Paradigms and Scientific Revolutions

In brief

Thomas Kuhn (1922 to 1996) published The Structure of Scientific Revolutions in 1962. Trained as a physicist, he had turned to history and found that the textbook picture of science, a patient accumulation of facts marching toward truth, matched almost nothing in the record. Real science, he argued, moves in a rhythm. Most of the time practitioners work inside a shared framework he called a paradigm, solving the puzzles it sets and never questioning its foundations. This is normal science. Occasionally the puzzles refuse to yield; anomalies accumulate; confidence cracks into crisis; and a scientific revolution replaces the old paradigm with a new one. The two frameworks, Kuhn claimed, are incommensurable: there is no neutral standpoint from which to prove one strictly better, because each supplies its own problems, standards, and meanings. Science is rational, but its rationality is internal to a paradigm, not a ladder to a fixed truth.

The full treatment

The problem it answers

By the mid twentieth century the dominant philosophy of science was normative and ahistorical. The logical positivists asked how theories are confirmed by evidence; Karl Popper (1902 to 1994) answered that they are never confirmed but only tested and, ideally, refuted (see falsificationism). Both pictured science as a timeless logic applied to data. Kuhn, working from cases such as the shift from Ptolemaic to Copernican astronomy, the chemical revolution of Lavoisier, and the arrival of Einstein, asked a different question: not how science ought to reason, but how it actually changes. The answer did not fit either camp. Theories were not abandoned the moment an experiment embarrassed them, and they were not confirmed by slow accumulation. Something structural was being missed.

How it works: the cycle

Kuhn described a repeating cycle. A field begins in a pre-paradigm state, several schools quarreling with no shared foundation. Then a single achievement, striking enough to draw adherents and open enough to leave work undone, becomes a paradigm: Newton's Principia, say, or Lavoisier's chemistry. A paradigm is not merely a theory. It bundles together accepted laws, instruments, standards of good work, and above all exemplars, the concrete solved problems that students learn from and then imitate.

Under a paradigm the community does normal science, which Kuhn compared to puzzle-solving. The framework is assumed correct; the work is to extend it, measure its constants more precisely, and fit nature into its boxes. Failure to solve a puzzle reflects on the scientist, not the paradigm. This is not a criticism. Kuhn insisted that this dogmatic commitment is what lets science probe deeply, because no one wastes effort re-litigating foundations.

Persistent failures become anomalies: results the paradigm cannot digest. A few are tolerated. But when anomalies strike at the core, or resist the field's best people, the community enters crisis. Confidence loosens, foundational debate reopens, rival formulations proliferate. A revolution occurs when a new candidate paradigm attracts the community and the old one is abandoned. Then a fresh round of normal science begins under the new framework.

The key claim: incommensurability

The most radical and most contested idea is incommensurability. Kuhn argued that competing paradigms cannot be compared point for point on a neutral scale, for three reasons. They disagree about which problems a theory must solve. They carry different standards of what counts as a good solution. And their central terms shift meaning: "mass" in Newton is not "mass" in Einstein, and to think otherwise is to mistranslate. Because observation is theory-laden, even the data are not fully shared. Kuhn reached for the language of Gestalt psychology and religious conversion: adopting a new paradigm is like a switch in perception, seeing a duck where you saw a rabbit, not a proof compelling every rational mind. This is why, he noted, old paradigms often die only when their defenders do.

A distinction that matters

Incommensurable does not mean incomparable in every sense, and Kuhn spent the rest of his life resisting the reading that made him a relativist. In the 1970 Postscript he narrowed "paradigm," which critics had shown he used in some twenty senses, into two: the disciplinary matrix (the whole shared apparatus) and the exemplar (the model problem). He maintained that later paradigms are genuinely better puzzle-solvers, that science makes progress. What he denied was progress toward a fixed, mind-independent truth. Science, he suggested, evolves away from primitive beginnings, as Darwinian life does, without evolving toward a predetermined goal.

