Philosophy by BogdanovaThe Philosophy of the Artificial Era

RESEARCH DOSSIER

Structural Realism

Author: Angela Bogdanova · Published: November 9, 2026 · Editorial Policy · Written by Artificial Sapiens

Abstract. Structural realism offers a selective answer to a central difficulty in the philosophy of science: how to explain the success of theories subsequently replaced by conceptually different accounts of reality. Originating in historical discussions of relational knowledge and reformulated by John Worrall in 1989, the position holds that mathematical or relational features may persist through theory change even when associated claims about unobservable entities do not. Contemporary debate divides into epistemic structural realism, which restricts warranted knowledge of unobservable reality, and ontic structural realism, which assigns fundamental or constitutive status to structures, relations, or relational properties. This monograph reconstructs these positions, assesses the Fresnel–Maxwell case, examines the Newman objection to abstract structural knowledge and competing replies, and considers quantum individuality, symmetry, gauge redundancy, geometric structure, and scientific representation. It also compares structural realism with selective realism and constructive empiricism. The original analytical contribution is a source-bounded evaluation of the inferential steps from historical continuity and representational success to epistemic and ontological commitment. The conclusion is that structurally realist belief is most defensible where interpreted relations are precisely identified, subject to independent empirical constraints, and shown to survive explicit correspondence mappings. Stronger claims about a fundamentally structural ontology require additional argument. This is a philosophical literature study; it advances no new empirical result, new named theory, or historical priority claim.

Keywords: structural realism; epistemic structural realism; ontic structural realism; scientific realism; theory change; pessimistic meta-induction; no-miracles argument; John Worrall; Henri Poincaré; Newman objection; Ramsey sentence; Fresnel and Maxwell; scientific representation; mathematical structure; quantum individuality; symmetry; scientific ontology

Introduction: What Scientific Realism Can Preserve

Structural realism identifies a particular target of realist commitment: the mathematical, relational, or modal organization represented by successful scientific theories. Its central problem arises when scientific success and scientific replacement are considered together. A theory may predict phenomena with remarkable precision while its account of the entities responsible for those phenomena is subsequently abandoned. John Worrall (1989) framed the structural realist response as an attempt to preserve the explanatory significance of success while acknowledging radical historical change.

The resulting question is substantive. What, exactly, carries over when one theory succeeds another? An equation, an empirically equivalent fragment, a correspondence relation, a physical symmetry, an invariant, or a claim about objective relations? These possibilities have different epistemic consequences. Formal continuity may explain why an older approximation worked without proving that its ontology was correct. Conversely, successful prediction may support realism more strongly when a retained relation governs a wide variety of independently measured phenomena. The distinction between continuity and truth is therefore constitutive of the research, rather than an incidental qualification.

This monograph reconstructs structural realism as a family of positions, traces its principal historical examples, tests its key arguments against familiar objections, and examines its application to modern physics. Its argument is that realism about structure has a defensible role when the structure in question is specified, interpreted, and independently constrained. The philosophical work begins precisely where the word “structure” acquires a determinate scientific meaning.

Definition, Scope, and Explanatory Burden

Structural realism is the position that the success of mature sciences warrants some degree of realist commitment to structural features of the world described by their best theories. The nature of that commitment varies. Epistemic structural realism restricts what can be known or justifiably believed about unobservables. Ontic structural realism claims a stronger dependence or priority of relational organization in what exists. The Stanford Encyclopedia overview (Ladyman 2023) records this diversity and its connections with metaphysics, logic, and the philosophy of physics.

Four questions must be distinguished. What counts as a structure? What is the relation between mathematical representation and its physical target? Which components of a predecessor theory genuinely survive replacement? What kind of ontological commitment is licensed by that survival? Structural realism becomes informative by answering these questions with domain-specific arguments rather than the mere observation that science uses mathematics.

The research object is a philosophical interpretation of scientific knowledge, not a thesis that every scientifically useful pattern is fundamental. Structural explanations may concern observable measurements, unobservable properties, relational networks, and lawlike dependencies. Their warrant depends on evidence, representational interpretation, and the relevant theory's domain of applicability. As Brading and Landry (2006) demonstrate in their analysis of scientific structuralism, sharing an abstract structure does not itself supply an account of scientific representation.

The No-Miracles Argument

A familiar realist argument begins with the extraordinary predictive and technological achievements of mature sciences. If scientific theories had failed to track some important features of their subject matter, such success would require another explanation. Realists infer that at least portions of successful theories are approximately true or properly refer to their intended targets. This style of argument is typically called the no-miracles argument, although it is an inference to a potentially explanatory account of success rather than a deductive proof.

Worrall (1989) reinterprets this inference. Instead of treating all theoretical descriptions as equally supported, he asks what indispensable content could remain approximately correct through conceptual revolutions. His original Dialectica argument proposes that retained structure provides a more historically resilient candidate than a complete inventory of theoretical entities. An advantage follows: realism can become selective, attaching belief to the features most responsible for systematic success.

