Chapter 3

Thomas Kuhn and Contemporary Discussions of Scientific Discovery,

At the end of the nineteenth century, scientists had reason to be confident about their achievements. Newton's theory of gravitation had proved successful. It not only gave accurate predictions of the movements of planets, but did so with an aesthetically pleasing, mathematically elegant set of equations. More important, it furnished a fundamental framework in which all further scientific investigation could be integrated. To be sure, scientists had still not thoroughly explored and mastered every potential area of investigation in physics, let alone every area of biology. But they could confidently expect that those yet-to-be-explored areas would harmonize with Newton's framework.

10. The scientific revolutions of the twentieth century

The twentieth century rudely shattered this complacency. Within a period of thirty years, two revolutions in physics overturned Newton's universe in a way that a nineteenth century physicist would have said was impossible. In 1905, Albert Einstein published his first paper on the special theory of relativity. This theory contradicted the fundamental assumption of Newton that measurements of length in space, length in time, and mass of particles are independent of the person's situation who does the measuring. Since space, time, and mass were fundamental to Newton's entire theory, the whole view of the physical universe had to be rethought.

The second revolution, the quantum revolution, began with Max Planck's papers on radiation in 1900.¹ Planck postulated that light was emitted in fixed quantities of energy, rather than being emitted in a simple continuous stream. Planck's idea remained an oddity in physics for more than twenty years. The corpuscular character of light that Planck's theory implied could not be fully reconciled with many other phenomena showing the interference patterns of waves. But to this oddity were gradually added other oddities showing a similar pattern. Neils Bohr in 1913 succeeded in explaining atomic spectra on the basis of quantum ideas, but this account was in one respect still odd: there was no framework capable of thoroughly reconciling particle and wave aspects of the behavior. Finally, in 1925 and 1926 Werner Heisenberg and Erwin Schrödinger produced formulations accurately predicting atomic energy levels. Schrödinger's formulation ("wave mechanics") depended on representing atomic electrons by waves corresponding to fixed quantities of energy, that is, quantized levels of energy. The universe that Newton had assumed to have continuous levels of energy was found to be discrete. Worse, this universe, at an atomic level, behaved not like a particle, not like a wave, but in a way showing features of both. Causality itself seemed to function oddly at an atomic level. In order for the equations of quantum mechanics to hold, it was argued, some events must be innately unpredictable, indeterminate.

Both special relativity and quantum theory invalidated Newton's equations. Furthermore, they showed that the basic intuitions behind Newton's universe were invalid. They produced a picture of the universe that went contrary to intuition. According to special relativity, events at high speed deviated in strange ways from what we are accustomed to in our everyday world. According to quantum theory, events on a very small scale deviate in strange ways. The quantum revolution proved, if possible, even more unsettling because it was impossible to picture the underlying realities. They could be accurately described only in terms of equations that corresponded to no good intuitive picture of the world. Relativity and quantum theory both spawned further developments that were nearly revolutions in their own right and moved the world of physical theory even farther away from the old world of Newton.

First, as a development of relativity, Einstein published in 1915 his "General Theory of Relativity." In this theory he expanded the theory of special relativity to include an account of gravitation. In the new theory mass and energy corresponded to curvature in the very structure of space and time. Newton's view of the universe, by contrast, had assumed that space and time were flat. And it had assumed that gravitation was a "real" force, not something that could be equated with the structure of space and time itself.

Second, in the area of quantum theory, Max Born and Werner Heisenberg introduced an alternative to Schrödinger's formulation, called matrix mechanics. The details of this proposal are not important for our purposes. At a deep level it was, in fact, mathematically equivalent to Schrödinger's formulation. But Heisenberg for the first time explicitly formulated the uncertainty relations of quantum mechanics, namely, mathematical statements implying that one cannot measure a particle's position and momentum simultaneously.

Reflection on the uncertainty relations and on the phenomena of wave/particle duality spawned a philosophical interpretation of quantum theory, the "Copenhagen interpretation." This school said that, in many atomic situations, key quantities like position and velocity were not fully defined in a classical sense until an experiment was performed measuring them. The measurements in the experiment in effect "forced" a particle to take a determinate position or velocity. This view was more radical than several alternatives. For instance, one could have said merely that we as observers did not know what the actual value was (as a Newtonian might have said). Or one could say that we could never in principle know the actual values of all the variables, because measurement of one value inevitably disturbed the others (this conclusion was a result of quantum theory, not of Newton, but it was still fairly safe). But the Copenhagen interpretation said that talk about definite actual values, independent of measuring them, was virtually meaningless.²

After Heisenberg and Schrödinger, refinements in quantum theory have continued to appear. These refinements have made the nature of quantum description ever more esoteric. Heisenberg's and Schrödinger's work has now been surpassed in turn by Dirac's relativistic quantum mechanics, quantized field theory, quantum electrodynamics, quantum chromodynamics. One does not know what theories will appear in the future on the border of knowledge.

