The Limits of Clicker Training

ORIGINAL: Scout in Canada

I'm sure IMPO I know what does and does not work. But you all have the right to think what you wish.

In your opinion, why has Steve White been so successful in his training demos? It seems he has found a way to do bite work, scenting, and other behavior with a clicker.

 
So, and I don't know where the quote came from, what doesn't work?  Teaching scenting?  I don't even have a "PO" but I could make it work.  Reinforcing natural behaviors isn't a difficult action.  It just seems to me that the person with the "IMPO" is afraid to try something new. 
 
But just to honor the scenting game, I will start teaching my hounds (probably only Xerxes since he is the most eager to learn anything-and we've played scent games before) and we'll work on scent discrimination.  If I'm successful that will be more credit to his breed. If I fail-which I won't-I'll just keep trying. 
 
 
Now I'm gonna go read the AKC rules for participation in tracking.  See yas!

well, but lots of people reward their agility and flyball etc. dogs with "the bite", i.e. a tug game, and they still mange to train good "releases" without using harsh corrections.

Excuse me, but let's not let this thread deteriorate into bickering and baiting.
 
Awsomedog I see where you are coming from with the bite work (the most rewarding activity for the dog being the bite and therefore there is nothing effective with which to follow up a marker such as a click).  However, could you concede that that is not the only way to train a dog in this field? 
 
It seems rather silly to say categorically "This will not work" if it is true that people within that field have had success with it.  The way you select and train dogs - no it won't.  But the way someone else has selected and trained dogs, yes it has.  I remember you saying on another thread that you didn't like someone else's posts because they always said/implied "thieir way was the only way"  You may have fallen into that trap yourself there.
 
You are entitled to your opinion I am sure, but there is a difference between opinion (even a professional one) and fact - and if its a fact that someone else has had success with it then that supersedes opinion (even a PO) in my view.  However it would be nice to see links, books, material, video links etc of this kind of thing being taught with clicker training, to see for myself just how it works and how well it seems to work.  Mind you, I imagine that is not easy to get hold of or you'd have idiots all over the place trying it!
 
Finally, I disagree pretty strongly about the scentwork.  The police officer I spoke to who trained and used his GSD in scent detection paired the dogs favourite toy with the scent he wanted the dog to find so that in essence, the dog thinks that by looking for this distinctive scent he will find his toy  and in a way he is right because when he has signalled that he has "found" it, he always gets a game with the toy afterward.  (This particular officer always made sure to make it look as if the toy had come from that hiding place but I don't know how widespread that practise is or how much effect it has on the dog's work.)  I also know of dog handlers in the miltary who use a very similar method.  In this case the dog's reward is not the scent detection itself - its the tennis ball at the end of it.  Perhaps you are saying that this makes the method less reliable, and that to use something similar in bitework would make the dog less reliable, well for that you have your opinion and others have theirs..... and neither is going to change I think.

There is good reason that clicker work is NOT wide spread though out the military and police that do scent or bite work

 
Yes, I think most of us can think of many reasons for that.....[][] And it has NOTHING to do with it being the "best" OR "only" way to teach it.
 

I spent years in that field of work and understand how it works and *one* person can't change that.

No, you're right, one person can't. But decades of research and hundreds of people CAN. [] And that's the point! There is nothing worse than somebody with experience in a field who flat-out refuses to realize that as times change, we learn new things, and new ways of doing things (darn those outdated vets...haha).
 

  But you all have the right to think what you wish.  And I'm cool with that.

You're right, we do. As do you, and if you continue to believe that "your" way is the "only" way, we're cool with that too. But I'm sure the rest of us would rather get back to what WE know works, and that is, the clicker method! (After all, it IS the clicker board.......).
[]
 
Kim MacMillan

Finally, I disagree pretty strongly about the scentwork. The police officer I spoke to who trained and used his GSD in scent detection paired the dogs favourite toy with the scent he wanted the dog to find so that in essence, the dog thinks that by looking for this distinctive scent he will find his toy and in a way he is right because when he has signalled that he has "found" it, he always gets a game with the toy afterward.

 
It's funny, because that's how a LOTof scent trainers I know train scentwork. The dogs think of their scentwork as a game, and in the end they find their toy - THAT's the reward, THAT is what they work for. And, well, the drug detection dogs, the contraband detection dogs, the police tracking dogs - they're pretty reliable!
 
Kim MacMillan

ORIGINAL: Awsomedog

ORIGINAL: ron2

I have doubts about using a clicker in bite work and field trials where noise level could obscure the sound. But that doesn't mean you can't click on a controlled bite or click on solid recall. Of course, a lot of cattle dog work relies on a whistle, something they can hear in full run around a flock of bleating sheep.

You can be in a field with no noise, and I'm telling you from years of experience, *a click* will not make a dog release that bite. And if the dog does release, that's not a dog I'd have in police work. The officers life is more important than how someone feels about "clicker training". Now if your talking bite work for fun where a soft dog can be used, well click away.