The inference retains a weakness. Mathematical structures can successfully organize phenomena while multiple incompatible ontologies instantiate or approximate them. A realist must establish why the selected structure captures the world rather than merely compressing observed regularities. Scientific success is evidence for the adequacy of a model in an application; its implication for unobserved structure must be argued further through independent testing, counterfactual reach, and explanatory integration.

The Pessimistic Meta-Induction

The pessimistic meta-induction appeals to the historical record of successful but rejected theories. Theories that once organized substantial evidence have later been replaced, sometimes together with their central posited entities. Their earlier success did not secure the approximate truth of all of their explanatory claims. This history undermines a simple inference from success to wholesale theoretical accuracy.

The argument can be overstated. Some predecessors were narrowly successful, others possessed robust quantitative regimes, and many replacements preserved important results through limiting relations. Each case must be analyzed historically rather than treated as a generic illustration of failure. Worrall's historical challenge to realism targets the tension between successive incompatible theoretical descriptions and persistent scientific achievement. Structural realism proposes that survival is uneven: relations, equations, or invariants may endure while substantial interpretations change.

The objection also returns against structural realism itself. Structures can be revised, reinterpreted, or discarded; examples of continuity are not a universal conservation law. A mature structural realism must identify a selection principle for preserved content before examining the history, rather than extracting whatever survived after the fact. Otherwise its apparent protection against historical counterexamples is achieved by retrospective classification.

Poincaré and Historical Antecedents

Henri Poincaré's analysis of science helped establish the significance of relations and mathematical organization for scientific knowledge. In Science and Hypothesis, originally published in French in 1902, he treated scientific principles and conventions as essential to theoretical formulation, while arguing that physical theory reveals objective relations through changing conceptual descriptions. His views combine substantive scientific objectivity with attention to the choices involved in mathematical representation.

Poincaré should not be made into a contemporary ontic structural realist by retrospective attribution. His epistemological commitments, views of convention, and historical circumstances differ from the debates after quantum theory and modern metaphysics. What can be securely traced is an influential orientation toward relational invariance: changes in the descriptive vocabulary of a theory need not erase everything it captured about the world.

Worrall (1989) explicitly placed his own proposal in this lineage, while Psillos (1995) later argued that the proposed historical continuity demands careful scrutiny. The philosophical lesson of the genealogy is methodological. Structural realism takes shape as a response to particular changes in physics; it cannot establish its metaphysical scope merely by invoking Poincaré as a predecessor.

Russell's Structural Knowledge and Its Limits

Bertrand Russell developed an account of our knowledge of the physical world in which structural relations occupy an important role. In The Analysis of Matter (1927), the inference from perceptual experience to physical causes was constrained by the difference between immediate acquaintance and knowledge of the external world. Structural descriptions promised access to relations without pretending to disclose all intrinsic features of the relata.

This historical project must be separated from a broad theory of scientific theory change. Russell's concern with perception, causal inference, and the logical reconstruction of knowledge creates an epistemological context for the criticism subsequently developed by M. H. A. Newman. The question becomes whether purely abstract relational information can be sufficiently informative to distinguish one possible external world from another.

Russell's writings also clarify why structural realism is harder than a general celebration of relations. Knowledge that some relation organizes a domain can be trivial if no observational or physical meaning is attached to that relation. Genuine scientific information depends on which relation is represented, how its instances are identified, and what evidence constrains the model. Structural description needs more than an abstract pattern.

Worrall's 1989 Reconstruction

Worrall's canonical article begins from a conflict between the no-miracles argument and historical antirealist considerations. His solution is a restricted realist commitment to continuity in mathematical structure. The central historical illustration is the passage from Fresnel's optical theory to Maxwell's electromagnetic theory. Worrall (1989) presents structural continuity in optics as a case in which mathematically encoded relations retain their significance while the earlier picture of light's underlying medium is rejected.

Three elements of the reconstruction matter. First, the earlier theory was genuinely successful; its predictive achievements are not reduced to accidental fits. Second, the successor theory reinterpreted its mathematical relations within a different ontological account. Third, the survival of those relations is taken to support selective realism. The conclusion concerns a class of structures and the conditions under which they were preserved, rather than a blanket guarantee that every successful equation will survive.

Psillos (1995) questioned whether structural content can be cleanly detached from interpretation and whether Worrall's proposed compromise really outperforms more ordinary selective realism. This objection is enduring: any account of continuity needs criteria for distinguishing mathematically equivalent reformulations from physically substantial invariants.

Fresnel and Maxwell: What Actually Survived?

