THE VIENNA CIRCLE AND LOGICAL EMPIRICISM VIENNA CIRCLE INSTITUTE YEARBOOK [2002]

Transcription

1 THE VIENNA CIRCLE AND LOGICAL EMPIRICISM VIENNA CIRCLE INSTITUTE YEARBOOK [2002] 10

2 VIENNA CIRCLE INSTITUTE YEARBOOK [2002] 10 Institut Wiener Kreis Society for the Advancement of the Scientific World Conception Series-Editor: Friedrich Stadler University of Vienna, Austria and Director, Institut Wiener Kreis Advisory Editorial Board: Rudolf Haller, University of Graz, Austria, Coordinator Nancy Cartwright, London School of Economics, UK Robert S. Cohen, Boston University, USA Wilhelm K. Essler, University of Frankfurt/M., Germany Kurt Rudolf Fischer, University of Vienna, Austria Michael Friedman, University of Indiana, Bloomington, USA Peter Galison, Harvard University, USA Adolf Grünbaum, University of Pittsburgh, USA Rainer Hegselmann, University of Bayreuth, Germany Michael Heidelberger, University of Tübingen, Germany Jaakko Hintikka, Boston University, USA Gerald Holton, Harvard University, USA Don Howard, University of Notre Dame, USA Allan S. Janik, University of Innsbruck, Austria Richard Jeffrey, Princeton University, USA Andreas Kamlah, University of Osnabrück, Germany Eckehart Köhler, University of Vienna, Austria Anne J. Kox, University of Amsterdam, The Netherlands Saul A. Kripke, Princeton University, USA Elisabeth Leinfellner, University of Vienna, Austria Werner Leinfellner, Technical University of Vienna, Austria James G. Lennox, University of Pittsburgh, USA Brian McGuinness, University of Siena, Italy Kevin Mulligan, Université de Genève, Switzerland Elisabeth Nemeth, University of Vienna, Austria Julian Nida-Rümelin, University of Göttingen, Germany Helga Nowotny, ETH Zürich, Switzerland Erhard Oeser, University of Vienna, Austria Joëlle Proust, École Polytechnique CREA Paris, France Alan Richardson, University of British Columbia, CDN Peter Schuster, University of Vienna, Austria Jan Šebestik, CNRS Paris, France Karl Sigmund, University of Vienna, Austria Hans Sluga, University of California at Berkeley, USA Elliott Sober, University of Wisconsin, USA Antonia Soulez, Université de Paris 8, France Wolfgang Spohn, University of Konstanz, Germany Christian Thiel, University of Erlangen, Germany Walter Thirring, University of Vienna, Austria Thomas E. Uebel, University of Manchester, UK Georg Winckler, University of Vienna, Austria Ruth Wodak, University of Vienna, Austria Jagiellonian University, Cracow, Poland Anton Zeilinger, University of Vienna, Austria Honorary Consulting Editors: Kurt E. Baier Francesco Barone C.G. Hempel Stephan Körner Henk Mulder Arne Naess Paul Neurath Willard Van Orman Quine Marx W. Wartofsky Review Editor: Michael Stöltzner, University of Bielefeld, Germany Editorial Work/Layout/Production: Hartwig Jobst Robert Kaller Camilla R. Nielsen Erich Papp Christopher Roth Editorial Address: Institut Wiener Kreis Museumstrasse 5/2/19, A 1070 Wien, Austria Tel.: +431/ (international) or 01/ (national) Fax.: +431/ (international) or 01/ (national) ivc.zuef@univie.ac.at homepage: The titles published in this series are listed at the end of this volume.

3 THE VIENNA CIRCLE AND LOGICAL EMPIRICISM RE-EVALUATION AND FUTURE PERSPECTIVES Edited by FRIEDRICH STADLER University of Vienna, and Institute Vienna Circle, Austria KLUWER ACADEMIC PUBLISHERS NEW YORK, BOSTON, DORDRECHT, LONDON, MOSCOW

4 ebook ISBN: Print ISBN: Kluwer Academic Publishers New York, Boston, Dordrecht, London, Moscow Print 2003 Kluwer Academic Publishers Dordrecht All rights reserved No part of this ebook may be reproduced or transmitted in any form or by any means, electronic, mechanical, recording, or otherwise, without written consent from the Publisher Created in the United States of America Visit Kluwer Online at: and Kluwer's ebookstore at:

5 EDITORIAL On the occasion of its anniversary, the Institut Wiener Kreis/Vienna Circle Institute, together with the Zentrum für überfakultäre Forschung/Center for Interdisciplinary Research of the University of Vienna, organized an international symposium on The Vienna Circle and Logical Empiricism. Re-Evaluation and Future Perspectives of Research and Historiography. This event was hosted by the Department of Contemporary History at the Campus of the University of Vienna, in Vienna July 12 14, The Institute Vienna Circle (IVC) was founded in 1991 as a non-profit society. It has been supported ever since by the Austrian Ministry of Science and Research and the City of Vienna. The institute is a member of the International Union of the History and Philosophy of Science Division of Logic, Methodology and Philosophy of Science, and has been working together with the University of Vienna since 1997 on the basis of a co-operation agreement. Since the very outset, the IVC has worked together closely with similar institutes and societies in Austria and abroad, focusing on the promotion, cultivation and dissemination of a scientific philosophy and history and philosophy of science in the tradition and spirit of the Vienna Circle/Logical Empiricism. The Institute s research activities also include the documentation, application and development of its results. IVC adheres to a pluralist and (post-)enlightened conception of science and philosophy of science that is committed to the democratization of knowledge and science and the critique of all forms of irrationalism, dogmatism, and fundamentalism. To this end, the IVC regularly organizes conferences and lectures in Austria and abroad, edits three book series (in English and German), and maintains a library and archives with materials by, and on, members of the Vienna Circle and associated philosophers as well as scientists. (cf. survey at the IVC s Website: As regards the most recent activities, one should draw attention to the Vienna International Summer University Scientific World Conceptions, which has been taking place every year (in July) since 2001 at the University Campus. Each summer university has had a different focus on research-related topics: 2001 (Unity and Plurality of Science), 2002 (Mind and Computation), and 2003 (Cosmological and Biological Evolution). Another current activity is our participation in the ongoing ESF-Network on Historical and Contemporary Perspectives of Philosophy of Science in Europe, from 2001 to 2003, which is planned to be extended as a follow-up program of ESF. The edition of Moritz Schlick s papers (Moritz Schlick Edition) is another international research and publication project, which is presently underway with a team of scholars from the University of Graz and the University of Rostock (Germany). The proceedings of the Third

6 VI EDITORIAL International History of Philosophy of Science Conference (HOPOS 2000) are to be found here in the Vienna Circle Institute Yearbook 9/2001. This Yearbook presents the contributions of invited lecturers as well as a selection of contributed papers of the aforementioned anniversary jubilee conference. It also features work from international research and historiography on Logical Empiricism and its influence, in addition to its further development by renowned scholars and a younger generation of philosophers. We have divided this yearbook papers into thematic chapters that focus on the origins, history and historiography, with such leading figures as Schlick and Reichenbach along with other members of this influential movement. The yearbook also addresses more topical issues such as the unity and plurality of science, contexts of science, epistemology and ethics, and some (long neglected) women of Logical Empiricism. The reception of the Vienna Circle/Logical Empiricism in the Soviet Union and Russia is dealt with in a special concluding report section. As usual, the volume also has a review section on recent publications dealing with scientific philosophy and philosophy of science. In this regard it is important to note that the selected papers on Rudolf Carnap have been published separately in the volume of the IVC series Vienna Circle Institute Library established this year. This volume, Language, Truth, and Knowledge, edited by Thomas Bonk, is to appear at the same time as this Yearbook published by Kluwer. Several other invited papers will be part of the forthcoming Cambridge Companion to Logical Empiricism, ed. by Alan Richardson and Thomas Uebel. Last not least let me thank to all who helped make the anniversary conference possible and contributed to the publication of the proceedings in these two volumes: my colleagues Elisabeth Nemeth and Eckehart Koehler as members of the Program Committee, the members of the Local Organizing Committee with Margit (Mischa) Kurka, Daria Mascha, Robert Kaller from the IVC and Marianne Ertl from the Department of Contemporary History. Here I would like to express my sincere gratitude to our financial supporters: the Austrian Ministry of Education, Science and Culture, the City of Vienna (Division of Culture) and the Bank Austria. Finally, I would like to thank our review editor Michael Stöltzner, and Camilla Nielsen, Christopher Roth and Hartwig Jobst, who, together with members of the Advisory Editorial Board, were involved in the production of this Yearbook. Vienna, October 2002 Friedrich Stadler (University of Vienna, and Vienna Circle Institute)

7 TABLE OF CONTENTS A. THE VIENNA CIRCLE AND LOGICAL EMPIRICISM FRIEDRICH STADLER: What is the Vienna Circle? Some Methodological and Historiographical Answers XI I. ORIGINS AND HISTORY ARNE NAESS: Pluralism of Tenable World Views PAOLO PARRINI: On the Formation of Logical Empiricism ANITA VON DUHN: Bolzano s Account of Justification DAVID JALAL HYDER: Kantian Metaphysics and Hertzian Mechanics II. MORITZ SCHLICK HUBERT SCHLEICHERT: Moritz Schlick s Idea of Non-territorial States MASSIMO FERRARI: An Unknown Side of Moritz Schlick s Intellectual Biography: the Reviews for the Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie ( ) HANS JÜRGEN WENDEL: Between Meaning and Demarcation DAGMAR BORCHERS: Let s Talk about Flourishing! Moritz Schlick and the Non-cognitive Foundation of Virtue Ethics III. HANS REICHENBACH CARSTEN KLEIN: Coordination and Convention in Hans Reichenbach s Philosophy of Space 109

8 VIII of Simultaneity in Historical and Philosophical Perspective 121 ROBERT RYNASIEWICZ: Reichenbach s IV. OTHER PROPONENTS AND PERIPHERY JUHA MANNINEN: Towards a Physicalistic Attitude WOLFGANG HUEMER: Logical Empiricism and Phenomenology: Felix Kaufmann ARTUR KOTERSKI: Béla von Juhos and the Concept of Konstatierungen PAOLO MANCOSU / MATHIEU MARION: Wittgenstein s Constructivization of Euler s Proof of the Infinity of Primes GRACIELA DEPIERRIS: Quine s Historical Argument for Epistemology Naturalized V. UNITY AND PLURALITY ELLIOTT SOBER: Two Uses of Unification CHRISTOPHER HITCHCOCK: Unity and Plurality in the Concept of Causation DIEDERICK RAVEN: Edgar Zilsel s Research Programme: Unity of Science as an Empirical Problem VI. CONTEXTS OF SCIENCE GREGOR SCHIEMANN: Criticizing a Difference of Contexts On Reichenbach s Distinction between Context of Discovery and Context of Justification GIORA HON: Contextualizing an Epistemological Issue: the Case of Error in Experiment JUTTA SCHICKORE: The Contexts of Scientific Justification. Some Reflections on the Relation Between Epistemological Contextualism and Philosophy of Science 265

9 IX VII. EPISTEMOLOGY DANIEL COHNITZ: Modal Skepticism. Philosophical Thought Experiments and Modal Epistemology 281 F.O. ENGLER: Structure and Heuristic: in Praise of Structural Realism in the Case of Niels Bohr 297 VIII. ETHICS UWE CZANIERA: The Neutrality of Meta-Ethics Revisited How to Draw on Einstein and the Vienna Circle in Developing an Adequate Account of Morals 313 IX. WOMEN OF LOGICAL EMPIRICISM DAGMAR BORCHERS: No Woman, no Try? Else Frenkel-Brunswik and the Project of Integrating Psychoanalysis into the Unity of Science MICHAEL BEANEY: Susan Stebbing on Cambridge and Vienna Analysis NIKOLAY MILKOV: Susan Stebbing s Criticism of Wittgenstein s Tractatus ADELHEID HAMACHER-HERMES: Rose Rand: a Woman in Logic B. GENERAL PART REPORT DOCUMENTATION OLESSIA NAZAROVA: Logical Positivism in Russia 381

10 X REVIEWS Ernst Mach s Vienna or Phenomenalism as Philosophy of Science. Edited by John Blackmore, Ryoichi Itagaki and Setsuko Tanaka. Kluwer Academic Publishers: Dordrecht (Erik Banks) 389 Herbert Hochberg, The Positivist and the Ontologist. Bergmann, Carnap and Logical Realism, Rodopi: Amsterdam/Atlanta (Erwin Tegtmeier) Liliana Albertazzi / Dale Jacquette / Roberto Poli (eds.), The School of Alexius Meinong (= Western philosophy series 57), Aldershot et al.: Ashgate, (Maria Reicher) M. Ferrari / I.-O. Stamatescu (eds.), Symbol and Physical Knowledge. On the Conceptual Structure of Physics, Springer: Berlin (Thomas Mormann) Uwe Czaniera, Gibt es moralisches Wissen? Die Kognitivismusdebatte in der analytischen Moralphilosophie, Mentis: Paderborn (Gabriele Mras) ACTIVITIES OF THE INSTITUTE VIENNA CIRCLE Activities 2002 Preview OBITUARY Eugene T. Gadol ( ) (Friedrich St adler) 417 Index of Names 419

11 FRIEDRICH STADLER WHAT IS THE VIENNA CIRCLE? SOME METHODOLOGICAL AND HISTORIOGRAPHICAL ANSWERS What is the Vienna Circle? is a question which is neither rhetorical nor trivial. It is perhaps an attempt to square the circle which is, meanwhile, mathematically possible, as Karl Menger described as early as This question might be also a problem of how the whole relates logically to the parts or the parts to the whole, which was already addressed by mereology (whole-part theory) according to Stanislaw (1916). 2 Of course, we are all familiar with the irritating fact that one and the same phenomenon can be described consistently by more than one theory (underdetermination of a theory by observation). A popular way for beginners to proceed is to check the lexical entries on a concept that might merit further discussion. But is this enough? If we look at the current definitions of the Vienna Circle, we quickly recognize the difficulty of providing a representative description of the circle and of Logical Empiricism in its entirety. Even the autobiographical accounts given since the classical period of the Schlick Circle show a remarkable variance depending on focus and underlying motivations. The locus classicus is and remains the manifesto of 1929 entitled Scientific Conception of the World. The Vienna Circle, published by the Ernst Mach Society and co-authored by Rudolf Carnap, Hans Hahn and Otto Neurath. 3 A second milestone in the diffusion of Vienna Circle ideas, especially in the Anglo-Saxon world, was the article written by Herbert Feigl and Albert Blumberg: Logical Positivism A New Movement in European Philosophy (1931) in the Journal of Philosophy. Here, both authors underscored the (anti-kantian) synthesis of rationalism and empiricism. 4 What is so striking about these representative text excerpts? The scientific world conception (as opposed to Weltanschauung) is directed against metaphysics as well as philosophy as a discipline in its own right. As an alternative, we can note a tendency towards a (physicalist) unified science that later culminated in an empiricist encyclopedia project and includes the principle of tolerance as applied to logic and scientific languages. Already here, the contours of epistemological options emerge: phenomenalism vs. physicalism, coherence theory vs. correspondence theory of truth, syntax vs. semantics as regards the logic of science, ideal vs. normal language as scientific language and, last not least, verification vs. confirmation (or falsification) of empirical statements.

12 XII FRIEDRICH STADLER If one also takes into account that the manifesto represents only one ( left ) wing of the Vienna Circle at the end of the twenties (one simply need recall the negative stance of Schlick under Wittgenstein s influence), then it becomes abundantly clear that there existed only a limited consensus. Is it possible, even so, to find a sort of basic agreement here one that unites the members of the Vienna Circle both the central figures and those on the periphery? 5 A simple theoretical framework can be found in the following description: a basic scientific orientation grounded in logical and linguistic analysis, an explanatory and epistemological monism in terms of methodology and research subjects, and finally a sort of fallibilistic epistemology with interdisciplinarity featuring as a program that opposes any sort of foundationalist system. First, let me point out that until the end of the Vienna Circle s existence the tensions between scientific philosophy on the one hand, and empirical science (dualism of philosophy and science) on the other, went unresolved. These differences were most pronounced in the discussions between Moritz Schlick and Otto Neurath. 6 We can conclude that neither the autobiographical accounts of contemporaries nor the historical accounts originating shortly after 1945 were able to provide an adequate picture of the Vienna Circle. Moreover, we can only recognize a partial, albeit broad, overlap with Logical Empiricism in general when we take into account the related movements of the Berlin Circle around Hans Reichenbach or the Warsaw Group around Alfred Tarski. 7 Here I think that a historical reconstruction is indeed called for, even if it ultimately proves insufficient. Such a reconstruction was already briefly sketched by Gustav Bergmann in his letter to Neurath in 1938: 8 Seen in this way, the important scientific movements which until now had their common center of radiation in Vienna psychoanalysis, the philosophy of the Vienna Circle, and Kelsen s legal and political philosophy, really belonged together and they determined the specific intellectual atmosphere of the Austria that vanished, just as did, in the artistic sphere, the authors Broch, Canetti and Musil. The importance of the Vienna Circle as a cultural phenomenon within the context of New Objectivity (Neue Sachlichkeit) is illustrated by contemporary figures and their more or less positive reactions to its program, e.g. the writers Hilde Spiel, Jean Améry, Bertolt Brecht and Robert Musil. 9 But what does present-day research and history have to say about all this? To be sure, the (hidden) history of the reception of Logical Empiricism ranks worldwide as a real success story in the field of history and philosophy of science. Moreover, there has been a remarkable renaissance in research of this area over the past two decades. There seems to be a real interest in the study of the emergence, development and impact of this specifically Central European tradition. 10

13 WHAT IS THE VIENNA CIRCLE? XIII As even recent entries on Vienna Circle and Logical Empiricism present overlapping and incomplete descriptions 11, my first suggestion is to attempt a historical and genetic reconstruction within the socio-cultural context, together with a description of individuals and theoretical positions from both a diachronic and synchronic perspective. This means dealing with the multi-facetted phenomenon of the Vienna Circle from a historical point of view, which can only be described here in very general terms: 12 The new historiography on Logical Empiricism sets in with the rediscovery of Ernst Mach ( ) as a precursor of Gestalt theory, evolutionary epistemology, (possibly radical) constructivism and the modern historically oriented philosophy of science. But already in Mach s reception of the Vienna Circle one can see not only a certain pluralism of views but also a polarization of the various positions (Mach s influence on Carnap s Aufbau / Logical Construction, the critical distancing to psychologism in the manifesto, the alternative to the principle of economy in Karl Menger, etc.) 13 Nevertheless, this research program, which was interpreted differently by the Vienna Circle, actually represented a sort of prototype for Logical Empiricism in the interwar years irrespective of whether one backs the bold claim as to the existence of a typical Austrian philosophy (as opposed to German idealism). Accordingly, Richard von Mises described Mach s impact as follows: In the transformation from the formal to the material mode of language, Mach s elements correspond to the protocol sentences. 14 This, of course, can easily be read as being a building stone of the neutral monism as represented by Bertrand Russell. In this context, a number of different studies can be seen as focusing on the impact of neo-kantianism, Russell and Mach on Carnap s Aufbau philosophy. 15 It is thus not so surprising that, already prior to World War 1, the proto-circle of the later Vienna Circle (the first Vienna Circle ) began to take shape both organizationally and philosophically. Within a discussion circle (inter alia, with Ph. Frank, Hahn, Neurath, R. v. Mises) at a coffeehouse, traditional academic philosophy grew more scientific. The First Vienna Circle met regularly as of 1907 to discuss the synthesis of empiricism and symbolic logic as modeled after Mach, Boltzmann and the French conventionalists (Pierre Duhem and Henri Poincaré). This was also regarded as an indirect answer to W.I. Lenin s polemical remarks against Mach in his book (Materialism and Empirio-criticism. 1909) which remained very influential up to the Velvet Revolution of 1989/ This early phase in the development of Logical Empiricism can also be interpreted as an anti-cartesian turn in epistemology and philosophy of science, which undermined both the synthetic apriori and the secure foundations of knowledge. In the middle of the permanent crisis of philosophy between reform and revolution in society and science, the further development of this scientific philosophy had, in any case, been initiated.

14 XIV FRIEDRICH STADLER I. Focusing on the center of the Vienna Circle we recognize the following development : with the conflict-laden appointment of the physicist and philosopher Moritz Schlick ( ) to Mach s chair for natural philosophy of inductivist sciences in Vienna in 1922 the heyday of scientific philosophizing in the post World War 1- period was prolonged. Even though Schlick felt committed to an epistemological realism in his main oeuvre General Theory of Knowledge (1918/1925), he began his inaugural lecture with a programmatic allusion to Mach under the sway of the Viennese tradition up to Wittgenstein: Almost all philosophy is natural philosophy. 17 Several years thereafter, the intellectual foundations for the formation of the Vienna Circle had been laid: the work of Frege, Russell / Whitehead and Wittgenstein s Tractatus, together with Duhem s and Poincaré s holistic theory of science, paved the way for the emergence of a unique, innovative scientific culture that has remained a subject of international scholarship to this very day and that addresses issues that still have a bearing on modern philosophy of science. In the phase during which the so-called Schlick Circle on Boltzmanngasse became a veritable institution, there was already a pluralism of positions that emerged in the field of tension between Wittgenstein s Tractatus (1918/22) and Carnap s Logischer Aufbau der Welt / Logical Construction of the World (1928). But in spite of all discrepancies between Carnap s rational reconstruction and the philosophy of ideal language (Wittgenstein), all those involved came to identify with a philosophical reform movement as opposed to academic traditional philosophy (Schulphilosophie according to Philipp Frank). Even the prevailing epistemological options critical realism (or constructive realism) in Schlick, Feigl, Kraft and later Popper, on the one hand, and phenomenalism / physicalism in Carnap, Neurath, Kaufmann, on the other could not prevent a reconciliation of rationalism and empiricism. As early as the twenties, a diversity of positions and methods had emerged (a theoretical pluralism which has usually been ignored in the corresponding historiography). The mathematician Karl Menger, who had spent time in Holland, formulated his principle of tolerance regarding the selection of logics and scientific languages (a principle which was only later ascribed to Carnap). He founded an additional platform, namely his influential Mathematisches Kolloquium ( Mathematical Colloquium ), operating in parallel to the Schlick Circle on Boltzmanngasse. 18 Long before Quine, he criticized the principle of economy, verificationism and the analytic / synthetic dualism which had become wellknown in uncritical historical accounts as the standard model of Logical Empiricism an interpretation which was reinforced by Quine s influential Two Dogmas of Empiricism. 19 The contacts that were initiated with the Polish logicians and Alfred Tarski (fostering the semantic conception of truth) ultimately

15 WHAT IS THE VIENNA CIRCLE? XV resulted in a further provocation of the Wittgenstein group in the Circle (Schlick, Waismann). But undoubtedly the greatest challenge for the early Vienna Circle was the Tractatus which experienced an uneven and fragmentary reception (as had the manifesto in 1929). 20 In sum we can say that, for this formative and non-public phase of the Vienna Circle, certain important components justify the use of the prefix neo. A conventionalist understanding of scientific theory, the philosophy of ideal language, and the Hilbert program and theory of relativity (especially Einstein s 1921 published Geometrie und Erfahrung / Geometry and Experience), as well, became the building blocks for the turning point in philosophy proclaimed with great optimism in the first volume of Erkenntnis (Schlick 1930). 21 Schlick, Waismann and other members of the Vienna Circle saw knowledge as the correlation of empirical facts with a system of symbols based on formal logic. Such a theoretical innovation was also an intellectual revolution under the sway of modernism, as Carnap formulated so eloquently in his preface to The Logical Construction. 22 This radical program, in turn, left an indelible mark on avant-garde art (constructivism associated with Gerd Arntz, the artist of Neurath s pictorial language), literature as well as architecture (Werkbund and Bauhaus) centering around Ludwig Wittgenstein, Paul Engelmann, Adolf Loos, and Josef Frank, as well as in the context of Neurath s efforts within the Congrés International d Architecture Moderne ( CIAM). 23 Clarity and precision as both ends in themselves and features of scientific philosophy bridged both Wittgenstein s cultural pessimism and the socio-culturally enlightened impetus of the Vienna Circle. With this convergence of various elements of philosophy of science, the dynamics of theory was accelerated in the phase in which the Vienna Circle made public appearances and expanded its international contacts. The latter development was accompanied by the disintegration and uprooting of Logical Empiricism in the German-speaking world. In this sense, the phenomenon of the Vienna Circle is a prototypical case study on intellectual emigration. 24 The gradual liberalization of verificationism and as Wittgenstein s influence diminished the transition from phenomenalism to physicalism (with Carnap and Neurath) was in full swing as already reflected in the publication and internal reception of the 1929 manifesto. The increasingly international orientation of the Vienna Circle had been evident since the late twenties in the contacts with Hans Reichenbach s Berlin Group, the Warsaw logicians around Alfred Tarski, and the American neo-pragmatist and semiotic movement (Bridgman, Tolman and most notably, Charles Morris). 25 By the beginning of the thirties, a research program had emerged with empiricist, physicalist and phenomenalist variants. (e.g., in Felix Kaufmann, the phenomenologist of the Vienna Circle, who had been influenced by Husserl, Kelsen and Carnap). 26 The themes discussed in Schlick s circle such as the principle of tolerance, the disputed foundational problem of empirical science, and the justification and confirmation method with explanation vs. understanding (a variant of the Metho-

16 XVI FRIEDRICH STADLER denstreit) all reflect the seminal methodological debates that were taking place in mathematics / logic (logicism, formalism, Platonism / intuitionism) and in the natural sciences and humanities. 27 Undoubtedly, the strongest variant had emerged by 1934 from the Unity of Science movement spearheaded by Neurath, Carnap and later Morris. Their efforts culminated in six well-attended international congresses and the (uncompleted) publication project of the International Encyclopedia of Unified Science. 28 This theoretical dynamics and process of differentiation can be illustrated in light of several of the few surviving authentic Vienna Circle documents. 29 To all appearances, we are confronted here with two diametrically opposed trends. While the international influence of the Vienna Circle was steadily growing, the group had been systematically marginalized in Austria and Germany starting in the early thirties. The murder of Moritz Schlick and the disgraceful, for the most part anti-semitic, reactions to this, brutally ushered in the process which I have elsewhere described as the demise of scientific reason. 30 This took place parallel to the general trend at universities, which at the time were increasingly coming under the influence of an growing anti-democratic and racist discourse dominated by clerical-fascist and national socialist forces. This development led to the Anschluss which culminated in systematic dismissals, banishment and annihilation. II. Thus far we have focused mainly on the core of the Vienna Circle the so-called Schlick Circle which we can describe as being a matter of applied complexity theory, so to avoid the usual clichés of positivism and neopositivism. The vitality and productivity of this group that transcended the boundaries of both disciplines and countries was, however, a product of the openness of the logicoempiricist program and the osmosis with the peripheral circles within the sociocultural setting. By way of illustration, I would like to mention just the three most important networks that have been neglected in the pertinent literature or have recurrently prompted highly divergent interpretations: 31 Karl Menger s Mathematical Colloquium ( Gödel s universe ) The Wittgenstein group, including primarily Ludwig Wittgenstein, Moritz Schlick, and Friedrich Waismann The Heinrich Gomperz-Circle, including Karl Popper and Edgar Zilsel, inter alia In this context, it is worth noting that members of the Vienna Circle in addition organized the Verein Ernst Mach (Ernst Mach Society) between 1929 to 1934 as a parallel movement serving to popularize scientific philosophy and The Scientific Conception of the World as part of the Viennese ( Red Vienna s ) adult education movement. 32 Moreover, Otto Neurath s Gesellschafts- und

