What Is Scientific Realism?

What Is Scientific Realism?

Scientific realism is the view that successful scientific theories give us approximately true descriptions of a mind-independent reality. When physics tells us about electrons and quarks, when biology tells us about genes and natural selection, when astronomy tells us about black holes and dark matter, the scientific realist takes these claims to be roughly correct. The entities that theories posit are real, and the relationships the theories describe actually hold in the world.

This might seem like common sense. Of course electrons are real — we use them every day in electronic devices. Of course genes exist — we can sequence them. But the philosophical position is more nuanced and more controversial than it first appears. The problem is that many of the entities science posits are unobservable. We cannot see electrons, genes, or black holes directly. We infer their existence from observable phenomena — the tracks in a cloud chamber, the patterns of inheritance, the gravitational effects on nearby stars. The scientific realist says that these inferences are warranted: the best explanation of the observable phenomena is that the unobservable entities exist and behave as our theories describe.

Anti-realists disagree. They argue that we should be more cautious. The history of science is littered with entities that were once posited and later abandoned — phlogiston, caloric, the ether, crystalline spheres. If scientists of the past were wrong about these entities, why should we think we are right about electrons? Perhaps our current theories, despite their empirical success, posit entities that do not exist. The anti-realist concludes that we should accept our theories as empirically adequate — useful for predicting observable phenomena — without committing to their truth about unobservables.

Scientific realism is not a single doctrine but a family of positions, varying in strength and scope. The most common formulation, due to Hilary Putnam and Richard Boyd, has three components. The metaphysical component holds that the world has a definite, mind-independent structure. The semantic component holds that scientific theories should be interpreted literally — the terms in theories refer to real entities, and the claims of theories are true or false. The epistemic component holds that mature, successful theories are approximately true. Each component can be challenged independently, leading to a spectrum of positions.

Historical Background

The debate between realism and anti-realism about science has ancient roots. Plato distinguished between the world of appearances and the world of Forms, suggesting that what we observe is not ultimate reality. In the modern period, the debate took shape with the rise of empirical science. John Locke distinguished between primary qualities (shape, size, motion), which he thought existed in objects themselves, and secondary qualities (color, taste, smell), which he thought existed only in the mind of the perceiver. This distinction raised questions about which aspects of the world science truly captures.

The instrumentalist tradition, associated with thinkers like Pierre Duhem and Ernst Mach, argued that scientific theories should be understood as instruments for predicting observable phenomena, not as descriptions of reality. Mach, for instance, argued that atoms were merely convenient fictions — useful for calculation but not necessarily real. This instrumentalist tradition influenced the logical positivists of the early twentieth century, who were generally skeptical of metaphysical claims about unobservable reality.

The modern debate was crystallized in the mid-twentieth century by several developments. The success of quantum mechanics raised acute questions about realism. The theory's probabilistic nature and the measurement problem seemed to challenge the idea that science describes a definite, mind-independent reality. The Copenhagen interpretation, associated with Niels Bohr, was widely interpreted as anti-realist, suggesting that quantum mechanics describes our observations rather than the quantum world itself.

Hilary Putnam's work in the 1960s and 1970s was pivotal in defending scientific realism. Putnam formulated the "No Miracles Argument," which remains the most influential argument for realism. Putnam argued that the success of science would be a miracle — a completely inexplicable coincidence — unless our theories were approximately true. The fact that our theories make accurate predictions about novel phenomena is best explained by the hypothesis that they capture something real about the world.

Bas van Fraassen's 1980 book "The Scientific Image" provided the most influential contemporary challenge to realism. Van Fraassen proposed "constructive empiricism," the view that science aims at empirical adequacy — saving the phenomena — rather than truth. He accepted that theories make claims about unobservables but argued that we should not believe those claims. We should accept a theory as empirically adequate without committing to its truth. Van Fraassen's argument forced realists to clarify their position and develop more sophisticated defenses.

The historical challenge to realism was strengthened by Larry Laudan's 1981 paper "A Confutation of Convergent Realism," which presented a long list of historically successful theories that turned out to be false. If empirical success does not guarantee truth in the past, Laudan argued, we have no reason to think it does now. This "pessimistic meta-induction" remains one of the most powerful arguments against scientific realism.

Key Concepts

The No Miracles Argument. The most famous argument for scientific realism. Putnam argued that the empirical success of science — its ability to make accurate predictions, especially novel predictions about phenomena not yet observed — would be inexplicable unless our theories were approximately true. If electrons did not exist and our theories about them were false, it would be a miracle that our electron-based theories successfully predict the behavior of cathode ray tubes, semiconductors, and particle accelerators. The best explanation of scientific success is that our theories are roughly right about the world, including the unobservable parts.

The Pessimistic Meta-Induction. The most powerful argument against scientific realism. Laudan compiled a list of theories that were empirically successful in their time but are now considered false — including the phlogiston theory of combustion, the caloric theory of heat, and the wave theory of light in a luminiferous ether. These theories posited entities (phlogiston, caloric, the ether) that we now believe do not exist. If past successful theories were false, the argument goes, we have no reason to believe that current successful theories are true. The success of a theory is not a reliable indicator of its truth.

