His drive to kill animals is frighteningly intense. At his worse he has turned on humans approaching a live animal he has bailed up. Didn't bite, but it's the only time in his life he's ever even threatened a person let alone made an aggressive move towards them. He's otherwise a very sweet dog with no resource guarding history. I don't know that he'd go through barbed wire, but if he can smell the animal and knows it's close and can't get to it, he will injure himself trying. Yet, now he will leave a live animal he was about to kill when called. It hasn't stopped him from hunting per se, but it does show the power of conditioning.

I do not think that it would necessarily work as well on other dogs. Pyry is a thinker and is not interested in chasing animals he won't catch. He's small, so that means he can't take off on exciting, high speed chases over the countryside like my mother's other dog can. He can, but the prey will leave him to eat dust, so he doesn't bother. I think that with chasers, they tend to switch off and not hear verbal commands when they go into chase mode. IME with Kivi, you can certainly build up so that they are far more likely to hear you when in chase mode, but Kivi doesn't put his all into a chase and would never injure himself, so I don't know how that holds with a more driven dog. I think that when they get glassy-eyed, whether they are chasers or killers, you HAVE to make your move the moment they even think about it, regardless of which method you're going to use.

I would have crated the chicken-chasing dog for the day. What's so bad about that? Seems a better deal than a few waps on the head with a broom and 30 minutes out of your day to wave the broom around.

I don't think there is a good outcome for a dedicated stock chaser if the owners don't want to put in any time or effort.

Edited to shorten.

Edited January 6, 2010 by corvus

[corvus]

" Of course you can. And I base that on pure fact. Exotic animal trainers do it all the time. "You can't spank a Killer Whale" they say. As Aidan has pointed out, it's not the reward that counts, but the conditioning. "

well i do see your point however the purpose of this thread is for DOGS only "killer whales" may have different "drives" or methods to achive an outcome e.g. we know dogs work in pack that for example some dog have some sort of prey drive or food drive etc i have no clue about killer whales and cant speak on that particular animals behalf unless you know for a fact that there similar to dogs phycologically

Well, they do call them the wolves of the sea. They are known to hunt in groups.

It doesn't matter, though. If you want to talk dogs, I can and have offered similar examples. But FWIW, drives and methods don't matter as long as you have the reward right. And that holds for any animal. Although there's a slight problem with building motivation in animals that graze or browse all day, don't engage in play or other social activities, and run at the first sign of danger. The principles are still the same.

[Aidan]

I'm with Monelite here.

Often these people want the problem fixed yesterday!

Just so we're clear - you know I'm with you both on this one, right? I've said so explicitly, it's a practical matter. I thought (possibly mistakenly) that MonElite was wondering how it might be done, if you were to try using purely positive methods so I was offering suggestions as to how this might be done given a rich, eccentric client

First up, the problem is not that killing chooks is so inherently reinforcing that we will never compete with it

I believe there are variables to this ie, depending on how often the action has been reinforced (self reinforced), depending on whether the dog has successfully achieved full predatory drive sequence and/or whether it's just the thrill of the chase with no killing involved. It would be extremely difficult (albeit not impossible) to compete with a dog which has fulfilled the entire predatory drive sequence.

Yes, I would think so too

If behaviour has hugely reinforcing consequences it might be very strongly conditioned. Conditioning a competing response might prove to be difficult considering time available, proximity or safety. This can also be true for behaviours that are linked to very weak reinforcers, sometimes you just need to stop something now before someone gets hurt.

The point I'm labouring on and on is to not confuse the issue of competing conditioned responses with the issue of competing reinforcers as these [reinforcers], on their own, do not prohibit the conditioning of an alternative response that is under stimulus control.

[MonElite]

I would have crated the chicken-chasing dog for the day. What's so bad about that? Seems a better deal than a few waps on the head with a broom and 30 minutes out of your day to wave the broom around.

The dog was crated for several days before it as time to let it out for some time

Edited January 6, 2010 by MonElite

[Misha]

Just for fun to add an easy one.

Let's say the requirement is invisible barrier. 100% proofing required. And to be fair unlimited time( the life of a dog minus couple of years) and unlimited resources ( money to be paid). Training is a must management not possible.

I will stress 100% reliability required. No matter what is on the other side. Possible training methods no compulsion allowed.

I will be the devil advocate

[Kavik]

Why though? I see no need for an invisible barrier, with or without the use of aversives to teach.

[MonElite]

I thought (possibly mistakenly) that MonElite was wondering how it might be done, if you were to try using purely positive methods so I was offering suggestions as to how this might be done given a rich, eccentric client

See the thing is that in THEORY I know how to do it, but I have not ever met a dog that the theory was sucessfully applied to.

[MonElite]

Im with Kelpie-I on the subject

I'm with Monelite here.

With Borats accent - Hi Five

[Aidan]

Just for fun to add an easy one.

Let's say the requirement is invisible barrier. 100% proofing required. And to be fair unlimited time( the life of a dog minus couple of years) and unlimited resources ( money to be paid). Training is a must management not possible.

I will stress 100% reliability required. No matter what is on the other side. Possible training methods no compulsion allowed.

I will be the devil advocate

Are we allowed to use management throughout the process (a leash, for e.g)? Can we settle for 99.9% reliable?

I think that for some dogs this would be very easy to achieve quickly, for other dogs it would take longer, and other dogs would never learn it reliably. For a workable model see: http://www.clickertraining.com/node/2409