What Is Constructive Empiricism?

What Is Constructive Empiricism?

Constructive empiricism is a philosophical position about the aim of science, developed by Bas van Fraassen in his 1980 book "The Scientific Image." Its central claim is that science aims at empirical adequacy — the ability to correctly describe and predict observable phenomena — rather than truth about the unobservable aspects of reality. When a scientist accepts a theory, according to constructive empiricism, they believe that it is empirically adequate, but they need not believe that it is true, particularly in its claims about unobservable entities.

The position is called "constructive" because it takes science seriously as a rational, successful enterprise that constructs models of the world. It is called "empiricism" because it restricts justified belief to what can be empirically observed. The combination is meant to capture the insight that science is enormously successful at predicting and explaining observable phenomena, without requiring the additional commitment that its theoretical claims about unobservables are literally true.

To understand constructive empiricism, consider a scientific theory that posits electrons. The theory makes various predictions about observable phenomena — the readings on instruments, the tracks in cloud chambers, the behavior of electrical circuits. The constructive empiricist says that we should believe these predictions. If the theory has been well-tested and its predictions have been confirmed, we have good reason to believe that the theory is empirically adequate — that it will continue to correctly predict observable phenomena. But the constructive empiricist stops short of believing that electrons exist as the theory describes them. The existence and nature of electrons are claims about unobservables, and the evidence — which consists entirely of observable phenomena — does not warrant belief in these claims.

This might seem like an unnecessarily cautious position. If the theory works, why not believe that the entities it posits are real? Van Fraassen's answer is that this move involves an epistemic leap that goes beyond the evidence. The evidence supports the theory's empirical adequacy — its ability to save the phenomena — but the step from empirical adequacy to truth is an additional inference that is not logically compelled. We can explain the theory's success without committing to the truth of its claims about unobservables: the theory works because it correctly models the observable regularities in the world, not necessarily because its unobservable posits correspond to real entities.

Historical Background

Constructive empiricism emerged in the context of the scientific realism debate that intensified in the 1960s and 1970s. During this period, scientific realism — the view that successful theories are approximately true descriptions of reality — was the dominant position among philosophers of science. The No Miracles Argument, formulated by Hilary Putnam and Richard Boyd, seemed to provide a powerful case for realism: the success of science would be miraculous unless our theories were approximately true.

Van Fraassen published "The Scientific Image" in 1980, and it immediately became one of the most influential works in philosophy of science. The book was based on van Fraassen's 1978 Immanuel Kant Lectures at Princeton and presented a systematic challenge to scientific realism. Van Fraassen did not deny that science is successful or that theories are useful. What he denied was that the success of science gives us reason to believe that theories are true in their claims about unobservables.

The position has roots in the empiricist tradition. David Hume's skepticism about causation and induction, the logical positivists' emphasis on observable meaning, and the instrumentalist tradition associated with Pierre Duhem and Ernst Mach all contributed to the intellectual background of constructive empiricism. However, van Fraassen's position is distinct from these earlier views. Unlike instrumentalism, constructive empiricism takes theoretical claims literally — it does not reinterpret them as shorthand for observational claims. Unlike logical positivism, it does not dismiss claims about unobservables as meaningless. Instead, it grants that theoretical claims have genuine content but argues that we are not epistemically warranted in believing them.

Van Fraassen's argument drew on several philosophical resources. His voluntarism about belief — the idea that belief is not entirely determined by evidence but involves an element of choice — was influenced by the pragmatist tradition and by thinkers like William James. His emphasis on the distinction between observable and unobservable entities drew on a long tradition in empiricism. His probabilistic framework for understanding confirmation and evidence drew on Bayesian approaches to epistemology.

The reception of constructive empiricism was immediate and intense. Scientific realists developed new arguments to counter van Fraassen's challenges, including more sophisticated versions of the No Miracles Argument and selective realism. The debate between realism and constructive empiricism became one of the central debates in philosophy of science, generating a vast literature that continues to grow. Van Fraassen himself has refined and extended his position in subsequent works, including "Laws and Symmetry" (1989) and "The Empirical Stance" (2002).

Key Concepts

Empirical adequacy. The central concept of constructive empiricism. A theory is empirically adequate if what it says about observable phenomena is true — if it correctly describes and predicts everything that can be observed. Empirical adequacy is weaker than truth: a theory can be empirically adequate while being false in its claims about unobservables. Van Fraassen argues that empirical adequacy, not truth, is the aim of science. Scientists seek theories that save the phenomena, and the success of science can be fully explained by the empirical adequacy of its theories without invoking their truth.

The observable-unobservable distinction. The line between what can and cannot be observed is foundational to constructive empiricism. Van Fraassen draws this line based on what unaided human senses can detect. We can observe planets through a telescope because the telescope merely extends our natural visual capacity. We cannot observe electrons, even through an electron microscope, because the "observation" involves a complex theoretical inference from instrument readings to the entity, not direct perception. This distinction has been challenged by realists who argue that it is vague and arbitrary. If seeing through a telescope counts as observation, why not seeing through an electron microscope? Van Fraassen responds that the distinction, while not perfectly sharp, is clear enough for practical purposes and is grounded in the human sensory apparatus.

