The Online Library of Liberty

A project of Liberty Fund, Inc.

• The Selected Works of Gordon Tullock
• Introduction
• Preface and Acknowledgments
• The Organization of Inquiry
• Chapter I: The Social Organization of Science
• Chapter II: Why Inquire?
• Chapter III: The Subject and Methods of Inquiry
• Chapter IV: Data Collection
• Chapter V: The Problem of Induction
• Chapter VI: Verification and Dissemination
• Chapter VII: The Backwardness of the Social Sciences
• Chapter VIII: Practical Suggestions

CHAPTER III

THE SUBJECT AND METHODS OF INQUIRY

The subject of inquiry is, quite simply, anything which anyone might be curious about or which might be practically useful. This formula sounds simple almost to the extent of simple-mindedness, but in fact it conceals a number of difficult problems. Let us start with curiosity, which is defined as “the desire to learn or know about anything.”1 It has been quite seriously questioned whether it is possible really to learn or know about anything. This position, which originated in religious speculation, has a number of variants. The extreme position, solipsism, is logically invulnerable. I cannot really prove to any adherent of this position that either I or this book exists and naturally have no chance to prove that anything else exists. All I can say is that I very much doubt if anyone really believes in solipsism.

A less extreme position, best stated by Bishop Berkeley, does not doubt the reality of the world, but points out that there is no proof that this real world corresponds to the sense impression which we receive from our necessarily imperfect sensory equipment. Again, the position is logically invulnerable; there is no way of demonstrating that the appearances and the realities coincide, since any evidence we present will refer only to the appearances. It should be noted, however, that the opposite position, that this paper is, in fact, white and that the ink is black, is equally invulnerable logically; it also cannot be disproved by any conceivable test. To the practical man the problem seems to be of little relevance. It makes no real difference whether the statement “If we do A then B will result” or the statement “If we do something which appears to be A, then something which appears to all of our senses to be B will happen” is true. To the pure scientist, however, the question is a relevant one.2

How, then, does he handle this important but apparently unanswerable question? It is hard for us to tell what someone else really thinks, but certainly all scientists act as though they believe that the appearances and the realities coincide. To most laymen this will appear so obvious that they will wonder at my spending so much time on it, but to some scientists it will appear to be wrong. Ever since the time of Galileo there have been scientists and philosophers of science who have held some variant of the views so well expressed by Bishop Berkeley. In the latter part of the nineteenth century this view rapidly expanded its influence, and in the first half of the twentieth it was very widely held. Poincaré, Duhem, and Mach are probably its most important modern proponents.3

The basic view of reality shared by these scientists can be illustrated by Toulmin’s condensation of Mach’s views:

Suppose we heat an enclosed vessel containing a gas and record the gas pressure at various temperatures. If we are careful in making our observations, the pressure will vary in proportion to the (absolute) temperature of the gas. The statement “The pressure varies as the absolute temperature varies” is certainly more compact than a table showing a vast number of readings of our instruments, but is it any more than that? The layman would immediately answer that it represents a scientific law and that its discovery was an appreciable increase of our knowledge. For a proponent of Mach’s position, however, it is really only another way of presenting the table of instrument readings, superior only because it takes up less space and is easier to remember. The real discovery, in this view, consists of the initial instrument readings.

The position of the believers in this version of Bishop Berkeley’s philosophy, although they generally do not realize it, is that of the Jesuit cardinal who was Galileo’s principal intellectual opponent. St. Robert Bellarmine wrote:

Galileo’s crime consisted in not taking the hint in the cardinal’s first sentence. He refused to act “prudently” and maintained that he had discovered a real law of nature rather than a simple system for tying a set of observations into a neat bundle.

In fact, of course, all scientists act as if they believe they are engaged in uncovering the real world. They show no signs of real doubts about the application of their theories to reality. “Albert Einstein has remarked that if you want to know what a scientist really believes, don’t listen to what he says, but observe what he is working on.”6

If the theories were merely devices for conveniently summarizing experimental results or mnemonic devices to make it easier to keep the results in mind, the interest of all scientists in improving them would be inexplicable. There are only a few experimental observations which can be treated more simply by the Einsteinian system than the Newtonian. If they were not interested in the truth of their theories, obviously the scientists would recommend the Einsteinian solutions for these few problems and use the Newtonian system everywhere else. The question of which was true would not arise, and the work of Einstein would be regarded as simply a minor step forward. The actual attitude of scientists on this problem is clearly contrary to Mach’s position. Einstein’s work is held to have disproved Newton, a conclusion obviously impossible if we assume that theories are simply abbreviated statements of observations. Nevertheless, the vast majority of all computations undertaken by physicists are strictly Newtonian. Only in a few special fields7 has the work of Einstein actually had any significant effect on the everyday work of the physicists.

Any economist can testify that many highly successful businessmen have great difficulty in explaining how they behave, even to themselves. They are apt to grab hold of various ill-conceived theories, but they obviously do not really understand their own behavior in an analytical sense. In part, this is simply an illustration of the principle of division of labor. The manager of a plant making small electric motors has all sorts of difficult and urgent problems, and working out a careful and consistent explanation of his own behavior in solving these problems is not among them. That is a matter for another specialist in the system of division of labor, the economist. If, after the problem has been solved by the economist, he presents the results to the businessman, he should not be surprised if they are rejected. The businessman has spent the day worrying about whether he should buy a set of new core-winding machines and still has not made up his mind. If his attention is directed to the problem of why this worries him, he is likely to seize upon some simple solution which will not distract him from his preoccupation with the core-winders and to maintain that economists are impractical theorists.

