Punishment/negative Reinforcement/both?

[Kavik]

Very interesting guys!

haven - I haven't had much to do with horse training, that was very informative.

Erny - I hadn't thought about heeling in that way of the threat of the punisher. Interesting.

[Staranais]

Haven:  If you use what some people refer to as 'active training' you give the command and use the correction before the dog complies, then wean off the corrections as the dog becomes reliable in response to the command. In that case it would be negative reinforcement as the dog avoids the punisher by staying in the heel.

Could someone please explaint this to me? If you give a command, then correct, the dog cannot avoid the correction even with speedy compliance. So aren't you just classically conditioning the dog to dislike the command? In fact, aren't you practically teaching him that speedy compliance is pointless, since it does not stop the correction?

Haven: I also think that once the dog has learned the skill and is aware of the repercussions of non compliance, then staying in the heel command to avoid a punisher could be considered negative reinforcement. However, as I'm typing this I wonder......you're not actually removing anything although you could argue that you are removing the threat of a punisher but then if the dog continues to stay in the heel command you never actually remove the threat, although they do avoid it.

I'd personally still call this positive punishment, I think. The aversive is only being applied when the dog moves out of position, right? So you're still only applying it to decrease the behaviour of being out of heel position, right? That's positive punishment.

Though I see Erny's point too. Once the dog understands that punishment will be applied if he moves out of heel position, I guess there is an element of negative reinforcement, since the very absense of punishment is in itself reinforcing. ;)

Haven:  Not sure about Amhailte, but she seems to know everything anyway!

My dog taught me everything I need to know. Incidentally, he runs correspondance courses too - he says $199.99 + GST, if anyone's interested? :D

PGM: Which is one of reasons that I don't care much for these kinds of discussions, it doesn't matter. What matters is what works, find what works and call it what you like.

That's so true. But the one benefit of knowing these types of theories and terminologies is that when you talk to other trainers, you can understand each other. If we both have different definitions for "punishment", for example, communication will pretty much grind to a halt between us. :D

[Akitagal]

Thanx Haven! Your post was funny as it was what you are "stroking your chin" about that got me to post in first place. I also have done some natural horsemanship and saw it as neg reinforcement but didnt know it was called that until i started reading up on all dog training info. I do have an open mind and get my advice from many sources but thought dol would be a good one for this type of question as i interested in all responses so i can then look and asses and question and decide. I like hearing all the differing views. But yes i try not to take all things as gospel and can see how easy it would be for someone who is just learning about all this to do so, so thankyou for the advice.

Thankyou everyone for responding it has helped me hopefully clarify some of how the scientific explanations relate to actuall dog training situations.

P.s Haven at some time i would be interested in hearing your thoughts on what you were thinking re: better when talking about humane/kind ways to train an animal.

[haven]

The 'passive training' way though, if the dog leaves the heel position and gets a correction IE tension and release, I would call it + punishment.

But what do you call the 'release' part of the equation?

If the tension is released before the dog complies with the command IE the correction is a quick tension and release, it is +punishment.

If you apply the tension but don't release it until the dog is back in the heel position, it is -reinforcement.

The problem I have with discussions like this is that people don't use the correct terminology. EG consider those who say they use only +reinforcement. However, if they remove attention or withold the reward, they use -punishment and if they use a head collar to apply pressure they use +punishment. Because some people misconstrue the terms positive and negative to mean bad and good and the term punishment to mean hitting your dog, discussions easily become confusing and hotly debated, even though everyone is essentially saying the same thing.

Isn't that right Kavik

[pgm]

The threat of the punisher becomes negative reinforcement. IE The threat the dog feels is removed because it didn't do the wrong thing. Similar to when we're driving on the roads. The threat of a speed fine is a negative reinforcement. The threat is removed when we've returned home and haven't sped (so we know we don't get the fine).

No Erny, I have to disagree. Understanding removes the threat. So long as I know what the speed limit is and so long as I understand what a speed limit means (it means that if I don't obey I might get punished) there is no threat, real or imagined.

This why, contrary to popular opinion, when a dog is taught using negative reinforcement (as an aspect of training) by somebody who knows what they are doing, the dog's confidence and attitude increases.

The popular opinion has it that if you use aversives your dog will become a robot and just go through the motions but without any enjoyment in the constant fear of correction.

