Question about Positive reinforcement

[Rotlover2604]

Hello all,
First post on the board. I've had dogs throughout my childhood and never heard of using positive reinforcement with the dog until reading a few books and this forum.. here's my question ;
How does positive reinforcement always prevail over "punishing" your dog? Say my dog nips at my hand and I want to teach him not to.. well normally I would yell "no" and maybe lightly tap him on the nose as soon as he did it. Using positive reinforcement would mean to praise or give treats when he does not bite my hands, right? Please excuse my ignorance if my understanding is wrong about they way positive reinforcement works.
Another example : I'm in my car and my rottie is barking at a car/motorcycle behind me. Rather than yelling at him to not bark, i'd wait until he paused between barks and praise/treat..correct? What I dont understand this - I would think that giving him treats after he stops doing something bad, rather than punishing him for doing something bad would give him the wrong idea..? I dont want to accidentally teach him that he gets treats and praise for barking at people and then stopping, rather than not barking at all. Someone please straighten me out on this.
Thanks,
Ryan

[Tywaknya'sBABY]

Timing is everything. I don't reward when she stops biting me. I walk away when she starts because she was starting to combine biting and rewards so I get your point.
My father used negative reinforcement so I was new to this also.
When you reward it makes them WANT to do what you want.
When you punish they do it because they HAVE to and what happens if one day they don't feel they have to?
You don't reward as soon as they stop doing it. You want to catch them when they are doing good.

At second glance this looks confusing. Sorry.
I'm sure theres someone who can explain it much better

[amontague]

My understanding of positive reinforcement is to try to reward them doing the right thing *before* they do the wrong thing. For example, I was having a problem with my dog barking at people whenever he saw them through the window. My trainer suggested I give him a treat the second he saw someone, but before he started barking. Continue giving him treats (and saying the command "quiet" in a pleasant voice) as long as he didn't bark. But stop the treats the moment he started to bark. It took a while, but he started to get the idea: no barking = getting treats, barking = treats stop coming. So for your nipping example, the idea would be to figure out what is happening *before* the dog starts nipping (e.g. vigorous play), and reward before the nipping starts. Then, if the dog nips, stop the rewards, put your hands in your armpits so he can't get at them, and turn your back and ignore the dog for at least 30 seconds. Then he learns that nipping ends all of the good things in life.

I hope that helps.

[spidey]

I think you might be misunderstanding the concept a bit, and the terminology has been misused so much in dog training conversations that I'm not surprised. There are four elements to operant conditioning (which is what most dog training is at least partly based on). While some positive trainers state that they ONLY use positive reinforcement, the truth is that most use at LEAST a combination of positive reinforcement (giving the dog something the dog wants to increase the incidence of a behaviour) and negative punishment (removing something the dog wants to decrease the incidence of a behaviour). Many trainers use those two plus positive punishment (you do something the dog doesn't like to decrease the incidence of a behaviour) and sometimes negative reinforcement (you STOP doing something the dog doesn't like to increase the incidence of a behaviour).

Here are examples:
- your dog sits, you give the dog a treat - positive reinforcement
- your dog bites you, you go away - negative punishment
- your dog pulls, you give a leash correction - positive punishment
- you call your dog, you apply zaps with an e-collar until your dog starts to come to you, then you stop zapping - negative reinforcement

It's very rare to find a trainer who is successful using ONLY positive reinforcement. If you want your dog to stop biting your hands, the easiest thing to do is precisely what another dog would do, which is negative punishment - you remove yourself from the situation (and maybe yelp). The dog wants to play, you are teaching the dog that playing ends when the dog bites - the dog learns that if she wants to play, she cannot bite. You can use positive punishment (pinching, bopping on the nose, whatever), but there is always the risk that this can elevate the problem by exciting the dog more (not to mention making the dog headshy), and simply removing what the dog wants (you) usually works well as long as you are consistent about it.

With regard to barking - keep in mind that it is MUCH easier to teach a dog what TO do, than what not to do. "Quiet" is a difficult concept for a dog, because it is the absence of a behaviour. What a clicker trainer friend of mine does (and it works extremely well if you are diligent and consistent) which is 100% positive reinforcement, is train "bark" on cue, once you have "bark" on cue (after a few sessions), you ask for a bark, reward it with a low-value treat (like kibble), and as the dog stops barking to eat the treat, you give your "quiet" cue (and ideally also a hand signal like holding your hand out in a "stop" gesture), since the dog is not barking, you click and treat with a VERY high value treat (roast beef, cheese, bacon, whatever the dog LOVES). If you do this enough times, the dog will learn what your "quiet" cue means, and because you reward it with a higher value treat every time, the dog will learn it pretty fast. As with any other training, you need to tune it up regularly, but it does work - I phased out the treats altogether for the "bark" part, but still reward the "quiet" part with very high value rewards, to keep the idea in my dog's mind that the "quiet" one is always worth listening to (barking is also pretty self-rewarding for some dogs). You need to train the two cues together, since the one you REALLY want (quiet) can only be trained when the dog is doing what you don't want (barking). Besides that, then you have "speak" on cue.

[Rotlover2604]

That really cleared things up for me. Thanks everyone!

[SABELLESMOM]

Originally Posted by Rotlover2604

I'm in my car and my rottie is barking at a car/motorcycle behind me. Rather than yelling at him to not bark, i'd wait until he paused between barks and praise/treat..correct?