The same phenomenon may affect scientists. A man who is extremely interested in the results obtained by high-energy accelerators and who is working night and day to put these results into some kind of theoretical framework is not likely to be highly motivated to study the problem of why he is so interested and why he chooses the particular methods he uses to solve his problems. Like the businessman, he is likely to pick up some apparently simple solution as a way to economize on his time. We need not object to this; the division of labor is one of the indispensable requisites for human life, but we need not, also, pay much attention to either the scientist or the businessman when they analyze their own activities.

A second reason for some scientists’ denying the possibility of truth in their own theories arises from the rather peculiar real situation of those theories. The history of science has been a history of the disproof of theories.8 Observations have tended to stand the test of time quite well (although even here we make improvements which sometimes disprove rather than improve an earlier observation), but theories seem not to last. As a result of this experience, scientists are trained to be skeptical of their theories.9 They are thought of as hypotheses, always living in the shadow of potential disproof. Every scientist must keep continually in the back of his mind the possibility that the theory on which he is currently working may simply be wrong. But if this is true of generally accepted theories, new theories are subject to even more suspicion. Until a theory has been in existence for some time and has been subject to considerable testing, the scientist can have little real confidence in its validity. Nevertheless, frequently the most fruitful line of search open to a man interested in a given field is to apply a new, and hence dubious, theory. Under these circumstances, it is clear that scientists can have little psychological confidence in their theories and that the development of a “theory of theories” which simply denies the validity of theories as anything other than a simplified set of observations is not an unexpected result.

The hollowness of this “theory of theories,” however, can be readily seen if we consider the situation when a new theory is proposed. If neither the new or the old theory which it purports to replace represents any deeper truth, why should we choose one over the other? The conventional answer is that we choose the simpler. When, however, this explanation is used to explain the demolition of the Newtonian system by Einstein, it is clear that the word “simple” must have some meaning in this context other than its dictionary definition. By almost no stretch of the imagination can the Theory of Relativity be considered simpler than Newton’s system.10

The scientists’ acceptance of the Einsteinian approach as “simpler,” however, can be justified if “simple” is given some other meaning than it bears in common speech (or that intended by Occam when he propounded his famous “razor”). If in this context “simple” means something else than it does in ordinary speech, then it would appear that anyone using it to justify the choice of one theory over another should tell us what he means by the word. He should also tell us why “simplicity” in this special sense is desirable. A theory which is simple in the normal meaning of the word clearly has some advantages over one which is complex, although whether this is a decisive advantage is a question to be determined in each case by considering other matters, too. If “simplicity” is assigned some other meaning, however, then we may ask whether this other “simplicity” is desirable. No decision on this matter can be reached, however, until we have an explanation of what “simplicity” means in this context.

It has been suggested by Michael Polanyi that what is really meant by the scientist when he says that one theory is simpler than another is that he feels it is closer to the truth.11 According to this view, the scientist is behaving like a businessman who explains that his prices are determined by adding a percentage mark-up to the cost. In both cases, the man is producing an erroneous rationalization for his own behavior, but his behavior is, in fact, highly rational. If Polanyi is correct, and I believe he is, then the scientists are not searching for something obscure which they choose to denominate by the word “simplicity” but are searching for truth.

The theory that scientists are in search of truth raises certain logical problems. One such which I regard as false has to do with the definition of “truth.”12 I do not believe that anyone really has any difficulty understanding what this word means, and I think that people who appear to have such difficulty are actually worried about other problems, particularly whether there is any real thing which corresponds to the concept and how we recognize truth. Both of these are real problems and worth discussing briefly. We may begin with the problem of whether science can hope to reach any real truth about the universe.13

We have seen that most theories now in use are regarded by scientists as of dubious validity, and we cannot, therefore, allege that they are true with any high degree of confidence that our own statement is true. But the view that present theories are to be viewed with skepticism does not necessarily imply that all are untrue. Mathematical propositions, to take an extreme case, are normally accepted as being really true. It is possible that some proofs and demonstrations will be found to be flawed in the future, but the bulk of mathematics will stand up to future critics as well as Euclid has stood up to his critics in the last two thousand years. Theories about the real world are obviously more subject to suspicion. Most of the theories propounded by the Greeks have been discarded. Some, however, have been retained. The theory of the lever remains as the Greeks left it, and a few other examples could be found. The theory of the lever may, of course, be disproved tomorrow, but the fact that it has withstood two thousand years of critical examination, much of it using tools which the Greeks could not even dream of, does raise some presumption that here we have a bit of theory which is absolutely true. It seems likely that somewhere in our present vast collection of theories there are others which are, in fact, true, that is, which will not be disproved at any time in the future. It is, of course, impossible to say which they are.

We must be skeptical about each theory, but this does not mean that we must be skeptical about the existence of truth. In fact, our skepticism is an illustration of our belief in truth. We doubt that our present theories are in fact true, and look for other theories which approach that goal more closely. Only if one believes in an objective truth will experimental evidence contrary to the predictions “disprove” the theory. We are skeptical of our present theories because we suspect they do not coincide with the truth. The progress of science consists of developing ever newer theories which approach ever closer to the truth. For the logical reasons we discussed earlier, we can never really be sure that the truth is there, but we have no choice but to act as if it is. Further, the success of our theories in predicting new observations is evidence, albeit not conclusive evidence, that they do have some relation to reality; and steady improvement in this respect can be taken as indicating that they are continually getting closer.