17 WHAT IS THE VIENNA CIRCLE? XVII Wirtschaftsmuseum in Wien (Museum of Society and Economics in Vienna, ) played an important role with its pictorial statistics ( Wiener Methode der Bildstatistik ) which were developed in exile as the Isotype movement as part of the International Encyclopedia of Unified Science project. 33 The intellectually fertile ground for all these activities was the remarkably strong participation of Vienna Circle members in the Viennese Adult Education institutions. 34 These external perspectives have no bearing on the various interpretations of the transition from Wittgenstein 1 to Wittgenstein 1½ to the late Wittgenstein of Philosophical Investigations where, apart from the anti-enlightenment thrust, there are remarkable convergences with Neurath s position in the Encyclopedia of Unified Science. Here I would like to mention the language game conception with its reference to habit, the so-called scholar s behavioristics and pseudorationalism, all of which are part of the model of an empirical orchestration of the sciences. 35 Since the first Vienna Circle, this fallibilistic concept represented for Neurath a basic motive of his critique of every type of unambiguous verification or falsification, as well as of a hierarchical system of sciences with their ultimate foundation. With such a relativistic motive Neurath formulated on a number of different occasions as the well-known boat metaphor, which was thus mainly directed at the difficult, but successful newcomer Karl Popper. It was also formulated in a more dramatic way to counter the semantic turn of his friend Carnap who stood under the influence of Tarski and Wittgenstein. It was here that the absolutely unbridgeable rifts between philosophical relativism and absolutism became manifest. 36 Here it should be mentioned that this controversy of philosophical relativism vs. absolutism continued even after the forced emigration of the Vienna Circle to the U.S, in Philipp Frank s Institute for the Unity of Science ; it was this dispute, by the way, that was reformulated with the so-called Science Wars in the late 1990s. 37 I would also like to make some similar claims with regard to the perhaps more complex and emotion-laden relationship between Karl Popper and the Vienna Circle, since it continues to inform the current debates. This perspective has long been perpetuated in an uncritical way, most notably in Popper s own accounts, in his teaching and in his research. 38 Against this historical background, the relationship between Karl Popper ( ) and the Vienna Circle becomes easier to understand. In his widely-read autobiography Popper presented and subsequently criticized the Popper legend in the chapter with the not so modest title: Logical Positivism is dead: Who killed Logical Positivism? 39 As already mentioned, the Vienna Circle did not represent a static, homogeneous school of philosophy, with one dominant figure, one main work and basic dogmas. The reception and appraisal of Popper s ideas by different members thus very varied considerably: from positive reactions to the Logic of Scientific Discovery (1934) in Carnap, Feigl, Schlick and Frank to the vehement principle

18 XVIII FRIEDRICH STADLER criticism in Neurath. In his above cited review essay of 1935 Neurath had mainly taken issue with the false search for a privileged system of statements as the paradigm of empirical sciences. 40 Popper found this decisive criticism rather flattering, although he did not, however, respond to Neurath s counter-interpretation. 41 Neurath had given preference to his meta-theoretical holism and an epistemological relativism as an alternative to the philosophical absolutism that even Popper himself conceded and this in spite of the fact that, within the Vienna Circle, there were some like Viktor Kraft, Karl Menger, Kurt Gödel and Herbert Feigl who expressed their allegiance to variants of a metaphysical (constructive) realism and intuitionism. Even Schlick, who apparently for personal reasons did not invite Popper to his circle on Boltzmanngasse, paid tribute to Popper s Logic of Scientific Discovery by including it in the series Schriften zur Wissenschaftlichen Weltanschauung, which he and Frank edited. 42 Thus the meaning criterion of verification, too, became increasingly liberal. From the mid-thirties on it hardly played a role anymore in the Encyclopedia of Unified Science. It had originally been intended more as a pragmatic concept of metaphysics (with Ockham s razor as an antidote) during Vienna s inter-war years. (Apart from that period, the history of verificationism was more representative of the status of normal science). 43 In the publications and protocols, the Vienna Circle inductivist and hypothetico-deductive methodologies seemed to co-exist. No strict distinction was made between context of justification and context of discovery, which allows us to draw the following conclusions which are also of general import for the interpretation of Vienna Circle and Logical Empiricism: The history of the development of Logical Empiricist theories since the turn of the century does not allow any clear canonization of a philosophical school in the strict sense, since what we are dealing with is a dynamic between center and periphery. The varying receptions of Wittgenstein, Tarski and Popper have influenced the development of various philosophies of science inspired by rational reconstruction, on the one hand, and by encyclopedic models on the other. Thus, if I were a police commissioner, I would give the following answer to Popper s rhetorical question: if it s really true that the patient died, then there is more than one perpetrator, especially from within the circle, both disciples and critics alike. But this, of course, is a completely different story which should be told in another context. III. Let me come to my tentative conclusions with an illustration of the issue at hand, which is to be elaborated on in forthcoming studies:

19 WHAT IS THE VIENNA CIRCLE? XIX THE VIENNA CIRCLE / LOGICAL EMPIRICISM ELEMENTS OF A RE-EVALUATION 1. Methods a) Intradisciplinary: Scientific Philosophy b) Interdisciplinary: Scientific Conception of the World Wissenschaftslogik Philosophy of Science Wissenschaftstheorie c) Transdisciplinary: Encyclopedia of Unified Science Foundations of the Unity of Science d) Philosophy, Science and Art: Werkbund and Bauhaus External Characteristic: Coffeehouse Science (Local and Global Networking) Science as Culture 2. Scientific Communication a) International: Urban Cultures of Science (Vienna, Berlin, Prague, Warsaw, Cracow, Budapest, Lvov, Copenhagen, Paris, Cambridge /Oxford, London...) b) Multi-ethnic: mainly enlightened Jewish and many Ethnics/Nationalities c) Multi-lingual: Communication /Publication in German, English, French, Italian, Polish, Turkish Characteristic: Cultural Exodus with Disintegration (from the German speaking World) Internationalization (into the Anglo-Saxon World) 3. (Self-)Organization a) Academic Field: Universities /Academy of Science b) Extra-academic Context: Adult Education and Viennese Cultural Movement of Spätaufklärung : Popularisation of Scientific World Conception Institutions: Ernst Mach -Society, Adult Educational Institutions ( Volkshochschulen ), Pictorial Statistics /Isotype with the Social and Economic Museum (Gesellschafts- und Wirtschaftsmuseum in Wien) c) Formal /Informal: Conferences, Lectures /Lecture Series, Seminars, Discussion Groups resp. Books /Book Series, Journals, Working Papers Characteristic: Philosophy of Science as a sort of Social Epistemology 4. Sociology of Science and Knowledge a) Men and Women: Else Frenkel-Brunswik, Olga Hahn-Neurath, Marie Jahoda, Marie Reidemeister-Neurath, Olga Taussky-Todd, Rose Rand, Hilde Geiringer-Mises, Susan Stebbing, Janina Hosiasson-Lindenbaum... b) Several Generations and Professions: Professors (Distingushed Chairs /Associate/ Assistant), Students and Visiting Scholars from Abroad Characteristic: Republic of Scholars (Neurath) Result Uprooting and Alienation from Central Europe! Americanization with Mutual Change of Theories

20 XX FRIEDRICH STADLER In addition to the well-known insights that still inform today s history and philosophy of science (e.g., the explanation and validation of scientific theories; foundations and methods of science between induction and deduction; unity and plurality of science, and so on), I would like to emphasize other topical features of the Vienna Circle which meriting further study. Generally, we see the realization of a common fallibilistic epistemology that points to the hypothetical status of scientific knowledge which is only gradually integrated in everyday experience and language (e.g., common sense). Therefore, the usage of symbolic logic did not contradict a scientific philosophy which is at the same time a study in human understanding (Richard von Mises 1951). And I am not reluctant to go back to the classical roots of the Vienna Circle by deliberately using the term methodological relativism. 44 First of all, it is a way of philosophizing based on a language-critical attitude and a great amount of problem-oriented, open-ended discussion. This is something experienced personally by Arne Naess, who as the last living member of the Schlick circle focuses in this volume on the Vienna Circle s thought style which, in (not only) his opinion, leads to an inherent pluralism of tenable worldviews. Secondly, the use of unambiguous language, together with exact methods is certainly a main legacy of the Schlick Circle and those associated with it: it is only given this formal approach that content and positions can be criticized and refuted a characteristic which most current modern and postmodern philosophies lack. At least the challenge of factual research, and in particular the admittedly, non-linear advancement of science (the humanities and natural science) is an essential reason why the tension between science and philosophy (empiricism and rationalism) cannot be resolved definitively. However, the reconciliation of research in different fields of science facilitates permanent reflection and reevaluation of the complex concept called scientific philosophy. This reconciliation began in the classical period with its international and interdisciplinary networking, without aiming at one more philosophical dogmatic tradition (which is, by the way, normally characterized by an authoritative head, some leading textbooks and a hierarchical organization covering a limited number of academic fields.) A serious alternative to this (Vienna-Circle oriented) open-minded position can hardly be formulated. I believe that the present volume documents the wide range, pluralism and diversity of the Viennese heritage and message, be it called scientific philosophy (as initiated by Schlick), scientific humanism (according to Carnap) or a republic of scholars (following Neurath), as a guide to a intellectual journey which continues through the present day and on into the future.

21 WHAT IS THE VIENNA CIRCLE? XXI NOTES Karl Menger, Ist die Quadratur des Kreises lösbar? ( Is it Possible to Square the Circle? ), in: Alte Probleme Neue Lösungen in den exakten Wissenschaften. Fünf Wiener Vorträge, Zweiter Zyklus. Leipzig & Wien 1934, Barry Smith, Parts and Moments. Studies in Logic and Formal Ontology. Munich: Philosophia Peter Simons, Parts. A Study in Ontology. Oxford University Press An abridged reprint in English translation in: Otto Neurath. Empiricism and Sociology. Ed. by Marie Neurath and Robert S. Cohen. Dordrecht: Reidel 1973, Albert Blumberg /Herbert Feigl, Logical Positivism, in: Journal of Philosophy 28, 1931, Members of the inner circle: Gustav Bergmann, Rudolf Carnap, Herbert Feigl, Philipp Frank, Kurt Goedel, Hans Hahn, Olga Hahn-Neurath, Béla Juhos, Felix Kaufmann, Viktor Kraft, Karl Menger, Richard von Mises, Otto Neurath, Rose Rand, Josef Schächter, Moritz Schlick, Olga Taussky-Todd, Friedrich Waismann, Edgar Zilsel. Members of the periphery are: Alfred Jules Ayer, Egon Brunswik, Karl Bühler, Josef Frank, Else Frenkel-Brunswik, Heinrich Gomperz, Carl Gustav Hempel, Eino Kaila, Hans Kelsen, Charles W. Morris, Arne Naess, Karl R. Popper, Willard Van Orman Quine, Frank P. Ramsey, Hans Reichenbach, Kurt Reidemeister, Alfred Tarski, Ludwig Wittgenstein. Regarding the relation of Schlick and Neurath cf.: Rudolf Haller (Ed.), Schlick und Neurath. Ein Symposium. Amsterdam: Rodopi As to the network with Berlin, Warsaw and Prague: Rudolf Haller / Friedrich Stadler (Eds.), Wien Berlin Prag. Der Aufstieg der wissenschaftlichen Philosophie. Wien: Hölder-Pichler- Tempsky 1993; Lutz Danneberg /Andreas Kamlah /Lothar Schäfer (Hrsg.), Hans Reichenbach und die Berliner Gruppe. Braunschweig-Wiesbaden: Vieweg Klemens Szaniawski (Ed.), The Vienna Circle and the Lvov-Warsaw School. Dordrecht-Boston-London: Kluwer 1989; Francesco Coniglione, Roberto Poli, Jan Wolenski (Eds.), Polish Scientific Philosophy. The Lvov-Warsaw School. Amsterdam-Atlanta: Rodopi 1993; Katarzyna Kijania-Placek and Jan Wolenski (Eds.), The Lvov-Warsaw School and Contemporary Philosophy. Dordrecht-Boston- London: Kluwer 1998; Jan Wolenski, Essay in the History of Logic and Logical Philosophy. Cracow: Jagiellonian University Press Gustav Bergmann, Memories of the Vienna Circle. Letter to Otto Neurath (1938), in: Friedrich Stadler (Ed.), Scientific Philosophy. Origins and Developments. Dordrecht-Boston- London: Kluwer 1993, 207. Cf. Volker Thurm-Nemeth (Hrsg.), Konstruktion zwischen Werkbund und Bauhaus. Wissenschaft Architektur Wiener Kreis. Wien: Holder-Pichler-Tempsky 1998; Wendelin Schmidt- Dengler (Hrsg.), Fiction in Science - Science in Fiction. Zum Gespräch zwischen Literatur und Wissenschaft. Wien: Hölder-Pichler-Tempsky To mention only two examples. Ronald Giere and Alan Richardson (Eds.), Origins of Logical Empiricism. Minneapolis: University of Minnesota Press 1996; Gary Hardcastle /Alan Richardson (Eds.), Logical Empiricism in North America. Minneapolis: University of Minnesota Press Cf. Michael Friedman, Logical Positivism, and Friedrich Stadler, Vienna Circle, both in: Routledge Encyclopedia of Philosophy. London-New York As general reference cf.: Friedrich Stadler, The Vienna Circle. Studies in the Origins, Development, and Influence of Logical Empiricism. Wien-New York: Springer Cf. Karl Menger, A Counterpart of Ockham s Razor, and Geometry and Positivism. A Positivistic Microgeometry, in: Karl Menger, Selected Papers in Logic and Foundations, Didactics, Economics. Dordrecht-Boston-London: Reidel 1979, , Richard von Mises, Positivism. A Study in Human Understanding. New York: Dover 1951, 99 Alberto Coffa, The Semantic Tradition from Kant to Carnap. To the Vienna Station. Cambridge University Press 1991, Michael Friedman, Rediscovering Logical Positivism. Cambridge Uni-

22 XXII FRIEDRICH STADLER versity Press 1999; Michael Friedman, A Parting of the Ways. Carnap, Cassirer, and Heidegger. Chicago and La Salle, Ill.: Open Court 2000; Alan W. Richardson, Carnap s Construction of the World. The Aufbau and the Emergence of Logical Empiricism. Cambridge University Press Thomas Uebel, Vernunftkritik und Wissenschaft: Otto Neurath und der Erste Wiener Kreis. Wien-New York: Springer Cited in: Stadler, The Vienna Circle, op.cit., 196. Karl Menger, Ergebnisse eines Mathematischen Kolloquiums. Ed. by E. Dierker and K. Sigmund. Wien-New York: Springer 1998; Karl Menger, Reminiscences of the Vienna Circle and the Mathematical Colloquium. Ed. by Louise Golland, Brian McGuinness and Abe Sklar. Dordrecht-Boston-London: Kluwer First Publication: Willard Van Orman Quine, Two Dogmas of Empiricism, in: Philosophical Review 60, 1951, Jan Wolenski and Eckehart Köhler (Eds.), Alfred Tarski and the Vienna Circle. Austro-Polish Connections in Logical Empiricism. Dordrecht-Boston-London: Kluwer Moritz Schlick, The Turning Point in Philosophy (1930), in: Schlick, Philosophical Papers. Ed by H.Mulder and Barbara F.B. van der Velde-Schlick. Dordrecht-Boston-London: Reidel 1979, Rudolf Carnap, The Logical Structure of the World. Berkeley and Los Angeles: University of California Press Translated by Rolf George. Cf.: Elisabeth Nemeth and Friedrich Stadler (Eds.), Encyclopedia and Utopia. The Life and Work of Otto Neurath ( ). Dordrecht-Boston-London: Kluwer Cf. Friedrich Stadler, Transfer and Transformation of Logical Empiricism, in: Giere/ Richardson (Eds.), op.cit, and The Wiener Kreis in Great Britain: Emigration and Interaction in the Philosophy of Science, in: Edward Timms / Jon Hughes (Eds.), Intellectual Migration and Cultural Transformation. Wien-New York: Springer Cf. Gerald Holton, From the Vienna Circle to the Harvard Square, in: Stadler, Scientific Philosophy op.cit., Friedrich Stadler (Hrsg.), Phänomenologie und Logischer Empirismus. Zentenarium Felix Kaufmann. Wien-New York: Springer 1997; Clemens Jabloner / Friedrich Stadler (Eds.), Logischer Empirismus und Reine Rechtslehre. Beziehungen zwischen dem Wiener Kreis und der Hans Kelsen-Schule. Wien-New York: Springer Cf. The Foundational Debate. Complexity and Constructivity in Mathematics and Physics. Ed. by Werner DePauli-Schimanovich, Eckehart Köhler and Friedrich Stadler. Dordrecht-Boston- London: Kluwer Otto Neurath / Rudolf Carnap / Charles Morris (Eds.), Foundations of the Unity of Science. Towards an International Encyclopedia of Unified Science. 2 Volumes. The University of Chicago Press: Chicago and London 1970f. Cf. the protocols of the Schlick Circle and the theses of the Vienna Circle as reported by Rose Rand in: Stadler, The Vienna Circle, op.cit., ch Friedrich Stadler/Peter Weibel (Eds.), The Cultural Exodus from Austria / Vertreibung der Vernunft. Wien-New York: Springer A reconstruction in detail in: Stadler, The Vienna Circle, op.cit., chapters Stadler, ibid., ch. 7.2, and John Blackmore (Ed.), Ernst Mach A Deeper Look. Documents and Perspectives. Dordrecht-Boston-London: Kluwer 1992: John Blackmore, Ryoichi Itagaki, Setsuko Tanaka (Eds.), Ernst Mach s Vienna or Phenomenalism as Philosophy of Science. Dordrecht-Boston-London: Kluwer Nemeth /Stadler (eds.), Encyclopedia and Utopia, op.cit., section III. The following members lectured and instructed courses in several institutions ( Volkshochschulen ): Rudolf Carnap, Herbert Feigl, Philipp Frank, Hans Hahn, Felix Kaufmann, Viktor Kraft, Otto Neurath, Moritz Schlick, Friedrich Waismann, and Edgar Zilsel. Otto Neurath, The Orchestration of the Sciences by the Encyclopedism of Logical Empiricism (1946), in: Neurath, Philosophical Papers Ed. by Robert S. Cohen and Marie Neurath. Dordrecht-Boston-Lancaster: Reidel 1983,

23 WHAT IS THE VIENNA CIRCLE? XXIII Otto Neurath, Pseudorationalism of Falsification, in: ibid., From an Popperian point of view cf.: Malachi H. Hacohen, Karl Popper The Formative Years Politics and Philosophy in Interwar Vienna. Cambridge University Press Cf. Friedrich Stadler, Challenging the Dogma of the Ahistorical Philosophy of Science: The Case of Relativism, in: Jutta Schickore / Friedrich Steinle (Eds.), Revisiting Discovery and Justification. Berlin: Max Plack Institute for the History of Science (Preprint 211). Regarding a contextual reconstruction of Heinrich Gomperz, Karl Popper and the Vienna Circle: Stadler, The Vienna Circle, op.cit, ch. 10, and Martin Seiler / Friedrich Stadler (Hrsg.), Heinrich Gomperz, Karl Popper und die österreichische Philosophie. Amsterdam-Atlanta: Rodopi In: The Philosophy of Karl Popper. Ed. by P.A. Schilpp. La Salle, Ill.: Open Court 1974, By the way, popper never replied to this decisive criticism, cf. Stadler, The Vienna Circle, op.cit., Interview with Sir Karl Popper (1991), Ibid., 487. Letter Schlick to Frank, June 9, Vienna Circle Archives, Haarlem, Holland. Cf. C.J. Misak, Veriflcationism. Its History and Prospects. London-New York: Routledge Philipp Frank, Relativity A Richer Truth. Foreword by Albert Einstein. Boston: Vienna Circle Institute Museumstraße 5/2/19 A-1070 Vienna Austria Friedrich.Stadler@univie.ac.at

24 ARNE NAESS PLURALISM OF TENABLE WORLD VIEWS I shall try to outline a philosophical point which I think might be acceptable from a combined logical and empirical point of view. From the world view (wissenschaftliche Weltauffassung) to the manifoldness (Mannigfaltigkeit) of tenable world views. In the following, I consider humans as not insignificant parts of the world. Different views about humanity therefore imply different world views. A second premise I assume increases the manifold of which I speak. By scientific I do not here mean something like implied by science but a weaker contention: compatible with science. What about religious views if a weaker contention is adopted? The formidable development of Christian theology since Søren Kierkegaard has reduced earlier tendencies to assume that there must be a conflict between Christian and scientific general views. The number of what were called Christian dogmas is reduced. Here I shall only suggest that this development plus the considerable increase in the number of people who favor a sort of Buddhism increases compatibility. But I find it premature to take up the many implied relevant questions. The only remark I tentatively make is that those questions suggest a possible scientific acceptability of certain world views which have a strong religious flavour. Logical empiricists, as I understood their Weltauffassung, suggested a view of the world derivable from scientific knowledge. Because scientific knowledge apart from mathematics and logic was clearly considered hypothetical, the status of a definite articulation of such a view would, by necessity, be rather hypothetical. One way to delimit the use of the term science is by assuming that it only comprises the natural sciences. In what follows, I include a large part of the humanities. Historical research is scientific in the fairly large sense to which I subscribe. A world view that clearly negates results of historical research is not a world view compatible with science. Historiography, the streamlined accounts of enormously complicated happenings, need not be compatible with scientific method. Two mutually incompatible accounts of a revolution may both be excellent, but not part of a science. Suppose somebody asks you: What is your world view?, suggesting that he or she would write an article or a book describing (and criticising?) it. It is fairly clear that what would be considered an adequate answer would have to take into consideration that person s value priorities. We should demand of a descripton of 3 F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

25 4 ARNE NAESS a world view actually held by a human being that it take into account his or her value priorities. The logical empiricists took this seriously. Especially when political issues were being discussed, they outlined value priority views which were simply dangerous. That is, the increasingly powerful national socialist movement made public assertions of opposite views likely to result in repressive attacks, sometimes even in incarceration in a concentration camp. The pertinent answer to the formidable question What is your world view? will assume different forms: if you and your family are politically repressed, or if you are engaged in a dangerous fight against oppressors, your world will be narrower. If you live in a peaceful country like Norway, and you are without severe political or personal problems, perhaps only then are you likely to take very broad questions such as those of cosmology seriously. Incidentally, descriptions of what is going on in the cosmos by professional cosmologists tend to be talk about vast, unimaginable explosions within vast areas of time. This may not give rise to dark or pessimistic, tenable world views. The formidable power of the Catholic Church in Vienna was felt in everyday life. Traffic came to a halt when some religious procession required considerable space in the streets. I remember such an occasion in which traffic came to a complete standstill because a relic, a piece of bone that might possibly have belonged to Christ, was triumphantly paraded through the city. At least one of the logical empiricists, Otto Neurath, had respect for and perhaps was jealous of, the power of the Church. He admired certain trends within the philosophical traditions of Catholicism. He insisted that especially Thomas Aquinas, but also other theologians developed their ideas on a remarkably high logical level. For certain axioms, they derived conclusions in a logically safe way. It was not by chance that Otto Neurath again and again warned against what he called excessive respect for mathematics and logic: Logic? Leave it to the Catholic experts!. He himself had great competence in logic and the history of logic, but having been active in politics not without serious consequences (prison) he warned against trying to formulate logical empiricist political views on a high logical level. These views were, as could be expected in the 1930s, mostly socialist and sometimes Marxist. But back to the question of limits to the diversity of world views compatible with the results of scientific research. More than most philosophers, I admire and cherish research rather than science. The endlessness of research, and the short life of definite scientific theories compared, for instance, to ethical norms, are for me a source of great admiration. It has reminded me of climbing: always risky to some extent, but manageable and mostly open to well-founded trust. I recommended in 1935 that the terms antimetaphysical (or even ametaphysical ) be left out of the formulations of basic characteristics of logical empiricism. Rudolf Carnap s view that philosophy in the future would essentially consist in the elaboration of the logical syntax of language, I found deplorable. I recommended that research attitude should be used as a central expression and slogan, rather than scientific world view. Our views should not be incompatible with a

26 PLURALISM OF TENABLE WORLD VIEWS 5 consistent research attitude. (After all, there are innumerable questions we pose which are not even very thinly covered by painstaking research.) There is not much vitally relevant science to point to. Propaganda for more such research, and a more consistent research attitude, are both essential. The metaphysics of Spinoza includes a definition of freedom and of free choice. Very roughly, a decision is free according to him when taken under no external pressure whatsoever. But God (Deus) is defined in such a way that it does not invite research. If we start with the last part of the Ethics, and not the formidable first part, we are open for research and redefinition of terms. This, in turn, facilitates an intense research attitude in relation to metaphysical texts in general. From the last the fifth part, we may proceed to the broad, fourth part. Every point there has a practical, a life import. In short, it is possible to maintain a research attitude, even when interpreting and applying the dense metaphysics of Spinoza. When I was staying in Carnap s home in California, this was one of the themes of our daily discussion. On entering the Schlick seminar in Vienna in 1934, I presented myself as a kind of Spinozist. Spinoza enjoyed, of course, a high standing among the seminar members, but philosophy the way Spinoza practised it was of course fully, totally, decisively a matter of the past. So it was considered touching and in a way, admirable, to be a kind of Spinozist, but philosophically it was centuries too late. What I am driving at is a complete acceptance of the kind of metaphysical formulations of a philosopher like Spinoza. That is, acceptance of their meaningfulness. But as for the kind of understandable contemporary articulations which would today be nearest to the meaning of his formulations is an open question. Different sets of answers would play a role in outlining different Spinozist world views. They would present examples of scientifically acceptable, mutually more or less incompatible world views. Studying Immanuel Kant s texts, we may arrive at similar, or at least analogous, conclusions. Considering the great number of mutually incompatible nineteenth century and later Kant interpretations, we might point to the possibility of a variety of modern Kantian world views. What about old Chinese and Indian philosophies? We may tentatively interpret the texts so as to make them relevant to the questions of scientific compatibility. The pervasive relevance of yoga in Indian philosophy has been studied extensively in the West as a practice, a strangely close combination of a seemingly abstruse and vague philosophy with very definite, clearly described practices. As an example of basic philosophical terms rather different from western ones, I would like to concentrate on emptiness, sunyata in Sanskrit. At an East-West international philosophical congress, an Indian participant left his chair in a quiet way and crawled under the table. When asked closely later what he meant by this, he tried to make us understand that his movement was relevant and philosophically understandable within the Indian yoga tradition he belonged to.

