Loading field map
Scientific change and theory choice is the subfield of philosophy of science that investigates how scientific theories are created, modified, rejected, and replaced, and how scientists decide which theories to accept. It sits at the intersection of descriptive questions about what scientists actually do and normative questions about what they should do. The field asks whether science progresses toward truth, whether there are rational standards for choosing among competing theories, and whether the historical record of science supports any particular account of how knowledge develops.
The subfield emerged from a tension between two observations. On one hand, science appears to be the most successful knowledge-producing enterprise humans have devised. On the other, the historical record shows that scientists have repeatedly abandoned theories that were once universally accepted, sometimes for reasons that look social, psychological, or political rather than strictly logical. The central problem is to determine whether this history can be reconciled with the claim that science is rational.
This problem has two connected parts. The first is descriptive: how does scientific change actually occur? The second is normative: how should it occur? A satisfactory philosophy of scientific change must answer both, because a purely descriptive account cannot tell us why we should trust science, while a purely normative account that ignores history risks describing an ideal that no real scientist has ever followed.
The modern subfield begins with the logical empiricists of the Vienna Circle and their successors in the mid-twentieth century. Their approach was shaped by the conviction that philosophy should emulate the rigor of formal logic and that the meaning of a statement lies in its method of verification. On this view, scientific theories are formal systems connected to observation through correspondence rules that give theoretical terms their empirical content.
For the logical empiricists, scientific change was primarily a matter of theory reduction and accumulation. A new theory could replace an old one if it explained everything the old theory explained, plus more. Newtonian mechanics, for example, could be derived as a limiting case of Einstein's relativity under certain conditions. This picture made scientific change look progressive and cumulative: later theories subsume earlier ones, and the growth of knowledge is monotonic.
The logical empiricist account had a serious weakness, however. It could not explain revolutionary changes in which the new theory is not a simple extension of the old. The shift from Ptolemaic to Copernican astronomy, or from phlogiston chemistry to Lavoisier's oxygen theory, involved changes in the very meaning of central terms. The logical empiricists had little to say about such episodes, and their account of theory choice was largely formal: choose the theory with greater empirical adequacy, simplicity, and explanatory power. These criteria were assumed to be unproblematic, and the historical messiness of actual scientific practice was treated as irrelevant to the logic of justification.
The decisive break came in the 1960s with Thomas Kuhn's The Structure of Scientific Revolutions and the work of other historically minded philosophers such as Paul Feyerabend, Imre Lakatos, and Stephen Toulmin. These thinkers argued that the logical empiricist picture was not merely incomplete but actively misleading. They insisted that any adequate philosophy of science must take the history of science seriously, and that history reveals patterns the formal account could not capture.
Kuhn's central concept was the paradigm: a shared framework of assumptions, methods, standards, and exemplary problem solutions that defines a scientific community's research agenda. Normal science, on Kuhn's account, is puzzle-solving within a paradigm. Scientists do not test the paradigm itself; they take it for granted and work on the problems it defines. Anomalies—results that do not fit the paradigm—are initially ignored or explained away. But when anomalies accumulate and resist resolution, a crisis develops, and the community may undergo a scientific revolution in which one paradigm is replaced by another.
Kuhn argued that paradigm choice cannot be fully explained by the standard criteria of accuracy, simplicity, and fruitfulness, because these criteria are themselves partly defined by the paradigm. Different paradigms are incommensurable: they do not share a common vocabulary, and their standards of evaluation differ. This led Kuhn to a controversial conclusion: the transition from one paradigm to another is more like a conversion experience than a logical deduction. It is not irrational, but it is not fully determined by evidence and rules either.
Kuhn's account was widely read as claiming that science does not progress toward truth, but only changes its frameworks. Kuhn himself resisted this reading, insisting that science does progress in the sense of solving more problems and achieving greater puzzle-solving power. But the damage was done: the image of science as a rational, cumulative enterprise had been challenged at its core.
Feyerabend pushed the historical critique further. In Against Method, he argued that there is no single scientific method, and that the history of science shows that progress has often required violating whatever methodological rules were current. He proposed an anarchist theory of knowledge in which "anything goes," not because all methods are equally good, but because no fixed set of rules can capture the creative and context-dependent nature of scientific inquiry. Feyerabend's position was more radical than Kuhn's, and most philosophers rejected his conclusion, but his historical examples forced the field to confront the diversity of scientific practice.
Lakatos attempted to rescue rationality from the historical turn by proposing a middle way. He argued that science should be understood in terms of research programmes: sequences of theories connected by a shared hard core of assumptions, protected by a belt of auxiliary hypotheses that can be modified in response to evidence. A research programme is progressive if it predicts novel facts and degenerating if it merely accommodates anomalies after the fact. Lakatos claimed that scientists rationally choose progressive over degenerating programmes, and that this account fits the historical record better than Kuhn's paradigm model.
The debate between Kuhn and Lakatos framed the field for decades. Kuhn emphasized the social and psychological dimensions of scientific change; Lakatos emphasized the rational reconstruction of scientific decisions. Both agreed that the logical empiricist picture was inadequate, but they disagreed about what should replace it.
