The Limits of Clicker Training

A-dog you said on the other thread:

ORIGINAL: Awsomedog

ORIGINAL: Chuffy

Why is this procedure not workable for bite or scent work?

That! I have already stated over and over. Now I answered your question on shaping, you or anyone can answer mine. Explain to me in detail how a clicker would be used to train *real bite work* or *teaching scent work*. Start with Bite, then scent. I'm looking forward to your answers?

I thought it made more sense to continue that here as the 2 threads are spilling over confusingly.  So here, we discuss where marker training might not work and why and in the Resistance!!! thread people discuss why they are hesitant to use marker training. 

Just to clarify, you said that if a dog released on a click it wouldn't be trained in bite work further - you need to get past the equipment used!  It is the procedure I am interested in, why the procedure wouldn't work, not why a click wouldn't work.  I proposed the use of a whistle replacing the click with a dog trained that way from scratch and why this would not be effective has not yet been explained.  Scout has also pointed out that the marker can be used a diff way (ie. the animal doesn't release when he hears it but waits for his release command and then comes back for his reward) and again you haven't explained why this wouldn't work.  I also can't find anywhere in your posts where you ecplain why it can't be used for scent work - perhaps I missed something.

Would also like to re-iterate that the marker is not used in real life where there is genuine danger, it us used in the training stages only.

I want to stress that I don't have a set idea of how it might be used (I'll work on it and get back to you with some ideas for you to tell me are completely useless!!) and as I am unlikely to work in either field I doubt I will ever have the experience necessasry to be assured I'm not just speculating, theorizing and at times talking out of my behind.  However mrv has put points forward re scent work.

Finally, I don't know about anyone else but I'm not trying to "catch you out" here - I'm genuinely interested to discover areas where this kind of training is not/might not be workable.

ORIGINAL: Awsomedog

ORIGINAL: Scout in Canada

ORIGINAL: Awsomedog
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.

It doesn't seem like you get clicker training. The 'click' is a marker that indicates to the dog - what you just did was good. It is NOT a command or "release" mechanism meant to make the dog release their bite. Just like I wouldn't click to tell the dog to lay down, you wouldn't click to tell the dog to release their bite. The click is a construction tool, used to mark the right behavior.

Oh good grief man, can't you remember what's said from one thread to another. I've aleady explained and gave ***in detail*** what it is, only I didn't use a clicker.

 
I know you gave an example of 'shaping', and it was fine. But you often bring up the limitations of clicker training in bitework because "a click will not make any decent K-9 dog release their bite". And what I don't get, and is quite relevant to this particular thread, is how you think the click is supposed to command "release bite".

ORIGINAL: Chuffy


Not strictly true (from what I understand - and I'd like to emphasise that bit!!)

A marker can mean, as I said above, "That's exactly what I wanted!  Job done, now time for the reward"  which is what A-dog is seeing it as.  OR it can mean "Yes thats exactly it, thats what I want, keep going, reward is on the way".  Which is what you are seeing it as.

Chuffy, thanks for having a open mind and actually reading what I posted.

 

I think it's important to point out that I don't think either of you are wrong.

Well at least I'm not. lol []

ORIGINAL: Chuffy

A-dog you said on the other thread:

ORIGINAL: Awsomedog

ORIGINAL: Chuffy

Why is this procedure not workable for bite or scent work?

That! I have already stated over and over. Now I answered your question on shaping, you or anyone can answer mine. Explain to me in detail how a clicker would be used to train *real bite work* or *teaching scent work*. Start with Bite, then scent. I'm looking forward to your answers?

I thought it made more sense to continue that here as the 2 threads are spilling over confusingly.  So here, we discuss where marker training might not work and why and in the Resistance!!! thread people discuss why they are hesitant to use marker training. 

Just to clarify, you said that if a dog released on a click it wouldn't be trained in bite work further - you need to get past the equipment used!  It is the procedure I am interested in, why the procedure wouldn't work, not why a click wouldn't work.  I proposed the use of a whistle replacing the click with a dog trained that way from scratch and why this would not be effective has not yet been explained.  Scout has also pointed out that the marker can be used a diff way (ie. the animal doesn't release when he hears it but waits for his release command and then comes back for his reward) and again you haven't explained why this wouldn't work.  I also can't find anywhere in your posts where you ecplain why it can't be used for scent work - perhaps I missed something.