27 6 ARNE NAESS What I am suggesting is the acceptance of the existence of a variety of different world views, all compatible with science. But it should also be added that the research centering on this diversity reveals the very serious difficulty of finding a vocabulary or a terminology that, at least not entirely superficially, describes the cognitive content of each of the different world views. The deeper the differences, the more difficult it is to find a terminology that makes a survey possible. Why is the richness of world views today perhaps more important for the future of humanity than ever before? Because, as I see it, vast globalization is inevitable. What by many people is believed to be on the whole a great advantage, might turn into a calamity in my view. War becomes more unlikely because war within a closed system of globalised economic liberalism will rightly be seen as a threat to that very economic system. But the likelihood of people and societies with a pronounced tendency to maintain clearly different world views decreases. The intensity of human interaction crossing all borders may foster a certain level of Gleichschaltung, a tendency for a decrease in the diversity of world views realised and acted upon at any definite time. In the 1930s there was in Europe a decrease in the diversity of world views because of the intense focus on (Soviet) communism versus anti-communism, and Nazism versus anti-nazism. Today, other factors tend to limit diversity. The globalization of technology and perhaps the intensification of interaction through and chatting on the Internet? In 1934 and 1935 it seemed paradoxical to me that logical empiricists stressed the term empiricism. They tended to reject the idea that painstaking empirical research could be a genuine part of philosophical inquiries. Philosophy as logical syntax of language did not, according to Carnap, need anything of the kind. It needed theoretical clarification, but he did not put up any list of questions of an empirical kind which had to be answered in order to discover the logical syntax of actual spoken languages. There are extremely different languages so different that the translation of one into another is a highly creative job very different from translations within the scope of English, French and German. The definite and complete rejection by Carnap of Heidegger s sentence Das Nichts nichtet did not, for Carnap, need substantial empirical support. I mean empirical research. Movements comparable to logical empiricism might invest quite a deep concern to avoid unnecessary rejections. Let me for a moment consider a special movement. Within the ecological movement there are groups often characterized as doomsday prophets. They are, as I see it, generally criticised in an unnecessarily harsh way. I propose that we say something of this kind: Yes, if present trends of unecological policies continue long enough, catastrophes are likely to occur, but we don t have reason to believe that such polices will actually persist long enough to make catastrophes likely. The opinion of Moritz Schlick and certain other logical empiricists that philosophy of the traditional kind was dead or would soon die did not destroy the importance of what the traditionalists were saying. The same applies to the

28 PLURALISM OF TENABLE WORLD VIEWS 7 doomsday prophets. Much of what they say is of some interest. It is only their radical negative predictions that need not be taken very seriously. Could the study of the remarkable influence of logical empiricism in the thirties help to instigate a comparable movement today? We talk about a global green movement. Within that movement a diversity of world views is powerfully presented. One is that of favoring a rich life with simple means. It presumes that people have fairly clear value priorities. Instead of mathematics and formal logic being studied with special intensity, there would be studies of life quality, how one feels about oneself and the world. Comparative studies of what makes people feel well, of course, including the study of feeling well in the economically poor countries, are relevant. Logical empiricism was acutely controversial from the first moment. To the extent that such a reception is inevitable, it is highly suspicious if a movement does not at least start as a controversial phenomenon. The green movement called deep ecology was for many years highly controversial. But at a fairly recent international congress in Melbourne, Australia, it appeared to be fairly, if not totally, uncontroversial. At the same conference, the following question became more pertinent: But what can be done to change politics in a green direction? That question refers to problems not directly attacked by the deep ecology movement as such. Logical empiricists certainly were fully aware of the necessity of being politically active. Are they comparably healthy movements today? I don t think there are any, but it is important for many of us to try to influence movements we wish well to reach the high level of creative dialogue and sober mutual respect in spite of differences of opinion which characterized the logical empirical movement. Centre for Development and Environment University of Oslo, Box 1116 Blindern N-0317 Oslo Norway k.f.nass@sum.uio.no

29 PAOLO PARRINI ON THE FORMATION OF LOGICAL EMPIRICISM 1. THE STATUS QUAESTIONIS During the discussion between D. Howard and T. Oberdan at the 1991 Conference, organized in Konstanz for the centenary of Carnap s and Reichenbach s birth, the following question was posed: is it possible to see Schlick s Allgemeine Erkenntnislehre as an anticipation of the theses that later characterized the Vienna Circle (which was formed around Schlick himself) and that were formulated by its members taking into consideration also Russell s and Wittgenstein s ideas? 1 This question refers to the existence of a first Vienna Circle and the nature of the relationship between this first group and the real Vienna Circle. T. Uebel has re-examined this second question in a paper recently presented at the Florentine conference on Logical Empiricism (1999). Looking at it from an interesting perspective, he aims at singling out the characteristics of Logical Empiricism as a distinctive school of philosophical thought. More precisely, he asks himself whether it was simply another version of neo-kantian philosophy and, if not, what other elements were present. In order to provide an answer, he discusses again the relationship between the Austrian and German-Kantian components underlying the movement and links his paper to the recent debate on the distinction between analytic and continental philosophy. According to Uebel, it was the co-operation of the members of the early circle of philosophically minded Austrian scientists the mathematician Hans Hahn, the physicist Philipp Frank and the economist Otto Neurath with the scientifically trained German philosophers Moritz Schlick and Rudolf Carnap that accounted for the distinctive force of Viennese Logical Empiricism. This can be clearly seen he states when we look at the positions taken by Schlick in the Allgemeine Erkenntnislehre (1918 1, ). Uebel explicitly says, though, that in his reconstruction he does not take into consideration the Berlin strand of Logical Empiricism 2. I intend here to discuss this very same set of questions showing how the analysis of the evolution of the ideas of the main member of the Berlin Circle H. Reichenbach and thus of the more German branch of the neo-empiricist movement, can bring to light some interesting elements. Important indications can be found in the philosophy of geometry he developed in Philosophie der 9 F. Stadler (ed.). The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

30 10 PAOLO PARRINI Raum-Zeit-Lehre (1928). More precisely, I will try to show that the analysis of the debate on Kantism that divided Reichenbach and Schlick in the early 1920s at the time of their discussion on the theory of relativity, and the analysis of Reichenbach s subsequent change of perspective on the nature of coordinative principles are enlightening on two relevant points that are closely related to Howard s, Oberdan s and Uebel s above-mentioned interpretations: (i) the relationship between Kantism and Conventionalism within Logical Empiricism; (ii) the role played by typically neo-empiricistic ideas such as the verification principle and the thesis of the tautological nature of logic ideas which are more of an analytic than of a continental nature regarding the possibility of using French Conventionalism in a strictly neo-empiricistic sense and not in a neo- Kantian one. My final goal is to show that we cannot understand Logical Empiricism s development if, as traditional interpretations did, we focus our attention only on the positions developed under the influence of Russell s and Wittgenstein s ideas, thereby ignoring the Kantian and Conventionalist heritage of the movement. On the other hand, in order to show the importance of these components, we must not overlook the other half of the apple : the conceptual reorganisation brought about by a reconsideration of some old problems in light of the new theoretical approach linked to the coming to the fore of some theses (verification principle and linguistic theory of the a priori) which are the consequence of the impact of Russell s and Wittgenstein s ideas interpreted from an empiricistic perspective. 2. SCHLICK S CRITICISM OF IMMANENCE PHILOSOPHIES AND REICHENBACH S CONCEPTION OF THE OBJECT OF KNOWLEDGE To illustrate my point, I will first focus on the criticism Schlick made in the first edition of Allgemeine Erkenntnislehre (1918) against immanence philosophies, in particular Schuppe s immanentism, on the basis of his conception of knowledge intended as coordination. In criticizing Schuppe s immanentism, Schlick says that it is quite clear that thought, in the sense relevant to knowledge, signifies nothing but the designation of objects. But that an object is not produced by our giving it a designation indeed, it is independent of it and can exist without our correlating some sign or representation with it is all contained in the very concept of designation. The above fallacy would never have been committed if the two meanings of the word thought had been kept apart by assigning different terms to them 3. In other words, Schlick considers his conception of knowledge to be antithetic to Schuppe s, because it encompasses the idea that the sense of to think inherent to knowledge means only to designate objects and that it has nothing to do with any kind of dependence of the objects on the thought designating them. As we know, in the Allgemeine Erkenntnislehre these objects are charac-

31 ON THE FORMATION OF LOGICAL EMPIRICISM 11 terized as transcendent objects and even called by the Kantian expression Dinge an sich (things-in-themselves). In Reichenbach s conception, which differs from Schlick s, the equation of the process of knowing to a process of unique coordination could be used to explain why the object of knowledge, constituting some sort of non-metaphysical transcendence, remains nevertheless immanent with respect to the presuppositions of the cognitive process. In his 1920 essay Relativitätstheorie und Erkenntnis Apriori, Reichenbach used the characterization of knowledge elaborated by Schlick partly to reject and partly to accept Kant s conception of the a priori. Here the coordinative conventions mentioned by Schlick are treated as constitutive principles. They are introduced and characterized in contrast to particular laws by calling into question their constitutive value with respect to the objects of knowledge: in contrast to particular laws (or axioms of connection [Verknüpfungsaxiome]), these constitutive principles (or coordinating principles, axioms of coordination [Zuordnungsaxiome]), do not say what is known in the individual case, but how knowledge is obtained; they define the knowable, [...] they show the order rules according to which knowledge is obtained and indicate the conditions the logical satisfaction of which leads to knowledge. For Reichenbach, in the process of knowing, one of the two sides of knowledge the real is defined by coordination with the equations. The object of knowledge thus constitutes itself in a way which is immanent with respect to the coordination principles. Its transcendence is due to the fact (devoid of any metaphysical connotations) that it is susceptible to a potentially unlimited number of empirical determinations 4. In Schlick s conception of knowledge, the elements of one of the ends of the cognitive connection (concepts and judgments) are coordinated with the elements of the other end (objects and facts) by means of the introduction of suitable connections of a definitional and conventional nature. It is by means of these kinds of connections (usual, concrete, implicit and coordinative conventions) that we make the association between the apparatus of symbols and the entities (objects and facts) that concepts and judgements have to designate. What remains valid in Reichenbach is Schlick s idea of knowledge as coordination or designation, and truth as unique coordination or designation. The elements of the two ends of the coordination, though, are not conceived as preexisting to the coordination itself: coordination connections have a function in the constitution of the objects of knowledge. Evidently, even though they both consider as valid the coordinative or designative conception of knowledge, these two ways of looking at the cognitive process are profoundly different. This difference affects and is made evident by Schlick s and Reichenbach s attitudes towards Kant s theory of synthetic a priori judgements.

32 12 PAOLO PARRINI 3. THE DISAGREEMENT BETWEEN SCHLICK AND REICHENBACH ON KANTIAN PHILOSOPHY AND CONVENTIONALISM This divergence of opinions on the object of knowledge refers back to their wellknown disagreement regarding the way in which they rejected Kantian synthetic a priori judgements. Reichenbach distinguished between two different meanings of the concept of a priori: 1) a priori in the sense of necessarily true or true for all times, and thus of absolutely independent of experience; 2) a priori, in the sense of constituting the concept of object, that is, the principle that forms the object of scientific knowledge, more specifically of physics. The theory of relativity shows the indefensibility of the first notion of a priori, but not of the second 5. On the contrary, since the years of the first edition of Allgemeine Erkenntnislehre, Schlick did not seem willing to concede anything to Kantianism. On this point he criticizes both Cassirer and Reichenbach, and maintains that the essential characteristic of the Kantian synthetic a priori is the connection, in the very same concept, between the apodeictic certainty and the constitutive value. Consequently, he does not make a distinction between the two meanings of Kantian synthetic a priori, thereby avoiding its adoption in any of its two senses. The more general and abstract component of scientific knowledge should be assimilated not in Kantian constitutive principles (even if devoid of any apodeictic validity) but to very general hypotheses or to conventions as defined by Poincaré 6. And if we ask ourselves what convention in Poincaré s sense means, the only sensible answer we can give is that we must be talking about conventions of a linguistic-definitional nature conventions the task of which consists in establishing scientific language regarding the formulation of both scientific facts and laws. Poincaré explicitly states this thesis in his pages on geometry and in his polemic against Eduard Le Roy s radical conventionalism ( nominalism ). 7 The terms of agreement and disagreement between Schlick and Reichenbach in the 1920s are now clear. They both took as their starting point a general interpretation of knowledge intended as coordination and the negation of Kant s theory of the synthetic a priori. Within this very same framework, though, Schlick states that in order to understand cognitive coordination we need admit only the existence of linguistic conventions, in full conformity with his idea that in the cognitive process we have a designation of objects and facts by means of suitable concepts and judgements on the basis of previously fixed conventions. According to Reichenbach cognitive coordination can be understood only if we admit the existence of coordination principles that constitute the objects of knowledge and that, as such, have to be assimilated to Kant s synthetic a priori principles in all features excepting that of their apodeictic validity.

33 ON THE FORMATION OF LOGICAL EMPIRICISM 13 The terms of their disagreement were thus as profound and strong as the terms of their agreement. Despite this, in the early 1920s both authors gradually developed a more similar perspective that aimed especially at defending in contrast to Kant and Cassirer s neo-kantianism the contingent nature of the cognitive synthesis, in other words, the idea related to the thesis of the independence of perception of reason that the very same possibility of coordinating concepts to experiences and obtaining a unique coordination (truth) depends on experience. This possibility has been proven by now, but we cannot demonstrate a priori its necessity 8. All this is indicated by the numerous letters they exchanged on this topic between September and December 1920, as well as by the conclusive comment on their discussion made by Reichenbach in Der gegenwätige Stand der Relativitätsdiskussion THE ANALYSIS OF GEOMETRY AND CONGRUENCE IN PHILOSOPHIE DER RAUM-ZEIT-LEHRE I think that the disagreement between Schlick and Reichenbach was not of an exclusively terminological nature. It was more related to the question of whether it is possible to understand the cognitive process by taking recourse to only two components, experience and linguistic conventions (including coordinative conventions), or whether besides these two we also have to admit some presuppositions of a synthetic nature constitutive not only of science s language (such as the more strictly linguistic ones) but also of the objects of knowledge. Disregarding our judgement on this matter from the theoretical point of view, from a historical point of view we have to underline that in the years following 1922, Reichenbach changed his original position and ended up embracing the one defended by Schlick. Even in his 1924 work Axiomatik der relativistischen Raum-Zeit-Lehre, Reichenbach adopts Schlick s perspective and terminology: he refers to implicit and explicit definitions (impliziten and expliziten Definitionen), but instead of speaking of constitutive principles or coordination principles he speaks of coordinative definitions [Zuordnungsdefinitionen], seen in contraposition with conceptual definitions [Begriffsdefinitionen] 10. Moreover, he already sees the problem of the choice of the best set of definitions as having nothing to do with the truth of a theory. He presents it in connection with descriptive simplicity, viewed in contrast to inductive simplicity as a principle of probability containing an important assertion about the physical world 11. In Axiomatik der relativistischen Raum-Zeit-Lehre he still considers the distinction between the different components (conventional and empirical) of scientific theories as one of the main tasks of epistemological analysis: some of these components, though, are no longer seen as constitutive principles, but simply as conventions that have the value of coordinative definitions.

34 14 PAOLO PARRINI A similar position is expressed in Philosophie der Raum-Zeit-Lehre (1928). While discussing the problem of the empirical determination of geometry, Reichenbach underlines the existence of a subjective arbitrary component in the scientific cognitive process without which he explicitly states objectivity cannot be attained. Now, though, he speaks of Zuordnungsdefinitionen, and these coordinative definitions are described in semantic-linguistic terms. If we change the coordinative definitions, we no longer constitute a different object. We simply give empirically equivalent descriptions for example, different geometrical descriptions of the physical world that differ only for the language in which they are formulated 12. What can we say about Reichenbach s relevant change of perspective? What prompted him to move closer and closer to Schlick s position according to which we only have either hypotheses or conventions? Until now, I have found no indicative references in either Reichenbach s works or in the archive material 13. So, even allowing that future studies might bring to light new documents, it seems to me that the only possible conclusion here is that neither Schlick, Reichenbach or Carnap felt the need to say even a few words about Reichenbach s change of perspective. Personally, I don t know how to interpret this silence. Is it a meaningful event or simply a twist of fate not worth any further consideration? An analysis of Reichenbach s text, though, allows us, I think, to advance a hypothesis and here we come to the main point of my paper: the re-evaluation of the role played by Russell s and Wittgenstein s ideas in the formation of what was considered the standard doctrine of Logical Empiricism, the way in which Logical Empiricism was viewed for a number of decades, and Quine s criticism thereof in his famous Two Dogmas of Empiricism. This possible answer can be explained on the basis of the fact that Reichenbach s argumentation about the problem of empirical determination of physical geometry, and the necessity to take recourse to a coordinative definition of congruence, are perfectly in tune with the ideas elaborated by the Logical Empiricists of the Vienna and Berlin Circles on the nature of logic and meaning. I am referring here to the ideas that more or less freely, and more or less legitimately, referred back to Russell s philosophy of logical atomism and Wittgenstein s Tractatus logico-philosophicus, in other words to that type of philosophy that traditional interpretations have considered, along with Mach s thought, to be one of the main origins of Logical Empiricism as it appeared on the philosophical scene between the late 1920s and the early 1930s. These philosophical ideas are the following: (i) the subdivision of statements in simple (elementary, atomic) and complex, and the dependency of the value of truth of the former on specific experiences associated to them; (ii) the tautological nature of some complex statements, in other words, the fact that they are devoid of any empirical-factual content; (iii) the argument expressed by proposition of the Tractatus ( To understand a sentence means to know what the case is when it is true ), understood by Logical Empiricists as stating (contrary

35 ON THE FORMATION OF LOGICAL EMPIRICISM 15 to what Wittgenstein wanted to say) that understanding a statement means understanding the conditions of its empirical verification. This formulation amounts to the verification principle, according to which the meaning of a statement is its means of empirical verification and implying that all those statements that cannot be verified empirically have to be considered as devoid of meaning. As I have already shown 14, both Reichenbach s final position and its underlying argumentation are far closer to Poincaré s geometric conventionalism than what Reichenbach admitted himself. What is relevant for the purposes of my paper, though, is that Reichenbach s argumentation (like that already presented by Poincaré) is explicitly posited in terms of meaning and lends itself to reconstruction taking into consideration the above-mentioned typically neo-empirical ideas. Reichenbach mentions neither Russell s nor Wittgenstein s names; he does not even quote the verification principle or the thesis of the tautological character and thus empirical and factual vacuity of certain statements. Nevertheless, when explaining why a certain assumption (the coordinative definition of congruence expressed through the elimination of universal forces) has to be seen as a definition without justifying his change of perspective with respect to the years when he considered similar assumptions as being constitutive principles analogous to Kant s synthetic a priori principles, he uses the concept of meaning and expresses his arguments by means of verificational terms and ideas. In Philosophie der Raum-Zeit-Lehre, in fact, the argument of the linguistic conventionality of the assumption regarding congruence is characterized by the following three fundamental aspects: (i) first of all, Reichenbach offers an analysis that aims at proving the absence of empirico-cognitive content (of a direct type, as I will elaborate later) in the questions regarding the geometric description of physical space or, alternatively, of the relation of congruence; (ii) secondly, he states that these questions are devoid of any sense or meaning because they are devoid of any empirical-factual content; (iii) finally, he concludes that it is only by taking recourse to suitable conventions (the coordinative ones) that we can give them meaning. In other words, it is only by means of the implicit adoption of a verificational-reductionist theory of meaning that we can qualify as devoid of any meaning a statement such as the one concerning the self-congruence or rigidity during transportation of the differentially corrected rod. Consequently, we can consider as a stipulation concerning the fixation of meaning the convention asserting the rigidity of this rod in the course of its transportation (that is the assertion of the non-existence of universal deforming forces) 15. To put it briefly, in order to justify the necessity of a convention concerning congruence and to assimilate it to a definitional or semantic-linguistic stipulation, Reichenbach first points out the lack of empirico-cognitive content of the assertion of congruence, then qualifies this assertion as devoid of meaning and, finally, states that we can talk about it in terms of truth or falsity only relative to a conventional assumption regarding linguistic rules. This argumentation seems to agree perfectly with a reductionist interpretation (which was establishing itself

36 16 PAOLO PARRINI in those years) of the two most characteristic doctrines of Logical Empiricism: the verification principle and the linguistic theory of the a priori. Reichenbach is in line with a reductionist formulation of the verificational theory for two reasons. First of all, because as was to become clear after Quine s criticism of empiricism s two dogmas he implicitly takes as a unit of empirical meaning a single assertion, individually considered. In other words, Reichenbach states that the assertion of the rigidity of the differentially corrected solid rod statement is devoid of meaning on the basis of the fact that, if considered individually, it does not have any specifiable empirico-factual content. Secondly, because he implicitly considers the absence of direct empirico-factual content tantamount to the absence of any empirico-factual content. According to the linguistic theory of the a priori, all a priori assertions, beginning with the logico-mathemathical ones, have to be considered as analytical assertions, in other words as assertions whose truth value is based only on their logical structure and the meaning of the terms used. Analytical assertions, thus, are true in all conceivable circumstances and their truth is independent of the contingent facts of the world. This is why they tell us nothing about the world and have no factual content. In the neo-positivistic interpretation, therefore, there is a link between the analyticity of an assertion and the fact that it is devoid of empirico-factual content. It seems evident to me, though, that even admitting such a link, the step from the absence of empirico-factual content of the assertion of congruence, viewed by itself, to the assertion that the corresponding conventional assumption is of a semantic-linguistic sort and so to the assertion of the analiticity or tautological nature of such an assertion once that the convention has been established is possible only by implicitly taking for granted the non-validity of the thesis from which reductionism springs. In other words, in order to pass from one assertion to the other, we need to assume that an assertion devoid of direct empiricofactual content cannot even have an indirect content deriving from the network of relationships that connect it to the other assertions belonging to the theoretical system of which it is a part. As we know, the very recognition of this fact would lead Hempel and Quine, in the 1950s, to recover Duhem s holism. They put aside any form of reductionism and defended the impossibility of formulating an empirical criterion of cognitive meaningfulness capable of making a clear distinction between a cognitively meaningful or meaningless discourse. The fact remains that, in 1928, Reichenbach discussed the problem of geometry using the verificational and reductionist terms typical of the classical neo-empiricist conception current during the active period of the Vienna and Berlin Circles. And these very terms allow him to justify the attribution to the coordination conventions of that linguistic-semantic value Schlick had defended since the early 1920s without having a fully developed philosophical justification of his position. Therefore, it is not surprising that at a certain point Reichenbach himself gave a classical formulation of neo-empiricist criticism of Kant s theory of synthetic a priori judgements. During the first International Conference on

37 ON THE FORMATION OF LOGICAL EMPIRICISM 17 Scientific Philosophy held in Paris in 1935, he presented a paper on the problem of the a priori in which we find again his criticism of Kant, but where there is no mention of the constitutive principles and all assertions are seen as either synthetic assertions deriving from experience or analytical assertions ( Tout ce que nous savons du monde est tiré de 1 expérience, et les transformations des données empiriques son purement tautologiques, analytiques l6 ). 5. INTERPRETATIVE CONSEQUENCES The path followed by Reichenbach between 1920 and 1928 (when certain ideas matured fully) and his change of perspective can help us I think to characterize the formation process of Logical Empiricism, its development and its crisis as a result of alternative theoretical perspectives. To illustrate this better, I will refer to Howard s, Oberdan s and Uebel s interpretations mentioned at the beginning. Uebel seems to be perfectly right in saying that one of the essential characteristics of Schlick s Allgemeine Erkenntnislehre was the anti-kantian use of French Conventionalism, and that such a use was in line with the desiderata of the most representative exponents of the first Vienna Circle (Frank, Neurath and Hahn). Here, though, we need to be more specific about two points. First, when using the term French Conventionalism, we must refer mainly to Poincaré s linguistic conventionalism. This regards both Schlick s position and, partly, the more clear-cut direction that prevailed within the first Vienna Circle. In Allgemeine Erkenntnislehre Schlick does not mention the names of other important French epistemologists of the same period and, more particularly, he does not seem to know Duhem s holistic conception. Frank, in his pages on the origins of Logical Empiricism and the first Vienna Circle, explicitly says that they looked at Poincaré s conception in order to account for the subjective and abstract tracts of the scientific cognitive process in order to avoid both an old form of Kantism and a form of Machian sensism. It is certainly true that among the epistemologists mentioned there, Frank also included Duhem, but with the exception of Neurath Duhem, at least initially, does not seem have had as great an influence on the development of Logical Empiricism as Poincaré 17. Secondly, the ideas formulated by Schlick in the first edition (1918) of Allgemeine Erkenntnislehre, if considered separately, were not such that they excluded an interpretation of the philosophical meaning of the theory of relativity that did not ascribe to Kant any merits beyond the negation of a naively empiricistic or radically sensistic conception of knowledge. As I said before, Reichenbach s position in Relativitätstheorie und Erkenntnis Apriori shows that Schlick s conception of knowledge as coordination could be used to negate Kant s theory of synthetic a priori judgements while still according a certain value to the idea of synthetic principles that, even though deprived of their apodeictic quality, still maintained a certain degree of constitutive value. If I inter-

38 18 PAOLO PARRINI pret Philosophie der Raum-Zeit-Lehre correctly, then we have to say that, in order to abandon this idea completely, it was necessary to adopt a general philosophical conception centered around two doctrines typical of Logical Empiricism in its classic Vienna and Berlin version: the verification principle and the linguistic theory of the a priori. This interpretation allows us to consider as still valid the traditional view of Logical Empiricism that has always ascribed great degree of anticipation to the ideas developed by Schlick in the first edition of Allgemeine Erkenntnislehre, without forcing us to say, as Oberdan feared, that virtually all the essential conceptions of Vienna and Berlin Logical Empiricism were contained in that work. Actually, in order to arrive to these conceptions it was necessary to adopt the verification principle and the linguistic theory of logical truths neither of which had been clearly formulated in Allgemeine Erkenntnislehre. All this is confirmed by two letters written by Schlick in 1926 and 1927 when he was presenting the second edition of Allgemeine Erkenntnislehre. In the first one (7 March 1926) addressed to Rudolf Carnap, Schlick moves closer to the position that Carnap was elaborating and that was fully expressed in Der logische Aufbau der Welt. When communicating this evolution of his thought, he says that Carnap and himself agreed more than one might imagine on the basis of the revisions contained in the second edition (1925) of Allgemeine Erkenntnislehre: an adequate formulation of his new ideas Schlick says would have required him to revise his work much more exstensively. In fact, after the impact of Russell s, Wittgenstein s and Carnap s ideas (Aufbau and Scheinprobleme), he changed his position and came to the conclusions expressed in the 1932 works Positivismus und Realismus and Form and Content. The second letter was addressed to Cassirer and was written one year later (30 March 1927) after the publication of Erkenntnistheorie nebst den Grenzfragen der Logik und Denkpsychologie, an essay where Cassirer also discusses the main theses of the Allgemeine Erkenntnislehre 18. Schlick responds to some of the criticism put to him by Cassirer, who in a letter dated 4 March 1927 l9 had admitted he had overlooked the changes made in the second edition of Allgemeine Erkenntnislehre. In this sense, Schlick s letter shows well the philosophical path that led him from the positions defended in that work to the ones formulated in the London lectures Form and Content (1932) where, following Russell and using the verification principle, he rejected as senseless the distinction between reality in itself and phenomenic appearence. So this letter shows very clearly the role played by Wittgenstein s and Russell s ideas on the development of Schlick s thought towards the classic formulations of the Vienna Circle that he had anticipated in the essay-manifesto Die Wende der Philosophie (1930). When his turn was complete, Schlick indicated as the starting points of his new philosophy the attention given to the problems of language, the argument regarding the tautological nature of logic and the linguistic foundation of the a priori, the representative conception of language and, finally, the verification principle which he described as the fundamental principle of philosophizing 20.

