Why positive only training?

aiw said:

REALLY? Dogs aren't pack animals? That seems pretty surprising. I've heard people say that dogs arent a whole lot like wolves (makes sense given the centuries of breeding) but I've never heard that they're not pack creatures... gives me something to think about.

Click to expand...

Yeah, as others have mentioned, dogs are social animals, not pack animals. Left alone to fend for themselves, feral dogs are generally solitary. When I visited Taiwan, I saw a lot of stray and feral dogs, and they were all solitary, with the exception of two huskies who seemed to follow each other. They were more concentrated in areas with human activity and food. All of my observations were consistent with research done by Ian Dunbar on feral dogs.

The idea of dogs as pack animals is subconsciously ingrained by stories, TV, and media. Everybody *thinks* they understand dominance, but really don't. Ask yourself how dominance or pack hierarchy works. Can you explain it? It's actually a complex and debated subject even in the academic world.

I think NEVER correcting a dog might require emotional control beyond my abilities, if someones in the garbage I think they should learn that makes their owner very unhappy. Most dogs aim to please so a verbal correction is plenty (firm but not scary angry). I would then follow up with teaching "leave it". That seems like a reasonable approach to me, it does include some correction though. If the bond between owner and dog is good then wouldnt the dog continue to respond to verbal corrections without needing the escalation?

Click to expand...

That dogs aim to please us is also a myth. They aim to please themselves. It's possible to use corrections with minimal fallout, but it's a very fine line and it's a risky thing to try to pull off. At first I thought the positive reinforcement stuff was just a new age mumbo-jumbo fad. As time went on, the fallout from aversive punishments was real and observable. Everything I read about the fallout from aversive/dominant training came true. It was clear I needed to shift my paradigm.

You can minimize your need for corrections by setting up the environment so that your dog does not have the opportunity to get practice unwanted behavior. Then as your dog gets more reliable, you can start adding stuff back into the environment. Reward methods actually require planning, creativity, and forethought. We're humans, so we're capable of it.

ohbehave said:

Dogs (and people) experience punishment every day. That is, they make direct contact with aversive stimuli every day.

There is nothing bizzare or particularly cruel about these things. Of course, aversive stimuli are certainly the most common ingredient in cruelty, but aversvie stimuli are entirely common and natural.

When a dog (or person) is laying in one position for too long... starts to feel uncomfortable like that and rolls to a new position for relief and renewed comfort, he has just confronted aversive stimuli and reacted to it. Not a particularly cruel scenario to my eyes.

When you are "being a tree" and a dog expends effort pulling on that leash (as well as the probable discomfort from the collar pressure on the neck), and then ceases ....voila... your dog has probably just encountered aversive stimuli and responded to it.

Walk outside in the glaring sun and then go back inside to find those sunglasses. Again, you have been punished: for going outside unprotected from the glare.

I do agree that these situations can be considered "cruel" (or extra, extra aversive) if the dog (or person) did not know how to escape the aversive stimuli... and I think it is here that the use of aversive stimuli requires some ethical and behavioral consideration.

Regardless of our desire to use 80% or 100% positive reinforcement, the world is utterly dripping with aversive stimuli that are unavoidable, daily events, and they come at our hands and at the hands of the natural environment.

What I like most about a positive reinforcement philosophy is that it might help make the dog a bit more responsive to mild corrections (mild aversives) without having to resort to heavy handed measures. Dogs that have been raised with lots of aversive techniques can get desensitized ... or so my theory goes... and require even more intense versions of aversives to get the same result. (assuming you must you punishment) Things can get ugly fast.

OhBehave

Click to expand...

