Clicker Training Research

Mom2Melli, thank you for detailed explanation of your training practices.
The goal of my posts is not directed at you changing them but as a benefit for people who are new to clicker and will be looking for information.
There are many parts where I would see potential problem but few things have become clearer. I think this article might help to see the difference
http://www.clickertraining.com/node/111
When creating chain I expect my dog to be responsible for knowing parts of behavior as I am neither cheerleeder telling my dog to go on, nor a teacher approving my dog´s behavior. Dog training is and always will be a dialog I have with my dog. I build focus for the task well before I start chaining...
The cues dog gets while performing a chain are largely enviromental. For example in agility tunnel after jump becomes reinforcer. The next cue itself is becoming a reinforcer as well as next behavior reinforces previous. Adding click at that point would be meaningless to me.
It is a news to me that clicker is allowed to use at rally trials or any trials for that matter. Former APDT currently Cynosport allows using treat at very certain particular parts of exercises. I think I would be miffed if anyone come and start clicking even outside of the ring while we compete. Not because I can not train my dog against some distraction but I would rather find it rude:rolleyes2:
 
Apparently, in rally novice (at least) you can click as well as verbalize and use body cues. I looked it up after watching a lady on Youtube with a pom click at every stationary task. She got a perfect score, too, I think.

I am sure in agility it might be odd clicking for different obstacles. We don't do agility because I have a bad knee and it would kill me so I really don't know much about it or how to train for it. I would maybe like to do treibball and the course description mentions continuing clicker use for this activity. Not sure how though.

I also suppose my school is into cheerleading and giving approval. It made sense to me. The trainer is always telling people "I don't hear a click". Our first exercise in the human-only class is to shape a person's behavior with hot-cold multiple clicks to get them to do a task. Too few clicks and they quit trying. Clicking when they do a move wrong really confuses them. Clicks with each move towards the right task encourages them and gets them there quickly.

Some examples of why to do it the way I described: The shepherd I have now is an insane worrier and if he doesn't get a click after a task his brow wrinkles and he jumps up with his "mommy do you still love me" thing seeking approval. It's not asking for a treat. It's that he is literally terrified he has done it wrong and I am not happy with him. It is the same jump up he does if there is something terrifying like gunshots, a helicopter low over our house, or if a strange dog lunges at him. My particular dog runs to me for approval all the time. In the case of the shepherd I really could train him without a treat ever. My approval means THAT much to him. Of course, the treats are great rewards and are definitely used, but he really really is motivated by my approval. The teacher told me to go ahead and click with parts of sequence and click/treat at completion like everyone else would to keep him focused and give him feedback until the sequence is mastered and he himself knows he is right. It worked. He quit jumping up after every part of a sequence. This dog needed the extra feedback. As for this sheltie who has come along recently, well, she understood only Spanish 3 months ago and I found that she was having a harder time than she should in class. The teacher suggested to try clicking like I had done before. The extra clicks shaped her behavior very quickly. By the end of class the extra clicks were gone.

Maybe the extra clicks are not needed for the average dog, maybe they are helpful with certain dogs who are timid or have other issues and need extra feeback. Who knows. Maybe it is something that particular trainer suggests sometimes and totally made up herself. I like her and have used her over and over for my various dogs and at various levels. Maybe my newest two dogs happen to be in the clicker training manuals' and videos' little-read "clicker training troubleshooting" sections. I just know it certainly helped these two.

In any case, I just hope I cleared up the idea I just run around clicking all the time for no reason with no rewards, skipping over failed parts of a sequence, as well as how this phase fit in in the overall long-term training scheme. I also wanted to put the whole thing en bloc so that someone reading the thread might not miss parts and get confused as to how it fit together.
 
Last edited:
thanks for clarifying
:D

I had no doubts since reading all your other posts, I generally concur how you train pretty much everything, I was sure I may have just missing something here.
and lol
I had to laugh at the picture in my mind of a person indiscriminate clicking everything and not putting it together made me think of that skid with Wil Ferrell
It kills me everytime I watch it!
http://www.youtube.com/watch?v=FuKa07Hsmu4
 
Husband HATES the clicker (not that I am training the husband mind you, but when working with the dogs in the house). There is some war movie (forget name) where they use clickers to ID the troops. You click, they click = one of own. Some guy hears the click and responds and promptly gets killed. It was a guy loading his rifle and it made the same click and so when he clicked back it totally gave his position away to the enemy.

Moral of post: Don't clicker train your dog in the middle of a war zone. However, teaching the dog hand signals in a war zone is much safer.
 
Last edited:
Mom2Melli, which venue did you find allowed to use clicker? I have competed in Rally AKC and APDT which is Cynosport and never seen anyone ever using clicker in the ring.
I believe UKC will allow using some devices but handler has to be deaf or hard or hearing...
 
Thank you for coming back to the discussion Mom2Melli and for giving such a detailed explanation of how your method of training works. It was very insightful and I do agree with the statement that you made - the click does carry a meaning. If it had no meaning then it would be pointless making the sound from a learning perspective. From your explanation I think your click just has more meanings than mine. I'll do a comparison based on your examples:

At Point A - no difference in the use of clicker to train.


