Steven Weinberg on scientific revolutions

These are excerpts from an article by Steven Weinberg published in the New York Review of Books, Vol XLV, Number 15 (1998).

I first read Thomas Kuhn's famous book The Structure of Scientific Revolutions a quarter-century ago, soon after the publication of the second edition. I had known Kuhn only slightly when we had been together on the faculty at Berkeley in the early 1960s, but I came to like and admire him later, when he came to MIT. His book I found exciting.

Evidently others felt the same. Structure has had a wider influence than any other book on the history of science.

... Structure describes the history of science as a cyclic process. There are periods of "normal science" characterized by what Kuhn sometimes called a "paradigm" and sometimes called a "common disciplinary matrix." Whatever you call it, it describes a consensus view: in the period of normal science, scientists tend to agree about what phenomena are relevant and what constitutes an explanation of these phenomena, about what problems are worth solving and what is a solution of a problem. Near the end of a period of normal science a crisis occurs--experiments give results that don't fit existing theories, or internal contradictions are discovered in these theories. There is alarm and confusion. Strange ideas fill the scientific literature. Eventually there is a revolution. Scientists become converted to a new way of looking at nature, resulting eventually in a new period of normal science. The "paradigm" has shifted.

To give an example given special attention in Structure, after the wide-spread acceptance of Newton's physical theories--the Newtonian paradigm--in the eighteenth century, there began a period of normal science in the study of motion and gravitation. Scientists used Newtonian theory to make increasingly accurate calculations of planetary orbits, leading to spectacular successes like the prediction in 1846 of the existence and orbit of the planet Neptune before astronomers discovered it. By the end of the nineteenth century there was a crisis: a failure to understand the motion of light. The problem was solved through a paradigm shift, a revolutionary revision in the understanding of space and time carried out by Einstein in the decade between 1905 and 1915. Motion affects the flow of time; matter and energy can be converted into each other; and gravitation is a curvature in space-time. Einstein's theory of relativity then became the new paradigm, and the study of motion and gravitation entered upon a new period of normal science.

Though one can question the extent to which Kuhn's cyclic theory of scientific revolution fits what we know of the history of science, in itself this theory would not be very disturbing, nor would it have made Kuhn's book famous. For many people, it is Kuhn's reinvention of the word "paradigm" that has been either most useful or most objectionable.

... But the quarrel over the word "paradigm" seems to me unimportant. Kuhn was right that there is more to a scientific consensus than just a set of explicit theories. We need a word for the complex of attitudes and traditions that go along with our theories in a period of normal science, and "paradigm" will do as well as any other.

What does bother me on rereading Structure and some of Kuhn's later writings is his radically sceptical conclusions about what is accomplished in the work of science. And it is just these conclusions that have made Kuhn a hero to the philosophers, historians, sociologists, and cultural critics who question the objective character of scientific knowledge, and who prefer to describe scientific theories as social constructions, not so different from democracy or baseball.

Kuhn made the shift from one paradigm to another seem more like a religious conversion than an exercise of reason. He argued that our theories change so much in a paradigm shift that it is nearly impossible for scientists after a scientific revolution to see things as they had been seen under the previous paradigm. Kuhn compared the shift from one paradigm to another to a gestalt flip, like the optical illusion created by pictures in which what had seemed to be white rabbits against a black background suddenly appear as black goats against a white background. But for Kuhn the shift is more profound; he added that "the scientist does not preserve the gestalt subject's freedom to switch back and forth between ways of seeing."

Kuhn argued further that in scientific revolutions it is not only our scientific theories that change but the very standards by which scientific theories are judged, so that the paradigms that govern successive periods of normal science are incommensurable. He went on to reason that since a paradigm shift means complete abandonment of an earlier paradigm, and there is no common standard to judge scientific theories developed under different paradigms, there can be no sense in which theories developed after a scientific revolution can be said to add cumulatively to what was known before the revolution. Only within the context of a paradigm can we speak of one theory being true or false. Kuhn in Structure concluded, tentatively, "We may, to be more precise, have to relinquish the notion explicit or implicit that changes of paradigm carry scientists and those who learn from them closer and closer to the truth." More recently, in his Rothschild Lecture at Harvard in 1992, Kuhn remarked that it is hard to imagine what can be meant by the phrase that a scientific theory takes us "closer to the truth."