Thing is, nature is full of aversive stimuli. One of my dogs learned that jumping up on the fence is a bad idea, because you can get your leg caught in the fence and break it. However, I don't particularly want to "help" the natural process by aversive stimuli coming from me. I can't control everything else, but I can control what I do. Plus there are different kinds of punishers. Trying to avoid pain or discomfort has a whole other meaning (and comes from a different place neurochemically) than mild frustration because you're not going to get the cookie right this moment.
Because I am the human with a larger brain, I can plan my lessons to be LIMA (least invasive, minimally aversive) and desensitize my dogs to things I do which could be considered aversive (like trimming toenails, necessary handling.) I don't know what the word "correction" means, so I can't say if my dogs are responsive to mild weasel words or not. I do try not to present my dogs with things worth avoiding to any degree, because avoidance is not my goal. I really like Emily Larlham's (the lady with all the cool Kikopup videos) definition:
http://www.auf-den-hund-gekommen.net/-/Progressive_Reinforcement_(+R_-P)_Training_Manifesto(E).html
I will admit that I am sometimes guilty of making eh-eh sounds at my dogs.

Pawzk9 said:

Because I am the human with a larger brain, I can plan my lessons to be LIMA (least invasive, minimally aversive) and desensitize my dogs to things I do which could be considered aversive (like trimming toenails, necessary handling.) I don't know what the word "correction" means, so I can't say if my dogs are responsive to mild weasel words or not. I do try not to present my dogs with things worth avoiding to any degree, because avoidance is not my goal. I really like Emily Larlham's (the lady with all the cool Kikopup videos) definition:
http://www.auf-den-hund-gekommen.net/-/Progressive_Reinforcement_(+R_-P)_Training_Manifesto(E).html
I will admit that I am sometimes guilty of making eh-eh sounds at my dogs.

Click to expand...

"Correction" usually refers to a punishment procedures.... often a pop on the leash or a zing from an e-collar... or a harsh "no!".

I do use the "eh eh" sound and it is merely a "no reward mark" (or more technically correct: an S-delta).

If "eh eh" has become the new "punishment!", then I am exiting the building with the positive-only crowd.

There are basic behavioral principles by which we all live by. Sometimes we get startled, sometimes we crunch our shin against a coffee table, and sometimes we get scolded.

I do recognize that a person (or dog) can have a relationship and when a human and dog get together, it is best that the human take responsibility for making the relationship as healthy as possible.

In a healthy relationship, a "correction" or "punishment" can be imposed and it is all taken in stride. For example, there are people who can correct me and their are no hard feelings at all... I appreciate the information and attempt to behave better.

There are others who may try to do the same thing, and the punishment creates problems and fails to improve my behavior.... and my produce unwanted side effects.
Murray Sidman (a radical behaviorist such as myself) wrote a book called "Coercion and it's Fallout" in the 60's)

Mis-timed reinforcement can create problems and mis-timed punishment can create problems. In addition, the intensity of the punishment can be inappropriate, too... producing problematic behavioral fallout.

There are times in which there are competing reinforcers (the running bunny) for which we have no alternative. Emily's example of the couch sitting dog, to my eyes, would fail miserably for any dog that loves the couch. Any solution that requires the focused attention and continual presence of the owner is not a solution at all.
And that wasn't just a "pure reinforcement" problem. If she had offered a solution that involves the same technique, but included punishment of couch sitting, THAT solution would require the constant presence as well. In the end, she recommends containing the dog ... I suppose in a crate.... in the absence of an owner.

To me, a better solution would be to contrive convincing scenarios in which the dog learns to never jump on the couch with me or without me. Once that training is done, the dog could live freely in the house and not in a crate! In my opinion, this is the most ethical training because it produces maximum and long term physical freedom. Yay for the dog!

People provide free health care, super healthy diets, play time, and protection from the elements... and on important and safety related occasions, the proper use of aversive stimuli seems fair (and perfectly, mundanely natural) to me.

In my opinion, I am being exceedingly, over-the-top, soft handed with my pup. However, I can identify a number of occasions in which my interactions with her include aversives.

A no-reward mark is aversive. It is a punisher (if it has been conditioned properly). Walking away from an attention-wanting dog is bad news from the dog's perspective and bad news is a punisher. These sorts of consequences are devastatingly aversive to certain types of dogs.

