Clicker = training???

ORIGINAL: ron2

Is she? What can she offer that she knows will be rewarded and not corrected or punished? She has two options. Execute another behavior and possibly get corrected again or execute a behavior, even though there is possibly no reward or incentive for doing so. Or, shut down and do nothing, which may be the desired response, but is not rewarded. She is not human. Yes dogs are capable of thinking and linking processes, especially in free-shaping. In fact, it's amazing. It's also highly motivating because it is rewarding.

Well the dog will stop after the correction, then he will wait for the owner to tell him what to do (if you want to think about it that way), the owner does nothing and turns back to receive the guest, dog stopped = no correction = stopping = good behavior, voila

ORIGINAL: ron2

But you will want her "decision" to be a behavior that you want, not necessarily what her first idea was.

Then it was not her "decision", it was yours, maybe her "decision" was walking away, but you wanted her to sit instead

In addition to sheep training, I train dogs to work chasing geese off various properties. I have to admit that when I read this:

You deliver a lot of corrections you end up with a behaviorally inhibited dog.

I thought, "Ok, and the bad part about that would be?"

Dogs that think twice before offering a behavior live to work another day. They don't dash across traffic lanes, chase down power equipment, fly off rooftops (yes, one of my dogs went to the DC area where part of his patrol area turned out to be a rooftop), harass injured wildlife, chase deer or cattle or racehorses (another one of my dogs works several thoroughbred farms in the NoVA area), they don't bother golfers playing through or boaters or fishermen out on the lake, they wait patiently on the cart or boat until asked to work.

AND on top of this they still have the ability to think through very complex problems set to them by escaping geese or other waterfowl. These fowl get acclimated quickly and the dogs have to adjust to the things the fowl do to avoid the dogs or discourage them.

I just did a little experiment. I realized that I've never tried clicker training with Cord (I've done a little treat and lure training for tricks, nothing formal). Cord will be seven next month, and has never been anything but traditionally trained (with the aforementioned exception). He IS rather handshy, thanks to some rough prior handling - of all my dogs I'd expect him most of all not to "get" free shaping.

There wasn't any difference between him and puppies I've done this with. I took some video but it's a lovely shot of my back, a pillow that was in the way, and the top of Cord's head. [8D] Still, you can see that we charged the clicker, and captured a head turn and a rear paw scratch within the two minute limit of my memory card. [] He doesn't have any problems experimenting and putting two and two together. I have a pitcher here that I'll play with after I clear out my memory card. And try to get a better angle with the camera [8|]

I was fairly confident about how this would go because my old Rick dog was introduced to the clicker for the first time when he was about four, after some very intense sheepdog training and competition. That dog would fly through behaviors at the slightest suggestion you wanted something - it was hard to capture anything, in fact, because stuff would go by so fast, it was easy to mark the wrong thing. But the key was that he'd wait for the context that meant we were playing.

Back later, maybe with video.

Well, I got a nice video but can't figure out stupid YouTube. I suspect my file size is too big, though YouTube isn't letting on. Anybody got a web site that will let me post a 300+ MB video for free? I can't edit it because IMovie thinks it's a bad file for some reason. It plays just fine straight from the file menu, however.

One of the big problems with using corrections is it's very hard for the dog to figure what, exactly, he did that caused the correction to occur. Example: owner perspective: dog lunges sideways, pulling on leash; owner gives a mild collar pop to correct the pulling on leash. Dog perspective: I was smelling cat scent, I was wagging my tail, I had my front feet on grass and my back feet on pavement, the leash was taut, I got a pop. The only way the dog can figure out which of the six behaviors he was engaged in at the time of the correction was the "bad" one is to experiment and see which one(s) earn him collar pops. When you're using +R to "show a dog what to do" that is exactly what dogs do-- they experiment to see what, exactly, it is that they are doing to get the reward. Dogs really aren't very motivated to "experiment" with corrections, and most likely will just stop doing any and all things they thought might have gotten them the correction. You deliver a lot of corrections you end up with a behaviorally inhibited dog. Unfortunately that is exactly what a lot of people want.

This is well-put. So is Rebecca's response.

