Realism and Instrumentalism

A brief clarification about one of the main differences between realism and instrumentalism in science.

Realism in science holds that our scientific theories are true or approximately true descriptions of not only the underlying hidden causal structures of phenomena but also how they might give rise to the observable features  of various phenomena. So, for instance, the quantum mechanics account of matter is not only a true or approximately true description of what goes on at the hidden causal structure level of matter, but is also a true or approximately true description of how the observable features of matter are causally produced by the hidden quantum mechanical structure of matter.

Instrumentalism, on the other hand, denies that our scientific theories are true or approximately true descriptions of things in the ways just described above. As a result, it holds that scientific theories are nothing more than theoretical tools or instruments whose job is to help scientists with the task of ultimately making predictions about the observable phenomena of the world. So, for example, in the history of science the original editor of Copernicus's book on The Revolutions of the Planets construed Copernicus's theory of the solar system as being not a literally true description of it. Consequently, for this editor, Copernicus's scientific theory of the solar system is nothing more than a useful instrument for making predictions about the orbital positions of the planets and stars. Thus, according to instrumentalism, the goal of science is useful predictions, not descriptive (and explanatory) truth, which is akin to Pierre Duhem's instrumentalism.

As a footnote to this discussion, this is how we should look at the instrumentalist nature of modern orthodox neoclassical economics: its methodological approach to a theoretical study of a real market economy like capitalism is one in which it does not (a la Milton Friedman) seek to give a true or approximately true account of it, but only a useful predictive account of it instead.