Reward Based Training: Understanding Intermittent Reinforcement

Ok, we are going to get a little technical as we dive into the next piece in our series on Reward Based Training (RBT)! “Intermittent Reinforcement” is a term coined by American psychologist and behaviorist, B. F. Skinner, in his studies on Operant Conditioning and Behaviorism. The shortest definition of intermittent reinforcement is that the reward (reinforcement) is only given part of the time that the subject gives the desired response. Although we will shortly get to how this applies with dog training, let’s give a human example to help explain this concept in layman’s terms.

People find gambling to be an addicting behavior. Why? It’s because of “intermittent reinforcement!” Let’s use slot machines to demonstrate this. If people knew that they would put $1 into a slot machine and receive 75 cents back every time, this game would quickly cause people to lose interest. However, because you put in the $1 knowing that most of the time you will not get anything back, sometimes you might make your $1 back, and rarely you will hit a jackpot, this makes the game exciting, fun, and something that you want to do over and over! This is “intermittent reinforcement” in action.

In terms of dog training, intermittent reinforcement is very important for RBT. Many who have not used and do not fully understand RBT worry that using food will cause a dog to become dependent on the food and that the dog will not work unless food is present. This is true if you reward every behavior and don’t properly move to intermittent rewarding in a timely fashion. With RBT, we use a “continuous reinforcement schedule” when first teaching a behavior, meaning we will reward for every correct response. However, once the dog has learned a behavior and the cue, we move to an “intermittent reinforcement schedule.”

What’s important is making sure you use a “variable ratio reinforcement schedule” when using intermittent reinforcement. The word “ratio” means that you have an average number of times you will reward a behavior – for example, 5 out of 20 correct responses. The term “variable” means that the reinforcement will come at unpredictable times. So, if asking for 20 “sits,” with the goal of rewarding 5 times, you will give at more random intervals. Instead of giving on the 5^th, 10^th, 15^th, and 20^th “sits,” maybe you will give on the 3^rd, 10^th, 16^th, and 19^th “sits.” If you reward on the same 5^th “sit,” every time, your dog will quickly pick up the pattern and the dog will know when to expect a reward, moving back to almost a “continuous reinforcement schedule.”

The key is to keep training like a “game” for your pup…make it as exciting as those slot machines are for humans! Next time, we will discuss how to go from “continuous reinforcement” to “intermittent reinforcement” for your pups, with real examples, common questions, etc!