Excrutiatingly Stubborn Dog

[Aidan3]

but the falling out from incorrect application is handlers fault not the method.

That's right, and fall-out from correct application is also the handler's fault.

so when the dog attacking other dogs in the park you ignore when he attack and reward him when he doesnt is good method that one yes?

Yes, now you're getting it! It's exactly like training with an e-collar, you take the vicious dog to the park, sic it onto another dog then zap the hell out of it. See, we're all experts on the interwebz.

[JoeK]

but the falling out from incorrect application is handlers fault not the method.

That's right, and fall-out from correct application is also the handler's fault.

so when the dog attacking other dogs in the park you ignore when he attack and reward him when he doesnt is good method that one yes?

Yes, now you're getting it! It's exactly like training with an e-collar, you take the vicious dog to the park, sic it onto another dog then zap the hell out of it. See, we're all experts on the interwebz.

Falling off doesnt happening if correction is done properly or maybe dog is alredy fallen off by the scruff his mother give him on the puppy, the bitch doesnt give him a treat, she growl and mouth him with complusion for bad behavior, so what happens on the fall off here?

If you using Ecollar, better making sure is not type of dog is ramping up the aggression on the stimulation and turn him into the serial killer.

Edited June 3, 2011 by JoeK

[Mrs Rusty Bucket]

so when the dog attacking other dogs in the park you ignore when he attack and reward him when he doesnt is good method that one yes?

Fall out - think nuclear fallout from a-bombs. Ie the bomb flattens this spot here but then contaminates with radioactive dust fallout everywhere downwind.

So I'm curious - JoeK - what strategies would you use with the aggressive dog? Would you take an aggressive dog like this to a park with other dogs without retraining it somewhere else first?

Edited June 3, 2011 by Mrs Rusty Bucket

[Aidan3]

Falling off doesnt happening if correction is done properly

Not always, and it's not always a problem when it does. Do you know much about the cause and effect fallacy or confirmation bias?

Bitches certainly can have a negative influence on their pups, just like human mothers. I tend to think that bitches have the advantage over us in terms of non-verbal communication, though.

If you using Ecollar, better making sure is not type of dog is ramping up the aggression on the stimulation and turn him into the serial killer.

Are you saying you would actually do what I suggested so long as the dog wasn't the "ramping up the aggression" type? Not around my dog you wouldn't!

[PetSitters]

Falling off doesnt happening if correction is done properly

Not always, and it's not always a problem when it does. Do you know much about the cause and effect fallacy or confirmation bias?

Bitches certainly can have a negative influence on their pups, just like human mothers. I tend to think that bitches have the advantage over us in terms of non-verbal communication, though.

If you using Ecollar, better making sure is not type of dog is ramping up the aggression on the stimulation and turn him into the serial killer.

Are you saying you would actually do what I suggested so long as the dog wasn't the "ramping up the aggression" type? Not around my dog you wouldn't!

Aidan, how is fallout determined scientifically when a dog for example that has been compulsively trained dispays symptoms when the same dog cannot be wiped of it's former training experiences to re examined after training in a different method from a clean sheet of paper so to speak? I ask on the basis that different dogs have different responses and some I would imagine to suffer varying levels of fallout where I don't see enough uniformity to make firm conclusions. Even litter mates can respond differently so perhaps the measurement of complusion applied to an extremely hard dog over an extremely soft dog would alter lab results dramatically?

[Aidan3]

Aidan, how is fallout determined scientifically when a dog for example that has been compulsively trained dispays symptoms when the same dog cannot be wiped of it's former training experiences to re examined after training in a different method from a clean sheet of paper so to speak? I ask on the basis that different dogs have different responses and some I would imagine to suffer varying levels of fallout where I don't see enough uniformity to make firm conclusions. Even litter mates can respond differently so perhaps the measurement of complusion applied to an extremely hard dog over an extremely soft dog would alter lab results dramatically?

Excellent questions, PetSitters. When we're working with a single animal this is what is known as a "single-subject design", which has a number of problems (and some advantages) which you can Google, there should be some accessible info on SSDs out there. In a nutshell, it would be very difficult for a dog trainer to correctly attribute an effect to a cause using a single dog. It would be very difficult for a research scientist or applied behaviour analyst, for that matter.

When you start talking about groups of dogs we have ways to address the problems inherent with SSDs. Nevertheless, statistical analyses makes several assumptions which we have to be confident that we can make about the dogs we have in front of us. This makes it difficult to work with very small groups of dogs, or dogs of different breeds with widely varying traits unless we're using tightly controlled conditions.

Luckily,the most basic laws of learning have been found to be highly generalisable. We can do things in tightly controlled environments, or with highly uniform samples (a sample is a group of subjects), or even with completely different species, then make observations and see if they apply in the real world, or with the species we are more interested in. This can enable us to confidently make some assumptions about, say, dog behaviour that are based on experimental data which can reveal causal relationships.

