Theory Change

Introduction: Theory Change as a Historical and Epistemic Problem

Angela Bogdanova
Published by Aisentica Development
Research Dossier · Philosophy of Science · American English (en-US)
Editorial source date: November 7, 2026
Reserved canonical for subsequent release: https://philosophybybogdanova.com/dossiers/theory-change/

Theory change is the revision, displacement, or reorganization of the concepts, explanations, models, and investigative commitments through which scientific research interprets phenomena. This definition includes local theoretical amendments and far-reaching disciplinary transformations. The central problem is to explain when such changes are rationally warranted and how they can yield scientific progress. Kuhn's account of scientific revolutions famously challenged the cumulative image of science, while the realist debate about theory succession asks how successful claims survive replacement (Kuhn, 1962; Psillos, 2022).

Historical descriptions must distinguish the occurrence of change from its normative evaluation. A scientific community may replace a set of explanations because of new observations, revised instruments, conceptual invention, organized criticism, or social circumstances. The mere fact of replacement establishes neither improvement nor the truth of the successor. Judging the transition requires specifying relevant aims, available alternatives and evidential reasons.

This monograph reconstructs the arguments of Duhem, Popper, Kuhn, Lakatos, Laudan and Feyerabend, compares contemporary debates on realism and observation, and interprets three historical cases. Its authorial contribution consists in analyzing the variable fates of earlier achievements during change. The inquiry advances an original comparison and justified conclusions within established philosophy of science; it establishes no new named theory or empirical law.

The Identity of a Scientific Theory

Scientific theories are organized explanatory commitments, not merely lists of individually testable sentences. Their identity may depend on theoretical principles, exemplifying models, mathematical representation, experimental practices, criteria of accuracy and a vocabulary for classifying the domain. Boyd and Bogen's study of theory and observation demonstrates that observations become evidence through processes and background information whose roles are open to investigation (2026).

At least five forms of change deserve separate description. Propositional revision alters the sentences held to be true. Representational revision changes models or idealizations. Semantic revision alters the meaning or extension of a category. Methodological revision changes the procedures judged appropriate for testing and inference. Institutional revision changes the collective organization of research. These forms often appear together; their analytical separation helps identify the actual causal and evidential history of an episode.

Suppose a physical theory keeps its central equation but its investigators replace an instrument's calibration. The change could improve measurement without replacing the underlying theory. Suppose instead the theory's fundamental concepts are redefined while a widely used approximation remains numerically accurate. That transformation is scientifically substantial even if many calculations remain familiar. The unit of description must therefore be chosen before a judgment about continuity is made.

Change and Progress: Two Different Judgments

A discipline can undergo major conceptual change without thereby improving its explanations. Conversely, an accumulation of reliable measurements can improve knowledge while leaving its leading theoretical principles stable. Niiniluoto's analysis of scientific progress distinguishes descriptive development from progress, which evaluates development relative to epistemic or other aims (2024). These different judgments explain why a single episode can be viewed as advancement in one respect and loss in another.

Candidate criteria include predictive accuracy, explanatory unification, resolution of significant empirical problems, resistance to error, truthlikeness and support for successful intervention. The criteria can diverge. A new theory may yield more precise forecasts in a restricted domain while making a formerly transparent explanation harder to construct. Another may explain more phenomena without yet providing accurate quantitative predictions.

The appropriate philosophical question is not whether progress is a universal label for scientific change. It is which achievements improved, by what standards, and on what evidence. Without that specification, historical applause for a successor replaces inquiry into its merits. Progress becomes an intelligible claim when the comparison identifies its object, reference class and relevant uncertainty.

Cumulative Accounts and Their Limits

The cumulative picture portrays science as an expanding store of facts and truths. This image captures the preservation of experiments, methods, observations and some increasingly precise mathematical relationships. It also makes intelligible the long-term success of scientific work across generations. The historical discussion of revolutions documents reasons to question a universal cumulative pattern, and the comparison of progress concepts distinguishes accumulation from other standards of improvement (Nickles, 2017; Niiniluoto, 2024).

The key difficulty is that later theories sometimes change what counts as the relevant object, property or explanation. An experiment performed under one vocabulary can later be described using incompatible theoretical categories. The observation may remain valuable while an earlier inference about what it disclosed is discarded. Accumulation of data is not equivalent to accumulation of every interpretive claim.