Just to clarify, the click does not elicit behavior, the click ends behavior (and tells the dog that the behavior he was exhibiting in that nanosecond is the one he is being rewarded for). 
I think this article may be useful for those who are interested in bitework or police k-9 training with motivational methods:
www.tinyurl.com/yo6sb2

ORIGINAL: mudpuppy

well, but lots of people reward their agility and flyball etc. dogs with "the bite", i.e. a tug game, and they still mange to train good "releases" without using harsh corrections.

My Aussie goes from swinging around me attached to a toy, to "outing" it immediately.  She may not be a Mal, but she's a high drive little girl and I have never had to punish her to teach her to "out". 

Mmm, I am still inclined to think there might be psychological limitations. How can I teach my bunny to roll onto her back if she never even rolls onto her side? She lies on her belly pretty much at all times. Sometimes, she goes onto her side while getting herself settled, or if she's super super relaxed, but it doesn't happen often enough to be able to work with it and she wouldn't be lured into that position. What I'm finding with her is that training sessions have to be very short and sometimes she's just not in the mood. Rabbits think the most important creatures in the world are rabbits and everyone else should be automatically doing their best to meet all their needs before they even think of them. So after a few minutes of clicker training, Bonnie starts to suspect that I want something of her and she assumes she won't like it because I'm not a rabbit so what would I know. At the moment at least, there's not a great deal of shaping she'll put up with before she goes and lies in a corner and studiously ignores me. Maybe it will get better with persistance, though. 

I devel pretty highly into protection work and scent work. First, most military and police dogs, are not the genetic gift to god. They usually get what's left over. Then the training is pretty shabby. The dogs are not that impressive. I reguarly train with Border patrol and U.S. Coast Guard. And someimtes with SDPD.

From what I gather The reasons they do not use clicker or marker training vastly, is resources. They do not have time to train handlers in the in's and outs of operant conditioning. The other thing is resitance to change. They have been using the yank and crank system since the 40's. they have recieved "good enough for goverment work" results. They are constantly asking me how I get my dog to be so obedient and yet preserve her drive.

Now the one thing that was posted....that the scent is rewarding? That got me. First what is so rewarding about smelling for a bomb, marijuana, human foot steps, an accelorator, or a cadevor? nothing. When I teach a dog to scent for a human. I have to give them a reason to work. The scent at first is a neutral stimulus. the dog has no reason to track. the scent means nothing. I have to pair the scent with something...cut up hotdogs work good. This gives the scent a meaning, it creates a biological response associtated with the scent. The dog is not tracking becuase it loves the smell. but because it associates the smell with maybe finding some food.

There is forced tracking but I find this to be highly counterproducative if not unethical.

as for bomb dogs. The dog is sniffing for powders, nitrates. the powders mean nothing to the dog unless you pair it with something.

where does clicker training come into this. if my dog is tracking a human and comes along an article, the dog indicates by sitting. how does the dog know found something. that the tracking produced. by giving a click or a marker then rewarding. or finding the box with the bomb in it? same thing.

Here is how clicker or marker training ties into bite work. at first I just play with the dog. having the play tug and letting them win. when they get to be about 7 mos. I start teaching out. This is when the foundation of clicker and marker trianing will be introduced. the dog already will know what the clicker or marker is from obedience and/or tracking. when the dog is playing I will freeze. the dog sooner or later will let go. I mark or click (i say mark or click becuae I do not use a clicker, i use a marker word....I need both of my hands) immediatly this lets the dog know that outing was the correct response and can get a bite. Once the dog has the out down. I can start teaching the dog to bark and guard a suspect. which again when the dog preforms correctly, i mark and the dog gets it toy.

Back to scent work and detection. the scent should never really be what the dog is after or the object with the scent on it. Because this is stuff generally dangerous and or evidence. I want the dog to leave it alone. dogs usually want to have rewards not just sit by them. and if they do not get them. The behavior will become extinct.

These are few dogs that have worked they way to world champioships....how clicker trianing

Klemm Vom Rotten
A'tim
Zico van de berlex-hoeve
D'jango ot vitosha
Turcados
Elgos

I have never met a protection or scent trainer, That has any type of "great" success. that does realize the value of marker training.


I do see where if one needs to click to tell a dog it's doing good....they best get another dog for the work. From every bite trained dog I have ever seen. the dog does not need to be trained to bite. it needs to be trained to contain itself and preform other behaviors to get the bite.

And harsh corrections for teaching out? This will cause out problems. the worst information out there is, if a dog does not let go, when you get the toy put it away. what? The dog is already possesive.  Now your just giving a reason for the dog to become more possesive. I teach the dog the out starts the game, not stops it. then whern the game is done. I straddle them and pick thier front legs off the ground by cradling them in my arms. They all let go after awhile. This way, the out did not end the game.