All of my other dogs have been quite naturally biddable and very Rin-Tin-Tin like. They were never trained by myself or anyone else. Granted, I behave and acted in ways that helped maintain and encourage the cool, livable behaviors, but I did not formally train them. My greyhound walked with me, off leash, to my University of Florida classes in the late 1980's. She would wait outside the building door. No leash. Untrained. Then, I would exit the building and she would be there waiting...and walk back to my apartment with me... through massive crowds of student pedestrians and bicyclists... across a busy main street (University Ave)... all the while, she'd walk so close when in crowds her front shoulder bone would begin to bruise the outside of my knee.

No training.

When I called her (Lucy!), nothing could prevent her from getting back to me.

If I were not aware of behavioral science and if my only experiences were with that adult dog, I could see how I could get caught up in the animal-rights, PETA type purely positive philosophy (dogma?). I didn't like the idea of that lovely dog getting the slightest flea bite... she was an angel that, to me, deserved only the comforts... fluffiest bedding and best food.

Not all dogs are like Lucy, though. On occasion, situations need to be carefully contrived that will directly reduce or eliminate serious problem behaviors.

A working **** Hound would not have given me such an easy time... and would test the last nerve of even the greatest of purely positive trainers. Try as they might, they will never be able to produce an apartment-friendly Cockapoo out of that Hound.

It's a reinforcement problem. A dog driven to smell every scent and chase every insect and mammal to the exclusion of your best edible or toy will be dicey, at best, even with excellent positive reinforcement training. Again, with no training, I could walk my Lucy through a field of squealing bunnies, children, or other dogs and she would be 100% reliable. You wouldn't get that from a positive only trained **** Hound. Ever.

It's a reinforcement issue... a dog with the DNA to seek and find prey or any other happy consequence that isn't in tune with our desires gives us a reinforcement problem.

So, I remain unconvinced about the purely positive only philosophy.

oh behave

ohbehave said:

A working **** Hound would not have given me such an easy time... and would test the last nerve of even the greatest of purely positive trainers.

Click to expand...

I haven't been around hunting dog training for about 20 years now but I always thought many people who did it were too caught up in the macho aspect of using potent aversives/punishments to train. The thought was that you had to beat/punish/choke/shock/whatever a hunting dog into submission to get it to perform. I think some of those trainers actually enjoyed the punishment they inflicted.

My Grandfather (a veterinarian), his brothers (2 of them were also veterinarians) and my cousins always used mostly positive training for hunting and herding dogs. They had great dogs that would follow whistle commands quite well in the field. (I recently posted a link to a dog training book from 1882 that is a lot like their training methods.) They never used choke chain, prong collars or hands on punishment to train dogs. By technical definition they probably did use some aversive methods like saying "No" but these weren't common because they were rarely necessary.

While it would be rare to have a scent hound or other very driven breed to operate untrained in a highly distracting environment, training that is 95-99% positive with only the mildest aversives can allow them to work effectively in spite of distractions.

You could train in dog language, though it may take a lot of learning on your part. Dogs don't care if the other chess a shoe, tears up the trash, or peed on the carpet. Dogs don't reward others for good behavior. They may intice one to play by giving a toy, but they won't give one a treat for sitting.

Some high drive dogs get great satisfaction out of certain behaviors. The coonhound for example, may get much more out of running a track than doing as told. So the reward must be greater than the distraction, and given so the dog understands it clearly. One of mine during a certification, couldn't help but chase a rabbit she stirred up working cadaver. She was stressed, though had what I thought to be 100% recall lol. Sometimes reactive dogs can't resist!

I have never seen a purely positive trained dog. I have seen minimal punishment used, and these dogs work great. Then some take more discipline. I do feel if raised as a young pup with lots of positive training, they do much better.

By "verbal correction" I essentially mean "Get off the table you jerk!". I wanted to make it clear that when I am talking about corrections I don't mean hitting the dog, just communicating my unhappiness. I am definitely on board with the idea that you want a strong, positive relationship with your dog. If you dont want that you probably have no business owning a dog. And, yes its best to teach good manners before bad habits crop up but there is often enough work to be done with things that ARE issues that sometimes things which arent get set aside.

