Reading focus
Syllabus: How Do Biological Systems Shape Motivation and Emotion?
Compare the response to an unexpected reward with the response to a predicted reward and an omission.
Why you’re reading it. The paper supplies a precise learning hypothesis to compare with the motivational functions developed next.
A question to consider. If a reward becomes entirely predictable but remains worth pursuing, what should happen to the reward-evoked prediction error, and what would that leave unexplained about behavior?
Why this reading
A signal that changes when a reward is unexpected carries different information from one that simply registers the reward. Schultz, Dayan, and Montague connect patterns of dopamine-neuron firing with a computational account of learning. The paper shows how a formal prediction can organize neural observations and generate further tests.
Before you read
A reward-prediction error is a discrepancy between what occurs and what was expected. A positive error means better than expected; a negative error means worse than expected. Temporal-difference learning updates a prediction using immediate reward and the change in predicted future reward. A value estimates future returns, whereas a prediction error indicates a needed update. A raster plot shows spikes on successive trials; a histogram summarizes their timing. The equations formalize these distinctions rather than define pleasure.
Read for
- Compare unexpected reward, predicted reward, and omitted reward. Ask why timing and expectation matter to the proposed signal.
- Keep the recordings, model assumptions, simulations, and new predictions separate. Matching a pattern is evidence for a model, but does not establish every biological component it assumes.
- Examine the unresolved questions about temporal representation, aversive events, stimulus identity, and local effects at dopamine terminals. How broad is the proposed account?
Consider afterwards
- Sketch the predicted signal when a familiar reward arrives earlier than expected. Which part of your sketch depends on how time is represented?
- Why would a reduced response to a predictable reward not establish that the reward has become less pleasurable?
Editorial context
This influential synthesis links primate recordings and learning theory; it is not a single experiment demonstrating every proposed function of dopamine. The model's explanatory success should be separated from claims about all dopamine cells, targets, and motivational processes. Salamone and Correa's effort review and Mohebi and colleagues' comparison of firing and release illustrate why region, measurement, task, and timescale matter. Reward-prediction errors remain a specific explanatory proposal, not a synonym for pleasure or motivation as a whole.