What Is Underdetermination?

Direct Answer

Underdetermination is the thesis that the evidence available at any time — and, in the strong version, all possible evidence — is compatible with more than one scientific theory. A famous example: for any theory that fits all observations of planetary motion, one can construct an alternative theory that makes exactly the same observational predictions while disagreeing about the underlying mechanics — perhaps by adding hidden planets, adjusting initial conditions, or changing the geometry. If two theories make identical predictions, observation alone cannot tell them apart; evidence underdetermines theory choice. The thesis does not say that evidence is useless, only that it cannot force a unique conclusion. It marks the limit of empiricism: the data constrain the space of acceptable theories, but they do not uniquely fill it.

Historical Context

The idea has two classical sources. The French physicist and philosopher Pierre Duhem, at the turn of the twentieth century, argued that in physics a hypothesis is never tested in isolation: it is always tested together with auxiliary assumptions about instruments, initial conditions, and background theory. When a prediction fails, the logic does not tell us which part of the bundle to blame — the hypothesis or the auxiliaries. W. V. O. Quine generalized the point in "Two Dogmas of Empiricism" (1951): our beliefs face the tribunal of experience not individually but as a whole web, and in principle any statement can be held true "come what may" by adjusting other parts of the web. This Duhem-Quine thesis turned underdetermination from a technical point about physics into a general thesis about the structure of empirical knowledge. The history of science offers vivid examples: the Ptolemaic and Copernican systems were observationally equivalent for centuries — they differed in simplicity and worldview, not in predictions.

Key Distinctions

Several distinctions organize the concept. Weak versus strong underdetermination: weak underdetermination says actual evidence leaves multiple options; strong underdetermination says all possible evidence does. The strong version is philosophically radical but hard to prove; the weak version is uncontroversial and historically common. Holist versus holistic-equivalence versions: the Duhem-Quine thesis is holist — it concerns how a whole theory-web confronts experience — while the classic examples concern empirically equivalent theories that differ in unobservable structure. Epistemic versus logical underdetermination: the question is whether evidence fails to determine a rational choice (epistemic) or merely fails to logically entail a choice (logical). Even where logic leaves alternatives, scientists may still be rationally justified in preferring one — which is where extra-evidential virtues enter: simplicity, scope, explanatory power, and fruitfulness. These virtues are real criteria, but they are values, not observations.

Contemporary Debates

Underdetermination is the deepest threat to scientific realism. If evidence can never force a unique theory, then the realist claim that mature theories are true descriptions of the world seems to outrun what evidence can support; the instrumentalist responds that this is exactly why theories should be treated as predictive tools rather than truth-claims. Defenders of realism offer replies: the empirically equivalent alternatives are often artificial — constructed by gerrymandering — and real scientists judge them by virtues such as simplicity and unification that carry evidential weight of their own. The debate also connects to Kuhn's account of paradigm choice: because paradigms are incommensurable, the evidence alone cannot adjudicate between them, and the decision involves values, judgment, and community dynamics. Popper's falsificationism is likewise affected: if a failed prediction can always be blamed on an auxiliary assumption, then naive falsification is untenable — which is why Popper himself came to speak of bold conjectures and of accepting refutations only after serious attempts to defend the theory.

Practical Implications

Underdetermination has sobering practical lessons. First, evidence rarely settles a question alone: when experts disagree, the disagreement is often about which theory the evidence best supports under different weighting of values — acceptable risk, simplicity, caution — not about the facts. Second, beware claims of "proof": a theory that fits all the data is not thereby shown to be the only theory that could fit them; "the data confirm my theory" is weaker than "the data rule out all rivals." Third, in your own reasoning, the lesson of holism applies: when a prediction fails, examine the whole bundle of assumptions — including your own — before assigning blame, and be alert to the temptation to protect a favored belief by adjusting everything else around it, which is the intellectual counterpart of immunizing a theory against refutation. Fourth, recognize that choosing among theories that fit the same facts is legitimate, but be explicit about the values (simplicity, scope, caution) that drive the choice.

Related Concepts

• Scientific Realism vs Instrumentalism — the debate underdetermination fuels
• What Is Falsification? — the standard underdetermination complicates
• What Is Inductive Reasoning? — the inference evidence underdetermines
• What Is a Paradigm Shift? — how communities choose among incommensurable frameworks
• What Is Knowledge? — what underdetermination shows knowledge cannot be

Further Learning

• Read the SEP entry on Underdetermination of Scientific Theory
• Consult the IEP entry on the Duhem-Quine Thesis
• See the web-of-belief argument in Quine's Two Dogmas of Empiricism — Stanford Encyclopedia of Philosophy
• Visit the Understanding Reality collection for the full learning path