This was my example. The dog is question is used to clicker training which targets the behavior at the moment of the click. If it were being used for teaching "sit," for example, the click would be timed to the butt hitting the dirt.

In the case of the barking at the mirrored black motorcycle helmet following us, the dog was clicked the moment there was no barking. It was difficult to time, of course.

Still, being used to the clicker, it was as if the dog stopped to think. "Now what was I doing when she clicked me? I'm going to see if it happens again." When it did, her behavior was reinforced. After a few more reinforcements, she decided that "not barking" was what I wanted from her and she complied.

By clicking the moment barking stopped is not rewarding the barking. Barking wasn't clicked. Silence was. So even tho barking occurred seconds before, it is not reinforcing the wrong behavior. Obviously, as the dog understood and stopped barking.

Trainer error (badly timed clicker) would not see the right results until the timing was corrected by the trainer. That is more quickly achieved than you think. The dog will learn when the trainer does it right.

Clicker Training Lessons - How to be a Good Leader
www.clickerlessons.com/lessons.htm

[spidey]

In the case of barking at the motorcycle, I'd distract the dog and ask for a reliable behaviour, then click/treat THAT instead. If you do this enough (and make it fun/rewarding enough), the dog will learn a default behaviour of switching his attention to you when he sees the motorcycle.

I'd also like to mention that "positive" and "negative" don't really mean what many people think they mean in terms of behaviour modification. "Positive" doesn't mean "fun", it more or less means "adding something" (hence "positive reinforcement" means "adding something the dog likes", "positive punishment" means "adding something the dog doesn't like"), and negative doesn't mean "not fun", it more or less means "removing something" ("negative punishment" is "taking away something the dog likes", and "negative reinforcement" is "taking away something the dog doesn't like").

[groupieindenial]

Ditto on what most said. You can't use ONE method only.

I had issues for a LONG time with my boy barking at motocycles. What worked is BEFORE he started barking, I would say "leave it" or "no" when I could see him thinking about barking at it in my rearview mirror. When I had him harnessed back there, I would toss a treat into the back. My aim was bad, poor guy would be too busy looking for the treat, he'd forget about the stupid motorcyle!! Hey, it worked, within a few weeks when he'd see a motorcyle he'd look to me first, so I had better have a treat handy!

I LOVE manipulating their minds

[Ayoka]

Thank you Spidey for the great explanation of reinforcement and punishment positive and negative...I have read about this before but it all seems clearer now.

In any class that I have taken, the positive reinforcement & negative punishment have been used (it works for us!!!) but they have never really been explained in the context of other methods.

In my opinion, this would make a great sticky in the Training section.

[Rotlover2604]

Yeah that was definitely a big help.. I'm so used just yelling "no" if the dog does something bad, rather than rewarding for doing something good. I'll try it

[spidey]

I'd also like to mention that the order I listed things in in my first post is generally the order I like to apply these methods in if possible (although I don't think I ever use negative reinforcement, that I'm aware of). At very least, when training a puppy or training a new behaviour to any dog, I feel you should start with positive reinforcement or negative punishment (depending on the behaviour, since some behaviours, like play biting, are best addressed with negative punishment straight away), long before you ever consider positive punishment. I only resort to positive punishment (usually verbal corrections but occasionally leash corrections) when I feel a behaviour has been trained and proofed to the extent that I have cause to believe that the dog actually does understand it, or if there is safety at stake.

[debbiej]

another thing I've learned is to refine the vocabulary used with the dog. even make a list so everyone (even you) uses the same word to mean the same thing. and "no" is meaningless to a dog. "no bark" means something, but choose commands that can have a specific meaning. (is this off topic?, if so, sorry)

[groupieindenial]

Originally Posted by debbiej

"no bark" means something, but choose commands that can have a specific meaning. (is this off topic?, if so, sorry)

Its funny how they don't put the two together unless you taught them together. I mean, they just understand a specific string of constanents. Blah blah, blah blah "go for a " and their ears perk.

I was trying to teach my dog not to bark (a family memeber inadvertantly tought him that barking got him a treat, even if he did'nt do what he was asked). So my dog knows "Speak", I thought if I say "no speak" when he's about to bark at what I don't want him to bark at, he'd get it. It was too funny. "nooooooooooo" - and he'd intensely pay attention...."speak" = WOOF! <----Vegas. DAMNIT. "no"............"speak" = WOOF! DAMNIT!! lol

Its funny how all they hear is the command that means to do something. I think dogs think "no" is not to do anything, stop what your doing etc...

Saying no, no, no, when I could see he is thinking about barking and rewarding him worked.

[debbiej]

LOL...
how about 'quiet'?

[spidey]

Originally Posted by Rotlover2604

Yeah that was definitely a big help.. I'm so used just yelling "no" if the dog does something bad, rather than rewarding for doing something good. I'll try it

So many people are in this boat. Ask yourself a) what incentive does the dog have to do the right thing if you never reward it? and b) punishment only provides one side of the equation to the dog - would you rather I told you what I wanted and paid you when you did it, or just shouted at you when you didn't do what I wanted, even though I'd never trained you to do it?

Dogs are very simple, they repeat behaviour which gets them something they want, and they eventually stop performing behaviour which doesn't get them what they want (or which gets them something they don't want). It's unfair and counterproductive to just tell a dog no, teach them what you DO want. I'm glad you're planning to try something different, you'll be surprised, and your dog will thank you.