Stop It!

February 20, 2010 at 9:55 pm 24 comments

Negative reinforcement is probably the most misunderstood concept in dog training. Many trainers equate negative reinforcement with punishment and then condemn its use as inhumane – while ironically some behaviorists have called to eliminate the distinction between positive reinforcement and negative reinforcement.

I wonder how many of these people ever went shopping with a tired, crabby toddler? Because when a cranky three-year-old throws a tantrum until that box of atomic sugar bombs goes into the shopping cart, he’s giving you a textbook example of negative reinforcement.

And the beauty of it is, you’ll rarely (if ever) need to use a stimulus as harsh as a tantrum when training your dog.

Animals are masters in the subtle, effective use of negative reinforcement. They use it all the time to establish and protect territory, maintain individual space, protect food resources and to teach boundaries to their young. Negative reinforcement teaches your dog to lay in the shade instead of the sun on a hot day. And if you’re lucky, it’s the tool your puppy’s mother and littermates used to teach him bite inhibition.

In Animals in Translation Temple Grandin writes:

Rewards and positive reinforcers are the same thing: something good happens to you because of something you did. Punishment and negative reinforcement are opposites. Punishment is when something bad happens to you because of something you did; negative reinforcement is when something bad stops happening to you, or doesn’t start happening to you in the first place, because of something you did. Punishment is bad, and negative reinforcement is good. Punishment makes you stop doing what you’re doing, although a lot of behaviorists believe that punishing a bad behavior isn’t as effective as rewarding a good behavior when it comes to getting an animal to do what you want him to do.

Negative reinforcement is the hardest to understand. Negative reinforcement isn’t a punishment; it’s a reward. But the reward is negative in the sense that something you don’t like either stops or doesn’t start in the first place. Say your four-year-old is screaming and crying and giving you a headache. Finally you lose your patience and blow up at him, and he’s shocked into silence. That’s negative reinforcement, because you’ve made the crying go away, which is what you wanted. Now you’re probably more likely to blow up at him the next time he starts a tantrum, because you’ve been negatively reinforced for blowing up at him during this tantrum.

And part of the beauty of negative reinforcement is that, unlike punishment, it’s highly effective even when very subtle, mildly aversive stimuli are employed. This can involve little more than being somewhat annoying or tiresome – you just need to apply enough pressure that the subject of your attention wants to do something to make you stop it.

Here’s an example of mild pressure being used to make a horse back up and move forward. Note how calm and relaxed the horse and trainer are. In spite of all the ugly baggage we humans associate with the term, like a smile or a frown, negative reinforcement in its milder forms is one of the foundations of human and animal body language.

Ironically, even though negative reinforcement is something that all animals (including humans) use without thinking about it – most humans need coaching to use it consciously. The video linked here (by the most excellent Lori Drouin) provides some nice examples of how to introduce negative reinforcement (or pressure and release) into lure / reward training. It’s simple, fair and effective and as Ms. Drouin points out, the use of pressure / release also gives us an excellent way to segue out of luring.

Transitioning from luring to negative reinforcement is valuable because it eliminates the dog’s reliance on treats. In addition, because negative reinforcement is an integral part of dog and human body language, using it teaches you to pay attention to how you use subtle cues like touch, eye contact and body posture to turn pressure on and off. This is totally awesome because, along with petting and praise, these are tools you’ll have with you anywhere you work with your dog.

Don’t fall into the trap of believing the propaganda that using negative reinforcement to communicate with your dog (or, for that matter, your toddler) is cruel and abusive. Negative reinforcement is a fair and natural part of the way animals communicate with each other. Practice it. Learn how to use it well. Your dog will thank you.

Entry filed under: behavior science, dog training, dogs.

