It’s your first day playing tennis. After many misses, you’ve just hit a serve in for the first time! I have three questions about what happens next. How are you able to learn which movements led to the outcome (the credit assignment problem), how do you do it again (the re-entrance problem), and how do you improve (the refinement problem)? I am interested in how your brain solves these problems by reinforcing its own neural activity which controls movement. To study this, we model how reinforcement acts on “neural population dynamics”: rules which dictate how neural activity transitions from one time point to the next. We test these models experimentally using novel brain-machine interfaces and targeted stimulation of neuronal ensembles. Ultimately, we hope to discover how reinforcement enables the brain to re-enter target neural activity patterns, and thus to learn.