Instrumentalism

Core Idea

Instrumentalism views scientific theories as useful tools for prediction and control rather than descriptions of reality. Theories are instruments: they organize experience and facilitate manipulation of nature, but make no claims about how reality really is. Instrumentalism avoids metaphysical commitments and epistemological risks. However, it struggles to explain why instrumental theories seem to describe objective regularities and why success of abandoned theories suggests theories track something real.

Explainer

You know from scientific realism that the default philosophical interpretation of science is broadly optimistic: our best theories are approximately true descriptions of reality, including its unobservable parts. Electrons, genes, spacetime curvature — scientific realism says these posits genuinely refer to things that exist. Instrumentalism starts from a skeptical challenge to this picture. How could we ever know that a theory describes unobservable reality, rather than merely organizing observable phenomena in a useful way? The instrumentalist answers: we cannot, and we should stop pretending we can.

The core instrumentalist move is to reinterpret what scientific theories *are*. Rather than descriptions that can be true or false about a hidden reality, theories are inferential tools — devices for moving from observed inputs to predicted outputs. Think of a map. A map is not a true-or-false description of the world in the way a sentence is; it is a useful tool for navigation. Different maps of the same terrain (topographic, road, satellite) serve different purposes without any one being 'the true map.' Similarly, Newton's mechanics and quantum mechanics may describe the same physical domain in deeply incompatible ways, yet both are 'correct' in the instrumentalist's sense: both license reliable predictions within their domain of application.

This position has real appeal in the history of science. Ptolemy's epicycles made excellent astronomical predictions for over a thousand years — they were, by any standard, useful instruments — despite being, according to our current understanding, deeply wrong about the structure of the solar system. Instrumentalism would say: the question of whether they were 'really true' is a pseudo-question. What matters is predictive success. Similarly, Niels Bohr adopted an instrumentalist interpretation of quantum mechanics: the wave function is not a description of physical reality but a calculational device for predicting measurement outcomes.

The deepest challenge to instrumentalism is the no-miracles argument (central to scientific realism): if our theories are not approximately true, why are they so successful? The fact that engineers use quantum electrodynamics to build microchips that work seems to cry out for explanation. The instrumentalist's best response is that this 'success' is just the selection effect of theory choice — we keep the theories that work and discard the ones that don't. There is no mystery beyond that. But the debate turns on whether predictive success across novel domains genuinely demands a realist explanation, or whether successful prediction is all that science ever provides and all it should be asked to provide.