Do you play piano? I bet you learned to finger the notes, and read the music, and even the basic theory fairly quickly. At that point would you say you have "learned to play the piano"? Were you ready to jump into Bach or Beethoven? Ready to perform at Carnegie Hall? Probably not.
"Learning," particularly when learning a physical skill, is much more than acquiring the basic understanding of how to do the behavior. It also includes developing physical proficiency, fluency, figuring out the cues and contengencies, and developing the ability to perform under stressful or distracting conditions. This is true when humans learn a physical skill, and it's true of our dogs. In order to be certain I systematically cover all facets of "complete" learning, I break my training into four stages.
In stage one, training is comprised of formal sessions with counted reps. This is the stage where I get exactly the behavior I want and add the cue. If the dog makes a "mistake" here, I ignore it and may make what I want more obvious or easier. I aim for a very high rate of reinforcement for doing what I want.
In stage two, I'm still in formal training sessions. This is where I add elements like distance, duration, and distractions. I might begin taking the behavior on the road. I am still ignoring mistakes and striving for a very high rate of reinforcement. I want the dog to experiment and make some mistakes, because that's how he figures out what I want. "Is it right if I do this? What about this? How about this?" Corrections are absolutely counter to what I want to do during either of these first two stages, because they suppress behavior. I want my dog to experiment, and he won't if there's an unpleasant consequence for doing so.
Stage three is the first time I cue the behavior OUTSIDE of formal sessions. I continue having formal sessions, but I begin to phase the behavior into "real life." The key here is that I cue the behavior ONLY when MY DOG wants something. Maybe he wants me to open the door to go outside. Maybe he wants his dinner bowl to be put down. Maybe he wants me to throw his bumper. If he does what I ask, I do what he wants. If not, I don't. This is how to use environmental rewards. I create a clear contingency between doing what I ask and getting what HE wants. This is the stage where my animals learn that not responding to what I ask has consequences, namely that they don't get what they want.
When stage three is very, very solid, I'll begin to use the cue in situations where my doesn't necessarily have anything obvious riding on his response. This is stage four. However, by this stage we have thousands of well-reinforced reps behind us, and fluency is aiding the reliability. I still try to reinforce, but sometimes at this stage, especially as life goes on, all they get is thanks and a pet.
I occasionally use negative punishment (removing something they want) at this stage if I'm not getting a response. I don't do that very frequently though. By stage four I simply shouldn't have to. If I'm getting non-responses, I rushed training.