Kuhn did not deny that there is progress in science, but he denied that it is progress toward anything. He often used the metaphor of biological evolution: scientific progress for him was like evolution as described by Darwin, a process driven from behind, rather than pulled toward some fixed goal to which it grows ever closer. For him, the natural selection of scientific theories is driven by problem solving. When during a period of normal science, it turns out that some problems can't be solved using existing theories, then new ideas proliferate, and the ideas that survive are those that do best at solving these problems. But according to Kuhn, just as there was nothing inevitable about mammals appearing in the Cretaceous period and out-surviving the dinosaurs when a comet hit the earth, so also there is nothing built into nature that made it inevitable that our science would evolve in the direction of Maxwell's equation or general relativity. Kuhn recognizes that Maxwell's and Einstein's theories are better than those that preceded them, in the same way that mammals turned out to be better than dinosaurs at surviving the effects of comet impacts, but when new problems arise they will be replaced by new theories that are better at solving those problems, and so on, with no overall improvement.

All this is wormwood to scientists like myself, who think the task of science is to bring us closer and closer to objective truth. But Kuhn's conclusions are delicious to those who take a more skeptical view of the pretentions of science. If scientific theories can only be judged within the context of a particular paradigm, then in this respect the scientific theories of any one paradigm are not privileged over other ways of looking at the world, such as shamanism or astrolgy or creationism. If the transition from one paradigm to another cannot be judged by any external standard, then perhaps it is culture rather than nature that dictates the content of scientific theories.

Kuhn himself was not always happy with those who invoked his work. In 1965 he complained that for the philosopher Paul Feyerabend to describe his arguments as a defense of irrationality in science seemed to him to be "not only absurd but vaguely obscene"... But even when we put aside the excesses of Kuhn's admirers, the radical part of Kuhn's theory of scientific revolutions is radical enough. And I think it is quite wrong.

It is not true that scientists are unable to "switch back and forth between ways of seeing," and that after a scientific revolution they become incapable of understanding the science that went before it. One of the paradigm shifts to which Kuhn gives much attention in Structure is the replacement at the beginning of this century of Newtonian mechanics by the relativistic mechanics of Einstein. But in fact in educating new physicists the first thing that we teach them is still good old Newtonian mechanics, and they never forget how to think in Newtonian terms, even after they learn about Einstein's theory of relativity. Kuhn himself as an instructor at Harvard must have taught Newtonian mechanics to undergraduates.

In defending his position, Kuhn argued that the words we use and the symbols in our equations mean different things before and after a scientific revolution; for instance, physicists meant different things by mass before and after the advent of relativity. It is true that there was a good deal of uncertainty about the concept of mass during the Einsteinian revolution. For a while there was talk of "longitudinal" and "transverse" masses, which were supposed to depend on a particle's speed and to resist accelerations along the direction of motion and perpendicular to it. But this has all been resolved. No one today talks of longitudinal or transverse mass, and in fact the term "mass" today is most frequently understood as "rest mass," an intrinsic property of a body that is not changed by motion, which is much the way that mass was understood before Einstein. Meanings can change, but generally they do so in the direction of an increased richness and precision of definition, so that we do not lose the ability to understand the theories of past periods of normal science.

