23d • Predation Tidbits
Reward Prediction Error - Wolfram Schultz
One way we can influence our dog’s and their dopamine levels with reward is something called the reward prediction error (RPE). This can be described in short as the difference between the expected reward and reward that is actually received. When a dog receives an expected reward the prediction error is zero (no change in behavior), when a dog receives a reward better than expected that is a positive RPE (increases behavior) and when a dog receives a reward worse than expected that is a negative RPE (decreases behavior).
The way that RPE can be applied to dog training is massive and gives you a great tool of communication to increase or decrease behaviors. This is where variable rewards, especially for more complex behaviors, can be of great benefit to us in communication to our dogs to let them know they did good and that they did great. This is hopefully something that we do consciously but can also be done unintentionally so it is important to remember that our dogs are always learning both inside and outside the home.
Reward prediction error is not limited to dog training, and it is something that all mammals experience and is how we make knowledgeable decisions. Dopamine neurons that are released both during the reward stimulus such as a marker word or clicker and during real rewards are similar and are individual to your dog. This is further reinforced with a reward the dog receives that is equal to or greater than expected for a desirable behavior, but timing is also very important and bad timing can taint your high value reinforcer. This is known as temporal discounting and while dopamine will still be produced it will be less and therefore your reinforcement weaker.
9
5 comments
Reward Prediction Error - Wolfram Schultz
powered by
skool.com/beyond-the-leash-9053
Go beyond obedience! Build engagement, reliability, and a deeper bond through canine psychology and proven force free methods.
Suggested communities
Powered by