The second problem connected with our position that science is based on a belief in the existence of a real universe susceptible to human understanding and knowledge concerns the test of the truth of any given proposition. The logical problems connected with this question have been exhaustively discussed by various philosophers; Karl Popper’s The Logic of Scientific Discovery14 may be taken as a summation of the present position. We need not, therefore, go into these problems, but it is necessary to discuss two possible “tests of truth”: “workability” and “consensus of the informed.” Before discussing the workability test, we must devote some time to the subjects investigated by applied science, for this test is closely connected with the practical use of science.

If pure science seeks truth, applied science seeks useful information. The difference may be seen most clearly by noticing how often engineering magazines run articles discussing ways of approximating various functions. Serious research is put into developing formulas which, although known to be wrong, are simple and give results approximating the correct result. Frequently an engineer, in undertaking a particular piece of research preliminary to designing some useful device, will be able to choose among three or four formulas, all of which approximate the correct formula with varying degrees of fit under various conditions and which are of different degrees of difficulty.

Nevertheless, note that these “theories” are referred to as approximations. Although they work very well and are simpler than the “true” theory, even the practical man who uses them agrees that they are incorrect. The situation may be considered as analogical to the following figure. The numbered points indicate observations; the small letters minor theories which bind together these observations into a coherent system; and the over-theory X connects these minor theories into a rational whole. Reversing our explanation, from X, a, b, and c may be deduced; from a, 11, 12, and 13 follow. Note, however, that there are other alternate minor theories available. Anyone interested only in 31, 32, and 33 could also use c′. If it was simpler, the applied scientist (or even the pure scientist) would have no hesitancy in so using it.15 Similarly in the 20 area, a scientist has his choice between theories b and b′′, which cover the whole area, and b′, which covers only part of it. For practical purposes, he chooses the most convenient.16

From the standpoint of the applied scientist, this raises no particular problem. He chooses the “theory” which works best in his individual problem, but note the theoretical situation. A number of theories fit each given set of facts, and for practical purposes they may be interchangeable.17 Workability, then, cannot be a decisive criterion in determining the correct theory. Nevertheless, the applied scientist using theory b′ for some practical purpose will be perfectly willing to agree that theory b is better. He will refer to b′ as an approximation, but in this meaning “approximation” is a little different from its ordinary usage. Scientific experiments are frequently difficult, and a certain vagueness in result is normal. Graphically, the results, instead of falling on a simple curve, vary around a sort of zone. Under the circumstances several equations are likely to be about equally good fits. Normally, in fact, a scientist in possession of a computer can get the best fit by instructing the machine to use one equation on one part of the data and another elsewhere. Applied scientists frequently do just that. Thus the “true” theory, to which the others “approximate,” may (or may not, of course) actually deviate from the measured data more than the “approximation.” What the scientist means, in this case, is that b′ cannot be deduced from X while b can. He therefore believes b to be correct and b′ incorrect regardless of the question of which fits the actual data (within limits) more closely.

The same attitude can be seen more clearly in those not uncommon situations in which a series of experiments have produced some data which an applied scientist needs, but which cannot be deduced from any existing over-theory. Under these circumstances, shown at the right of our diagram, the applied scientist may use an equation d′ in his work, but refer to it as an approximation in spite of the fact that there is nothing for it to approximate. In essence, he is assuming that it is incorrect in spite of its agreement with his observations because it does not fit into the general theoretical picture. He believes that another theory, d, will eventually be invented which fits the data in the 40’s and which can be connected to the over-theory X or, as in our diagram, to a new and higher over-theory Y. This belief is obviously not empirically based and can be considered either as an act of faith or as based on the same general picture of the universe held by the pure scientist.

The pure scientists sometimes do the same thing. Gregor Mendel introduced the “genes” into his theory of heredity not because he believed that they existed, but because he thought they made it easy to understand the data. They served much the purpose of d′ in Figure 2. In this case Mendel, who was right in so much else, turned out to be wrong. The genes do exist, although it took almost forty years for the biologists to realize this.

The applied scientist agrees with the pure scientist that it is impossible to reach the truth and that the great general theories represent a closer approach to that truth than the small theories which bind together small parts of experience. With regard to the small theories, he uses them more or less regardless of their relation to the grand theories and with heavy emphasis on their simplicity. The grand theories may well have no close relation to practical work. The minor theories deduced from them may, of course, but these are not really dependent on the higher theories. Practical engineers have gone on using the Newtonian mechanics for fifty years now in spite of its destruction by Einstein. They are not perturbed by the lack of any formally worked out relationship between the minor theories they use, which work well, and the new mechanics.

This is not said in a spirit of criticism. Both applied and pure science have important roles to play in our struggle to improve our position. The roles are, however, different. The applied scientist is mostly interested in minor theories covering the small segments of reality in which he is operating. The pure scientist is interested in the grand amalgamations which bind all of these little theories together. This difference arises naturally from their different basic motives, but the two tasks supplement rather than contradict each other. The applied scientist will attempt to deduce minor theories with practical applications from the grand synthesis of the pure scientist, and this is a valuable check on the accuracy of the theories. Similarly, the numerous minor theories developed by the applied scientist provide data and stimulation for the theorist. Thus “workability” is a necessary, but not a sufficient, condition for a valid theory.