Point B - At this point we diverge. I would not use the clicker at this stage because the actions (heel, down, call etc) are learned and this would just be practice. I do not progress with the clicker beyond 'Point A'. For me, the job of the clicker is ONLY to shape(teach)/perfect a particular behaviour. So for instance teaching heelwork, first use the clicker to indicate the correct position when stationary. Once the position is established I raise the bar and move a step and click for dog moving with me and when this is solid then click after a couple of steps in the right position and holding heel position when I stop moving. The behaviour the dog has learned is therefore 'hold heel position.' (I don't do more than 4 steps before clicking because once the dog knows to maintain heel position on the move then the number of steps becomes immaterial and clicking after 10 steps would only be telling him that he was right on the 10th step, not the preceeding 9.)
Once these actions are learned and solid I could do them in a sequence and reward at the end. I would not need to click because the dog now knows what is expected but where you are using the clicker for encouragement/feedback I would be using 'Yes'. If one of the behaviours deteriorated THEN like you I would go back a step and do the 'Point A' steps again to reinforce the correct action.

Point C - This would be where I would use a 'jackpot'. So if Epic has gone round an agility course or Yogi done his obedience test they get a extra big play or handful of treats in the end. These are what I regard as my bonus payouts. The dog has worked super hard over an extended period so has earned something exceeding ordinary at the end. (I pretty much exclusively reserve this for shows and have found this motivates a bit more under stressful conditions. The show environment has come to mean they will get a really special reward.)

So in summary the clicker has a greater number of meanings for your dogs than for mine as you have used it to teach a behaviour (click=instant treat) and to subsequently maintain the behaviour (click=doing great, reward to follow shortly.) For my dogs it is used only to teach the behaviour. The click 'captures/pinpoints' the action the dog has done the instant it hears it.

The difference I feel here also boils down to timing. For your dogs, the click is referring to something they have done, for mine it refers to something they are doing. For example teaching Epic to walk backwards I have to (try) and click as he is stepping back, not once he has gone back and stopped because what I'm aiming to tell him is that the going back is correct, not going back and stopping. C&T Yogi as he finishes his heelwork would tell him the sit at the end (or even the release) was the important action. This only becomes important really if someone needs to perfect a behaviour eg sit fast, straight and at heel as opposed to sit at heel.

I 100% agree with you that feedback is necessary. I wouldn't want to go 'cold turkey' from using the clicker to no information, it seems rather unfair and would probably set training back with an anxious dog.

With the random rewarding you mention, this is something else we agree on except I would be making the click the random rather than just the reward. So the dog still knows every time he hears a click he gets the treat but the click doesn't happen on every step, it might be on the second or fourth. He'll work that bit harder in expectation it'll come soon for sure. Same principal just applied slightly differently.

This leads into the click/treat at the end vs reward at end. Again same principal, different application. Our dogs probably work just as hard to the end point because we've both set an expectation of a reward for a task sucessfully done. :biggrin2:

On a technical point, I start introducing the 'yes' feedback at the point the dog is starting to consistently and confidently perform the behaviour in response to a cue and I say it just prior to clicking so it develops positive associations. As I begin fading the clicker I will be using the feedback and still rewarding quite frequently but not after every 'yes' so keeping motivation and encouragement while building the concept of a delayed reward.

I would be interested to know whether you have used the clicker to 'fine tune' a behaviour e.g. faster down/tighter heel position and if so how did you do this?

Please be assured I'm not being critical. Far from it :wink2: -now I've read your posts I can see the merits of your method, particuarly in that it seems to place less dependance on perfect timing - something I struggle with! Most of the pet owners we train at our club would probably get on better with a method that doesn't requre hair trigger reactions that take practice so rewarding on completion (probably also using lure) would be preferable to shaping. I also imagine your method is quicker, but possibly less presice? Again that would suit our owners because they are not concerned with presicion so long as they get the behaviour. (Presicion is also a very big de-motivator in my experience.:no:) Plenty of food for thought anyway! :smile2:
 
I believe it was AKC novice rally that the clicker was in. In reading the rules it seemed to indicate that was OK along with other verbal or physical cues. I personally wouldn't be clicking in competition either, only in trainng and only at the step going from learning the tasks in isolation to beginning chain the tasks. Once the dog was comfortable with running through courses confidently I would transition with some good/yes verbally and then eventually nothing except the jackpot at end reward kind ofthing

"I would be interested to know whether you have used the clicker to 'fine tune' a behaviour e.g. faster down/tighter heel position and if so how did you do this?"

Not nec. tighter heel because the treats in the hand cover that pretty well and clicking and heeling is like juggling an octopus -- but faster down and straighter/closer sits and such. Yes. Don't click for imprecise postion. Click for desired position. Our trainer told us to simply not click/treat for anything we didn't like the looks of. Do over or fix as best seems to fit the task. As the dog gets closer/straighter or faster withhold for more minor of defect to tighten after correcting on these tasks the dog sort of seems to know to get the click to keep trying. They scoot closer or try something else. They are thinking about what you might be asking during the neutal no response wait it out time and when they adjust themselves for the click it makes even more of an impact on them. Just keep asking it to look better and better before the click and they will start to become thinkers and triers. At least that is what the trainer was pointing out and we saw with a BC headed for full on obedience (rather than rally).
 
Last edited:
You must log in or register to reply here.
Back
Top