Behind the Kennel Door Distorted Perspective

24 Comments Add your own

1. Christina | February 20, 2010 at 11:24 pm

Thank you VERY MUCH for posting this! A topic I am very interested in, as it is important to move away from TREATS TREATS TREATS all the time. In addition, I don’t always have pieces of liver or cheese in my pocket, so with what I have learned here, I can train anywhere, with or without the cheese. Add a little praise or a pet, and I am in business! Thank you Smartdogs…..
2. Viatecio | February 20, 2010 at 11:56 pm

The question is, when does R- become P+?

If I give a leash correction for lunging and the dog no longer lunges on leash, is it because I have applied an aversive to stop the behavior or has the dog been reinforced to walk on a loose leash by avoiding the negative stimulus? If the dog is wearing a bark control collar, is it being reinforced for staying quiet by not being stimmed, or is it being punished by receiving a stim when it barks?

Very interesting read. However, it’s not something I’m actively thinking about when training…I just DO and DO what works. Psychology and behaviorism stuff is a great standard to have to talk about training and it should be discussed, but it just confuses me when that’s ALL some trainers talk about. Worry about the forest first, then look at the trees…
3. Mongo | February 21, 2010 at 12:42 am

I think its also fair to mention that nearly all dogs earliest life lessons are shaped by negative reinforcement.
A good dam raising a litter is a busy beast delivering growls, cuffs and an occasional “pin it down till it squeaks” to teach her pups manners.
Removing fair, sane negative reinforcement completely from training does dogs a huge disservice. Granted, lessons can be taught with P+ but require more TIME- not just human patience- TIME.
For creatures who have at most, ten good years of life to live, stretching the time line to learn the basic training leassons to be good companions at our side out , just so trainers can feel good about themselves is deeply unfair.
just my .02
4. Wild Dingo | February 21, 2010 at 2:34 am

YES! as a post grad (very old post grad) psych major, who’s hero was good ‘ole BF Skinner, I couldn’t agree MORE with this! thanks for the video link. awesome. I’m so tired of hear the “positive-only’s” spew that we shouldn’t do somethng that makes our dogs uncomfortable. Life is full of STRESS. stress is how we learn (people and animlals). no of course we shouldn’t compulse to teach a new behavior. Smacking someone because they don’t know how to sit, or shoving their butts in a chair is just rude if they don’t know the meaning. but after learning the meaning of a command, putting pressure on the dog and releiving pressure is simple. I don’t see how freaky ‘positive only’ people get so upset about this.

In fact, the very essence of using lures and clicker training itself can be construed as stressful because the dog can’t get what it wants until it guesses the behavior! stress builds frustration and frustration builds drive –well most times. it can shut down drive but you learn how far you can frustrate the dog before you have to give him something to stay in the game of learning new behavior.

thanks for sharing the vids. can’t wait to watch them!

oh and i just read one of the comments above and thought of the use of e-collar. there are different ways of using e-collar. one is through “escape”…the dog is stimulated and learns to do a behavior to escape stimulation (i feel is barbaric). i wonder if this is considered negative reinforcement or punishment. I would consider it punishment tho the pressure occurs BEFORE the dog does the right behavior (which sort of defines neg reinforcement). versus the dog is given a command and opportunity to exeucte command and if not, given a short stimulation of the collar similar to a leash correction. which of coures is considered “compulsion” or “punishment”. in the e-collar situation, if using one, it seems punishment the lesser of the evils!

it seems both punishment and negative reinforcement have some form of pressure applied. punishement occurs after behavior, negative reinforcement happens before behavior. in which case the escape ecollar stim is still quite horrifying.
5. Ed | February 21, 2010 at 1:21 pm

Wonderful job of presenting what should be easily understood ideas and I’m excited to see someone who has the scholarly chops for the job and dog-training experience go through Temple Grandin thoughtfully (as here and in some of your previous posts). Although her books have value I’ve always been disturbed by the uncritical way people thumb through one of her books or read a magazine article and then jolly it all up into a gooey ball of magic.

Like Wilddingo, “I don’t see how freaky ‘positive only’ people get so upset about [negative reinforcement] while they are fine with confusing a dog for ages under the guise of “shaping a behavior” and then, when months of not-really-training leaves then with a confused and difficult dog, they resort to the allegedly pure positive of NILIF.