We have a problem in our house with a self-reinforcing behaviour, jumping up on the kitchen table. Generally we keep unattended food out of reach but a couple of times our dogs found some. They are even jumping up there when we are in the room! I'm not prepared to 'just ignore' that behaviour or essentially relinquish our food to doggy appetites. If I don't set a negative consequence with getting up on the table wont their motivation for our food win out every time? What if its both, a negative consequence ("NO!) for jumping up and reward for leaving it alone? I think its really important to keep the vast majority of your interactions positive but surely the world cannot be engineered to be all positive all the time.

ohbehave said:

"Correction" usually refers to a punishment procedures.... often a pop on the leash or a zing from an e-collar... or a harsh "no!".

Click to expand...

Then why not just call it a punisher? The reason is because "correction" sound nicer. I believe that if you (general you) are going to punish, you should have the courage to admit it at least to yourself. And if you can admit it to yourself, you can admit it to others. I have more respect for someone who states that they use some punishment than someone who dances around the issue with weasel words like "correct" that have no real definition in behavioral or training terms.

ohbehave said:

"
I do use the "eh eh" sound and it is merely a "no reward mark" (or more technically correct: an S-delta).

If "eh eh" has become the new "punishment!", then I am exiting the building with the positive-only crowd.

Click to expand...

And yet, later in the same post you write:

ohbehave said:

"A no-reward mark is aversive. It is a punisher (if it has been conditioned properly). Walking away from an attention-wanting dog is bad news from the dog's perspective and bad news is a punisher.
.

Click to expand...

So which is it? It is or it isn't. I will certainly admit that walking away from a dog is a negative punisher. But it also isn't generally a no reward marker. I guess I don't condition mine correctly, because I use eh eh very rarely simply to interrupt a bad idea (not just a wrong choice). No primary aversive attached. And I think you'll find the majority of people talking against some mythical positively only crowd are people who are arguing against a red herring. The people who use primarily positive reinforcement (I suppose you could use primarily positive punishment and still call yourself "positive only"?) will usually admit to occasionally thoughtfully using something in another quadrant.

ohbehave said:

"In a healthy relationship, a "correction" or "punishment" can be imposed and it is all taken in stride. For example, there are people who can correct me and their are no hard feelings at all... I appreciate the information and attempt to behave better. .

Click to expand...

If someone points out to me where the equation went wrong, and gently leads me to an understanding of the right way to solve it, I'd be fine with that. If they yell at me for being stupid and rap my knuckles with a ruler, they are likely to poison any affinity I might have for math. If I make a social faux pas and you correct me for it instead of tell me a better way to handle it, I'm likely to have an aversion to future social events of the same sort, because I am anxious and don't know what to do instead.

ohbehave said:

"There are others who may try to do the same thing, and the punishment creates problems and fails to improve my behavior.... and my produce unwanted side effects.
Murray Sidman (a radical behaviorist such as myself) wrote a book called "Coercion and it's Fallout" in the 60's).

Click to expand...

I'm familiar with it. I'm not sure I'd consider you in the same league as Murray Sidman anymore than I'd claim that Karen Pryor or Bob Bailey is "like myself"

ohbehave said:

"There are times in which there are competing reinforcers (the running bunny) for which we have no alternative. ).

Click to expand...

Odd. My alternative to bunny chasing is a recall. Because my dogs have a great history of reinforcement and I have created a habit, it works, even though my cookie is not as exciting as killing and eating the bunny would be.

ohbehave said:

" My greyhound walked with me, off leash, to my University of Florida classes in the late 1980's. She would wait outside the building door. No leash. Untrained. Then, I would exit the building and she would be there waiting...and walk back to my apartment with me... through massive crowds of student pedestrians and bicyclists... across a busy main street (University Ave)... all the while, she'd walk so close when in crowds her front shoulder bone would begin to bruise the outside of my knee.

No training.

Click to expand...