An adequate account accordingly asks what persists. Does a quantity retain a measurement procedure? Is a numerical prediction approximately reproduced? Does the successor conserve a causal relation or merely offer new equations for the same phenomena? Precise continuity is a historically grounded relation among specific achievements, not a presumption that whole theories accumulate unchanged.

Duhem and the Structure of Empirical Testing

Pierre Duhem argued that physical hypotheses generally confront experiments in networks of assumptions. A predicted result follows from the focal theory together with initial conditions, assumptions about instruments and supporting principles. A failed prediction therefore establishes a difficulty for the conjunction but does not logically identify a unique offending member. Ariew's analysis of Duhem's philosophy and Stanford's account of empirical underdetermination clarify this methodological difficulty (Ariew, 2026; Stanford, 2023).

The practical implications are substantial. A scientist facing an anomaly may recalibrate apparatus, alter an auxiliary hypothesis, refine an idealization or reconsider the central explanatory structure. These choices are not automatically equivalent. Some repairs possess independent experimental support and lead to new risky tests; others preserve conformity to observations by reducing the possibility of future criticism.

A historical explanation must reconstruct this dependency structure. It should record which assumptions were needed for the inference, which were independently checked, and what each proposed revision explained. Duhem's insight supports methodological care without requiring the conclusion that empirical evidence is useless. A result can strongly challenge a theoretical system even while leaving several rational routes for revision.

Popper: Criticism and Falsifiability

Karl Popper placed public criticism and possible falsification at the center of methodological appraisal. A theory that makes informative predictions exposes itself to evidence that could count against it. The history of Popperian accounts of progress provides the context of his critical rationalism, while the discussion of test dependencies identifies the challenge that auxiliary hypotheses pose for simple refutation (Niiniluoto, 2024; Stanford, 2023).

A prediction is usually derived from a package rather than a single law. Failure of the derived statement does not establish whether the theory, background conditions or observational report is responsible. Popper's logical point about incompatibility should therefore be distinguished from the methodological work of verifying an apparent counterexample. Evidence gathering and critical discussion are integral to deciding which claim to revise.

The most durable Popperian contribution to theory change is the demand for exposure to error. A modification should make the theory better answerable to empirical criticism. At the same time, the timing of replacement remains a historical question. Scientists often have reasonable grounds to retain a productive framework while assessing an anomaly and developing an alternative. Risky testing provides a norm of criticism rather than a universal timetable for abandonment.

Kuhn and the Historical Turn

In The Structure of Scientific Revolutions Kuhn presented scientific development through normal and revolutionary episodes, emphasizing research traditions as the context in which observations and puzzles acquire their significance. Bird's reconstruction of Kuhn's argument describes normal science as disciplined problem solving whose success depends on shared exemplars and expectations (2025). Kuhn's book on revolutionary science changed the agenda of historical philosophy of science by questioning how fixed methodological rules could describe actual transitions.

The historical turn challenged a narrative in which scientists confront neutral data and immediately discard the first falsified proposition. Scientific training gives researchers reliable techniques, standards for successful work and expectations about legitimate problems. These intellectual resources make cumulative improvement within a disciplinary framework possible, even as they can make unfamiliar alternatives harder to comprehend.

This analysis does not require that all sciences experience an identical sequence of stages. It does require careful reconstruction of communities, disciplinary achievements and changes in what counts as a useful solution. A verdict about rationality that ignores the problems available to historical actors risks grading their decisions according to evidence and concepts acquired much later.

Normal Science and the Productive Force of Stability

Normal science is often caricatured as intellectual conformism, yet in Kuhn's account it represents demanding, cumulative puzzle solving within a framework regarded as promising. The essays on tradition and scientific change show why sustained commitments can make technically difficult inquiry possible (Kuhn, 1977). Research teams do not ordinarily reinvent basic categories whenever a measurement diverges from expectation.

Stability makes reliable calibration, reproducible technique, shared education and specialized mathematical work possible. Over many cycles, a community may produce predictions and experiments of exceptional precision. The same investment can create vulnerability: anomalous phenomena may be regarded as isolated puzzles even when they suggest that the framework's deepest commitments require revision.

