INTRODUCTION

§1. The problem of truth is one of the great classical problems of philosophy; and philosophy in its long history has given many different answers to the question of what truth is. The following considerations do not, however, concern the history of this question; rather, they examine the problem of truth from a systematic point of view. To begin with, the question at issue can be formulated in the following manner: What are the conditions for calling a proposition true; that is to say: How can one decide whether a given proposition is true or false? The examination of this problem will also permit us to respond to questions such as that of knowing whether there exist propositions which are absolutely and definitively true or whether each truth possesses a relative and provisional character – and to other questions of this kind, which are ardently discussed in philosophy. Furthermore, the envisaged inquiry will lead us to consequences which perhaps can contribute to clarifying certain points of the discussion opened in this periodical by Neurath and Petzäll – and which concerns the theory of science of the Vienna Circle.

The method by which we will attempt to establish the criteria for truth will not be that of synthetic and speculative philosophy; it will rather consist in a logical and methodological analysis of the procedures which are employed, in science, to verify a proposition; for it is only by such an analysis that one can hope to determine the character of truth in the only sense that matters for our scientific and everyday knowledge.

INTRODUCTION

This symposium is dedicated to the ideas of Moritz Schlick and Otto Neurath. I am grateful for being able to count them among those thinkers who had an essential influence on my own philosophical development.

In the fall of 1929,I participated with great interest in Schlick's seminar and in his lecture on the philosophy of nature, and I heard the debates between him and Neurath in the lively meetings of the Vienna Circle. I visited lectures given by Neurath in Vienna and later in Berlin and Chicago, and I had many stimulating, sometimes exciting, discussions with him. I think of these very different thinkers, different in temperament and philosophical style, with deep respect and gratitude.

In my contribution to this symposium, I will not evoke personal memories; rather, I will investigate the opposing views that Schlick and Neurath held on one of the major problems of epistemology.

The problem I have in mind is the question under which conditions – or given which reasons – should we accept empirical statements, especially the statements of the empirical sciences. I will call Schlick's and Neurath's views foundational and coherentist, for short.

Both Schlick and Neurath were empiricists. In the brochure “The Scientific Conception of the World: The Vienna Circle,” which was published in honor of Moritz Schlick by Carnap, Hahn, and Neurath in 1929,2 the basic conviction of the group is said to be “empiricist and positivist: there is knowledge only from experience, which rests on what is immediately given. This sets the limit for the content of legitimate science.”

INTRODUCTION

Science is often said to be a search for truth, a quest aiming at the formation of true beliefs about our world – beliefs concerning particular facts was well as general laws that may connect them.

But however appealing this conception may seem, it has, first of all, fundamental logical flaws, and second, it fails to do justice to a group of considerations that govern the critical appraisal and the acceptance or rejection of hypotheses and theories in science.

I propose briefly to elaborate this claim and to suggest an alternative way of characterizing science as a goal-directed endeavor.

The following considerations are informed both by the ideas of logical empiricism and by the more recent methodological explorations in a pragmatic-sociological vein undertaken by Thomas Kuhn and by kindred thinkers.

SCIENTIFIC THEORIZING: INVENTION AND CRITICAL APPRAISAL

Scientific theories are introduced in an effort to bring order into the diversity of the phenomena we encounter in our experience. How are such theories arrived at, and how are they supported?

Devising of effective theories is not a matter governed by systematic rules of discovery. As Einstein was fond of saying, scientific hypotheses and theories are arrived at by free invention, by the exercise of a creative scientific imagination. But their actual acceptance in science is subject to a critical evaluation by reference to the results of experimental or observational tests and to certain important additional criteria, which will be considered shortly.

Rudolf Carnap was the leading figure among the originators and the moving spirits of the stream of philosophical thought known as logical positivism or logical empiricism. Carnap preferred the latter name: The appellation “empiricism” rather than “positivism” was to set the movement apart from earlier forms of positivism, and the term “logical” was to call attention to the great importance this new empiricism attributed to the concepts and methods of contemporary logic as tools of philosophical analysis.

In Carnap's work, logic loomed large not only as a tool but also as a subject of philosophical investigation. A great deal of his research was devoted to problems in logic and metalogic, including the fields of logical syntax and semantics; the results of this work comprise substantial contributions to deductive logic as well as the most comprehensive and rigorous system of inductive logic yet devised.

The papers of my colleagues on this symposium address themselves to Carnap the logician; let me, therefore, attempt a brief appreciation of Carnap the logical empiricist philosopher.

Carnap's work outside the field of logic was devoted almost exclusively to epistemology and the philosophy of science, and his principal contributions to these fields are united by a common leitmotiv – namely, the search for ever more careful and philosophically illuminating reformulations, or explications, of the basic idea of empiricism that all our knowledge of the world ultimately derives from what is immediately given to us in the data of our direct experience.

