Good Morning Everyone,
I hope everyone had a great weekend. My discussion here is about the four quadrants of operant training. Everyone clearly can grasp the difference from Positive Reinforcement and Positive Punishment (correction). I for one, and I know I’m not alone, always had a hard time grasping the difference between Negative Reinforcement & Negative Punishment. So I starting using examples using both Humans (Two-Legged) and Dogs (Four-Legged) to further describe the differences and offer more clarity.
This is a prelude to my next blog about the Marker/Clicker Commands that I found most useful. FYI, I did not come up with the term “Operant Conditioning.” I am not here to bore you with the science, but I must mention two people. Edward Thorndike was the first extensively studied this topic. However, B.F. Skinner is ultimately referred to as the “Father” of Operant Conditioning. The information regarding the four quadrants discussed here, were obtain from his book published in 1938 titled “The Behavior of Organisms: The Experimental Analysis.”
In order to discuss the Marker/Clicker Commands (next blog), the quadrants of Operant Conditioning will need to be discussed first.
The first quadrant of Operant Conditioning is Positive Reinforcement. If a dog sits, he is given a treat as a form of reinforcement to ensure the behavior (dog sitting) will be MORE LIKELY to occur in the future.
Two-Legged Example: Growing up playing Little League Baseball, every time we won a game, my father would treat us all IMMEDIATELY after the game to ice cream cones at Dairy Queen. We quickly learned when work harder and win, we will be reinfored with ice cream.
The second quadrant of Operant Conditioning is Negative Reinforcement, which means we remove the aversive pressure from the dog. Therefore, we are going to pull on the dog's collar until the dog sits. Once the dog sits, the pressure is removed. By removing or TAKING AWAY the pressure from the dog’s neck as soon as the dog sits, he is MORE LIKELY to sit in the future.
Two-Legged Example: The child that is grounded before the new school year ever begins. The grounding is promised to be lifted once the child proves they can do well and produce good grades.
The third quadrant of Operant Conditioning is Negative Punishment. This is sometimes refer to a “Negative Reward Marker.” It communicates to the dog the behavior the dog just performed is incorrect and to try again. No form of pressure is used.
Two-Legged Example: When I was in high school, my Aunt Robin would always bake for our soccer team and bring the cookies to every soccer game. We knew if we won the soccer game and performed correctly, we would immediately get to eat the cookies. If we lost the game, and performed incorrectly, my Aunt Robin would say “next time,” and hold on to the cookies until next game. (I personally think she let the other parents eat the cookies after the game.)
The fourth quadrant of Operant Conditioning is Positive Punishment and simply put this is a correction. It is an aversive consequence of non-behavior or because of performing a behavior incorrectly so that the incorrect behavior is less likely to occur in the future. Re-read that last sentence please. Positive Punishment makes it less likely the INCORRECT behavior to occur in the future. Does this help the dog realize what the CORRECT or desired behavior is we want?
Remember this, effective Positive Punishment has two parts. First, the correction, or aversive consequence, should not be given until after the dog or human has demonstrated a clear understanding of the what is being asked of them. The second part is the timing of the correction. Timing is so critical for effectiveness when you apply both Positive Reinforcement and Positive Punishment.
Two Legged Example of the First Requirement: Liam is a two year old boy that knows how to count to ten on his fingers. Liam can add to any number up to ten because he can use his fingers to help himself visually. Every time Liam gets a math equation right, his parents give him an M&M®. Every time Liam gets a math question wrong, his parents give him a slap on top of hand. If Liam’s parents were to ask him what nine times ten equals. Liam would have no idea what the answer is because he has never been taught multiplication yet. By slapping him on top of his hand, and using Positive Punishment, it will not reduce the likelihood of him getting it wrong again in the future. In fact, by slapping him, he could become unmotivated to try the next time his parents ask him a math question. His parents cannot expect Liam to know the answer if he has never learned the skill. The same concept goes for a dog. You cannot expect the dog perform a behavior or skill either unless we take the time to teach the skill first.
Four Legged Example of the Second Requirement: You tell your dog to “DOWN” and you walk over to grab your cup of coffee. As you pick the cup up, your dog gets up and walks over to you. You tell him “No,” and when he gets to you, you give him a correction for getting up and you walk him back over to have him lay back down. You may even do this one or two more times, and each time your voice may get louder and the correction more stern.
Was the correction the dog received as a result of getting up? Was the correction for coming to you? From the dog’s point of view, he may think the correction is for coming to you. This is just a minor example why it is hard to apply corrections in a timely manner. When in reality, the distance between the dog and the coffee was just one foot over threshold. If we had stopped and reinforced the dog one or two feet closer, the dog may have stayed in a “DOWN,” and we would only be using positive reinforcement.
Two-Legged Example: Let us go back to the example we used with the little boy Liam that was learning math. Liam is now learning to count to twenty, and he knows the sum of six plus six. He has his addition and subtraction up to fifteen down pat. Mom asks him what ten plus seven equals? Liam is not sure and he guesses sixteen. We know this is wrong and his mom knows that he is not sure of his answer. Does she slap him on the top of the hand OR say, “Almost, try again. You are so close,” and withhold the M&M® until he answers correctly. By saying, “Almost, try again. You are so close,” Liam’s mom is using a form of Negative Punishment.
Four-Legged Example: Now let us apply this to the “PAW” and “SHAKE” game. Since, we humans shake with our right hand I teach dogs that “SHAKE” means give me their right paw, and “PAW” means give me their left paw. Liam thinking about the answer to his math question maybe the equivalent to a dog trying to learn “PAW” and “SHAKE.” If Fido offers me his left paw after I say, “SHAKE,” do I correct him? Do I use a Negative Reinforcement and withhold the reward until he completes the desired behavior. When Liam’s mom tells him, he almost has it and to try again, she is doing just that. Let us say Fido offers the wrong paw again, would he be corrected at this point? If a correction is given, what does Fido learn from this correction? For trying, for getting it wrong twice, for looking at us because he was not sure what we were asking of him?
While the timing of the mark with Positive Reinforcement is important, it is even more important for Positive Punishment. A missed mark with Positive Reinforcement will not affect the motivation of the dog. The dogs are still motivated to work. If we miss the mark on Positive Punishment (a correction), we are possibly killing their motivation for the wrong reason. We asked for a “PAW” and Fido sat and looked up at us. We just corrected a dog for looking at us, and that certainly is not going to help strengthen the trust and bond we have together. This is a big reason why Positive Punishment should just be avoided all together; timing is too critical.
Thank you all for reading this! I really hope you found it extremely educational and I encourage you to share with all your colleagues. A special thank you goes out to Demaree Clay of “K9U (and felines too).” She has been so helpful and instrumental in so many ways!!!
Disclosure: I am not an English major so grammar is not my best attribute. Please email me if you find any grammatical errors. Please email me if you have any comments, concerns, or questions. I can be reached at firstname.lastname@example.org