The philosophical challenge is to distinguish fruitful persistence from institutional insulation. The answer depends on what the tradition continues to accomplish and whether its defenders consider serious alternatives. A prolonged unresolved anomaly is not proof that continued research is irrational. Neither is the historical productivity of a framework proof that every proposed defense is well supported.

Anomalies, Crisis, and Alternative Frameworks

An anomaly becomes decisive through a history of observation, theoretical response and rival development. Many discrepancies are removed by improved measurements, refined approximations or newly specified conditions. Others persist and undermine a central expectation. Bird's explanation of Kuhnian crisis and Nickles's assessment of revolutionary variety caution against assigning one universal anomaly threshold (Bird, 2025; Nickles, 2017).

A competing framework changes the meaning of the anomaly. A result that once appeared merely inconvenient may become valuable evidence because an alternative predicts it naturally. Conversely, a rival may initially leave other well-confirmed results unexplained. The rational decision is thus comparative and temporal; investigators must weigh unresolved liabilities against emerging strengths.

To reconstruct a crisis, historians should identify when an observation became reliable, which theoretical options were available, and how scientists justified their assessments. Retrospective narratives often make eventual victory look inevitable. Reintroducing the uncertainties of the actual decision period clarifies why thoughtful investigators might reasonably have disagreed.

The Problem of Scientific Rationality after Kuhn

Kuhn's account is often interpreted as undermining rational comparison because scientists judge theories through values that do not mechanically determine a unique decision. His discussion of objectivity and theory choice addresses the importance of accuracy, consistency, scope, simplicity and fruitfulness. The analysis of Kuhn's mature position explains why shared standards can guide reasonable judgment without operating as an algorithm (Kuhn, 1977; Bird, 2025).

Values can conflict in application. A more accurate theory may be mathematically less tractable, while a conceptually elegant successor may initially possess fewer verified predictions. Scientists may also judge differently how much importance to give to future discovery. Such disagreement is compatible with a common commitment to objective argument about evidence.

The conclusion is neither automatic relativism nor guaranteed uniqueness of appraisal. Reasons are public, discussable and improvable even when several courses of action remain rational. A historian who records only the winner erases this structure; one who records reasons, alternatives, disagreements and subsequent tests explains how a rational transition developed.

Incommensurability and the Changing Language of Nature

Incommensurability concerns limits to the translation and comparison of conceptual frameworks whose terms are embedded in different classifications and explanatory commitments. Oberheim and Hoyningen-Huene's analysis of taxonomic incommensurability distinguishes this problem from a blanket claim that rival theories cannot be compared at all (2024). Kuhn's later work concentrated on localized difficulties of translation, especially when categories cross-classify natural kinds.

Scientific terms can retain their spelling while changing inferential roles. Newtonian and relativistic uses of mass and time do not carry identical theoretical commitments. In chemical history, names for gases, elements and combustion processes changed along with assumptions about their composition. Historical reconstruction must interpret such terms in their original problem settings.

A partial failure of translation does not prevent scientists from comparing predictions, shared measurements, experimental successes and problem-solving capacities. It may demand a careful account of how the rival descriptions relate to particular observations. The philosophical gain is precision: comparison is neither a simple synonym substitution nor an impossibility. It is an argument whose translation assumptions need to be stated.

Feyerabend and the Epistemic Value of Alternatives

Paul Feyerabend defended theoretical pluralism because entrenched concepts can influence what researchers are disposed to see or accept as evidence. The intellectual history of Feyerabend traces his development from critical rationalism to challenges against universal methodological rules (Oberheim, 2025). The comparison of incommensurability accounts documents his important differences from Kuhn (Oberheim and Hoyningen-Huene, 2024).

The constructive contribution of pluralism is that rivals can generate experimental questions a dominant theory leaves unasked. A competing conception may expose hidden assumptions in familiar measuring practices or suggest an unfamiliar interpretation of an old result. Inquiry benefits from imaginative variation because the accepted contrast class is rarely complete.

Pluralism does not in itself establish which theory is correct. The existence of alternatives creates occasions for criticism; their eventual epistemic standing depends on evidence, explanatory reach and contestable reasoning. The stronger lesson is to permit the search for rivals while maintaining serious standards for their appraisal. Such openness can preserve the possibility of progress when normal practice has lost the ability to interpret a central anomaly.

