Training Out An Unwanted Behavior

[hankdog]

Dobie mum there's a method called CAT that dies basically what you did for dog aggressive dogs. Quite interesting reading, they put the dog in front of another dog and just ignore the carry on. You need a succession of bomb proof tester dogs because obviously it's pretty stressful for them. Apparently at some point and it can be after two days the aggro dog will give up and try a pro social behaviour.

I did have a trainer start something similar with mine but we really needed more dogs, time and a really large training area because it's loud. We didn't get a pro social response but did get a flight response, for which I am so grateful.

Time is the thing Huski. How long has the dog been doing the behaviour I'm sure must also influence how long it will take to stop.

I think there's also an element of superstitious type belief in a lot of annoying behaviours. Along the lines of " if I pull my ear drop bears don't fall out if trees on me." Until you stop doing the ear pulling you can't prove they won't. If you do it once and the drop bears don't fall how do you know that wasn't an exception to the rule. You pulled your ear for 3 years and it worked, this one time you didn't and were ok could have been a lucky break. So in my case I might walk past 10 dogs and nothing happens but on dog 11 it rushes across the road and has a go. Now my dogs who was starting to think maybe a could be right has irrefutable evidence I'm wrong.

That folks is why I spit the dummy at off leash dogs.

[raineth]

Hankdog, I might be really off in this, but do you have a psychology background (I have a vague memory that you do)?

Because I've been thinking the same thing, and wondering whether the OCD treatment (behaviour response prevention) might be why punishment does sometimes work (from what I hear) in correcting reactive behaviours? By preventing the reactive behaviour the dog learns that they don't have to have a tanty to keep themselves safe.

Which is just along the very similar lines to what you were just saying about superstitious beliefs.

I also read another report in which they did something very similar with a few reactive Great Danes. They put them in a special enclosure and filled it with grain until only their necks and heads were above the grain. They then brought in a non-reactive dog. The dogs in the grain could bark etc but they could no lunge or move forward, and therefore much of their reactive behaviour was prevented. It was effective in curing them of their reactivity.

But obviously very difficult for the average owner to implement!

Edited March 26, 2014 by raineth

[hankdog]

Yes I do Raineth but it's purely academic which I suppose is why I need to understand what I'm doing. Practice is very different though. I think learning theory is great but there's definitely an interaction between learning and personality and I think when you have both the dogs and the handlers personalities to factor in it become less clear cut.

I'm also thinking that there's a qualitative difference between training a skill and getting rid of a bad behaviour. Initially I was old school check chain type training because that's what I learned as a teenager. Since then I Woukd like to think positive training methods would work but I'm starting to believe that's not so and I'm interested in how to figure out the minimum amount of aversive method needed.

Time frame is as Huski says important, but what else?

I would so put Mactub in a bath of grain, how good would that be? I did at one point consider erecting a run in the middle if the dog park and just waiting it out!!!

[disintegratus]

I too am following this thread with great interest, having a large reactive dog myself. I don't think the grain option would work for him though, they got into the house when we weren't home once, between the (then) 4 of them, they ate about 8kg of bird seed in one sitting. They were pooping it out for days, it was terrible.

[raineth]

I too am following this thread with great interest, having a large reactive dog myself. I don't think the grain option would work for him though, they got into the house when we weren't home once, between the (then) 4 of them, they ate about 8kg of bird seed in one sitting. They were pooping it out for days, it was terrible.

:laugh:

[hankdog]

I am quite conscious that at times the peanut butter I oil Jakes brain with is either a distraction, a lure or a reward and I guess in the early days it was also an internally applied muzzle- it's very difficult to bark with a mouth crammed with peanut butter. So the grain would be a physical barrier or for your dogs disintegratus also a reward.????

[Amax-1]

Training out unwanted behaviour with positive methods works ok with dogs who have inherent drives in the right place, that is dogs who lend themselves well to handler focus regimes, dogs of high drive who are easily adapted to handler induced reward which includes many of the agility champion's training techniques used on the right dogs is sensational but on the wrong dogs is next to useless where aversives need to come into play. Unfortunately in the modern training era where factions are at work banning equipment like prong and Ecollars, thus trainers are not learning the benefits of this equipment and how to use them which are so much further advanced than a check chain ever was and far easier to use in customising the right level of aversive for the given application, I can only see training out unwanted behaviour proving more difficult as time evolves with dogs who's drive levels don't support the modern era approach.

