Dopamine: From Prediction to Error

Today, if you open a textbook to a chapter about dopamine, you can expect to read that dopamine acts as a specific type of learning signal: a temporal difference reward prediction error. The fact that this idea offered a simple and principled explanation for a wide range of experimental observations at the time it was introduced, and the fact that its predictions continue to be refined and tested to this day, have made it one of the most successful marriages of theory and experiment in all of neuroscience.

But textbooks change. Before P. Read Montague and Peter Dayan introduced the reward prediction error hypothesis in the 1990s, dopamine was seen as a facilitator of movement and reward chemical. In the years since, while many predictions of the theory have been borne out, some new observations do not seem to fit cleanly.

To learn where the reward prediction error theory came from, how it has evolved, and where it might be heading, drag the ⬤ below to explore a timeline of major milestones. Watch as the dopamine field's centre of gravity, represented by the red dot ⬤ on the chart, swings back and forth between the contributions of theorists and experimentalists, and between learning- and behaviour-centric views of dopamine function. Click on the headings to the right of the chart for a summary of each research contribution.