Lakatos and Scientific Research Programmes

Imre Lakatos treated sequences of theories, rather than isolated hypotheses, as important units of appraisal. His research-programme methodology distinguishes a relatively protected theoretical hard core from an adjustable protective belt of auxiliaries. Musgrave and Pigden's explanation of empirical progressiveness shows why a programme advances through new content that receives corroboration, rather than simply by accommodating already-known evidence (Lakatos, 1970; Musgrave and Pigden, 2026).

A programme may remain worth pursuing despite anomalies because particular adjustments develop a sequence of successful novel tests. A degenerating sequence repeatedly explains past surprises without producing comparably informative corroboration. These are judgments about trajectories rather than single observations, and they invite a more historically sensitive account of scientific persistence.

The challenge is prospective timing. A research programme may look unpromising before an important discovery, and a successful programme may later stagnate. Lakatos offers no mechanical expiration date for unproductive research. A responsible application therefore documents a chronology: what was predicted, when the prediction was made, what was observed, and how the programme's rivals performed. Without that chronology the labels progressive and degenerating risk becoming retrospective praise and blame.

Laudan and the Evaluation of Problem Solving

Larry Laudan's book on scientific growth proposed that the rational progress of science can be assessed through the empirical and conceptual problems that research traditions solve (Laudan, 1977). Niiniluoto's survey of problem-solving accounts discusses both the attraction of this approach and questions about how problems should be weighted (2024).

A problem-oriented appraisal begins with historically significant difficulties rather than a speculative measurement of distance from ultimate truth. A tradition can improve by explaining an anomaly, resolving inconsistency among its assumptions or integrating previously unrelated phenomena. It can also lose standing by generating new conceptual conflicts or abandoning successful explanations without adequate replacements.

The philosophical issue is how to count success fairly. Not every solved problem has equal importance; some are narrow technical puzzles and some demand reconstruction of fundamental assumptions. A later historian should identify problems recognized by the participants, describe the criteria used for success and avoid assigning achievements retroactively. Laudan's approach shows how progress may be rationally argued from documented accomplishments, while leaving room to contest what counts as a satisfactory solution.

Evidence, Bayesian Updating, and Historical Context

Bayesian confirmation offers a formal language for comparing the impact of evidence on competing hypotheses. A result is especially discriminating when it would have been considerably more likely under one available hypothesis than under its rivals. This insight strengthens the demand for comparative tests. Yet the analysis of unconceived alternatives explains why the hypotheses considered at a historical moment may be incomplete (Stanford, 2023).

The approach also faces choices about prior probabilities, likelihoods and the interpretation of evidence. Scientific revolutions can change the categories through which outcomes are described, making direct comparison technically difficult. A historical analysis that merely applies Bayes's theorem after the event does not show which contrasts scientists actually entertained.

A more informative reconstruction asks what each contender predicted beforehand, which measurements were feasible and how results changed contemporary judgments. Probabilistic analysis can reveal assumptions and clarify evidence, but it cannot replace the documentary history of creating, specifying and testing rival theories. Its contribution is the structure of discriminating support rather than an algorithm that guarantees adoption.

Scientific Realism and Historical Replacement

Theory change raises a serious question for realism: how can science be trusted if many previously successful theories have been rejected? Psillos's discussion of realist responses examines the pessimistic meta-induction and arguments for selective preservation of the constituents that contributed to success (2022). The historical puzzle becomes more informative when one distinguishes the claims discarded from the operations, relations and mechanisms retained.

Successful theories can contain elements later judged false. Sometimes they survive as approximations in restricted regimes; sometimes a mathematical relationship retains predictive utility while its original explanation is rejected. Selective realists attempt to explain why successful components deserve continued confidence. Such claims require detailed case evidence about the relationship between the retained component and the earlier success.

The historical skeptic rightly cautions that even today's most sophisticated interpretations are revisable. Yet revisability does not make all present theories equally supported. Realist argument is strongest when it identifies the exact scope of continuity and refrains from treating an old theory's success as confirmation of every proposition it contained.

Underdetermination and Unconceived Alternatives

Evidence often fails to select a unique theory from every possible alternative. Stanford's discussion of underdetermination distinguishes different forms of this problem and emphasizes the historical role of alternatives scientists had not conceived (2023). An accepted account can be superior to known rivals and still face future challenge from a theory that organizes the domain in an unforeseen way.