I did much the same with my first "own" dog, an Aussie named Demian back in the early 70s. I'd take him to college with me and he'd wait till I came out, and I frequently walked him on sidewalks near busy streets off leash, and play with him off leash in the park bordering one of the busiest retail areas of Kansas City, MO. I was very young and stupid and he was very lucky to survive my cluelessness. By the way, if you and the dog are in the same space and you are both awake, training is going on. Either you are training the dog or the dog is training you.

ohbehave said:

" It's a reinforcement problem. A dog driven to smell every scent and chase every insect and mammal to the exclusion of your best edible or toy will be dicey, at best, even with excellent positive reinforcement training. Again, with no training, I could walk my Lucy through a field of squealing bunnies, children, or other dogs and she would be 100% reliable. You wouldn't get that from a positive only trained **** Hound. Ever.

It's a reinforcement issue... a dog with the DNA to seek and find prey or any other happy consequence that isn't in tune with our desires gives us a reinforcement problem.

Click to expand...

It depends on your reinforcer. If you've studied behavior, I'm sure you've heard of Premack? Sue Ailsby teaches her stud llamas to walk nicely on the way to breed the girls by Premacking it. Sex is the reinforcer. It does involve a bit of negative punishment, though. Also, "go sniff" is a great Premack for many dogs. Interestingly, if you give them permission to do something it become much less interesting than if you try to keep it forbidden fruit.

ohbehave said:

" So, I remain unconvinced about the purely positive only philosophy.

oh behave

Click to expand...

Well, I don't think anyone is going to hold a gun to your head and tell you you have to depend solely on R+/-P (as I said, purely positive probably doesn't exist except as an ideal.) But, if you haven't tried it, or at least seen it done well, I'll remain unconvinced by your lack of convincedness (is that a word?)

aiw said:

By "verbal correction" I essentially mean "Get off the table you jerk!". I wanted to make it clear that when I am talking about corrections I don't mean hitting the dog, just communicating my unhappiness. I am definitely on board with the idea that you want a strong, positive relationship with your dog. If you dont want that you probably have no business owning a dog. And, yes its best to teach good manners before bad habits crop up but there is often enough work to be done with things that ARE issues that sometimes things which arent get set aside.

I think its really important to keep the vast majority of your interactions positive but surely the world cannot be engineered to be all positive all the time.

Click to expand...

Ah, but when other people say "correction" they may mean other, very different, things.
That's why it isn't a very useful word when describing behavior and what you do with it. Punishment means something that reduces the likelihood of that behavior occuring again. (so, unless I want my dog to be less likely to sit, I'm not going to punish a sit if I asked for a down). Unfortunately, people who haven't educated themselves about operant conditioning don't necessarily understand that (or that "negative" isn't "bad" - it simply means you took something away from the situation)
It's too bad that both punishment and negative have other meanings in common speech. I could come up with "nicer" terms probably. But then I'd be in the same situation as using the word "correction" which doesn't mean anything to the majority of people I'd be talking to. If you don't understand the meanings of the words, you know, the information is readily available at the click of a key. Look up operant conditioning.
The world can't be controlled as far as aversives go. But I can engineer my training to control MY use of aversives. That's the point.

Very interesting thread.

There's just two points I'd like to add.

The first point I'd like to add is about the term "correction". I don't really get hung up about terminology. But the reason why- speaking only for myself- I use the term "correction" rather then "punishment" is twofold.

1. While "punishment" has a clearly defined meaning in operant conditioning theory - and "correction" has no such meaning - most people misunderstand that word. That's because the word punishment also has a common meaning which does not necessarily line up with learning theory, and very few people who handle dogs are versed in the theory.
   