Raining today – maybe I’ll try getting started on the yo-yo game today in the barn aisle.
6. Viatecio | February 21, 2010 at 2:15 pm

Wild Dingo, I like your thinking!

I think the training you were referring to with the e-collar would be negative reinforcement, but similar to how a dog must “guess” a behavior in order to be rewarded by a pure-positive trainer, the dog must “guess” the behavior to turn off the stim. Both dogs are reinforced, and both will be stressed again: what EXACT behavior brought that goody, vs when will that stim be turned on AGAIN and what EXACT behavior turned it off.

For the record, I am not the biggest fan of using this type of negative reinforcement to train with e-collars…and as for Parelli, he has the right idea but I tend to go with the mindset and technique of Clinton Anderson–more realistic and practical IMHO 🙂 Just the things I’ve seen and read about PNH don’t impress me whatsoever and even seem downright dangerous sometimes.
7. stone soup diaries | February 21, 2010 at 2:24 pm

It all goes back to the parsing of language. ‘Pressure’ can be applied many ways.

The fact that most people, trainers alike, have been conditioned to use the leash and collar to prevent behaviors because that’s how it has been used for years.

If they break it down into its component parts, ANY TRAINING TOOL should only appear as a way to limit options.

The application of pressure, socially, environmentally, physically or otherwise is always what generates the appropriate response.

The withholding of a treat or game can be construed as pressure in that context. Turning your back on a jumping dog is a punishment. It is also pressure.

In regard to the ecollar, again, the language is coarse and subject to misinterpretation. Tortora used the terms escape, avoidance and punishment to describe what he was seeing in his experiments with electronics.

Escape, avoidance and punishment occur in the natural world.

Pressure is applied with a leash and collar in the same way.

When using an ecollar, halti or regular collar, one applies the principals, not necessarily in a putative way in order to acquire an acceptable or rewardable behavior.

The term “escape” was applied until a dog found an alternative to a behavior in order to relieve discomfort. The pressure was applied and the dog volunteered behaviors until he discovered the one that relieved the pressure. (Positive Punishment)

The term “avoidance” was applied as a dog volunteered a behavior in order to avoid discomfort. Having been conditioned using the principal of escape, the organism learned to avoid the circumstance/behavior/action that CAUSED the pressure. (Neutral unless the dog did not respond appropriately, then it became Positive Punishment)

The punishment phase is the extinction process. Pressure is applied at a level of extreme discomfort to the dog at a time that it appears as a consequence to an act. With Tortora’s experiments, punishment was applied as a direct result of the organism’s deliberate attempts to engage in an act that had been preconditioned AGAINST. (Also Positive Punishment)

Appropriate application of ecollars are not painful during the escape phase of the training, they become even less so as the dog learns to ‘avoid” the stim of the collar.

Any way you look at it, it is Positive punishment as it is defined by the talking heads.

What you describe is the use of pressure in its varying forms, “escape” to introduce a behavior, “avoidance” to teach a behavior, and punishment for the permanent extinction of a behavior..

Current technology in electronics allows for the handler to select a level that is as mildly uncomfortable as an insect landing on your hand. The deliberation in creating collars that have a virtually infinite number of settings allows for the stimulation to be sufficient for the individual on the end of the leash. The current collar manufacturers are the direct benefactors of Tortora’s experiments, allowing for the creation of collars that provide for a stimulation level so neutral and innocuous that at the lower levels, they cannot be construed in any way as painful.

But still the language does color one’s attitude towards the tool.

If you read his study, you will find that the language of escape, avoidance and punishment predate him as well.
8. SmartDogs | February 21, 2010 at 2:47 pm

It doesn’t have to be any more horrifying than the leash pressure Lori Drioun uses in her video. Escape / avoidance training with the ecollar can be done very effectively using nothing more than mildly annoying levels of stimulation. This can either be ‘shock’ turned to low levels or vibration. And – in my experience, most dogs are a lot more stressed out by the vibration than very low ‘shocks.’

