The purpose of this paper is to provide a methodological rather than, strictly speaking, a philosophical discussion of its subject, the logic of concept formation in psychology. But even a treatment of this kind cannot entirely avoid matters of a more general nature, some of them logical, some epistemological. By insisting on the limitations of this essay I merely wish to caution the reader in three respects. First, those more general matters, logical and epistemological, will be kept at a minimum. Second, no attempt will be made to state them with the degree of precision and all the qualifications which are in order in a paper that addresses itself exclusively to logical analysts. Third, I shall for the most part content myself with stating them, without defending them in the way and in the sense in which a technical philosopher who speaks to his colleagues must defend what he asserts.

Preliminary Remarks: Before entering into the subject matter of the paper, it may be useful to present American readers with a sketchy outline of present-day European philosophy of science, as it appeared to me at the International Congress of Philosophy of Science, held in Paris at the Sorbonne on October 15–22, 1949. The sections of which I can give a first hand impression and which are of particular interest for our subject, are those of physical science and of probability.

In a paper recently published in this quarterly I argued that modern quantum physics, as an integrated system of laws, supplements and completes in purely quantitative terms the fragmentary order of first-person experience, removing in a unique way ambiguities otherwise encountered at the level of common-sense things; and I contended that the choice of a different selective operator—purpose or value, let us say, rather than quantity—might entail an entirely different range and system of order. It is now my intention to follow up this contention and to indicate at least in outline form some of the implications and possibilities of such a program; for it is possible that this undertaking may bring into new perspective some of the crucial problems which have vexed philosophers at all times when they have dealt with normative propositions and moral judgments.

If, as may be generally agreed upon, the term science is to be taken to mean verified knowledge, then it has three attributes: (1) the logical and methodological; that is, how we arrive at verified knowledge; (2) the epistemic; that is, the bodies of verified knowledge that have been arrived at; and (3) the sociological; that is, the organization of men by means of which the bodies of knowledge have been arrived at and the method prosecuted.

Scholasticism, with its intense rational method and its fervent intellectual spirit, virtually disappeared from the general consideration of Natural Philosophy at the coming of the Reformation and the subsequent Newtonian Revolution. Only in this century has it seemed to merit any reconsideration.

May I suggest that there is an angle to the problem of purposeful and non-purposeful behavior which has been overlooked in the recent discussions in this journal (Philosophy of Science, Vol. 17, Nr 4, pp. 310–332), namely, the psychological side of the problem.

The task of the philosophy of science, according to Churchman, is twofold. It is, one, “to determine the ideal of science by scientific methods,” and, the other, “to describe the manner in which science can most efficiently approach its ideals” (190). The primary purpose of this book on experimental inference is to deal with these two subjects as they bear upon what is commonly called material or empirical science, rather than upon formal or analytic science, though it is recognized that no absolute separation can be made between these two different types of science, or between the performance of the above tasks in connection with each. The kind of science, then, whose objective and methods are to be elucidated is the kind in which in some broad sense of these terms decisions are necessarily made on the basis of empirical observation and testing. Some observation or series of observations is made, and it is the task of the scientist to make an inference based upon these observational data. What is his aim in doing this, and what are the most efficient ways in which he can proceed?