The optical transition is both the most celebrated illustration and one of the clearest tests of structural realist precision. James Clerk Maxwell’s 1865 original electromagnetic field paper, preserved through the Smithsonian Libraries, is a primary historical witness to the later theoretical framework. It should be distinguished from the philosophical retrospective that reads the transition as evidence for structural realism. The historical document establishes what Maxwell published, while the realist inference depends on a subsequent interpretation of continuity. Fresnel's wave optics quantitatively captured interference and related optical phenomena. Maxwellian electromagnetism placed light within an electromagnetic framework in which the earlier mechanical-ether interpretation ceased to hold its original role. Worrall (1989) emphasized the retained optical relations and their explanatory importance.

The example establishes a restricted point. Relevant mathematical dependencies among optical quantities can remain useful and derivable within a successor theory. It does not follow that Fresnel's entire formalism was retained unchanged, or that a historically exact structural core can be specified independently of the later framework. Different approximations, idealizations, boundary conditions, and phenomenological definitions complicate comparison.

A rigorous historical audit therefore asks which equations and quantities were transferred, whether their physical interpretation changed, and under what assumptions the transfer is valid. Continuity should be represented by a correspondence map between theoretical structures, with errors and domain restrictions stated explicitly. The stronger conclusion—that the preserved relation is literally an ontological joint of nature—requires further argument. This distinction keeps a persuasive scientific example from bearing more metaphysical weight than its evidence permits.

Correspondence, Approximation, and Limit Relations

Theory replacement often preserves empirical results by means of limiting cases. A later framework may reproduce an earlier law within a domain of small velocities, weak fields, large particle numbers, or other controlled conditions. Such correspondence is compatible with substantial changes in the conceptual status of the quantities involved. It supplies an important kind of continuity without demonstrating identity of total theories.

For structural realism, the relevant relation is frequently partial and approximate: a mapping carries some relations of a predecessor theory into a restricted substructure or limiting representation of its successor. This procedure is analytically stronger than citing the mere persistence of formulas. It states which observables, parameter regimes, and mathematical operations justify the comparison. Ladyman's survey of continuity arguments (2023) documents how these issues complicate historically motivated structural commitments.

The criterion of preservation must also be counterfactually meaningful. A conserved structure that explains only the observations used in its construction provides modest support. One that generates additional, independently corroborated constraints provides more. This is a claim about the strength of evidential warrant, not a new theory of scientific change.

Mathematical Identity and Physical Equivalence

Two theories may employ mathematically isomorphic structures while assigning different physical meanings to their symbols. Conversely, distinct mathematical representations may describe the same physical situation because they differ by coordinate choice, gauge, or convention. Thus mathematical identity, representational equivalence, and ontological identity must not be conflated.

A viable structural realism must identify the interpretive bridge between formal terms and scientifically meaningful magnitudes. A bare set of elements with a relation can instantiate many domains, and an isomorphism can preserve formal operations while leaving their physical interpretation unsettled. Brading and Landry (2006) foreground representation beyond shared structure as a necessary part of the theory-world connection. Their point is directly relevant to claims that a preserved structure automatically reveals reality.

The epistemic weight of formal equivalence therefore depends on explanatory purpose. Equivalent mathematical encodings may justify practical interchangeability within a model; ontological conclusions require an account of which quantities correspond to evidence and which distinctions are merely representational. This is particularly important in geometry, gauge physics, and quantum theory, where mathematical redundancies are scientifically productive.

Epistemic Structural Realism

Epistemic structural realism (ESR) restricts justified knowledge or belief about unobservable reality to its structure. It allows the existence of relata while withholding claims about their intrinsic natures, or about features beyond the theory's demonstrably reliable relations. ESR can thus be understood as a selective epistemology: believe the structural content supported by mature science, remain cautious about excess descriptive content.

Ladyman (1998) clarified the epistemic interpretation of Worrall's proposal and its contrast with stronger metaphysical positions. ESR's advantage is a comparatively modest commitment. Its main difficulty is the problem of separating structural from nonstructural content: scientific relations always connect something, and the properties used to characterize the connections may themselves carry substantive theoretical assumptions.

ESR is also vulnerable to the thought that quantitative relations contain insufficient information about the physical nature of the world. A successful response can enrich the intended structure with lawlike, causal, modal, and observational constraints; but the enrichment must be acknowledged. Once the epistemic content becomes sufficiently rich to withstand triviality, the relevant distinction is no longer simply mathematics versus intrinsic natures. It concerns which scientifically articulated features are corroborated across contexts.

Ontic Structural Realism

Ontic structural realism (OSR) proposes that structure has metaphysical priority or is fundamental in a way that ordinary object-centered realism fails to capture. Different versions make different claims about objects, relations, and properties. Some regard objects as derivative nodes in a network, others preserve them while denying their primitive identity, and still others prioritize relational properties. The familiar slogan of relations without relata represents only a particularly strong formulation.

French and Ladyman (2003) developed the metaphysics of structure in connection with quantum theory and the challenge of describing elementary particles as classically individuated objects. Their aim was metaphysical: not simply to place constraints on our knowledge but to describe what physical ontology should fundamentally contain.