After these two revolutions, one could still claim in retrospect that Newton's theory worked as a first approximation. Relativity was a refinement of Newton in the domain of high velocities. Quantum theory was a refinement in the domain of very small physical systems. This qualification helped people to preserve the idea that scientific advance consists simply in adding to the body of known truths.³

But it was difficult to deny that some other things were going on as well. Both relativity and quantum theory challenged not primarily some poorly established hypothesis or some theory just beginning to be established, but the very best and mostly firmly established physical theory (Newton's). And they offered the challenge at the very basis of the theory, by disputing the very ideas of measurement and reality interwoven with every single experiment.

Hence the existence of these revolutions raises questions about the naive inductive view of scientific research discussed in the previous chapter. Are scientific data and scientific laws atomistic? Does scientific progress consist simply in adding more data and adding more laws to the list of approved laws? Or if such a picture is not quite right, is it enough to add a footnote to the effect that occasional pruning of old laws may replace them with more accurate versions of the same? The revolutions produced by the theory of relativity and quantum theory, however, included changes in the shape of physical theory of a most radical nature.

11. Thomas Kuhn: the revolution in history and philosophy of science

The watershed in thinking about scientific progress occurred in 1962. In that year Thomas S. Kuhn published The Structure of Scientific Revolutions, in which he rejected the classic view of science, the view associated with Baconian scientific method. Kuhn argued that science did not advance merely by a step-by-step inductive method. ⁴ Research on specific problems always took place against the background of assumptions and convictions produced by previously existing science. In mature science, this background took the form of "paradigms," a cluster of beliefs, theories, values, standards for research, and exemplary research results that provided a framework for scientific advance within a whole field. Since the word paradigm has come to be used in several different senses, we will instead use the phrase disciplinary matrix.⁵ Newton's fundamental work, Philosophiae naturalis principia mathematica, generated just such a disciplinary matrix within the field of natural science in general and physics in particular. Newton's work was an exemplar, a concrete research result that suggested a way of problem-solving for a large number of unsolved problems. At the same time, as people reflected on the implications of Newton's work, they obtained from it not only a prime example of a successful theory but a framework that suggested further questions, experiments, and generalizations building on and within the overall theory. Newton's theory evolved, then, into a disciplinary matrix for subsequent research.

In chapter 2, we analyzed Baconian scientific method as involving six fundamental assumptions. Over against these assumptions, we may summarize Kuhn's view in a series of counterassumptions:

1. Data are never "hard facts," completely independent of any theory. What counts as data depends on the disciplinary matrix, or framework of assumptions, that scientists use. All data is "theory-laden." It already presupposes, in its very status as data for a given experiment or a given theory, that the universe is organized in a way compatible with the assumptions of the science as a whole. The current disciplinary matrix affects how scientists make observations, what they think the observations actually measure, and what kinds of data or experiments are relevant to the outstanding open questions in their field.
2. Hypotheses do not arise from making a generalization in a vacuum. Rather, they arise from the combined influence of the overall disciplinary matrix in the field, detailed experimental results, the structure of theories in related areas that may suggest analogous solutions in the area currently under scrutiny, and expectations generated by the ruling disciplinary matrix as to what types of theory are most likely to be successful.
3. One cannot simply deduce a prediction from an isolated hypothesis. Predictions from a hypothesis depend not only on the hypothesis itself but on a surrounding body of theory specifying how the hypothesis is related to any particular experimental setup. One must also include here "observation theories," theories about any specialized apparatus used in measurements and the meaning of those measurements.
4. Discarding or retaining a hypothesis is almost never easy. Experiments can go wrong for a large number of reasons. Some unforeseen extra interference may not have been excluded from the experiment. Any one of a group of hypotheses or laws helping to relate the given hypothesis to the experiment may be invalid. One of these may have to be discarded, but a single experiment, or frequently even a whole series of experiments, does not indicate which of a series of connected, mutually dependent hypotheses is incorrect.
5. Most important, science does not advance merely by adding confirmed hypotheses to an existing atomistic list of laws. The laws of a given field of science are related to one another in a coherent way. Additions and subtractions affect the whole. Moreover, there can be times of "revolution" when the whole body of knowledge is recast.

Kuhn is particularly stimulating on the subject of this fifth point, the question of scientific progress. In this area, Kuhn distinguishes between at least three kinds of situation in the development of a particular scientific field. The first is "immature" science. In this situation, the field of investigation for the science is poorly defined. Different workers in the field dispute the kinds of data that are relevant to their field, the purpose of the investigation, the shape a finished theory will have, and the kind of tests that confirm or disconfirm the theory. Investigators are casting about for a fundamental insight that will bring order into a disparate field. In immature sciences, it is not clear how one measures progress. People do experiments and gather data, but because of the unsettled character of the field, it is seldom clear whether their work will make a lasting contribution. Kuhn thinks that the social sciences, for the most part, may still be in this state.