This situation does not license equal confidence in any imaginable hypothesis. Available explanations differ in their tested consequences, internal coherence, simplicity, explanatory power and response to counterevidence. Rational comparison is conducted within an actual evidential and conceptual setting. Philosophical fallibilism concerns the possibility of further improvement, not the suspension of every current judgment.

An important historical consequence is the need to reconstruct the available contrast class. Scientists of a period should be assessed according to what was known and could reasonably be investigated, not according to alternatives formulated decades afterward. Acknowledging unconceived rivals also supports practices of critical dialogue and conceptual exploration that may widen the range of future tests.

Measurements, Observation, and Cross-Theory Evidence

Observation is made through instrumentation, calibration, selection and interpretation. Boyd and Bogen's account of experimental mediation examines these processes and their implications for the relation between theory and evidence (2026). The fact that a measurement depends on assumptions does not mean its recorded pattern disappears when those assumptions change.

A geological magnetic record, an astronomical observation or a chemical mass measurement can be reinterpreted as theoretical language develops. The historical task is to identify which parts of the recording procedure remain stable, which calibrations change, and how newly available theories explain the resulting pattern. Independent cross-checks can make measurements robust despite important conceptual differences.

There is a danger of circular appraisal when a theory receives support solely from procedures whose validity presupposes the disputed theory in the relevant respect. The remedy is a clear dependency audit and, where possible, independent instruments or alternative measurement pathways. The significance of data rests in their capacity to constrain competing explanations; the existence of theory-dependent interpretation is a reason to investigate how that constraint operates.

Mathematical Correspondence and Domain-Limited Retention

When a successor theory reproduces successful predictions of a predecessor under specified conditions, continuity can be scientifically valuable even if the two theories make incompatible foundational claims. The discussion of conceptual discontinuities cautions against treating numerical agreement as proof of semantic identity (Oberheim and Hoyningen-Huene, 2024).

Classical physics provides a familiar case. Newtonian methods remain useful for many calculations where their approximations are adequate, even as general relativity reorganizes the theoretical description of gravitation and spacetime. A restricted approximation has a definite domain; calling the older framework simply true or false without naming that domain loses an important part of scientific practice.

A serious continuity claim should identify the variables related, the limits of the approximation and the observations recovered. When a historical description supplies none of these, the assertion that a successor 'contains' its predecessor remains ambiguous. Preserved mathematical performance is a significant achievement, but it must be distinguished from preservation of the earlier ontology or full explanation.

Theory Change in Chemistry: Oxygen and Phlogiston

The chemical revolution transformed explanations of combustion and the classifications by which substances were studied. The historical discussion of oxygen chemistry documents the importance of phlogiston, changing experimental interpretations, Lavoisier's account of oxygen and the persistence of caloric in his thinking (Weisberg, Needham, and Hendry, 2026).

Historical actors did not suddenly acquire contemporary chemical ontology. Experimental records produced under phlogiston-based accounts retained significance after their interpretive scheme changed. Weighing and chemical preparation acquired new explanatory roles, while the new chemistry still contained commitments later revised. The revolution therefore combined genuine advances with historically bounded errors.

This case supports a distinction between measurement retention and theoretical retention. A mass measurement can remain a useful observation even when the explanation attached to it is replaced. It also shows why a binary description of revolution is inadequate. Lavoisier's work established a new conceptual organization without immediately eliminating all obsolete assumptions. The reasons for preferring a framework must be assessed within the evidence and vocabulary then available.

Theory Change in Geology: Seafloor Spreading

The twentieth-century reorganization of geoscience brought marine magnetic data, crustal age and large-scale movement into an integrated explanation. The U.S. Geological Survey's description of magnetic stripes records alternating magnetic polarity patterns, while its account of seafloor development relates them to the processes of spreading. The National Research Council's historical account of plate tectonics identifies a conceptual revolution in the 1960s (1993).

The value of marine magnetism did not arise from one isolated pattern. Its interpretation depended on independently developed ideas about field reversals, remanent magnetization, rock dating, crust formation and lateral motion. When those lines of work supported one another, the same observations constrained rival geological accounts more strongly.

This transition illustrates the combined effect of improved instruments, new empirical patterns and explanatory integration. It also shows why no single date should substitute for a history of development and acceptance. Proposals can precede general agreement by years, and the evidential case may emerge through cumulative contributions by distinct researchers. A philosophical reconstruction that names only a revolutionary theory misses the research practices that made it persuasive.

