Underdetermination in Science: Why Certainty Remains Elusive

Evidence often supports multiple theories. This article explains the roots and implications of underdetermination for science’s credibility and progress.

Science is often portrayed as an impartial and cumulative process where evidence steadily narrows the range of possibilities until a single theory emerges as the “truth.” This image reassures the public and policymakers, but it is overly simplistic. The actual process of science is far messier and less linear than textbooks sometimes suggest.

The concept of underdetermination captures this reality. It refers to situations where available evidence is compatible with multiple, even incompatible, theoretical explanations. Underdetermination is not merely an abstract philosophical problem; it is a recurring feature of real-world science. Whether in cosmology, quantum physics, or climate modeling, scientists often face situations where data can be interpreted in more than one way.

This article examines the origins of the underdetermination problem, its appearance in major historical and contemporary scientific cases, and its implications for scientific objectivity and credibility. Recognizing underdetermination does not weaken science seen as an instrumental project; but it does make the idea of scientific realism impossible — a fact that few scientists want to acknowledge.

1. What is Underdetermination?

Underdetermination arises when empirical evidence alone cannot select one theory as the uniquely correct explanation. Different theories may make the same predictions, or their discrepancies may lie outside the scope of what can be observed with current methods.

For instance, a set of cosmological observations may be equally compatible with a universe dominated by dark matter or one governed by modified gravity laws. Neither theory can be confirmed or refuted by the available evidence. This is not unusual; it is the norm in many scientific debates.

When evidence fails to discriminate, scientists turn to criteria beyond empirical adequacy. These include simplicity, explanatory power, and coherence with other accepted theories. However, these criteria are not objective in the same way as empirical tests. What one scientist values as “simple” or “elegant,” another may find unconvincing.

Thus, underdetermination highlights the limits of evidence-based reasoning. It reminds us that even when experiments are well-designed and data are plentiful, theory choice may remain ambiguous, shaped by values, preferences, and broader paradigms.

2. Historical Roots: Duhem and Quine

The modern understanding of underdetermination owes much to Pierre Duhem and W.V.O. Quine. Duhem argued that hypotheses are never tested in isolation. Every prediction relies on a network of auxiliary assumptions: background theories, measurement assumptions, and even mundane factors such as instrument calibration.

When an experiment fails to produce the expected result, we cannot be sure whether the problem lies with the central hypothesis or with one of the auxiliaries. This “holism” makes scientific testing far more complex than a simple confirm-or-refute model suggests.

Quine took Duhem’s insight further. He argued that our entire web of beliefs faces the tribunal of experience as a whole. Because there are countless ways to adjust auxiliary assumptions, there is no unique way to respond to conflicting evidence.

The Duhem–Quine thesis reveals underdetermination as a structural feature of science. It is not a temporary obstacle solvable with more data but a permanent characteristic of how theories relate to evidence.

3. Historical Example: Astronomy and Cosmology

Underdetermination is not a modern invention; it has shaped science for centuries. Consider the competition between geocentric and heliocentric models of the solar system. Before the advent of telescopic evidence, both models could explain planetary motions by introducing additional epicycles and adjustments.

Supporters of the geocentric view could tweak parameters to fit new observations, just as heliocentrists did. Evidence alone could not definitively decide between the two until broader shifts in astronomy and physics tipped the balance.

A modern parallel exists in cosmology. Observations of galactic rotation curves can be explained by positing vast amounts of unseen dark matter or by proposing that gravity behaves differently on large scales (modified gravity). Both hypotheses fit the evidence, albeit with different assumptions.

This persistence of competing frameworks illustrates how evidence can remain ambiguous even in highly quantitative sciences. Additional data can help, but often it simply sharpens debates rather than eliminating them.

4. Example: Quantum Physics

Quantum mechanics provides a striking contemporary example. Several interpretations — the Copenhagen interpretation, the Many-Worlds hypothesis, Bohmian mechanics — make exactly the same experimental predictions.

Because these interpretations differ only in their metaphysical commitments, no experiment can distinguish among them. Scientists may prefer the conceptual clarity of Bohmian mechanics or the elegance of Many-Worlds, but such preferences are philosophical rather than empirical.

This situation has persisted for decades and may endure indefinitely. Underdetermination here is not a failure of technology or imagination; it is built into the theory itself. The mathematical formalism works regardless of which interpretation one favors, leaving theory choice to aesthetic or philosophical considerations.

