I think the problem with corrections (positive punishment is the behavioral psych term, applying a punishing stimulus) is you have to ensure the dog pairs the punishment with the thing you want them to stop doing. That is actually really difficult to do.

For instance imagine this chain of events:

Your dog is wearing a shock collar

Dog sees another dog

Dog feels anxious and barks

Dog receives a shock and stops barking

Success! Right? Your dog paired the bark with the shock! Maybe, maybe not. Maybe it paired seeing the other dog with the shock. You repeat this dozens of times, now your dog thinks whenever it sees other dogs it gets shocked. Congrats, now your dog is either scared of other dogs, aggressive toward other dogs, or both. It's basically luck of the draw if your dog will respond well to positive punishment AND it is hard to time positive punishment in a way that improves your dog's chances of responding well - even if the timing is instant like in the case of a bark collar.

Imo, it is best to go positive reward only because the risk of you messing up your dog's training is MUCH lower. And then if it doesn't respond to positive reward, you can try positive punishment. But always go with the minimally aggressive/aversive training method first. Then escalate if you have to.

Edit: another problem with punishment is it can teach your dog to hide healthy communication behaviors. Say you correct your dog when it snarls/growls at a child or dog that is getting up in its face. It learns not to snarl/growl, but it doesn't learn to not be bothered by things up in its face. Now instead of communicating that it is pissed off, it goes from seeming fine (i.e. not snarling or growling) to escalating to snapping or biting. Whenever I see people with shock collars, they often correct healthy behaviors like this and it freaks me out because they could be teaching their dog not to telegraph its aggression.

My take? Corrections of the type you're asking about are really only useful when not stopping a behavior immediately is harmful.

Most things, you have the time to step back and start over with whatever is going on, so it's more efficient to rediect, distract, or otherwise engage the animal (because it doesn't just apply to dogs) into a *desiredz behavior instead.

Yeah, a dog starts doing something that could lead to injury, you act fast, give a leash correction and maybe even full on stop them by whatever means works if the situation is severe enough. But how often is a dog going to be trying to pick up a lit firework, or pulling the leash to drink antifreeze?

Because if it isn't something that dangerous, you have the time to give a gentle redirection instead. Long term, that's the best option for an otherwise healthy animal.

And, barring circumstances beyond your control, we take the dogs out into the world, so we can choose to avoid situations where safe and gentle controls aren't going to work most of the time. Not always, for damn sure, because life throws curveballs. But if you're taking your puppy or new adult dog into new situations where you aren't going to be able to use gentler methods as the default, that's on the human. Yeah, you gotta go to vets, andb they need walks and potty time, you can't avoid that kind of thing.

So, until training is pretty far along, just pick your situations as best you can.

There's always exceptions, where a specific animal just won't respond to redirection unless it's forceful. But that shouldn't be the default

It isn't even a matter of philosophy, it's a matter of choosing outcomes. There's a shit ton of evidence that points towards minimum "negative" stimuli and maximum positive resulting in more relaxed, obedient, and generally healthily behaved animals. It just works better, and it isn't just dogs. Humans get the same outcomes, cats do, chickens do, I wouldn't be surprised to find someone had trained flatworms and gotten better results.

Now, if you want the animal to be unsure around people, but still obedient, I guess there's an argument to be made that using more negative stimuli will get that result better and more efficiently than positive as default. But that tends to break down in emergencies because instead of looking to the handler as a source of support and guidance, most animals are going to freeze up instead, worried that because they don't know what to do, they're going to get an unpleasant response. Yeah, again, exceptions exist, but we're talking about long term goals in general, not specific animals and situations.

To me, the default is touch, or sound. You literally just either touch them, or use a sound to get their attention. That's the basis of all training anyway; you get them paying attention to you, so that they can detect the actual command. If they're over at the fence sniffing, they aren't going to hear "come". So you whistle, right? You break through whatever they're focused on, and they orient on you, and you tell them what you want.

If they're on leash, and pulling towards the fire hydrant, and you want them to stop, you get their attention before anything else. That may mean stopping all movement, shortening the leash, and literally blocking their path/nose/eyes, but until they're oriented on you, you can't rely on anything. You can shout til you're blue in the face if a dog is nose blind, won't do anything useful.

The key is that in the early training they get the reward for reorienting. You train them that any time they're looking to you (or listening in the case of a blind animal) when you call their name, or whatever signal you prefer, it's a good thing. That's the fundamental necessity of training any critter, even little humans. You build the control connection where paying attention to you is the best thing possible when you ask for it/demand it.

That's how off leash work works. You set up that connection between paying attention to you as the default and it being a good thing, a thing that makes everyone involved feel good.

Once you get that connection built, it doesn't matter what's going on, because as soon as they don't know what to do, they look to you. That's how you eventually build up to a "stay" that they won't even scratch during. They scratch, you redirect their attention to you, give the command again, and eventually they get the drift. No need to tap a nose. You just tap their shoulder, and they know you want their attention. They give it freely and with no tension.

It may not seem like a huge difference, tapping on the shoulder vs the nose. If it doesn't, have someone you know tap both of them on you, and reevaluate.

By the time you're working advanced obedience or complex "tricks", if you laid down the fundamental understanding of redirection, you won't even need to verbally correct. You can just jiggle the lead or snap your fingers, and that's enough.

Minimum force leads to bare minimum effort in the long run. Think about it, if someone pinches you to get your attention for months, and then just taps you, are you going to connect the tap with the same desire from them? No, and most likely you'll react to their mere presence with tension because you'll expect the pinch.

If they start off with the tap, you're less tense to begin with, and when they switch to just a hand placed on your spinner shoulder, the decrease in force matters less, you'll still understand they want your attention. If they pair the process with a sound, then eventually you wouldn't need to be touched.

That's why a leash jiggle is effective as the eventual attention getter. It's as gentle as a touch can get, and carries sound with it. It takes longer! There's more little steps in between the first attempt and doing something like an obedience trial, but it's worth it because even if you drop the leash, they'll look to you before anything else. That's safety right there.

And, again, yeah there's exceptions. Some dogs can get so nose blind, or so excited that the bare minimum force needed at the beginning may need to involve special tools like a face lead, or even a pinch collar if they're big enough. It'll take way more steps, and some dogs just don't do subtle at all until they're so worn out by play/exercise that it just isn't practical to start small. But that type is not the norm, and tbh, it tends to only happen with dogs that weren't trained early.

I wasn't joking about it working as a principle with anything. We trained our hen to walk herself inside on command using the concept. You can't tap hens unless you want to start a fight or scare them. You can't bark commands either. You gotta be soooo gentle. But if you pair a sound with a reward when they orient on you? Boom, you unlocked the door to their understanding. They'll stop scratching and hunting for bugs, look at you from across the chicken run, and see what you want. Even the rooster will do that, and he's an idiot with too much testosterone to think about anything but food, fighting, or the other f word unless he's in the mood for cuddles.

Works with toddlers, and they're not half as smart as a rooster ;)

But it's also slower, ngl. Building up that initial connection where they always look to you, give you their attention takes months sometimes, and may not be perfectly reliable for the first year or so, if the dog is super prone to being nose blind. But it really is the most important "command" in the chain, and it's best achieved without what most people think of as a correction