39 ON THE FORMATION OF LOGICAL EMPIRICISM 19 It is only in light of the importance given to these ideas that we can understand the process that led to the most typical formulations of neo-empiricism. In Reichenbach s 1928 work on the philosophy of space and time, these arguments are never mentioned explicitly nor do we find any mention of their connections with the problem of geometry. Even so, the argumentation in favour of the necessity of a coordinative assumption of congruence and its linguistic-semantic value seem to indicate clearly enough that it was only by means of an implicit reference to the verification principle and the tautological nature of the a priori that it was possible to abandon the idea of the constitutive a priori, as formulated in his 1920 work on the theory of relativity. Whether this was a theoretically valid move or not, whether this involved the emergence of a tension with Einstein s perspective and, finally, whether this in some ways concealed a vision of the relationship between theory and experience less defensible than Cassirer s, is a completely different story 21. NOTES D.Howard, Einstein, Kant, and the Origins of Logical Empiricism in: W.Salmon & G. Wolters (Eds.), Logic, Language and the Structure of Scientific Theories. Proceedings of the Carnap-Reichenbach Centennial, University of Konstanz, May, 1991, Konstanz / Pittsburgh: University of Pittsburgh Press / Universitätsverlag, 1994, pp ; T. Oberdan, Einstein s Disenchantment, in op. cit., pp T.Uebel, On the Austrian Roots of Logical Empiricism: The Case of the First Vienna Circle, a paper presented at the International Workshop Analytical and Continental Aspects of Logical Empiricism: Historical and Contemporary Perspectives, Florence, November 2-4, 1999, forthcoming in: P.Parrini, M.Salmon and W.Salmon (Eds.), Logical Empiricism. Historical and Contemporary Perspectives, Pittsburgh: University of Pittsburgh Press, pp. 1, 6 and 18 of the manuscript. M.Schlick, Allgemeine Erkenntnislehre (1918 1, ), English edition, General Theory of Knowledge, Translation by A.E. Blumberg, With an Introduction by A.E. Blumberg and H.Feigl, Wien & New York: Springer-Verlag, 1974, pp H.Reichenbach, Relativitätstheorie und Erkenntnis Apriori (1920), English translation with an Introduction by M. Reichenbach, The Theory ofrelativity and A Priori Knowledge, Berkeley & Los Angeles: University of California Press, 1965, pp and for the quotations; see also pp. 40, 42, 88. H.Reichenbach, Relativitätstheorie und Erkenntnis A priori, English edition cit., pp See M.Schlick, Kritizistische oder empiristische Deutung der neuen Physik? (1921), English translation in M.Schlick, Philosophical Papers, vol. I ( ), ed. by H.L.Mulder and B.F.B.Van De Velde-Schlick, Dordrecht: Reidel, 1979, pp and 333. If I understood correctly the answer Friedman gave me in Vienna during the conference, he did not seem to agree with my interpretation of Poincaré s ideas. I am highly convinced, though, that the numerous passeges I quoted in the anthology Fisica e geometria dall Ottocento a oggi, Torino: Loescher, 1979, III/7-9, clearly indicate that Poincaré gave a linguistic twist to his conventionalism. Please note that the passages quoted in section III/8 entitled Il convenzionalismo di Poincaré e la critica di Duhem: convenzioni linguistiche e contenuto teorico, testify to Duhem s criticism of Poincaré s argument according to which whatever a scientist creates in a scientific fact is the language used to express it. This aspect of the question has often been overlooked (see, for example, D. Gillies, Philosophy of Science in the Twentieth Century. Four Central Themes, Oxford: Blackwell, 1993, II/4-5); it is, on the contrary, a very important aspect of French epistemology between the late and early century.

40 20 PAOLO PARRINI On this point see my paper Reason and Perception. In Defence of a Non-linguistic Version of Empiricism, in: P.Parrini, M.Salmon and W.Salmon (Eds.), Logical Empiricism. Historical and Contemporary Perspectives cit. See H.Reichenbach, Der gegenwärtige Stand derrelativitätsdiskussion (1922), English translation in Hans Reichenbach, Selected Writings: , Vol. II, ed. by M. Reichenbach and R.S.Cohen, Dordrecht: Reidel, 1978, pp. 38 and 44. For the letters, see P.Parrini Origini e sviluppi dell empirismo logico nei suoi rapporti con la filosofia continentale. Alcuni testi inediti, Rivista di storia della filosofia, Vol. 48, See H. Reichenbach, Axiomatik der relativistischen Raum-Zeit-Lehre (1924), now in H. Reichenbach, Gesammelte Werke, ed. by A.Kamlah and M. Reichenbach, vol. 3: Die philosophische Bedeutung der Relativitätstheorie, Braunschweig/Wiesbaden: Friedr. Vieweg & Son, 1979, pp.1 (=13) and 5 (=17); English edition by M. Reichenbach with a foreword by W. Salmon, Axiomatization ofthe Theory ofrelativity, Berkeley & Los Angeles: University of California Press, 1969, pp.3 and 8. H. Reichenbach, Axiomatik der relativistischen Raum-Zeit-Lehre cit., pp. 8-9 (=20-21); English edition Axiomatization ofthe Theory ofrelativity cit., pp H. Reichenbach, Philosophie der Raum-Zeit-Lehre (1928), English edition, The Philosophy of Space and Time, ed. by M. Reichenbach and J. Freund, with Introductory Remarks by Rudolf Carnap, New York: Dover, 1958, p. 37. For even clearer statements see also H. Reichenbach, The Philosophical Significance of the Theory of Relativity (in: P.A.Schilpp [Ed.], Albert Einstein: Philosopher-Scientist, Evanston, Ill.: The Library of Living Philosophers, 1949, Second edition, New York: Tudor, 1951, pp ) and The Rise of Scientific Philosophy ([1951], Berkeley & Los Angeles: University ofcalifornia Press, 1966, pp ). Unfortunately the space at my disposal does not allow me to further elaborate on my archival work. See P. Parrini, Empirismo logico e convenzionalismo, Milano: Angeli, 1983, in particular Section 2.2. H. Reichenbach, Philosophie der Raum-Zeit-Lehre (1928), English edition, The Philosophy of Space and Time cit., Sections 3 and 4, in particular, from the final lines of Section 3, p.14, to p.16. H. Reichenbach, L empirisme logistique et ladésagrégationde l a priori, in Actes du Congrès International de Philosophie Scientifique (Sorbonne, Paris 1935), Paris: Hermann, 1936, I: Philosophie scientifique et empirisme logique, p. 32. See P. Frank, Modern Science and Its Philosophy, Cambridge, Mass.: Harvard University Press, , , pp.8-12 and In Jahrbücher der Philosophie, Vol. 3, 1927, pp , see, in particular, Sec. 3, pp The numbers indicating the day and the month have probably been reversed. Regarding Schlick s letter to Cassirer and the entire issue, see P. Parrini, Origini e sviluppi dell empirismo logico nei suoi rapporti con la filosofia continentale. Alcuni testi inediti cit., in particular pp M. Schlick, Form and Content. An Introduction to Philosophical Thinking, in M.Schlick, Philosophical Papers, Vol. II ( ) cit., pp.286, 309; see also pp I dealt with this theoretical question (discussed in the paper by Howard quoted above) in Knowledge and Reality. An Essay in Positive Philosophy, Dordrecht: Kluwer, Department of Philosophy University of Florence I Florence, Via Bolognese 52 Italy Parrini@unifi.it

41 ANITA VON DUHN BOLZANO S ACCOUNT OF JUSTIFICATION * Bolzano investigated the following problem. How can we determine whether or not a certain truth is basic without recourse to subjective criteria based on intuition or immediate perceptual knowledge? For him, the criterion of self-evidence is not a means for justifying propositions because it does not provide us with a scientific proof presenting the objective reasons for a proposition, reasons that hold independently of our knowledge. 1 Bolzano intended to provide a workable alternative to the criterion of intuitive self-evidence, and claims that we have to search for proof even of self-evident propositions at least until it becomes clear that and why no proof could be required. 2 I reconstruct Bolzano s account of justification, which is designed to replace the criterion of self-evidence and provide a scientific basis for the demonstrative sciences. I then argue that although Bolzano succeeded in devising a procedure for grounding truths, his theory fails on the account that it implicitly reintroduces an epistemological problem. 1. BOLZANO S TWO RULES AND THE GROUND-CONSEQUENCE RELATION Bolzano s position on justification is close to the foundationalist view, according to which there are basic beliefs or, in his case, basic truths which are justified without being based on other beliefs or truths and upon which other beliefs or truths are based through valid inferences. His notion of justification has a dual purpose: it must show not only that there are basic truths, but also why they are basic. It serves as a criterion for establishing certain true propositions as axioms without resorting to intuition and, in addition, provides a means for discovering truths which are not accessible by means of insight. 3 Hence Bolzano s notion of justification has two aspects, an epistemological and a logical one. The epistemological aspect is that a justification is a mark which distinguishes knowledge or true belief from opinion and the logical aspect is that a justification is a strict derivative proof which excludes all relevant alternatives as well as alien intermediate concepts. The question is, did Bolzano succeed in combining these epistemological and logical aspects, and did he provide a foundation for the demonstrative sciences? The young Bolzano (1804) provides two rules for this purpose, both with regard to mathematics: 4 21 F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

42 22 ANITA VON DUHN (1) (2) We must search for proof even of self-evident truths, at least until it is clear that and why no further proofs could be required. Even with a completely strict proof, we can allow ourselves to be satisfied if and only if it neither begs the question nor uses alien intermediate concepts or kind-crossing. The first rule is designed to eliminate the criterion of intuitive self-evidence and to solve the epistemological problem of how basic truths can be known, if not through intuitive insight. Once we have clearly grasped the primary truths, we can extend a science by deriving further and new truths from the basic ones: by examining the reasons why a proposition we hold to be true is true, we not only increase our confidence in it, but by drawing out the elements of an implicit deduction, we obtain the key to new truths which were not clear to common sense. 5 For example, the geometric proof of the truth that the angles at the base of an isosceles triangle are equal leads to the discovery of the truth that, in an equilateral triangle all angles are equal. The second rule warns against probationem per aliena et remota and kind-crossing, such as proving propositions about angles and relations between straight lines by referring to planes, or using the concept of movement to provide geometrical truths. 6 Bolzano claims that a truth is justified or scientifically proved if and only if it is shown to be objectively dependent on other truths. 7 That is to say, the relation between true propositions holds if and only if a primitive truth is the basis for a complex truth which follows from it as a consequence. 8 He introduces an axiomatic order among truths: primitive truths or axioms are at the basis of more complex truths or theorems, and these two kinds are related by a dependence relation which Bolzano calls ground-consequence (Abfolge). If a truth p is based on another truth q, then p is a consequential truth. If we can prove that p is not based on q or a set of truths Q, then we can establish that p is a basic truth or axiom. Unlike the relation of derivability, which today is called logical consequence, Abfolge is a relation that holds only between true propositions which are compatible as ground and consequence. 9 If p is based on a truth q (or a set of truths Q), p depends on q (or the set of truths Q), so p is true on the grounds of q (or Q). 10 The ground-consequence relation is irreflexive 11 because no truth can be based upon itself, and it is assymmetric 12 because grounds are simpler than their consequences. Unlike the relation of derivability, ground-consequence is an intransitive 13 relation with regard to the nearest grounds of a proposition. Bolzano claims that (i) if a proposition p is the nearest ground of a second proposition q, then q cannot be the nearest ground of p and (ii) if p is the nearest ground of q and q is the nearest ground of a third proposition r, then p is only a partial or auxiliary ground of r. 14 In addition, Bolzano distinguishes between subjective and objective grounds and consequences as two kinds of implications: (1) confirmations or truths which, as premises, produce a cognition (Erkenntnis) or follow from a cognition

43 BOLZANO S ACCOUNT OF JUSTIFICATION 23 as consequences, and (2) justifications where objective grounds and consequences hold among truths independently of our knowledge. 15 In a confirmation (Gewissmachung), our aim is to obtain the certainty that a proposition is true and we obtain what he calls a ground of knowledge (Erkenntnisgrund). In a justification (Begründung), the truth of a proposition is derived from its objective base and we obtain an objective ground. 16 We can confirm logical and mathematical truths by testing them empirically, or by repeatedly testing their derivation, but in ajustification we can have no recourse to experience. 17 In expounding upon the distinction between confirmation and justification, Bolzano follows his two rules for providing a sound basis for the demonstrative sciences; namely that we must search for proof even of self-evident truths, at least until it becomes clear that and why no further proofs could be required. This distinction is made explicit by inverting the order between the propositions which stand in a ground-consequence relation. In a justification, p is true on the grounds of q and q justifies p. In a confirmation, I know that q is true on the grounds of p and p confirms q. 18 I know that q is true on the grounds of p because p explains why q is the case, whereas in a justification, p is true on the grounds of q and q justifies p, because it explains why p is the case. With regard to the first rule and his distinction between confirmation and justification, Bolzano takes up the Aristotelian distinction between and and criticizes the latter for distinguishing between two kinds of knowing or understanding rather than between an objective justification and a subjective confirmation. 19 Aristotle distinguishes between (1) knowing that p on the grounds of q where q explains why p is the case and (2) knowing that p on the grounds of q where q does not explain why p is the case. In the first case we understand a fact through its explanation, and in the second case we become acquainted with a fact without any scientific demonstration. 20 Bolzano, however, distinguishes between objective or logical grounding of truths and their subjective or epistemological grounding: 21 (1) p is true on the grounds of q and q explains why p is the case, so qjustifies p and (2) I know that q is true on the grounds of p because p explains why q is the case, so p confirms q. He rejects the Aristotelian view which bases the distinction between Erkenntnisse or true judgments and Wissen or scientific knowledge on whether we are acquainted with their objective grounds. But this criterion for knowing something is itself a cognitive or subjective one: the distinction is not simply between unjustified and justified knowledge, but between true judgments or cognitions which we know to be justified and true judgments which we know to be not

44 24 ANITA VON DUHN justified. 22 This raises another epistemological problem, as Bolzano recognized: how do we become acquainted with the grounds of a cognition other than through intuition or insight? How do we know whether a true judgment is justified or not? Bolzano remarks that if our distinction is between two kinds of cognitions, the words grasping (Begreifen) or insight (Einsehen) would be more appropriate than knowing (Wissen), in order to qualify such Erkenntnisse as clear insights. 23 However, his own distinction is problematic as well, at least with regard to the justification of basic truths, where the epistemological question remains. I discuss this point below in section 3. A confusion between a confirmation and a justification violates the second rule that our proofs may neither beg the question nor use alien intermediate concepts or kind-crossing: If a subjective confirmation is mistaken for an objective justification due to a failure to note the reverse order of the propositions, what results is not a grounding proof. Bolzano explains that geometers did not state the correct ground of the truth to be demonstrated because they inverted the order between the propositions in the proof of the first theorem in Euclid s Elements. Consequently, some scientific proofs are not grounded proofs (justifications), but only confirmations. For example, the truth that an equilateral triangle is possible contains the objective ground of or justifies the truth that two circles intersect, but if one only pays attention to the mere knowing, then the converse relation might quite easily occur. 24 Two circles intersect or have a point in common, because for every two points a and b there must be a third, c, which has the same distance ab from both a and b, so that ca = cb = ab. We can easily mistake a confirmation for a justification, however, by inverting the order of the propositions in the demonstration. We know that an equilateral triangle is possible by confirming this on grounds of the truth that two circles intersect. 2. BOLZANO S JUSTIFICATION OF AXIOMS Bolzano examines the requirements for an axiomatic method (1804) and claims that (1) more general and simpler truths are the ground of the more particular and more complex ones and that (2) it is not enough for an axiom to be certain or self-evident. In addition, we must follow the first rule and search for a proof of it, at least until we become clearly aware of the reasons why no proof could ever be required. 25 Axioms are unprovable propositions composed of simple concepts, 26 but it does not follow from their unprovability that they are self-evident. He wants to establish axioms on logical grounds and therefore rejects the psychological criterion of self-evidence by claiming that we cannot justify the truth of a conceptual proposition by assuming that it is intuitively self-evident. For this reason, he introduces a distinction between the objective justification (Begründung) of basic conceptual truths and their subjective ascertainment (Gewissmachung), which he opposes to the traditional view that basic conceptual truths

45 BOLZANO S ACCOUNT OF JUSTIFICATION 25 are immediately obvious or intuitively self-evident. 27 We can reconstruct his claim as follows: (1) Conceptual truths, unlike perceptual truths, cannot be known (wissen) immediately or be self-evident, 28 although we can come to know them (erkennen) without inferring them from other truths but in this case we do not have a clear Erkenntnis. We come to know a truth clearly (deutlich) if and only if we are aware of its ground. 29 (2) Conceptual truths may be subjectively plausible (einleuchtend) and even certain, but we need to prove their objective grounds. 30 We can be immediately aware of conceptual truths, but we are not immediately aware of their objective grounds. 31 This notion of justifying or grounding axioms seems to originate with Leibniz, but Bolzano does not mention this. Leibniz uses his principle of sufficient reason as a call to ground axioms. 32 Even though Leibniz explicitly states the need for a grounding of Euclidean axioms, he does not elaborate a deduction as Bolzano does. For both Leibniz and Bolzano, we have immediate knowledge of the primary truths of reason, or conceptual truths, but whereas Leibniz explicitly states that such knowledge is intuitive, 33 Bolzano does not specify the nature of this knowledge, except for saying that the principle of sufficient reason is the first expression of our awakening reason (Vernunft), to always inquire about the ground. 34 Rather, he assumes that there must be primary propositions, whose truth we recognize from them themselves 35, or by understanding the meaning of their components, for otherwise we could neither make any grounded inferences, nor show (erweisen) that some statements cannot be grounded. 36 For Bolzano, a proposition cannot be both an axiom and a theorem, since axioms are not arbitrary propositions, but basic truths from which the other propositions in a theorem are derived. 37 Moreover, inverting the order of ground and consequence in a proof violates the second rule: Bolzano points out that the Euclidean proposition a straight line is the shortest distance between two points cannot be inferred from propositions referring to triangles or planes, since these latter are either proofs per aliena et remota, because they use concepts not contained in the axiom, or they are begging the question because they presuppose the basic truth rather than containing it as a component and thus inverting the order of ground and consequence. 38 This claim, and the intransitivity of the ground-consequence relation with regard to nearest grounds, commits him to saying that theorems cannot be undecidable. If a proposition p is the nearest ground of a proposition q and q is the nearest ground of a proposition r, then q cannot be the nearest ground of p, nor can p be the nearest ground of r. Since the ground-consequence relation is intransitive with regard to nearest grounds, axioms cannot be both grounds and consequences. This is problematic from a modern point of view, since some theorems are undecidable. If we admit that axioms have an arbitrary nature, rather than claiming that they are basic propositions underwriting other propositions, then any indemonstrable proposition in a theorem can function as an axiom

46 26 ANITA VON DUHN but Bolzano would reject this idea. Bolzano did, however, take some steps towards modern mathematical logic by re-defining axioms as non-intuitive propositions and by determining them through an implicit definition, through their relationships to other propositions: axioms are true propositions which are not consequences of other propositions. The problem, from a modern point of view, is that he attempts to define the nature of these propositions by defining them as basic conceptual truths, albeit on the criteria of independence and consistency. 39 How does Bolzano determine which propositions are axioms or primitive truths and which propositions are theorems or consequences? And how does he justify axioms? He determines the nature of axioms to be unprovable by showing that they are primary truths which cannot be derived from any other propositions as consequences. p is a basic truth if and only if p is composed of simple concepts, which are indefinable, and if there is no truth q or set of truths M,N,O, upon which p is based. One such basic truth is the axiom There is something. 40 Another is the principle of sufficient reason. He uses a regressive proof or backward demonstration from consequences to ground which he calls a deduction or derivation (Herleitung). 41 This is an exact derivation which eliminates redundant premises or redundant components of premises, and in which grounds must be simpler than their consequences. A proposition p is justified as an axiom if and only if we can show that for a given proposition A is B, there are no two propositions of the form A is X and X is B from which A is B could be inferred. 42 In other words, in order to justify a proposition as an axiom, we must rule out all relevant alternatives. 2.1 Bolzano s claim that the self-evidence of basic truths is not immediately obvious (einleuchtend) but must be established Bolzano s justification of axioms has a further purpose, which is to refute the claim that basic conceptual truths are intuitively self-evident and need no further justification. In 1837, he objects to the thesis held by older logicians who, instead of giving a proof for basic truths, merely state that there is a truth die für sich selbst einleuchtet. 43 Such intuitive self-evidence of a proposition is no proof of its truth. In Religionswissenschaft (1834), he says that it is easy to convince someone of the truth that it is colder in winter than in summer, merely by referring to das bloße Gefühl. Likewise, he adds, it is easy to convince anyone, even the most stupid person, of the truth that a straight line is the shortest distance between two points, merely by asking him to span a thread between two points and to let him notice that, as he stretches the thread, the distance between the two points becomes shorter as the thread takes the position of a straight line. But is this how we make him understand the reason (Grund) why the straight line is the shortest? 44 Surely not, for all we have done is to confirm the axiom

47 BOLZANO S ACCOUNT OF JUSTIFICATION 27 without providing a justification. Similarly, the drawing of a triangle confirms that the sum of 3 angles is equal to 180, but it does not prove the theorem. In 1810, he objects to the claim that the characteristic of an axiom lies in its intuitive nature (Anschaulichkeit). This property, he says, is hardly suitable to provide a firm basis for the classification of all truths into two classes, that is, into axioms and theorems. He gives three reasons for the unsuitability of intuitiveness. First, this property allows for differences in degree, so we will not be able to determine precisely what degree would be sufficient for an axiom. Second, the intuitive nature depends on fortuitous circumstances, such as our education or experience, which brings us to recognize it more frequently or less. Third, the degree of intuitiveness differs from one person to another and what is obvious for one may appear obscure to another. 45 Further on, he gives a fourth reason, which he also mentions in Clear propositions, if they are conceptual, are not necessarily intuitively self-evident, but rather, they are clear to us because they are inferred from other, more basic truths. I think that what he objects to is an inherent confusion in the word Evidenz, which may be due to the fact that it means both visibility and clarity or transparency. Bolzano distinguishes between visibility or intuitive selfevidence and the clarity or transparency of a proposition, as it becomes obvious to us. Perceptual truths can be intuitively self-evident, though even in their case Bolzano restricts the self-evidence to immediate judgments, where something is predicated on the first-person subject. 47 Conceptual truths, however, appear clear to us only because they are inferences. For this reason, he says, the proposition you should not lie may appear much clearer to us than its objective ground, the axiom you should promote general well-being, because it presupposes the latter. 48 So inferred propositions often have a higher degree of intuitiveness than axioms, which is why we are bound to confuse the order of ground and consequence in a proof. 49 For although we can be immediately aware of conceptual truths, we are not immediately aware of their grounds. In addition, he claims in the Anti-Euclid 50 that in a scientific presentation, the insight (Einleuchten) of a true proposition is not immediate, but must be established by means of derivations from simpler truths. We must be able to perceive the correctness of such an inference without seeing the figure to which it refers. In the Wissenschaftslehre 51 he states explicitly that conceptual truths are not immediately self-evident or obvious, but are testable (prüfen) either through experience or repeated derivation. A proposition is more likely to be obvious if it is inferred from others, or tested empirically. This claim, however, leads to another problem, which I will now examine, in section 3. For testing is confirming, but not justifying. It seems that Bolzano s claim that the self-evidence of basic truths is not immediately obvious but must be established, takes him back to the problem of confirming truths and results in an epistemological problem in his theory: we can logically justify truths, but we may need non-logical knowledge for having access to them.

48 28 ANITA VON DUHN 3. BOLZANO S VIEW ON CONFIRMING OR TESTING TRUTHS AND AN EPISTEMOLOGICAL PROBLEM IN HIS THEORY How do we learn or come to know that a proposition is true or that a logical relation, such as that of ground and consequence, holds between true propositions? Bolzano claims that ultimately all subjective grounds of knowledge have an objective base, 52 but how do we come to know that base? The obvious answer is, through logical knowledge or knowledge by means of concepts, by understanding the components of the conceptual propositions. He establishes axioms as conceptual truths which are unprovable but not intuitively obvious and claims that the domain of axioms stretches as far as that of pure simple concepts which are indefinable. 53 Axioms implicitly define the items referred to by concepts such as [point] or [line] and we must have an understanding of what these items are, in order to prove a geometrical theorem. Here we are faced with a dilemma: either we say that we cannot know what [point] or [line] are without knowing a geometrical theory, or we accept that we have a naïve understanding of these concepts which we then clarify by means of a geometrical theory. So even though we may assume that we come to know axioms because we have already understood their components, a problem remains, concerning the nature of this understanding. Even if we appeal to conceptual understanding, or to conventions, how do we explain our grasping the meaning of these components without an ultimate recourse to insight or intuition? Here is a possible Bolzanian reply: Bolzano says that there is a peculiar connection between the components of a subjective proposition or judgment, which must be combined with one another in a certain quite distinctive way. He is not very clear about the nature of this connection; all he says is that it must be a sort of mutual influence these ideas have upon each other. 54 Further on he explains that a judgment is not merely a sum of ideas, but a certain efficacious combination (wirksame Verbindung) of ideas. 55 Bolzano calls the degree of this efficacy the degree of confidence (Zuversicht) with which we make a judgment or the degree of certainty with which we hold something to be true. Perhaps he means that we understand a statement when we grasp its components with a sufficiently strong degree of confidence, that is, our conviction is the glue that holds the components of a judgment in a certain order: when we hold a statement to be true, we provide an efficacious combination of its components. So we understand the components of a basic truth by holding their order to be true. If the components of a proposition are determined by their order, which Bolzano holds, 56 then their meaning is determined by this order, and we understand their meaning by understanding their order and holding it to be true. It follows that for Bolzano a statement implicitly defines its components, these latter being related in a certain order, which we hold to be true. So our understanding has a rather pragmatic, if not to say dogmatic, character.

