Constructive Empiricism

Core Idea

Bas van Fraassen's constructive empiricism proposes that the aim of science is to produce empirically adequate theories (theories whose observable consequences are true), not to describe reality truly. We should believe that observed phenomena occur as the theory says, but need not believe claims about unobservables. This preserves empiricism while avoiding commitment to a theory-independent reality and explains scientific success without positing mysterious accuracy about unobservables.

Explainer

From your study of scientific realism, you know the realist's core claim: mature, successful scientific theories are approximately true descriptions of reality, including its unobservable parts — electrons, quarks, spacetime curvature. Realism explains scientific success by pointing to truth: theories work because they correctly describe the world. Bas van Fraassen's constructive empiricism accepts the empiricist challenge to this picture. It asks whether we really need to believe in unobservables to account for everything science achieves.

Van Fraassen's answer is no. He proposes that the aim of science is not truth but empirical adequacy: a theory is empirically adequate if everything it says about *observable* phenomena is true. The critical distinction is between the observable and the unobservable: observable things are things we could perceive directly under favorable conditions — a bacterium viewed through a microscope qualifies, but an electron does not. We should believe our best theories only to the extent of their empirical adequacy. We accept that they correctly describe what we can observe, but we remain agnostic — not disbelieving, merely withholding belief — about their claims concerning unobservables.

This position is more subtle than it first appears. Van Fraassen is not an instrumentalist who treats theories as mere calculation tools with no real content. He thinks the claims of a theory are genuinely true or false. He simply argues that *rational belief* should track what is observable. The analogy he offers: a navigator might rely on a map because it correctly depicts observable coastlines, while remaining agnostic about whether the map's depth contours correspond to underwater formations no one has ever seen. Acceptance of a theory — using it to guide action and belief about observables — is rationally separable from full *belief* in its unobservable ontology.

The most powerful objection is the no-miracles argument: if electrons didn't exist, it would be miraculous that theories positing electrons make such extraordinarily accurate predictions. Realists argue the best explanation of scientific success is approximate truth about unobservables. Van Fraassen's response is that this argument itself illicitly presupposes inference-to-best-explanation: it assumes we should believe the best explanation, but whether we should extend belief beyond observables is exactly what's at issue. He argues that empirical adequacy is sufficient to explain predictive success, and that the history of radical theory change — where formerly successful theories were later abandoned — counsels humility about any unobservable commitments we make today.