Against my better judgement, I've made a comment on a Boston Globe article about the Dog Whisperer controversy. There are two particular things that keep on grabbing my attention in the discourse. First is that the idea that a punishment is a superior mechanism of behavior change in comparison to positive reinforcement. I wrote about this in a previous blog post. The second relates to how the public discourse has moved from excoriating "academic elitists" toward a general disdain for knowledge derived from books and school rather than life experiences.
I realize this sounds like a big topic for a little puppy to blog about. Jason put me up to this.
In the Globe article Mr. Millan was quoted as saying "my school was animals, not books." This is such an unnecessary polarization. Both academic knowledge (book learning) and real life experience are necessary. Either in isolation are useless.
A great body of academic literature relating to operant conditioning exists. Here is what it says:
Does punishment work? Yes, when presented without delay; when consistent; when limited in duration and intensity; when the consequence is logical; limited to the specific situation at hand; when no mixed messages are sent; and when negative punishment is used.
This last part is important. Pay attention here. A negative punishment is the removal of an attractive stimulus after a response. An example of this to try at home with a puppy? Puppy is playing tug, gets too excited, and nips your hand. A negative punishment would be removing the toy and stopping play with the puppy for a period of time. A positive punishment would be the application of an aversive (unpleasant) stimulus. Back to the same example, puppy gets rambunctious when playing tug and nips the hand. Owner swats dog on the nose. Shock collars are another example of positive punishments. Dog leaves the yard and an electric shock is administered by a collar attached to its neck.
What are the risks of punishment? Here are a few: effectiveness of the punishment usually disappears when threat of punishment is removed; rewards can override the punishment; punishments can trigger escape or aggression; teaches that aggression is a legitimate way to influence others; can inhibit learning better alternative responses; is often applied in an unequal fashion.
B.F. Skinner, known as the father of operant conditioning, wrote that people "work harder and learn more quickly when rewarded for doing something right rather than when punished for doing something wrong." Where did he learn this from? Experience in the lab with animals and people.
What has your experience been--whether with animals or humans? Do punishments work? Are rewards more effective tools for shaping behaviors? What does real life experience show you? Feel free to post your experiences here.
Dr. Mihalko,
ReplyDeleteI, too, commented on that Boston Globe article. Unlike you, I am very familiar with César Millan's actual methods from having watched every episode of Dog Whisperer. You are correct in your criticisms of punishment used to influence a dog's behavior. However, Millan doesn't use punishment. He uses "corrections" ranging from a spoken "sh" sound (similar to the use of a clicker) through sudden but harmless touching with the hand or foot. He only uses these corrections when a dog becomes excited and fixated, immediately prior to "red zone" behavior such as barking, biting, attacking other dogs, and so forth. While all of your criticisms are correct, they do not actually apply to what Millan actually does. I urge your readers to take a look at Millan's website to see what he actually does, believes, and teaches. Thanks for letting me post this.
David, I appreciate your response--both in its thoughtfulness as well as in its civility. A huge difference from the tone that many comments seem to be relating to the article on boston.com. Just a few points of clarification. I am actually quite familiar with Millan's techniques. Many of those techniques, including the two you mentioned, are punishments. "Sudden but harmless touching with the hand or foot" as well as a spoken "sh" sound are both examples of a mild positive punishment--a mild aversive sensation is delivered after behavior to reduce the likelihood of the behavior occurring again.
ReplyDeleteMy comment in the article had more to do with the increasing public discourse against academic knowledge and the preference for real life experience (Millan's quote of his learning coming from real life experience, not books). I think both are necessary because together they create the possibilities that either in isolation cannot create.
Clicker training, by the way, is not at all like a spoken "sh" sound. The technique you mentioned that Millan uses--which is one I use--is mildly annoying to my dog and redirects her attention. On the other hand, a clicker is used to mark a desired behavior. Further, the clicker is always initially paired with a reward (treat, play, pet, excited speech). A click is something a dog wants to hear--it's becomes a reward.