Dont Say No To Your Dog?

[gsdog2]

Yeah full trial...not just the bite work.

I will be very interested to see a video of your schutzhund dog trained using praise/correction alone, with no toys or food, Longcoat, if you would care to post a video of the trials you compete in? It will be interesting to see if you can gain results, and scores, comparable to those teams using food and toy motivators.

Agreed

Seriously LC, could you show us a video of your dog in competition?

[Longcoat]

When a distraction is greater than the reward, that's when purely positive methods fail in my experience.

Depends on the reinforcement history and how well the response has been conditioned. I'm not 'purely positive', but the value of an individual reinforcement v level of distraction becomes a non-issue even at my level of expertise.

It certainly does depend upon reinforcement history but in the reliability crunch, there will be some distraction somewhere greater than the reward on offer and the dog will bolt, or disobey especially off leash.

Diva was right, it is not that the distraction is "greater than the reward on offer" but that the dog is not conditioned, through it's reinforcement history, to respond in that environment. Believe it or not, it is not a competition between reward and distraction (unless we are discussing bribery). Otherwise how would you explain dogs who recall off live prey, not knowing whether there is a reward available or not?

That's were a dog trained to recall to avoid punishment works in any environment. I agree with what you are saying here Aidan, but conditioning a dog to respond in so many different environments, there will be one environment missed being the time the dog finds it's fate. That's were E Collars etc come into play, drop means drop so to speak.

[Longcoat]

Yeah full trial...not just the bite work.

I will be very interested to see a video of your schutzhund dog trained using praise/correction alone, with no toys or food, Longcoat, if you would care to post a video of the trials you compete in? It will be interesting to see if you can gain results, and scores, comparable to those teams using food and toy motivators.

Agreed

Seriously LC, could you show us a video of your dog in competition?

No problems, I will organise one

[Aidan]

A few have tried to train police and security K9's with food and positive reinforcement and what suffered was reliablity.

That is not universally true. Selecting a few non-specific examples does not the truth reveal.

It has been demonstrated repeatedly that positive reinforcement will produce extraordinarily reliable behaviours over thousands upon thousands of trials. Punishment does not produce the same level of reliability in the long-term.

Practical considerations, not ultimate reliability, dictate the use of +P and -R in the field.

What some of the K9 trainers have told me Aidan, is that some behaviours cannot be corrected effectively with positive reinforcement and punishment in those circumstances creates a better result???.

I could say with some certainty that those behaviours could be corrected with +R, but as a practical matter it doesn't always make sense to take what might be the long way around (and sometimes it does, but we don't always see that either!)

[Staranais]

My girl does "round the house" obedience such as drops, stays, off leash walking, etc, for just praise/correction. I refuse to shovel treats into her for easy, routine requests. It's just not necessary.

But I can't imagine training her for her SAR work with no extrinsic rewards. I have never seen, or heard, of a SAR dog working for praise alone. All our dogs get a huge play and praise reward when they find the victim or end of the trail (except for the few dogs who prefer, and get given, food and praise instead). We reward because we want them working extremely enthusiastically for long periods of time. It's not sufficient to have a dog working half heartedly, going through the search routine just because the handler says so. We want dogs who live to work, and that requires huge rewards.

It's the same with all the schutzhund dogs I've seen - all good schutzhund dogs I've seen work in either prey or food drive in the obedience phase (and food drive in the tracking phase). They do not work just to avoid correction, or gain praise.

I will be very interested to see a video of your schutzhund dog trained using praise/correction alone, with no toys or food, Longcoat, if you would care to post a video of the trials you compete in? It will be interesting to see if you can gain results, and scores, comparable to those teams using food and toy motivators.

You have just confirmed that a "few" SAR dogs are rewarded with food obviously not all which is the point.

Yes, the rest are trained with toys. None are trained with praise/correction alone.

William Koehler's were................how did he manage that???

You're joking, right?

Koehler was training dogs well before the advent of modern SAR theory, e.g. by the Syrotucks in the 70s. To my knowledge, and please correct me if I'm wrong, he didn't train any dogs in SAR airscent at all.

And his tracking method, "force tracking" based on the force retrieve, is out of favour for several very pertinent reasons. I would think you would find it difficult to find any modern search trailing/tracking team using Koehler tracking methods.

Why? Because reward based methods generally work better in producing enthusiastic, focused scenting dogs. We want to train really good search dogs. Not ones that do an average job.

Again, I'll be genuinely interested to see your trialling videos of a dog trained using no food and no toys, if you're intending to trial, and see how they compare to teams who have used toy and food motivation.

[Aidan]

When a distraction is greater than the reward, that's when purely positive methods fail in my experience.

Depends on the reinforcement history and how well the response has been conditioned. I'm not 'purely positive', but the value of an individual reinforcement v level of distraction becomes a non-issue even at my level of expertise.

It certainly does depend upon reinforcement history but in the reliability crunch, there will be some distraction somewhere greater than the reward on offer and the dog will bolt, or disobey especially off leash.

