NILIF - we don't do that here, not the classic way. It's more like a child learning rules - you don't get your food if you are clowning around at the table, you don't go outside unless you are wearing a coat, you don't leave the bathroom until you've flushed and washed hands. Most of this stuff is shaped before the child is verbal, but you never formally
train a child to do them. They imitate you and find out what you want when they are at the age that they naturally want to please.
I start out with all their fun stuff happening in the crate, so by the time they are three months' old they will slam themselves into a crate so hard it hits the wall, if I've got a food dish or if we are preparing to go for a ride.
They learn respect simply by learning what "No" means. A lot of it is centered on their behavior when walking around the farm or training the recall.
Their formal training up to their first exposure to stock, has two parts. The first part is "Come always means come." This teaches both a command, and a correction of some kind. You start with calling the pup (name and recall), using running away games and happy encouraging voice and praise, to teach what you expect when you call.
Then you start enforcing. My mentor says God made their legs short to make this stage easy. Most Border collies have already learned a verbal correction because they like to use their teeth as babies. [
] If they ignore the verbal correction, mama walks them down and gives them a wee scruff, then walks away and gives them another chance to get it right.
Gradually you fade both the praise and the corrections, and simply walk off when your dog is committed to returning for you. Make a fuss when your dog catches up. Walk a little ways then say, "Go play!" or throw a ball or whatever. Repeat about a million times.
This transitions to the next part, "Where did she go?"
You wait until your pup is busy (but not too far from you), then walk off. No recall. If your dog doesn't respond at all or runs the other way, say "HEY! That'll do!" Um, I mean, whatever your recall is. Keep walking in the opposite direction.
Repeat until you can fade the command and the dog knows that if you correct, he's supposed to come in that circumstance. Then repeat until you can simply leave the area and your dog will follow.
ALWAYS reward compliance by releasing your dog to go play after a few steps, or a game of ball - change it up. But no praise and no treats - it's important that the dog come because you expect him to, not because his mind is filled with the prospect of snuggles and luscious goodies - we're past that at this point.
And once a dog is attuned to you on this level, he's really past NILIF - I don't care if these dogs walk in front of me (it's needed so my dog can "scout" for sheep on the hillside), and they would never think of busting out a door because I'm not there. They wouldn't snatch a treat from my hand, because they have learned to control their impulses. They don't jump on people because if the person is not me, they don't really care. They are calm because they know they will only get what they want that way.
We do it this way because it's important that the working dog learn to work through problems under pressure. This brings us to the second part of your question, about operant conditioning. Notice that although this is very P+ based, the dog is working through what's expected of him on his own.
I've got a dog that was P+ trained his whole life, but his problem solving skills are top notch. He has none of the "crossover" hesitation or problems I've heard other people describe. He eats up free shaping sessions and gives me behaviors so fast that I've stopped clicker training him, because I have to improve my reflexes before we can continue! In fact, I've not seen the "crossover problems" in any dog I've had here, from trainers I respect and emulate. This makes me think that possibly it's not corrections per se that cause a dog to stop thinking, but a lack of freedom while training - or abusive corrections, of course.
Notice, however, that this is highly specific to Border Collies (works in Aussies, shelties, and any of the more people-oriented breeds, too). And you really can get the same results with R+, but you don't get the working under pressure part that's so important for what we do.