Would also like to re-iterate that the marker is not used in real life where there is genuine danger, it us used in the training stages only.

I want to stress that I don't have a set idea of how it might be used (I'll work on it and get back to you with some ideas for you to tell me are completely useless!!) and as I am unlikely to work in either field I doubt I will ever have the experience necessasry to be assured I'm not just speculating, theorizing and at times talking out of my behind.  However mrv has put points forward re scent work.

Finally, I don't know about anyone else but I'm not trying to "catch you out" here - I'm genuinely interested to discover areas where this kind of training is not/might not be workable.

Ok, but I'm out of time so I'll come back to this later.

Would also like to re-iterate that the marker is not used in real life where there is genuine danger, it us used in the training stages only.

That is a very key point that is often misunderstood about clicker training!

I want to stress that I don't have a set idea of how it might be used (I'll work on it and get back to you with some ideas for you to tell me are completely useless!!) and as I am unlikely to work in either field I doubt I will ever have the experience necessasry to be assured I'm not just speculating, theorizing and at times talking out of my behind. However mrv has put points forward re scent work.

I believe clicker training would be useful in bitework just like it is with any other type of training - it marks the precise behavior that you want and communicates that to the dog. Whatever bite task you are teaching, you need to be able to tell the dog "yes that's what I want". erhaps in small increments leading to the final full behavior. Either way, the clicker accurately captures the behavior you want so it can be repeated and assigned a verbal cue or visual signal.

I cannot see the application of clicker training having any limitations. 
 
To teach scent discrimination one could use clicker training to teach the dog which scent or scents get the marker.  It seems pretty basic to me.  Have 2, then 3 then 4 or more of those items with different scents and one with the target scent.  It seems to me to be a pretty basic training exercise.  If an untrained pekingnese can walk into a pile of 30 tennis balls and pick out the one I tossed, then with a little training any other scent discrimination or detection shouldn't be that difficult.
 
I will defer with reference to bite training, since I know absolutely nothing about it, even though I find it very interesting.

Scout - I imagine that a dog that might be written off as "too soft" by A-dog may benefit from someone working with a clicker to "build drive", but this is only speculation on my part and I would like to learn more about it.  As you say, it just communicates to the dog what you want in a very precise manner.  I don't want to teach it to my dogs!!!!  I'm just nosy and I like to learn.
 
A-dog I would also be curious to learn how you train the bite work.... 1.  do you use shaping as in your example or something else?  2.  Do you think its possible to shape this kind of behaviour or is that where we are diagreeing?  3.  Do you think your voice would be as ineffective as a click in shpaing this bhvr and that is why you say it won't work?  In which case, I'll bring up the whistle again.... I think, like me you have an allergy to gadgets and like to keep training simple and not fix or alter what is already adequate, but if (3) is true then perhaps it's time to open up to the idea that a gadget as a marker can sometimes be more effective than your voice?  You said yourself elsewhere that the whistle is used to break the dogs concentration where verbal release commands failed.  ???

I also have no direct experience with bite work, but I don't see why you couldn't use a clicker. Basically "bite work" is an elaborate tug game, right? the dog is taught to seize and hold the decoy and release on command. Well, I use a clicker to teach dogs to play tug-- to seize and hold the toy, and release on command. Once the dog understands the "release" command, you don't need your clicker anymore. So I don't see why you couldn't train your future bite dog to release the jute sleeve instantly on command with a clicker, and then as per normal training protocol just gradually proof the behavior until the dog is releasing instantly even when fighting a real person.  If the dog isn't releasing under these circumstances I'd say you went too fast and skipped some proofing steps.
 

Right with you there MP.... would also like to hear from anyone experienced in the field whether it really is possible to "build drive" with a clicker or whether this is a limit to its potential.  I imagine its possible, although I don't know how I would go about it..... That will need some more thought I think.

OK, aside from things that are impossible from a physiological point of view, what do you think are the things that can't be taught with clicker training?

 
The only thing that "can't" be taught with a clicker, is a behaviour that the particular species you are training is not physically capable of performing, because of physiology.
 