Turning now to methods of recognizing the truth, the consensus of the informed is often urged as decisive. This is a natural result of the fact that men are not infinitely capable. We can hardly hope to investigate more than a very small part of reality ourselves and therefore must rely on the division of labor for the bulk of our information about the universe. In areas where we have not taken the trouble to become good judges ourselves, we must, of necessity, accept someone else’s judgment. Naturally, we should try to select the best authorities for our information about the area which we do not intend to investigate ourselves, but we should keep in mind that the experts have often been wrong.18 But let us inquire how the “informed” themselves judge the truth or falsity of a position. Clearly, they cannot base their judgment on their own position, which is what consensus of the informed would mean in this situation. They must, then, have some other, superior, method of determining the truth, and this method is simply themselves investigating and reaching personal judgments on the truth of the matter. An intelligent outsider who has the time and interest in the problem should investigate, himself, since only in this way can he reach the level of certainty of the experts themselves. Personal knowledge is always superior to hearsay, and, paradoxically, the fact that we seek out the best experts in each field for information is itself proof that we consider it to be so.

To elaborate further, it is safe to assume that each investigator would like information (whether his basic motive is curiosity or the desire to make use of the information in a practical way is irrelevant here) on a vast number of matters. We can arrange this desired information in an array so that the most desired information is at the top and the least desired is at the bottom. This would roughly represent the proportionate amount of effort he would be willing to put into a search for each type of information. To this schedule we must attach a second which shows the individual’s view of the difficulty of obtaining each bit of information. From this schedule, if it were at all realistic, it would be immediately obvious that any individual in his lifetime could investigate only a comparatively few problems. Further, leaving aside mathematics, obtaining absolute certainty would probably be outside the real possibilities even if a full lifetime were devoted to the task.

Returning to our earlier discussion of general and particular curiosity (the same point can be made for applied science), a man has a general curiosity in a certain area; as a result of this curiosity and his estimate of the difficulties of investigation, he develops a particular curiosity in some given subarea and undertakes an investigation. This is unlikely to lead to theories which are absolutely true in our present state of knowledge, but normally at least some improvement will result. The improved knowledge in this area changes the schedule of priorities for the remaining problems and also changes the estimates of difficulty (it will possibly also add some new problems to the schedule). On the basis of this new schedule, the investigator now turns to another problem, which may, of course, be a development of his previous one.

So far, however, we have described the behavior of an investigator dealing with problems near the top of his priority schedule.19 Only a few problems, even only a few of the problems relative to his particular research, are within the scope of the possible investigation of one man. He must therefore depend on others for much information.20 As we work down the priority schedule, we proceed from information which the investigator ranks high enough to determine for himself to information that he is willing to take on trust. This latter field, however, is also ranked. Here the relevant criteria are the importance of the information to the investigator and the ease of obtaining someone else’s findings. Here again, the investigator will economize on the effort he puts into obtaining different types of information. In some areas he will read numerous reports and carefully judge their merits. In others he will simply consult a single “standard authority.” And in the ultimate case, he simply accepts what information happens to come his way without effort or verification on his part.

All of this is a fairly good description of how scientists really act. The fact that such a description can be derived from our basic postulates is evidence that they are true. The similarity to economics is, of course, very strong. In both, men are treated as attempting to reach desired ends with limited means. Thus science has a relation to economics above and beyond the trite observation that it must be paid for and is one kind of economic activity.

We are now in a position to discuss the methods actually used by scientists in deciding which theory is to be provisionally accepted. Out of the infinite universe of possible theories, those which clearly conflict with the evidence are first ruled out. This is simply requiring that the theory fit the real world. Unfortunately sometimes no theory has been invented which does not conflict with at least some data. In such cases a judgment must be made as to which existing theory is closest to the real world. If a number of theories which fit the real-world data are now available—and this also will frequently be the case—the one (or several) of the higher order of generality will be selected. If we still have more than one theory, we will choose the simplest. Thus simplicity returns to our picture of the behavior of scientists, but in the ordinary, everyday meaning of the word. Further, simplicity appears in subordination to resistance to disproof and generality.

This pattern of behavior can be most readily explained by a belief on the part of the scientist that the universe is logically ordered, and that both the nature of the universe and its order are comprehensible to the human mind. Scientific progress thus would approach, although not necessarily reach or even get very close to, a single grand theory explaining everything in the universe. In the late nineteenth century this ideal was actually held by most scientists, who thought of each theoretical or experimental advance as bringing us closer to the distant goal. Now, in the latter half of the twentieth century, few scientists consciously think of their work in these cosmic terms, but their behavior patterns still match those of the earlier investigators.

In addition to pure curiosity and the desire to make practical use of new discoveries, there is another reason for undertaking research, which we have denominated “induced curiosity.” An investigator wholly motivated21 by induced curiosity is different in many ways from one motivated by either curiosity or a desire to make practical application of new knowledge. In the first place, from the standpoint of the man whose curiosity is induced, scientific concern with the real world is secondary to other matters. If he could establish and maintain his reputation, and hence his job, by reporting completely fictional discoveries, this would accomplish his end. While an investigator motivated by curiosity or practical utility must, of necessity, concern himself with real phenomena, the man motivated by induced curiosity could, if the risk of discovery were not great, simply ignore reality.