INTRODUCTION

The following inquiry attempts to contribute to the logical analysis of the concept of probability. Philosophical efforts to clarify this concept are as old as its history, and the basic ideas of the two main groups of theories which seek to determine its meaning go back to the beginnings of this history. For a first orientation, these may be contrasted as the “aprioristic” and “empiricist” views of probability.

On the aprioristic views, probability theory rests on a priori judgments of proportions in empirical or logical space. On the empiricist views, probability theory concerns the relative frequencies with which certain results occur in statistical series of empirical observations.

At present, these two views are no longer on an equal footing. Instead, the progressive refinement of the logical analysis of scientific concept formations has made it increasingly clear that an adequate logico-descriptive theory of the probability concept can be found only in the direction in which the empiricist interpretation seeks it.

A consistent realization of the empiricist view, however, leads to certain logical difficulties, which have not been completely overcome even in the latest articulations of the empiricalstatistical probability theory. The present study examines these difficulties and develops a proposal for their removal on the basis of a “finitistic” turn of the statistical interpretation of the probability concept.

LOGICAL DIFFICULTIES OF THE STATISTICAL INTERPRETATION

The basic problem of the logical analysis of the probability concept – the problem in response to which the theory groups just mentioned differ from each other – is the question of the meaning of probability statements (i.e., of statements which attribute a certain probability to an event).

INTRODUCTORY REMARKS ON THE TWO EMPIRICIST GROUPS

The central ideas of logical, or scientific, empiricism as it developed during the twenties and early thirties in Vienna and in Berlin grew out of collaborative efforts of scientifically interested philosophers and philosophically interested scientists. Those thinkers noted that while the claims made by the physical sciences were amenable to objective test by experiment and observation, the pronouncements put forward by metaphysics were incapable of any such objective critical appraisal. And while hypotheses advanced in the physical sciences would eventually be accepted or rejected and thus lead to the growth of a body of objective scientific knowledge, the problems and pronouncements of metaphysics, inaccessible to objective appraisal, kept reappearing over and over again.

Moritz Schlick, the founding father of the Vienna Circle, addressed this situation in an article entitled “Die Wende der Philosophie,” which appeared in 1930 in the first issue of the empiricist periodical Erkenntnis. Rejecting “the anarchy of philo-sophical opinions,” he solemnly declared, “We stand in the middle of a definitive turn of philosophy and are justified in regarding the fruitless strife of philosophical opinions as finished.” The means for avoiding all such fruitless strife he saw in the clear and precise thinking exemplified by the exact sciences and informed by the principles of exact formal, or symbolic, logic.

Broadly speaking, the Vienna Circle held that the purported problems of metaphysics constitute no genuine problems at all and that in an inquiry making use of an appropriately precise conceptual and linguistic apparatus, metaphysical questions could not even be formulated.

1. 6 [5] “Über den Gehalt von Wahrscheinlichkeitsaussagen,” Erkenntnis 5 (1935), 228–60. Translation by Christoph Erlenkamp, revised by Olaf Helmer.

2. 7 [14] “On the Logical Form of Probability-Statements,” Erkenntnis 7 (1937), 154–60. [16] “Supplementary Remarks on the Form of Probability-Statements, Suggested by the Discussion. (Participants: Braithwaite, Clay, v. Dantzig, Ditchburn, Lindenbaum, Mannoury, Storer, Waismann.), Erkenntnis 7 (1937), 360–3.

3. 8 [24] With P. Oppenheim, “A Definition of ‘Degree of Confirmation,’” Philosophy of Science 12 (1945), 98–115.

Essay 6 is a translation of an article based on Hempel's Ph.D. dissertation. The essay argues for a view of probability as determined by empirical frequencies, while offering sympathetic but ultimately destructive criticism of the particular frequentist theories floated by Reichenbach and von Mises. The positive contribution of the article is a “finitization” of Reichenbach's approach.

The brief critical essay 7 also derives from the dissertation. It was read at the Fourth International Congress for the Unity of Science at Cambridge University in 1938. There, in a friendly spirit, Hempel identified an intractable difficulty that vitiates both von Mises's theory and Reichenbach's.

In the early 1940s, collaborating with Olaf Helmer and Paul Oppenheim, Hempel saw the project of logical empiricism as urgently needing a satisfactory account of how sentences can confirm or disconfirm other sentences. Unlike Carnap, the H2O trio sought an explication of degree of confirmation in partially empirical terms. The high-water mark in their hopes for the project (1945) is marked by essay 8 and by a parallel publication by Helmer and Oppenheim.