49 BOLZANO S ACCOUNT OF JUSTIFICATION 29 But this still leaves us with the problem of how we get to the primitives: Bolzano s Herleitung of axioms is a deductive procedure, whereas his regressive proof for getting to the primitives would seem to require an inductive procedure. Let us take Bolzano s above-mentioned example of a basic truth there is something : in order to establish this statement as a basic truth we must understand what there is and what something is. How can we come to know this by means of a deductive procedure? The statement is uninformative, so we cannot unpack its semantic content and lack criteria for proving that it cannot be derived from any other propositions as a consequence. The regressive procedure from consequence to ground calls for an inductive procedure which enables us to discover and thus understand the statement. Once we have discovered several truths in this way, we can then derive other statements from them as consequences. Bolzano s deductive method for justifying axioms does not provide us with criteria for determining which truths are primitive because it does not contain an explanation of how we are supposed to understand primitive truths and the meaning of their components. 57 Unfortunately there is a further problem within his own theory. Bolzano seems to say that a priori or conceptual truths cannot be justified through experience and that the objective ground of conceptual truths cannot lie in observational propositions. He also points out that a justification is possible above all in conceptual sciences, but more rarely in empirical sciences, because very often the objective reason is completely or partly unknown to us. Hence we cannot provide a justification by deriving them from their objective reasons, and in our attempts to support their proof we only increase their degree of probability. 58 Bolzano s claim that conceptual truths can be tested either through experience or by repeatedly testing their derivation 59 has provoked Husserl s and Stove s accusation that he is an empiricist who holds that we believe logical generalizations because they are true observation-statements. 60 I think, however, that their accusation is partly unjustified, for Bolzano argues, on the contrary, that we cannot generalize from observation or empirical intuition: we can come to know, by observation, that a single line can be extended, but we cannot infer from this the general truth that all straight lines can be extended. For this generalization we need to appeal to the principle of similarity. So we do not learn general truths through intuition, but through conceptual principles. 61 Bolzano s claim about our empirical knowledge of logical and mathematical truths is much weaker: he claims that we can learn or become acquainted with logical truths by experience and we can confirm such truths by testing them empirically. For example, we can learn the syllogistic form of barbara, that x is F and all F are G implies x is G, for all x and all F and all G, by testing it against experience. We discover that in all observed instances, if both the premisses of the barbara form are true, then the conclusion is true and that there are no instances of barbara where the premisses are true and the conclusion is false. This implies that our verification not only grounds our confidence in logical rules and mathematical propositions, but also the justification for the applicabil-

50 30 ANITA VON DUHN ity of these rules. And here the accusation has a point, for Bolzano s attempt to combine epistemological and logical aspects leads to an inconsistency in his account of justification. 62 However, the point here is not so much that Bolzano is an empiricist, but rather that such a combination inevitably results in problems, at least as long as it includes formal logic. To answer the question put at the beginning of this paper, I think that Bolzano s notion of justification is probably as adequate a foundation for the demonstrative sciences as you could get in the century, and its shortcomings may be due to the problem of justification itself, rather than to Bolzano s account. NOTES * This text is a revised version of a paper presented at the symposium Vienna Circle/Logical Empiricism, July 2001 and is closely related to a larger project I am presently working on, concerning Bolzano s epistemology. I would like to thank Michael Beaney and Ivan Kasabov for their critical comments. References to Bolzano, unless otherwise indicated, refer to the Bernard Bolzano Gesamtausgabe, ed. E. Winter, J.Berg, F.Kambartel, J.Louzil, B.van Rootselaar; Stuttgart Bad-Cannstatt, F. Frommann-Holzboog Verlag, 1969-(BBGA) , 3 (Betrachtungen über einige Gegenstände der Elementargeometrie (1804) in Bolzano s early mathematical works, Czechoslovak Studies in the History of Science, Prague, partial English translation by S. Russ in W.Ewald, From Kant to Hilbert, vol. 1, OUP, 1996); Beyträge II (1810), 2, 11, 12, 21 (Beyträge zu einer begründeteren Darstellung der Mathematik (1810) in Bolzano s early mathematical works, op.cit. (Beyträge). English translation by S. Russ in W.Ewald, op.cit.); 1817 (Purely analytic proof...), 1; English translation by S.Russ in W.Ewald, op.cit; WLIV, 525; (Wissenschaftslehre (1837), 4 vols, Aalen, Scientia Verlag, 1981 (WL)) and the Anti-Euklid, a manuscript in Bolzano s Nachlass edited by Karel Vecerka, Sbornik, Prague, 1967, pp , who dates the text around Jan Sebestik, however, situates the text closer to , 3. Cf. WLIV, 401, 525. Cf. on this Ali Behboud (2000), Bolzanos Beiträge zur Mathematik und ihrer Philosophie, ch, 1, Bern Studies in the History and Philosophy of Science, Bern. Erstlich stellte ich mir die Regel auf, dass ich mich durch keine Evidenz eines Satzes von der Verbindlichkeit los zähle, noch einen Beweis für denselben aufzusuchen so lange, bis ich deutlich einsähe, dass und warum sich durchaus kein Beweis fernerhin fordern lasse. [...] so muss man das Bestreben alle Wahrheiten der Mathematik bis auf ihre letzten Gründe zu entwickeln, und dadurch allen Begriffen dieser Wissenschaft die möglichste Deutlichkeit, Berichtigung und Ordnung zu verschaffen, [...]. 1804, Preface, pp.6-7. Cf. also 1810, II, 2, 1817, 1; Anti-Euklid, p , Preface, p. 8; 1810; Beyträge II, 2; WLIV, 401. Cf. 1804, Preface, pp Beyträge II, (1810), 12. cf. also 1817, 1 where he says that scientific proofs should bejustifications and introduces the distinction between Grundwahrheiten and Folgewahrheiten. WLII, 221. note. ein sehr merkwürdiges Verhältnis, vermöge dessen sich einige derselben zu andern als Gründe zu ihren Folgen verhalten. WLII, 162; 221.note: der Begriff einer solchen Anordnung unter den Wahrheiten, vermöge deren sich aus der geringsten Anzahl einfacher Vordersätze die möglich grösste Anzahl der übrigen Wahrheiten als blosser Schlusssätze ableiten lasse. Cf. WLII, 198. WLII, 204.

51 BOLZANO S ACCOUNT OF JUSTIFICATION Cf. WLII, 209. Cf. WLII, 213. Cf. WLII, 213, note. WLII, 198. ML, 13 (Von der mathematischen Lehrart ( ) in BBGA, II, A, vol.7, Grössenlehre. (ML)); WLII, 162. The distinction between confirming and justifying is not the same in the formal and the natural sciences, because in the latter, it is a causal relation which must be justified. In the natural sciences, both confirmation and justification are related to experience, but although a justification may involve an observational element, it must have a non-empirical base. In the formal sciences, only the confirmation may occur through experience. Cf. WLIV, 525; ML 13. Cf. WLII, 198, cf. also ML, 14, WLIV, 525. Aristotle, Posterior Analytics, A13, 78a-b, trsl. Jonathan Barnes (1975), Clarendon, Oxford. Cf. Barnes commentary, p In addition, Bolzano mentions Leibniz s liaison de vérités as a predecessor of his own notion of connection between truths as ground and consequence. There is a clear parallel between the two authors, since both Leibniz and Bolzano distinguish this logical relation from the epistemological connection betweeen truths as cognitions. Cf. WLII, 198, note. As Bolzano points out: the Greeks required that we only say of something that we know it, if we have come to know it from its ground. Cf. WLIII, & note. 1. WLIII, Cf. WLIV, 525; Anti-Euklid, pp Bolzano explains that geometers did not state the correct ground of the truth to be proved because they inverted the order between the propositions in the proof of the first theorem in Euclid s Elements. Therefore some scientific proofs are not grounded proofs (justifications), but confirmations. Cf. on this Paolo Mancosu (1999), Bolzano and Cournot on mathematical explanation in Revue d histoire des sciences, vol. 52-3/4, pp Cf. 1804, preface, op.cit. note 4. Beyträge,II, 19. Cf. Beyträge, appendix on the Kantian construction of concepts through intuitions (Anhang über die kantische Lehre über die Construction der Begriffe durch Anschauungen (1810) (appendix to Beyträge), in Bolzano s early mathematical works, op.cit. English translation by S.Russ in W.Ewald, From Kant to Hilbert, op.cit.; WLIII, 305.5; Bolzano s attack is directed in particular against the Kantian view that mathematical concepts are constructed through intuitions and that axioms are indemonstrable because they are intuitively self-evident. Cf. KrV, A732/B760. Cf. WLIII, Cf. WLIII, 300.9, Cf. WLIII, 305, 314; Anti-Euclid, pp Frege has a similar claim in Logic in Mathematics, (1914), The Frege Reader (1997), Michael Beaney ed., p Cf. WLIII, 313, 315.6, 316.note 1. Cf. G.W.Leibniz, Die philosophischen Schriften, vol.4, 2. Abtheilung, Berlin 1880, p Cf. on this J. Mittelstrass (1969) Die Begründung des principium rationis sufficientis, in Studia Leibnitiana, Supplementa, vol.3, pp For instance, when referring to definitions, Leibniz asserts that they include intuitive knowledge because they include a primary truth of reason. New Essays on the Human Understanding (1765), trsl. P. Remnant & J. Bennett, CUP, Cambridge, bk. IV, ch. ii, p Cf. WLIII, Bolzano qualifies our capacity to come to know a truth or make a true judgment as a feeling for truth (WLIII, 316.3, 316.note 1) through which we can come to know a truth without coming to know its ground. Cf. on this Anita Konzelmann (2001), Intuismus oder Empirismus? David Stoves Bolzano-Kritik forthcoming. Fortunately for otherwise his critical view of self-evidence would collapse Bolzano does not say that this is how we come to know primary truths. Rather, this feeling for truth applies to grounded truths, cf. WLIII, Beyträge, II, 19; WLIII, Cf. WLIII, 300.9, 316.note 1.

52 32 ANITA VON DUHN Cf. Beyträge (1810), , 10, Anti-Euclid, pp.210, ff. An axiom is independent if and only if it cannot be derived from the remaining ones and it is consistent if and only if no contradiction can be derived from it. WLII, 214. Cf. 1810, Beyträge, II, 21. Behaupten wir also, a) dass ein Satz eine Grundwahrheit sey: so müssen wir (...) darthun, dass (...) keine einzelne, auch kein Inbegriff mehrer Wahrheiten angeblich sey, aus denen er sich wie eine Folge aus ihrem Grunde ableiten lässt. (WLIV, 577). Dieses wird in den meisten Fällen eine eigene Betrachtung erfordern, die ich zum Unterschiede von einem eigentlichen Beweise (oder einer Demonstration) mit dem bestimmten Rahmen einer Herleitung (oder Deduktion) belege. Grundsätze werden also zwar nicht bewiesen, wohl aber deduciret, und diese Deductionen sind ein wesentlicher Bestandtheil des wissenschaftlichen Vortrages, indem man ohne sie niemals gewiss seyn könnte, ob jene Sätze, deren man sich als Grundsätze bedienet, dieses auch wirklich sind. (Beyträge II, 21, my italics). WLII, 214.note. RWI, 3. Lehrbuch der Religionswissenschaft, 3 Teile, Sulzbach, 1834, BBGA, I, vols.6-8. (RW) 1810 (Beyträge II, 10). Cf. WLIII, 300, ff. Cf.WLIII, (Beyträge II, 21; Appendix, 10). Cf. Appendix, 1810, 10. Cf. Anti-Euclid, p WLIII, ; cf. also Appendix 1810, 10, where he claims that one can very easily test the results of mathematics by intuition and experience and that the axiom of the straight line being the shortest between two points is proved by everyone by innumerable experiments long before they can prove it by deductions. He adds that the well-known obviousness of mathematics gradually disappears where experience is lacking. Cf. WLII, 221.note; ML 13;WL 198. Beyträge II, 22. Wenn nämlich alle mathematischen Begriffe erklärbare Begriffe wären, so könnte es auch keine Grundsätze in den mathematischen Disziplinen geben. Da es aber einfache Begriffe gibt; (...) so muss man (...) auch wirkliche Grundsätze in [der Mathematik] anerkennen. WLIII, 291. WLIII, 293. Bolzano claims that the components of a proposition are related in a certain order: a subject A has a certain attribute b. The canonical form of propositions is A has b. Cf. WLII, 126, 127. Perhaps the problem lies in the fact that he himself does not seem very clear about precisely which propositions are the basic axioms on which the demonstrative sciences are based for (to my knowledge), he makes no explicit statement to that effect. He only assumes that there must be more than one basic truth. WLII, 215, WLIV, 525. Bolzano s view implies that non-logical knowledge is necessary for coming to know the logical grounds of a proposition, at least in empirical sciences, because we cannot justify empirical laws without observational knowledge. Bolzano could object that for a justification in the natural sciences, my observation statements must be related by a conceptual relation of ground and consequence. Physical truths are truths of reason which, for their derivation, require the clear presentation of a purely conceptual truth, even if they also require a perceptual premiss. Cf. WLIII, But when we confront Newton s law that there is a force of attraction between any two particles in the universe with observation statements, we also give valid reasons for it and, in so doing, we justify it without having to recur to the logical relation between propositions. If we have not tested the truth of a proposition either by experiment, or by repeated checking of its derivation, we do not give it unqualified assent (...). Does not experience teach us that we make mistakes in mathematical judgments, and that we make these mistakes more easily the more we trust what that philosophy calls by the high-sounding name of pure intuition? (...) The

53 BOLZANO S ACCOUNT OF JUSTIFICATION only reason why we are so certain that the rules barbara, celarent, etc. are valid is because they have been confirmed in thousands of arguments in which we have applied them. (WLIII, 315.4, my italics). [Bolzano] would have firmly rejected the idea of a pure logic in my sense and, even more, in the sense of a pure logic to be clarified epistemologically. (...) No one has taken notice of the fact that Bolzano s epistemology rests upon the foundation of an extreme empiricism. (...) there is here no difference between Bolzano and his contemporary, John Stuart Mill. Husserl (1913) Introduction to the Logical Investigations, Martinus Nijhoff, The Hague, 1975, p.48. David Stove (1986) The rationality ofinduction, part 2, Clarendon, Oxford, pp Cf. ML, 14. Cf. on this Mark Textor (2001), Logically analytic propositions a posteriori?, in History of Philosophy Quarterly 18, 1, who argues that non-logical knowledge is necessary for determining the truth or falsity of logically analytic demonstratives. This fallacy in Bolzano s theory pervades even his logic. He claims that logical relations hold between true propositions independently of our knowledge. But in some cases we can only come to know that a relation of derivability holds between truths if we have non-logical knowledge. Cf. WLII, 197.3, 155. Institute for Literature Bulgarian Academy of Sciences Blvd. Shipchenski prohod, No. 52, block Sofia Bulgaria anita@ibl.bas.bg

54 DAVID JALAL HYDER KANTIAN METAPHYSICS AND HERTZIAN MECHANICS * l. INTRODUCTION People interested in the history of the philosophy of science generally know about Heinrich Hertz, the discoverer of electric waves, and the author of a work (the posthumous Principles of Mechanics) whose picture-theoretical account of science was adopted or criticised by Mach, Boltzmann, Wittgenstein, Hilbert and many others. Despite the Principles great influence, there is no consensus on what exactly Hertz was up to when he wrote that peculiar book. How did Hertz understand its significance? As a work in theoretical physics? As a philosophical treatise? If both, what is the connection between these two components? In this paper I shall argue that Hertz s Principles of Mechanics is an instance of what Kant called for in the introduction to his Metaphysical Foundations of Natural Science, and claimed to deliver in the book itself, namely a metaphysics of corporeal nature. A metaphysics of corporeal nature is an a priori construction of the concept of matter. Such a construction lays the groundwork for the unified system of science by rigorously relating metaphysical principles to the intuitions of space and time. When I say that Hertz s book is an instance of this species, I do not mean to suggest merely that one can think of it that way, for I believe that there is evidence to suggest that Hertz thought of it that way too. Some of that evidence has become available to us recently in an edition of a series of lectures which Hertz delivered at the university of Kiel in 1884, immediately before he began the series of experiments which led to his landmark works on electromagnetism. The title of the manuscript of these lectures, which appears to have been intended for publication as a book, is Über die Konstitution der Materie. 1 Most of the philosophical ideas one associates with Hertz his picture-theory, his scepticism regarding action at a distance are articulated here, and we find the same mix of scientific realism and neo-kantian idealism that makes Hertz so interesting a figure for philosophers. Hertz had been reading Kant s Metaphysical Foundations in the previous semester, and his lectures are directly concerned with the proper way of framing the concept of matter. This is in part what he takes the science of the constitution of matter alluded to in the title of the manuscript to be: a specification of the concept of matter which will prepare it for drawing up the pictures which scientific cognition requires to go about its work. Although it would be mislead- 35 F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

55 36 DAVID JALAL HYDER ing to claim that Hertz s early lectures are a simple extension of Kant s project, they still clarify the extent to which Hertz used Kant s philosophy to guide his philosophical thinking. Unlike many of his contemporaries, unlike even his teacher, Helmholtz, Hertz s debt to Kant does not stop at theses concerning the unknowability of things in themselves, or the aprioricity of time and space, but extends to his ruminations on the concept of matter and on the notion of an a priori construction of the latter. And although Hertz, even at this early stage, was inclined to reject the Kantian claim that force was essential to matter, he went about rejecting that claim, in his posthumous work, by trying to show that an a priori construction of matter could be achieved without recourse to that notion, that a complete mechanics did not require force of matter. Even where Hertz argues against Kant, he uses Kantian methods to do so. Hertz s understanding of the term matter in these lectures is not quite what one might expect. He explains what he means with a long analogy between matter and paper money, calling matter a currency issued by our understanding in order to settle its relations with things. Paper money, Hertz explains, is a sign for something else, and this alone constitutes its meaning and its value, in that it does not matter to this end how it is physically made. Hertz uses this somewhat Helmholtzian analogy to distinguish between three kinds of sciences concerned with this intellectual currency. The first of these disregards the semiotic character of matter, and considers it as a thing in itself. Such are the sciences of physics and chemistry in their everyday applications. The second kind of science is more directly concerned with the essence of matter, much like a science concerning the laws of circulation of paper money. This science, although it is conscious of the sign-character of the concept it analyses, does not need to consider specific details of the fabrication of these signs. It comprises the theoretical parts of physics and chemistry, as well as the science which Hertz is trying to situate, namely, that of the constitution of matter. The last kind of science is one concerned with the concept in itself, and is like a science which studies the fabrication of money the kinds of paper to use, the engraving techniques and the inks. This is the domain of philosophy, whose responsibility it is to descend into the workroom of our understanding, and to determine how it comes to make such concepts. 2 In his ensuing discussion, Hertz goes on to present the state of contemporary atomic, electromagnetic and ether theories, as well as attacking the concept of force and considering the claim to metaphysical priority of a number of material concepts. It is here that the picture-theory makes its first appearance in Hertz s works. I believe that these early discussions represent a thoughtful reaction to Kant, and that much of Hertz s later work is best understood from that point of view. Above all, many particularities of the structure and argument of the Principles of Mechanics are easier to understand if one approaches the work from this angle. My approach in the following is loosely chronological, for I want to argue for a continuity in Hertz s thought which has been overlooked, largely because the early work I will be discussing has been unavailable. I begin with a brief review

56 KANTIAN METAPHYSICS AND HERTZIAN MECHANICS 37 of Kant s notion of a metaphysics of corporeal nature, turning from that to a treatment of the picture-theory of scientific cognition as it appears in Hertz s Constitution of Matter and his Electric Waves. I conclude with a section on the Principles of Mechanics, where I claim that the Principles should be read as providing a science of the constitution of matter in the sense already envisaged in the Constitution of Matter, and that this is reflected in formal analogies between the Principles and Kant s system. 2. KANT S METAPHYSICAL FOUNDATIONS These connections are easiest to see when one recalls that Kant s Metaphysical Foundations represents a very particular kind of science of matter a science which is in essence concerned with constructing pictures in the imagination. Kant s philosophy of science requires such a construction of material pictures in order to bridge the gap between his transcendental philosophy and normal empirical science. Without them, we could explain neither how pure mathematics and the categories become binding in the realm of empirical experience, nor why the science of nature must be unified. In discussing both Hertz and Kant, it will be useful to keep in mind that Kant s concept of matter is exceedingly general: matter is whatever is involved in an intuition which does not belong to its form, and so sensations qualify as matter. The pure science of matter which Kant develops in the Foundations is concerned with matter in its most abstract sense, that is, as the content of any possible intuition or product of the imagination. Obviously this definition is in close accordance with Hertz s pecuniary model. In the Introduction to the Metaphysical Foundations, Kant argues that all natural science, if it is to be apodictic knowledge [Erkenntnis], and not merely so-called knowledge, must contain an a priori part. For if all the concepts and laws in a body of knowledge were arrived at empirically and inductively, the science in question would not be a rational science. And so it follows that there must be transcendental principles at the foundations of the natural sciences, principles which are on the one hand metaphysical, in that they concern the relations between pure concepts of the understanding, and on the other mathematical, because they relate to the construction of these concepts in intuition. However, since the natural sciences are also concerned with existing things, there must also be metaphysical principles which specify the kind of existence one is dealing with, which will be either the soul, for psychology, or corporeal matter, for physics. This demand entails that there be a special metaphysics for physics itself, which bridges the gap between the transcendental principles of the sciences, and the empirical laws which make up the body of physics. This metaphysics must relate the concept of matter to our intuition of space. Now, as Kant emphasises, the properties of matter which are successively investigated in the Metaphysical Foundations are necessarily empirical. Whether matter is defined as the movable in space, as that which fills a space, or as that

57 38 DAVID JALAL HYDER which has a moving force (to run through Kant s successive definitions of the concept), nothing can guarantee that things satisfying these definitions must exist. For physics is concerned with the material world, and the material world is that given to us in sensation. Concepts that have to do with possible sensation are, in turn, empirical concepts, and nothing guarantees their metaphysical, as opposed to merely logical possibility. At the same time, however, the propositions developed out of these concepts by means of metaphysical analysis are apodictic. They pronounce neither upon the various kinds of matter or forces that do in fact exist, nor upon the exact form that physical laws might take. In order that the application of mathematics to the science of bodies, which can only by this means become a natural science, may become possible, one must first develop the principles of the construction of those concepts which belong to the possibility itself of matter. Therefore one will have lay down a complete analysis of the concept of matter, which is a business of pure philosophy. It will need to avail itself of no special experiences, but only ofthat which it itself finds in particular (although in themselves empirical) concepts as they pertain to pure intuitions in space and time that is, according to laws depending essentially on the mere concept of nature. This complete analysis of the concept of matter is thereby an actual metaphysics of corporeal nature. 3 So we get the following equation: A priori principles + Empirical Concept(s) of Matter = Metaphysics of Corporeal Nature. The metaphysics of corporeal nature, in turn, when it is combined with specific data concerning the observed behaviour of natural systems will yield physics as we know it. A priori principles + (pure forms of the understanding, time, space) Empirical concept(s) of = matter (phoronomic, dynamic, mechanical) Metaphysics of corporeal nature Metaphysics of + corporeal nature Empirical laws = Natural science (Newtonian physics,...) One way to think of the place of applied metaphysics in Kant s system is to regard it as a schematism of empirical concepts. It resembles the schematism of the first Critique in its linking of the categories to pure intuition. But whereas in the schematism we move solely within the realm of the a priori pure forms of the understanding are constructed by connecting them with the pure a priori intuition of time in the Metaphysical Foundations pure empirical material concepts, such as a movable something in space, are constructed by connecting them with the intuitions of both time and space. In schematism itself, Kant contrasts the schemata of pure concepts of the understanding to schemata of sensual concepts, and explains their differing relations to what he calls pictures [Bilder]. A sensual concept is a rule for drawing [verzeichnen] arbitrary pictures

58 KANTIAN METAPHYSICS AND HERTZIAN MECHANICS 39 in the imagination. The schema of a sensual concept is the product of the pure imagination by means of which such pictures are at all possible, and which furthermore connects the pictures to the concepts they characterise [bezeichnen]. The schema of a pure concept, on the other hand, is a pure synthesis of the productive imagination according to a rule for generating a conceptual unity it can never yield a picture, for it has no sensual content. 4 To make this somewhat easier to visualise: the concept of a dog is a rule for producing various images of four-legged animals; the schema of the concept is the operation of the imagination in accordance with the rule the set of drawing operations which lead to pictures of dogs; and a pure schema is one of the operations underlying the latter, one of the operations involved in acts of drawing generally, one might say. Just as general drawing techniques do not produce a picture of anything, so the schema of a pure concept does not produce a picture in the imagination. This can only be the case once an empirical sensual concept is given as a rule. According to this division, the Metaphysical Foundations describes the schematisation of the empirical concept of matter. Such a schematisation determines the set of images of material systems which are in accordance with that concept. These are the most general images of the natural world that we can produce, and any particular representation of a physical system will be a refinement of a picture drawn from that image bank. These pictures are necessarily mathematisable, for they have been schematised within the extensive magnitudes of space and time. Furthermore, they connect to experience by means ofthe categories and pure intuition, which are the conditions of the transcendental unity of apperception, thus of any objective experience we may have, whether scientific or everyday. Finally, the requirement that all special sciences partake in the metaphysics of corporeal nature which means nothing more than that they will draw their images from that image bank entails that science will also take the form of a unified mathematical theory: each scientific theory and its images will be a more determinate, filled-in version of the basic schemes given by Kant s corporeal metaphysics. With these points in mind, let us return to Hertz s Constitution of Matter. 3. HERTZ S ON THE CONSTITUTION OF MATTER In the opening lectures of this book, Hertz makes the first documented use of his picture-theory of science. The theory is introduced to deal with a fundamental paradox confronting us when we try to think about the microscopic world, for instance when we describe heat or elasticity by means of an atomic theory. The paradox is straightforward: we cannot use the atomic conception without ascribing properties to atoms properties which, by definition, the atoms cannot have. All these properties will be taken from our experience of the macroscopic world they will be sensual properties, Hertz argues - but atoms are by definition not sensible. Every sensual representation of the atoms includes an absurdity, every

59 40 DAVID JALAL HYDER translation of sensual properties of matter onto atoms contains a logical error. But what then remains? Hertz answers that... there is always something left over once we have cancelled out all thought. There remains a system of conceptually defined magnitudes, which are connected among themselves and to the macroscopic properties of matter by means of rigorously defined mathematical relations; even if it is not allowed to take the latter at face value and to ascribe them imaginable meanings [Bedeutungen], they still retain their value as ancillary magnitudes to those relations. 5 Now, Hertz goes on to argue, one might take a very strict view of the matter, and choose to describe perceptual facts in the simplest possible manner, rejecting everything beyond the phenomena as a fiction. This is, he observes, the opinion of many physicists regarding the theory of the constitution of matter. The only constraints placed on our concept of matter from this point of view are that they be logically consistent, and that they lead to simple equations. But Hertz rejects this phenomenological approach, arguing that although physics has every right to go this road, it has no necessary superiority. For every intuition of a concept, whether in everyday life or in scientific practice, has properties that do not actually hold of things in themselves. Not only is there is no principled difference between a phenomenological approach and one in which in which we imagine microscopic entities, we have no alternative to constructing images of our scientific concepts. Even if we try to eliminate as much as possible from these images, a core will remain, which contains the essential properties with which we are concerned. [These] are not false representations, but rather the conditions of representation itself; we cannot remove them and replace them with better ones, rather we must either add them in or give up on all representations in this field. 6 Of course this immediately raises the question as to precisely which properties are essential to our concept of matter. Hertz devotes the first chapter of the second part of The Constitution of Matter to this question, treating at length the prime candidates for metaphysically necessary properties extension, movability, impenetrability, and indestructibility while referring in passing to the Metaphysical Foundations. The one metaphysical property obviously missing from this list, namely force, he treats at length in the preceding chapters, in what can only be described as a lengthy attack on the coherence of this concept. There, Hertz goes so far as to suggest that all actions at a distance, including gravitation, are best represented by means of a field theory, and that such a field should itself be understood as propagating in time. 7 Hertz characterises his discussion of the various properties of matter as the search for a fixed pole, a property which we must ascribe to matter in order for it to be suitable for picturing the world. 8 Despite the obvious parallel to Kant s undertaking (for all of these properties are analysed in the course of the Metaphysical Foundations), Hertz s result is somewhat different from Kant s. He concludes that all of these properties are a

60 KANTIAN METAPHYSICS AND HERTZIAN MECHANICS 41 mixture of empirical and a priori components, and that none of them can be taken as constitutive of our concept. He concludes with the lengthy analogy to paper money outlined above, which can only be interpreted as a form of conventionalism. The specific form we give to our concept of matter does not essentially affect the meaning of the scientific pictures which employ that concept; however, it is a grave error to project the properties that we have given it onto the things to which it refers. What is essential to these pictures, he concludes, is that they allow us to represent the mathematical relations between macroscopically observable magnitudes. It follows, a fortiori, that only the spatial and temporal relations within these pictures are essential, since only they can be the source of mathematical relations. 4. CONSTRUCTIONS AND PICTURES IN ELECTRIC WAVES As we saw, Hertz first introduced the picture-theory while considering atoms. How, he asks, and under what circumstances, are we justified in thinking about the microscopic world? The question and its answer are of great generality. But Hertz posed it at a specific moment in his research. At the time he gave his lectures on the constitution of matter, he was struggling to develop a picture of Maxwell s theory of electromagnetism which he could use to guide his experimental work. He describes that struggle in the introduction to his Electric Waves as holding between four competing pictures of electromagnetic phenomena, ranging from one based on pure action at a distance to a pure field-theoretical approach, in which all action at a distance is replaced by the polarisation of the ether, and in which the bearer or original cause of the actions at a distance (namely the electric fluid) is eliminated completely. 9 In this context, the questions he raises in the lectures concerning the status of basic properties of matter, concerning the existence of the ether and the legitimacy of action at a distance are not merely of philosophical interest. They are questions concerning the legitimacy and implications of the basic elements of the competing pictures. Hertz had to ask himself whether it was legitimate to regard electricity as a substance which exerted a force; he had to inquire into those characteristics which could be legitimately ascribed to the ether; indeed, he had to ask himself whether the very notion of an electric force was legitimate, or whether it ought instead be eliminated in favour of ether polarisation. At the same time, these questions were directly connected to the legitimacy of Maxwell s electromagnetic equations. For although Hertz, like his mentor Helmholtz, believed that Maxwell s equations were phenomenologically valid, this conviction did not tell him how to apply those equations in all cases. In the absence of a picture giving them a physical meaning, the mathematics could not be unambiguously applied. Hertz described this quandary as follows in the introduction to Electric Waves:

