Concerning this relatively familiar thesis
None of this means that there are no sufficient reasons for being persuaded, or that such reasons are ultimately not decisive for the group. Nor does it mean that the reasons for choice differ from those usually enumerated by philosophers of science.
Accuracy, simplicity, fruitfulness, and so on are such reasons. What it does suggest, however, is that these reasons function as values, and therefore may be applied differently, individually or collectively, by those who hold them in common.
For example, if two people differ in their opinions about the relative fruitfulness of theories, or if they do not agree about the range to which those theories extend, neither can be said to be wrong. Neither is unscientific. There is no neutral calculus for theory choice, nor is there any systematic decision procedure which, properly applied, must lead every individual in the group to the same conclusion. In this sense, it is the community of specialists, rather than the individual members of the scientific community, that establishes a valid conclusion. In order to understand why science has progressed as it has, we need not strip bare every individual’s personality and the path by which he arrived at a particular choice, however fascinating a topic that may be. But we do need to understand how the shared values of a particular professional group interact with the particular experiences shared by that group, so as to ensure that the majority of its members will ultimately find one set of arguments more decisive than the rest.
That process is the path of persuasion.
But it raises a deeper problem. Two people who perceive the same situation differently and yet use the same vocabulary in discussion are clearly using words differently.
In other words, they are arguing from viewpoints I have called incommensurable. How can they hope to share anything, much less dare to hope to be persuaded? Even a preliminary answer to this question requires a more explicit account of the nature of the difficulty. I believe that, at least in part, it takes the following form. The practice of normal science depends on the ability to classify objects and situations into similarity sets derived from exemplars, and here the similarity sets are primitive in the sense that such classification is made without answering the question, “Similar in what respect, exactly?” Then a central feature of any revolution is that some part of the similarity relations changes. Objects that had previously been grouped in the same class are, after the revolution, classified into different sets, and the opposite also occurs. Think of the sun, moon, Mars, and Earth before and after Copernicus. Think of free fall, pendulum motion, and planetary motion before and after Galileo. Since, even in the altered classification, most objects continue to be grouped together, the names of the sets are largely retained. But shifts at the lower level often become part of a profound change in the network of interrelations among them.
The transfer of metals from the class of compounds to the class of elements played a decisive role in the emergence of new theories of combustion, acidity, and physical and chemical combination. In an instant, such changes spread throughout the whole of chemistry. It is therefore no surprise that, when such rearrangement is under way, two people whose conversation had previously seemed perfectly intelligible should suddenly find themselves responding to the same stimuli with mutually contradictory statements and generalizations. Such difficulties will not be felt in every area of their scientific discourse, but they are bound to arise, and when they do, they will cluster most densely around the phenomena on which the choice of theory most centrally depends.
Although such problems are at first conspicuous in the exchange of opinions, they are not merely linguistic, and the difficulties cannot be resolved simply by stipulating definitions for the troublesome terms. Because the words around which the difficulties gather and become entangled have in part been learned through direct application to exemplars, the parties confronted with a breakdown in communication cannot say, “I use the word ‘element’ (or ‘mixture,’ or ‘planet,’ or ‘unconstrained motion’) in the manner determined by the following criteria.” In other words, they both use it in the same way, and they cannot rely on a neutral language suited to both their theories or to the empirical consequences of both theories. Part of the difference is prior to the application of the languages in which, nevertheless, it is reflected.
But those who suffer such a breakdown in communication
must have something to rely on. The stimuli that come to them are the same. Their general neural apparatus is also the same, though it is wired differently. Moreover, except for the most important but minute areas of experience, even their neural programming must be almost identical, precisely because, apart from the immediately preceding past, they share a single history. As a result, their everyday world, most of their scientific world, and their language are all shared. Given what they have in common, they should be able to discover a great deal about how they differ. But the technique required for this is neither simple nor comfortable, nor is it part of the scientist’s regular arsenal. Scientists rarely truly realize what this technique is, and they very rarely employ it longer than the time necessary either to induce conversion or to confirm that conversion will not take place.