Understanding removes the threat.

[Staranais]

Perhaps threat is the wrong word then. I don't think Erny means that the dog crawls through heeling, dreading the possibility that he will be punished. If the dog fully understands what earns punishment and what doesn't, there is no need for him to dread or fear anything, since he has the choice whether to be punished or not.

Perhaps a better word is 'possibility'. The dog understands that there will be a possibility of punishment should he fail to keep in the heel position. Therefore he makes the choice to stay in heel position, since he knows this will mean he is not punished.

(Correct me if I'm wrong, Erny. Not trying to put words in your mouth. )

[haven]

Haven:  If you use what some people refer to as 'active training' you give the command and use the correction before the dog complies, then wean off the corrections as the dog becomes reliable in response to the command. In that case it would be negative reinforcement as the dog avoids the punisher by staying in the heel.

Could someone please explaint this to me? If you give a command, then correct, the dog cannot avoid the correction even with speedy compliance. So aren't you just classically conditioning the dog to dislike the command? In fact, aren't you practically teaching him that speedy compliance is pointless, since it does not stop the correction?

Personally Amhailte, I think it is playing a dangerous game, one that could certainly lead to learned helplesness, handler aggression etc.

The premise i believe is to get the dog working to avoid the punisher with the goal of speed and reliability. It is considered by some to be a 'better' way of training rather than allowing the dogs actions to dictate your response, but I thought that was the essence of operant conditioning!

I wouldn't condone it for most dogs and certainly not the majority of handlers!

Haven: I also think that once the dog has learned the skill and is aware of the repercussions of non compliance, then staying in the heel command to avoid a punisher could be considered negative reinforcement.

Amhailte: Once the dog understands that punishment will be applied if he moves out of heel position, I guess there is an element of negative reinforcement, since the very absense of punishment is in itself reinforcing. 

That's exactly what I was saying.

Haven at some time i would be interested in hearing your thoughts on what you were thinking re: better when talking about humane/kind ways to train an animal.

There are a few threads around where it has come up. I'll try and find them and supply the links so I don't have to hijack this one!

Cos we never hijack do we guys

[Staranais]

Ah but the hijacks are always interesting, and I always learn from them - hijack away!

And having said that, here's a mini-jack all of my own...

The premise i believe is to get the dog working to avoid the punisher with the goal of speed and reliability. It is considered by some to be a 'better' way of training rather than allowing the dogs actions to dictate your response, but I thought that was the essence of operant conditioning!

If you're aiming for a speedy response, wouldn't it be better to either:

1) use negative reinforcement (e.g. e-collar set VERY low), turned off when dog completes requested task so that it is in dog's best interests to comply very quickly; or

2) selectively positively reinforce only the fastest responses with a prey drive type reinforcer?

Why do people think this 'active training' is a better route than these two options?

[Akitagal]

Thanx pgm for reply...sorry i am slow typer and not keeping up with pace of thread! I started this particular discussion as i am learning the therory behind dog training. I have trained dogs before and have been working with dogs and horses a long time but was struggling on matching actual training situations to the terms that were being used in training books and info. In particular neg reinforcement and positive punishment. I wanted to get it clear in my head as i think it will lead to a better understanding of what i am using when training and also as someone posted, more effective communications with other trainers.

[Kavik]

I think active trainers (feel free to correct me if I am wrong K9) would say that the first option with the e-collar IS one sort of active training. If I understand it correctly.

Active - Command - Correct - Comply - Reward

I guess clicker training would be the other extreme of training, where everything is dependednt on what the dog does first - has to offer the response?

[haven]

Honestly Amhailte it's not a topic I know much about but it just doesn't seem sound to me. At least it wouldn't be my prefered method and I'd have serious reservations about using it on certain dogs and teaching it to certain handlers.

However it was introduced to me by an experienced and well respected trainer here on DOL.

I personally think that the whole point of operant conditioning is to teach the dog the consequences of his/her behaviour. So the dog doesn't comply to a command, they get a correction. They comply, the get reinforcement. Some consider this an inferior way to train, I think it is the whole point of training.

For an sound dog who knows the command and with a competent handler I can see that the method has merit. You give the command, give the correction and next time you get immediate compliance and the behaviour is -reinforced in the future because the dog believes (and rightly so) that his compliance avoided the punisher.