... [S]cientists who come of age in a period of normal science find it extraordinarily difficult to understand the work of scientists in previous scientific revolutions, so that in this respect we are often almost incapable of reliving the "gestalt flip" produced by the revolution. For instance, it is not easy for a physicist today to read Newton's Principia, even in a modern translation from Newton's Latin. The great astrophysicist Subrahmanian Chandrasekhar spent years translating the Principia's reasoning into a form that a modern physicist could understand. But those who participate in a scientific revolution are in a sense living in two worlds: the earlier period of normal science, which is breaking down, and the new period of normal science, which they do not yet fully comprehend. It is much less difficult for scientists in one period of normal science to understand the theories of an earlier paradigm in their mature form. I was careful earlier to talk about Newtonian mechanics, not Newton's mechanics. In an important sense, especially in his geometric style, Newton is pre-Newtonian. Recall the aphorism of John Maynard Keynes, that Newton was not the first modern scientist but rather the last magician. Newtonianism reached its mature form in the early nineteenth century through the work of Laplace, Lagrange, and others, and it is this mature Newtonianism--which still predates special relativity by a century--that we teach our students today. They have no trouble in understanding it, and they continue to understand it and use it where appropriate after they learn about Einstein's theory of relativity.

Much the same can be said about our understanding of the electrodynamics of James Clerk Maxwell. Maxwell's 1873 Treatise on Electricity and Magnetism is difficult for a modern physicist to read, because it is based on the idea that electric and magnetic fields represent tensions in a physical medium, the ether, in which we no longer believe. In this respect, Maxwell is pre-Maxwellian. (Oliver Heaviside, who helped to refine Maxwell's theory, said of Maxwell that he was only half a Maxwellian.) Maxwellianism--the theory of magnetism, electricity, and light that is based on Maxwell's work--reached its mature form (which does not require reference to an ether) by 1900, and it is this mature Maxwellianism that we teach our students. Later they take courses on quantum mechanics in which they learn that light is composed of particles called photons, and that Maxwell's equations are only approximate; but this does not prevent them from continuing to understand and use Maxwellian electrodynamics where appropriate.

In judging the nature of scientific progress, we have to look at mature scientific theories, not theories at the moments when they are coming into being.

... Nor do scientific revolutions necessarily change the way that we assess our theories, making different paradigms incommensurable. Over the past forty years I have been involved in revolutionary changes in the way that physicists understand the elemntary particles that are the basic constituents of matter. The greater revolutions of this century, quantum mechanics and relativity, were before my time, but they are the basis of the physics research of my generation. Nowhere have I seen any signs of Kuhn's incommensurability between different paradigms. Our ideas have changed, but we have continued to assess our theories in pretty much the same way: a theory is taken as a success if it is based on simple general principles and does a good job of accounting for experimental data in a natural way. I am not saying that we have a book of rules that tells us how to assess theories, or that we have a clear idea of what is meant by "simple general principles" or "natural." I am only saying that whatever we mean, there have been no sudden changes in the way we assess theories, no changes that would make it impossible to compare the truth of theories before and after a revolution.

... It is important to keep straight what does and what does not change in scientific revolutions, a distinction that is not made in Structure. There is a "hard" part of modern physical theories ("hard" meaning not difficult, but durable, like bones in paleontology or potsherds in archeology) that usually consists of the equations themselves, together with some understandings about what the symbols mean operationally and about the sorts of phenomena to which they apply. Then there is a "soft" part; it is the vision of reality that we use to explain to ourselves why the equations work. The soft part does change; we no longer believe in Maxwell's ether, and we know that there is more to nature than Newton's particles and forces.

The changes in the soft part of scientific theories also produce changes in our understanding of the conditions under which the hard part is a good approximation. But after our theories reach mature forms, their hard parts represent permanent accomplishments. If you have bought one of those T-shirts with Maxwell's equations on the front, you may have to worry about its going out of style, but not about its becoming false. We will go on teaching Maxwellian electrodynamics as long as there are scientists. I can't see any sense in which the increase in scope and accuracy of the hard parts of our theories is not a cumulative approach to truth.

... Kuhn's view of scientific progress would leave us with a mystery: Why does anyone bother? If one scientific theory is only better than another in its ability to solve the problems that happen to be on our minds today, then why not save ourselves a lot of trouble by putting these problems out of our minds? We don't study elementary particles because they are intrinsically interesting, like people. They are not--if you have seen one electron, you've seen them all. What drives us onward in the work of science is precisely the sense that there are truths out there to be discovered, truths that once discovered will form a permanent part of human knowledge.

Back to Lifschitz's web page