Causal relationships are links between cause and effect, which are often very difficult to see, much less prove. Humans are notoriously bad at getting them right, we learn all sorts of things that influence our decisions and even our observations (not the YouTube vid with the balls). We often confidently make false attributions, e.g my child became hyper after drinking red cordial, completely missing the facts that my child did not become hyperactive after drinking red cordial yesterday, and that my child was at a party playing with lots of very excited kids when she drank red cordial this time.

Right from the earliest days of behaviourism, punishment was observed to have all sorts of unintended consequences that could be very difficult to predict. Sensation, perception, prior learning, biology and cognition influence the way that animals respond to aversion. Skinner studied punishment deeply and eventually decided that we should avoid it. Things have changed a bit since then, we know a bit more about punishment, but Skinner's observations are still relevant.

All the things that I have mentioned in earlier posts are things that have been observed in the lab (and outside the lab), and it is prudent to consider them when working with dogs. We need to keep an open mind and accept the possibility that our actions, no matter how skillful, may have unintended consequences that we might not see or causally link to an effect. How much we err on the side of caution is probably a personal choice, and will probably change for each person over time.

[Mrs Rusty Bucket]

Even litter mates can respond differently so perhaps the measurement of complusion applied to an extremely hard dog over an extremely soft dog would alter lab results dramatically?

Given that the fallout from the same punishment can be unpredictable - ie it depends on how the dog perceives things and that depends on each individual dog and its life experiences - seems enough of a good reason to avoid it.

If it's hard to control, manage and measure in a laboratory, consider how much harder it is to control in a normal training environment.

While I hate referencing wiki, this seems to explain all the techinical terms and the science really well.

http://en.wikipedia.org/wiki/Reinforcement

In particular

As Skinner discussed, positive reinforcement is superior to punishment in altering behavior. He maintained that punishment was not simply the opposite of positive reinforcement; positive reinforcement results in lasting behavioral modification, whereas punishment changes behavior only temporarily and presents many detrimental side effects.[2]

[m-j]

Falling off doesnt happening if correction is done properly or maybe dog is alredy fallen off by the scruff his mother give him on the puppy, the bitch doesnt give him a treat, she growl and mouth him with complusion for bad behavior, so what happens on the fall off here?

In short the pups can learn to ignore her i.e. when the pups want to feed and she won't let them but they insist and blow her reprimands off, so she has to get up and move. When the payoff it great it is worth taking the risk of the correction to gain the reward. Human drug couriers do it a lot, they risk the ultimate punishment of death to gain the reward. Greys at work that blow off an electric fence to get a bunny, whereas the same fence if touched by the same dog in different circumstances sends the dog back up the other end of the 200m paddock/pen screaming, at a very fast pace.

Also mum's do "give treats" they play with and groom their pups more often than they reprimand them (well this what I have experienced with the few litters I have raised) I don't know if mum had asked for something other than sometimes pinning the pup down initially so it can be groomed, but I don't believe dogs tell each other to do things anyway, only not to do things and this is only if they are prepared to get into a conflict, the exception could be co-operative hunting, although I'm not really sure about this ???

[Mrs Rusty Bucket]

Human drug couriers do it a lot, they risk the ultimate punishment of death to gain the reward

I don't think a punishment that has never been experienced has any effect on the rate of desired response. The drug courier does the behaviour because they have experienced the reward of lots of money or drugs. The punishment has to be applied consistently, every time they do the behaviour to have effect in reducing the rate of the undesired response (eg transporting drugs). In the case of death being the punishment - it will completely reduce the undesired response but have little to no effect on other couriers.

And yes based on learning science, most of our prison system is counter productive.

As far as the mother correcting the puppy. Hopefully she gives some warning (cue) first and follows that up with the correction - which may only be to limit the puppy's opportunity to behave undesirably (eg bite mum). So puppy learns to back off if it hears the growl ie there will be no reward for continued approach. And it learns that if there is no growl on approach, the opportunity for the reward of a feed is there. But the lesson that lasts longest - is the one of reward for approach (for feed).

[m-j]

Human drug couriers do it a lot, they risk the ultimate punishment of death to gain the reward

I don't think a punishment that has never been experienced has any effect on the rate of desired response. The drug courier does the behaviour because they have experienced the reward of lots of money or drugs. The punishment has to be applied consistently, every time they do the behaviour to have effect in reducing the rate of the undesired response (eg transporting drugs). In the case of death being the punishment - it will completely reduce the undesired response but have little to no effect on other couriers.

In many cases this probably is true, but we have an innate desire to stay alive and if we know this is a consequence I imagine it would be considered. If I was/had ever been given the opportunity to have to make that decision I believe it would put me off, even at the stage in my life where money was short and risk taking not such a determining factor in making any decision. :rolleyes:

What about the example of the Greys that have experienced the fences reprimand, but still choose to blow it off. I know that these examples are anecdotal but when training using compulsion aren't dogs weighing up consequences to avoid a reprimand, by not doing a desired behaviour anyway?

