Reward Prediction Error in Machine Learning

Rich Sutton's liberal difference learning is recognized as the algorithm that the brain is using in reinforcement learning. Beauregard Smith: As we understand a little bit more about what works in machine learning, we go back to the brain and look for it. The principle is that you have to have information about error somewhere in your system and you have to get it to the right place at the right time. That's the principle now that we're working with in neuroscience.