Reward prediction errors can update the learned value of actions and influence subsequent reward-seeking choices.

This page isolates one claim from the wider argument of The Human OS Manual.

The classification describes the evidence supporting this claim, not the Human OS framework as a whole.

Pessiglione and colleagues used a controlled human instrumental-learning task with pharmacological manipulation and fMRI, finding that dopamine-dependent changes in striatal reward-prediction-error signals tracked changes in reward-seeking choice.

This experiment does not show that every completed task produces the same dopamine response, that dopamine alone explains motivation, or that all forms of action-outcome learning use the same reward-learning mechanism.

The chapter uses the narrower learning principle to explain why small, completed actions can generate new evidence. The Evidence Map does not convert that into a universal dopamine claim.

Developing findings or mechanisms that help explain possibility without establishing universal outcomes or direct real-world effects.

Sources are listed according to the role they play in this claim. One source may provide direct support, another may explain a mechanism, establish a limitation, or prevent an interpretation from becoming broader than the evidence allows.

The printed book preserves the claim inside the argument that gives it meaning. These references connect the current digital record to the first-edition evidence trail.

This claim connects to the following parts of the Human OS architecture. The module labels locate the claim inside the framework; they do not imply that the research source validates the architecture itself.