[corvus]

You hit the nail on the head when you mentioned safety and avoidance. It baffled scientists for a while that animals kept on performing an avoidance behaviour after a signal linked to an aversive even when the signal no longer predicted that aversive. A fellow called Mowrer came up with the two-factor theory, which said that the animals would come to avoid the signal rather than the aversive experience itself, but eventually it was shown that the same thing could be seen without a signal at all, so the animal couldn't be avoiding the signal itself. The current view is that there is no prediction error. The animal never learns the avoidance behaviour is not necessary because they always do it and always get the expected result: successful avoidance. There is also some recent work on safety as positive reinforcement. I just wrote an article about all this kind of thing after several discussions with some 'force free' trainers that seem to have got it in their heads that avoidance (negative reinforcement) is BAD and teaching dogs successful and appropriate avoidance behaviours is therefore unethical. http://blog.creatureteacher.com.au/ This is quite relevant at the moment as the Australian APDT has decided they are not comfortable with negative reinforcement, which means they may not support some new methods for tackling fear-related problem behaviours such as BAT and CAT (incidentally, CAT is supposed to be done under-threshold like BAT is, so the dog shouldn't be so close they are barking and lunging). The APDT are still deciding whether it contravenes their ethics policy or not.

Aaaanyway, your question was broader than avoidance behaviours. I think the answer is probably in the Brelands' article "The Misbehavior of Organisms", which talks about how animals are more likely to do some behaviours than others despite the best application of operant conditioning techniques. Animals have a suite of usually species-specific behaviours they will tend to fall back on in all sorts of situations. For example, dogs bark (and dig, and chase, and mouth), birds peck, cats scratch and so on. These behaviours come naturally and serve a purpose. It is hard to convince an animal that what comes naturally does not get them as close to their goals as something else that does not come naturally. And some of those behaviours just inherently feel good because they are so critical for finding food. But it can be done, and shouldn't take years. You just need consistency and the right conditions. The latter is the big problem for reactive dogs, of course, and the reason why people do end up working on it for many years. There's a second part to the article I wrote that talks about training safety behaviours for dogs. You probably already have a couple with Jake. I think they are wonderful, but kind of depend on being able to get that arousal down pretty fast and having your dog aware that they have just entered the 'safety zone', which also means they have to truly believe they are safe. You can teach them a signal that means they are safe, but they won't believe it if they are highly aroused. In fact, high arousal will sink many (not all) attempts at training with operant conditioning. That's a whole other discussion, though. :)

[BC Crazy]

Corvus Welcome back... You've made one reactive dog owners' day :D

Sorry folks slightly OT.

Edited April 3, 2014 by BC Crazy

[alpha bet]

We all accept that the dog's personality is set, it is the behaviours that we can alter. That being said, of course the personality may affect the likely behaviours we will see. It is just up to us to modify these behaviours we don't want and encourage the behaviours that we do want. So to keep things simple we have to understand that we aren't actually trying to 'train' out an unwanted behaviour.

The objective is to BLOCK behaviours we don't want - a BLOCK can be made by using good Management, verbal, mental and if need be physical. The more ingrained the unwanted behaviour the more dramatic the BLOCK might need to be.

This can be very difficult for the inexperienced as they are reluctant to appear rough at all - also in this day and age with our 100% Positive Reinforcement mentality, it is not politically correct to manhandle a dog in any fashion. The advantage an experienced trainer has, is that often they have encountered unwanted behaviours in many different dogs thus have built up an ability to catch the behaviour at an earlier stage and place the block, often with very little sign of doing much.

e.g. a dog who likes to hump peoples leg...... (I use this example as had one of these dogs yesterday).... Generally people wait to do anything until the dog is actually attached to the leg..... whereas when the dog arrived here he came up to me to say hello and immediately moved his nose into my crutch.... he was being rude so he coped a quick knock away from me... dog then watched and made much more polite moves to get to know me. However 15 minutes later another person arrived and he went into crutch mode and within perhaps 10 sec was into humping (now harder to stop). We then gave him different people to meet and each time he went to the crutch he got a knock away. Dog now starting to understand that he should not invade the humans space.