Diva was right, it is not that the distraction is "greater than the reward on offer" but that the dog is not conditioned, through it's reinforcement history, to respond in that environment. Believe it or not, it is not a competition between reward and distraction (unless we are discussing bribery). Otherwise how would you explain dogs who recall off live prey, not knowing whether there is a reward available or not?

That's were a dog trained to recall to avoid punishment works in any environment. I agree with what you are saying here Aidan, but conditioning a dog to respond in so many different environments, there will be one environment missed being the time the dog finds it's fate. That's were E Collars etc come into play, drop means drop so to speak.

Behaviours taught with -R don't necessarily generalise any better than behaviours taught with +R. The e-collar has practical benefits in the field but I wouldn't say that a dog wearing an e-collar was a more reliable dog than a dog who was taught the same thing with food if both were then put into a trial situation or maybe even a highly distracting situation (with the e-collar dog wearing the e-collar - they aren't magical!)

If you were to compare two dogs in a Skinner Box, one trained with food and the other trained with shock, then removed the food and removed the shock I would put money on being able to train the one with food to perform more trials until extinction than the one trained with shock. I'd put that money down because I know the probability is higher from repeated experiments.

Of course we don't train (or even trial) dogs in a Skinner Box, so practical considerations come into play. However, every year we see more and more dogs on the podium trained with more and more +R, we're finding ways around those practical considerations and even discovering some false economies in using aversives along the way.

[bedazzledx2]

Well said Aiden! My money's on you!!!!

When a distraction is greater than the reward, that's when purely positive methods fail in my experience.

Depends on the reinforcement history and how well the response has been conditioned. I'm not 'purely positive', but the value of an individual reinforcement v level of distraction becomes a non-issue even at my level of expertise.

It certainly does depend upon reinforcement history but in the reliability crunch, there will be some distraction somewhere greater than the reward on offer and the dog will bolt, or disobey especially off leash.

Diva was right, it is not that the distraction is "greater than the reward on offer" but that the dog is not conditioned, through it's reinforcement history, to respond in that environment. Believe it or not, it is not a competition between reward and distraction (unless we are discussing bribery). Otherwise how would you explain dogs who recall off live prey, not knowing whether there is a reward available or not?

That's were a dog trained to recall to avoid punishment works in any environment. I agree with what you are saying here Aidan, but conditioning a dog to respond in so many different environments, there will be one environment missed being the time the dog finds it's fate. That's were E Collars etc come into play, drop means drop so to speak.

Behaviours taught with -R don't necessarily generalise any better than behaviours taught with +R. The e-collar has practical benefits in the field but I wouldn't say that a dog wearing an e-collar was a more reliable dog than a dog who was taught the same thing with food if both were then put into a trial situation or maybe even a highly distracting situation (with the e-collar dog wearing the e-collar - they aren't magical!)

If you were to compare two dogs in a Skinner Box, one trained with food and the other trained with shock, then removed the food and removed the shock I would put money on being able to train the one with food to perform more trials until extinction than the one trained with shock. I'd put that money down because I know the probability is higher from repeated experiments.

Of course we don't train (or even trial) dogs in a Skinner Box, so practical considerations come into play. However, every year we see more and more dogs on the podium trained with more and more +R, we're finding ways around those practical considerations and even discovering some false economies in using aversives along the way.

[Jeff Jones]

I've not heard of a police dog anywhere in the world trained successfully with purely positive training, but that's not what we're talking about.

I guess it depends on your definition, but Steve White was doing what I would consider to be purely positive training.

I have seen Steve train dogs completing tracking and obedience but have never seen or heard about Steve completing bitework using purely positive methods.

[OSoSwift]

I would imagine that the dog being allowed to actually do the bite work could be seen as a positive reinforcement to the dog performing it? Therefore it would be reward enough and no other treats/toys would be necessary especially for a highly prey driven dog.

[bedazzledx2]

Absolutely! The same could be said of the herding breeds doing what they are bred to do. You don't see farmers handing out tit bits for the dogs when they round up the sheep! They don't need to as the work is inherently rewarding to dogs that are bred to herd.

I would imagine that the dog being allowed to actually do the bite work could be seen as a positive reinforcement to the dog performing it? Therefore it would be reward enough and no other treats/toys would be necessary especially for a highly prey driven dog.

[4Kelpies]

Absolutely! The same could be said of the herding breeds doing what they are bred to do. You don't see farmers handing out tit bits for the dogs when they round up the sheep! They don't need to as the work is inherently rewarding to dogs that are bred to herd.

I agree. Positive training is all about using rewards that the dog finds rewarding - not just food. My dogs all love their food and I've used treats to teach sit, drop, come, stay, loose leash walking and a whole range of behaviours but I wouldn't dream of getting between my dogs and a sheep with a piece of food. They just wouldn't see it. I love ice cream but if I was about to serve for a match in tennis and someone asked me if I wanted an ice cream I'd tell them to get lost.