The ecoller example you gave is pretty much 'active' training. I believe the second example is what is used while the dog is learning the command, but like I said I'm not an expert on it.

NB: In my earlier post I wrongly refered to the opposite of 'active' training being 'passive' training, when it is actually 'active' and 're-active.' Am going to edit my brain bad.

[pgm]

use negative reinforcement (e.g. e-collar set VERY low), turned off when dog completes requested task so that it is in dog's best interests to comply very quickly;

I would call that an example of active training.

as i think it will lead to a better understanding of what i am using when training and also as someone posted, more effective communications with other trainers.

Actually, most of the discussions I have witnessed amongst trainers discussing theory has resulted in disagreeement and confusion. As I said, a lot can depend on how you describe what is going on - different descriptions lead to different interpretations.

As I said before Amhailte's explanation of the terms is all you really need to know.

Positive = adding something

Negative = removing something

Reinforcement = increasing behavior

Punishment = decreasing behavior

Write these on your hand if you have to sit an exam and you'll do fine.

[Staranais]

Kavik:

I'm a bit confused by all this. Can you explain further?

I understand how 'active training' would work with the e-collar, it's simple negative reinforcement (escape training). The collar turns on, you give the command, the dog complies, the collar turns off. The dog has learned that he must obey you, and only then will the collar turn off. It is in his interest to obey quickly, because that means the collar turns off quickly. Right?

But how does this apply to a prong collar correction, for example? You give the command, prong the dog, he complies. But what has he learned?

And what if you give the command, prong the dog, and he does nothing? Just prong him again? Then what was the point of pronging him the first time?

I think I must be missing something here, sorry.

[pgm]

For an sound dog who knows the command and with a competent handler I can see that the method has merit. You give the command, give the correction and next time you get immediate compliance and the behaviour is -reinforced in the future because the dog believes (and rightly so) that his compliance avoided the punisher.

The correct sequence for teaching the sit for example:

1: press the ecollar (low level stim)

2: give the command

3: physically place dog in the sit position

4: when bum hits the ground take your finger off the button

5: praise the dog/give food reward

Total duration of stim about 1 second. The stim from the ecollar is not a correction: negative reinforcement yes, but not a correction.

The method is very simple and could be mastered by just about anyone with the proper instruction.

[Staranais]

Or could you answer my question about active training with a prong collar, PGM?

[haven]

It is something that will achieve clarity with time, education and experience Akitagal. I struggled at first with the seemingly paradoxical +punishment/-reinforcement thing too.

Basically it's all in the timing. If the aversive comes before the behaviour and is then removed as a result of that behaviour it is -reinforcement.

Maintaining pressure on a headcollar until the horse steps forward is -reinforcing the forward movement.

When you start up the car and it beeps at you until you put your seat belt on, that's -R.

When you pinch a dog's foreleg until he picks it up and then release, lifting the leg has been -R.

When you are handling your dogs paws and he mouths you and you keep holding on until he is quiet and calm, thats -R.

[haven]

We're on the same page Amhailte. I can't explain it because I have reservations about the theory behind it.

I believe the idea is to give an effective correction IE one you don't have to repeat. In a perfect world you could do it every time, but in reality?

The above example I agree with PGM although it is not personally the way I would choose to teach a command. However, when the method was described to me (active training in general) it was not using an Ecollar in the manner you suggested, it was command, correction, compliance, reward.

I have issues with using corrections on dogs during the teaching phase.

I have issues with using corrections without giving the dog the opportunity to comply.

I have issues with giving a punisher and reinforcement for the same command.

I'm not saying dogs can't be trained this way, I just don't think it is the most ideal for most dogs and most handlers and could have potentially disasterous consequences.

Edited August 24, 2005 by haven

[Kavik]

The way K9 describes it (using a sit as an example) if I get it right

Give sit command

give small corrections almost as you are placing the dog in the sit and until the bum goes on ground (I think)

Then reward for compliance.

As the dog becomes quicker, reduce the number of corrections?

I'm not totally sure. Maybe K9 will need to explain.

[Akitagal]

Thanx Haven.

[Kavik]

I haven't used the method, just heard K9 explain it, so if it doesn't make sense it is because I am still trying to figure it out too