How Humans and Dogs Can Learn -- and Suffer -- From the Use of Negative Reinforcement

In this Nov. 26, 2012 photo, Dill, a veteran assistance dog in training, looks on as inmate John Barba walks away after comma
In this Nov. 26, 2012 photo, Dill, a veteran assistance dog in training, looks on as inmate John Barba walks away after commanding him to sit and stay during a demonstration at Western Correctional Institution in Cresaptown, Md. Dill is one of three dogs assigned since September to inmates at the maximum-security prison for basic training as service dogs for disabled military veterans. The inmates, who are also veterans, are among the state's first prisoners to join a national trend of training service dogs in correctional institutions. Professional trainers say prison-raised dogs tend to graduate sooner and at higher rates than those raised traditionally in foster homes because puppies respond well to the consistency and rigid schedules of prison life. (AP Photo/Patrick Semansky)

Every time I read about water-boarding, I cringe. Because I'm a dog trainer, I can't help but think about how similar this reviled torture method is to some of the things we do to dogs.

I've written a good deal on my site, TheDogs, about some of the problems involved with using shock collars in animal training. But mostly, I've discussed the use of shock collars to stop behaviors -- behaviors such as running out of the yard, or barking, or going in the wrong direction while assisting a hunter. In these instances, the dog usually receives a quick but detrimental shock to the neck when he is out of bounds, or barking, or going in the wrong direction. If there are multiple shocks, they come one at a time. Fun stuff. The goal, in these situations, is to stop behavior. For this, shock collars may work... but the fallout is often not worth the benefit.

But there's an even worse application of this form of training: Using the shock to elicit desired behaviors. In this kind of training, the shock does not cease until the dog does the correct thing. For instance, I once met a poodle whose owner had taught him to go to his dog bed by giving him shocks whenever he was anywhere in the room that wasn't on his bed. The bed was the safe zone, and it was as if the rest of the room were filled with swarming snakes. The dog learned the task. But I wouldn't say it was a happy home.

All behavior -- including every behavior in which you are engaged right this moment! -- has a consequence that falls into one of the operant conditioning quadrants: positive reinforcement, negative reinforcement, positive punishment and negative punishment. Technically this kind of shock collar training is not considered punishment -- it's negative reinforcement, "negative" because the presence of the desired behavior (sitting on the bed) takes away the bad thing (pain), which encourages the likelihood that the behavior will happen again in the future. (Eight most important words to remember when training a dog? Reinforced behaviors are more likely to happen again). Guilt and nagging are also negative reinforcement cohorts: If the behavior (say, calling your grandmother) results in the taking away of the annoying thing (your mother texting you hourly to tell you to call your grandmother), then I may be more inclined to do the behavior in the future. You might not enjoy calling granny, but the 10 minute conversation is less annoying than the constant notifications on your phone from mom mom mom mom mom mom.

Lili Chin shows how her dog Boogie's recall behavior could be affected by punishment and reinforcement.

Another example of negative reinforcement is water-boarding, the torture procedure that involves holding a wet cloth over someone's mouth until they beg for mercy. Apparently, the experience feels a lot like drowning.  Here is the late Christopher Hitchens going through the water-boarding procedure for Vanity Fair magazine:

The desired behavior (telling the persecutors what they want to hear) will result in the taking away (hence, negative) of something bad (the feeling that one is drowning).  It works. And every time it works, the torturers are reinforced. They got what they wanted. So they do it again! Because...reinforced behaviors are more likely to happen again.

But there's usually an unintended result to any kind of negative reinforcement: When all is said and done, your subject is probably not going to be to keen on you. The detainees are never going to buy their torturers a beer. Even if it isn't clear to your subject that you're specifically the one determining the presence or absence of the bad thing, he is still probably going to associate the bad stuff with you in some way, just because you are a part of the picture when it's happening. I guess that isn't such a big deal if you're in the CIA and you're using negative reinforcement against the Taliban. If you are waterboarding someone, you're most likely torturing someone you probably thinks should die anyway. Either that, or you're a terrible big brother. But dogs are, in theory, creatures we choose to have around because they love us. And we love them. So why the hell are we shocking them at all?

Of course, most of the collars we're talking about operate via remote, so you need not be that close to your dog feels the shock. But I don't think that will necessarily help him feel good about you. He might not directly associate your presence with the pain, but he may trust you less in any case. After all, it's because of you that he's at the darn training facility to begin with. Next time you try to get him in the car to go somewhere, he might not feel like getting in there with you.

In an excellent post on the problems of shock collars by Eileen of the site Eileen and Dogs, we see the process of using electric shocks in basic training. Fortunately, we're spared from seeing an actual dog try to maneuver this stressful situation. But the stuffed animal stand-in does a pretty good job. Someone get that dog an agent!

If you can stomach it, Eileen has a host of other videos on her site -- videos in which we can see the process as used on real dogs (some are promotional clips from Sit Means Sit, a training chain that is big on this type of training). Negative reinforcement can be effective in getting desired behaviors. These videos show that. But they also show countless dog body language positions that I'd say translate into something like:  Oh my god what the hell is going on! Make it stop!

So how do you get a dog to do something like stand on a turned-over bin without using a shock collar? So glad you asked! Here, Eileen shows us how to use the kind of smart positive reinforcement training I teach to my dog training students in order to manipulate behaviors quickly, easily and painlessly. This time, the dog she uses is real, and that wagging butt translates roughly into this: I love training, I love my owner, and I love standing on bins.