One of the things I've learned over the past couple years is that corrections are essential to the learning process. Once the dog learns what's expected, they need to be held to that standard.
Example. Brooks knows that sit means sit. His butt hits the ground and he doesn't move until I release him or give him another command. If he moves, I give him a correction, be it an AH-AH, a quick pop on the leash, or a nick from the e-collar. Sit means Sit. Here means Here. Down means Down, etc. There are rewards for doing something correctly, a treat, physical or verbal praise, a happy release, and consequences for failure to comply with a known command.
When I'm teaching a new behavior, there are no corrections. Only positive reinforcement when he performs the behavior correctly. An enthusiastic "YES" or a click from a clicker follow IMMEDIATELY by an edible treat. Once he's performing the behavior correctly, I add the verbal command, "YES", click, treat. Once he is performing the behavior by verbal command (or non-verbal signal), which tells me he understands what he is being told to do, I will correct him verbally or physically as appropriate.
An example. We have been dong a lot of casting drill for field work. Through weeks of training Brooks knows a single blast of the whistle means "SIT", regardless of where he is relative to my position, or what he may be doing at any given moment. If I give him a whistle and he doesn't immediately sit, I nick him with the e-collar, and his butt will immediately hit the ground. Mind you I use level 2 on the collar, which I've used my myself to see how it feels, and it's almost imperceptible. But to Brooks, who's been conditioned that this is a correction, it's enough to make him comply with the command. It's a small correction, but it works.
I am not advocating any training methods or tools, I am merely trying to show how, why, and when corrections can be beneficial to the training process.