Trick Challenge #1-March to the Music

[SheltieChe]

"If it aint broke, don't fix it." If Sally is getting it and having fun it is a total win-win! No need to change anything!

totally agree, if you having fun it is all that counts. Beauty of training silly tricks is oif you screw it up it is only silly trick. And eventually one learns enouhg to know what works for them and their dog so mistakes are less for " important" behaviors.

 

[SheltieChe]

I've never heard that before. I always thought that a click was the bridge between the behavior and a reward.

I feel like that would make it almost impossible to capture a behavior, for instance, with a dog like Lexi. Clicker in one hand, one hand out to get her to paw at it. She paws at it, and I click but as I'm clicking she is offering tons of other things. The point of the clicker is to mark the actual desired behavior, I thought. Lexi would never learn anything if I had to treat at the same time as the click and not allow anything in between.

That is where placement of reinforcement and your mechanics come is. Mine still atrocious. Famous chicken camp at Bailey's first day all they do is teach you to hold your clicker and bowl of grain in your hands right.
Another thing is placement of reinforcement itself. For example I am teaching boys to put their head down. So once I click for extension of the neck forward and down I will put the food just slightly in front of the head on the floor so to be able to reach it he needs to repeat same movement with a neck. Just like you did with Lexi doing "around"- you put your food in front of her face but just slightly out of reach so she needs to make another small forward move and thus moves around the circle.

 

[Justicemom]

Clicker training as such has evolved in the past 20 years. Traditional clicker training thought out to be very linear. I.E. you want to teach dog to pick up a dumbell so you take dog to the obedience ring and put that dumbell in the middle of the room and click for successful approximations , i.e. dog looked at the area, dog turned head, dog blinked , ear turned in right direction so you would click like 30 times a minute for any smallest thing you can reward that will lead you to right behavior until you eventually have that dog sniffing and licking and touching and eventually taking that dumbell into mouth. It is more like complicated game of "hot and cold" that kids play.
What I do is different that entire behavior split into many parts which shaped separately and often "speed up". For example I want my dog to jump onto the table. I have a behavior of dog going to the dog bed. So I will put dog bed onto the table and shape "going on the table" this way then remove dog bed and ask for same behavior and because dog is likely to do latest behavior he was reinforced for I will get same jump on the table without dog bed on it. So insted of linear shaping step by step you start thinking outside of box looking for solutions by manipulating environment as well as reward system. That requires more thoughtful dog.
I do not think linear shaping is Corbinam problem though, although it would be great to see visual, you are so good with camera! It more sounds like overarousal in anticipation of reward to me.
What I mean by slowing behavior down is manipulation of environment- take dog into bathroom with nothing else happening there, put dog on the leash to prevent from running off sniffing, shape while dog in crate i.e. remove distractions as much as you can. You might need to move that bowl with food further or put it on higher table so dog has calmer/ more thoughtful offers. Sometimes you just wait a little and once this frantic stuff slowed down reward for closest criteria you see and start there. Sometimes you get couple behaviors but you only want one- lets say I want nose touch but not paw touch so I put dog into box because response cost of lifting paw out of box is much higher than lifting it on flat surface so dog is likely to offer nose touch first.

So instead a starting from scratch, your doing something or using something the dog knows to jump start the behavior. That's what you mean but non linear clicker? To me that is awlfully close to starting a behavior with a lure. The lure starts the behavior and then is removed quickly when the dog starts offering the behavior the behavior close to what you want. It might not be the end result you are looking for but it is a place to start. Difference the lure has no initial connection to the behavior that comes from understanding were using a behavior that the dog already knows does.

Still thinking on slowing behavior- using environment to limit choices the dog wants to offer. or decreasing arousal is that about right? Terms are confusing sometimes and just trying to wrap me head around them.

 

[Justicemom]

Ok I have decided that training tricks with 4 dogs is going to be too time consuming with all the other training I am also doing. So I am going to train Birch, since he knows almost nothing compared to the other dogs and he is my clicker guy.

Started last night, didn't get very far only a few paw lifts when I touched his paw with my fingers and that was it. I got lots of nose touches though. I am not saying anything or naming any behavior yet.

Day 2- started with a few paw lifts when paw was touched, then just presented my hand and waited for him to lift his paw towards my hand. Offered that within 10 secs, first try. Seems he is right handed. as he offered to lift that paw 4x more than the left. Then he got kind of bangy and started flinging his feet around pawing the air but with no purpose. So I decided I would reward him for only hitting my palm. 3-4 clicks and he was calming down and focusing on my palm. 3-4 more clicks and he was placing his paw on my palm. Did about 10 more clicks. Offering paws when hand presented but not always paw on correct side. Quit for the night. Tomorrow will try to get him to use both paws and hopefully offering correct paw when I change hands.

 

[SheltieChe]

So instead a starting from scratch, your doing something or using something the dog knows to jump start the behavior. That's what you mean but non linear clicker? To me that is awlfully close to starting a behavior with a lure. The lure starts the behavior and then is removed quickly when the dog starts offering the behavior the behavior close to what you want. It might not be the end result you are looking for but it is a place to start. Difference the lure has no initial connection to the behavior that comes from understanding were using a behavior that the dog already knows does.

