An important idea in learning is the notion that dopamine signalling from the midbrain represents a prediction error that can guide associative learning. This idea has been around for a while, but evidence so far has been correlational rather than causal. Moreover, some data has suggested that dopamine represents reward and/or reward saliency rather than prediction error. A new report from Steinberg et al provides some causal evidence supporting the idea that dopamine release from the midbrain really does signal prediction error (23708143). Specifically, the lab used optocogenetics to enable light-activated midbrain dopamine release. They then used activation to overcome the classic effect of blocking. The primary paper is above as well as a good summary (23799468).