Table of Contents
Underdetermination
Underdetermination is a relationship between evidence and theory in which the available evidence is logically compatible with more than one mutually incompatible theoretical conclusion. The term is most commonly used in philosophy of science, where it names the thesis that observational data alone are never sufficient to uniquely determine which of the competing theories that account for that data is correct. The problem has roots in ancient skepticism but received its modern formulation primarily through the work of Pierre Duhem and W.V.O. Quine, whose contributions are frequently combined under the label the Duhem-Quine thesis. Underdetermination bears directly on debates about scientific realism, instrumentalism, theory-ladenness of observation, and the rationality of scientific theory choice.
Scope and Formulations
Philosophers distinguish several versions of the underdetermination thesis, which differ in strength and in the conclusions they are taken to support.
Transient underdetermination (also called weak or practical underdetermination) holds that at any given moment, more than one theory is compatible with the evidence then available. This is widely accepted; the controversy concerns how much philosophical weight this observation can bear.
Holist underdetermination, associated with Duhem, holds that scientific hypotheses cannot be tested in isolation. Any test of a hypothesis H also implicates auxiliary assumptions - background theories, calibration procedures, instrument reliability, and so on. When an experiment yields an anomalous result, logic alone cannot identify which element of the tested conjunction (H plus auxiliaries) is false. Scientists must make a judgment about where to place the blame, and that judgment is not itself dictated by the evidence. This is sometimes called the Duhem problem or the problem of auxiliary hypotheses.
Quine's extension generalized Duhem's point to all knowledge, including logic and mathematics. On Quine's web of belief model, any statement can be retained in the face of recalcitrant experience by making sufficiently far-reaching adjustments elsewhere in one's belief system. This more radical thesis has been contested by philosophers who argue that it proves too much - that if everything is revisable, the thesis itself loses determinate content.
Strong or permanent underdetermination - the claim that for any body of evidence, however complete, there will always remain empirically equivalent but logically incompatible theories - is the most contested formulation. Its defenders argue it shows that scientific realism is untenable in principle; its critics argue that empirical equivalence is rarely genuine, and that when it does obtain, it is a defect in theory formulation rather than a feature of the world.
Historical Background
The problem was anticipated in ancient Pyrrhonist skepticism and appears in early modern form in discussions of astronomical models - notably in Andreas Osiander's preface to Copernicus's De Revolutionibus (1543), which argued that the heliocentric model need only save the phenomena rather than describe physical reality. The Tychonic system and the Copernican system made identical predictions from the available data, illustrating that observational equivalence does not resolve questions of truth. The modern philosophical treatment begins with Duhem's La Théorie physique: son objet et sa structure (1906) and was extended by Quine in “Two Dogmas of Empiricism” (1951) and Word and Object (1960). For a fuller treatment, see History of Underdetermination.
Current State of Debate
The underdetermination thesis is accepted in some form by most contemporary philosophers of science, but its precise scope and significance remain disputed. Several lines of response have been developed against the stronger versions of the thesis.
Critics of strong underdetermination argue that genuinely empirically equivalent theories - theories that agree on every possible observable consequence, not merely on observations made to date - are rare or non-existent outside of artificially constructed examples. The standard illustrations (such as the hypothesis that the world was created five minutes ago with all apparent evidence of a longer history in place) are acknowledged to be empirically equivalent to their rivals, but critics contend that they are not genuine scientific alternatives because they lack independent theoretical virtues.
A second line of response appeals to theoretical virtues - simplicity, explanatory scope, internal coherence, fruitfulness, and unification - as legitimate grounds for theory choice that go beyond brute consistency with the data. Scientific realists argue that these virtues are truth-conducive: theories that have them are more likely to be approximately true. Critics reply that the evidential status of theoretical virtues is itself underdetermined.
A historically grounded form of underdetermination has been developed by P. Kyle Stanford, who argues that the recurring failure of scientists to conceive of alternatives that were later recognized constitutes evidence for a problem of unconceived alternatives distinct from the Duhem-Quine thesis. On Stanford's account, the history of science shows that even our best theories have repeatedly faced rivals their proponents could not imagine, giving inductive grounds for humility about current theories.
The thesis also intersects with debates about social epistemology and the sociology of scientific knowledge, where some theorists argue that underdetermination creates space for social, cultural, or political factors to influence which theories scientists adopt. This application of the thesis is itself contested; see Underdetermination and Social Epistemology - Debate.
Consensus Status
There is broad consensus among philosophers of science that some form of underdetermination - at minimum, the Duhemian holist version - poses a genuine and non-trivial problem for naive falsificationism and for simple inductive accounts of scientific confirmation. There is no comparable consensus on the stronger thesis that underdetermination permanently prevents theory choice or undermines scientific realism. See Underdetermination - Philosophy of Science Consensus.
Viewpoints
Scientific Realist Viewpoint - Scientific realists hold that the underdetermination problem, while real, does not preclude warranted belief in the approximate truth of well-confirmed theories. They argue that inference to the best explanation and the track record of mature science provide grounds for realism that the underdetermination thesis does not defeat.
Instrumentalist Viewpoint - Instrumentalists take underdetermination as support for the view that scientific theories are tools for organizing experience and generating predictions rather than descriptions of a mind-independent reality. On this view, questions about which of two empirically equivalent theories is “really true” are ill-formed.
Constructive Empiricist Viewpoint - Constructive empiricists, following Bas van Fraassen, argue that science aims at empirical adequacy rather than truth. Underdetermination is taken to show that belief in the truth of theories (as opposed to their empirical adequacy) exceeds what the evidence warrants.
Social Constructivist Viewpoint - Social constructivists argue that underdetermination is not merely an abstract logical problem but an actual condition of scientific practice, and that the gap between evidence and theory is routinely filled by social, institutional, and ideological factors. This view is associated with the Strong Programme in the sociology of scientific knowledge.
Pragmatist Viewpoint - Pragmatists, in the tradition of John Dewey and later Richard Rorty, hold that underdetermination dissolves as a problem once the demand for a theory-independent correspondence between belief and world is abandoned. On this view, the question is not which theory is true but which serves inquiry best.