The Training
Positive Reinforcement training is totally reliant on the stimulus, response feedback loop that animals use. When training a dog to sit, we have a reinforcer (a stimulus that causes a behavior to be repeated) (With our 5 year old doberman,Wolfie, a mix of his own dogfood, some store bought treats, and leftovers (all in small amounts)) a target behavior, and a plan for eliciting that behavior. (For our first experiment, the behavior I wanted to elicit was to return a dog toy, his 'teddy', to my lap, without ever having to put his nose, paws or even say a word to him about our target.) To work correctly the plan must include short (very short) jumps from one portion of the task to another.
For Wolfgang to be able to find, place his nose on, pickup, hold in his mouth, carry the teddy to where I am, and drop it gently on my lap, is a non-natural sequence of events for Wolfie. Until now, he did not 'have' to share his teddy, but to make the experiment worth something, we needed something with a bit more bite, this was it. My plan on feedback was to start using a clicker (a sound cue) and food as a reinforcer: the first stage consisted of just clicking and treating (1 SMALL treat) him when he approached the target (teddy). Even if he didn't go within 4 feet of it, if he walked in the direction of the target, I clicked and treated. After about 10 minutes he would consistently walk from me (the giver of treats) to 5ft away (the teddy right under foot).
AT NO TIME in this stage did I yell, scream, touch him, pull on him, or touch the teddy. Nothing but click and treat. That's it. If he didn't go near the teddy, I did nothing. I didn't look at him, I didn't speak to him; I only responded with a click and a treat IF he walked towards the teddy, If not... nothing happened. (Meaning NO PUNISHMENT for mistakes.)
Now that I was getting 100% response from the 'nearness' behavior (near that area), it was time to move on. (For most people and dogs, 5-10 minutes is THE most training you want to do in the beginning, if you or the dog are getting frustrated or bored, take a break, come back in 30 to 60 minutes, and do 5 to 10 more minutes.) Our house has very little (except people and a few toys) to keep Wolfgang's interest, so I continued on with the session. My next 'behavior' mark was to get his nose near the teddy. He didn't have to touch it yet, just look at it (this was the slowest part of this trick for Wolfie.)
At first he walked towards the toy (as had worked before) and I DID NOT click and treat him. I was looking for a walk AND a nose towards the toy. Several more times he did the 'required' behavior, but no treat. 3 minutes into the exercise, he seemed to react to my nonchalance (nonverbally) as if he were 'doing' something wrong. He was confused. It had worked before.
Instead of assuming he couldn't learn what I wanted, or that he was being bad, I assumed I was the problem, and I was correct. Because I was dealing with a large working dog breed, as he grew up I had physically intimidated him: "I'm bigger than you, and don't you ever forget it." (My fear response for my family, "I will never keep a dog who has bitten me or my family out of aggression", so instead of being afraid of my dog, my dog is terrified of me.)
Thinking that it was this intimidation between us, and NOT HIM BEING BAD, I gave him minute alone, I left the room, and when he came to greet me when I returned, I gave him verbal and physical reinforcement ("Good Boy, Good Dog" and pet under the chin {he loves that}.) A moment later he went at it again, but this time, when he got the nose closer to the teddy, in addition to ONLY the treats, I gave him a quick (but warm) 'Good Boy' or a 'That's it' and even a 'That's what I'm looking for'. Almost immediately, his 'hit' ratio shot thought the roof. He had been intimidated by the lack of reinforcement for a behavior that had worked before. Now he was getting a 100% response from me saying that not only was this related but different behavior good, but it was even better than the first one.
Now we were getting somewhere. At ~97% success rate for 'nose over teddy', I felt we were ready to go on to the next level, touching. It is true that touching teddy with his mouth is a learned response for Wolfie, but only in play situations, it had never been in a work or a training situation, so he was dealing with conflicting conditions. Now I was clicking and treating him for any touch on the teddy. If he knocked with his foot, treat. If he moved it with his nose looking for a treat, jackpot (3 to 5 treats.) After 5 more minutes and about 3 jackpots, we were getting about 80% perfect response on touching the teddy with his nose. At this point he was physically moving the teddy around the room and it was in a corner so I stopped, moved the teddy to the opposite side of the room, and started again.
After a few only regular (single treats) he picked teddy up. Jackpot! Literally the moment he finished the treats, he walked back to the teddy, picked it up in his mouth, and looked at me with it in his mouth. Jackpot! He did it again. Jackpot! This time, instead of changing the treat ratio, I changed the environment, I carried the teddy into the living room, and carried the treats too.