While the historical debate raged, another tradition developed that approached scientific change from a more formal direction. The semantic view of theories, associated with Patrick Suppes, Bas van Fraassen, and Frederick Suppe, treats theories not as sets of statements but as families of models. A theory is defined by a class of mathematical structures, and empirical claims are made by asserting that some part of the world is isomorphic to one of these models.
This approach has important consequences for understanding theory change. If theories are models rather than statements, then the question of whether one theory reduces to another becomes a question of whether one family of models can be embedded in another. The semantic view also makes it easier to understand how theories can be applied to different domains, since the same abstract structure can be instantiated in many ways.
The semantic view does not directly answer the question of how scientists choose among theories, but it provides a framework for understanding what is being chosen. It also connects philosophy of science to the practice of modeling in contemporary science, where mathematical models often play a more central role than explicit laws or axioms.
A different challenge to the rationalist tradition came from the sociology of scientific knowledge, which emerged in the 1970s and 1980s. Sociologists such as David Bloor and Bruno Latour argued that the content of scientific theories should be explained in the same way as any other cultural product: through social interests, power relations, and institutional structures. On this view, the distinction between internal (rational) and external (social) factors in scientific change is untenable, because all scientific decisions are social decisions.
Philosophers responded to the sociological challenge in several ways. Some, like Larry Laudan, argued that rationality and social explanation are compatible: a scientist can be rationally persuaded by evidence while also being influenced by social factors, and the two explanations operate at different levels. Laudan proposed a problem-solving model of scientific rationality, according to which theories are chosen because they solve more conceptual and empirical problems than their rivals. This model was designed to be historical and empirical while preserving a normative role for rationality.
Others, like Philip Kitcher and Helen Longino, developed naturalistic approaches that treat philosophy of science as continuous with empirical science. Kitcher argued that the social structure of science—its division of labor, its reward systems, its patterns of communication—can be studied empirically and that such study reveals how rational consensus emerges from the interactions of imperfectly rational individuals. Longino emphasized the role of critical dialogue and social norms in ensuring that scientific knowledge is objective, even when individual scientists are biased.
The naturalistic turn did not resolve the debate between rationalists and sociologists, but it changed the terms. The question was no longer whether social factors influence science—everyone agreed they do—but whether such influence undermines the rationality of scientific change or is instead part of how rationality is achieved.
The current state of the subfield is characterized by pluralism and a willingness to combine insights from multiple traditions. Few philosophers today defend the pure logical empiricist picture, and few defend the radical relativism that some attributed to Kuhn and Feyerabend. The field has instead settled into a set of overlapping debates.
One major area of research concerns scientific realism and antirealism. Realists argue that successful scientific theories are approximately true and that the entities they posit—electrons, genes, black holes—really exist. Antirealists, following van Fraassen's constructive empiricism, argue that science aims at empirical adequacy, not truth, and that we have no reason to believe in unobservable entities. This debate is closely connected to theory choice, because if realism is true, then choosing a theory is choosing a claim about the world; if antirealism is true, then theory choice is about finding the best instrument for prediction and control.
Another active area concerns evidence and confirmation. Bayesian approaches model scientific rationality as the updating of prior probabilities in light of evidence. These approaches provide a precise account of how evidence should change our confidence in theories, but they face difficulties in specifying priors and in handling the holistic nature of theory testing. The Duhem-Quine thesis—the claim that any theory can be maintained in the face of contrary evidence by adjusting auxiliary assumptions—remains a central problem for any account of confirmation.
A third area concerns scientific pluralism. Some philosophers argue that the diversity of scientific approaches is not a temporary stage on the way to a single unified theory, but a permanent feature of science. Different models may be useful for different purposes, and there may be no single best theory of a complex domain. This view has been influential in biology, where multiple models of the same phenomenon often coexist, and in the social sciences, where methodological pluralism is common.
The field has also become more attentive to the practice of science. Ethnographic studies of laboratories, historical studies of scientific instruments and techniques, and philosophical analyses of modeling, simulation, and experiment have all contributed to a richer picture of how scientific change actually occurs. This work has blurred the boundaries between philosophy, history, and sociology of science, and it has made the field more empirical in its methods.
Despite these developments, the core questions of the subfield remain recognizable. How do scientific theories change, and what makes some changes progressive? Are there rational standards for choosing among theories, or is theory choice ultimately a matter of values, interests, or incommensurable frameworks? Does science converge on truth, or does it merely produce increasingly useful models? How should we understand the relationship between the social organization of science and the epistemic quality of its products?
These questions have no settled answers, and the field is characterized by ongoing disagreement. But the disagreement is productive: it has generated a rich body of philosophical work that is historically informed, methodologically diverse, and attentive to the actual practice of science. The subfield has moved from a confident formal account of scientific rationality to a more cautious, pluralistic, and empirically grounded set of approaches. What remains constant is the conviction that scientific change is not a simple accumulation of facts, but a complex process that deserves careful philosophical scrutiny.