Acceptance versus belief. Van Fraassen distinguishes between accepting a theory and believing it. Acceptance involves a practical commitment: using the theory for prediction, explanation, and further research. Belief involves a cognitive commitment: holding that the theory is true. The constructive empiricist accepts theories — uses them, relies on them, takes them seriously — without believing them in full. Acceptance involves belief only in empirical adequacy, not in truth. This distinction allows the constructive empiricist to be a fully engaged participant in scientific practice without making the metaphysical commitments of the realist.

Voluntarism about belief. Van Fraassen holds that what one believes is not entirely determined by evidence. Two rational agents with the same evidence can legitimately hold different beliefs, because belief involves a volitional element — a choice about how much to commit beyond what the evidence strictly requires. This is not a license for irrationality. The evidence constrains belief: you cannot rationally believe something that the evidence contradicts. But the evidence often does not determine a unique belief, and in such cases, the agent has latitude. The constructive empiricist chooses to believe only what the evidence strictly warrants — empirical adequacy — while the realist chooses to go further and believe in truth about unobservables. Both choices can be rational, but the empiricist's is more epistemically cautious.

The critique of inference to the best explanation. Scientific realists often defend their position using inference to the best explanation (IBE). The best explanation of a theory's empirical success, they argue, is that the theory is approximately true. Van Fraassen challenges this use of IBE. He argues that IBE is a legitimate form of inference in some contexts but not in the context of inferring truth about unobservables from empirical success. The reason is that the "explanation" offered by realism — that the theory is true — is not the only available explanation of success. Empirical adequacy also explains success, and it does so without positing unobservable truth. Since the realist's explanation is not uniquely best, IBE does not warrant the inference to truth.

The epistemology of instruments. Van Fraassen's position raises questions about the epistemic status of instrument-based observations. When we use a thermometer, we read a number off a dial and infer a temperature. Is this observation or theory-laden inference? Van Fraassen argues that instrument readings can count as observations when the instrument is sufficiently transparent — when we can understand its operation without relying on the very theory being tested. A thermometer can be calibrated and checked independently, so its readings can count as observations. But a cloud chamber's "observation" of an electron track relies on theoretical assumptions about ionization and particle physics, making it theory-laden rather than purely observational.

Contemporary Relevance

Constructive empiricism remains one of the most influential positions in philosophy of science, and the debate between it and scientific realism continues to generate productive philosophical work. The position appeals to philosophers who are impressed by the success of science but skeptical of metaphysical commitments that go beyond the evidence. It offers a way to respect science without embracing what some see as the excesses of scientific realism.

The position has practical implications for how we think about scientific authority. If constructive empiricism is correct, then when scientists tell us about unobservable entities — dark matter, the Higgs boson, gravitational waves — we should understand these claims as empirically adequate models rather than literal truths about hidden reality. This does not diminish the practical value of these claims — the models work, and we can rely on them for prediction and control — but it changes their epistemic status. This perspective can be valuable in contexts where overconfidence in theoretical claims has led to problems, such as in economic modeling or climate prediction.

The rise of computational science and machine learning has new relevance for constructive empiricism. Machine learning models are essentially black boxes that make accurate predictions without providing theoretical explanations. Constructive empiricism provides a natural framework for understanding these models: they are empirically adequate instruments for prediction, and we need not commit to their internal representations as descriptions of reality. This perspective is increasingly relevant as AI systems take on larger roles in scientific research and practical decision-making.

The position also intersects with debates about scientific pluralism — the view that multiple, incompatible theories can each be legitimate scientific tools. If science aims at empirical adequacy rather than truth, then two theories that are both empirically adequate but make incompatible claims about unobservables can both be acceptable. This is harder to reconcile with scientific realism, which holds that theories aim at truth and that incompatible theories cannot both be true. Constructive empiricism thus provides a philosophical foundation for a more pluralistic understanding of science.

Critics of constructive empiricism have raised various objections. Some argue that the observable-unobservable distinction is untenable — that it rests on contingent facts about human physiology that have no principled epistemological significance. Others argue that the position is too restrictive — that it ignores the explanatory role of unobservable entities in scientific practice. Still others argue that van Fraassen's voluntarism about belief is problematic — that it makes belief too much a matter of personal choice and not enough a matter of rational constraint. These criticisms have led to refinements of the position and to the development of intermediate positions that try to capture the insights of both realism and empiricism.

Sources

• van Fraassen, B. C. (1980). The Scientific Image. Oxford: Clarendon Press.
• Stanford Encyclopedia of Philosophy. "Constructive Empiricism." https://plato.stanford.edu/entries/constructive-empiricism/
• van Fraassen, B. C. (2002). The Empirical Stance. New Haven: Yale University Press.
• Musgrave, A. (1985). "Realism versus Constructive Empiricism." In P. M. Churchland & C. A. Hooker (Eds.), Images of Science. Chicago: University of Chicago Press.
• Ladyman, J. (2002). Understanding Philosophy of Science. London: Routledge.

Related Topics

• What is scientific realism? — The opposing view that science aims at truth
• What is falsifiability? — Popper's demarcation criterion
• What is pseudoscience? — The demarcation problem
• How to evaluate scientific claims — Practical implications of the realism debate
• How to evaluate evidence — Evidence standards in science
• What is epistemology? — The broader philosophical context