The Boston Globe featured a great article promoting positive reinforcement-based training. This method, although widely used by animal trainers in zoos, trainers of working and performance animals, and pet dog trainers, receives its fair share of critics who are die-hard, card-carrying practitioners of methods originally espoused by trainers like William Koehler.
Today’s Koehler-advocates argue that punishment-based training “has always worked.” Well, if it did, why is it that there are so many dogs with serious issues, and I’m not taking about dogs that are born with personality disorders–such as fear aggression– or whose brains are simply not wired correctly, but have never been exposed to punishment based training. That’s the subject of a whole ‘nother blog article.
I’m talking here about dogs that have had their choke collars popped, or been hung by their collars everytime they see a human or another animal nearby, and are now terribly dog or people-aggressive. I own one such dog, my rottweiler Busta. When I started out training, I used these same harsh methods because I didn’t know any better. My method of teaching Busta to ignore other dogs was to yank his prong collar and yell “leave it” everytime a dog appeared on the horizon. Well guess what, Busta associates other dogs with pain, so he goes over the top when he sees them. He’s terrible on walks; he’s terrible in the car. I don’t carry him in the car much because he’ll destroy the backseat to get to another dog.
The Boston Globe makes some interesting points, and a few that need clarifying. First of all, postive reinforcement does not mean permissive. It is not acceptable for a dog to jump on people, go potty in the house, chew up furnishings, or bark for hours on end. Trainers do not simply ignore unwanted behaviors. We, instead, do one of two things: distract the dog so it will discontinue the behavior, or ask it to offer another behavior that are incompatible with the unwanted one.
In a training situation, where I’m training specific behaviors, like heeling. I reward the dog with a click and treat when it’s walking on my left side, shoulder in line with my knee. The clicks and treats cease once the dog forges ahead, lags behind, or decides to walk on my right side. Sometimes I’ll call the dog back, but very often the dog realizes that I’m no longer interacting with it and will return on its own accord. Pretty soon the dog gets the idea: “heel” means walk on Mom’s left side, because really good things happen there.
Positive reinforcement doesn’t not mean that you have to spend a really long time teaching an animal a behavior. I found that to be true with punishment-based methods. If you are teaching a behavior for the first time, like sit, or chase your tail, or lick your lips, and you have the right approach, the dog will learn these behaviors very quickly. My puppies learn new behaviors in a few days.
What takes long is changing a dog’s emotion to some stimulus. If you have an over-the-top aggressive dog, yes, a positive reinforcement approach to behavior modification will take a long time. Why? Simply because you are working against a behavior that the dog has learned over a period of time, and has been rewarding to him in the past. An aggressive dog has learned that snarling and lunging at people will make them back off. The perceived threat has gone, the dog is rewarded, so the dog adds this behavior to its repertoire.
Think of human behavior. A child has a really bad experience during a medical procedure; every subsequent visit to the doctor elicits behavior that makes carrying out the procedure difficult or impossible. Positive reinforcement has to work against the child’s emotional reaction to the procedure, and it’s subsequent behavior, and that is very hard to change.
According to Dr. Nicholas Dodman, Cesar Millan’s approach to dog training creates a “…quick fix that will unravel.” When a dog experiences punishment for lunging/snarling at humans, it learns to avoid the punishment by not engaging in the behavior in the first place. The dog no longer reacts. Unfortunately, though, the dog’s emotions remain the same: it still feels as though humans are a threat.
Now that’s dangerous. To the unsuspecting human, the dog appears “cured” of its aggression. The dog tolerates people walking by it, or being petted. Then one day, when it’s owners feel that those aggression days are long behind them, a veterinarian leans over the dog during an examination, or a child hugs the dog. Disaster strikes. The dog looses it, and savages the vet or the child. The dog is labeled “dangerous” and is then euthanized, simply because fear got the better of it and it just couldn’t think of another appropriate behavior.
What gives positive reinforcement an edge over punishment-based training is that it is based upon science. Clicker training, as the Globe article points out, developed out of B.F. Skinner’s work on operant conditioning. It is applicable to all species of animals–from fish to humans–because it is based upon the science of learning. True, all clicker trainers sometimes have to go on gut instinct when working with a dog, because like humans, not all dogs learn in the same way. But their decisions made on the gut level have the science of learning theories behind it. That, to me, is enough to make me try this method on both my canine and human students.
To read the article, click here.