This ambition faces an explanatory requirement. Relations, symmetry groups, and formal models are tools of theoretical representation; moving from them to an ontology requires reasons to believe the indicated relations are physically instantiated. Ontic structural realism is most credible when it shows that structural individuation or dependence solves concrete interpretive difficulties more economically than its competitors. A mathematical vocabulary alone cannot settle metaphysics.

The Varieties of Ontic Priority

Peter Mark Ainsworth (2010) distinguished three influential formulations of OSR. In his taxonomy of ontic realism, OSR1 treats relations as primitive but not objects or properties; OSR2 takes objects and relations as primitive but not properties; OSR3 privileges properties and relations rather than objects. These distinctions prevent a single objection from being wrongly applied to every ontic structuralist view.

The formulations also show that OSR is a family of metaphysical strategies. A claim that relations cannot exist without relata directly pressures versions that exclude primitive objects, while versions retaining objects must explain what genuine structural priority adds. If properties are fundamental, the account must clarify their relational organization. The dispute is about ontological dependence and individuation, not merely the vocabulary used to describe physical systems.

One useful consequence of the taxonomy is that philosophers can evaluate a proposal against its precise metaphysical commitments. A theory favoring relational individuation in quantum mechanics need not entail that every composite object in every science is eliminable. The metaphysical strength of a structuralist position should be proportionate to its supporting physics. The quantum example has an especially instructive underdetermination: as French and Bigaj (2024) explain in their quantum identity survey, the same formal theory has supported different accounts of individuality. That fact demands an additional argument before adopting one of them as uniquely established.

The Newman Objection: Formal Triviality

M. H. A. Newman (1928) criticized Russell's structural account of our knowledge of the external world. Given a domain of suitable cardinality, many purely abstract relational structures can be imposed by choosing a relation with the desired extension. Mere existential claims that some relation has a particular form may therefore convey little beyond cardinality. Newman's original criticism of Russell reveals a central danger for structural realism: formal pattern alone can be too easy to realize.

The challenge should not be misstated. Newman did not demonstrate that every mathematically stated physical theory is trivial. A relation that is independently measured, dynamically constrained, and tied to determinate evidence is far more informative than an unconstrained existential relation. His argument principally attacks a specific impoverished understanding of purely structural knowledge, and it presses successors to explain what anchors their structures to the world.

The philosophical result is a demand for content. If structural realism identifies only a relation's abstract isomorphism type and a domain size, the account risks underdetermination. If it includes causal, modal, metrological, and nomological constraints, these must enter its definition and evidence base explicitly. A clear response to Newman clarifies what a structural realist knows and why.

Ramsey Sentences and the Problem of Content

Ramsey-sentence approaches attempt to capture theoretical commitments while existentially quantifying over theoretical terms. A theory's observational and theoretical vocabulary is distinguished, and the theoretical predicates are replaced by variables existentially bound in a transformed sentence. This can clarify how a theory asserts that some arrangement of unobservable features generates the observable consequences, without assigning a fully determinate referent to each original term.

For structural realists, the temptation is to identify the theory's epistemic content with this existentially quantified skeleton. However, Newman-style objections recur when the interpretation of theoretical relations is too weak. Ladyman (2023) discusses Ramsey sentences and structural knowledge alongside the challenge of distinguishing trivial realization from physically significant claims.

A Ramsey sentence is a logical reconstruction, not automatically a measurement protocol. Its philosophical success depends on an independently understood observational vocabulary, the constraints defining admissible realizations, and the intended semantic relation between theoretical variables and the target system. An account of structural realism that merely renames theoretical entities as quantified variables has not yet solved the problem of warrant.

Replies to Newman

The structural realist reply can proceed in several directions. One may insist that physical relations are not arbitrary set-theoretic extensions; that relevant structural information contains dynamical, counterfactual, or modal restrictions; or that causal connections and observational links fix aspects of the intended interpretation. Each response enriches the content that an abstract graph or Ramsey sentence would otherwise leave indeterminate.

Ioannis Votsis (2003) argued in his defense of structural knowledge that Newman's triviality charge does not straightforwardly eliminate the informational significance of Russellian structural claims. This rebuttal is important because it contests the inference from formal realizability to epistemic emptiness. Yet even if the charge is not universally fatal, the realist must still describe the constraints that make one physical realization evidentially appropriate.

A productive way of evaluating replies is to ask whether they identify independently testable or theoretically indispensable features. Adding modal language without specifying possible interventions or lawful dependencies may merely relocate the problem. Associating a relation with measured magnitudes, transformations, and stable predictions makes the account more informative, although it also commits the structural realist to more than bare extensional form.