Theory Change in Physics: General Relativity

The replacement of Newtonian gravitation as a fundamental account involved conceptual reorganization together with new empirically relevant consequences. The American Physical Society's history of Mercury and eclipses describes the calculation of Mercury's perihelion advance and the 1919 eclipse test. NASA's history of eclipse observations provides additional institutional context for the expedition and its scientific reception.

The eclipse event was important in the history of scientific acceptance, but its observations had technical uncertainties. A careful account should not represent a single expedition as deductive proof of an entire theory. The case becomes more informative when considered alongside other tests, developments in mathematical structure and later measurements.

Relativity also preserved much earlier calculation in restricted conditions, producing a layered transition. Fundamental concepts of gravity and spacetime changed, while tested numerical relations continued to have a legitimate operational role. This coexistence of novelty and correspondence challenges both the picture of a clean intellectual rupture and the picture of simple accumulation. It supports a selective, explicitly qualified concept of continuity.

Comparative Lessons from Chemistry, Geology, and Relativity

The three historical cases differ in the nature of the theories displaced and the evidential networks that supported their successors. Oxygen chemistry substantially reorganized the language of substances and combustion. Plate tectonics integrated new geophysical observations with a unifying causal account. General relativity combined new theoretical concepts with discriminating astronomical tests and limiting relations to earlier equations.

All three also preserve something from earlier science. Chemical procedures and quantitative records survived changed classifications. Marine geophysical records became more informative through a new interpretation. Newtonian calculations remained serviceable within understood limits. Preservation thus concerns distinct objects: records, techniques, mathematical approximations and research problems.

The original comparative finding of this dossier is that the question 'What survived?' is as informative as the question 'What was rejected?' for understanding a rational transition. It reveals when a successor explains former achievements and when it merely displaces a vocabulary. The comparison uses published scholarly and institutional accounts; it constitutes an interpretation of documented episodes rather than new historical source discovery.

Revolution, Local Adjustment, and Research Continuity

Scientific change can range from parameter refinement to restructuring of accepted entities and standards. The analysis of revolutionary diversity documents why actual history resists a single pattern of revolutionary transition (Nickles, 2017). A local model correction may preserve almost every theoretical commitment; a disciplinary reorganization can change the questions and instruments by which inquiry proceeds.

The scale of change affects the burden of explanation. A small adjustment may be assessed by repeat measurements. A semantic transformation may require explicit translation and comparative reconstruction. A change of research practice may require attention to a community's training, institutions and accepted problems. The same word 'theory' can obscure these distinctions when applied across cases.

Calling every significant improvement a revolution reduces that term's analytical value. Describing every revolution as ordinary addition conceals real conceptual changes. A better historical vocabulary describes the specific continuity and displacement, and then examines whether the new configuration improved the field's ability to learn from evidence.

The Difference between Acceptance, Use, and Pursuit

Scientists may accept a theory as currently well supported, use a model for an applied calculation and investigate a competing theory as a possible future improvement. These activities involve different commitments. Kuhn's essays on scientific judgment explain how multiple criteria enter appraisal; Lakatos's account of research programmes offers reasons for continuing to work on an imperfect but potentially developing approach.

An accepted theory may contain known anomalies, yet scientists can continue using it where its predictions remain accurate. A rival may deserve exploratory resources without already justifying the same degree of acceptance. A third model may be convenient for engineering because of computational tractability, even when it is less fundamental than another account.

This distinction improves historical description. A scientist's development of an alternative is not by itself evidence that the prevailing theory had already been abandoned. A field may include overlapping traditions, partial conversions and a prolonged division of labor. Reconstructing theory change requires evidence about the commitments actually made, not merely a final retrospective map of winners.

Ad Hoc Adjustments and Independent Empirical Content

An ad hoc repair is often understood as an adjustment proposed mainly to protect a favored hypothesis from known contrary evidence without yielding independent testable support. Lakatos's criterion of progressive changes makes the distinction between accommodating past failures and generating corroborated novel content central (Musgrave and Pigden, 2026).

This judgment is historically sensitive. An auxiliary assumption introduced to explain one anomaly may later gain independent confirmation, changing its evidential standing. Another adjustment may improve fit while reducing explanatory reach or inviting still more exceptions. The term 'ad hoc' should therefore be tied to an explicit chronology and substantive justification.