The objective is to BLOCK the behaviour when he is still in the thought process - as in the above example, when he is sniffing - this is the behaviour that pre-empts the unwanted behaviour. Once he is in the mode of humping the leg he is in mindless mode and the lesson is not as obvious to the dog. With repetition a dog can start to altere his behaviour then there is a chance to TEACH him a different way to respond. This process can be used for all behaviours - and not one piece of food insight.

In our school we dealt with many different problems over the last 20+ years - our objective is not to just provide a distraction to the problems but get the dog and handler to face the problems and learn how better to deal with issues - and it is true to say that nowdays I am much more subtle with my Blocking Response.

We have many dogs who are reactive to other dogs and there are different approaches that we use depending on the personality of the dog we are dealing with. However our classes are run with dogs off leash for most of the class - reactive dogs might stay on-leash which is a form of Management and also allows us to use a physical block. Once the dog has learnt to relax in the company of others then they have the chance to learn that other dogs can be fun..... not all dogs are able to adjust and hence become a permanent form of Management but a greater percentage learn how to be social with different dogs and hence become much easier dogs to live with.

But the subject of reactive dogs is a topic all of its own.....

[hankdog]

Amax-1 that's so true, and how many of the "untrainable" dogs will end up being PTS because the owners can't or won't access aversives. Not a great outcome.

Corvus I wonder if these dogs would also be low on a risk taking measure, they get stuck just doing a behaviour in that same thoughtless manner. This I think may be why increasing confidence and especially the clicker training where they start to generate their own "new" behaviours can be helpful. It was a game changer for Jake.

Alpha bet I'm in exactly that head space where I'm trying not to be rough but from watching how effective a bit of manhandling is I realize I have to be. Jakes in kennels this week and having training daily, I had started to realize at this point my trainers could take him through this next phase far better than I could. Circumstances just put me away from home this week but I'm thinking it wasn't a bad thing.

I don't know how much research there is on human dog personality interaction but I'm sure there would be some interesting work to be done there. Wouldn't it be great if we could have a dog RSVP service.

[Aidan3]

Dobie mum there's a method called CAT that dies basically what you did for dog aggressive dogs. Quite interesting reading, they put the dog in front of another dog and just ignore the carry on. You need a succession of bomb proof tester dogs because obviously it's pretty stressful for them. Apparently at some point and it can be after two days the aggro dog will give up and try a pro social behaviour.

I did have a trainer start something similar with mine but we really needed more dogs, time and a really large training area because it's loud.

Just to clear something up, being well acquainted with CAT, you really don't just put the dog in front of another dog and ignore the carry on. It might happen by accident, but it shouldn't happen by design. Going well over threshold doesn't give you much to work with and carries a risk of the strategy back-firing in the same way that flooding can back-fire.

[Aidan3]

By that stage they never chased the sheep without cue because we'd spent 2 years teaching the rules for approaching a flock, and that unless that cue is given the sheep are not available, but the ball IS available. Management was a huge part of that in the beginning but not once it was trained and proofed. Our older dog will frequently duck into the sheep paddock to retrieve a ball then come straight back. Again it wasn't a quick process but it's a thing that works on some dogs.

So dealing with the OP's question, you never trained out an unwanted behaviour. Rather, you trained in the one you wanted. Best possible approach but not what the OPs question is asking.

How would you have dealt with an experienced sheep chaser?

I think it's important to note that punishment doesn't suppress sheep-chasing indefinitely. I would suggest that it should be followed up with training calm, affiliative behaviours around sheep.

