The view that a scientific theory is to be regarded as an instrument for producing new predictions or new techniques for controlling events, but not as itself capable of literal truth or falsity. The most famous example of this claim is that of Andreas Oseander (or Osiander, 1498–1552), whose Preface to De Revolutionibus Orbium Coelestium advocated that Copernicus’s heliocentric theory of the solar system should be accepted as a device for predicting eclipses and tides, but not regarded as true (and therefore potentially in conflict with Church doctrine). Instrumentalism diminishes the difficulty over our right to confidence in scientific theory, since it is easier to suppose that we have a right to adopt a theory as an instrument, than that we have a right to regard it as true. A tempting reaction to deep theoretical stress, such as the apparently incompatible wave-particle duality of quantum mechanics, is to suggest that each view serves as an instrument within its own proper sphere, and thereby to sidestep the theoretical urge to reconcile them. It is also tempting to take an instrumentalist (sometimes called heuristic) attitude to the use of devices such as sets, numbers, or possible worlds, that seem to facilitate our thinking in important ways, but not to deserve a place in our ontologies.

A difficult question in the philosophy of mind and language is to tell what distinguishes acceptance in a purely instrumentalist spirit from true belief. Some philosophers of a pragmatist bent, whilst sympathetic to instrumentalism, will be especially prone to deny that there is a real distinction here, since in such a philosophy all belief is simply acceptance into the system deemed most useful.