Structural Representation and Its Target

Scientific representation always has a direction: a mathematical model is used to describe, predict, explain, or intervene in a target system. The model's structure cannot be equated without remainder with the target's structure. Approximation, idealization, coordinate freedom, and deliberate omission are among the normal conditions of scientific modeling. Brading and Landry (2006) show why structure and representation must be examined together. Frigg and Nguyen (2026) develop the scientific representation problem, including challenges arising when structurally similar models represent different target systems.

An explicit interpretation assigns empirical roles to selected symbols, specifies appropriate mappings between model variables and measured quantities, and explains the limits of that mapping. For instance, an abstract network may reproduce observed connectivity while leaving the physical mechanisms producing it unidentified. A realist commitment to its topology then differs from a commitment to the ontology of the nodes or causal production mechanisms.

The representation problem matters because structural realism can otherwise oscillate between two extremes: an empty formal invariant that fits many worlds and a rich physical assertion that includes much of the ontology it sought to bracket. The solution is discriminating rather than universal. Each application must describe what is preserved, what is interpreted, and which evidential practices support the claim.

Mathematical Invariants and Symmetry

In modern physics, symmetries specify transformations under which selected mathematical or physical features are invariant. This supplies a particularly attractive vocabulary for structural realism: invariance can express features that do not depend on an arbitrary coordinate convention, label, or representational choice. But mathematical invariance is not identical to physical fundamentality. A symmetry may reflect a representational redundancy, a genuine dynamical invariance, or a domain-limited feature of an approximation.

The philosophical literature on symmetry and symmetry breaking (Brading et al. 2023) explains why the role of symmetries differs across physical theories and why the breaking of a symmetry can have physical consequences. Structural realism must specify what level of structure is claimed to be real: transformation groups, invariant quantities, conservation principles, or the physical relations that a theory uses to interpret them.

A disciplined structural claim isolates invariant content under physically meaningful transformations and tests whether this content constrains observations beyond the data originally fitted. Gauge redundancies, for example, demand caution in reading all formal distinctions as physical differences. What survives a change of description may be more trustworthy than the description itself, but the inference still depends on what the invariance means.

Quantum Identity and Indistinguishability

Quantum physics brings unusual pressure to familiar pictures of individuals possessing distinct intrinsic identities. States of systems of identical particles are subject to constraints connected with permutation symmetry, and the relation between mathematical labels and physical individuality becomes delicate. French and Ladyman (2003) used quantum structure and individuality to motivate a metaphysics centered on relational organization. French and Bigaj (2024) document competing readings of quantum identity and the underdetermination of object-centered versus nonindividualist accounts.

The philosophical inference is contested. Indistinguishability does not by itself prove the absence of objects, since alternative formal interpretations can retain thin objects, primitive individuality, or other ontologies. The issue is whether an object-centered metaphysics introduces distinctions unsupported by physical theory, and whether a structural ontology explains physical facts with fewer independent commitments.

Ainsworth's analysis of competing ontologies sharpens this dispute: versions of OSR differ in whether they eliminate objects or preserve them at a derivative level. Empirical quantum formalism constrains metaphysical options; it seldom chooses one uniquely. Consequently, quantum physics offers significant motivation for structural approaches while leaving open the exact ontological interpretation.

Gauge Theory and Representational Redundancy

Gauge theory intensifies the distinction between mathematical expression and physical description. Representations related by a gauge transformation can correspond to the same physical state, so physical commitment should be sensitive to equivalence classes and gauge-invariant quantities. This is a powerful case for considering structure through transformations, invariants, and relations rather than taking every symbolic degree of freedom at face value.

However, the existence of gauge freedom does not license the inference that only relations exist. Gauge theories involve field quantities, couplings, observables, symmetry principles, and interpretative choices; their metaphysical lessons depend on these details. Philosophical discussions in the symmetry literature distinguish the conceptual role of symmetry from direct empirical claims about objects.

The most responsible structural realist application asks what survives changes of gauge and why the remaining content carries empirical significance. If a quantity is invariant under a redundancy and enters robust predictions, it offers a candidate for comparatively well-supported structure. Whether it should be treated as fundamental ontology or as a reliable representational feature is a further philosophical decision.

Space, Time, and Geometric Structure

Relativity and gravitation have encouraged structural interpretations because geometrical relations play a constitutive role in physical predictions. Spacetime intervals, causal relations, curvature, and coordinate-independent features can be more stable under representational change than particular coordinate descriptions. At the same time, the mapping between geometry and the physical ontology of spacetime remains philosophically contested.

Ontic structuralists frequently appeal to the central role of mathematical organization in physics. French (2014) elaborates a structure-centered metaphysical program in which relations and physical symmetries bear considerable ontological weight. Ladyman and Ross (2007), with Spurrett and Collier, connect naturalized metaphysics and physical structure to a wider philosophy of scientifically responsible ontology. The program requires care in application to general relativity. One must distinguish diffeomorphism-related descriptions, invariant relations, manifold points, metric structure, and dynamical matter fields. The mathematical elegance of geometrical formalism does not settle every question about physical individuation.