Quantum physics thus demonstrates how underdetermination can be deeply embedded in even the most successful scientific frameworks. Evidence alone cannot always “force” consensus, and this reality must be acknowledged rather than ignored.

5. Auxiliary Assumptions and Flexibility

One reason underdetermination persists is that scientific testing depends heavily on auxiliary assumptions. Experiments do not test hypotheses in a vacuum; they rely on assumptions about instruments, background conditions, and other theories.

When evidence contradicts predictions, researchers can modify auxiliary assumptions to preserve the core theory. For example, a failed particle detection might be blamed on a malfunctioning detector rather than on the theoretical model predicting the particle.

This flexibility is not inherently bad. It allows scientists to troubleshoot and refine their methods rather than discarding useful theories prematurely. But it also means that evidence can rarely deliver a decisive verdict. Competing theories may adapt to fit the same evidence, prolonging debate and uncertainty.

Understanding the role of auxiliary assumptions helps explain why underdetermination is an enduring feature of science rather than a rare anomaly.

6. Theoretical Virtues: Beyond Evidence

When evidence is insufficient to discriminate between theories, scientists turn to theoretical virtues—qualities that make a theory appealing beyond its empirical adequacy. These virtues include simplicity, coherence with established theories, explanatory scope, and the ability to generate fruitful new research.

The problem is that such virtues are not objectively measurable. One scientist may favor simplicity, while another values explanatory richness. These subjective preferences shape theory choice as much as, or sometimes more than, the available data.

For instance, advocates of the Many-Worlds interpretation of quantum mechanics value its conceptual clarity, while critics find the idea of infinite universes extravagant. Neither position is determined by evidence.

This reliance on theoretical virtues reinforces the point that science is not purely mechanical. Human judgment plays a central role in theory choice, and underdetermination ensures that this judgment can never be fully replaced by data.

7. Paradigms and Frameworks (Kuhn)

Thomas Kuhn’s theory of scientific paradigms adds another dimension. Paradigms establish shared assumptions, methods, and standards of evidence for a scientific community. Within a paradigm, anomalies are often set aside or explained away rather than prompting wholesale revision.

Paradigm shifts occur only when anomalies accumulate to the point that the existing framework becomes untenable. The new paradigm reinterprets old evidence in a different conceptual light, often rendering it impossible to directly compare theories.

This paradigm dependence shows how underdetermination operates at a systemic level. Competing paradigms can both explain the same evidence, but their differing assumptions make cross-comparison difficult.

Kuhn’s insights highlight the sociological dimensions of underdetermination: theory choice is shaped not just by data but by community norms, intellectual traditions, and historical contingencies.

8. Can More Evidence Solve the Problem?

It is tempting to think that gathering more data will eventually resolve underdetermination. In some cases, new evidence does help narrow the range of plausible theories.

However, the problem is deeper. Theories can often be adjusted to accommodate additional evidence. Dark matter models can add new particle candidates, while modified gravity theories can tweak equations. The same data set may support multiple explanations indefinitely.

This does not mean that science is futile, in the sense of solving practical problems. Rather, it means that theory choice is a dynamic process involving both evidence and judgment. More evidence can reduce uncertainty, but it rarely eliminates underdetermination entirely. Recognizing this helps temper unrealistic expectations about scientific certainty. In other words, scientific realism is an impossible dream.

9. Implications and Conclusion

Underdetermination challenges the idea that evidence alone determines scientific truth. But it does not mean that science is arbitrary or unreliable, seen as an instrumentalist project.

Scientific communities use peer review, replication, and cross-disciplinary scrutiny to stabilize knowledge despite underdetermination. These social and methodological safeguards ensure that theories are tested rigorously, even if they are never proven conclusively.

Acknowledging underdetermination fosters humility about scientific claims and clarifies the provisional nature of theories. Science progresses not by achieving final truth but by continually refining its models through debate and critical testing.

By understanding the limits of evidence, we can appreciate both the strengths and the constraints of science. Underdetermination reminds us that science is a human endeavor, guided by evidence but also shaped by values, assumptions, and historical context.

In other words, science is fallible and can, in principle, never deliver any conclusive metaphysical truths about which “reality” we are in; science is an instrumental project only, and scientific realism is impossible.

Chris Bocay

---