Briefly put, what the parties in a breakdown of communication can do is regard one another as members of different language communities and then become translators.17) Taking as a subject of study the very difference between discourse within the group itself and discourse between groups, they can first attempt to identify the terms and usages that are employed without much trouble within each group but become the focus of trouble in discussions between groups. (Usages that do not raise such difficulties can be translated as they sound.) Once they have singled out the troublesome areas in scientific communication, they can rely on the everyday vocabulary they share in an attempt to clarify their difficulties further. In other words, each will try to discover what the other would see and say when given various stimuli, in response expressed in that other’s own language. If they can sufficiently restrain themselves from explaining anomalous behavior as the result of mere error or confusion, they will soon learn to predict one another’s conduct with insight. Each will learn to translate the other’s theory and its logical consequences into his own language, while also learning to describe, in his own language, the world to which that theory applies. This is precisely the task that the historian of science regularly performs—or ought to perform—when dealing with an outmoded scientific theory.
If accomplished, translation enables the parties in a breakdown of communication to experience vicariously the strengths and defects of each other’s views.
For that reason, translation becomes a powerful means on both sides for persuasion or for conversion. But persuasion need not succeed, and even if it does, it need not involve conversion or be followed by it. The two experiences are not the same, an important distinction that I have only recently come fully to understand.
As I see it, to persuade someone
is to make him believe that my own view is superior and therefore ought to replace his own. That much is often accomplished without recourse to anything like translation.
Without translation, many of the explanations and problem-statements obtained by members of one scientific group will appear obscure to others. Yet from the start, each language group can usually produce some concrete research results, and these results, though they can be described in words understood in the same way by both groups, cannot be explained by the other group using its own terms. If the new viewpoint endures for a time and continues to achieve success, research results expressible in this way are bound to increase in number. For some people, such results alone will be decisive. They may say: I do not know how the advocates of the new view succeed, but I must learn. Whatever they are doing, it is clearly right. Such a response appears especially readily among those who have just entered the specialty, because they have not yet acquired the special vocabulary and commitments of either group.
But arguments described in vocabulary that the two groups use in the same way are usually not decisive, at least not until almost the final stage in the emergence of opposing views. Among views already accepted within a specialty, few are persuaded without some reliance on the more extended comparison made possible by translation. Though the price to be paid is often extremely long and complicated sentences—think of the debate between Proust and Berthollet conducted without borrowing the term “element”—many additional research results can be translated from the language of one scientific community into that of another. Moreover, as translation proceeds, some members of each group will begin to understand vicariously how statements that had previously been obscure could appear as explanations to members of the opposing group. Of course, the availability of such a technique does not guarantee persuasion. For most people, translation is an intimidating process, and it is wholly foreign to normal science. Nevertheless, as argument is piled upon argument, and as challenge after challenge is successfully met, only blind stubbornness ultimately remains as the reason for continued resistance.
Then, for both historians and linguists,
a second aspect of translation, long familiar, becomes decisively important. Translating a theory or worldview into one’s own native language does not make it one’s own. One must discover that one is not translating, but thinking and acting in what had formerly been an alien language. But such a transition, even when there are quite plausible reasons for wishing to make it, is not a change that an individual may or may not deliberately choose. Rather, somewhere in the process of learning translation, he realizes that a change has occurred, that he has slipped into the new language without any conclusion having been reached. Or else, like many of those who first encountered relativity or quantum mechanics in middle age, he finds himself completely persuaded by the new view and yet unable to internalize it, unable to be at ease in the world constructed by that view. Such a person has completed his choice intellectually, but the conversion required for it to become effective does not occur in him. He can use the new theory, but he will do so as a foreigner in an unfamiliar environment, and it becomes an option available to him only because natives exist there. His work is, in a sense, parasitic upon that of the natives, because he lacks the coordinates of the intellectual elements that future members of that community will acquire through education.
Therefore, the conversion experience, which I likened to a Gestalt
shift in the unified whole of experience, lies at the heart of the revolutionary process.
Good reasons for choice provide the motive for conversion and create a climate in which conversion is more likely to occur. In addition, translation provides the essential input for the neural reprogramming which, though for the present still unintelligible, clearly underlies conversion. But neither good reasons nor translation constitutes this collective conversion, and that is precisely the process we must explain if we are to understand the essential pattern of scientific change.