This, of course, presents a problem for those administering a system of induced research. They must make certain that the investigators are induced into research. They must make certain that the investigators are induced to pay attention to the real world. As we have seen, the actual system used by administrators in our present setup is simply to count the number of papers published by a man in journals of various degrees of reputation. The reputation of the journals, again as we have seen, is determined by their readers. Now, a man motivated by induced curiosity reads a journal not because he is interested in the facts reported, but because he hopes to find a suggestion for work of his own. It will be seen that a self-perpetuating process might be set in motion in which a journal read only by people motivated by induced curiosity gradually slipped away from reality in the direction of superficially impressive but actually easy research projects.

In most sciences this does not happen, and we may profitably consider why. First, journals are normally read by applied scientists as well as by pure scientists. The applied scientists are apt to seize on any idea and try to make some use of it. If the failure of their application to work leads them to doubt the original research, they may protest strenuously. On a lower level, the applied scientists are likely to be mostly interested in articles which promise some real advance in the area in which they work. They are normally under considerable economic pressure to confine their reading to things which might lead to practical advances. They are therefore less likely to be interested in ideas and formulas for purely aesthetic reasons and more interested in reality than are less practical people.

The curiosity-motivated investigators also are a major hazard to any man who is simply trying to make money through having articles published and who would rather not do the research. The barriers to false research would thus appear to be high, and a man whose only motive for research is induced curiosity is usually held to fairly high standards. We can, however, see certain conditions which might conduce to a sharp reduction of the quality of induced research. The first of these conditions would be a lack of likely practical applications for research in a given field. If people whose sole objective is to find something out of which they can make money are not likely even to read the journals, then the investigators in that field are subject to much less pressure.

Further, if the field is one in which there are vastly more people (as a result of the necessity of staffing teaching posts in each field according to the number of students) than would appear justified by the likelihood of making discoveries of any significance, then there will be more pressure to make false discoveries or to present trivial discoveries as major. This kind of situation is one in which all of the people in the field are apt to be looking primarily for an opportunity to do something which can be made to look like research, and the reputation of journals is consequently likely to be dependent on the aid they give in this endeavor. One symptom of the existence of this condition is the development of very complex methods of treating subjects which can be readily handled by simple methods. Calculus will be used where simple arithmetic would do, and topology will be introduced in place of plane geometry. In many fields of social science these symptoms have appeared.

The people really interested in the truth could, of course, prevent this development if they were present in sufficient numbers. If, however, they are a small minority in any given branch of investigation, then they are not likely to be able to set the tone of investigation. In areas where the number of teaching positions is vastly greater than the apparent likelihood of discovery, people who are simply curious tend to be a small minority, since they are generally attracted to areas where discoveries in the real sense are likely. It also seems probable that people who are genuinely curious are apt to have higher IQ’s than those who are only induced (and by comparatively small salaries in the present-day United States) into research. Those areas of research which tend to have lower average intelligence among their workers also should be suspected of having relatively few persons who are genuine pure scientists.

Lastly, in an area where motives other than seeking the truth are important, the induced researcher is unlikely to be held to a high standard. If it is widely believed that the function of the researcher in a given field is to uphold some special point of view or if it is doubted whether anything approximating “truth” is really existent in a given field, then the standards to which an induced researcher must conform may be deplorably low. Simply presenting a rationalization for some position chosen on other grounds may be acceptable as an objective of research, and the principal criterion in judging journals may become their points of view. The concern with reality which unites the sciences, then, may be absent in this area, and the whole thing may be reduced to a pseudo-science like genetics in Lysenko’s Russia. Again, these symptoms may be found in some of the social sciences.

So far, I have discussed science and inquiry as though they were the same thing. In one of the general uses of inquiry, this is true, but in other meanings of this term they are different. Investigations may be started which are not motivated by either curiosity about reality or the desire to make practical use of knowledge of the real world, but by some other motive. A lawyer building up a brief for his client, for example, may be much more intelligent, more learned, and more ingenious in his research methods than most scientists, but his investigation is not scientific, because he is not searching for the truth. He looks for an argument, based on factual information to be sure, which he thinks will persuade. Whether it is true or not is not of his concern. In fact, in the Anglo-Saxon adversary type of legal proceedings, he is prohibited from expressing his personal opinion on this point in court.

Consider, for another example, an advertising man trying to produce copy which will sell toothpaste. The higher ranking advertising copywriters, certainly highly intelligent and ingenious men, devote large amounts of time and effort to the search for good slogans, but this again is not a search for truth. Truth-seeking, however, may be found even here. If we have several slogans, none of which, shall we say, is true, the question of which of these will convince the most people is subject to investigation which aims at obtaining a truth (i.e., which of these false statements will sell the most toothpaste).

Even if we limit ourselves to investigations aimed at reaching the truth or at least an approximation of it, it is obvious that many such investigations would not normally be called scientific. A jealous husband who hires detectives to determine whether his wife is adulterous is engaging in an investigation aimed at reaching the truth, but this is hardly part of science. Popular usage, however, is of little help in distinguishing between scientific and non-scientific fields. Archaeology, for example, is considered a perfectly respectable science, while history normally is not. Since the difference between archaeology and history is precisely that we have better information in historical fields,22 it would appear that the popular definitions are unclear.

Left to myself, I should like to define science in such a way that only fields in which fairly elaborate theoretical structures have been developed, like physics or economics, would be included. Other areas, like biology (minus genetics and a few other specialties), would not be called sciences, because they have not yet attained the theoretical stage.23 This, however, would be a sharp deviation from customary usage, and I will, accordingly, confine myself to distinguishing between two types of science, the theoretical and the empirical. Since both of these words are “plus” words, I hope that scientists in the two categories will not object.

