How does the brain learn to predict rewards? In this issue of Nature Neuroscience, Qian, Burrell et al. show that understanding how dopamine guides learning requires knowledge of how animals interpret tasks — what they believe is happening and when. By carefully manipulating cue–reward contingencies, the authors show that dopamine responses track belief-state reward prediction errors. These findings reaffirm — against recent challenges — that mesolimbic dopamine neurons signal prediction errors in line with the temporal difference learning rule, a core algorithm that bridges neuroscience and artificial intelligence.
- Eleonora Bano
- Steven Ryu
- Adam Kepecs