Why can a scientific theory make accurate predictions even if one of its claims is false?

In the context of scientific realism, a theory can generate accurate predictions even if some of its claims are false because its predictive success may rely only on specific, approximately true components rather than the theory as a whole.1

Philosophers often distinguish between the "instrumental" reliability of a theory—its ability to "save the phenomena" or predict observable outcomes—and the truth of its underlying theoretical claims. For example, Augustin-Jean Fresnel’s theory of light was highly successful in predicting phenomena like diffraction and interference, despite being based on the false hypothesis of an all-pervading ether. Contemporary science retains the mathematical treatment of wave propagation from Fresnel’s work while discarding the ether, demonstrating that the parts of a theory responsible for predictive success can be isolated from those that are incorrect.231

From an instrumentalist perspective, theories are viewed as tools for generating predictions about observables, meaning their truth or falsehood is secondary to their empirical adequacy. Conversely, realists argue that the success of these theories is best explained by the approximate truth of the specific mechanisms they identify.2314