Back to the OP's orginal question. The only thing you cannot teach a dog with a clicker is true aviodance behaviors.

This dog's name is RIK this was shot by a friend who went to belgium to pick up his new pup. Rik, lost his leg in a training accident. But was a world champion. This dog bites for real, and has been through a korung 3 (that is the highest pressure, most demanding protection test on the planet, thier is only 1 dog alive, who carries this title A'tim, Rik has sinced passed on) the korung 3...is more demanding of the dog than any real police work I have ever seen.
[linkhttp://ebackpack.guhsd.net/dct990/Sites/Videos/Chomsky/rik2.wmv]http://ebackpack.guhsd.net/dct990/Sites/Videos/Chomsky/rik2.wmv[/link]
 
You guessed it clicker trained!

Thank you VERY much for that post.  I have a feeling that was the info we were all striving after.  Thank you thank you thank you - very informative.  Brilliant post.[sm=clapping%20hands%20smiley.gif]
 

  Back to the OP's orginal question. The only thing you cannot teach a dog with a clicker is true aviodance behaviors.

 
This is probably the singular most relevant thing so far posted on this thread!!  Could you expand on that, or would anyone like to take that further.....?

Now if your talking bite work for fun where a soft dog can be used, well click away.

 
Such as field trials and hunting retrieval. I knew someone that provided that kind of training and he used remote collars, set very low, just strong enough for a tingle. Another member here, who has been here in a while due to computer troubles, used remote collar to field train her dogs. It was a surer thing than any sound you could come up with.

The clicker I got for $4.08 is adjustable and even on the softest setting is quite loud, and perfect for close quarter work. My only doubts about it being heard are on, in the case of a hunting dog, running a 100 yards out, pointing prey, then running 50 yards lateral and return, they won't hear the click so well but you don't use the click as the command. You use it upon the completion of the desire behavior, which is usually the obedience of a command from your voice or a whistle, or even a tingle from a collar. Point being, when the dog recalls all the way back to your heel position, then c & t. The patterns that you run can be trained in close quarters and expanded later. IMHO.
 

I devel pretty highly into protection work and scent work. First, most military and police dogs, are not the genetic gift to god. They usually get what's left over. Then the training is pretty shabby. The dogs are not that impressive. I reguarly train with Border patrol and U.S. Coast Guard. And someimtes with SDPD.

 
Wow thanks so much for posting all that. We've all been hoping to hear some detailed descriptions from someone in that field of training, and here you are! Your post is really interesting, and I wasn't surprised at all to hear that scenting in itself is not the reward.
 
It's sounds like you do some very interesting work!

gunny, I knew you were using positive training, but thanks for the additional information on the clicker trained dogs.  This is precisely what I was getting at.  Dogs do what works for them, and what works for them is usually a primary reinforcer that gets paired with the not so hot stuff they don't care about, but which we want them to care about.  Dogs are trained to alert to drugs, contraband vegetation, cancer cells, humans, and incendiaries.  Do they care about that stuff on their own?  Nope.  But, if we pair the discovery of those scents with something the dog wants, well then....we get a dog that is happy when one of those scents pops up, and will alert us to "collect"  his prize (toy, hot dog, whatever).   I taught "out" (although her word is "drop") by simply stopping the game until my pup released the object, then reinforcing the response by starting the game again - took her about 2 seconds to figure out how to play.  I never have any problem getting her to "out" even when she's all worked up and nuts - she knows what keeps the fun happening.  That's really all it is to the dog.  The reward is the game, the food, or something the dog wants.  It isn't usually the scent itself (possible exception, roast beef LOL, but who tracks that?). The excitement about finding the scent is not the reward, it's the fact that the "find" PREDICTS the reward. []

Thank you gunny for that very interesting post. I'll be the first to admit that I don't know anything at all about bite or protection work and I'm only beginning on the scenting, but my main concern with the scent work is, how do I communicate with Marlowe that he's on the right scent? He can smell everything, yet I want him to track only one particular scent. How is he to know which one unless I show him? And how can I show him unless I provide a clear reward structure and use some kind of system to mark each performance that earns a reward? Most information I've read on beginning tracking (AKC regs) recommend that the track-layer bait the articles with a food reward at first so the dog can make a clear connection between finding the articles and getting rewarded.

And it's not like his instict to hunt doesn't contain a reward as well. He scent tracks small animals because at the end of the track he may get the reward of completing the kill. Many coonhounds hunt in order to get the carcass at the end. Very few hunters actually eat the game--they shoot it out of the tree and give it to the hounds as their reward. So, in training Marlowe to track human beings, that is what I have to compete with, his previous training of being rewarded with a delictable racoon carcass. How can I do that without employing rewards of my own, and I think that the rewards can be much more precisely given if I also use a marker of some kind to indicate just what exactly he is getting the reward for. It makes perfect sense to me.