Kuhn's Structure of Scientific Revolutions: Paradigms and Normal Science
Thomas Kuhn's 1962 account of how scientific communities solve problems, respond to anomalies, and change the frameworks that guide their research.
PhilosophyAll chapters Reading tools Notes & comments
Normal science and conceptual change › What Makes a Result a Surprise?Full chapter · Sections in this chapter1 / 8 ↓
What Makes a Result a Surprise?
Imagine two researchers trying to explain why a substance gains weight when it is heated. They agree on what the balance reads. They may disagree about what is gaining weight, whether part of the surrounding air has entered the reaction, and what the apparatus needs to keep track of. A reading on the balance can be surprising because the people using it expected something specific. Changing the explanation can change which features of the apparatus are worth controlling in the next experiment.
Thomas Kuhn makes this relationship between expectation and experiment central to The Structure of Scientific Revolutions. Published in 1962, with a postscript added to the second edition, the book asks how established scientific work produces both detailed knowledge and some of the reasons to revise the framework under which that work is done. Kuhn investigates historical episodes in which scientists had to change what they considered a well-posed problem, a useful explanation, or a successful result. His account of oxygen and combustion gives us a case to return to as those distinctions get clearer. The Structure of Scientific Revolutions, preface, chs. I and VI, and 1969 postscript
Learning to Solve the Right Sort of Problem
Kuhn uses normal science for research based on achievements a scientific community takes as the starting point of further work. An achievement becomes important for this purpose if it attracts people from other ways of working and leaves them a great deal to do. It may show how to connect a theory with a certain kind of observation or how to obtain a measurement worth using in other investigations. The new researchers can devote time to problems that make sense within this achievement without reopening the question of its standing each morning. The Structure of Scientific Revolutions, chs. II–III
Normal science can be demanding. Suppose a laboratory knows how to estimate a rate from the change in the reading of an instrument. The first estimate may be too crude to distinguish two predictions. The laboratory can refine the method, control disturbances it previously ignored, and specify more carefully the conditions in which the instrument measures what it is meant to. Each improvement can enlarge the set of questions the researchers can answer. Kuhn calls this work puzzle-solving because the shared achievement helps set both a problem and standards for its solution. A puzzle is expected to have a solution within the practice, although finding it can take skill and ingenuity. The Structure of Scientific Revolutions, chs. III–IV
It is helpful to distinguish a puzzle from any problem a group might want to solve. Kuhn gives the example of a pile of pieces selected from different jigsaw puzzles. Nothing about the resulting pile gives the player a reason to expect that there is a picture containing precisely those pieces. Scientific researchers have reasons to expect that some problems can be solved. Their prior achievements help them identify questions for which they have ways of proceeding. This is one way a research tradition can direct their efforts productively. It can also leave a problem outside ordinary work because no one can yet say what it would take to solve it. The Structure of Scientific Revolutions, ch. IV
What Scientists Share
In the original text, Kuhn calls achievements that serve as models paradigms. In the postscript he distinguishes two uses of the term. A community can share a broad set of commitments including formulas, models, values, and examples of successful work. He calls that ensemble a disciplinary matrix. He uses exemplars for shared concrete solutions that teach a person how to approach other cases. An agreed equation and an example of how to use it belong to the same practice. They do different work in teaching someone to solve a problem. The Structure of Scientific Revolutions, 1969 postscript, §§2–3
An instructor might provide students with a formula and several worked exercises showing how it applies to moving bodies. The difficult step in a new problem can be deciding which interval matters, what counts as the body being studied, and which effects can be set aside for this purpose. Watching someone solve an earlier problem can make the family resemblance visible before the student can state a rule that captures it. In Kuhn's account, treating problems as similar is part of what scientists learn from examples. It helps explain how people can become adept at a common practice even when no one has supplied a complete description of the rules. The Structure of Scientific Revolutions, ch. V and 1969 postscript, §3
A disciplinary matrix is a community's set of shared commitments. An exemplar is a worked achievement that researchers can use as a model for recognizing and solving another problem. Kuhn distinguishes these senses in the postscript. The Structure of Scientific Revolutions, 1969 postscript, §§1–3
How an Anomaly Becomes Important
