Instrumentalism and Anti-Realism

Core Idea

Instrumentalism denies that scientific theories aim at truth; they are instruments for predicting observations and organizing experiments. Unobservable theoretical entities are merely convenient calculational fictions. This view sidesteps worries about whether theories match reality but sacrifices the idea of science as truth-seeking.

Explainer

From scientific realism, you know the realist's claim: successful scientific theories are approximately true, and the unobservable entities they posit — electrons, spacetime curvature, quarks — really exist. Realism is a natural interpretation of science's success story. Instrumentalism challenges that story at its root by rejecting the question of truth altogether for theoretical claims about unobservables.

The instrumentalist position, associated historically with Mach and Duhem and with van Fraassen's constructive empiricism in its modern form, says a scientific theory is an instrument for organizing and predicting observations. The Bohr model of the atom is not a picture of tiny orbiting particles — it is a calculation device that generates correct predictions about spectral lines. Whether there *really are* electrons in the sense the realist intends is a metaphysical question that instrumentalism sets aside as unanswerable. What matters is empirical adequacy: does the theory save the phenomena?

The move gains force from the history of science. Realists face the pessimistic meta-induction: every theory that has been abandoned (phlogiston, caloric, the luminiferous ether, Ptolemaic epicycles) was once predictively successful. If past successful theories had radically mistaken ontologies, why trust that current ones have the right ontology? Instrumentalism sidesteps the induction entirely: if we never claimed our theories described hidden reality, we cannot be embarrassed by ontological revision. The theory worked, and that is all theories are supposed to do.

The cost of instrumentalism is significant. It struggles to explain *why* theories work — if electrons do not exist, why does treating them as if they do generate such remarkably accurate predictions across wildly different experimental contexts? The realist's explanation is simple and natural: theories work because they are approximately true. The instrumentalist must either deny the question (pragmatic success is all that matters) or postulate some weaker notion of representational success. Van Fraassen's response is to accept that theories may be true, but argue we have no reason to *believe* the parts about unobservables — only the observable consequences.

Instrumentalism thus raises fundamental questions about scientific explanation and what science is *for*. If theories are instruments, does science explain the world or merely organize predictions about it? Is the distinction between observable and unobservable principled enough to sustain the asymmetric epistemic attitude instrumentalism recommends? These questions connect directly to constructivism and relativism, and they challenge the assumption — built into most science education — that learning science means learning how nature really is.