An editorially auditable reconstruction identifies the initial anomaly, the exact change introduced, its novel consequences, independently available evidence and later tests. That reconstruction can distinguish responsible repair from mere retrospective accommodation. It also prevents the historian from praising every successful innovation as though its success was foreseeable from the moment of proposal.

Values, Institutions, and Social Criticism

Scientific change is organized by communities whose institutions influence which evidence is collected, which alternatives receive resources and how criticism is distributed. Longino's account of social scientific knowledge argues that critical practices and attention to contextual assumptions matter to scientific objectivity (1990). This does not imply that the social origin of a claim settles its truth.

A working intellectual community needs standards for criticism that can address central premises rather than merely refine accepted examples. Such criticism may occur through rival research traditions, revised instruments, replication, theory-specific predictions and debate about appropriate methods. Institutional structures can promote or impede these activities.

The normative conclusion is that theory change is more accountable when reasons and data can be inspected and challenged by qualified investigators. Published disagreement is an important resource when it exposes hidden assumptions. The epistemic quality of change depends on how a field responds to arguments, not simply on the rate at which it adopts new ideas.

Historical Reconstruction and the Problem of Hindsight

After a successful transition, the evidence that eventually supported the successor can appear inevitable. Historians must resist this retrospective distortion by reconstructing the sequence of discoveries, alternatives and uncertainties. The interpretation of Kuhn's historical approach and critical review of scientific revolutions make clear that historical philosophy draws its claims from episodes whose explanations remain contestable (Bird, 2025; Nickles, 2017).

A useful case record distinguishes four dates: the initial formulation of an alternative, the first relevant observation, substantial disciplinary acceptance and later confirmation. These points are frequently separated by years. It also identifies the standpoint of primary actors, who often lacked theories and instruments available to subsequent generations.

This attention to chronology protects scientific judgment from anachronism. A rational scientist may decline to abandon an established framework when an alternative has not yet achieved comparable empirical support. The eventual success of the alternative gives the historian an outcome, but it cannot replace the earlier evidential situation. Good historical explanation recovers how the choice looked to those making it.

A Documentary Procedure for Comparing Transitions

A disciplined comparison of historical changes can proceed through several related inquiries. Identify the problem field and achievements of the earlier framework; reconstruct the anomalies that became salient; list alternatives actually available; distinguish new predictions from later accommodations; and establish which measurements, equations or practical techniques survived. This is a synthesis of established methodological questions, not a newly proposed philosophical system.

The procedure helps test broad claims. If a supposed revolution preserved all explanatory categories while changing instruments, its degree of conceptual discontinuity requires reconsideration. If a supposed cumulative advance eliminated an earlier classification system, the history may contain an unacknowledged semantic transformation. If the successor's empirical triumph is asserted without a contemporary prediction record, the inference is too strong.

In the cases examined here the reconstructed changes involved different mixtures of conceptual revision and evidential consolidation. The procedure makes comparisons explicit without pretending that a chemical revolution, a geological reorganization and a new physical theory share a single quantitative measure of progress.

Counterexamples and Limits of Generalization

Selected successful theories make a poor basis for universal historical laws. To test a sweeping claim about rational scientific progress, an investigator would also need cases of abandoned programmes, unresolved disputes, misleading data and periods of stagnation. The analysis of competing progress accounts and assessment of scientific revolutions emphasize that philosophical disagreement extends to what should count as progress or revolution (Niiniluoto, 2024; Nickles, 2017).

The present examples were chosen for conceptual contrast and documentary accessibility, not as a random sample of scientific transitions. They demonstrate the usefulness of distinctions and supply plausible interpretations of specific histories. They cannot estimate the frequency of revolutions, demonstrate a universal mechanism of theory replacement or resolve every realist dispute.

This limitation also applies to sweeping pessimistic arguments. An archive of superseded theories may neglect scientific achievements that remain reliable across successive explanations. Philosophical inference from history requires attention to both failure and retention. The burden is to identify which conclusions the available cases genuinely warrant.

Computational Science and Contemporary Theory Change

Modern inquiry uses simulations, large observational datasets, automated classification and computational hypothesis generation. These practices alter how alternatives are explored and how evidence is processed. The study of evidence and observation clarifies why data-processing operations and assumptions remain relevant when a scientific output is evaluated (Boyd and Bogen, 2026).