I have a dog that will never run across the street because when he was a pup he ran across the street and came extremely close to being hit by a car. It was a scary experience for him. This is a good thing. I don't want a dog that is unafraid of the street.

Most situations that my dog can get into, though, are not that serious. I am glad, in a weird way, that he scared the poop out of me by running into the street, because it was an extremely effective correction.

But I don't see the value in him learning most things that way for two reasons:

1. In order for a correction to work, it has to be intense and dramatic enough to really change the dog's behavior. This is a fundamental of the Monks/Koeheler method, a method that I have used. I don't enjoy scaring/irritating/hurting my dog enough to change his behavior.

2. If you're not truly correcting with the correction, then you are nagging the dog with regular pssts or pokes or pulls or whatever, and that seems as pointless to me as screaming a dog's name louder in the hopes that he honestly can't hear you or will understand you better if you scream.

Either you get a dog that is inhibited (not offering behaviors in hopes of avoiding correction, which is what you want when you are doing correction-based training), or a dog that has no reason to listen to you because all you do is pull and poke at it.

What really gets me when we talk about correction-based training (which does work--I have used it myself and understand Brookcove's take on it when the stakes are high!) is that nobody is being honest about what a correction is.

Either a correction is strong enough to correct the behavior (ie, it hurts or scares or otherwise negatively impacts)

Or it's not effective. Even the monks of New Skete will tell you that it is abusive (and that's their word, not hyperbole or "clicker extremist dogma") to correct a behavior repeatedly. I like the monks because they are honest about the fact that a correction negatively impacts the dog. That's the whole point.

Woo hoo, Fisher! Thank you for making a great ending to a long and crappy week. Thank you for being so clear and so precise in your post.
 
My sentiments--and my beliefs--exactly.
 
I realize I'm not a big-time trainer, but seeing the results of clicker training (and I see it as training) have been more than anything I could have ever read. Totally. For years I was a Koehler handler, and I was pretty darn good at it, too, and now I try to stay as far away from Koehler as I possibly can.
 
Yep, I gave it up. And I've received so much more in return. Amazing.

Well the dog will stop after the correction, then he will wait for the owner to tell him what to do (if you want to think about it that way), the owner does nothing and turns back to receive the guest, dog stopped = no correction = stopping = good behavior

 
So, you stopped the dog with a correction. Where is her motivation to do something polite or to not continue the behavior? Maybe you don't want the dog walking away and getting into something else she decides to do. In some situations, the ideal behavior would be to sit, down, or stand in the heel position. In any case, you have already directed her away from her original decision to jump on a guest.
 
It's not that I'm particularly against a correction, even a mild verbal negative marker. I am for positive motivation to behave in the manner that is best suited for the situation. I have jump on cue, for showing off. And showing "off." And I'm still not seeing how commanding a motivated obedience does not allow the dog to think. Whereas, a correction could lead to a cessation of any thinking in order to avoid further correction. A shut down.
 
Just issuing a correction without following with a positively motivated command is like asking them to read your mind.

Dogs that think twice before offering a behavior live to work another day. They don't dash across traffic lanes, chase down power equipment, fly off rooftops (yes, one of my dogs went to the DC area where part of his patrol area turned out to be a rooftop), harass injured wildlife, chase deer or cattle or racehorses (another one of my dogs works several thoroughbred farms in the NoVA area), they don't bother golfers playing through or boaters or fishermen out on the lake, they wait patiently on the cart or boat until asked to work.

I don't understand why you think that clicker trained dogs cannot be taught to wait on a cart until asked to come off, not chase or harass livestock, avoid jumping from rooftops, or dash across a road.  If you ask a clicker trained dog to "stay", she does it, just like a traditionally trained dog.  But, if you want to teach your dog a complex behavior like go get me a beer from the fridge, my money is on the clicker trained dog to learn it faster. 
Also, my sense is that you are correcting your dogs by voice (since they work away from you), and that is not the same as a physical correction.  In fact, to dogs with a long history of positive associations with you, it may be interpreted as a NRM, not so much as a correction, except in those circumstances where you have been very stern.  So, your experience, and that of your dogs, may be quite different from the obedience dog who was taught by "Koehler" type methods, or even the more moderate Monks of New Skete.
The truth is, I can see, even in the posts on this forum, that there are simply some people who want a dog to be "seen but not heard", and don't have the time or patience to engage in clicker training.  I'm glad that ron did, because I think he and his dog are so much happier.  He is JQP to some extent - someone who wanted a good companion dog.  So, to muddy the waters with what is, or is not, appropriate for working dogs might overshadow the fact that most people simply want what ron now has, and he got it easier and faster with clicker training than with anything else he tried.  God love him for having such an open mind.  If he could do it, anyone on idog can surely do it.

