more on Aversives, an article by Terry Ryan

[Top Barks]

Aversives in Training

Terry's Tips > Aversives in Training

Legacy's classes stress reward-based, dog-friendly training techniques. We hesitate to incorporate aversives in our training because
AN AVERSIVE:

    * IS AN INCOMPLETE PROGRAM. It only teaches the dog what NOT to do.
    * CAN MAKE BEHAVIOR BASED IN FEAR OR ANGER WORSE. It adds yet another dimension of fright or arousal, compounding the problem.
    * CAN CAUSE CONFUSION AND REDUCE TRUST. It's best to earn your dog's respect by consistent leadership and kind training, not by intimidation, force, or physical abuse.
    * SUPPRESSES ACTION, NOT INTENT. For example, a dog punished for growling may stop the growling. If growling is inhibited, the next time the dog may skip the warning growl go right into action.
    * MIGHT LEAD TO THE WRONG CONCLUSION. If your dog chases the family cat and you punishes her for doing so, does your dog think she’s bad for chasing the cat? She might think you’re angry at her for not running faster and catching the cat!
    * NEEDS TO BE CONSISTENT. An aversive should be administered at the FIRST and at EVERY subsequent time the behavior us shown. This is rarely possible in real life! Since you’re apt to miss some of the time, your dog will learn to take his chances and play the lottery.
    * MAY NOT GENERALIZE WELL. If you punish your dog for chewing the dining room table leg, she might think "Maybe the dining room chair is OK" or "Maybe the table in the living room is OK."
    * MIGHT SHUT THE DOG DOWN. The proper term for this is LEARNED HELPLESSNESS. The dog decides that everything she does is wrong, so she does nothing. Such a dog isn’t learning anything -- except to endure punishment.
    * COULD CAUSE THE DOG TO SUBSTITUTE. Many problem behaviors are repetitive actions such as barking, digging, chewing. Such behaviors produce calming chemicals in the dog’s body and therefore reward the dog for the repeated action -- much like rocking a baby settles the child. If you interrupt or stop one repetitive behavior, the dog might simply resort to a substitute behavior.
    * REQUIRES PERFECT TIMING. ONCE. To be effective, an aversive needs to be delivered within a range of two seconds after the behavior. The aversive is not having the proper effect if you need to do it more than once or twice. After that, it constitutes abuse.
    * NEEDS TO BE INTENSE. If the punishment isn’t strong enough to stop the behavior the first time, the owner typically escalates the punishment next time. The dog develops a "punishment callus". Eventually the level of punishment needed to work will have to be stronger than what you started with.
    * MIGHT SEEM LIKE A REWARD. If the owner doesn’t choose the aversive properly, the dog might consider it a reward. Example: "Bad Dog! No!" Dogs like attention. The owner has spoken to him, looked at him, and perhaps got closer and touched him as well.
    * MIGHT BECOME OWNER CONTINGENT: If the dog associates the punishment with the owner, the effect (if any) will not carry over to times when the owner is gone.
    * SHOULD NOT BE USED FOR HUMAN SATISFACTION: The goal of an aversive is to quickly and permanently change the unwanted behavior. Is that happening or are the humans simply letting out their frustration onto the dog?

The fact is, most people that share their lives with dogs do not want to punish them, however they do want a mannerly dog. There are humane and effective alternatives to punishment.

    The above information was taken from the experience of Terry Ryan. Facts and research taken from the works of: M. Breland-Bailey, R. Bailey, B. F. Skinner, K. Pryor, M. Sidman, M. Burch, B. Schwartz & others.

[tallulah0710]

Great article thanks for sharing 

[Top Barks]

Glad someone liked it

[harveyroan]

I completely agree with all those tips in the article. But I don't think you can only use positive reinforcement, you have to use some negative punishment other wise how does the dog learn whats right and wrong. I think the problem with using any aversive is knowing when its appropriate, how much, and timing.

Nice little article

[Ralu A]

Great article Thanks.

[Top Barks]

I completely agree with all those tips in the article. But I don't think you can only use positive reinforcement, you have to use some negative punishment other wise how does the dog learn whats right and wrong. I think the problem with using any aversive is knowing when its appropriate, how much, and timing.

Nice little article

Do dogs really know right from wrong?

[Juno]

A very interesting read, thanks!  I showed it to my dad who, although is only ever super-kind to Juno, had suggested a couple of things to me which I wasn't happy with (ie tapping the nose, making a loud noise to interrupt a behaviour).  Also showed him the aversives sticky and think I might have got through to him! (He is veeeery stubborn and old fashioned in his thinking ). Thanks, Mark!

I observed a puppy class that we're joining in June last night and one of the owners said they had been shaking a rattling tin to stop their pup jumping up at the sofa. The trainer mentioned a lot of the reasons above as to why this isn't effective in the long run - she seems like a good'un! 

[harveyroan]

I completely agree with all those tips in the article. But I don't think you can only use positive reinforcement, you have to use some negative punishment other wise how does the dog learn whats right and wrong. I think the problem with using any aversive is knowing when its appropriate, how much, and timing.

Nice little article

Do dogs really know right from wrong?

No they don't I just used that as a figure of speech. I'm no expert like yourself but learning so please correct me if what I've read/learned is incorrect. From what I've learned is that if you do use a aversive to discourage a behaviour you must provide an alternate behaviour.

From what I've observed in behaviours have been allowed to escalate there is no alternative but to use a aversive to change the behaviour but again you must provide an alternate behaviour.

Some very respected trainers I know of have had to use aversive, but they only have to use it once or twice to change that behaviour.