The second kind of situation is that of normal "mature" science. A particular science becomes mature when some investigator or group of investigators advances a fundamental theory, including supporting data, that proves clearly superior. This theory becomes an exemplar, a key research result that also largely determines the whole disciplinary matrix for subsequent research. It suggests a whole line of experiments, a "research program."⁶ The theory explains and organizes a significant body of data. In addition, it confirms its promise by engendering a whole line of experiments that refine, extend, and confirm the theory and that link it with other existing theories. The success of the new disciplinary matrix inaugurates a period of "normal science," devoted to "puzzle solving."⁷ Most of the scientists working in the field devote themselves to small puzzles, the remaining areas of investigation where the overall disciplinary matrix already suggests lines of questioning and the forms of hypotheses that might solve the puzzle.⁸

As long as the scientists in a field continue to solve the puzzles that they find for themselves, they go forward in a way that superficially resembles the Baconian inductive ideal. They add small bits of generalization to the existing body of generalizations. There are always some remaining anomalies, or areas where explanations have not been produced. Here and there are some potentially embarrassing data that do not seem to be compatible with the existing disciplinary matrix. Nevertheless, as long as people are making progress in the puzzle solving, they assume that incremental advances in the field (or in some neighboring field) will eventually enable them to see the compatibility of the anomalies with the disciplinary matrix.

The third situation is that of "extraordinary" science leading to scientific revolution. Revolution occurs when an existing disciplinary matrix is replaced by a new one incompatible with the original. A revolutionary situation first arises when anomalies in a particular field cannot easily be ignored. The anomalies begin to fall into patterns that show an order of their own. More and more tinkering with the disciplinary matrix is necessary in order to produce any kind of rational summary of the anomalies. Inelegant, complex, unmotivated hypotheses arise to account for the anomalies. As more and more energy is devoted to working on the anomalies, tinkering with the reigning disciplinary matrix leads people to produce different versions of the disciplinary matrix. The disciplinary matrix itself no longer looks so unified as it once did.⁹

In this situation people are willing to search about more broadly, looking for better solutions. In the process they are willing even to challenge ideas traditionally associated with the existing disciplinary matrix. Eventually they stumble upon an alternative approach, inexact at first, but appearing to offer some possibility of dealing with the anomalies. This approach is refined, enhanced, and reformed in order to increase its accuracy and the scope of data accounted for. If the process continues, this new approach generates a disciplinary matrix of its own. A fight then ensues between adherents to the old disciplinary matrix and those holding the new as to which matrix is to be used in the future development of science. In this period, it is difficult for adherents of the two disciplinary matrices even to communicate well with one another, because they may have different standards for what counts as data and different standards as to what sorts of explanation have the most promise.¹⁰

Kuhn also notes that, if no satisfactory solution arises, even after prolonged effort and radical attempts to generate alternative explanations, people may fall back on the existing disciplinary matrix and treat the anomalies as an intractable area reserved for future generations. In this case, the period of extraordinary science has not generated a revolution but collapsed back into normal science working with essentially the same disciplinary matrix as before.

12. A specific illustration of Kuhn's theory

The study of electricity provides an example of the process of change in science. According to Kuhn, in the first half of the eighteenth century there was no standard theory of electricity. There was no clear exemplar to bring coherence to the progress of research. Instead, "there were almost as many views about the nature of electricity as there were important electrical experimenters, men like Hauksbee, Gray, Desaguliers, Du Fay, Nollett, Watson, Franklin, and others."¹¹ In such a stage of immature science, there is as yet no standard disciplinary matrix in the field. Some of the theories of electricity of the time regarded "attraction and frictional generation as the fundamental electrical phenomena." Others regarded attraction and repulsion as equally fundamental. A third group regarded electricity as a fluid that ran through conductors. In each case the idea of which phenomena were fundamental directed the concentration and goal of the research.

This preparadigm stage came to an end with Franklin's work Electricity, which became the exemplar for future research. It proved its superior promise by encompassing all the phenomena within its scope.¹²

Subsequent to Franklin's time, the field of electrical research represented a field of normal science. Franklin's theory was elaborated, refined, enhanced, and extended. Scientists no longer debated about the fundamental nature of electricity or the fundamental directions that research should take. They could therefore concentrate more on esoteric phenomena; they could study in great detail the phenomena that the theory indicated were of greatest significance. No individual scientist needed to return and reconstruct the whole field from its foundations up. The resulting specialization made the literature on electricity less accessible to the general public but meant efficiency in making progress within the specialization.

The theory of electricity has since been normal science. However, according to Kuhn, normal science may at times be interrupted by revolutions in theory. These revolutions may take place within a small specialty (such as studies of diamagnetism) or within a broader field. Kuhn is not explicit, but inspection of the history of electricity subsequent to Franklin suggests a series of mini revolutions in small areas. Kuhn does mention explicitly a revolution introduced by James Clerk Maxwell in the second half of the nineteenth century.¹³ Maxwell's electromagnetic theory introduced "displacement current" and other ideas difficult for his contemporaries to digest. The triumph of his theory therefore took time, during which some adherents to older views were converted and some were displaced by a younger generation.¹⁴