It must be pointed out that the distinction is one of degree, although most disciplines are mainly in one category or the other. Even mathematics in practice concerns itself to some extent with the real world. Recently, in fact, a sort of experimental method has been introduced under which various problems, such as commensal work, are tried out on computers instead of solved in the conventional manner. At the other extreme, even a biologist engaged solely in collecting specimens in a previously unsurveyed area does have some theories and hypotheses. The difference between a “theoretical” science like physics and an “empirical” science like most botany is a difference in emphasis. It is also, probably, a difference in stage of development. The “theoretical” sciences can, I think, justly claim to be more highly developed than the “empirical” ones. We can hope that the future will bring general theories into the presently empirical areas.

Although we now have a principle for distinguishing between two kinds of science, we are still lacking one for distinguishing between science and other types of inquiry. It is rather widely held that the difference is one of “method,” but this seems unlikely. There is no evidence that the brains of scientists work differently from those of other men. So far as we can see, the primitive caveman had as good mental equipment as the modern man, and some of the cavemen certainly had the inherent equipment to become Nobel prize laureates. Not all of the most intelligent men of modern days are engaged in science, and there is no particular reason to suppose that highly intelligent men engaged in investigating some matter without the scope of science would refrain, either from ignorance or desire, from using any technique of investigation known to the scientists. It would appear, then, that there must be something special about the situation in which the scientist finds himself which resembles other, non-scientific structures.

Our language, at this point, has a false implication. The place where a scientist works is generally called a laboratory and his work research, regardless of what he is doing. Since the word is the same, a feeling that the work is also somehow the same has developed. Actually this is quite untrue. Scientists carry on the most diverse activities. Probably the category of “all things done in laboratories” is larger than the category of “all things done in structures which are not laboratories.” Certainly, the two classes are of the same order of magnitude. Laboratories where radically different problems are being studied will normally resemble each other no more than they resemble other, non-scientific structures.

Similarly, “research activities” is very likely a wider category of action than “non-research activities.” There is a sort of order and tendency to repeat in non-scientific activities, but the active and ingenious scientist is always thinking up some completely new approach to a problem which has never been tried anywhere else. The view that the nature of the activity engaged in is the defining characteristic of science is a fairly widely held one, but is clearly wrong. It is perfectly possible for two men, one a scientist and the other not, to do exactly the same things, but the scientist will still be doing scientific work while the layman will not.

But if science is not distinguished from other activities by the methods it uses or by the fact that it seeks the truth, what does so distinguish it? There are, I think, two answers to this question. The first, which is a little doubtful, is that the scientist seeks general truth.24 He is not interested in the truth or falsity of a proposition about some specific person, place, or thing, but of more general propositions. The biologist is basically uninterested in the particular monkey he is studying; what he is trying to unveil are general “facts” which are true of all monkeys (or, all male rhesus monkeys one year of age who have been infected with a certain virus). Even when he concerns himself with individual variations, he will turn out to be interested in general measures of the range of variation to be expected in the species, not in some specific individual.25

A close look at what scientists actually do, however, indicates that while they are normally interested only in general truth, they will sometimes be concerned with particular truth. If we accept archaeology as a science, it is almost always concerned with the particular rather than the general. Geology, which is clearly a science, is quite frequently concerned with the particular truth of the underlying structure in some specified region. Even the more general sciences like physics and chemistry may occasionally be concerned with singular facts. Recently astronomers and physicists became much interested in the discovery of bright and temporary red spots on the surface of the moon. These were thought to indicate some sort of volcanic activity. Clearly the presence or absence of volcanos on the moon is a particular fact, albeit one of considerable interest. Nevertheless, most scientists, most of the time, are searching for general truth, not specific truth.

The second distinguishing characteristic of science is, in my opinion, the membership of all scientists in the scientific community. It is not anything special about the individual scientist, or his work, which distinguishes him, but the special human environment in which he operates. This environment is, in many ways, most peculiar. A scientist specializing in some narrow field may almost never meet the other members of the community with whom his intellectual contacts are most intimate. His important colleagues may live in other countries, be the products of violently different cultures, and speak no language that he knows. Nevertheless, his activities are controlled and shaped by these others, while the people whom he meets regularly, his family, friends, and colleagues on the local university faculty (if he teaches for a living), have almost no effect on his work.

Consider a scientist interested in a given problem. He may be the only man interested in it anywhere in his geographical neighborhood. There may, of course, be other people also interested in the same problem who live or work near him, but in any event, the vast majority of all the other people interested in this general field will live and work far away. Many of them will be foreigners. Nevertheless, he will depend on information about their previous work in his research, and his work will ultimately be judged and made use of by them. It is this far-flung community which is important to him as a scientist, not the local community in which he works and which has such a dominating influence on his non-scientific activities. The presence of this community distinguishes the scientific from the non-scientific world.

The rest of this book is devoted to a discussion of this community, the organization which controls inquiry, and although I cannot briefly define it, it is possible to point out certain of its characteristics here. In the first place, it is a system of voluntary co-operation. Every individual simply seeks his own ends, but the organization is such that this leads him also to serve the ends of others.26 As we shall see, the organization is not as precise and elegant as that of the market, but the problems dealt with are even more diffuse than the economic problem of properly distributing scarce resources among innumerable desirable ends.27 Under the circumstances, it is highly unlikely that a greatly improved system can be developed. It is my hope that scientists with a greater understanding of the community can use it better, and I have appended at the end of this book a few recommendations for minor improvements in the organization of the community, but the present system is clearly an efficient one, and I see no real prospect for major improvements.