Having taught my pups all the above behaviours I've then been able use them when working stock but working sheep has always been the reward. I've used a clicker when teaching contacts in agility where the reward was a tunnel or another jump. The dog would get to the correct position and be quivering with excitement waiting for the click and release to race off to the next obstacle. I used food when laying tracks for my old dog when I was teaching him to track. In a very short time he would track over the top of the food without pausing to eat any. I would lay the track and leave some toys and food at the end but he wasn't particularly interested in them. He was following my scent but wasn't finding me as I was behind him when he was tracking. The only reward here seemed to be the sheer joy of tracking. It's impossible to know exactly what dogs are thinking but it seems to me that rewarding my dogs helps them to learn that it's worthwhile being around me and co operating with me because good things happen when we're working together.

[centitout]

My girl does "round the house" obedience such as drops, stays, off leash walking, etc, for just praise/correction. I refuse to shovel treats into her for easy, routine requests. It's just not necessary.

But I can't imagine training her for her SAR work with no extrinsic rewards. I have never seen, or heard, of a SAR dog working for praise alone. All our dogs get a huge play and praise reward when they find the victim or end of the trail (except for the few dogs who prefer, and get given, food and praise instead). We reward because we want them working extremely enthusiastically for long periods of time. It's not sufficient to have a dog working half heartedly, going through the search routine just because the handler says so. We want dogs who live to work, and that requires huge rewards.

It's the same with all the schutzhund dogs I've seen - all good schutzhund dogs I've seen work in either prey or food drive in the obedience phase (and food drive in the tracking phase). They do not work just to avoid correction, or gain praise.

I will be very interested to see a video of your schutzhund dog trained using praise/correction alone, with no toys or food, Longcoat, if you would care to post a video of the trials you compete in? It will be interesting to see if you can gain results, and scores, comparable to those teams using food and toy motivators.

You have just confirmed that a "few" SAR dogs are rewarded with food obviously not all which is the point.

Yes, the rest are trained with toys. None are trained with praise/correction alone.

I start my trailing dogs off with food,and every now and again i 'Jackpot them" to keep them motivated.None of mine have been play driven except for the one in Sydney with another trainer,and his reward is his tennis ball.

When they associate the "victim" with the food/praise reward,i start to use more basic food rewards like dried liver etc,and at the end when they are back at the car-they get their ultimate-a chicken frame,roo tail etc,bu ti still jackpot them occaisionally to keep them motivated.

I want my dogs to be motivated,enthusiastic and food seems to do that-you cant have a half arsed search dog.

[Prydenjoy]

Selection played a very large part. Well bred dogs who perform the tasks that they are bred to perform are getting a TON of reinforcement. And if they didn't perform, they didn't get to hang around.

Good point!! The rest were dubbed "dumb" or "untrainable". We now know that to be far from the truth, we were just too "dumb" to work out how to motivate them!!

[huski]

I would imagine that the dog being allowed to actually do the bite work could be seen as a positive reinforcement to the dog performing it? Therefore it would be reward enough and no other treats/toys would be necessary especially for a highly prey driven dog.

My understanding with Schutzhund bite work is that the dog sees the sleeve as a toy.

[Prydenjoy]

If you were to compare two dogs in a Skinner Box, one trained with food and the other trained with shock, then removed the food and removed the shock I would put money on being able to train the one with food to perform more trials until extinction than the one trained with shock. I'd put that money down because I know the probability is higher from repeated experiments.

Sometimes I wonder why people even bother doing research when so many people just ignore the results and think that they know better!

[Kavik]

The obedience phase of Schutzhund is made of up several activities using different drives and as we mentioned previously, you can't use any motivators in a trail except voice and praise. Anything used in training has to be weaned off. Generally you take the dog from the crate onto the field and can't use motivating devices on the day.

I have seen SchH dogs tugging with their lead with their handler after their obedience routine in a trial

[Kavik]

You are not going to get results like this with praise and correction alone! Having watched the DVD - WOW! Amazing weave pole entries and performance!

Nor are you going to get a seesaw performance like this!

[Diva]

Aidan, I have no idea what gender or age you are, but sometimes I think I need to either adopt you, hire you or arrange to have you marry into my family! You almost always say what I believe or know, but so much more succinctly and wisely.

(note to self, never meet Aidan in real life, now too embarrassing )

Edited April 5, 2010 by Diva

[Aidan]

If you were to compare two dogs in a Skinner Box, one trained with food and the other trained with shock, then removed the food and removed the shock I would put money on being able to train the one with food to perform more trials until extinction than the one trained with shock. I'd put that money down because I know the probability is higher from repeated experiments.

I should clarify that I am talking specifically about a punished response here, not any alternative response brought about by negative reinforcement through removal of the shock. These depend entirely on the magnitude of shock and the schedule of reinforcement used.

[Aidan]

Aidan, I have no idea what gender or age you are

If we ever do meet how will you know it's really me?