I'm glad that ron did, because I think he and his dog are so much happier. He is JQP to some extent - someone who wanted a good companion dog. So, to muddy the waters with what is, or is not, appropriate for working dogs might overshadow the fact that most people simply want what ron now has, and he got it easier and faster with clicker training than with anything else he tried. God love him for having such an open mind. If he could do it, anyone on idog can surely do it.

 
Another nutshell moment. Everything you said is a square solid hit on the nailhead.
 
I am not a dog trainer, either self-defined or certified. Though, yesterday, in public, someone thought I was as I was showing what Shadow could do and how I accomplished and how to use a clicker. Granted, I do have a science and math background but others here don't and accomplish the same goals in about as much time.
 
You don't have to be Karen Pryor to know and understand this system and appreciate the results. You don't have to read through her groundbreaking papers from the 70's. Accept the results of us JQPs that have tried it. Though I have always lure and reward with Shadow, I have also used body blocking and scruffing. Now, I don't body block but I will step into his line of view, if necessary, and then command a positively trained obedience. And I don't scruff now and that is not because I find the technique distasteful but because the clicker training has been more effective for me and commanding an incompatible behavior gets rewarded and does, in fact, keep him thinking. He may decide to go ahead and down and I will reward that, too. Having seen him in different states of mind, a down is a sign of being relaxed and confident, which is what I need him to be. And most JQP are like me. We have a companion animal that must obey and we live in a litigious society where we are financially responsible for what our pets do. My insurance company pays settlement rewards if you walk onto my lawn and trip over a blade of grass. So, I don't need my dog "deciding" to jump and mouth you and accidently draw blood or make a bruise.
 
And I have found the marker training to be more effective for Shadow because it keeps him motivated. I also feel that I could train him for a working dog role with for the same reason. I've been free-shaping one behavior into a retrieve and drop it. Without remote collars, 15 foot lead, or collar corrections. He'll bring me Jimmy Hoffa for a treat.[]
 
For those that still doubt, approach from my initial perspective that if I fail, at least I had another reason to give him treats and do something fun. And then be amazed at how well it works. Word of advice, though, get the mechanics and sequence down tight so that you don't accidently reward the wrong thing.
 

I've been thinking about this whole "tell the dog what not to do, and he can do anything else" approach. Ok, lets say we have the dog, Big Joe. We tell Joe he can't leap up onto arriving guests. So next time guests arrive he greets them by ramming his nose into each crotch, very hard. We firmly tell him he can't do that either. Next time he grabs the arriving guest's handbag and runs off with it. We tell him he can't do that either. Next time he grabs the guest's pantleg and tugs on it. We tell him no way. By now he's learned arriving guest= Big Joe gets corrected, so he decides next time to take advantage of distracted owners and sneak into the dining room and eat all of the refreshments. We don't catch him doing that, so we can't tell him to not do that. Next time, we shut the doors and force him to stay in the hallway as the guest arrives, to protect our food. Big Joe is afraid of being corrected, so he barks, growls, and lunges at the guest.
So much easier to just tell Big Joe what you want him TO DO when guests arrive.
 
And in any other situation-- if you gently teach your dog what you want him TO DO in most situations, and have some default behaviors (when in doubt, SIT), and a few rock-solid obedience commands (recall, down, heel) you don't end up with dogs who run out into roads, dart out doors, fly off roofs, eat poisoned bait, etc. Your dog knows what to do-- wait for permission to go through doors, stay near owner instead of bolting off, drop goodies found outside on command, etc.