I believe that the conflict in decision making can be also considered as fallout to training with adversives, if you desire an immediate response i.e. life threatening situation of dog not choosing to drop immediatly instead of running across the road to chase cat when a car is coming.

As far as the mother correcting the puppy. Hopefully she gives some warning (cue) first and follows that up with the correction - which may only be to limit the puppy's opportunity to behave undesirably (eg bite mum). So puppy learns to back off if it hears the growl ie there will be no reward for continued approach. And it learns that if there is no growl on approach, the opportunity for the reward of a feed is there. But the lesson that lasts longest - is the one of reward for approach (for feed).

Exactly and if the pup is determined, the warning and the reprimand will be blown off and if there are several pups doing the same thing the bitch will get up and move. I always provide a way of letting the bitch escape if she feels the need, during the weaning period. It doesn't mean they are neccessarily hungry , they just want mum's food not mine

[Aidan3]

If it's hard to control, manage and measure in a laboratory, consider how much harder it is to control in a normal training environment.

Just to be clear, the problem isn't that fallout is hard to measure, it's hard to correctly attribute to a cause and often gets overlooked because trainers (or teachers, or parents) are looking elsewhere.

[PetSitters]

Aidan, how is fallout determined scientifically when a dog for example that has been compulsively trained dispays symptoms when the same dog cannot be wiped of it's former training experiences to re examined after training in a different method from a clean sheet of paper so to speak? I ask on the basis that different dogs have different responses and some I would imagine to suffer varying levels of fallout where I don't see enough uniformity to make firm conclusions. Even litter mates can respond differently so perhaps the measurement of complusion applied to an extremely hard dog over an extremely soft dog would alter lab results dramatically?

Excellent questions, PetSitters. When we're working with a single animal this is what is known as a "single-subject design", which has a number of problems (and some advantages) which you can Google, there should be some accessible info on SSDs out there. In a nutshell, it would be very difficult for a dog trainer to correctly attribute an effect to a cause using a single dog. It would be very difficult for a research scientist or applied behaviour analyst, for that matter.

When you start talking about groups of dogs we have ways to address the problems inherent with SSDs. Nevertheless, statistical analyses makes several assumptions which we have to be confident that we can make about the dogs we have in front of us. This makes it difficult to work with very small groups of dogs, or dogs of different breeds with widely varying traits unless we're using tightly controlled conditions.

Luckily,the most basic laws of learning have been found to be highly generalisable. We can do things in tightly controlled environments, or with highly uniform samples (a sample is a group of subjects), or even with completely different species, then make observations and see if they apply in the real world, or with the species we are more interested in. This can enable us to confidently make some assumptions about, say, dog behaviour that are based on experimental data which can reveal causal relationships.

Causal relationships are links between cause and effect, which are often very difficult to see, much less prove. Humans are notoriously bad at getting them right, we learn all sorts of things that influence our decisions and even our observations (not the YouTube vid with the balls). We often confidently make false attributions, e.g my child became hyper after drinking red cordial, completely missing the facts that my child did not become hyperactive after drinking red cordial yesterday, and that my child was at a party playing with lots of very excited kids when she drank red cordial this time.

Right from the earliest days of behaviourism, punishment was observed to have all sorts of unintended consequences that could be very difficult to predict. Sensation, perception, prior learning, biology and cognition influence the way that animals respond to aversion. Skinner studied punishment deeply and eventually decided that we should avoid it. Things have changed a bit since then, we know a bit more about punishment, but Skinner's observations are still relevant.

All the things that I have mentioned in earlier posts are things that have been observed in the lab (and outside the lab), and it is prudent to consider them when working with dogs. We need to keep an open mind and accept the possibility that our actions, no matter how skillful, may have unintended consequences that we might not see or causally link to an effect. How much we err on the side of caution is probably a personal choice, and will probably change for each person over time.

Thanks for your explanation Aidan appreciated

[jackie_a1]

Lack of consistency and really, what incentive is there to work for you? If he doesn't listen whats the big deal.

He's an adult GSD now you need to 'boot camp' him a little if he's not interested in food dont feed him out of a bowl anymore.

Nekhbet I love you and couldn't agree more

[Mrs Rusty Bucket]

What about the example of the Greys that have experienced the fences reprimand, but still choose to blow it off. I know that these examples are anecdotal but when training using compulsion aren't dogs weighing up consequences to avoid a reprimand, by not doing a desired behaviour anyway?

This one I've heard described as pavlov beating skinner. Ie the conditioned response or the one they do without thinking - beats the trained response ie the one by operant conditioning. So - it is extremely difficult (tho not impossible) to train a dog to go against its instinct. Ie for a greyhound, chasing the rabbit thing is going to win over any training to stay inside a fence. Or maybe just more training is required?

I had a horse, that learned from another horse - how to test an electric fence with one's whiskers, but also that a quick zap from the fence was well worth the payoff of long grass on the other side. So both of these horses would put their heads under the fence strap and run.