6. Revolutions and Relativism
One consequence derived from the position just outlined has been especially troubling to many of my critics.18) They see my view, particularly as developed in the final section of this book, as a relativistic position. The advocates of different theories are like members of different language-culture communities. Recognizing the similarity suggests that, in some sense, both groups may be right. When applied to culture and its development, such a position becomes relativistic.
Yet when applied to science, it may not
be so, and in any case, from a perspective that its critics have failed to see, it is far from mere relativism. I have argued that, whether as a single group or within groups, the practitioners of a mature science are fundamentally puzzle-solvers.
The values they deploy at times of theory choice may be derived from other aspects of their research, but the demonstrated ability to set and solve the puzzles presented by nature becomes, in cases of value conflict, the most conspicuous criterion for the majority of members of a scientific group. Like any other value, puzzle-solving ability proves ambiguous in application. Two people who share it may nevertheless differ in the judgments they draw from its use. But the behavior of a scientific community that makes it preeminent will be altogether different from that of one that does not. I believe that the high value assigned in science to puzzle-solving ability has the following inevitable consequence.
Imagine, for example, an evolutionary tree that begins from their common origins in primitive natural philosophy and
the crafts and develops into the modern scientific specialties. A line drawn along that genealogy, never doubling back from the trunk to the tip of any branch, would trace a lineage of theories connected by descent. If, choosing points not too close to the origin of those theories, we consider any two theories, even a relatively unconcerned observer will find it easy to devise a list of criteria by which the later theory may be distinguished from the earlier one. To name a few of the most useful: accuracy of prediction, especially of quantitative inference; balance between esoteric and everyday subject matter; and the number of different problems solved. Less useful for this purpose, though still important elements in scientific activity, would be such values as simplicity, scope, and harmony with other specialties. Such lists are not yet definitive, but I have no doubt that they can be completed. If they can, then the development of science, like biological progress, becomes a unidirectional and irreversible process. Later scientific theories are better than earlier ones at solving puzzles in the often quite different environments to which they are applied. This is not the position of a relativist, and it reveals the sense in which I am a convinced believer in scientific progress.
Compared, however, with the concept of progress most widely held among both philosophers of science and ordinary people,
this position lacks a crucial element. A scientific theory is usually felt to be superior to its predecessors not only because it is a better instrument for discovering and solving puzzles, but also because, in some way, it better represents what nature is truly like. It is often said that successive theories come ever closer to the truth, or approximate the truth more and more closely. Clearly, generalizations of this sort refer not to puzzle-solutions or to concrete predictions derived from a theory, but rather to its ontology, to the correspondence between the entities with which the theory populates nature and what is “really there.”
Perhaps there are other ways to recover a concept of “truth” applicable to whole theories, but this one will not serve. I do not think there is any theory-independent way to reconstruct phrases like “really there.” The notion of a fit between the ontology of a theory and its “real” counterpart in nature now seems to me, in principle, deceptive. Moreover, as a historian of science, I am struck by the improbability of that view. For example, I do not doubt that Newton’s mechanics improved upon Aristotle’s theory, and that Einstein’s theory improved upon Newton’s as an instrument for puzzle-solving. But I can see no consistent direction of ontological progress in their succession. On the contrary, in some important respects, though by no means in all, Einstein’s general theory of relativity is closer to Aristotle’s theory than either Einstein’s or Aristotle’s theory is to Newton’s. The temptation to describe this position as relativism is understandable, but to me the expression seems mistaken. Put differently, if this position is relativism, I cannot see that the relativist loses anything required for explaining the nature and development of science.