61 42 DAVID JALAL HYDER In my experiments, I could not... be guided by Maxwell s book directly. Instead I let myself be guided by the works of Helmholtz, as is clear from the account of the experiments. But in the special limiting case of Helmholtz s theory which leads to Maxwell s equations, and to which the experiments lead, the physical meaning of Helmholtz s theory unfortunately disappears, as it indeed generally disappears, if one wants to disregard action at a distance. I thus tried to construct to myself the indispensable physical representations without contradiction [Ich versuchte deshalb mir die unentbehrlichen physikalischen Vorstellungen widerspruchsfrei selbst zu konstruieren...] by departing from Maxwell s equations, but also by simplifying Maxwell s theory as much as possible by eliminating or simply leaving out all those elements which I didn t understand and which were dispensable, in that they could have no influence on possible experiences. 10 Hertz s method was therefore to pare down the concepts he had inherited from Helmholtz and Maxwell by constructing their physical concepts without contradiction. He presents the results of this construction in two papers from 1890, the first on the electrodynamics of stationary bodies, the second on that of moving bodies. 11 He describes the necessary intuitions of the first paper as consisting in positing a single directed magnitude for each point in space, whose changes depended only on the values of the immediately adjacent magnitudes, thereby excluding all actions at a distance. 12 In the introduction to the second of the papers, he asks whether these intuitions can also be used to model the electrodynamic appearances of moving bodies, and concludes that this question cannot be definitively answered without an understanding of the connections holding between the motions of ponderable matter and the motion of the ether. 13 In short, what he arrived at was a purified form of Maxwell s theory, in which all one has is a polarised medium. Regarding the ether, he observed that within the intuition he had developed, We regard... polarisations as the only actual existents; they are both the causes of the movements of ponderable matter and of the remaining appearances, which cause us to regard the latter as being changed. We defer an explanation of the essence of polarisation, its constitution [Zusammenhang] and its effects, or we look for it in mechanical hypotheses THE PRINCIPLES OF MECHANICS Thus the gap in Hertz s picture of electromagnetism was occupied by the ether: How are we to imagine its polarisation? More generally, how are we to conceive of what Hertz had called in The Constitution of Matter the modification of space 15 [Veränderung des Raumes] which produce those appearances we subsume under the concept of distance forces? In those early lectures, Hertz had identified this as one of the tasks of the science of the constitution of matter, arguing that if there is a change in a space [Besteht in einem Raume eine Veränderung], we cannot conceive of this otherwise than by there being something

62 KANTIAN METAPHYSICS AND HERTZIAN MECHANICS 43 which is changed, and which we can call matter. 16 To fill the gap one would need a picture of these hidden material systems. Hertz s last book, The Principles of Mechanics Presented in a New Form, 17 attempted to do just this. The Principles of Mechanics achieves its end by positing a class of unseen systems composed of hidden masses. Within a system of masses, the hidden ones are connected to the observable ones in such a way that the total energy of the system remains constant: all physical systems obey a single empirical law, which Hertz termed his Fundamental Law. Motions in which the kinetic energy of a system appears to change as a result of the action of forces are accounted for by referring the lost and restored energy to the motions of the hidden masses. In Hertz s system, all energy is kinetic and all connections between mass-points are rigid, so that no causes are required to explain why a mass-point takes the path it does. Quite aside from the physical picture it presents, the Principles of Mechanics is also structured in a most peculiar fashion. The work is divided into two books, the first of which develops a geometrical account of the motions of systems which is then applied to actual experience in the second book by means of a series of coordinative definitions and the fundamental law. The propositions of the first book, Hertz emphasises repeatedly, are purely a priori, being either definitions or following from the laws of our intuition and thought. These propositions comprise a generalised analytic geometry of n-dimensional space, as well as a peculiar definition of matter, which defines the latter as consisting of massparticles, each of which is a characteristic [Merkmal] by means of which we unambiguously correlate a definite point of space at a given time to another definite point of space at each different time. 18 In the second book, the propositions of the first book are given an experiential content by means of coordinative definitions, and the class of what Hertz calls the thinkable motions of systems described in the first book is constrained to those of natural motions, which are those motions of systems which are consistent with the fundamental law. In other words, the a priori picture of material systems, considered as collections of paths in space-time, which is developed in the first book is filled in to yield an empirical picture in the second book. This picture is a picture of physical experience as we know it so far. So we get the following equations: Generalised + geometry (Geometry of systems of points) Definitions of = material systems (mass-particle, -point, system of masses) A priori kinematics of systems of masses (Book 1) Book 1 + Fundamental law, = Coordinative definitions (Book 2) Empirical system of mechanics

63 44 DAVID JALAL HYDER Book 1 is the focus of our interest at present, for it comprises the same basic elements, and fulfils the same function as Kant s metaphysics of corporeal nature: it contains pure mathematical principles depending on our notions of space and time, and on strict phoronometric definitions of matter, i.e. definitions that appeal to nothing beyond our subjective correlation of one point in space with another at different points in time. At the same time, Book 1 of the Principles corresponds well to the philosophical part of Hertz s science of the constitution of matter. It seeks to give a form to the basic signs of our scientific pictures which makes them suited for regulating our dealings with the external world, without, however regarding them in their signifying relation, which is given only in Book 2. Taken together, Book 1 and Book 2 provide the means of picturing any system we might find in nature, so long as we can draw in the necessary details, while distinguishing rigorously between the apodictic and empirical parts of this picture. 6. CONCLUSION I said by way of introduction that even where Hertz disagreed with Kant, he used Kantian methods to prove his point. We are now in a position to see how this is so. Hertz claimed that his Principles allowed one to reconstruct the full content of mechanics without appealing to the notion of force. In doing so, he aligned himself with physicist-philosophers like Fechner and Kirchhoff in denying the claims of dynamicists, among whom one may count the young Helmholtz, who argued with Kant that the concept of force was in some sense essential to our concept of matter. As we saw, Hertz was already leaning strongly in this direction as early as 1884, when he held his lectures on the constitution of matter. To prove that this claim was false, he offered an axiomatic reconstruction of mechanics employing only a kinematic notion of matter as something movable in space, along with a priori geometrical propositions. Building on this foundation, Hertz was able to construct a picture of mechanics without appealing to the notion of force. From the point of view of Kant s Metaphysical Foundations, Hertz s Principles uses an exclusively phoronomic definition of matter, and Hertz refuted the claim that the concept of force is necessary to empirical science by completing Kant s constructive project without appealing to this concept without moving on, one might say, to the construction of either dynamical or mechanical concepts of matter as Kant himself had done. However, as Simon Saunders has argued, Hertz was deceived in his belief that his picture of mechanics was equivalent to those it was to replaced, because it failed to give criteria for physically isolating inertial frames. 19 Hertz overlooked, in other words, Kant s fundamental ground for claiming that a move from phoronomy to dynamics and mechanics (both in Kant s sense) was necessary, namely that only in making that move could experience become fully determinate and intersub-

64 K ANTIAN METAPHYSICS AND HERTZIAN MECHANICS 45 jective, and in consequence he failed to see that his alternative construction of the concept of matter his alternative picture of fundamental natural science 20 did not so much refute Kant, as vindicate him. NOTES * This paper develops the philosophical portion of a short piece on Hertz s Die Constitution der Materie that I wrote together with Heinz Lübbig (David Hyder and Heinz Lübbig, Questions Suspended in the Ether, in Nature, Vol. 404, No. 6775, 2000, pp ). Aside from a great debt to Prof. Lübbig for our discussions while working on that piece, I was given advice on my reading of Kant by Konstantin Pollok. Jesper Lutzen made a number of incisive comments on an earlier version of the present paper, and was so kind as to show me portions of the manuscript of his forthcoming book on Hertz s Principles, to be appearing under the title Heinrich Hertz s Principles of Mechanics. Along with his earlier Denouncing Force; Geometrizing Mechanics. Hertz s Principles of Mechanics. Copenhagen: Københavns Universitet Matematisk Institut 1995, these helped me to achieve a balance between science-historical and philosophical concerns Heinrich Hertz, Die Constitution der Materie. Berlin: Springer With the exception of this new and untranslated text, all citations of Hertz s works in the following are from the German edition of his collected works. The reader may be helped by the cross-referencing of the various editions, in English and German, of Hertz s works given in D. Baird, R.I.G. Hughes, and A. Nordmann (Eds.) Heinrich Hertz: Classical Physicist, Modern Philosopher. Boston Studies in the Philosophy of Science, Vol Dordrecht: Kluwer p Ibid, pp This and all subsequenttranslations from German texts are mine. Immanuel Kant. Metaphysische Anfangsgründe der Naturwissenschaft. Akademie edition IV.472, lines Emphasis in the original. Cf. Immanuel Kant, Kritik der reinen Vernunft, B Hertz, Die Constitution der Materie, loc. cit., pp Ibid., p. 36. Ibid., pp Ibid., p Heinrich Hertz. Untersuchungen über die Ausbreitung der elektrischen Kraft. Gesammelte Werke von Heinrich Hertz. Vol. 2. Leipzig: Johann Ambrosius Barth 1894, p. 21ff. Ibid., p. 22. Heinrich Hertz, Über die Grundgleichungen der Elektrodynamik für ruhende Körper and Über die Grundgleichungen der Elektrodynamik für bewegte Körper both in Hertz, Untersuchungen über die Ausbreitung der elektrischen Kraft, loc. cit. pp , Hertz, Über die Grundgleichungen der Elektrodynamik für ruhende Körper, loc. cit. p Hertz, Über die Grundgleichungen der Elektrodynamik für bewegte Körper, loc. cit. p Hertz, Untersuchungen über die Ausbreitung der elektrischen Kraft, loc. cit., p. 27. Hertz, Die Constitution der Materie, loc. cit., p. 62. Ibid. Emphasis in the original. Heinrich Hertz. Die Prinzipien der Mechanik in neuem Zusammenhange dargestellt. Gesammelte Werke von Heinrich Hertz. Vol. 3. Leipzig: Johann Ambrosius Barth Heinrich Hertz. Die Prinzipien der Mechanik, loc. cit., p. 54. See Jesper Lutzen, A Matter of Matter or a Matter of Space? in Archives Internationales d Hisloire des Sciences, Vol. 49 No. 143, pp , for a detailed discussion of Hertz s concept ofa mass-particle. Simon Saunders, Hertz s Principles, in D. Baird et al. (Eds.) Heinrich Hertz, loc. cit., pp See above all the discussion of Kant, p. 146.

65 46 DAVID JALAL HYDER 20. Cf. Hertz, Die Prinzipien der Mechanik, loc. cit., p. xxix, All physicists are unanimous that it is the task of physics to reduce all appearances of nature to the simple laws of mechanics. Fachbereich Philosophie Universität Konstanz D Konstanz Germany

66 HUBERT SCHLEICHERT MORITZ SCHLICK S IDEA OF NON-TERRITORIAL STATES 1. THE DOCUMENT In 1952, a small booklet by Moritz Schlick appeared. It was entitled Natur und Kultur, and was edited by Josef Rauscher, a former student of Schlick. Its preface identifies it as an unfinished manuscript of what Schlick had intended to become his main work. It may have been written around Schlick died in As far as I know, the booklet has not been translated into any other language. I am not sure as to how much Rauscher manipulated the text, but I think that it can basically be regarded without suspicion. The booklet features Schlick s thoughts on topics like culture and morality, the concept of fate in history, war and military service and other political problems. Although it is only a draft, and thus not ready for publishing, it shows a rather unknown side of Schlick. This side, one would be forced to say, does not really fit with the usual description of Logical Positivism. None of the key terms of Logical Positivism like meaning, verification, falsification, basic sentence, proposition, the possibility of feeling other persons pain, etc., appear in the booklet. Here we have, as it were, a Wittgenstein-free Schlick. Of course, Schlick was not schizophrenic; he simply was more of a philosopher than the standard historiography of the Vienna Circle and Logical Positivism seem to claim. 2. NEGATIVE ASPECTS OF THE STATE In the following I will only focus on a few remarks made by Schlick on the state. The term remarks must be stressed, because the whole issue was never really elaborated upon by Schlick. Schlick does not use the term social contract, but this concept obviously forms the backdrop for what he is saying. As is well known, this term is used to theoretically reconstruct what rational people could or would do in order to overcome a completely anarchic situation, a situation which usually is described as unsafe, poor, and miserable. According to this theory, rational people would agree to establish some kind of rulership or sovereignty by signing a social contract. Thus a state is created in order to guarantee peace and security. However, as Schlick writes, 49 F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

67 50 HUBERT SCHLEICHERT (1) There is nothing in our European civilization that causes more grief than the state. Under it we suffer most. Since there is nothing wrong with the original idea of the state, i.e. an association for help and protection through reasonable institutions, the mistake must lie in its execution The first question would be: Who participates in the unification? According to which principle shall the selection be made? (p. 67) This is an unusual and surprising question. According to the standard version of the social contract, such a question does not arise at all. There is no selection to be made the whole population, all the people together, agree to instate a certain government. Should somebody disagree, then at least in some constructions he would be forced to give his consent, or else he would be considered to be an enemy of the state. The state rules over a certain territory; everything and every person within it, even unborn children and babies come under the state s control. On the other side of the border, there are other states, which means potential enemies. From the very beginning, the state plays a double role. It guarantees peace within its territory and prevents civil war, while at the same time preparing for war outside its precincts, against other states. But this latter feature of every state is not discussed in the theories of social contract. Schlick makes a bitter remark about civil wars, which to Hobbes, for example, were the most terrible of all political situations: (2) Terrible as they are, one has to acknowledge the fact that civil wars usually do not claim as many victims as wars between states with separate territories have, i.e. between hostile countries. This is an argument to not separate contradictory tendencies by space, but to mix the enemies. Then the unavoidable balance will lead to only minor catastrophes, (p. 101) I will let the last sentence stand for a moment without commenting on it. However, it is probably true, as a body-count would reveal, that the number of people killed during official, regular wars conducted by states is much higher than those killed during civil wars. From this point of view, the whole ideology of a social contract is highly problematic. But let me come back to the last sentence of Schlick and his idea of separating hostile groups in space. 3. THE SOCIAL CONTRACT AND TERRITORIAL STATES For historical reasons, it is obvious that states are defined by a territory. But is this the only possibility? As Schlick says, states are constructions that are made by us and can be changed by us, and we should make these constructions as safe and rational as we can. That states start wars against others, is not contained in the idea of a social contract. It is rather a degeneration of that idea. Unfortunately, this degeneration is the most common and so to speak normal thing in history. As we shall see in a moment, Schlick looked for an alternative conception of the state in order to prevent such degeneration.

68 MORITZ SCHLICK S IDEA OF NON-TERRITORIAL STATES 51 There are still other difficulties in the usual theory; but in one respect the standard theory of a social contract comes quite close to political reality. In order to give the state a moral justification, free persons must consent to it. Yet even within contractarian theories, there is not much room for freedom of decision. There is only one state that can be established, and people can only say yes to it. Those who say no become outcasts. Once this is done, it is very difficult to revise this decision; the only way to do it would be by emigration. Practically speaking, the vast majority of all people on this earth become citizen or subjects of their state by birth and must remain in the state where they have been born. 4. AN ALTERNATIVE: SCHLICK-STATES At this point, Schlick introduces an alternative kind of social contract, which allows people to make a real choice between several possibilities, a choice that later also can be revised without great problems. He argues: (3) According to our notion of the state, it is an association with the purpose of protecting all vital necessities. This conception leaves it completely open whether the boundaries of the state i.e. the group of citizens ofthe state are determined by living together on the same territory, or by some other principle. (p. 101) It is not a priori necessary to define a state as sovereign over a certain territory. A state could also be defined as a sovereign over a certain group of human beings. In this case, several states could coexist within the same territory. Schlick tries to describe such a situation as follows: (4) Suppose that the separation according to political convictions replaces the separation by geographical states. In such a case there would be no countries in the usual sense, but political organisations, the members of which would live scattered over all continents. Each of these invisible communities could have its own laws and costumes, its courts, police, and state form. There could be invisible republics and monarchies, but the presidents and kings would not rule over territories, but only over such people as voluntarily belong to their state. Since human convictions can change, it follows from the very principle that one can at any given time move from one organization to another. (p. 102/3) Let us refer to this as Schlick s Principle and Schlick states. Clearly, all this is utopian thinking. The question is, is it an interesting utopia? Is it worth consideration? It is a theoretical reflexion about how modern, rational people, after having been through all the negative experiences mankind has had with territorial states, would or should construct their state or rather, their commonwealth of states. Unfortunately, Schlick does not provide any examples from history. Perhaps he thought that his ideas were too far removed from reality. Therefore, we have to look for examples ourselves.

69 52 HUBERT SCHLEICHERT One example that comes close to Schlick s ideas is that of the Roman Catholics. It is a large transnational, transterritorial group and everybody is invited to join the group. Theoretically, everybody belongs to this group only by his free decision, not simply by birth, and everybody is free to leave the group at any time. Theoretically, this group has its own civil law, especially for marriage and divorce, rules stating what forms of sexuality, birth control or genetic engineering are allowed or forbidden, what books are recommended and those that have to be burned, what days allowed for work, certain meat-eating days with fish as a substitute on other days, etc. Those who belong to this community are prepared to obey the whole set of these prescriptions wherever they live. This is done, more or less, without the help of the police. Even penalties are paid without control. One need only think of the exercises a believer has to do as a compensation for his sins after confession. Catholicism has a well-known transnational organization, including an absolute sovereign, and it has millions of followers all over the globe. It does not, however, claim a territory of its own, except for a very small piece of land called the Vatican state. For many reasons, of course, this is not a perfect illustration of Schlick s ideas. It is enough to mention the notorious intolerance of the Roman church. During its heyday, it never tolerated other religions on the same territory and persecuted those who wanted to secede from the one and only true church. This attitude openly contradicts Schlick s ideas. A better example for Schlick s ideas is the phenomenon called Diaspora, especially as it has been interpreted by the British historian Arnold J. Toynbee (see Toynbee 1966). The Greek term Diaspora means dispersed, scattered. It refers to communities scattered over a large area, even over the whole globe. A typical example are the Jews, yet there are many other groups, like the Armenians, or perhaps the Bahai. A diaspora is in a minority locally, wherever it may be. At the same time, a diaspora, unlike a local community, is ubiquitous (cf.p.81). Toynbee called the former Ottoman empire an association of diasporas, each of which was distributed over the whole empire and none being in exclusive possession of a local province (Toynbee 1966, p. 84). I have no idea as to whether Toynbee s interpretation is plausible. It does, however, point to an interesting possibility. The Ottoman empire, he says, was like a shot-silk robe. Due to later nationalistic ideologies, it was cut into more or less monochrome small pieces, which have been rewoven together into a coarse patchwork. If somebody wishes a more recent illustration, he need only compare a map of the Balkans from 1950 to a present-day one. The phenomenon of diaspora played an important role even in ancient times. Toynbee even writes that diasporas are coeval with the Sumerian city-states. However, as Toynbee believes, that role could become much more prominent in modern times, as geographical distances cease to be important due to modern communication technologies.

70 MORITZ SCHLICK S IDEA OF NON-TERRITORIAL STATES 53 Schlick does not want his groups or states to be defined by inborn and timeless properties like race or language, or by an intolerant religion that claims to be the only possessor of truth and salvation, but only by rational decisions of adult people. In Schlick s model, one subscribes to a state as one would to a healthinsurance company. If it does not fulfil one s expectations, one can change to another one. It might be costly, it might entail practical problems or inconveniences, but in principle it is always possible. Toynbee and Schlick shared the same basic interest in ensuring world peace. Both considered the existence of traditional local states to be something antiquated and dangerous. 5. THE MINIMAL WORLD STATE Schlick s plurality of states on the same territory, even if we remain on an utopian level, is riddled with numerous practical problems. He nevertheless thinks that they can be solved: (5) However, such a situation would only be stable, if there were specific rules determining the mutual relations between the members of the different organizations. (I purposely do not say: between the organizations themselves.) One should have agreed upon a minimum of supranational or international law. If you like, one could say that this amounts to establishing a world state This world state, however, would be very diluted ; it would be constituted by very simple rules It would by no means be difficult to establish such global rules. (p. 103) I think it would by no means be easy, but it would be possible. Such a world state could not, for example, be dominated by one group or group-state, as was the case with all former multiethnic empires like the Ottoman or the Habsburg empire. In fact, in every federal system the organization of the central power and its relation to the single states is of the utmost importance. But let us leave this aside. What Schlick envisages is a kind of United States of the World, a federal system of government. Only the single states are not to be defined by geographical borders, but by people who by their free decision (and not just by birth) become citizens of their respective states. These people can live close to each other in a given space, or may be scattered over a vast area, like the Jews in a diaspora. Actually, Schlick prefers this case to the concentration of a group in one country or region. If a group were not concentrated in one territory like traditional nations, it would reduce the possibility of aggression towards other groups. Citizens of a Schlick state would share a certain common feeling and have a kind of basic sympathy towards each other, because they would all share the same basic political and moral ideals. Therefore it is a rationally justified sympathy, whereas traditional patriotism, i.e. sympathy for other people only

71 54 HUBERT SCHLEICHERT because they happen to be of the same nationality, is not very rational. In our real world, the difference or discrepancy between nationalistic, territorial loyalty and the loyalty of diaspora-people toward their respective diasporas has often been noticed. It was, incidentally, one of the standard arguments for anti-semitism, especially during wars. What is true is that for an aggressive state, a narrowminded local patriotism is much more useful than loyality towards an international community living in diaspora. Schlick speaks about a world state. This would be the simplest solution. In Schlick s states without territory, no territorial conflict could ever arise. Even on a less utopian scale, Schlick s idea makes sense. We may think of former Yugoslavia, which had clear borders and no territorial conflicts with its neighbours. According to Schlick s principle, it was a mistake to break up this territory and hand it over to every nationality in a certain geographical area. Those who think that their nationality is the most important thing may form a state, but not a territorial one, and elect their own rulers and judges and educational system, etc. If they like, they can completely ignore other groups. But others, who have different political ideas, may form other political groups or states and live on the same territory as well. No group should be allowed to dominate another. No group has a right to any special aid or support from the common minimal state, with the exception that the safety of each group is guaranteed. 6. DESCRIBING SCHLICK-STATES IN DETAIL Schlick states should be made up of humans having the same political ideals, regardless of their language, colour or place of birth. Let us try to be more detailed. There could be a very socialist state and an extremely capitalistic one neighbouring each other. It is easy to imagine different groups sharing the same territory, and yet having quite different civil laws. There could be a state where divorce, birth control, abortion, human cloning, and euthanasia are allowed. This state could share its territory with a state that obeys the Pope. There could be a state where polygamy is the standard of family life, and even a Platonic state I mean a state of homosexuals. There might be quite different practices for healthinsurance and pensions, for education and culture. Different economic systems for different people, even different monetary systems within the same territory or across the whole globe are not impossible, somewhat like different telephonecompanies. Everybody could choose the state in which he wanted to live. Those who like it, might be governed by a fascist state, or live under a religious dictatorship, while others would prefer a system of direct (i.e. non-representative) democracy. There could be multiethnic states and those with only a single ethnic group, all on the same territory. The main condition would be non-interference and free movement between the states at any given time. There could and should be peaceful competition between states in order to win more citizens, like a competition to win customers in a free market.

72 MORITZ SCHLICK S IDEA OF NON-TERRITORIAL STATES 55 I must admit that there is a certain element of anarchism in Schlick s conception. But this serves to reduce the danger that goes with all traditional territorial states. It could also be an antidote to unlimited globalization. Schlick states are clearly related to political parties in democracies. One feels some sympathy for a certain party, but one can at any time switch to another party or even try to found a new one. The main difference is, if a party is in the opposition, its followers have to obey the law made by the ruling party. They have no chance of living as they see fit. It is quite possible for 49% of the population in a democracy to have to live according to the idea of a good life as is determined by the remaining 51 %. 7. A WAY TO PEACE. NATIONALISM The starting point of Schlick s utopian construction was the question of how to reduce the probability of wars between states. One of the main causes for conflicts and wars are quarrels and fights over territorial claims. If a state is not defined by a territory, territorial questions disappear. It is worth considering an application of Schlick s principle to calming or solving conflicts that take place between more or less hostile groups within a state. According to Schlick, such problems should not be treated by separating groups, with each group in its own closed territory. Rather, one should mix them. Here he offers a combination of practical and moral arguments: (6) Again, the essential presupposition is that the members of the groups live intermingled, for as soon as there is a territorial separation, new interests and complications arise. (p. 103) Attempts at secession and isolation prevent peace and the development of an international morality. Morality is always a product of living together. (p. 107) It is interesting to note that already Voltaire, the great philosopher of Enlightenment, proposed while writing on religious tolerance, a multiplication of the number of groups, in order to eliminate or, at least, reduce aggression and intolerance between hostile groups. In his famous Philosophical Dictionary, in the article on tolerance (in part I), he wrote: If you have two religions in your midst, they will cut each other s throat; if you have thirty, they will live in peace (Voltaire 1972, p. 390). Voltaire s favorite example was the peaceful development of Amsterdam, a city in which many different religions coexisted. Schlick did not speak about nationalism. But one can ask if the same methods could be used to reduce hatred, aggression and nationalistic hysteria. It is really tempting to interpret Schlick s ideas in this way, i.e. to take them as a suggestion for solving problems of nationalism. It is obvious that he would have rejected any kind of ethnic cleansing ; I think he also must have been against what was going on after the First World War, namely the establishment of all the national states in Eastern Europe. Schlick saw the problems clearly:

73 56 HUBERT SCHLEICHERT (7) Living together on the same territory, taken to be the principle of belonging together, gives rise to all such evils which dog our divided world the most. (p. 97) Today, the most realistic examples of several groups living together on the same territory are those of areas with ethnically mixed populations. The world is full of such places. Unfortunately, most of these examples do not serve to further Schlick s ideals. If we look at former Yugoslavia, we are presented with a shocking counterexample, which makes Schlick appear extremely naive. As he lived in Vienna, the former capital of the multi-ethnic Habsburg empire, Schlick must have known of enough discouraging examples of several nationalities living together within the same territory. In fact, this empire broke into pieces because of nationalistic problems. Schlick cannot have been so blind. The only way to peace seems to be a complete separation of nationalities, a division of the countries into several ghettos, each surrounded by very high fences, like wild animals in a zoo. In the Middle East one can often hear the slogan high fences make good neighbours, but what is true is that they do not make neighbours at all. Schlick had quite a different view of politics. (8) What is so unnatural about our states are their boundaries. Every borderline in space is unnatural, because there is never any plausible reason for calling something good which the other side calls bad. (p. 104) And Schlick repeats that separation, secession, isolation make it impossible for any kind of international morality to materialize. Morality grows out of living together. The more removed you are from others, the higher the fences are, and the less you learn to think about the others feelings and problems and to show consideration for them. There are a few examples more convincing than that of the Balkans. One could mention the island and state of Mauritius. As an island, it has no territorial problems with other states. And it is inhabited by a lot of different nationalities: Hindu, Tamil, Creole, Black, Indian Muslim, Chinese. At least till now, it has been a peaceful state. Another example, on a larger scale, can be taken from India. Here there are different religious groups living more or less within the same territory. But each group lives under its own specific laws, especially a personal law. Theoretically, people can also change their religion and move from one group to another. There is, of course, the federal state of India, regulating conflicts between the groups, which is a difficult and complicated task. And it is not clear how stable or unstable the whole construction is. Over the last few years it has often been said that there will never be peace as long as each nationality claims its own territorial state. How many territorial states must be established until there is finally peace in the Balkans? It is well known that most governments are against secessionist movements. When this question arises, the international community of states shares the attitude of the