Yet another fabulous Buster Keaton routine! []

Yet another fabulous Buster Keaton routine

thankyouverymuch. [sm=elvis.gif]

ORIGINAL: mudpuppy

I've been thinking about this whole "tell the dog what not to do, and he can do anything else" approach. Ok, lets say we have the dog, Big Joe. We tell Joe he can't leap up onto arriving guests. So next time guests arrive he greets them by ramming his nose into each crotch, very hard. We firmly tell him he can't do that either. Next time he grabs the arriving guest's handbag and runs off with it. We tell him he can't do that either. Next time he grabs the guest's pantleg and tugs on it. We tell him no way. By now he's learned arriving guest= Big Joe gets corrected, so he decides next time to take advantage of distracted owners and sneak into the dining room and eat all of the refreshments. We don't catch him doing that, so we can't tell him to not do that. Next time, we shut the doors and force him to stay in the hallway as the guest arrives, to protect our food. Big Joe is afraid of being corrected, so he barks, growls, and lunges at the guest.
So much easier to just tell Big Joe what you want him TO DO when guests arrive.

And in any other situation-- if you gently teach your dog what you want him TO DO in most situations, and have some default behaviors (when in doubt, SIT), and a few rock-solid obedience commands (recall, down, heel) you don't end up with dogs who run out into roads, dart out doors, fly off roofs, eat poisoned bait, etc. Your dog knows what to do-- wait for permission to go through doors, stay near owner instead of bolting off, drop goodies found outside on command, etc.

The scenario described in the first paragraph may sound funny, but when you train hundreds of humans and dogs each year, you start to think not.  These things really do happen in real life...
So much easier to teach the dog all the things you want him to do. []

I don't understand why you think that clicker trained dogs cannot be taught to wait on a cart until asked to come off, not chase or harass livestock, avoid jumping from rooftops, or dash across a road.

I'm not saying that! This was a response to the suggestion that the inhibited dog is a bad thing. I need a certain amount of inhibitedness - but it's not incompatible with thinking, either. In fact, the dog is actually thinking quite well in that state if he's been trained right (I'm not talking about a dog that's been abused or purposely shut down - I do know trainers like this and I leave them out of this discussion). The ideal working dog is calm, engaged in his environment, ready for anything, and thinking through his choices to fit a suitable action to the moment. He also is not afraid to experiment if experience or instinct is silent on the particular situation. The boundaries of grace.

It's not an either/or proposition for me. I fit the methodology to the goal. I don't dismiss any approach or technique just because it doesn't fit an ideal, short of abusiveness. There are certain things I'm abandoning more and more as less dog friendly - techniques that involve crowding the dog's space, physically manipulating them, pushing or blocking them. That's just a matter of taste, however, and there's still times I need to cut to the chase and guide a dog somewhere, for instance. I hope experience will change that someday.

I practice clicker training less for the dogs and more to improve my own (really pathetic) timing and my focus on the dog. I use it extensively with the rescues, of course. I use a ton of R- with Zhi because it's most effective on her. Cord was trained with a lot of P- so I'm having to use it some to communicate with him. And Doug the Dog, No Fuss Gus, and the pups were trained with the "make what's right easy and what's wrong difficult" P+ based method.

I'm not going to mince words and pretend I'm actually using a NRM when I know very well I'm not. The purpose of my spoken word is to stop the behavior - P+. I want the dog to go, "Oh, what?" Not, "Oh no! I don't get sheep!"

Also, I want to try very hard to never take the sheep away unless there's good reason (truth in advertising, I'm actually very bad about accidentally getting in the dog's way - trying harder, trying harder . . .) - so the reinforcer is always there. It's one of the peculiarities of good stockdog training that I've realized recently, that a good trainer overlays P+ right over the R+. It's like a favorite aunt who cooks amazing food, but corrects your manners while you eat. Maybe there's a hint of the R- at work there, but you'd have to assume projective thought on the part of a highly stimulated animal. I dunno about that.

Ok, lets say we have the dog, Big Joe. We tell Joe he can't leap up onto arriving guests. So next time guests arrive he greets them by ramming his nose into each crotch, very hard. [etc, etc . . .]

To mudpuppy, it's hard to know what to say. That's not my experience at all. I train a dog not to jump up with a variety of ways depending on the dog. For most I just turn away and say, "Ah-ah!" I have other more aversive techniques, none of which involves touching the dog with foot, knee, or anything else potentially dangerous. My goal is for the dog to say, "Well! That sucked! Maybe I'll not try that again."