A good collar (TriTronics is my favorite brand) can be turned down so low you can’t even feel it and it’s easy (at least for me) to find a level that a dog notices, but isn’t more than minimally stressed by. I started Audie on ecollar work when he was three months old. You can ask his breeder and other friends how I’ve ruined my boy.

Like most tools, the ecollar doesn’t work in a vacuum. You have to be able to tell the dog what he needs to do to turn it off, so in early work you must either have the dog on a leash/long line or have already trained the dog to respond to directional cues. The classic ‘three action’ introduction to ecollar work where you teach a dog to come to you, go away to a target and hold a stay takes advantage of this.

The “hammer the button until the dog manages to do what you want” method used by some ecollar trainers is far too stressful on dogs for my taste. I’ve seen it generate a lot of ugly, frantic escape behavior. I prefer to use it to play an electronically enhanced game of hot and cold.
9. stone soup diaries | February 21, 2010 at 2:55 pm

EgggZacly.

I openly despise some of the ways I see ecollars being used.

I like to think of myself as a trainer who uses ecollars as opposed to an ecollar trainer, since I trained traditionally long before I began studying and using them.

PS. I’m a DogTra user after the G2 debacle a few years ago.
10. SmartDogs | February 21, 2010 at 3:05 pm

Sigh. Yeah. All my TTs (except for the Sport Junior) are older models. I loved the G2-80M. Damn them for discontinuing it. I’ve got a few Dogtras around and a largish collection of crap collars I’ve confiscated to use w shoto clients why they are torture devices.

And hey GF – I didn’t know you had a blog! I’ll be adding you to the blogroll directly.
11. stone soup diaries | February 21, 2010 at 3:23 pm

Thanks for the add. I had another one on google’s blogger but it got hit along the same time my website got hit, so I ported over to wordpress which I like infinitely better.

I still have my ancient 500 Pro XL to which all other collars are measured. I switched to Dogtra when I had to replace almost 20 of the G2 Sport 50’s.

I make the clients take a hammer to their Innotecs. I HATE those cheap A$$ things.

I love your blog btw, I am here almost every day to see what juicy stuff you have uncovered!
12. SmartDogs | February 21, 2010 at 3:34 pm

I wondered what happened to that one. I was going to add it, but it quit getting updated and I hate to post links unless they’re actively updated.

Yes, the old 500 Pros rock. Mine is still my go-to collar (and would double nicely as a bludgeon if needed).

And – thanks 😮
13. Viatecio | February 21, 2010 at 4:08 pm

I’ve never had a high opinion of Innotek (and it’s been even lower since I learned that they and PetSafe are basically the same company now). I’m a Dogtra person myself. My collar was bought used and even at the lowest setting, I can still feel it, so I just use the pager most of the time and save the stim for serious infractions and avoidance training, like to stay out of the street.

And then it IS negative reinforcement, because I will give multiple stims while guiding her back to the curb so she knows where the “safe” area is and how to avoid the stim next time.

I’d be interested to know more about your collection of torture devices. Are they newer models from sub-par companies or are they serious old-school one-stim-fits-all collars? I’ve considered buying a JASA Force collar just to have on hand so people can see that, hey, maybe the prong collar really IS a more humane option: even Koehler himself said that it was too extreme, and people call HIM inhumane!
14. SmartDogs | February 21, 2010 at 4:22 pm

I’ve got a mix of one to ten year old collars. A few Innoteks (I hate them too) as well as lots of cheap models that local stores sell. Would love to have one of the JASA’s (or old TT’s where you change points to change levels) for grins and one of the cheap TENS type units sold on late night teevee would be fun too, but I am not willing to actually pay for them.