7. The Nature of Science
I will now conclude with a brief discussion of two frequent reactions to the text of my first edition. The first is critical and the second favorable, but in my view neither is entirely correct. The two are not related to what I have been saying so far, nor are they related to each other, but both reactions are sufficiently widespread that they seem to call for at least some response from me. Some readers of the first edition will have noticed that I repeatedly move back and forth between descriptive and normative modes. Such shifts were especially conspicuous in the frequently used sort of sentence that begins, “But that is not what scientists do,” and ends by asserting, “Scientists should not do that.” Some critics charge that I confuse description with prescription, thereby violating the venerable philosophical maxim: “Is” cannot imply “ought.”19)
In fact, that maxim has become so commonplace that it is no longer a respected principle anywhere. Many contemporary philosophers have found important contexts in which the normative and the descriptive are mixed so that they cannot be distinguished.20) “Is” and “ought” are by no means always separate in the way they have been thought to be. But there is no need to rely on the subtleties of contemporary philosophy of language to clarify the apparent confusion in my position on this point. The earlier part of this book presented a viewpoint or theory about the nature of science; like other philosophies of science, that theory bears responsibility for how scientists ought to behave if their activity is to succeed. It need not be any more correct than another theory, but it provides a legitimate basis for repeated “oughts” and “shoulds.” Put differently, one set of reasons for treating the theory seriously is that scientists, who have developed their methods and selected them for success, in fact behave as this theory says they must. My descriptive generalizations are evidence for the theory because those generalizations can also be derived from the theory, whereas for other views of the nature of science they constitute anomalous behavior.
I do not think the circularity of this argument is
a defect. The logical consequences of the viewpoint under discussion are not exhausted by the observations to which that view was originally directed. Even before this book was first published, I had found that parts of the theory it presents were useful tools for the study of scientific behavior and scientific development. A comparison between this postscript and the first edition should suggest that it has continued to play such a role.
Not just any circular point of view provides such guidance.
Concerning one final
reaction to this book, my answer must be somewhat different. Many people who found the book welcome and satisfying did so because they grasped its main theses as applicable to many other fields besides science, rather than because of what it reveals about science. I know what they mean, and I do not wish to discourage their attempts to extend the position. Nevertheless, their reaction also leaves me somewhat puzzled. Insofar as this book portrays the development of science as a succession of tradition-bound periods punctuated by non-cumulative breaks, its theses will undoubtedly have broad application. But the reason they naturally should is that those themes were borrowed from other fields. Historians who study literature, music, art, political development, and many other human activities have long described their subjects in the same way. Periodization divided by revolutionary breaks in form, taste, and institutional arrangements has reigned among their standard tools. If I have been original with respect to these concepts, it is chiefly by applying them to science, where the idea that development occurs in a different way had been widely accepted. And I regard the concept of concrete achievement, of the paradigm as exemplar, as a secondary contribution. I suspect, for example, that the troublesome difficulties surrounding the concept of style in art might disappear if paintings could be regarded as having been completed by taking one another as models, rather than as being produced in accordance with some abstract canon of style.21)
But this book also, of another kind of
There is a point I wish to make explicit, one that has not been readily apparent to many readers of this book. The development of science may resemble development in other fields more closely than has often been supposed, and yet it is also entirely different. For example, to say that science, at least after a certain point in its development, has progressed in a way that does not occur in other fields is not altogether wrong, whatever progress itself may be. One of the aims of this book has been to examine those differences and to explain them.
Consider, for example, the repeated emphasis above on the fact that in the mature sciences there are no competing schools, or that they are quite rare—this, I think, is how I must now put it.
Or recall what I have said about the extent to which the members of a given scientific community constitute the sole audience and the sole judges of that community’s research. Think again, too, about the peculiar character of scientific education, about puzzle-solving as an aim, and about the value system that groups of scientists develop in times of crisis and decision. This book brings to light other characteristics of precisely the same kind; none of them need be unique to science, yet they are involved in distinguishing scientific activity.
There remains a great deal to be discovered about all these characteristics of science. Since I began this postscript by emphasizing the need to study the structure of scientific communities, I shall now end by emphasizing the need for similar studies of corresponding groups in other fields, and, above all, for comparative studies. In any particular community, scientific or otherwise, how are members chosen, and how does one come to be chosen as a member? What is the process, and what are the stages by which one is socialized into the group? What, collectively, does the group regard as its purpose? What deviations will it tolerate, whether individual or collective, and how does it regulate deviations it cannot accept? A more thorough understanding of science requires this and more; there is no other domain in which the need is so acute. Like language, scientific knowledge may be intrinsically the common property of a group—or else nothing at all. To understand this point, we shall need to know the distinctive characteristics of the groups of scientists who create and use scientific knowledge.]