Aussies and labs, oh my gracious. I've worked with both of those, giant Golden Retrievers, two Husky/GSD mixes (one an adolescent male pup), a Malamute/Aussie mix, an ACD/GSD mix, the famous Dane-a-saur, a boxer mix, a BC/Walker Hound and of course those sly, agile, Border Collies who teach most trainers the utter futility of knees-to-the-chest and body blocks. None of them said, "Hmm. Jumping not allowed. Let's try invading space in some other way!"

No, I think what's going on in their heads is not, "I'm jumping" but "I'm getting in your face." I believe dogs do understand basic concepts of space and so forth - and they don't invade my space by accident, or without a sense of potential conflict. We talk about how important interwhelping socialization is for bite inhibition - there's some other things that are instilled during that period also.

Now, once I've got feet on the floor, I don't just go on with my life. I'll assume a relaxed posture and say, "Thank you." I don't know why "Thank you" helps but most dogs I've worked with read something positive in that, but not reinforcing enough to seek the whole chain again (as I've found is problematic with "ignoring and training the incompatible behavior"). Some dogs will experiment a bit, no problem - I don't make a big deal of it. I encourage it. I keep stepping back and stay relaxed - "Let's have a quick little chat about this" my posture says, "Let's find a workable compromise - you want attention, I don't like stupid - we can surely find a happy medium." Again, it's the game of "red light/green light."

I have seven high-drive dogs loose in the house at any one time. According to your scenario, I should have no more house left. Or I should be doing nothing but running around all the time saying, "No, no, no, no, not that either." No, they know their place is with me and relax until there's something to do. That's not to say they do nothing. Cord's the Water Bowl King - he comes and stares at me if the water dish is empty. Ben brings me all the bowls lying around before breakfast time as a subtle reminder that mealtime is approaching. Maggie lets me know if the sheep are in the garden.

Again, just want to reiterate - it's not that I think a dog can't be trained in a variety of other ways very successfully (done it). It's that I keep seeing this stuff about the poor shut down dogs trained with P+ when I don't see it in my own experience in or outside the working evironment.

I'm not saying that! This was a response to the suggestion that the inhibited dog is a bad thing. I need a certain amount of inhibitedness - but it's not incompatible with thinking, either. In fact, the dog is actually thinking quite well in that state if he's been trained right (I'm not talking about a dog that's been abused or purposely shut down - I do know trainers like this and I leave them out of this discussion). The ideal working dog is calm, engaged in his environment, ready for anything, and thinking through his choices to fit a suitable action to the moment. He also is not afraid to experiment if experience or instinct is silent on the particular situation. The boundaries of grace.

 
And that's the aim of +R, like clicker training. But it has been suggested, not by you, that commanding an incompatible behavior does not allow the dog to think, that commanding a +R trained response does not allow the dog to think. That the way should be to correct the dog and they will just automagically know what we want them to do, with no particular guidance as to what that next move should be.
 
I agree with on our needs for our dogs. There are times when I need my dog to obey and that's the fact, Jack. And I've used -R. Shadow jumps and I turn away. I command "off" and he obeys and then gets attention. That is, -R, turn away. Command, reward obedience, +R. Shadow's job is to get along in our world. He is not, at this time, a working dog. I'm not working sheep, I don't hunt, we're not sniffing out drugs or bombs, or running off the geese from an airport. We're not even pulling a cart. But I think I could train him to do whatever job we needed with the clicker because the danged method works. Oh yeah, I went there. Oh yeah, I did.

Brookcove, what you are talking about is Self control, where the dog learns to control himself in exciting situations in order to Get what he wants-- in your case, to work sheep. Your working dogs don't appear to be "shut down" and I suspect most of your "corrections" are perceived by the dogs as just information, not corrections.
Many people mistake the Inhibited dog for a calm, self-controlled dog, because he's behaviorally inhibited. You hear reports of such confusion all the time-- the reactive dog in a raging hysterical fit that the owner "holds down until he stops reacting" is reported to be magically calm for the next thirty minutes, which thrills the owner. But the dog hasn't learned how to control himself from within, he's just stopped behaving on his own initiative.