Scientific Method

Inductive Methods and Their Postulates.

All the methods of science start from observed facts, and usually end in some kind of generalization relating to a whole class of facts or events, or a large group of them. The number of instances actually ob served is never more than a very small percentage of the whole class or group to which the generalization is extended. The ques tion therefore arises as to the justification of such generalization from a limited number of observations. Even if we assume that there is law and order among natural phenomena this does not yet answer our question. It is probably impossible to offer a complete justification of scientific induction in this respect. As a matter of fact, certain assumptions are usually made, wittingly or unwittingly, both in science and in practical life. And all one can do is to state these assumptions or postulates explicitly, and hope that, to some extent at least, they are the result of a long process of adaptation of human thought to the nature of things. The first of these assumptions is that it is possible with reasonable care to select as samples from a group, or a class, that will fairly represent the whole group or class. This assumption is known as the Principle of Fair Samples. In order to obtain a fair sample, the selection must be made in a way that is calcu lated to avoid onesidedness or bias. In that sense such a sample is often described as a random sample; but the selection of such a sample usually calls for much knowledge and insight as to the best way of avoiding bias. The reliability of a sample can never be more than probable, and the degree of this probability varies with the size and variety of the sample—the actual number being less important than the variety, whenever there is reason to sus pect variation in the whole that is to be judged by the sample. Much depends on our experience of the phenomena in question. In some cases a single instance (observed, say, under conditions of the method of difference) may be deemed to be a fair sample, in other cases even a large sample may fail to inspire confidence. This brings us to the second assumption. In cases of strict generalizations or uniformities (as distinguished from partial statistical laws), inferred inductively by means of one of the more reliable inductive methods, what happens is really this : The conditions under which the observations are made are such as to show that in the particular cases observed certain antecedents are causally connected with certain consequents. The generalization that all antecedents of that kind are causally connected with consequents of that kind is made so spontaneously that most people hardly realize that they have made it. But, strictly speaking, the generalization is an additional inference over and above the inference concerning the connection in the particular cases observed. And here again one cannot strictly justify it, but only state explicitly the assumption on which it rests. The assumption is that what is found to be a sufficient reason in one instance must be a reason in all instances of that kind. This assumption I have called the Principle of the Uniformity of Reasons.

Scientific Method and Scientific Explanation.—All the methods of science are essentially methods of discovering order in natural phenomena. For it is the ultimate aim of science to discover order in nature. To discover order in any class or group of phenomena is to explain them; that is, to make them clearer, more intelligible. Conversely, to explain anything is to indicate its place in some orderly system. In a broad sense it may be said that explanation generally takes the form of tracing the one in the many, or identity amid differences (see the article EXPLANA TION). Still, there are different types of explanation, and their differences are worth noting. As might have been expected, there is a general correspondence between the several kinds of scien tific method and the several types of explanation, as will appear from what follows. The main types of explanation may be sum marily indicated under a few heads. (I) Reference to a Class. Sometimes an object is explained when it is allocated to its class, especially if the class is already known. If, e.g., one is in doubt

about the character of a plant, it is explained by finding out to what species or variety, etc., it belongs, or if one is puzzled by a certain flash of light it may be explained as lightning, etc. (2) Reference to an Evolutionary Series. Sometimes an object (or class of objects) is explained by assigning to it a place in a possible evolutionary series. Thus, e.g., the rival explanations of the South African skull of an alleged ape-man assigned to it different places in the chain of biological evolution. (3) Reference to Mediating Conditions. Sometimes the problem is, how certain apparently remote or different facts or events come to be con nected nevertheless. In such cases the explanation takes the form of indicating certain intermediate facts or events which bridge the gulf. Thus, e.g., the perception of sound, light, etc., is explained by the mediation of aii -waves or ether-waves, etc., between the source of stimulation and the percipient or receiver. (4) Refer ence to Laws. The commonest type of explanation consists in referring events to certain relevant laws. Thus, e.g., the bent ap pearance of a stick partly immersed in water is explained by ref er ence to Snell's law of refraction. The position of a planet may be explained by reference to Kepler's laws. And Kepler's laws themselves may be explained by reference to Newton's law of gravitation. (5) Reference to Purpose. In the study of human conduct, and of certain other biological phenomena, it seems impossible to dispense with all reference to purpose, even when other kinds of explanation are made use of at the same time. The most familiar way of explaining human actions is by referring them to some purpose that is pursued either consciously or un consciously. Such explanations are suggested by our own felt experiences on similar occasions, and have always been found so satisfying to many people that they have been applied also to most other phenomena at one time or another. Hence the animism, fetishism, and anthropomorphism characteristic of the early his tory of human thought. This kind of extravagance was a serious obstacle to the progress of science during many centuries—to say nothing about the evil influences of theological anthropo morphism. Hence the reaction in modern science, which has not unnaturally displayed a strong hostility to all such teleological explanations, that is, explanations by reference to purposes. But the just revolt against one extreme does not justify another extreme. Such teleological explanations, too, have their proper place in certain limited fields of research. And it seems difficult to conceive how even the most violent opponent of teleological explanation could regard his own researches, writings and other higher activities as aimless and purposeless.

What has just been said may help to account for the tendency of modern science to disclaim all attempts to explain things, and to confine itself to the more modest task of describing them. The contrast is usually expressed by the statement that science only tries to answer how things happen, not why. But to say how things happen is also to explain them. Indeed, what is called scientific description often includes much that cannot be considered to be anything but explanation, right or wrong. The opposition to the question: Why? and the identification of all explanation with answers to such questions, betrays what is really at the back of the mind of the opponents of explanation as a legitimate aim of science. They really object to teleological explanation being applied indiscriminately to all kinds of natural phenomena. This is quite right. But it is a very short-sighted policy to give this extremely narrow meaning to the term "ex planation," and so to deprive science of its legitimate and honour able claim to explain things, even if it does not explain every thing, nor any one thing completely. The unsophisticated man of science who is not addicted to shibboleths certainly thinks not only that science explains things, but explains them most cor rectly. (See articles INDUCTION, LOGIC, EXPLANATION, PROBA BILITY, LOGIC, HISTORY OF.)