Well-developed expectations help make unexpected results recognizable. The laboratory in our example can identify a reading as a surprise because its methods and theory suggest an outcome. The researchers might spend days checking the settings and using a different batch of material. In Kuhn's account, this is ordinarily work on a puzzle. People have reason to explore a difficulty inside a framework that helps them recognize and solve others. The fact that one calculation or reading does not yet agree with the framework leaves open what the problem is and whether a small change will resolve it. The Structure of Scientific Revolutions, chs. VI and VIII
An anomaly can take on a different standing when attempts to explain it keep running into the same difficulty. Kuhn identifies cases in which a mismatch concerns a particularly important expectation or works against an established application. Repeatedly trying to solve that problem helps show just how much would have to change if the anomaly could not be accommodated. In a crisis, researchers may offer more competing articulations of their shared account. Confidence in some of its rules can loosen. The sequence leaves scope for several responses, including solving the problem with the existing framework or setting the difficulty aside for another generation. The Structure of Scientific Revolutions, chs. VII–VIII
Kuhn gives an account of discovery in which noticing an anomalous result is a beginning. Researchers have to work out what the result is a case of. He discusses oxygen to show how reporting that someone discovered a gas can compress several developments. Preparing a sample, being able to distinguish it from other gases, and giving it a place in an account of combustion are distinguishable achievements. Assigning discovery to a single moment depends on which of those tasks one decides the word should name. The Structure of Scientific Revolutions, ch. VI, oxygen episode
Kuhn contrasts Joseph Priestley's identification of the gas within the vocabulary of phlogiston with Antoine Lavoisier's role in a different theory of combustion. A person could observe some of the same features of the gas and still understand what the experiment had produced in a different way. Kuhn also says the discovery of oxygen was insufficient by itself to bring about the change in chemistry. Lavoisier already had reasons for questioning some of the established account. The episode illustrates how a novel result can be shaped by the expectations under which it was detected and then become part of a different set of expectations. The Structure of Scientific Revolutions, ch. VI, oxygen and changes in chemical theory
What Changes in a Revolution
A scientific revolution, in Kuhn's account, changes the commitments that organize a research tradition. An alternative has to offer a way of working on problems as well as a response to a particular anomaly. Researchers can ask whether the alternative preserves results they have relied on, opens problems they can investigate, or makes sense of a difficulty that had been left hanging. In some episodes, a problem that had been crucial within the earlier framework becomes less pressing in the new one. The change can affect which questions help determine the success of the account. The Structure of Scientific Revolutions, chs. VIII–IX and XII–XIII
Consider Kuhn's distinction between discovering an error and accepting a replacement. If a community were to stop relying on its established way of doing the work because a serious case had gone badly, it would still need to decide what to do next. Kuhn argues that rejection of a paradigm in the episodes he studies involves considering a rival way of working. Deciding about the rival brings in its relation to the evidence as well as how it compares with the existing approach. The challenge to a theory and the availability of an alternative can matter at different points in the same investigation. The Structure of Scientific Revolutions, ch. VIII
Kuhn also considers whether a revolution can simply be described as a newer theory adding precision to the old one. For Newtonian and relativistic dynamics, he distinguishes recovering forms of equations under limiting conditions from keeping the same meanings for all the terms that occur in them. A relation that gives a successful approximation in some applications can be worth keeping. What the relation says about time, space, or mass may have to be read within the commitments of the account that gives it a place. For Kuhn, the mathematical relationship does not erase every change in the concepts the equations express. The Structure of Scientific Revolutions, ch. IX, discussion of Newtonian and relativistic dynamics
How Rivals Can Be Compared
Kuhn's term incommensurability concerns difficulties in bringing rival frameworks under one measure. The difficulty can arise because researchers who use the same word put it to work in different networks of problems and solutions. It can also arise because shared standards leave room for differences in which results count most. One group may value an alternative for making a wider range of problems accessible. Another may value the precision achieved by the current approach in a particular set of cases. Both can point to scientific considerations while disagreeing about how much each should weigh. The Structure of Scientific Revolutions, chs. IX and XII, and 1969 postscript, §§2, 5