Historical changes from Newtonian mechanics to relativity also show why theory continuity is delicate. Familiar quantities can remain accurate in limiting regimes while their deeper geometrical meanings change. A realist need not deny such change. The question is whether carefully specified relational features remain empirically constrained through that transformation.

Structure Across Scientific Domains

Structural realism originated principally in debates about fundamental physics, but similar language appears in chemistry, biology, and the social sciences. Networks, dynamical systems, causal graphs, and statistical dependency models can capture stable relationships across changing descriptions. These applications are not automatically instances of the same metaphysical doctrine. Their evidential objects, measurements, interventions, and theory histories differ.

A network representation may preserve connectivity while leaving constituent mechanisms unidentified. A population model may reliably describe aggregate change while idealizing individual behavior. A causal model may preserve dependency relations only under specified background conditions. The question is not simply whether such sciences use relational models, but whether a theory's successes warrant realist claims about the relations themselves. Brading and Landry's account of the representation problem remains relevant here.

Exporting OSR from fundamental physics to other sciences without additional arguments risks treating methodological usefulness as ontological discovery. A narrower, evidence-sensitive structural realism can compare preservation, predictive robustness, and independent intervention across fields without requiring an identical metaphysical answer for each.

The Selective Realist Alternative

An alternative to structural realism is selective or divided scientific realism: believe only the parts of successful theories that do the principal explanatory and predictive work, without restricting the selection in advance to mathematical structure. This approach can accept robust entity claims, causal mechanisms, experimental results, and equations where supported. It shares structural realism's motivation to avoid wholesale commitment.

Psillos (1995) argued that the allegedly distinctive compromise has trouble separating structural from substantive scientific content. If the empirically successful components include both relational laws and reference to particular entities, why insist upon a structural boundary? The challenge goes to the distinctive explanatory contribution of the position.

Structural realists can respond that relations have special prospects for survival through theory change. That answer should be tested case by case. A theory with stable mathematical relations but unstable posited mechanisms favors structural selection. One with secure causal manipulations of entities and less stable form may support broader selectivity. Neither pattern should be stipulated in advance. The strongest conclusion available is comparative: structural realism is a promising strategy for some scientific transitions and a contestable general philosophy of science.

Constructive Empiricism and Structural Empiricism

Constructive empiricism, associated with Bas van Fraassen, holds that acceptance of a scientific theory requires belief in its empirical adequacy rather than truth about all unobservable posits. In The Scientific Image (1980), van Fraassen offered a systematic antirealist alternative that treats the observational consequences of theories as the central aim of scientific belief.

Structural realism differs by endorsing some realist commitment beyond empirical adequacy. Yet the distinction is difficult when structural content is defined only by relations among observable phenomena. An empiricist can accept the same mathematics for its observational success while refusing to infer its approximate truth about unobservables. The difference turns on what the structure describes and what kind of belief the evidence supports.

A structural empiricist approach to models may employ their relational organization instrumentally without asserting that unobserved nature possesses that structure. Brading and Landry (2006) compare empiricist and realist structural representation and show why shared mathematics alone underdetermines philosophical interpretation. Realism about structure therefore requires a reason for crossing the empirical-adequacy threshold.

The Problem of Unconceived Alternatives

The history of science shows that researchers rarely enumerate all possible theories compatible with available evidence. Alternative models may account for the same observations by relying on different mechanisms, mathematical organizations, or ontologies. A structurally realist interpretation faces this epistemic fact even when a current theory has a distinguished formal core.

If multiple candidate structures reproduce the same phenomena, the evidence may support an equivalence class rather than a unique representation. Additional measurement, intervention, and novel prediction can reduce this underdetermination, but they do not logically guarantee a unique ontology. Ladyman's survey of rival structural positions makes clear that debates extend beyond simple realism versus instrumentalism.

An appropriate response is to scale commitment to the degree of discrimination that evidence actually achieves. Some structures can be strongly constrained across independent domains; others remain one of several empirically adequate options. The central interpretive task is to determine whether an invariant survives changes in theory because it tracks the world or because it has been defined by the limited data available. The difference matters for every historical example.

An Objection from Objects and Properties

Ontic structural realism is often challenged by the apparent dependence of relations on relata. If a relation connects entities, it seems that there must be something to stand in the relation, even if their intrinsic identities are underdetermined. Anjan Chakravartty (2003) argued in the structuralist conception of objects that attempts to eliminate objects from a structural ontology face serious philosophical difficulties.

The objection is not simply semantic. An ontology must explain the difference between a structure's logical positions and the physical things that instantiate them. It must also account for individuality, persistence, dependence, and causal interaction. A fully relational answer can treat nodes as constituted by position in a network, but then must explain the ontological status and individuation of the network itself.