A computational predictor can improve without making causal claims clearer. An automated search may propose a useful hypothesis without establishing it as true. Shared data infrastructure may allow old measurements to be reinterpreted under a new theoretical framework. These statements describe methodological possibilities; this monograph reports no new experiment with computational systems.

The historical lessons continue to apply: determine what changed, specify which tests discriminate rival interpretations and identify inherited achievements. The involvement of advanced computation changes the technical means of investigation but does not eliminate the demand for reasoned comparisons. The epistemic status of a theoretical transition remains grounded in the relation among evidence, claims and methods of criticism.

Original Synthesis: The Evidential Advantage of a Successor

The historical and philosophical materials support a comparative conclusion. Rational theory change is best explained when an account identifies the evidential advantages obtained by a successor and the accomplishments that remain available from its predecessor. This inquiry is richer than declaring that one theory defeated another, because success can appear in some domains while earlier techniques remain indispensable in others.

Duhem makes the dependency of tests visible; Popper emphasizes exposure to refutation; Kuhn explains the historical organization of disciplinary standards; Lakatos assesses sequences of theoretical work; Laudan emphasizes problem solutions; Feyerabend defends productive alternatives. Each contributes to understanding what must be reconstructed, while their disagreements about rationality and progress remain substantive.

My authorial interpretation is that evidential comparison and historical retention should be examined together. The first asks why change became warranted. The second asks which earlier achievements the successor explains, preserves or confines to a restricted application. Their conjunction supplies a clearer account of particular transitions. This is an original analytical synthesis of established arguments and documented examples, without a new named principle or priority claim.

Implications for Writing and Reviewing Scientific History

A rigorous account of theory change identifies the conceptual object of revision, reconstructs rival explanatory commitments, and documents the source and chronology of important evidence. It distinguishes primary authorial claims from secondary scholarly interpretations and institutional historical summaries. Where a restricted book is available only through publisher records, conclusions attributed to its full argument must be checked independently and reported at the appropriate level of access.

Scientific review should ask whether a claim about revolution relies on a genuine change of concepts, whether a claim about progress specifies a relevant standard and whether a claim about continuity identifies what exactly was preserved. References should support propositions near where those propositions are made, with direct contextual links permitting the reader to inspect their basis.

For philosophical publication the same practices support responsible original interpretation. An author can argue strongly for a synthesis while maintaining the distinction between the evidence available and what further historical research would be needed to establish. Such transparency gives the argument greater explanatory precision and preserves the documentary basis for later criticism.

Conclusion

Theory change is a multi-level transformation of scientific explanations, concepts, representations and practices. Its rationality is assessed through evidence, alternatives, error correction, problem solving and the standards relevant to a field's aims. Historical transitions can retain useful measurements and approximations even while rejecting the explanatory framework in which those achievements were first interpreted.

The traditions associated with Duhem, Popper, Kuhn, Lakatos, Laudan and Feyerabend illuminate different facets of this phenomenon. The realism debate investigates what can survive theoretical replacement; social epistemology shows why public criticism matters; historical case studies explain why no single moment or rule exhausts the development of a successor.

The chemical revolution, plate tectonics and general relativity each combine conceptual change with specific continuities. The resulting authorial conclusion is that the strongest account of a rational transition documents both a successor's evidential advantage and the identifiable achievements retained or revised from earlier inquiry. This conclusion is an argued philosophical interpretation of verified scholarship and historically described cases. It is offered without a claim to new empirical discoveries, new scientific laws or a new philosophical system.

Source Scope and Research Integrity

This manuscript provides comparative philosophical analysis of published arguments and historically documented cases. The existence and metadata of selected restricted primary books and chapters were inspected through publishers and scholarly catalogs; complete licensed editions of those works were not represented as fully read. Open scholarly reference articles and the National Academies report chapter were consulted for the historical and conceptual claims they support. Some USGS institutional pages were available through indexed text but direct retrieval was restricted. Case comparisons are analyses of published accounts, not independent archival discoveries. Sources and URLs are recorded in the private EVIDENCE.json ledger; all used research links are present in References. The author's original conclusions are explicitly distinguished from the claims of historical source authors and do not assert a new empirical finding or independently founded theoretical system.