1. The Problem. The concept of confirmation of a hypothesis by empirical evidence is of fundamental importance in the methodology of empirical science. For, first of all, a sentence cannot even be considered as expressing an empirical hypothesis at all unless it is theoretically capable of confirmation or disconfirmation, that is, unless the kind of evidence can be characterized whose occurrence would confirm, or disconfirm, the sentence in question. And second, the acceptance or rejection of a sentence which does represent an empirical hypothesis is determined, in scientific procedure, by the degree to which it is confirmed by relevant evidence.

The preceding remarks, however, are meant only as accounts of methodological tendencies and are not intended to imply the existence of clear-cut criteria by means of which the scientist can decide whether – or, in quantitative terms, to what degree – a given hypothesis is confirmed by certain data. For indeed, no general and objective criteria of this kind are at present available; in other words, no general definition of the concept of confirmation has been developed so far. This is a remarkable fact in view of the importance of the concept concerned, and the question naturally suggests itself whether it is at all possible to set up adequate general criteria of confirmation, or whether it may not rather be necessary to leave the decision in matters of confirmation to the intuitive appraisal of the scientist.

One of the most important and most discussed problems of contemporary philosophy is that of determining how psychology should be characterized in the theory of science. This problem, overflowing the limits of epistemological analysis and leading to heated controversy in metaphysics itself, is brought to a focus by this familiar disjunction: “Is psychology a natural science, or is it one of the sciences of mind and culture (Geisteswissenchaften)?”

The present article attempts to sketch the general lines of a new analysis of psychology, one which makes use of rigorous logical tools and which has made possible decisive advances toward the solution of the above problem. This analysis was successfully undertaken by the Vienna Circle (Wiener Kreis), the members of which (M. Schlick, R. Carnap, Phillipp Frank, o. Neurath, F. Waismann, H. Feigl, etc.) have, during the past ten years, developed an extremely fruitful method for the epistemological examination and critique of the various sciences, based in part on the work of L. Wittgenstein. We shall limit ourselves essentially to the examination of psychology as carried out by Carnap and Neurath.

The method characteristic of the studies of the Vienna Circle can be briefly defined as a logical analysis of the language of science. This method became possible only with the development of an extremely subtle logical apparatus which makes use, in particular, of all the formal procedures of modern logistics.

1. 13 [70] Rudolf Carnap, “Logical Empiricist,” Synthese 25 (1973), 256–68.

2. 14 [89] “Der Wiener Kreis und die Metamorphosen seines Empirismus,” in Norbert Leser, ed., Das geistige Leben Wiens in der Zwischenkriegszeit (Wien: Bundesverlag, 1981), pp. 205–15. Translated by Christian Piller.

3. 15 [106] “Hans Reichenbach Remembered,” Erkenntnis 35 (1991), 5–10.

4. 16 [109] “Empiricism in the Vienna Circle and in the Berlin Society for Scientific Philosophy: Recollections and Reflections,” Scientific Philosophy: Origins and Developments, in Friedrich Stadler, ed., Kluwer (Dordrecht: Academic Publishers, 1993), pp. 1–9.

In the end, Carnap and Neurath had more influence than his Doktorvater, Reichrenbach, on Hempel's thought; and although the interaction with Carnap went on much longer, it was Neurath whose influence proved to be the more enduring. Hempel tells the story here in memorial essays (13, 15) on Carnap and Reichenbach – and in two historical essays (14, 16) in which Neurath figures prominently on the career of logical empiricism.

1. In the following remarks, an attempt will be made to analyze the logical form of the concept of probability and of the statements in which it occurs, and to apply the result of this analysis to some problems concerning the language of empirical science. Our considerations will be based on the modern statistical theory of probability, which is, at present, certainly by far the most satisfactory and consistent basis for a theoretical account of the concept of probability as it is actually used in empirical science as well as in everyday language.

2. According to the statistical theory, each particular probability statement refers to an infinite sequence k of events and to a property P; the probability statement then has the form: “The probability for an element of k to have the property P is p,” which, by the statistical interpretation, is equipollent to “The relative frequency of those among the first n elements of k which have the property P converges toward the limit p as n goes toward infinity.” Thus, according to the statistical theory, probability is a three-termed relation between a property (or class), a sequence, and a real number; and each particular probability statement may be given the form “Prob (P, k, p)” (“The probability of P in k is p”). Here, “p” is a real number constant with o ≦ p ≦ 1; “k” is a functor whose arguments are the natural numbers and whose values are tetrads of real numbers which may be considered as determining the spatio-temporal coordinates of events.

1. 9 [3] “Analyse logique de la psychologie,” Revue de Synthese 10 (1935), 27–42. Wilfrid Sellars's translation, “The Logical Analysis of Psychology,” appeared in Herbert Feigl and Wilfrid Sellars, eds., Readings in Philosophical Analysis (New York: Appleton-Century-Crofts, 1949), pp. 373–84.

