AI teaches neuroscience about reward signalling.
Computer scientists have successfully developed reinforcement learning algorithms that present outcomes and rewards along a probability distribution. Previously, such algorithms simply presented an average outcome and associated reward. A spread of possible outcomes is of course truer to life. Neuroscientists have taken a leaf out of the AI book and discovered that the brain works in a similar fashion. Thus, the neurons in the dopaminergic reward prediction error network are all different. They too react to outcomes along a distribution curve.
Link to paper: https://www.nature.com/articles/s41586-019-1924-6
You might also like to browse other neuroscience papers: https://www.thesentientrobot.com/category/neuroscience/neuroscience-academic-papers/