Bite work, herding, rescue, detection, tracking, protection work - it's all been done already. And Steve White DOES use a clicker.....it has been said as much by him before in different interviews in different places. That's where the "positive reinforcement" part comes from. []

Yep, Steve White does use a clicker, and I have him on tape saying that he does. []
He also admits that it has been particularly hard for him to break down the macho kind of resistance to using them. 
The Massachusetts State Police take their dogs to the Topsfield Fair for a demo each year, and part of the commentary to the crowd includes a description of the training, which they indicate is based on toy training - to the dog, the bitesuit is just a huge tug toy, and dogs are often rewarded by a game of tug with the handler. 
The type of stability and drive necessary to do bite work does not come with every breed - you'll probably never see Cavalier King Charles Spaniels doing bite work.  But, the drivey dog can be trained to grab an object and tug until called off.  In the case of my drivey herder, it's a rope tug, which she will "tug" and "out" on command.  In the case of police K-9's, it's a bite suit, representing a human.  The difference is that the police K-9 might also have to be proofed to "hold" when the "tug toy" fights back.

I heard a story on the radio once about how dogs are trained for bomb detection work by the government. There was no mention of a marker, but it is definately done with +R.

You take one adolescent dog who has been specifically handled from a very young pup to view a game of tug on a white towel as the ultimate in fun. There are kennels out there who actually do this--breed dogs and imprint them from a young age on this game. Not every individual dog is going to show the necessary drive for it, but the ones who do are sold to the government for beaucoup bucks. Prospective dogs are brought in by the breeder and given a number of tests to see how, basically, obsessive they are about the towel. The ones who are completely focused and driven by that towel pass the test and are admitted into the program.

Anyway, that game of tug on the towel with the handler is the reinforcer. The game itself has nothing to do with the work they're doing. They aren't bite dogs, they're scent detection dogs. It's just that the tug on the towel is a reinforcer that is easily brought out in any situation, and for that reason they specifically select dogs for whom that game is a massive reinforcer. The scent training then proceeds as per normal, with the tug game as the reward for successful completion of the tasks.

ORIGINAL: houndlove

I heard a story on the radio once about how dogs are trained for bomb detection work by the government. There was no mention of a marker, but it is definately done with +R.

You take one adolescent dog who has been specifically handled from a very young pup to view a game of tug on a white towel as the ultimate in fun. There are kennels out there who actually do this--breed dogs and imprint them from a young age on this game. Not every individual dog is going to show the necessary drive for it, but the ones who do are sold to the government for beaucoup bucks. Prospective dogs are brought in by the breeder and given a number of tests to see how, basically, obsessive they are about the towel. The ones who are completely focused and driven by that towel pass the test and are admitted into the program.

Anyway, that game of tug on the towel with the handler is the reinforcer. The game itself has nothing to do with the work they're doing. They aren't bite dogs, they're scent detection dogs. It's just that the tug on the towel is a reinforcer that is easily brought out in any situation, and for that reason they specifically select dogs for whom that game is a massive reinforcer. The scent training then proceeds as per normal, with the tug game as the reward for successful completion of the tasks.

Part of the demo at this year's fair was a guy with a Belgian Malinois hanging off the end of the tug as he swung it around him.  BF and I looked at that and both said "I bet the little Anasazi bean could do that", so we went home and tried it with Sequoyah and her favorite soft frisbee - it worked - she hung on.  Drivey little beanhead speckled monstah.

I 've only a moment here so I'll make this a simple and as quick as I can. There is good reason that clicker work is NOT wide spread though out the military and police that do scent or bite work, and it has nothing to do with macho. One person can't be right and everyone wrong. The reward is the bite. In teaching scent work, the object with the scent is the reward. I only need to teach the dog the odor I want them to find, the dog already knows how to hunt (so I use that), then I teach pattern. In order to explain every little detail on how it all works I'd have to right a book. Sorry but I don't have that kind of free time. I spent years in that field of work and understand how it works and *one* person can't change that. I've seen many techniques used, and I'm sure IMPO I know what does and does not work. But you all have the right to think what you wish.  And I'm cool with that.

Later 

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.