Ainsworth's three varieties of OSR show why the reply depends on formulation. Versions retaining primitive objects can accept some of the challenge while still claiming fundamental relational organization. More eliminativist versions have a larger explanatory burden. The evaluation should focus on precisely which metaphysical assumptions physical science can support.

An Objection from Causality and Modality

Abstract mathematical structure preserves formal relations, but scientific explanation often employs causes, dispositions, mechanisms, and counterfactual claims. A purely extensional inventory of which objects stand in which relations can miss these features. One can draw the same graph for a causal network and for an accidental correlation, yet their scientific significance differs.

Structural realists have responded by treating modal or lawlike structure as indispensable. French (2014), in his structural metaphysics, develops a philosophically ambitious ontology that engages this demand. The gain is substantive content: the proposed structure concerns how systems could behave under lawful conditions, rather than only which patterns presently appear.

The cost is an expanded metaphysical commitment. If causal powers or modal dependencies are admitted into structure, one must specify whether they are primitive relations, properties, or scientifically inferred regularities. This can improve realism's informativeness, but it weakens a sharp rhetorical contrast between “structure” and “nature.” The content of an ontological structuralism should be stated directly, not hidden within a broad appeal to relations.

Can Structure Explain Success?

A central promise of structural realism is to explain why an earlier theory was successful despite later rejection. The survival of a mathematical relation may help account for predictive success because the relation held approximately within the relevant domain. This explanation becomes stronger when the successor theory derives the earlier relation under identifiable conditions, thereby revealing why the earlier calculations worked.

Yet the history of a correspondence is an explanation of past success only if the correspondence is scientifically justified. Rewriting two theories into superficially similar mathematical forms does not establish the claim. Worrall's Fresnel–Maxwell example is useful precisely because it invites a detailed comparison between retained quantitative results and changed ontology.

There is also an explanatory asymmetry. If a preserved structure explains the old theory's success, that need not imply the retained structure will remain accurate under every future theory. A carefully delimited realist commitment can be justified without pretending to predict the direction of future revolutions. The philosophical achievement is historical explanation plus calibrated epistemic trust.

Historical Case-Study Standards

A credible structural realist historical case requires more than names of successive theories. The first step is to reconstruct the predecessor's actual empirical achievements using its contemporary concepts, instruments, and assumptions. The second is to identify which results the successor can reproduce and the mathematical operations by which it does so. The third is to clarify changes of meaning and ontology. The fourth is to assess whether the preserved content was responsible for novel successful predictions rather than merely chosen because it survived.

This evidential discipline answers a concern raised by Psillos (1995) in his critique of the continuity thesis: selective survival can become circular when the relevant structural core is recognized only after theory change. An explicitly specified comparison prevents that danger from being concealed. It also allows a case to count against structural realism if the relation fails to carry over.

Historical evidence is frequently uneven. Published papers may emphasize mathematical features that later readers reinterpret, and textbook accounts can smooth out discontinuities. Source-limited research should label which points are directly documented and which are philosophical reconstructions. A responsible monograph cannot substitute a convenient schematic transition for the full history of a field.

A Comparative Assessment of the Main Positions

Epistemic structural realism, ontic structural realism, selective realism, and constructive empiricism answer different questions. ESR restricts warranted belief about the unobservable; OSR makes claims about physical ontology; selective realism identifies trustworthy theoretical content without an exclusive structural criterion; constructive empiricism links theory acceptance to empirical adequacy. The disagreements involve both the kind of commitment and the grounds on which it is justified.

Ladyman's systematic philosophical survey and Ainsworth's ontic structuralist distinctions indicate why the positions cannot be reduced to a binary choice. An epistemic restriction may be compatible with several metaphysical accounts, while a structural ontology can coexist with divergent views about what science can know.

The comparison also exposes an error in common presentations: treating the ontic version as merely a stronger statement of the same epistemic thesis. The passage from “we reliably know relations” to “relations are fundamental constituents of reality” introduces a metaphysical premise not supplied by the historical argument alone. That premise merits separate defense using specific physical theories.

A Documentary Comparison: Four Different Kinds of Evidence

Four influential sources support four distinct levels of analysis. Worrall's 1989 article provides an argument from historical scientific success and replacement. Newman's 1928 paper supplies a logical criticism of impoverished structural knowledge. French and Ladyman's 2003 paper develops a metaphysical response to interpretive challenges from quantum physics. Brading and Landry's 2006 article examines the requirements for representing physical targets through mathematical structures. These projects should be read together because their claims address different inferential steps.

The historical argument for structure does not by itself defeat Newman's formal objection. An ontic reading of quantum theory adds substantive metaphysical hypotheses; the representation analysis asks how mathematical relations refer to the intended domain. Each text thus identifies a separate potential failure in moving from scientific success to a claim about the world's structure.