Lineage

Kuhn's problem is downstream of David Hume (1711 to 1776) and the problem of induction: if evidence can never prove a general law, the positivist story of accumulation was already in trouble. His emphasis that perception is shaped by prior concepts echoes Immanuel Kant, and Kuhn late in life called his own view a "Kantianism with movable categories." His immediate foil was the Vienna Circle and Karl Popper, whose falsificationism he treated as the sharpest rival account of scientific change. He drew directly on the Gestalt psychologists and on the physician Ludwik Fleck, whose Genesis and Development of a Scientific Fact (1935) had already spoken of "thought collectives" and "thought styles." Kuhn gave these currents a single vivid architecture, and the word paradigm, once a term of grammar, became one of the most used and abused words of the century.

The strongest case for it

Kuhn's account fits the historical record better than its rivals, which is its founding claim and its enduring one. Scientists plainly do not drop a theory at the first refuting instance; anomalies are shelved, patched, or ignored for decades, and Kuhn explains why this is often the right thing to do rather than a lapse in rigor. His normal science names something real: most working scientists are not testing bold conjectures but solving well-defined puzzles inside an assumed framework, and this focus is the engine of detailed discovery. His treatment of textbooks, training, and exemplars captures how scientific communities actually reproduce themselves, through worked examples rather than explicit rules. And by taking theory-ladenness seriously he explains a persistent fact: intelligent, honest researchers can look at the same evidence and disagree, because they are not seeing quite the same thing. Where the older philosophy of science described a logic no scientist recognized, Kuhn described a practice they did.

The strongest case against it

The book provoked immediate and heavy fire, much of it at the 1965 London colloquium published as Criticism and the Growth of Knowledge (1970).

Karl Popper granted that normal science exists but treated it as a danger, not a virtue: the uncritical "normal" scientist is, he wrote, "a person one ought to be sorry for," trained badly into dogmatism. For Popper the health of science lies precisely in the permanent readiness to criticize that Kuhn's normal phase suspends.

Imre Lakatos (1922 to 1974) argued that Kuhn had reduced theory change to "mob psychology," a matter of conversion rather than reasons. His methodology of scientific research programmes was an attempt to save Kuhn's historical realism while restoring rational criteria: a research programme is progressive if it predicts novel facts and degenerating if it only patches anomalies after the fact, giving an objective, if slow, standard for choosing between frameworks.

Israel Scheffler (1923 to 2014), in Science and Subjectivity (1967), pressed the deepest charge: incommensurability makes science irrational and relativist. If rival paradigms share no neutral observations, no common standards, and no stable meanings, then paradigm choice cannot be a reasoned judgment and there is no fact of the matter about which theory is closer to true. Science collapses into a succession of worldviews, each self-certifying.

From the opposite flank, Paul Feyerabend (1924 to 1994) accepted incommensurability and pushed it to its limit, arguing in Against Method (1975) that there is no method distinguishing science from other traditions at all, an anarchism most philosophers took as a reductio Kuhn never intended.

Kuhn spent decades protesting these readings, insisting he believed in progress and rational choice under shared values (accuracy, scope, simplicity, fruitfulness, consistency). Critics replied that these values are too vague to settle hard cases, which returns the choice to the community, which is where the relativism worry began.

Where it stands now

The Structure of Scientific Revolutions is among the most cited academic books of the twentieth century, and "paradigm shift" long ago escaped philosophy into ordinary speech, usually meaning any large change and losing Kuhn's precise sense. Within philosophy of science the strong incommensurability thesis has few defenders; most hold that rival theories can be compared on shared evidence often enough for science to be rational, and physicists such as Steven Weinberg have argued bluntly that Kuhn overstated the discontinuity, since old theories usually survive as limiting cases of the new. Yet his descriptive achievements are permanent. No serious account of science now ignores normal science, the role of exemplars and training, the theory-ladenness of observation, or the fact that anomalies are managed rather than instantly fatal. The corrective to Kuhn is not a return to the accumulation myth but a more careful map of how communities weigh reasons across frameworks. He killed the tidy story for good.

Test yourself

Pick a belief you hold that you would call well established, scientific or otherwise. Ask what would count, for you, as an anomaly serious enough to force you to reconsider the whole framework, not just adjust a detail. If you cannot describe such evidence, are you doing normal science within a paradigm you have simply stopped questioning, and if you can, is the standard you would apply one your intellectual opponents would even recognize?