It is the existence of the scientific community which distinguishes between science and non-science, and this fact is implicitly relied upon by scientists in judging whether some field of study is or is not “scientific.” Thus a physicist is able to realize that Karlgren’s reconstruction of the pronunciation of archaic Chinese is “scientific,” not because he is able to judge either the correctness of the conclusions or the evidence or the methods used by Karlgren, but because he can see in the work and in its discussion by various other linguists the existence of a scientific community. We here can explain the apparent riddle that archaeology is a science while history is not. The social studies, of which history is one, have their own communities, but they operate rather differently from those in the sciences. The problem is discussed in detail in Chapter VI, but the difference in the attitude and approach is clear to even the casual observer (and the physical scientists tend to be very casual observers of the social sciences).

I may, perhaps, be permitted to introduce a personal experience to indicate the extent to which this suspicion is carried. Once in conversation with a minor member of the physics fraternity, I happened to mention the law of diminishing returns. He replied, “I don’t believe in general laws.” When I expressed surprise at such a statement from a physicist, he modified it: “I mean laws like that.” The “that” obviously referred to laws in economics, and his expression of distrust had nothing to do with this particular law, but with a general feeling that economists are not to be trusted. It proved quite impossible to shake him at all in argument; in fact, it was clear that he simply was not listening. He knew that my position was “unscientific.” He also knew that there were many plausible arguments for various false positions in his field which an amateur could not answer; so he simply assumed I was wrong because I was not a scientist. The situation is particularly ironic because the law of diminishing returns is really a physical law, and the evidence for it is very much stronger than that for most of the laws of physics.

As an introduction to my discussion of the scientists and their functioning in the scientific community, I must say a few words about scientific method in the strict sense. This is really a branch of logic or philosophy, not of the social sciences, but a little background is necessary for the rest of the book.28 The advance of scientific knowledge involves the accumulation of more data and the development of new theories. The relationship between the facts and the theories is a complex one. Normally, it is easy to demonstrate that a theory is based on facts which were known to its originator and that most of the information that we have was developed as the result of a hypothesis held by the original discoverer. Thus we gain little by asking which came first, the hypothesis or the data. In all relevant cases, a chain of data-hypothesis-data-hypothesis stretches back to our earliest records.

In this chain I have chosen to start with data collection, proceed to formulation of hypotheses, and then turn to data collections; but this is merely a convenient order for exposition. Logically, I could have started with the hypothesis just as readily. My present arrangement puts the hypothesis in the center of the process, which gives it the importance which I think it deserves. It also has the advantage of giving me a separate chapter in which to make one more assault on the problem of induction. Because solution of this problem has escaped investigators since Hume first propounded it, the odds are against success in my attempt, but I can at least try.

In the view of scientific method which I learned from Popper, the method by which we reach our hypothesis is less important than the question of whether the hypothesis is true, and this latter question can be answered only by testing it. Thus the logical problems of scientific method revolve around testing hypotheses, not around how we get them originally. This scheme is, I think, a realistic description of the actual behavior of scientists. It neatly solves as well a host of logical problems raised by “inductive” reasoning. Efforts to prove that we can reach conclusions about general laws by induction from specific instances have always failed. Popper points out that the truth of the general laws depends not on how they were derived but on how they pass tests once they have been invented. The crucial problem of science is not: Was this proposed law derived according to proper procedures, but: Is it true? This question can be most readily answered by testing it.

[1. ]American College Dictionary (New York: Random House, 1955).

[2. ]For a general discussion of the importance of such philosophical problems to science and the importance of science to philosophy, see K. R. Popper, “The Nature of Philosophical Problems and Their Roots in Science,” British Journal for the Philosophy of Science, 3, No. 10 (1952), 124–56.

[3. ]The philosophical issues are very well presented by J. Agassi in “Duhem versus Galileo,” British Journal for the Philosophy of Science, 8, No. 31 (1957), 237–48. The version of operationalism upheld by P. W. Bridgman comes very close to Mach’s position, “the proper definition of a concept is not in terms of its properties but in terms of actual operations. . . . concepts can be defined only in the range of actual experiment, and are undefined and meaningless in regions as yet untouched by experiment.” The Logic of Modern Physics (New York: Macmillan, 1951), pp. 6–7.

[4. ]Stephen Toulmin, The Philosophy of Science, An Introduction (London: Hutchinson, 1953), pp. 105–6. The whole of Toulmin’s Chapter IV (pp. 105–39) is a concise and lucid critical discussion of the general view of reality we are concerned with here. The four or five pages devoted to Mach give his position far more clearly than Mach ever did.

[5. ]Giorgio de Santillana, The Crime of Galileo (Chicago: University of Chicago Press, 1955), p. 99. The Catholics and the scientists have, in a way, exchanged positions on this issue. Since 1893 Galileo’s position has been the doctrine of the Catholic church. As a result a modern Catholic scholar can say: “It is a curious and paradoxical circumstance . . . that as a piece of scriptural exegesis Galileo’s theological letters are much superior to Bellarmine’s, while as an essay on scientific method Bellarmine’s is far sounder and more modern in its views than Galileo’s.” J. Broderick, The Life and Work of Blessed Robert Francis, Cardinal Bellarmine, S. J. (1928), vol. 2, 360, quoted in de Santillana, p. 101.

