Reference library
A Neural Substrate of Prediction and Reward
Paper summary
This author synthesis relates earlier primate dopamine recordings to temporal-difference prediction theory. Selected evidence includes responses to unexpected appetitive outcomes, transfer toward predictive cues after training, and decreased firing at the expected time of omitted reward. The computational account adds a prediction target, temporal representation and learning assumptions; it is not an experiment demonstrating multilayer chain-rule gradients.
Relevance to the project
Useful for distinguishing a reward-prediction teaching signal from a receiver-specific parameter gradient. It motivates stating what a learning signal represents and which temporal credit mechanism uses it, without importing external reward into the adopted design.
Classic appetitive-task synthesis, not a universal description of all dopamine circuits. Later modeling discussion and reference evidence were not independently audited. Adjacent article material at the beginning of the PDF is excluded.
Source record
- Reference ID
- doi-10-1126-science.275.5306.1593
- Archived version
- Published article, Science 275:1593–1599 (14 March 1997); institutional PDF includes adjacent preceding article material.
- Summary provenance
- reference-library/papers/doi-10-1126-science.275.5306.1593/metadata.json · summary
- Review status
- Targeted source-grounded review; no experimental reproduction or exhaustive proof/figure audit
- Reading scope
- Published full-text PDF acquired and identity verified. Read article opening, reward/prediction evidence and Figure 1 caption, selected temporal-difference formulation and assumptions (principally article pages 1593–1595); not every subsequent section, figure or reference.