FWIW – TriTronics models with momentary settings (not all have it) are a lot softer than Dogtras. And you are, apparently, a lot more sensitive than I am. I can’t feel a Dogtra until its set at at least 15 or so (out of 125). Below that, nada. I also have an old high-end PetSafe collar that is amazingly soft even at medium levels. It’s great to get people over the hump of the OMG ONOZ phase and to work with some very sensitive dogs.
15. stone soup diaries | February 21, 2010 at 5:14 pm

I had an opportunity to handle an old TT A1-90 where you have to change the contacts.

The lowest level feels like about a 35 or so a DogTra 3500.

Jasa’s are still used on dogs now, with upland gun dogs and occasionally retrievers. I have seen one or two. Today, most handlers will blunt the ends of the spikes with a hammer before they use them.

You can still buy them, as they were originally made, through hunting dog magazines. The TT A1-90’s I have seen for sale pretty cheap at http://www.collarclinic.com

That’s odd about the stim settings between the different collars. I can’t feel a DogTra until about 20 or so, but I can feel the stim of a TT at the 1/2 setting.

A DogTra at about 35 is pretty uncomfortable to me, but a TT Sport Basic can fry me on a 2.

Perspectives?

I used to have one of the original SD 1200’s and I really liked the smoothness of the stim. I just couldn’t keep the damn thing working and it didn’t have enough top end for the real hammerheads I used to get a lot of.

DogTra came out with a few new collars to replace the old 175 and the old 200. I have not seen them yet.
16. Viatecio | February 21, 2010 at 7:09 pm

I don’t think I’m that sensitive, I just think the rheostat is a bit wonky. I have to test the stim EVERY TIME I use it to make sure it’s low enough for my dog, and it is a sharper stim…but it’s all I have right now and she’s so tortured that she enjoys having it be put on. Too bad the pager thing is exclusive to Dogtra, because I’d like to have that function on one of the newer SD models; I’ve heard that stim is very smooth as well. I can’t afford to do any extraneous buying though (even refurbished, and I only have 1 dog anyway), so I work with what I have. I can take up to 40 on it.

I look forward to the day when I can experiment and play around with different collars and feel how each one has a different stim texture and quality, as well as how different dogs work with each one. I’m intrigued that the SD gave you so much trouble…that’s the brand SitMeansSit uses and they’d not use something that didn’t work to their specifications. No experience with TT or SD though, so can’t offer any perspectives on why you get fried…
17. SmartDogs | February 21, 2010 at 7:16 pm

The wonky low end rheostat issue is one reason I like TT over Dogtra for low-level work. With a lot of the Dogtras you also have to turn the transmitter all the way down to zero after you turn it up, or it doesn’t turn down properly. Wazzup with that?

RE: SportDog – Some of their higher end collars are nice units and some, as you note, have wonderfully smooth stimulation. But… the low end cheap ones sold in pet stores and hardware stores all over are complete and total crap. VERY noticeably delay between button and stimulation, irregular button action and all levels are very painfully high. I absolutely refuse to work with them.

I don’t know anything about the SMS collars. I was excommunicated some time ago.
18. stone soup diaries | February 21, 2010 at 7:58 pm

Viatecio how old is your DogTra? I have not had that problem and I go through a lot of them. I have a 280 (replacement phase model for the 200 and 175) for the little guys that come in and a 3500 that I use on virtually everything over thirty pounds or with an attitude. I have either used or sold all of the current models with the exception of the 300 series they just released.

I have not had much experience with Fred’s collar. I was under the impression that it was the same unit as the 400.

I thought it was too limited in it’s range, the batteries were worthless and the buttons were mashy.
19. Pooch Professor | February 21, 2010 at 9:44 pm

I have been using pressure to teach for a while, and it is quite effective. I use it a lot on the shelter dogs I train, as I don’t want to use training treats with them (any treat that is valuable enough for most of them is too rich/too messy/hard to store). Plus, the shelter dogs can get very bully-ish around the volunteers when they know treats are neaby, even biscuits.

