Underdetermination and the Duhem-Quine Thesis

Core Idea

The Duhem-Quine thesis states that no individual observation can conclusively test an isolated hypothesis because theories are tested holistically. When an experiment disagrees with prediction, we can always revise auxiliary hypotheses or background assumptions instead of rejecting the main theory. This means empirical evidence underdetermines theory choice: multiple incompatible theories can be consistent with all available evidence.

How It's Best Learned

Examine historical cases: Fresnel vs Newton on light, or the perihelion of Mercury (Newton vs Einstein). Show how ambiguous observational data can support competing theoretical frameworks.

Explainer

From your study of the theory-observation distinction, you know that observations are never purely neutral: what we see is always interpreted through prior theoretical commitments. The Duhem-Quine thesis pushes this further, making a logical point about how scientific testing actually works. When a prediction fails, we cannot simply conclude that the hypothesis being tested is false — because that hypothesis was never doing the work alone. Any experiment tests a whole web of beliefs simultaneously.

Here is the formal structure: to derive a testable prediction, you need your main hypothesis H plus a set of auxiliary hypotheses A (assumptions about instruments, background conditions, other known laws). The prediction P follows only from H ∧ A together. When P turns out false, logic tells you that H ∧ A is false — but it does not tell you which component to reject. You could abandon H. Or you could blame one of the auxiliaries: the instruments were miscalibrated, the experimental conditions were not controlled, a background assumption was wrong. Pierre Duhem, the physicist-philosopher, first noticed this in the context of optics; W.V.O. Quine extended it into a global thesis about the entire web of belief.

The history of science is full of instructive cases. When Uranus's orbit deviated from Newtonian predictions, astronomers did not abandon Newton's law of gravitation. Instead, they adjusted an auxiliary: perhaps there is an undiscovered planet perturbing the orbit. Le Verrier calculated where Neptune must be, and the telescope confirmed it. That was the right call. But when Mercury's perihelion refused to cooperate with Newton, generations of astronomers tried the same strategy — a hypothetical planet Vulcan — without success. Eventually, Newtonian mechanics itself had to go. The underdetermination thesis captures why both responses were logically available: no observation *forces* you to give up any particular belief.

The philosophical implications are significant. If theory choice is underdetermined by evidence — if multiple incompatible theories can always be made consistent with all available data by adjusting auxiliaries — then what determines which theory we should choose? The answer must appeal to criteria beyond pure empirical adequacy: simplicity, coherence, explanatory power, conservatism (minimal departure from prior belief), fruitfulness. Quine argued this means there is no sharp distinction between empirical science and speculative philosophy; both are parts of the same web of belief, differing only in their distance from the observational periphery. The Duhem-Quine thesis thus motivates holism: confirmation and refutation always involve whole theoretical systems, never isolated hypotheses.