Selective realism. A response to the pessimistic meta-induction. Selective realists argue that the pessimistic induction is too blunt. Not all parts of a theory are equally responsible for its success. The posits that drive the theory's empirical success — its "working posits" — are the ones we should believe in, while the idle or speculative components can be abandoned. Philip Kitcher and Stathis Psillos have developed sophisticated versions of this argument, showing that in many historical cases of false-but-successful theories, the false parts were not essential to the theory's success. The structural realist variant, associated with John Worrall, argues that what is preserved across theory change is not the entities but the mathematical structure, and that this structure is what we should be realists about.

Underdetermination. The thesis that the evidence available to us cannot determine which of several competing theories is true. For any body of evidence, there are multiple theories that are consistent with it. If the evidence cannot distinguish between theories, then we cannot justifiably believe that any particular theory is true. This challenge is particularly acute for claims about unobservables, since we can only observe the observable consequences of theories, not the unobservable entities they posit. Underdetermination does not refute realism directly but weakens the epistemic grounds for believing that our theories are true rather than merely empirically adequate.

The distinction between observable and unobservable. A central issue in the debate. Van Fraassen draws the line between what can be observed with the unaided senses and what requires instruments. Electrons are unobservable because we can never see them directly, only their effects. Planets, on the other hand, are observable — we could see them through a telescope, and in some cases with the naked eye. Realists challenge this distinction, arguing that the line between observable and unobservable is vague and arbitrary. If seeing through a telescope counts as observation, why not seeing through an electron microscope? The distinction seems to depend on contingent facts about human physiology rather than principled epistemological considerations.

Truth and approximate truth. Realists claim that theories are "approximately true," but what does this mean? Exact truth is too strong — we know that our theories are not perfectly accurate descriptions of reality. But approximate truth is difficult to define precisely. How do we measure the distance between a theory and the truth? If we cannot define approximate truth, the realist's claim becomes vague. This technical challenge has generated significant philosophical literature, with various proposals for how to formalize the notion of closeness to truth.

Contemporary Relevance

The debate over scientific realism has implications that extend beyond philosophy of science. It bears on questions about the nature of knowledge, the relationship between theory and reality, and the limits of human understanding. If scientific realism is correct, then science gives us genuine knowledge about the fundamental nature of reality, including aspects we cannot directly observe. If anti-realism is correct, then science is a powerful tool for prediction and control, but we should be cautious about treating its theoretical claims as descriptions of reality.

The debate is relevant to public understanding of science. When scientists tell the public about dark matter, climate change, or viral evolution, they are making claims about unobservable entities and processes. If the public adopts an anti-realist stance — treating these claims as useful fictions rather than descriptions of reality — this could undermine support for science-based policy. On the other hand, a naive realism that treats every scientific claim as certain truth can lead to disillusionment when theories are revised or overturned.

The replication crisis has added a new dimension to the realism debate. If many published scientific findings cannot be replicated, this suggests that the empirical success of at least some scientific claims is less robust than assumed. This does not directly challenge scientific realism — the crisis is primarily about methodological issues, not the truth of well-established theories — but it does complicate the realist's appeal to the success of science as evidence for truth.

Advances in physics continue to challenge and refine the realism debate. String theory posits entities and dimensions that are far beyond any conceivable observation. Is string theory science? If it is not falsifiable (as some critics argue), can it be approximately true? These questions push the boundaries of both realism and the demarcation problem. Similarly, the interpretation of quantum mechanics remains deeply contested, with some interpretations (many-worlds, Bohmian mechanics) more realist-friendly than others.

The rise of data-driven science and machine learning raises new questions for realism. When a machine learning model makes accurate predictions without providing a theoretical explanation, what is the epistemic status of its outputs? Is the model describing something real about the world, or is it merely a predictive instrument? The instrumentalist tradition in philosophy of science may provide a framework for understanding these models, but the question of whether their success implies truth about underlying mechanisms remains open.

Sources

• Stanford Encyclopedia of Philosophy. "Scientific Realism." https://plato.stanford.edu/entries/scientific-realism/
• van Fraassen, B. C. (1980). The Scientific Image. Oxford: Clarendon Press.
• Psillos, S. (1999). Scientific Realism: How Science Tracks Truth. London: Routledge.
• Laudan, L. (1981). "A Confutation of Convergent Realism." Philosophy of Science, 48(1), 19–49.
• Putnam, H. (1975). "What Is Mathematical Truth?" In Mathematics, Matter and Method (Vol. 1). Cambridge: Cambridge University Press.

Related Topics

• What is constructive empiricism? — The leading anti-realist alternative
• What is falsifiability? — Popper's criterion for scientific status
• What is pseudoscience? — The demarcation problem
• How to evaluate scientific claims — Practical implications of the realism debate
• How to evaluate evidence — Evidence standards and realism
• What is epistemology? — The philosophical context