I suspect the main failure of differential reinforcement of incompatible behaviours for stock-chasers is that the trainer is always in the picture. The dog fails when the trainer isn't there. There's nothing magical about aversives or reinforcers that makes them better or worse at training out unwanted behaviours in the long run. Lots of small reinforcers, applied with skill, will beat a big one - it doesn't matter if it's a positive reinforcer or a negative reinforcer, this result holds true. The benefit of the e-collar in this situation is that you can make it a really big reinforcer (and you don't need to be in the picture), thus negating the need to use lots of small reinforcers.

This, of course, must be weighed up in an "ends justifies the means" ethical consideration. There will be fall-out.

Edited April 3, 2014 by Aidan2

[Aidan3]

Amax-1 that's so true, and how many of the "untrainable" dogs will end up being PTS because the owners can't or won't access aversives.

It's a furphy. There are many aversives that trainers can use. The issue is that many trainers (of all types) are trying to train dogs that have problems above their ability to train them.

[megan_]

Dobie mum there's a method called CAT that dies basically what you did for dog aggressive dogs. Quite interesting reading, they put the dog in front of another dog and just ignore the carry on. You need a succession of bomb proof tester dogs because obviously it's pretty stressful for them. Apparently at some point and it can be after two days the aggro dog will give up and try a pro social behaviour.

I did have a trainer start something similar with mine but we really needed more dogs, time and a really large training area because it's loud.

Just to clear something up, being well acquainted with CAT, you really don't just put the dog in front of another dog and ignore the carry on. It might happen by accident, but it shouldn't happen by design. Going well over threshold doesn't give you much to work with and carries a risk of the strategy back-firing in the same way that flooding can back-fire.

Yeah that sounds a lot like flooding. - and you'd be hoping your dog doesn't redirect.

[corvus]

Corvus I wonder if these dogs would also be low on a risk taking measure, they get stuck just doing a behaviour in that same thoughtless manner. This I think may be why increasing confidence and especially the clicker training where they start to generate their own "new" behaviours can be helpful. It was a game changer for Jake.

Well, we would expect that if they are frequently getting scared or upset they are probably not in a great place emotionally and will be expecting more bad things to happen to them and less good things. I wouldn't take risks if I expected things to turn out poorly, and I guess they wouldn't either. On the other hand, if they are the kind of dog that is inherently risk averse, maybe highly emotionally reactive, and not skilled at solving problems, then it's like a whole lot of things are conspiring against them, really. I am not surprised that clicker training with Jake has been so positive for him. A while ago I wrote a series on risk aversion in dogs for the blog I linked to earlier. The third instalment has some hints for helping dogs learn to be more persistent, resilient, confident, and optimistic.

Sorry, I'm not sticking around. :) Just wandering through the ether and just happened to see something I know a bit about.

[BC Crazy]

Oh bummer Corvus Can I ask where I can the blog you have written regarding risk aversion if you do pop back in. I would be very interested to read it especially the 3rd instalment.

Some very interesting posts on this thread. IMO I believe whatever method you use would very much depend on the nature of the dog you are working with to begin with. Flooding for example with my girl would be a total disaster. She is way too fragile. She would just have a complete mental meltdown. Redirection & reward seems to work best for her.

[alpha bet]

BC Crazy - Flooding is a term often misunderstood.... and at times misused, however can be very effective if worked in stages.

We get loads of reactive border collies as we live in a semi rural area and BC's are often the dog of choice for many such families. If placed in an environment with lots of other dogs and just ignored they start to become more interested and curious - hence they start to adjust their behaviour and become more accepting of new situations.

[Aidan3]

Just a point, if you're doing it in stages then it's either systematic desensitisation or habituation. Agree that flooding is not necessarily a bad thing, though. It depends a bit on whether the dog is likely to cope and adapt favourably or not.

[hankdog]

I just read the CAT page and yes it was more of a "flooding" rather than a true CAT that we tried with Jake. I guess his sight and sort of on or off at that stage meant that keeping him under threshold was not an option. If he thought it was a dog then he lost it. The only time he went for me was on a prong strangely. Even when doing the flooding I still gave him peanut butter to lick off my hands...eek now I think of it.

If a dogs reactivity was more of a phobic response than a learned strategy then I think flooding might be effective.

Could dogs have a phobia? How would a phobic response look different?