I know when I got my dog I was only ever positive and ignored bad behaviour. What a mistake if you only ever ignore the behaviour that's not acceptable how is the dog to know it's not acceptable.

[tallulah0710]

I completely agree with all those tips in the article. But I don't think you can only use positive reinforcement, you have to use some negative punishment other wise how does the dog learn whats right and wrong. I think the problem with using any aversive is knowing when its appropriate, how much, and timing.

Nice little article

Are aversives positive punishment though? Not negative punishment.

[harveyroan]

I completely agree with all those tips in the article. But I don't think you can only use positive reinforcement, you have to use some negative punishment other wise how does the dog learn whats right and wrong. I think the problem with using any aversive is knowing when its appropriate, how much, and timing.

Nice little article

Are aversives positive punishment though? Not negative punishment.

Sorry yes they are. What I meant was is that you can't be just positive you have to use negative punishment e.g(turning your back on a dog jumping up).

I think if you have to use a aversive it 'has' to be used very carefully.

[praia]

Dogs don't know right from wrong in human sense, but they understand that certain behaviors earn rewards and others don't.  They learn this by first reinforcing those behaviors we want and punishing those we don't want.  When I say punishing, I'm talking in terms of operant conditioning where an event occurs which decreases the likelihood of that behavior being performed again.  Even if you're only denying a dog a treat, your attention, or praise, you're still technically punishing the dog.  Some clicker trainers use verbal cues a no reward marker, which could just be a neutral cue to some dogs, but could as be viewed as a low to mild verbal punishment to those with softer or fearful temperaments.  By definition of operant conditioning, all punishments are going to be aversive.

It's impossible to proof any sort of behavior without some form of punishment, though what I think the punishment/aversives this article is strictly referring to is coercive positive punishment and perhaps even negative reinforcement?

The article is okay in that it shows how easy it is to fudge up a dog if you don't know what you're doing, but I do have issue with how it seems as if they're claiming that it's possible to train dogs without any form of punishment, which is technically not true at all.

[Mel]

I can't agree with your last statement.

A dog learns that a certain behaviour pattern gets a positive response and/or treat from the owner/trainer and other behaviours illicit no response and/or treat.

My dog decides to bark at the lift. I know from accidental experience he looks to me to do something, anything about the noise, inbetween his barks. I do nothing. So he jumps up at the piano to try to grab something which gains my attention. By correcting myself, there is no longer stuff left there, so he can't steal it. Next port of call is the bathroom. Out he comes with a pack of Always, (sorry guys), and runs up and down infront of me. He gets no response. Pack of Always gets dropped on the floor, Tali sighs and lies down.

The reversal of the problem initially caused by the postie is going to take a long time. However, as long as he gets no response from me, he will, in time, realise not to bother to bark.

It may take Tali months or a few years to really lose this behaviour, but it will have been done with only positive reinforcement for good behaviour and no punishment.

Turning your back on a dog is a calming signal to help relax a stressed dog. Even walking away from a puppy who is biting is not punishing them. They just learn the games stops with certain behaviours, so they eventually stop doing them.

[harveyroan]

Dogs don't know right from wrong in human sense, but they understand that certain behaviors earn rewards and others don't.  They learn this by first reinforcing those behaviors we want and punishing those we don't want.  When I say punishing, I'm talking in terms of operant conditioning where an event occurs which decreases the likelihood of that behavior being performed again.  Even if you're only denying a dog a treat, your attention, or praise, you're still technically punishing the dog.  Some clicker trainers use verbal cues a no reward marker, which could just be a neutral cue to some dogs, but could as be viewed as a low to mild verbal punishment to those with softer or fearful temperaments.  By definition of operant conditioning, all punishments are going to be aversive.

It's impossible to proof any sort of behavior without some form of punishment, though what I think the punishment/aversives this article is strictly referring to is coercive positive punishment and perhaps even negative reinforcement?

The article is okay in that it shows how easy it is to fudge up a dog if you don't know what you're doing, but I do have issue with how it seems as if they're claiming that it's possible to train dogs without any form of punishment, which is technically not true at all.

Exactly what I wanted to say thanks. Sorry I'm not good at long posts The article IS good but I think like praia has said you can't train wholly positive. I do think that dog training has become a bit PC just as child care, you need to teach your dog manners without excessive force.

Sorry mel but turning your back on a dog jumping up at you is a negative punishmet your denying the dog attention. And if my dog had pinched a pack of always(NOT mine may I ad ) I would encourage him to bring it too me I wouldn't ignore it(if he started to rip up the always rather than drop it what would your next move be?). He's only doing whats natural to him and that's retrieving.

[Top Barks]

I do agree with a lot of what praia says but with reward based training the idea is set the dog up to be rewarded and not to fail in the first place.
yes we all with hold treats and ignore the dogs when attention seeking etc and for some dogs this may be more punishing than any sort of physical correction but i think the crux of this article is that aversion can be fudged up, even by people who claim to know what their doing.
it can also do a lot of damage to the dog /owner relationship.
I also think the article just points out the pit falls of using aversives, do you believe any of it not to be accurate? If so why?
I don't see where it says you can only train without any sort of punishment in the article?
Indeed there is plenty you can train without any form of punishment if you make the criteria for reward achievable and keep your rate of reinforcement high.

[tallulah0710]

Just googled the article....this was missing off the bottom...hope it clarifies! 

In this article, the words punishment and aversive mean delivering a harsh "correction". No attempt is made at categorizing the examples into formal learning theory.