[6. ]Joseph Berkson, “Smoking and Lung Cancer,” American Statistician, 17, No. 4 (October, 1963), 15–22, at p. 19. I have been unable to find the original statement.

[7. ]The fields, such as particle physics, where Einsteinian computations are regularly made, tend to have higher social prestige among physicists.

[8. ]“The new . . . does not rise on the back of the old; the two are incompatible and incommensurable; and transfer of allegiance means working in a new world.” “What Are Scientists Made Of?” Times Literary Supplement, October 25, 1963, p. 850.

[9. ]The discovery that the “noble gasses” could combine, first reported in the October 12, 1962, issue of Science (p. 136), was a particularly striking example of the insecurity of present knowledge. In the lead editorial, Philip H. Abelson pointed out that “For perhaps 15 years, at least a million scientists all over the world have been blind to a potential opportunity to make this important discovery. All that was required . . . was a few hours of effort and a germ of skepticism.”

[10. ]Clearly the simplest way to explain the experimental results reported by Dr. Rhine and his followers is to accept ESP as a fact. Most scientists, however, have insisted on much more complex explanations involving allegations of various complex experimental errors on the part of Dr. Rhine. For a survey of the viewpoint held by most scientists, see Waldemar Kaempffert’s “Science in Review” column in the New York Times for July 22, 1956, sec. E, p. 9.

[11. ]This is one of the major themes of Michael Polanyi’s Personal Knowledge.

[12. ]J. Agassi’s review article “A Hegelian View of Complementarity,” British Journal for the Philosophy of Science, 9, No. 33 (1958), 57–63 discusses a particular variant of the problem, the statement that “truth is relative,” and demonstrates the impossibility of producing a consistent explanation on this ground.

[13. ]In a personal letter dated July 25, 1956, on scientific philosophy, Waldemar Kaempffert, science editor of the New York Times, expressed the then fashionable skepticism in an extreme form. “We must not forget that the universe is a hypothesis. . . . There is no certainty; hence there can be no approach of reality.”

[14. ]New York: Basic Books, 1959.

[15. ]See W. Ross Ashby, Design for a Brain (2nd ed.; New York: Wiley, 1960), pp. 17–19, for an excellent example.

[16. ]“The general point then, is that even when it is possible to formulate a physical problem in exact mathematical language it is often practical for an engineer or scientist to neglect some terms in the mathematical formulation in order to expedite the solution. The mathematician would not make this simplification and would persist in attempting to solve the original problem even if he had to hand the problem on to successive generations of mathematicians.” Morris Kline, Mathematics and the Physical World (New York: Crowell, 1959), pp. 62–63. This attitude is characteristic of all pure scientists, not just mathematicians.

[17. ]In some cases, there may be only one theory available, a on our diagram, for example, and this may be interconnected with the higher-level theory X.

[18. ]They are sometimes also very badly biased. See Walter Hall Wheeler, “The Unitatheres and the Cope-Marsh War,” Science, 131 (April 22, 1960), 1171–76.

[19. ]It should be noted that the difference between a scientist and a non-scientist is largely the relative position of the desire to know, i.e., our priority schedule for investigation compared with other desires. The man who puts great weight on such things will likely become an investigator. The man who puts relatively little emphasis on obtaining further information will turn to other lines of endeavor.

[20. ]This reliance must, of course, be based on the assumption that the others upon whom he depends are in fact themselves doing research and not simply relying on still others.

[21. ]It is probably worth repeating that human motivation is usually complex and that people with a single motivation are probably uncommon. We can, however, discuss the effects of some single motivation and draw conclusions as to the effect it may have on the behavior of a man who also has other motivations.

[22. ]A common distinction holds that the presence of written records is the difference between history and archaeology.

[23. ]Evolution is sometimes considered as a “theory of biology.” That evolution is one of the grand theories can hardly be denied, but in the present state of knowledge, it is impossible to connect most biological information with this great theory. There is an almost complete absence of the chain of minor theories which should connect the grand theory with the individual observations. In the circumstances, the existence of this particular grand theory has little effect on the concrete research of individual biologists. See Anthony Standen, Science Is a Sacred Cow (New York: Dutton, 1950).

[24. ]The point is particularly well made by H. C. Loguet-Higgins, “Portrait of the Scientist as Artist,” Times Literary Supplement, October 25, 1963, p. 856.

[25. ]For an excellent example, see Roger J. Williams, Biochemical Individuality (New York: Wiley, 1959).

[26. ]In The Origins of Scientific Economics (New York: Doubleday, 1965), William Letwin shows that the precursors of modern economics were not dispassionate scholars, but very practical men who were mainly engaged in making propaganda for various special points of view. Frequently they were motivated by a prospect of the crassest sort of material gain. The discussion process, however, forced gradually rising standards of information and coherence upon them, and the eventual outcome was a genuine science.

[27. ]To be more precise, it is a special and especially difficult subproblem within that area.

[28. ]Karl Popper’s Logic of Scientific Inquiry provides the best general statement of the view I hold of scientific method in the strict sense. This is natural since I learned the subject under his guidance. It would be possible, however, to accept either Dr. Popper’s position or the position of this book without accepting the other, since they really refer to different aspects of science.