Pressure is easier to use, and frankly, I want to send home dogs who are comfortable being touched. People are going to touch the dog to make it sit; the least I can do is teach the dog that pressure near his hips, or pressure upward on the leash, means “sit.” The better he gets, the less pressure is needed. Then, if he is distracted when the owner says sit, and they just barely touch him, he will respond (as opposed to the dog who has never been pressured who whips around and mouths a hand on his butt).

The application of pressure becomes a reminder. It’s not forcing the dog into position; he still has a choice. But sitting is more comfortable.

I also like it because the concept of pressure is natural to dogs. When you use direct pressure, dogs need to learn to move away from it, not into it, and learning this can make their lives better in so many ways.
20. Viatecio | February 22, 2010 at 6:52 am

I bought the collar in July 2005 from a friend of a friend in Germany, and I don’t know how long she’d had it…maybe a year or so. She was upgrading to the two-dog model and didn’t need the one-dog 175NCP anymore. You’re welcome to have a look at it if you’re up for it, it’s not a primary collar and isn’t used too often.

Pooch Professor, that’s amazing you’re able to work with shelter dogs like that, and kudos. The one here in town where I volunteered for a short time had SO MANY volunteers (never a bad thing in itself(, but everyone was so inconsistent about what was allowed and what wasn’t, not to mention that, like most humane societies, it was positive-only or else. It’s the same here in the school kennels, everyone is pretty inconsistent and while most of the dogs leave knowing how to do some basic tricks, their command of obedience is nothing like what you’d expect after approx 2 years of work. (They’re great in vet’s offices though, after getting a lifetime of exams in that period!) We have the same problem with treats too, so I choose to not use them. I will not go into our choice of tools, but suffice to say that a lot of the TYPE of pressure exerted on the dogs is not something I would voluntarily choose to use, even if it IS appropriate negative reinforcement.
21. stone soup diaries | February 22, 2010 at 7:25 pm

Viatecio,
I do know that there was a real problem with the 175 and 200 models taking a huge jump from just below 20 and higher.

The sensitivity has since been addressed somewhat with the newer models from about 2007 and onwards. The 280 NCP Platinum is a rheostat dial with a LCD display and was originally designed to replace the 200 as the pet and obedience collar. The 280’s have levels from 0 to 127 like the higher end ‘pro’ collars and there is not the substantial jump from around 20 like the older models. I liked the 175 for small dogs and for owners who had larger breeds who wanted a smaller collar for their dog, but really disliked the stim level jump. The 280’s seem to have resolved that, plus they can be operated with one hand and you can actually ‘see’ where the levels are at.

Anyway. The 175 is a pretty decent collar. I understand what you are saying about the stim levels. It’s still a great tool for a good portion of dogs. My only issue with it is the range is pretty limited and my eyesight is not great, so the LCD display really makes a difference for me.
22. Jacob L'Etoile | February 24, 2010 at 12:49 pm

I am interested in the author you mentioned. The amazon link you provided only showed a list od used books. I am interested in presure and negative reenforcement as it may apply to training wild hawks. Do you have any sugestions as to what book may be best?
23. SmartDogs | February 24, 2010 at 1:07 pm

Heini Hediger – the ones on my shelf (and a couple of my favorites) are:

The Psychology and Behavior of Animals in Zoos and Circuses; and
Wild Animals in Captivity. Both are long out of print but usually available online as inexpensive used copies.

Another good resource is Pat Parelli’s video on The Seven Games. You can usually find an inexpensive copy on ebay.

Let me know if you blog about this.
24. Jacob L'Etoile | February 25, 2010 at 2:54 pm

Thank you for responding. I don’t blog much anymore, I discovered I really don’t like to write. I will drop a comment on this blog after I read the books though. I am interested in the idea of space and how to negotiate through it. I am a falconer and I prefer to train wild birds, as opposed to captive reared ones. One of the problems I have is aproching a freshly traped bird without stressing it. I don’t like the idea that we start off our first couple of training sessions on the wrong foot as it were. Thanks again