2. I actually don't even use the specific word "correction" very often in real life training. The terms I use more often are "correct that..." or "fix that..." and ALWAYS followed by a specific action that should be done. As in, for example, "let's try to fix that wide turn by a pop just as you go into it" (there are other ways of doing that, of course - this is just one way)

The second thing is about restraint and removal. There are times when dogs have to be restrained from doing something - often when doing something that could be dangerous to themselves or to others. There are also times when you have to remove something that a dog likes - maybe for the same reasons. I'm not going to get into the whole "nurture vs. nature" argument, just describing a fact. Now you can simply do the restraint or removal, and not attach anything to it. Or you can use the situations as a teaching opportunity. I prefer the latter - even though these are clearly punishments in the operant conditioning sense- because I always want to be teaching my dogs. But it's up to you.

ohbehave said:

"There are times in which there are competing reinforcers (the running bunny) for which we have no alternative. Emily's example of the couch sitting dog, to my eyes, would fail miserably for any dog that loves the couch. Any solution that requires the focused attention and continual presence of the owner is not a solution at all.
And that wasn't just a "pure reinforcement" problem. If she had offered a solution that involves the same technique, but included punishment of couch sitting, THAT solution would require the constant presence as well. In the end, she recommends containing the dog ... I suppose in a crate.... in the absence of an owner.
oh behave

Click to expand...

Ooorrrr . . . you can teach the dog that HIS bed is absolutely the best place in the world. Sometimes when you are on it, treats fall from the sky. In early stages of training, management is key. But once you've created a habitual behavior you don't need the management or the presence of the human. I've taught all my dogs a very good leave it. True story, we have some awful hailstorms in OK lately. In the last two years we have had storms with tennisball sized hail, having to replace glass in cars and roofs. It's also very hard on any birds around to be hit by something that big. Kills a bunch of them. About three days after the first storm, Ray came trotting around the house with a rotting starling. I told him to drop it (he did, without any attempts to keep his prize) and I told him and his mom and auntie (who were also out) to "leave it." I went into the house to get stuff to dispose of the mess. When I got back out, all three dogs were sitting a respectful distance from the carcass and it had not been moved. Because I had trained the behavior I didn't have to be right there to enforce it.

aiw said:

I'm not sure I buy the idea that dogs are entirely self-interested. I think that there is an emotional bond which extends both ways. If a close friend is upset I experience that as an aversive, it would be all the more so if they brought me my dinner each night! Dogs don't experience the world in the same way as humans but kindness and emotional bonds arent solely the province of humans. "Aversive" events happen in relationships (between both people and animals) all the time and it doesnt destroy the bond, I've never had a relationship that didnt involve at least one disagreement.

Click to expand...

Even if you break down altruistic human behavior, you will still trace the roots back to basic selfish needs.

- If my boss is in a bad mood, I'm going to stay out of his way so that I don't F myself over somehow.
- If my best friend is upset, I want to cheer him up so that I can be back to joking around. You could say, "but I'm concerned about my friend's well-being". Why? Because you like your friend and you don't want to lose him/her.
- If my dog does a bunch of cute stuff at me that makes me laugh, there is a good chance I will give him food or play with him or pet him.

As kids, we're taught selfishness is a bad thing, but actually, if you think about it, it's kind of the driving force behind society. It's just, there's proper selfishness which helps your society and ultimately yourself, and improper selfishness which helps yourself in the short term but harms everybody in the long run. Everybody wants to do whats best for themselves because that increases your rate of survival, and as it turns out, for social creatures like dogs and humans, what's best for ourselves correlates with what's best for everybody. So if we are able to tap into our dogs' selfish needs for survival, we have about a 100% chance of reaching his brain.

Without getting philosophical let me just say I don't agree with the premise that society and human nature is based in selfishness. That said, it makes sense to create incentive for good behaviour, unfortunately sometimes bad behaviour has stronger innate incentives (food on the table is a good example). In that situation you would have to either come up with something more motivating than the bad behaviour or find a way to lessen the bad incentive... or both.

Aversive is the operant conditioning term but I prefer the idea of consequence. A predictable system that offers both good and bad outcomes, I think the system should be set up to create the maximum good outcomes but occasionally its in the best interest of everyone to teach that a certain choice is a BAD one. In those cases the discomfort of the aversive might be WAY less than the possible real-world consequence (eg. a dog that chases animals into the street who wont respond to treats but will respond to an ecollar). Mostly though I think we all agree that minimizing discomfort and unhappiness is the way to go, the world's not perfect though and there is only so much we can control.