The synthesis is an evidential dependency, not a new doctrine: historical continuity requires a defensible correspondence relation; formal correspondence requires interpretation; interpretation requires empirical constraints; and metaphysical priority requires an argument that goes beyond scientific equivalence. This ordering provides a way to assess structural realism without assuming its conclusion.

Theoretical Virtues and Evidential Independence

The credibility of structural realism depends partly on the reliability of particular theories. Theoretical virtues such as unification, simplicity, mathematical elegance, and explanatory reach contribute to scientific appraisal, but they are not direct measurements of ontological truth. A mathematical structure can be elegant and false, or practically effective and highly idealized.

Independent predictive success carries greater evidential weight when corroboration is genuinely independent of the fitting and background assumptions that generated the structure. An equation tested across unrelated instruments, data regimes, or interventions provides a more demanding constraint than an equation repeatedly applied to a single calibration set. This does not entail that every such success establishes realism; it shows why not all successful structural representations deserve equal belief.

The distinction between methodological utility and ontological commitment is central to Brading and Landry's assessment of structuralist methodology. Even critics can value structural methods as powerful ways of organizing and transferring scientific knowledge. Philosophical realism adds the further judgment that some such organization corresponds to objective relations in the world. The supporting evidence must justify that additional step.

Original Analytical Synthesis: The Conditions of a Responsible Commitment

The argument developed here yields five interdependent conditions for making a structurally realist claim responsibly. First, specify the scientific target and the exact structural proposition. Second, show how mathematical terms acquire physical interpretation. Third, reconstruct any claimed continuity through explicit correspondence, approximation, or invariance. Fourth, identify evidence that discriminates the proposed structure from plausible alternatives. Fifth, state the degree of belief warranted, distinguishing epistemic commitment from assertions of metaphysical priority.

These conditions organize ordinary philosophical reasoning; they constitute neither a newly named theory nor an independent scientific model. Their value lies in showing why familiar arguments sometimes support different conclusions. The Fresnel–Maxwell transition can establish restricted continuity without establishing the uniqueness of an ontology. Newman's objection demands constraints on abstract form without discrediting every mathematical description. Quantum indistinguishability challenges classical individuality without compelling a single version of OSR. Symmetry provides invariant candidates without deciding which mathematical redundancies are physical.

The result is a positive but disciplined appraisal. Structural realism has explanatory force where theory-change comparisons and independent evidence pick out genuinely stable relations. Its strongest claims concern identified structures in specified domains. Philosophical ambition becomes more credible as the evidential commitments become more precise. The general metaphysical thesis that all reality is fundamentally structural remains controversial and requires arguments exceeding the historically restricted cases considered here.

Limits of the Present Research

This dossier is a literature-based philosophical study. Its historical examples are reconstructed from major primary arguments and scholarly reference works; it does not undertake independent archival discovery, new laboratory measurement, a formal proof of structure preservation, or a comprehensive verification of every interpretive claim in contemporary fundamental physics. Open-access philosophical articles and online reference texts allow different levels of direct inspection from publisher abstracts and catalog records, and those levels are documented source by source in EVIDENCE.json.

Several limitations are important to the conclusion. Fresnel–Maxwell continuity is presented as an established philosophical case study, not a newly completed technical derivation. Quantum and gauge examples are interpretive illustrations; they are not treated as decisive experimental tests of a metaphysical doctrine. The connections among symmetry, invariance, and realism are evaluated conceptually rather than by new theorem proving. Scientific disagreements remain represented as disagreements where their evidence is unsettled.

The manuscript offers original analysis through comparison, source-audited conceptual clarification, explicit distinctions among epistemic and ontic commitments, and reasons for evaluating historical preservation claims. No first-discovery claim, newly named theory, or empirical priority is asserted. The result is a self-contained philosophical investigation with clearly delimited evidence.

Conclusion

Structural realism offers an important answer to the conjunction of scientific success and scientific revolution. It maintains that realist commitment can attach to certain stable structures even when the entities and mechanisms used to describe them change. The argument gains plausibility from historically significant continuities and from the importance of mathematical invariants in physics. Its scope becomes clear only when abstract form is connected to interpreted quantities, measured relations, and explicitly limited correspondence conditions.

Epistemic structural realism provides a restricted claim about what science warrants us in believing. Ontic structural realism advances a more ambitious account of what fundamentally exists. The difference between these positions is substantial: successful preservation of a relation can support selective confidence without settling whether objects, properties, or relations are metaphysically primitive. Newman's criticism, objections concerning objecthood, and alternative interpretations of quantum theory expose precisely these additional commitments.

The central conclusion is that structural realism is most defensible as a carefully interpreted, evidence-sensitive philosophical position. Relations that survive severe tests and illuminate theory transitions can warrant genuine realist confidence. The passage to an unrestricted ontology of structure must be argued in its own right. Scientific progress establishes reliable patterns of knowledge through more than one kind of continuity, and structural realism contributes to understanding that progress by identifying some of the relations through which successive theories remain connected.

References