Model validation
Also: out-of-sample validation
Testing whether a model predicts data not used to fit it well enough for its intended decision.
Validation inspects residuals and task metrics across declared conditions and compares against alternative models or baselines.
Example A motor model fitted at 2 V and 4 V is checked with a separate 3 V trial.
Common misconception Training fit and numerical precision are not validation.