74 MORITZ SCHLICK S IDEA OF NON-TERRITORIAL STATES 57 Roman church towards divorce: it is of no importance whether partners love or hate each other; the only separation that is allowed is by death. Allen Buchanan, one of the prominent authors on secession, is not quite as rigorous, but he also insists that nations should have a right to self-determination only under very restricted conditions e.g., if otherwise they really are in danger of being exterminated. But as such, i.e. just because it is a nation, no nation should have the right to self-determination. (Buchanan 1996 p. 298) 8. THE RIGHT TO SELF-DETERMINATION Can we put aside the right of national self-determination, including the right to secession, so easily? Is it not a fundamental human right? Does it not follow from the basic ideas of democracy? Who is entitled to tell nations what they have to do, and what they must not do? Who could call himself a good democrat, and yet object if the majority of a population in a plebiscite decides to secede from their present state and establish a new one, usually a smaller one? Yet the right to self-determination does not have a very good reputation. Even philosophers prefer to look at states as if states, once established, are eternally fixed. But this contradicts the most basic principles of liberalism. It follows from these very principles that any group should be given the right to secede from a state if this group does not want to live in it any longer, and if it wants to establish a new state. But even the most liberal writer (Beran, 1984) found it necessary to make some restrictions, saying that a group must be territorially concentrated if it is to be given the right to self-determination. Why? Because every secession causes territorial problems. Think, for example, of the USA: they could not easily be divided into a white state and a coloured state, because both groups would have to claim the whole territory and, as this author writes (Beran 1984, p. 24) it is not practicable at present for two states to share legal sovereignty over the whole territory. Nearly all arguments against a liberal interpretation of self-determination refer to territorial questions. Schlick wanted to introduce his own ideas at exactly this point. I think that Schlick would have agreed with all modern critics of an unrestricted, unlimited, universal right to self-determination, as long as this right means what it usually means namely, territorial secession, new borders etc. If two nationalities want to exterminate each other, and each wants to occupy the whole territory alone, then according to Schlick one should not separate them. Territorial separation is the favoured method of every nationalism, but if all possible conflicts were to be solved simply by separating the hostile parties, our globe would soon be too small. Besides, it would create endless territorial claims and wars. It is because of these bitter, practical experiences that today many political philosophers deny the right to separation and secession. (See e.g. Moore, 1998). The biblical procedure if you go right, I shall go left, if you go

75 58 HUBERT SCHLEICHERT left, I shall go right is a negation of morality; but what is worse, it will not lead to peace, because to the left and to the right are already other people. Moritz Schlick s model offers quite a different solution. First, from a moral point of view the right to political self-determination should be granted to any nation. But Schlick s perspective is much wider than that. To him, the right to self-determination should be acknowledged to any group, not only to those that define themselves as a nation. What s so special about nations?, that they can ask for self-determination, asked Allen Buchanan (Buchanan, 1996) in a polemic argument. Schlick would agree, but draw a different conclusion. Schlick s conclusion would be that not only nations, but any group of adult human beings must be given the right to self-determination, including the right to secede from an existing state and try to create a new state. Actually, history knows only few secessionist movements which were not nationalistic. But the first movement to have been called a secession was a social, not a national movement. Thrice (494, 449, 287 BC), so the story goes, did the Roman plebeians withdraw themselves from the city-state of Rome in protest against the behaviour of the upper classes. This behavior received the Latin name of secessio. During the first secession, in 494 BC, the plebeians literally left the city and went to a hill called mons sacrum outside the urban territory. In Schlick s model, self-determination does not lead to chaos or war, because self-determination does not include an exclusive right over a certain territory. In contrast to territorial states, Schlick-states can be founded, changed or abolished without disturbing peace and justice. All groups or Schlick-states must live on the same territory, and no Schlick-state is allowed to dominate another. Each group gets its own government, but no group has a territory of its own. Only individuals, if anyone at all, are landowners. As no political group, no Schlickstate, governs a territory they are all forced to live together on the same globe, or island, or continent. So they must find some sort of a modus vivendi. 9. CONCLUSION We should not forget that Schlick s ideal states are not defined as nations or nationalistic groups, but by groups that share some more abstract, political or moral principles. He writes: (9) Will I not prefer a thousand times more to cooperate with a reliable Chinese of good character than with an egotistic, insincere European? (p. 99/100) And by cooperation, Schlick means political cooperation. Let me summarize. Schlick anticipated, perhaps not fully consciously, an alternative solution to one of the most urgent problems of political philosophy of our own time. His

76 MORITZ SCHLICK S IDEA OF NON-TERRITORIAL STATES 59 proposals are even more progressive than most philosophers are today. He developed a model for practically unlimited self-determination which does not lead to all the problems associated with nationalistic territorial self-determination. In a nutshell, it also points to a reconciliation between globalization and individual political freedom. It might be surprising to find such utopian thoughts coming from the central personality of Logical Positivism. But if so, it means only that the usual, by now canonical historiography of the Vienna Circle is too one-sided. It could even mean that Schlick s own interpretation of philosophy, as given in the first article of the journal Erkenntnis (Schlick 1930), was not a complete description of what he himself did. But all this does not present any great difficulty. There are, however, a few words in Schlick s booklet which really are confusing within the framework of political philosophy. The text, in two different places (pp and 107), makes use of the religious term civitas dei, taken from St. Augustine. There Schlick states that race, religion, political conviction or profession are all not so important as a basis of a natural state, whereas (10) the only reliable basis is the character of humans, their moral qualities (not their convictions ). People of good character, the kind and peaceful, belong together by nature ; they form the invisible state of god, civitas dei. (p. 99/100) Belonging together in this context can only mean: they should establish a state for themselves. And once more we read: (11) Only good will can be the ultimate principle of unification; the state that is established in this way is the true state of god, civitas dei (p. 107) Schlick here leaves it open what shall happen with the rest of mankind, with the vast majority of people, those who are of only average or less than average moral quality. I have no idea how to integrate such pseudo-religious, unpolitical remarks into Schlick s political ideas. Maybe something went wrong during the editing, maybe Schlick wanted to make a stand against Kant s statement that even a population of devils would be able to have a state, maybe it is only a shorthand for something more complex, or maybe Schlick was too much under the influence of Wittgenstein. I am unable to solve this riddle. So I shall leave it, together with my best wishes for the next ten years, to the Institut Wiener Kreis.

77 60 HUBERT SCHLEICHERT REFERENCES Beran, H. (1984): A Liberal Theory of Secession, Political Studies 32, p Buchanan, Allen E. (1991): Secession: The Morality of Political Divorce from Fort Sumter to Lithuania and Quebec. Boulder, Color., Westview Pr. Buchanan, Allen E. (1996): What s So Special About Nations? in Couture, Nielsen, and Seymour (eds.): Rethinking Nationalism, Univ. of Calgary Pr., p Moore, Margaret (edit.) (1998) National Self-Determination and Secession, Oxford UPr. Schlick, Moritz (1930): Die Wende der Philosophie, Erkenntnis Bd.1, p Schlick, Moritz (1952): Natur und Kultur. Aus den Nachlaß des Autors herausgegeben von Josef Rauscher. Humboldt-Verlag Wien-Stuttgart pg. (Sammlung die Universität, Bd. 30). All quotations are translated from this edition; page numbers refer to this edition. Toynbee, A.C. (1966) : Change and Habit. The Challenge of Our Time. Oxford UPr. (German translation: Menschheit woher und wohin? Plädoyer für den Weltstaat. Stuttg ) Voltaire (1972): Philosophical Dictionary, transl. Th. Besterman, Penguin Classics. SCHLICK-ZITATE (1) In unserer ganzen europäischen Kultur ist nichts, was dem Menschen größere Daseinsnot bereitet, als der Staat. An ihm leiden wir am meisten. Da nun an der ursprünglichen Idee des Staates Zusammenschluß zu Schutz und Hilfe durch vernunftgeschaffene Institutionen nichts auszusetzen ist, so muß der Fehler in der Durchführung der Idee liegen Die erste Frage wäre: Wer nimmt denn an der Vereinigung teil? Nach welchem Prinzip findet die Auswahl statt? (S. 67) (2) Bürgerkrieg So schrecklich solche Ereignisse auch sind, so ist doch zu bemerken, daß sie ungleich weniger verlustreiche und blutig zu verlaufen pflegen, als Kriege zwischen räumlich getrennten Staaten, also zwischen feindlichen Ländern. Dies spricht wiederum dafür, daß man entgegengesetzte Tendenzen, nicht auch räumlich voneinander trennen soll, sondern die Gegner untereinander mischen. Dann erfolgt der unvermeidliche Ausgleich in geringeren Katastrophen. (S. 101) (3) Unser Begriff vom Staate ist: Zusammenschluß zum Schutz aller gemeinsamen Lebensnotwendigkeiten. Bei dieser Definition bleibt es gänzlich offen, ob die Grenzen des Staates das heißt: der Umkreis der Bürger, die zu ihm gehören durch örtliches Zusammenwohnen bestimmt werden,, oder ob die Trennung durch ein anderes Prinzip erfolgt. (S.101) (4) Stellen wir uns einmal vor, daß die Trennung nach politischen Überzeugungen an die Stelle der Einteilung in geographische Staaten träte. Dann gäbe es keine Länder im üblichen Sinn, wohl aber politische Organisationen, deren Mitglieder durch alle Erdteile und Gegenden zerstreut ihre Wohnsitze hätten. Jede dieser unsichtbaren Gemeinschaften könnte ihre eigenen Gesetze, ihre eigenen Sitten, ihre eigene Rechtspflege und Exekutive und auch ihre eigene Staatsform haben. Es könnte unsichtbare Republiken und Monarchien geben, aber die Präsidenten und Fürsten würden nicht über Territorien herrschen, sondern nur über Menschen, die ihrem Staat freiwillig angehören. Denn da die Über-

78 MORITZ SCHLICK S IDEA OF NON-TERRITORIAL STATES 61 zeugungen des einzelnen sich ändern können, so liegt es im Prinzip der Sache, daß der Übertritt von einer Organisation zur anderen jederzeit stattfinden kann. (S. 102/3) (5) Allerdings wäre eine solcher Zustand nur dann von innerer Festigkeit, wenn es auch besondere Regeln für das gegenseitige Verhältnis zwischen den Mitgliedern verschiedener Organisationen (ich sage absichtlich nicht: zwischen den Organisationen selbst) gäbe; man müßte sich also auf ein gewisses Minimum von überstaatlichem oder interstaatlichem Recht geeinigt haben und, wenn man will, kann man sagen, daß dies eben auf die Konstitution eines einzigen Weltstaates hinausliefe Der,,Weltstaat wäre unter den gedachten Umständen sehr verdünnt, er wäre durch relativ sehr einfache Regeln konstituiert, Es wäre durchaus nicht schwer, derartige Weltstatuten aufzustellen. (S. 103) (6) Die wesentliche Voraussetzung ist allerdings wieder, daß die Mitglieder der Gruppen miteinander vermischt leben, denn sowie eine räumliche Trennung eintritt, entstehen neue Interessen und Komplikationen. (S.103). Streben nach Abtrennung, Isolierung verhindert die Entwicklung eines Zustandes friedlichen Zusammenlebens, es verhindert die Entstehung einer Völkermoral. Sittlichkeit ist immer das Produkt des Zusammenlebens. (S. 107) (7) Wenn das räumliche Beisammenwohnen zum Prinzip der Zusammengehörigkeit gemacht wird, so entstehen alle jene Übel des Gegeneinander, unter denen die in Staaten zerstückelte Welt am meisten leidet. (S. 97) (8) DasUnnatürlichean unseren Staaten sind ihre Grenzen. Jede räumlichegrenze ist unnatürlich, weil es nie einen vernünftigen Grund gibt, daß auf der einen Seite gut heißen sollte, was auf der andern schlecht heißt. (S. 104) (9) Werde ich nicht tausendmal lieber mit einem Chinesen gemeinsame Sache machen, den ich als zuverlässig und von gütiger Gesinnung erkannt habe, als mit einem Europäer, der unaufrichtig und selbstsüchtig ist? (S. 99/100) (10) die einzig zuverlässige Grundlage ist der Charakter der Menschen, ihre ethischen Eigenschaften (nicht,,überzeugungen ). Die Menschen von Charakter, die Gütigen und Friedfertigen, gehören,,von Natur zusammen, sie bilden die unsichtbare Civitas dei (S. 99/100) (11) Der gute Wille allein kann letztes Prinzip des Zusammenschlusses sein; der so gebildete Staat ist die wahre Civitas dei (S. 107) FB Philosophie UniversitätKonstanz, Fach D17 D Konstanz Germany hubert.schleichert@uni-konstanz.de

79 MASSIMO FERRARI AN UNKNOWN SIDE OF MORITZ SCHLICK S INTELLECTUAL BIOGRAPHY: THE REVIEWS FOR THE VIERTELJAHRSCHRIFT FÜR WISSENSCHAFTLICHE PHILOSOPHIE UND SOZIOLOGIE ( ) Moritz Schlick s intellectual biography constitutes a very interesting or more precisely unavoidable issue for the scholarship of the history of logical empiricism. Nevertheless, it still lacks extensive inquiry and has seldom been studied 1, whereas in the last years some research on the leading figures of the movement such as Neurath, Carnap and Reichenbach has begun. At least count, there is a lot of material which will enable us in the near future to satisfy the desideratum of biographical reconstruction 2. And yet this exciting argument by no means signifies simply archeological research. On the contrary, Schlick s scientific and philosophical development represents a paradigmatic aspect of the Vienna Circle: it exemplifies in the best way both the intricate roots of the scientific worldview and the relationship between continuity and discontinuity within the history of logical empiricism. Similar research is particularly illuminating when we consider Schlick s intellectual adventure before Vienna 3, namely the long period ending with the publication of the first edition (1918) of the General Theory of Knowledge 4. The terminus a quo of Schlick s early theoretical philosophy may be considered his Habilitationschrift of 1910, which concluded his development of thought from the apprenticeship as a physicist at the school of Max Planck until he began to devote himself to philosophy, in particular to ethical problems 5. Now, it is especially our knowledge of Schlick s philosophical biography in the years from his call to the University of Rostock (1911) to the conclusion of his opus magnum that still seems to lack many details, also as a result of the fact that Schlick published few contributions in this period. Schlick s philosophical work after the publication both of the Habiltationschrift devoted to The Nature of Truth in Modern Logic and of the nearly contemporaneous study on The Boundaries of Scientific and Philosophical Concept-formation 6 is above all documented in four essays apart from the unpublished lectures he gave in Rostock, only partially available in English translation 7, namely the papers on intuitive knowledge, on the philosophical significance of the principle of relativity, on the ideality of space and, finally, on the epistemological consequences of the general theory of relativity for the concepts of space and time 8. To be sure, these studies are of F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands. 63

80 64 MASSIMO FERRARI greatest importance for an understanding of Schlick s early epistemology as well as of the background of the General Theory of Knowledge. Nevertheless, they are not the only available ones giving an account of Schlick s philosophical framework during these years. From 1911 to 1916 Schlick wrote a number of critical reviews which appeared in the Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie : more precisely, some 30 reviews, which, to the best of my knowledge, have not been cited in any bibliography of Schlick, although they deserve careful analysis. Indeed, Schlick s activity as contributor to an influential German philosophical journal offers not only an exhaustive survey of his philosophical and scientific opinions, but also the opportunity to trace Schlick s intellectual development within both the historical context and the philosophical debates of his time. Schlick had a good relationship with the review founded in 1877 by Richard Avenarius and later edited by Paul Barth with the collaboration of the neo- Kantian Alois Riehl. Schlick s papers of 1910 as well as the essays of 1913 and 1916 quoted above appeared in this tribune of scientific philosophy, which sought at a kind of harmonization of positivism and neo-kantianism. 9 Moreover, as we can see from some of the letters Riehl addressed to Schlick, Schlick had been well acquainted with Riehl since 1910 and had an opportunity to visit him in Berlin before he moved to Rostock 10. Riehl was, on the other hand, one of the thinkers Schlick had studied more intensely in the years of his training, especially as regards Riehl s fundamental work Der philosophische Kritizismus, the main outcome of which was a realistic interpretation of the Kantian theory of knowledge which certainly influenced Schlick s conception of critical philosophy 11. No wonder, then, that at the very beginnig of his academic career Schlick contributed to the philosophical journal edited by Riehl and published a series of reviews in the Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie. Schlick s attitude as reviewer is based on the conviction that the most important aspect is to reach an understanding about matters of principle despite all the different points of view. 12 However, this doesn t imply that Schlick s reviews lose their strength or do not take a critical position. The examination of authors such as Natorp, Enriques, James, Frischeisen-Köhler, Dingler, Driesch or Cornelius, in particular, makes it clear that Schlick grounded his examination of the main issues of epistemology in his own philosophical achievements and, to some extent, on the point of view he later endorsed in the General Theory of Knowledge. Apart from some reviews which today have only bibliographical significance 13, documenting, for example, Schlick s interest in the history of ancient philosophy 14, it seems useful to divide this unknown aspect of his philosophical activity into thematic areas. Schlick reviews a number of books and contributions dealing for the most part with problems of psychology, logic, epistemology and philosophy of science; but from the philosophical point of view he pays special attention, on the one hand, to both the problem of realism and the critique of the so-called empirio-criticism 15, and, on the other, to the

81 AN UNKNOWN SIDE OF MORITZ SCHLICK S INTELLECTUAL BIOGRAPHY 65 critical interpretation of Kant s philosophy within the more general framework of an enduring discussion of Neokantianism. From these quite different reviews Schlick emerges as a typical supporter of the scientific philosophy in the German culture at the beginning of the century; and so it is very well understandable why Schlick became a regular contributor of the philosophical journal founded by Avenarius, playing a role of some importance in the contemporary debates about both epistemological questions and puzzles posed by the developments of science to the scientific worldview. 16 But it is also worth noting that Schlick discussed authors and epistemologic-philosophical problems which are at the core, in the period from 1907 to 1917, of the activity of the so-called first Vienna Circle, as is easy to see from the many reviews of Mach, Dingier, Duhem, Enriques, Zilsel, Rey, Planck, Stallo and Poincaré written by Philipp Frank and Hans Hahn 17. Schlick, for example, introduced the readers of the Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie to the work of Paul Volkman (Erkenntnistheoretische Grundzüge der Naturwissenschaften) and, later, to Hugo Dingler s Grundzüge der Naturphilosophie, although he criticized both the inadequacy of the way Volkman dealt with physical problems and Dingler s unrigorous argumentation 18. Even more important, though, is Schlick s review of Federigo Enriques main work, which was published in German in 1910 as Problems der Wissenschaft. To be sure, Schlick admired the encyclopedic outlook of contemporary sciences offered by Enriques in his valuable book; nevertheless, he raised a fundamental objection to the relativism endorsed by Enriques on the basis of his positivistic point of view, namely against the idea according to which nothing absolute exists, it being impossible to put an end to any infinite series of relative elements or degrees. But Schlick argued that this argument has no validity, for the good reason that it contradicts the spirit of critical philosophy by concluding from the failure of a definition the nonexistence of the object to define 19. Schlick thus rejected the fundamental point of Enriques theory of knowledge and he praised, on the contrary, the other parts of his work (for today s reader, indeed, the most questionable), which offer a survey of logic, methodology and scientific development up to the present 20. Let us now turn our attention, first of all, to Schlick s discussions of psychological books. It should be noted here that Schlick seemed interested in the classical topics of contemporary psychology, for example in the theory of feelings and reflexes 21 ; but he was altogether attracted by the recent results of Gestaltpsychologie, and in his opinion the concepts of quality and form allow a broader potential for application than is usually assumed 22. On the other hand, Schlick engaged himself in the defence of the scientific character of psychology against the unjustified demands of traditional philosophy 23. This tension between philosophy and psychology is the main topic dealt with in the review of William James s Psychologie, that is of the German, abridged edition of his main work. Schlick considers it an excellent handbook, having great merits particularly in its argument on the basis of real experience in everyday life, although this by no

82 66 MASSIMO FERRARI means is intended to damage the scientific nature of the book. Nevertheless, Schlick adds a critical and meaningful remark: The philosopher James, he says, fades quite into the background; and this is to the best advantage of the work 24. Here, it is not difficult to recognize Schlick s polemical assessment of James s philosophy which he had already expressed in his Habilitationsschrift where he completely rejected James s unacceptable concept of truth inasmuch it is essentially characterized by its mutability 25. This objection represents, of course, the last consequence of Schlick s general critique of pragmatism, the unscientific view of which to him seemed destined to collapse because of its untenable concept of truth, according to which the truth of a proposition depends upon its consequences. And a similar conception of truth is indeed diametrically opposed, in spite of the principle of verification Schlick himself endorsed, to Schlick s definition of truth as univocal coordination of judgements with states of affairs 26. The other side of Schlick s reviews concerning contributions to psychology can be summarized as an attempt to call attention to the boundary questions of logic, psychology and theory of knowledge. This is, for example, the case of Joseph Klemes Kreibig s book devoted to the study of intellectual functions, which Schlick evaluates as a typical result of the Austrian way of philosophical thinking promoted by authors such as Bolzano, Meinong, Höfler and so on 27. Schlick s arguments in this context are the same ones he developed in 1910 in his paper on the nature of truth as regards the relationship between logic and psychology. It is well known that Schlick discussed here Edmund Husserl s Logical Investigations and questioned his refusal of psychologism, arguing that it is right for logic to abstract from the individual-psychical aspect, but it is misleading to ignore the psychological element as such. We cannot [ ] regard the logical propositions, he says, as structures devoid of any mental character 28. In order to contrast Husserl s theory of pure logic as well as Bertrand Russell s platonism, Schlick refers to the concept of the organisation of our minds (Organisation des Geistes), that is, by the way, to a concept which had been elaborated, most notably, by Friedrich Albert Lange in his History of Materialism 29. In Schlick s opinion, the necessity of truth by no means implies that it can subsist independently from our psychological organisation or as something external of the mind. Our truths this is Schlick s claim are dependent upon the organisation of our minds, though it is precisely for that reason that they are necessary for us 30. Starting from this background, Schlick reviewed, for example, Adolf Stöhr s Lehrbuch der Logik in psychologisierender Darstellung and others contributions to logical inquiry such as Johannes von Kries s Logik 31 considering, more generally, the great problem of the relationship between logic and reality 32. Two main points seem to emerge here. On the one hand, it is a comprehensive image of logic which can be derived from Schlick s critical notices. So it is remarkable that Schlick, on the occasion of the publication of Ernst Schröder s Abriss der Algebra der Logik, expressed some doubts regarding its importance

83 AN UNKNOWN SIDE OF MORITZ SCHLICK S INTELLECTUAL BIOGRAPHY 67 for the great revolution within 20th century logic. Schlick, by contrast, did not believe that the idea of a logical calculus was a very fruitful one, although already the great Leibniz the hero of modern symbolic logic placed all his hopes in the calculemus! offering a solution to every philosophical problem. The reason for Schlick s skepticism was his conviction that a similar optimistic point of view presupposes the entire set of the fundamental concepts representing the reality and the relations underlying the reality itself. Yet the solution of such a problem Schlick comments probably presupposes the solution of all the greatest problems and the algebra of logic would come too late. In the field of mathematics, the algorithmic language of signs is possible and successful because here the basic concepts and the axioms are very few in number. It follows that up to now the algebra of logic has found extensive application in only one field, namely in the field of mathematics itself, which does not require its aid despite the assurances of some spokesmen of logistics (Logistik) 33. Schlick s polemical attitude towards the logistics doubtless represents the premise of the more general critique of modern logic he formulated later in the General Theory of Knowledge, in which he stressed, in fact, the still enduring validity of Aristotele s syllogistic34. To be sure, it is astonishing that in the age of Frege, Peano, Russell and Couturat one of the most influential figures of the future Vienna Circle held a similar idea. And yet the philosophical framework of Schlick s logical conception must still be clarified. This is exactly the second point emerging from Schlick s reviews in the period before the publication ofhis main work. It is thus necessary in this context to recall some others aspects of Schlick s early thought, especially as regards his theory of knowledge revealing the great influence or, to some extent, the complex mixture both of Poincaré s philosophy of science and of the heritage of Kant s critical philosophy. Two elements here seem to be meaningful. First of all, Schlick maintained that the philosophy of logic cannot give up the fundamental difference between analytic and synthetic judgements as well as between ideal truths and factual truths 35. According to Schlick, a similar difference, that is, the difference posed by Leibniz and Hume between vérités de raison (or relations of ideas) and vérités de fait (or matters of fact), is indispensable for constructing logic in the spirit of Johannes von Kries s book as the critical and formal doctrine of judgements 36. This point of view, also stressed in the General Theory of Knowledge 37, of course is related to the studies on the criterion of truth Schlick had already pursued in his Habilitationsschrift. Here the distinction between vérités de raison and vérités de fait seeks to prove that, in spite of every difference of principle, both the fields of truth require, at any rate, a procedure of verification consisting in the establishment of identity (Gleichheit) of judgements 38. In order to fully understand Schlick s thesis we have to add another piece to this puzzle. In his reviews as well as in the Habilitationsschrift, Schlick acknowledged again and again Poincarés s great merits and raised no doubts as to the latter s scientific genius and the importance of his epistemological work 39. According to Schlick, Poincaré definitively demonstrated, on the basis of a

84 68 MASSIMO FERRARI general Kantian point of view, that the present-day philosophy of mathematics has abandoned radical empiricism 40. Moreover, Schlick pointed out that Poincaré s nominalism (as Enriques defined it) 41 played a leading role within the debates on the relationship between logic and mathematics; in particular, his objections against the logistics and the reduction of mathematics to an enormous tautology lie at the core of contemporary disputes on the nature of mathematics, casting a new light upon the classical problems of epistemology since Kant. 42 How are all these different remarks Schlick makes in his short contributions to the Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie to be understood? Once again, the answer may be found in his Habilitationsschrift, where both themes, namely the distinction between ideal truths and factual truth as well as the philosophy of science endorsed by Poincaré constitute the centre of gravity of Schlick s early philosophical views. In fact Schlick is convinced that all judgements having a purely logical or mathematical dignity are absolutely true and universally valid, whereas this is not the case with empirical judgements or factual truths. But the question is precisely if, and, if so, in what sense, the apriori status of the first kind of judgements and the absolute validity of the first kind of truths enable any a verification of this sort of proposition. Now, according to Schlick, the solution of this problem can be found by conceiving the verification as a procedure that occurs in our inner experience, thus enabling mental experiments or, as in the case of geometry, resorting to intuitive representations. Surprisingly enough, Schlick concluded from these assumptions that inner experience constitutes the ground for every verification of apriori, logical or mathematical judgements and that, therefore, their apodictic character rest upon the organization of our own understanding. So Schlick says whereas in empirical truths our conviction of their correctness is based upon the experiences in which we find them confirmed, in a priori propositions it rests upon the structure of our own mind, which makes possible inner experience and finds its expression therein. Thus Schlick noted in this context that he fit in in the field of Kant s philosophy, and in this sense the laws of the mind are nothing but what Kant called forms both of understanding and sensibility 43. It also becomes clear why Schlick, for the first time in his philosophical writings, referred to Poincaré s authority. Schlick by no means shared Poincaré s separation of mathematics from logic, insofar as both are subjected to verification through inner experience or induction; and yet Schlick is wholly at one with Poincaré regarding the nature of complete induction as raisonnement mathématique par excellence. This kind of reasoning is valid in the field of logic as well as in the field of apriori judgements in general, and taken as a whole it makes it clear the subjective or a priori origin of eternal truth 44. Schlick s epistemological interest in this connection can also be defined as an attempt to use Poincaré for support in finding the criterion of truth in the spirit of Kant s philosophy (in a broad sense). Poincaré s conventionalism and his philosophy of geometry, on the contrary, have a very marginal role in this context, unlike what characterized

