Jump by Dr. Miles Arakaki
When I was a kid, our family dog was a retriever mix named, Dee. Dee never received any training and was pretty rambunctious as a year-old adolescent. She had no manners, and one of her annoying behaviors was jumping up on people in greeting. I happened to be flipping through a random magazine one day, and noticed an article on dog training, with a specific section on correcting unwanted jumping. This particular article recommended correcting the behavior by bending down and petting the dog on the head before she jumped. A little intrigued, I put Dee outside, then walked through the door and pet her on the head before she could jump. I repeated the drill for 10 minutes, after which I lost interest until the next day, when I ran another 10-minute session.
Lo and behold, after that second training session, Dee stopped jumping, and not just on me, but also everyone else. The unwanted behavior had been extinguished. The magazine article did not go into detail as to the reasoning behind the technique, and I never thought much about it until decades later, when I began to get more seriously into dog training.
Today, I am a big believer in understanding current scientific behavioral theory prior to implementing a training plan. Each dog is an individual with specific personality traits. There is no one standard training program that will work with all individuals. When a trainer does not understand behavioral theory, he does not have the flexibility to adjust the plan to meet the unique needs of the individual. Ineffective training techniques are repeatedly forced on the dog, confusion and frustration smolder, unintended unwanted behaviors develop, and ultimately the plan fails to meet expectations.
When we train dogs, and other animals (including people), we are essentially manipulating and shaping behaviors. Modern training is based on the theory of operant conditioning. Although related foundational concepts date back to Edward Thorndike’s 1898 Law of Effect principle, the ‘operant’ terminology was first presented in 1938 by American psychologist, B.F. Skinner. In operant conditioning, behavior is modified through reinforcements and punishments (positive and negative), and extinction. The terminology itself is confusing, and understanding it is the first hurdle towards developing a rational training program.
A reinforcement is anything that increases the probability of a behavior in response to a stimulus. A punishment is anything that decreases the probability of a behavior in response to a stimulus. ‘Positive’ refers to the application of a reinforcement or punishment, and ‘negative’ involves the removal of a reinforcement or punishment. People first get confused in associating the terms ‘positive’ and ‘reinforcement’ with something good or desirable, and ‘negative’ or ‘punishment’ with something bad or unpleasant.
When Dee jumped up to greet me, I would push her down and away. In my mind, I was trying to discourage her jumping, but to Dee, the physical interaction was a desirable response. My attempts to discourage her were actually positively reinforcing the bad behavior. By leaning down and petting her before she started jumping, I was applying a response that was decreasing the undesirable behavior; i.e. positive punishment. The terminology has nothing to do with goodness or unpleasantness. People who don’t recognize this concept, induce bias into their training which can affect the outcome.
When I train a dog to sit by pressing down on her butt, removal of that pressure upon execution of the task increases the probability she will respond correctly to the command in the future; i.e. negative reinforcement. When the dog sits, I pet her and praise her. If she becomes excited and starts to break from the sit, I immediately stop the praise, and that decreases the probability that she will break the sit in future; i.e. negative punishment. Extinction involves neither reinforcement nor punishment to decrease a certain behavior (see earlier post on Walter).
There is more than one way to skin a cat. When a trainer uses a technique I am unfamiliar with, I try to categorize it within the operant conditioning reinforcement/punishment criteria. I also try to judge whether the application is appropriate based on the individual dog, stage of learning, level of distraction, and the desired end-state. If it all makes sense, then cool; I’ve learned something new, and can add it to the toolbox. If not, the cynic engine begins humming, and I start anticipating problems coming further down the road. But it all starts with sitting down and wrapping one’s head around the meaning of operant conditioning, and its application in actual training.
Lo and behold, after that second training session, Dee stopped jumping, and not just on me, but also everyone else. The unwanted behavior had been extinguished. The magazine article did not go into detail as to the reasoning behind the technique, and I never thought much about it until decades later, when I began to get more seriously into dog training.
Today, I am a big believer in understanding current scientific behavioral theory prior to implementing a training plan. Each dog is an individual with specific personality traits. There is no one standard training program that will work with all individuals. When a trainer does not understand behavioral theory, he does not have the flexibility to adjust the plan to meet the unique needs of the individual. Ineffective training techniques are repeatedly forced on the dog, confusion and frustration smolder, unintended unwanted behaviors develop, and ultimately the plan fails to meet expectations.
When we train dogs, and other animals (including people), we are essentially manipulating and shaping behaviors. Modern training is based on the theory of operant conditioning. Although related foundational concepts date back to Edward Thorndike’s 1898 Law of Effect principle, the ‘operant’ terminology was first presented in 1938 by American psychologist, B.F. Skinner. In operant conditioning, behavior is modified through reinforcements and punishments (positive and negative), and extinction. The terminology itself is confusing, and understanding it is the first hurdle towards developing a rational training program.
A reinforcement is anything that increases the probability of a behavior in response to a stimulus. A punishment is anything that decreases the probability of a behavior in response to a stimulus. ‘Positive’ refers to the application of a reinforcement or punishment, and ‘negative’ involves the removal of a reinforcement or punishment. People first get confused in associating the terms ‘positive’ and ‘reinforcement’ with something good or desirable, and ‘negative’ or ‘punishment’ with something bad or unpleasant.
When Dee jumped up to greet me, I would push her down and away. In my mind, I was trying to discourage her jumping, but to Dee, the physical interaction was a desirable response. My attempts to discourage her were actually positively reinforcing the bad behavior. By leaning down and petting her before she started jumping, I was applying a response that was decreasing the undesirable behavior; i.e. positive punishment. The terminology has nothing to do with goodness or unpleasantness. People who don’t recognize this concept, induce bias into their training which can affect the outcome.
When I train a dog to sit by pressing down on her butt, removal of that pressure upon execution of the task increases the probability she will respond correctly to the command in the future; i.e. negative reinforcement. When the dog sits, I pet her and praise her. If she becomes excited and starts to break from the sit, I immediately stop the praise, and that decreases the probability that she will break the sit in future; i.e. negative punishment. Extinction involves neither reinforcement nor punishment to decrease a certain behavior (see earlier post on Walter).
There is more than one way to skin a cat. When a trainer uses a technique I am unfamiliar with, I try to categorize it within the operant conditioning reinforcement/punishment criteria. I also try to judge whether the application is appropriate based on the individual dog, stage of learning, level of distraction, and the desired end-state. If it all makes sense, then cool; I’ve learned something new, and can add it to the toolbox. If not, the cynic engine begins humming, and I start anticipating problems coming further down the road. But it all starts with sitting down and wrapping one’s head around the meaning of operant conditioning, and its application in actual training.