Still thinking on slowing behavior- using environment to limit choices the dog wants to offer. or decreasing arousal is that about right? Terms are http://sheltieforums.com/newreply.php?do=newreply&p=223904confusing sometimes and just trying to wrap me head around them.

yep, albeit I would rather say bringing arousal down to workable level
And yes, I would manipulate my environment and do what works. Since I did not teach first behavior/ cue as luring, it is all offered responses but I really only enough into science to know what works and not looking into positive R, negative P or antecedent
For example someone offered to teach as a trick putting head under the pillow. Traditional linear shaping would be put that pillow and click for any interaction with it- looking, walking, sniffing, chewing , eventually building response of dog touching it with nose, then touching with nose on the side, then touching with nose under etc.
I would get a smelly yummy treat, put it under the pillow and say cue" find it" / my dogs know cue/, c/t for head under the pillow, repeat twice, then pretend to put treat, c/t for putting nose under, then wait for offer doing nothing. Since my dogs are operant and as science goes are likely to repeat last behavior that was heavily reinforced I would expect easy offer dog putting head under the pillow. Click, reward, big party! Balance break, run/tug, repeat in different room with different pillows, with different distance running etc- I got a solid behavior under 5 minutes of total training time.

 

[romeosangiovese]

Romeo doing better than Juliet. He's getting the paw lifts and not touching my feet for the most part.

Juliet is more all-over-the-place.. Lol! I think with our recent "walk on my feet trick" she ended up wanting to put BOTH front paws on my one foot! But she's getting there.

I had a blister on my foot before I started. Now I don't. It's just a painful wound! The kids keep using their nails!

 

[Aearoniel]

Though I said I'd only be doing this with the girls Pip decided he wanted in and during play offered paw all by himself, used my marker word and high value reward. Did a few repeats holding my hand out infront of me and marking/rewarding paw on hand. Then switched hands (and OH delivered clicker!) no reward for left paw on my left hand... got the brain whirrling around a bit, several more left on left then bingo slight lift of the right paw "click" then next time full on right paw to left hand... brilliant few more repeats and end the session.

Callie and Indi already know "wave" with either paw so am moving wave seated to wave standing... at the minute we're at about 10% correct standing waves for Indi and Callie is at about 25% so lots more practice. Both girls think down is a better position for this trick but we shall see!

 

[corbinam]

I'm seeing similar results after about 3 sessions.

Bentley is excellent at lifting his right paw, and last night lifted his left paw about 75% of the time when I put out my right hand (Yay!) He is also lifting his paw without smacking my hand, which I'm taking as a good sign. The way he is doing it, I may not need to transfer it to my feet, then take the feet out of it.

Lexi is getting slightly less frantic (Shelli, again it sounds like Lexi and Diva are long lost relatives) and knows it has something to do with her paws. Sometimes she forgets to think and throws both her feet at me. She hasn't quite figured out the correct paw to correct hand thing, but I'm backing up and making sure she understands it's just 1 paw, not both (she wants to stand on my palm, I think. Also, she tried to do a flyball turn off my chest--fun!)

I had a blister on my foot before I started. Now I don't. It's just a painful wound! The kids keep using their nails!

I have to laugh because I have scratches up and down my arms from frantic pawing! (Yes, they need their toenails trimmed but I don't dare do it now because it will make it worse!!)

 

[corbinam]

Here's a short video of our earlier session:

http://www.youtube.com/watch?v=dEUoB5px7ic

 

[labgirl]

You have a 5 second window of time after clicking to give a treat. This is more than enough time to walk to a treat pot, get a treat and bring it back.

Click should equal, "that moment of behaviour was perfect and you will get a treat for it." The dog should not use the treat as the marker, but the click.

On occasion I have muddled my treats, or not been able to pull them out of the bag, 5 seconds have elapsed before I have given it without ruining the behaviour. The important part is 'when' you click.

I need some help with specifics. So, Koji can shake, but he doesn't have separate cues for left/right paw shakes. Should I first teach that cue? He usually will use the correct paw accordingly to which hand I have up. Usually, but not always.

So, let's say I don't need a cue, but put my hands out, as I move away. It seems like if "shake" is followed by a literal shake, wouldn't that confuse him now, because I'm not shaking? Or do you think that's not part of the actual command, and the command is really for him to raise his paw?

Lastly, if I use the shake command, wouldn't that reinforce shake to eventually become march? at what point do I need to stop using shake?

Depends how you want to cue him, with Cadbury, because he knew 'paw' already I am using verbal cues, with Merlin I am just pointing out my feet and getting him to touch them with the correct paw (left = his right, right = his left). I pointed at my foot first, then just offered the foot. When I first changed foot he still tried to use his right paw, but I waited until he lifted his left and then clicked. I use no words at all, but when he is doing the full behaviour I may include 'march' as a cue, as I would use 'heel'.

If Koji actually 'shakes' on the command I would not use it while teaching 'march' as you will lose that behaviour if you keep using it for a different process. Try a new cue.

I think everyone is doing so well! Keep it up guys, I love hearing how you are getting on!