2. 10 [92] “Schlick und Neurath: Fundierung vs. Köharenz in der wissenschaftlichen Erkenntnis,” Grazer philosophische Studien, vol. 16/17 (1983), 1–18; Christian Piller, trans., “Schlick and Neurath: Foundation vs. Coherence in Scientific Knowledge.”

3. 11 [102] “On the Cognitive Status and the Rationale of Scientific Methodology,” Poetics Today 9 (1988), 5–27. Hempel notes that Part I is an annotated and slightly revised version of [95] and that Part II contains passages from [97].

4. 12 [100] “Provisoes: A Problem Concerning the Inferential Function of Scientific Theories,” in Adolf Grunbaum and Wesley Salmon, eds., The Limitations of Deductivism (Berkeley: University of California Press, 1988), pp. 19–36. Also in Erkenntnis 28 (1988), 147–64.

Essay 9 is a period piece from the early heyday of physicalism, when it could seem plausible both that any meaningful statement must be translatable into a sentence in the language of physics and that the possibility of such translation is a purely logical matter. In it, Hempel's proposed dissolution of the mind-body problem was a version of the identity theory according to which it is a matter of logic that “Otto has a toothache” will be equipol-lent to some physical sentence. Which physical sentence? The answer is an empirical matter.

This paper has been suggested by a recent discussion between Prof. Schlick and Dr. Neurath, made public in two articles which appeared in volume 4 of Erkenntnis, which mainly concerns the positivistic concept of verification and truth.

For the following exposition it may be advantageous to refer to the well-known crude classification which divides the different theories of truth into two main groups – that is, the correspondence theories and the coherence theories of truth. According to the correspondence-theories truth consists of a certain agreement or correspondence between a statement and the so-called facts or reality; while according to the coherence theories truth is a possible property of a whole system of statements (i.e., a certain conformity of statements with each other). In extreme coherence theories truth is even identified with the mutual compatibility of the elements of such a system.

The logical positivists' theory of truth developed step-by-step from a correspondence theory into a restrained coherence theory.

Let us shortly consider the most important logical phases of this process (which do not exactly correspond to the historical ones).

The philosophical ideas, which L. Wittgenstein has developed in his Tractatus Logico-Philosophicus and which represent the logical and historical starting point of the Vienna Circle's researches, are obviously characterized by a correspondencetheory of truth.

According to one of Wittgenstein's fundamental theses, a statement is to be called true if the fact or state of affairs expressed by it exists; otherwise the statement is to be called false.

The working title of this collection was “Philosophical Essays, Early and Late” – that is, earlier and later than the essays of Hempel's middle period, which are collected in his Aspects of Scientific Explanation and Other Essays in the Philosophy of Science [54]. They were published in two decades (roughly, 1945–1965), during which logical empiricism evolved from an arcane neo-Kantian or counter-Kantian sect into something that would seriously be called “The Received View” – a view whose shape and reception owed much to those very essays. They were essays in the sort of logical constructivism that Russell had framed in Our Knowledge of the External World as a Field for Scientific Method in Philosophy (1915). Russell's book was the inspiration for Carnap's Der logische Aufbau der Welt (1928), and for the Logische Syntax der Sprache (1934), which incorporated Carnap's version of “physicalism.” In those two decades Hempel could still regard the failings and limitations of that work as mere bumps on the road to logical empiricism. This book of Hempel's essays charts his course into and out of the methods and concerns of that middle period.

Having been one of the first logical empiricist graduate students, Hempel would become the last survivor of the Berlin and Vienna circles. Throughout, he was in close intellectual contact with the major figures in the movement – and a movement it was, defined for its activists by no doctrine but by a certain morale and discipline.

METHODOLOGY OF SCIENCE: DESCRIPTIVE AND PRESCRIPTIVE FACETS

Two Conceptions of the Methodology of Science

In the course of its long history – most strikingly in recent centuries – scientific inquiry has vastly broadened man's knowledge and deepened his understanding of the world he lives in, and the remarkable successes of predictions and technological applications based on those insights are widely acknowledged as eloquent testimony to the soundness or the “rationality” of scientific methods of research. Yet, there is no unanimity among students of the methodology of science as to whether or to what extent it is possible to formulate clear and precise rules of procedure which are characteristic of scientific research and which make science the exemplar of rationality in the pursuit of reliable knowledge. Nor is there unanimity concerning the grounds on which such rules of scientific inquiry could be established and how, accordingly, their cognitive status is to be understood: Is the methodology of science to be viewed as a descriptive study of actual scientific research or as a quasi-normative discipline aimed at formulating – in an analytic, aprioristic manner – standards of proper or rational scientific inquiry?

These questions have in recent years been the focus of an intense and fruitful debate between two schools of thought, which I will briefly refer to as the analytic-empiricist and the historic-sociological or pragmatist schools. By the former, I mean a body of ideas that, broadly speaking, has grown out of logical empiricism and the work of kindred thinkers.