Lack of a single decision procedure leaves substantial work to do. Kuhn discusses what researchers can learn by trying to translate claims into a vocabulary another community uses. It may be necessary to use a description longer than the term it was meant to replace, and the difference itself can be informative. He mentions the dispute between Joseph Louis Proust and Claude Louis Berthollet as a case in which reasoning across vocabularies could require avoiding the shared word element. Working through such a description can help an investigator see how someone else came to take a result as evidence or regard a particular sentence as an explanation. The Structure of Scientific Revolutions, 1969 postscript, §5, discussion of translation
This also gives a reader a way to understand disagreement about standards. Suppose one account can handle results from more experiments, while another offers very close quantitative agreement in a well-established range. Choosing what to investigate next requires attending to both achievements and to what the research community could learn by pursuing either one. If later results show that the first account's wider scope can come with the precision formerly achieved by the second, the practical balance changes. Kuhn's point is that the role of a shared value such as precision has to be seen in the particular argument over how a theory meets it. An agreed list of values does not supply a guaranteed result of the comparison. The Structure of Scientific Revolutions, 1969 postscript, §§2, 5
Why the Historical Cases Matter
Think again of the balance in the first example. An isolated record of its reading can tell you something that happened. It cannot by itself explain which expectations made the result surprising, how researchers worked out what the apparatus had produced, or what an alternative gave them to do later. Those questions ask for different evidence. A lab record might show the first reaction to the measurement, while a revised procedure could show what researchers learned to control. The fact that the terms of inquiry change gives us reason to attend to what the records meant to people at the time. The Structure of Scientific Revolutions, chs. I and VI
There is another reason to look beyond a textbook account. A textbook can use past discoveries to teach students the concepts and practices that are now accepted. That is good for learning to solve the next exercise. It can make the older work look as though its purpose were to acquire the results presented in the current chapter. In Kuhn's account, textbooks can make a revolution difficult to see because they keep the parts of the earlier history that contribute to present problems and pass over others. A report written before the change can show why a question that no longer seems urgent once called for a great deal of work. That is historical evidence even when it offers a poor model for a student who is learning the current practice. The Structure of Scientific Revolutions, ch. XI, “The Invisibility of Revolutions”
A criticism often made of this argument can be posed directly. If standards and problems can change, is there any sense in which a field can improve? Kuhn identifies two grounds on which researchers can take a proposed change seriously. They may find that the new account resolves a recognized problem and preserves much of the earlier practice's ability to solve concrete problems. Keeping that ability can matter even where the theory now explains it differently. This leaves room for progress in precision and range as well as questions about what was lost in the transition. The historical narrative does not have to treat every change as an addition to a single inventory of discoveries. It can show which tasks became easier or harder to perform, and how the field changed in deciding which tasks to count. The Structure of Scientific Revolutions, ch. XIII, “Progress through Revolutions”
The distinction between a surprising result and a reorganization lets Kuhn explain part of why an achievement can be productive enough to bring its own limitations into sight. Researchers improve instruments and make predictions more precise in the course of solving familiar problems. That makes some patterns of failure easier to notice. A new framework can build on some of what was achieved and still change the description of how it was achieved. To follow the difference in a historical episode, track what people expected, what they learned to see, and what became a problem they knew how to investigate. The Structure of Scientific Revolutions, chs. III, VI, and VIII
Further Reading
- Thomas S. Kuhn, The Structure of Scientific Revolutions, second enlarged edition. Read Chapters III–VI for normal science and discovery, then the postscript for the distinctions among paradigms, exemplars, and disciplinary matrices.
- University of Chicago Press, The Structure of Scientific Revolutions, fourth edition. The edition adds an introduction by Ian Hacking.
Join the discussionAsk a question or share what you learned.
Keep exploring
Reference
Citation details
Cite or share this article.

Comments
Add to the conversation
A question, a correction, or an example from your own work.
No comments yet. Be the first to share your thoughts!