Page type: Consensus | Domain: Philosophy of Science

—

Scientific realism - broadly, the view that successful scientific theories are approximately true descriptions of a mind-independent world, including its unobservable components - is among the most contested foundational questions in philosophy of science. No field-wide consensus exists either for or against scientific realism. What does exist is a qualified, defeasible majority position among professional philosophers of science favoring some form of realism, alongside sustained, institutionally distributed dissent. Inter-community disagreement is substantial: working scientists, philosophers of science, historians of science, and sociologists of scientific knowledge hold systematically different views. Within academic philosophy of science specifically, surveys suggest a plurality lean toward realism, but the question remains genuinely open and the debate is considered live by its participants.

—

The PhilPapers Survey (2020), a large-scale poll of professional philosophers, found that approximately 75% of philosophers of science leaned toward or accepted scientific realism, compared with roughly 12% accepting anti-realism, with the remainder undecided or holding other views.¹⁾ This represents a rough plurality position within academic philosophy rather than a settled consensus; the survey itself is designed to document opinion distributions, not establish truth, and the realism/anti-realism question remains one of the survey's more contested items. The plurality is also sensitive to how realism is defined: structural realism, entity realism, and explanationist realism each attract different levels of support.

The most widely discussed argument for scientific realism - the “no-miracles argument,” associated principally with Hilary Putnam - holds that realism is the only explanation that does not make the predictive success of science a miracle.²⁾ This argument has broad recognition as the strongest positive case for realism and is reproduced in most introductory and advanced treatments of the debate. However, there is no consensus that it is decisive. Its force is acknowledged even by many anti-realists, who respond with the pessimistic meta-induction and other counterarguments rather than denying the argument's framing entirely.

The pessimistic meta-induction, developed most systematically by Larry Laudan, observes that the history of science is populated by theories that were empirically successful but whose central posited entities are no longer accepted - phlogiston, caloric fluid, the luminiferous ether.³⁾ Among historians and philosophers of science who engage seriously with historical case studies, there is broad agreement that this pattern is real: mature, predictively successful theories have been abandoned and their theoretical ontologies discarded. There is no consensus, however, on what this implies for realism. Selective realists argue that the posited entities which did do genuine explanatory work tended to be retained (the “divide et impera” strategy); anti-realists dispute this selectivity.

Structural realism - the position that what science correctly tracks is the relational or mathematical structure of the world rather than the intrinsic nature of unobservable entities - has achieved a notable degree of uptake within the specialist philosophy of physics community. Epistemic structural realism (ESR), associated with John Worrall, and ontic structural realism (OSR), associated with James Ladyman and Don Ross, are both recognized as serious positions addressed in the primary literature.⁴⁾⁵⁾ No consensus exists on whether structural realism succeeds, but there is broad agreement among specialists that it represents a genuine response to the pessimistic meta-induction.

Among working scientists, there is no organized or surveyed consensus on the philosophical question of realism. The informal or unreflective attitude of many practicing scientists has been described by philosophers as “naïve” or “common-sense” realism - the assumption that successful theories are approximately true - but this attitude is not a product of philosophical argument and is not treated by philosophers of science as evidential for the realist position.

—

The following questions remain genuinely open and unsettled within the expert literature:

• The reference problem: Whether the terms of superseded theories successfully referred to entities in the world, and whether reference is preserved across theory change, is disputed. Causal theories of reference (Kripke, Putnam) are relevant but not agreed to be decisive.
• What “approximate truth” means: Most realists invoke approximate truth but few have provided a rigorous semantic account of it. The absence of a generally accepted definition is an acknowledged weakness of the realist program.
• The selective realist program: Whether it is possible to identify, in advance or retrospect, which theoretical posits are “working” in the way that justifies realist commitment is disputed. Critics argue that the selection criterion is post hoc.
• Structural realism's coherence: Whether ontic structural realism is genuinely coherent - whether relations can exist without relata - is a live metaphysical dispute.
• Bas van Fraassen's constructive empiricism: Van Fraassen's position - that the aim of science is empirical adequacy rather than truth, and that belief should be withheld regarding unobservable entities - is widely regarded as the most technically sophisticated anti-realist alternative and has not been refuted to the satisfaction of its adherents.⁶⁾
• Entity realism: The view, defended by Ian Hacking and Nancy Cartwright, that realism is warranted for entities we can manipulate but not for the theoretical laws governing them, occupies a distinct position that does not map cleanly onto the realism/anti-realism divide.⁷⁾⁸⁾
• Social and historical dimensions: Sociologists of scientific knowledge and historians of science frequently challenge the presuppositions of the debate, questioning whether the history of science supports the cumulative-convergence picture that realists typically invoke.

—

The following viewpoint pages address positions that challenge, qualify, or reject scientific realism. The existence of this consensus page does not adjudicate between them.

• Scientific Realism - Constructive Empiricism Viewpoint - Bas van Fraassen's constructive empiricism; science aims at empirical adequacy, not truth
• Scientific Realism - Entity Realism Viewpoint - Hacking and Cartwright on realism for entities but not laws
• Structural Realism - Scientific Realism Viewpoint - Worrall, Ladyman, and Ross on structure as the proper object of realist commitment
• scientific-realism-pessimistic-meta-induction-viewpoint - Laudan's argument from the history of scientific theory change
• Philosophy of Science - Scientific Realism and Instrumentalism Viewpoint - Scientific instrumentalism; theories as tools rather than descriptions
• scientific-realism-social-constructivism-viewpoint - Sociological and constructivist challenges to realist accounts of scientific knowledge
• scientific-realism-pragmatism-viewpoint - Pragmatist approaches that reframe or dissolve the realism debate

—

• Scientific Realism - Main topic
• Scientific Realism - History - History of the realism debate in philosophy of science
• Scientific Realism - Debate - Overview of the realism/anti-realism debate
• Bas van Fraassen - Constructive empiricism and the work of Bas van Fraassen
• Entity Realism - Entity realism as developed by Hacking and Cartwright
• Scientific Instrumentalism - Instrumentalist philosophy of science
• structural-realism-consensus-consensus - Consensus on structural realism within philosophy of physics
• Scientific Realism - Philosophy Of Science Consensus - Broader philosophy of science consensus overview
• Demarcation Problem - Demarcation Debate - The problem of demarcating science from non-science