85 AN UNKNOWN SIDE OF MORITZ SCHLICK S INTELLECTUAL BIOGRAPHY 69 Schlick s later reception of these topics in the framework of his epistemological analysis of Einstein s theory of relativity. It is not possible here to offer an extensive account of Schlick s gradual revision of his former commitment to the field of Kant s philosophy and to illustrate his new, surely more anti-kantian perspective in the General Theory of Knowledge. At any rate, the fact that Schlick moved away from Kantianism and neo-kantianism is more complicated than is usually maintained, and is well documented by some reviews he published during these years. Schlick, for example, pointed out in his 1916 discussion of Hans Cornelius s Transzendentale Systematik, that this excellent publication has doubtless the merit of aspiring to a new orientation of the transcendental method, in particular as a result of the new laws of consciousness Cornelius was able to show (e.g. recognizing, remembering, distinguishing and so on) 45. But much more interesting is perhaps Schlick s agreement with Cornelius on the necessary transformation of the Kantian thing in itself, namely for the good reason, as Schlick says, that the transcendent, uncognizable thing in itself has to be substituted by the empirical cognizable thing in itself 46. As is well known, the theory of things in itself as cognizable things is precisely the main point of the General Theory of Knowledge 47 ; and yet it is instructive to see how Schlick gradually elaborates this argument during the period of his activity as a reviewer, combining, so to speak, Riehl s suggestions for a realistic interpretation of Kant with other contemporary sources and philosophical arguments. This is clearly the case of Max Frischeisen-Köhler s book Wissenschaft und Wirklichkeit, which Schlick reviewed in 1913, agreeing with the proofs in favour of the realism exhibited by this pupil of Wilhelm Dilthey. Schlick had been, of course, deeply interested in this topic since his training with Planck, whose physical realism he for the most part accepted 48. Now Schlick found in Frischeisen-Köhler s book, as well as in some arguments already proposed by Dilthey, the stimulating idea that the proof of the reality of the external world as independent from us does not follow from any theoretical reason, but is only the outcome of the practical, immediate experience (Erlebnis) of something resisting us (Hemmung) 49. On the other hand, Schlick suggested that Frischeisen-Köhler s objections against the neo-kantian idealism both of the Baden and the Marburg Schools are absolutely very well selected and perfectly right 50. To be sure, concerning Windelband and especially Rickert, this opinion does not represent any new, considering that Schlick had already expressed it in his Habilitationsschrift, where, once again starting from an essay by Frischeisen-Köhler, he developed a stringent critique of Rickert s conception of truth as a total reversal of the customary and established relationship between judgements and reality 51. But in the case of the Marburg school, Schlick was very happy to see that Frischeisen-Köhler criticized, on the basis of successful arguments, the conception of the object of knowledge as an infinite task, as the X of knowledge determined absolutely through the pure thought. On the contrary, Schlick

86 70 MASSIMO FERRARI says by quoting Frischeisen-Köhler himself that the object must have a determination already before the act of knowing 52. The polemical attitude towards the logical idealism of the Marburg school also figures centrally in an extensive review Schlick wrote in 1911 on Paul Natorp s famous book Die logischen Grundlagen der exakten Wissenschaften: a review inaugurating, moreover, Schlick s engagement as contributor to the Vierteljahrschrift 53. To understand the more general point of view underlying this very meaningful discussion of Natorp s main epistemological work, it is useful to quote first of all the last passages of Schlick s review, which include an ironical laudatio of the Marburg neo-kantianism: To be sure, even the rival of this logical idealism will find in every chapter of this book, which is indeed very difficult for the reader, a lot of stimulating themes, and it is a pleasure to follow the bold paths covered by the enthusiastic author, although one will, in the end, abandon these lands of pure thought with a sigh of relief 54. Similar skeptical remarks about the role of pure thought, and Schlick s own conception of the object of knowledge as something quite independent from thought, also underlie the critical examination of the Marburg neo-kantianism in which Schlick engaged still later in his General Theory of Knowledge. And there can be no doubt that for Schlick, in 1911 as well as in 1918, the forms of thought by no means had the status of creative forms asserted by Cohen and Natorp; on the contrary, they are purely forms of the given. 55 However, what seems noteworthy in the review of Natorp is that Schlick juxtaposed Kant s pure intuition to the Marburg theory of pure and creative thought as the exclusive ground of knowledge. Whereas Natorp attempted to provide the logical foundation of mathematics by transforming Kant s transcendental logic, Schlick, by contrast, emphazised that Kant s critical project depends upon the distiction between transcendental aesthetics and transcendental logic 56. According to Schlick, this means that Natorp s unfortunate conception about space and time as mere products of pure thought had to be abandoned. Schlick, in fact, protested expressly against the (in his opinion) Hegelian misinterpretation of Kant proposed by Natorp. Instead he drew the attention to the circumstance that assuming space and time as mere conceptual orders implies the ignorance of their differentia specifica, that is, the difference based upon pure intuition and distinguishing space and time as mathematical concepts 57. Schlick thus informed the readers of the Vierteljahrschrift of Natorp s failure to determine the content of thought by resort to intuition; and he added that in the reviewer s opinion, the other chapters of the book also fail to acknowledge the role intuition has played by achieving some results 58. To be sure, Schlick certainly did not aim at being a kind of orthodox Kantian 59, as he says elsewhere. But, at least at this time, Schlick was convinced that it was still possible to defend Kant s philosophy of geometry based upon intuition. In 1911, for example, he remarked that the geometrical axioms are characterized by intuitive necessity, but not by conceptual necessity; from this it follows, according to Schlick, that it is possible to reconcile Kant with non-

87 AN UNKNOWN SIDE OF MORITZ SCHLICK S INTELLECTUAL BIOGRAPHY 71 Euclidean geometries insofar as geometries belong only to the field of possibilities of thought 60. Here Schlick seemed to be acquainted with the strategies of immunization put forward by most neo-kantians and, more precisely, with Riehl s influential solution to the puzzle about Kant and non-euclidean geometries 61. But this point of view was to be completely given up in the General Theory of Knowledge, in which Schlick no longer considered geometrical space as intuitive representation and, in spite of his former neo-kantian way out, he defined the space in terms of pure conceptual idealisation 62. This collapse of spatial intuition in a Kantian or neo-kantian sense is surely connected, on the one hand, to Schlick s more general refusal of intuition within the process of knowledge and philosophical inquiry (as was endorsed, by contrast, by Husserl, Bergson and other thinkers) 63, and, on the other hand, to Schlick s epistemological analysis of Einstein s theory of relativity since The last aspect already can be traced back to the review of Natorp64 but emerges most prominently in the discussion about both Natorp s and Richard Honigswald s attempts to provide a neo-kantian foundation to the principle of relativity, according to which pure space and pure time may still play the role of transcendental presuppositions of every empirical measurement of fully relativized spaces and times: i.e., exactly the solution Schlick believed to be incompatible with modern relativistic physics 65. And yet in 1915 Schlick was still convinced that Kant s theory of knowledge and, in particular, his doctrine of space and time had not to be abandoned, but only modified with regard to essential parts. 66 For Schlick, in any case, a similar modification did not damage the correct point of view endorsed both by Kantian philosophy and neo-kantianism, that is, the conception of philosophical analysis as critical inquiry into the ultimate principles of the special sciences, which together makes possible the epistemological justification of all the newly discovered principles emerging from their historical development 67. This attitude also characterized Schlick s reviews during the period from 1911 to and constituted, more generally, the background of his philosophical activity up to the General Theory of Knowledge. At the end of this long path much of which involved his contributions to the Vierteljahrschrift fur wissenschaftliche Philosophie und Soziologie, Schlick was on the threshold to a new development of his thought, which is surely best known and studied largely by the scholars of Schlick s philosophy. But the period beginning with both the epistemological analysis of relativistic physics and the General Theory of Knowledge of 1918 belongs to another stage of Schlick s intellectual biography. And this story, we hope, will be told elsewhere.

88 72 MASSIMO FERRARI APPENDIX: SCHLICK S REVIEWS FOR THE VIERTELJAHRSCHRIFT FÜR WISSENSCHAFTLICHE PHILOSOPHIE UND SOZIOLOGIE P. Natorp, Die logischen Grundlagen der exakten Wissenschaften (Berlin/Leipzig: Teubner 1910), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 35, 1911, pp (Schlick 1911a). A. Voß, Über das Wesen der Mathematik (Leipzig/Berlin: Teubner 1908), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 35, 1911, pp (Schlick 1911b). L. Goldschmidt, Zur Wiedererweckung Kantscher Lehre (Gotha: F.A. Peters 1910), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 35, 1911, pp (Schlick 1911c). A. Bilharz, Descartes, Hume und Kant. Eine kritische Studie zur Geschichte der Philosophie (Wiesbaden [unknown publishers] : 1910), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 35, 1911, pp (Schlick 1911d). G. Mannoury, Methodologisches und Philosophisches zur Elementarmathematik (Harlem [unknown publishers]: 1909), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 35, 1911, pp (Schlick 1911 e). E. Schröder, Abriss der Algebra der Logik (bearbeitet im Auftrag der Deutschen Mathematiker-Vereinigung von E. Müller, I. Teil, Leipzig: Teubner 1910), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 35, 1911, pp (schlick 1911f). F. Enriques, Probleme der Wissenschaft (German transl. by K. Grelling, 2 volls., Leipzig/Berlin: Teubner 1910), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 35, 1911, pp (Schlick 1911g). A. Stöhr, Lehrbuch der Logik in psychologisierender Darstellung (Leipzig/Vienna: Fr. Deuticke [without year]), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 35, 1911, pp (Schlick 1911h). A. E. Haas, Die Entwicklungsgeschichte des Satzes von der Erhaltung der Kraft (Vienna: Hölder 1909), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 35, 1911, pp (Schlick 1911i). R. Eisler, Grundlagen der Philosophie des Geisteslebens (Leipzig: W. Klinkhardt 1908), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 35, 1911, pp (Schlick 1911l). W. Wundt, Die Prinzipien der mechanischen Naturlehre (Stuttgart: Enke 1910), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 35, 1911, pp (Schlick 1911m). O. Boelitz, Die Lehre vom Zufall bei Emile Boutroux. Ein Beitrag zur Geschichte der neuesten französischen Philosophie (Leipzig: Quelle & Meyer 1907), in: Vierteljahrschrif für wissenschaftliche Philosophie und Soziologie, 35, 1911, p. 441 (Schlick 1911n).

89 AN UNKNOWN SIDE OF MORITZ SCHLICK S INTELLECTUAL BIOGRAPHY 73 J. Petersen, Kausalität, Determinismus und Fatalismus (Munich: J.F. Lehmanns Verlag 1909), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 35, 1911, pp (Schlick 1911o). C. Stumpf, Philosophische Reden und Vorträge (Leipzig: Barth 1910); Die Wiedergeburt der Philosophie (Leipzig: Barth 1908); Vom ethischen Skeptizismus (Leipzig: Barth 1909), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 35, 1911, pp (Schlick 1911p). P. Volkmann, Erkenntnistheoretische Grundzüge der Naturwissenschaften und ihre Beziehungen zum Geistesleben der Gegenwart (Leipzig/Berlin: Teubner 1910), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 36, 1912, pp (Schlick 1912a). P. Volkmann, Die Eigenart der Natur und der Eigensinn des Monismus. Vortrag (Leipzig/Berlin: Teubner 1910), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 36, 1912, pp (Schlick 1912b). W. James, Psychologie (german transl. by M.Dürr, Leipzig: Quelle & Meyer 1909), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 36, 1912, p. 295 (Schlick 1912c). J. K. Kreibig, Die intellektuellen Funktionen. Untersuchungen über die Grenzfragen der Logik, Psychologie und Erkenntnistheorie (Vienna/Leipzig: Hölder 1909), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 36, 1912, pp (Schlick 1912d). M. Rubner, Kraft und Staff im Haushalte der Natur (Leipzig: Akademische Verlagsgesellschaft 1909), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 37, 1913, pp (Schlick 1913a). K. C. Schneider, Vorlesungen über Tierpsychologie (Leipzig: Engelmann 1909), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 37, 1913, pp (Schlick 1913b). M. Frischeisen-Köhler, Wissenschaft und Wirklichkeit (Leipzig / Berlin: Teubner 1912), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 37, 1913, pp (Schlick 1913c), S. Lourié, Die Prinzipien der Wahrscheinlichkeitsrechnung. Eine logische Untersuchung des disjuntiven Urteils (Tübingen: Mohr 1910), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 38, 1914, pp (Schlick 1914a). H. Lüdemann, Das Erkennen und die Werturteile (Leipzig: M. Heinsius Verlag 1910), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 38, 1914, pp (Schlick 1914b). H. Dingier, Die Grundlagen der Naturphilosophie (Leipzig: Verlag Unesma 1913), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 39, 1915, pp (Schlick 1915a). E. Becher, Weltgebäude, Weltgesetzte, Weltentwicklung (Berlin: Reimer 1915), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 40, 1916, pp (Schlick 1916a). H. Driesch, Die Logik als Aufgabe. Eine Studie über die Beziehung zwischen Phänomenologie und Logik, zugleich eine Einleitung in die Ordnungslehre

90 74 MASSIMO FERRARI (Tübingen: Mohr 1913), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 40, 1916, pp (Schlick 1916b). F. Raab, Die Philosophie von Richard Avenarius. Systematische Darstellung und immanente Kritik (Leipzig: Meiner 1912), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 40, 1916, pp (Schlick 1916c). A. Lehmann, Die Hauptgesetzte des menschlichen Gefühlslebens (Leipzig: Reisland 1914), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 40, 1916, pp (Schlick 1916d). W. von Bechterew, Objektive Psychologie oder Psychoreflexologie, Die Lehre von den Assoziationsreflexen (Leipzig/Berlin: Teubner 1913), in: Vierteljahrschriftfür wissenschaftliche Philosophie und Soziologie, 40, 1916, pp (Schlick 1916e). J. Burnet, Die Anfänge der griechischen Philosophie (german transl. Leipzig/Berlin: Teubner 1913), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 40, 1916, pp (Schlick 1916f). R. Herbetz, Prolegomena zu einer realistischen Logik (Halle: Niemeyer 1916), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 40, 1916, pp (Schlick 1916g). J. von Kries, Logik. Grundzüge einer kritischen und formalen Urteilslehre (Tübingen: Mohr 1916), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 40, 1916, pp (Schlick 1916h). H. Cornelius, Transzendentale Systematik. Untersuchungen zur Begründung der Erkenntnistheorie (Munich: Reinhardt 1916), in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 40, 1916, pp (Schlick 1916i). NOTES Cf. the contributions of Herbert Feigl, Moritz Schlick, in: Erkenntnis, 7, 1937/38, pp , transl. in: Henk L.Mulder/Barbara F.B.van de Velde-Schlick (Eds.), Moritz Schlick: Philosophical Papers, Vol. I ( ), Dordrecht: Reidel 1979, pp.xv-xxxviii; Friedrich Waismann, Vorwort, in: Moritz Schlick, Gesammelle Aufsätze , Vienna: Gerold & Co (reprint Hildesheim: Olms 1969), pp.vii-xxxi; Eugene T.Gadol (Ed.), Rationality and Science. A Memorial Volume for Moritz Schlick in Celebration of the Centennial of His Birth, Vienna-New York: Springer 1982; Anthony Quinton, Vor Wittgenstein: der frühe Schlick, in: Brian McGuinness (Ed.), Zurück zu Schlick. Eine Neubewertung von Werk und Wirkung, Vienna: Hölder-Pichler-Tempsky 1985, pp A short biography of Schlick together with an extensive bibliography of his writings as well as of his courses at the University of Vienna is offered by Friedrich Stadler, The Vienna Circle. Studies in the Origins, Development, and Influence of Logical Empiricism, Vienna-New York: Springer 2001, pp (see also pp for an impressive collection of documents concerning the murder of Schlick). Finally, I allow myself to refer the reader to my forthcoming essay Der junge Schlick. Ein Stück intellektueller Biographie (lecture delivered in Vienna and Graz, March 2000). Cf. for example Karin Gerner, Hans Reichenbach. Sein Leben und Wirken, eine wissenschaftliche Biographie, Osnabrück: Phoebe-Autorenpress See also the critical review of this book by Reinhard Siegmund-Schultze in: Miklós Rédei/Michael Stölzner (Eds.), John von

91 AN UNKNOWN SIDE OF MORITZ SCHLICK S INTELLECTUAL BIOGRAPHY Neumann and the Foundations of Quantum Physics, Dordrecht/Boston/London: Kluwer 2001, pp Alberto Coffa, The Semantic Tradition from Kant to Carnap. At the Vienna Station, Cambridge: Cambridge University Press 1991, pp Moritz Schlick, Allegemeine Erkenntnislehre, Berlin: Springer, 1918 revised ed (reprint Frankfurt/Main: Suhrkamp 1979), English transl. General Theory of Knowledge, with an introduction by Albert E.Blumberg and Herbert Feigl, Vienna-New York: Springer-Verlag Moritz Schlick, Über die Reflexion des Lichtes in einer inhomogenen Schicht, Berlin: Inaugural Dissertation 1904; Lebensweisheit. Versuch einer Gliickseligkeitslehre, Munich: Beck 1908; Das Wesen der Wahrheit nach der modernen Logik, in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 34, 1910, pp , reprint in: Moritz Schlick, Philosophische Logik, ed. by Bernard Philippi, Frankfurt/Main: Surhkamp 1986, pp (English transl. The Nature of Truth in Modern Logic, in: Schlick, Philosophical Papers Vol. I, op. cit., pp ). A brief survey of the young Schlick s early philosophical attempts may be found in Massimo Ferrari, Der junge Schlick. Ein Stück intellektueller Biographic, pp Moritz Schlick, Die Grenze der naturwissenschaftlichen und philosophischen Begriffsbildung, in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 34, 1910, pp , reprint in: Schlick, Philosophische Logik, op. cit., pp (English transl. The Boundaries of Scientific and Philosophical Concept-formation, in: Schlick, Philosophical Papers Vol. I, op. cit., pp ). Cf. Schlick, Philosophical Papers Vol. I, op. cit., pp Cf. Moritz Schlick, Gibt es intuitive Erkenntnis?, in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 37, 1913, pp , English transl. Is There Intuitive Knowledge?, in: Schlick, Philosophical Papers Vol. 1, op. cit., pp ; Die philosophische Bedeutung des Relativitätsprinzips, in: Zeitschrift für Philosophie und philosophische Kritik, , pp , English transl. The Philosophical Significance of the Principle of Relativity, in: Schlick, Philosophical Papers Vol. 1, op. cit., pp ; Idealität des Raumes, Introjektion und psychophysisches Problem, in: Vierteljahrschrift für wissenschaftliche Philosophie und Soziologie, 40, 1916, pp , English transl. Ideality of Space, Introjection and the Psycho-Physical Problem, in: Schlick, Philosophical Papers Vol. I, op. cit., pp ; Raum und Zeit in der gegenwärtigen Physik, in: Die Naturwissenschaften, 5, 1917, pp , , English trans, of the enlargededition (Berlin: Springer 1922) Space and Time in Contemporary Physics, in: Schlick, Philosophical Papers Vol. I, op. cit., pp See Klaus Christian Köhnke, Entstehung und Aufstieg des Neukantianismus. Die deutsche Universitätsphilosophie zwischen Idealismus und Positivismus, Frankfurt/Main: Suhrkamp 1986, pp We refer to Schlick s Nachlaß, Inv. Nr See once again Schlick s Nachlaß, Inv. Nr (it is about Schlick s notes from vol. II of Riehl s book). For a survey of Riehl s neo-kantianism cf. Wolfgang Röd, Alois Riehl und der Herbartianismus in Österreich, in: J.C. Nyiri (Ed.), Von Bolzano zu Wittgenstein. Zur Tradition der österreichischen Philosophie, Vienna: Hölder-Pichler-Tempsky 1986, pp ; Massimo Ferrari, Introduzione a II neocrilicismo, Rome-Bari: Laterza 1997, pp.36-40, 90-92; Manfred Pascher, Einführung in den Neukantianismus, Munich: Fink 1997, pp ; Renato Pettoello, De Herbart à Kant. Quelques considérations sur le réalisme de Alois Riehl, in: Revue de Mêtaphysique et de Morale, 102, 1998, pp See Schlick 1916c, p We quote Schlick s reviews only giving the year of publication and following, within the same year, an order indicated by alphabetic letters. The complete list of the reviews can be found in the appendix to this paper. See Schlick 1911d, 1911i, 1911l, 1911n, 1911o, 1911p, 1912b, 1913a, 1914b. Schlick 1916f. Schlick 1916c, p See e.g. Schlick 1916a. Cf. Thomas Uebel, Vernunftkritik und Wissenschaft. Otto Neurath und der erste Wiener Kreis, Vienna-New York: Springer 2000, p Schlick 1912 and Schlick 1915.

92 76 MASSIMO FERRARI Schlick 1911g, p Ibid, pp See Schlick 1916d and Schlick 1916e. Schlick 1912d, p. 297; see also Schlick 1916b, p. 258, Schlick examines Gestaltpsychologie still later, during a seminar at the University of Vienna in the years (cf. Gestaltpsychologie, ed. by F.Stadler, in: Conceptus, 21, 1987, 53-54, pp ). Also see Schlick, General Theory of Knowledge, op. cit., pp , concerning especially a brief discussion of Ehrenfels. Schlick 1913b, p Schlick 1912c, p.295 Schlick, The Nature of Truth in Modern Logic, loc. cit., pp Ibid., pp , 88,94 ff. Schlick 1912d, p.296. Schlick, The Nature of Truth in Modern Logic, loc. cit., p. 54. Schlick expresses his esteem for Lange s masterpiece elsewhere too: see The Boundaries of Scientific and Philosophical Concept-formation, loc. cit., p. 30 and General Theory of Knowledge, op. cit., p Schlick, The Nature of Truth in Modern Logic, loc. cit., p.98. Cf. also pp , where Schlick considers perfectly correct Herbert Spencer s psychological interpretation of the principle of contradiction (a point sharply contradicted by Husserl s critique of psychologism). Schlick 1911h and Schlick 1916h. Schlick 1916g, p Schlick 1911f, p Schlick, General Theory of Knowledge, op. cit., p See also Warren Goldfarb, The Philosophy of Mathematics in Early Positivism, in: Ronald N. Giere/Alan W. Richardson (Eds.), Origins of Logical Empiricism, Minneapolis/London: University of Minnesota Press 1996, p.215. Schlick 1916f, p Schlick 1916h, p Schlick, General Theory of Knowledge, op. cit., p Schlick, The Nature of Truth in Modern Logic, loc. cit., pp.80, (translation slightly modified). Schlick 1912a, p. 294; Schlick 1913, p.486. Schlick s admiration for Poincaré concerns at the same time his contribution to the theory of probability: Anybody who looks for clarity about the principles of the calculus of probability has to consult the books of a mathematician such as Poincaré. One will find therein, although condensed in few propositions, more correct insights into the essence of probability than are offered by the long and to some extent metaphoric arguments of Lourié [Schlick was referring here to Samuel Lourié s Prinzipien der Wahrscheinlichkeitsrechung, i.e. the book which he isjust reviewing] (Schlick 1914a, p. 277). Schlick 191 le, p Schlick 191 lg, p Schlick 191 lb, p Schlick, The Nature of Truth in Modern Logic, loc. cit., p. 84 (translation slightly modified). Ibid.,p. 85. Schlick 1916i, p.385. Schlick 1916i, p Schlick, General Theory of Knowledge, op. cit., pp. 88ff., 194ff., 235ff. Cf. Joia Lewis Turner, Conceptual Knowledge and Intuitive Experience: Schlick s Dilemma, in: Ronald N. Giere/Alan W. Richardson (Eds.), Origins of Logical Empiricism, op. cit., pp ; Ferrari, Der junge Schlick. Ein Stück intellektueller Biographic, pp Schlick 1913c, p See General Theory of Knowledge, op. cit., p. 176 and Coffa, The Semantic Tradition, op. cit., p Schlick 1913c, p Schlick, The Nature of Truth in Modern Logic, loc. cit., pp.45 ff., especially p.49. See also Max Frischeisen-Köhler, Über die Grenzen der naturwissenschaftlichen Begriffsbildung, in: Archiv für systematische Philosophie, 12, 1906, pp

93 AN UNKNOWN SIDE OF MORITZ SCHLICK S INTELLECTUAL BIOGRAPHY Schlick 1913c, p Schlick 1911a. Schlick 1911a, p.260. Schlick General Theory of Knowledge, op. cit., p See also ibid., p. 366: In our view, facts stand fixed even without being captured by concepts. Schlick 1911a, p Schlick 1911a, p Schlick 1911a, p.258. Schlick remarks, furthermore, that Natorp s claim to draw all the determinations of thought from the principle of the pure origin (Ursprung) is untenable: Indeed, the concepts at work here are so abstract, that it seems impossible to employ them without building unvoluntarily intuitive images, and the language must be therefore a metaphoric one in order to describe similar concepts (Schlick 1911 a, p. 257). Schlick 1911c. Schlick 1911b, p.261. See Alois Riehl, Helmholtz in seinem Verhältnis zu Kant, in: Kant-Studien, 9, 1904, pp (esp. pp ) and Alois Riehl, Der philosophische Kritizismus. Geschichte und System, vol. II, Die sinnlichen und logischen Grundlagen der Erkenntnis, (1879), 2nd revised edition, Leipzig: Kröner 1925, pp Schlick General Theory of Knowledge, op. cit., pp See Schlick, Is There Intuitive Knowledge?, loc. cit., especially p. 149 for a similar critique of Kant s pure intuition. Schlick 1911a,p.260. Schlick, The Philosophical Significance of the Principle of Relativity, loc. cit., p On the discussion between Schlick and Hönigswald see Massimo Ferrari, Eine Diskussion über die Relativitätstheorie. Richard Hönigswald und Moritz Schlick, in: Wolfdietrich Schmied- Kowarzik (Ed.), Erkennen-Monas-Sprache. Internationales Richard-Hönigswald-Symposion Kassel 1995, Würzburg: Königshausen & Neumann 1997, pp Schlick, The Philosophical Significance of the Principle of Relativity, loc. cit., p Ibid., pp See for example Schlick 1911m, p Dipartimento di Storia University of L Aquila I L Aquila,Via Roma 33 Italy mferrari@cc.univaq.it

94 HANS JUERGEN WENDEL BETWEEN MEANING AND DEMARCATION 1. A CRITERION OF DEMARCATION DOES NOT DISTINGUISH BETWEEN MEANING AND NONSENSE BUT BETWEEN CONCEPTUALLY DIFFERENT KINDS OF KNOWLEDGE IN THIS RESPECT, IT IS A CRITERION OF MEANING. In his Critique of Pure Reason, Immanuel Kant s main intention was to demarcate justifiable and legitimate cognitive claims within metaphysics in order to place philosophical knowledge on secure foundations. His much-discussed conclusion was that all metaphysical cognitive claims that go beyond subjective conditions of possible knowledge remain unjustifiable. As a result of the breakthroughs in the empirical sciences and the advances in the quest to clarify the foundations of logic and mathematics in the 20th century a need for demarcation arose not only with respect to exaggerated philosophical deliberations but also to all non-science and thus to philosophy as a whole. In the process, possible factual knowledge has been restricted to the domain of scientific knowledge alone. Logical Empiricism employs verifiability as a criterion of meaning primarily to demarcate a meaningful use of language with rich empirical content from empty and therefore meaningless use of language by stating the basis of the validity of experience. In their characterization of scientific statements, the members of the Vienna Circle took a methodological demarcation of scientific method as their starting point, relying on Bacon s characterization of the natural sciences as inductive sciences. In their view, general statements about nature can only be justified through repeated observations or experiments. This inductive method seemed to be a positive feature of all science. By resorting to this supposed characteristic of empirical science, the members of the Vienna Circle wanted to demarcate it from metaphysics. Simultaneously, however, this demarcation meant to them a demarcation between the meaningful and the meaningless. It should be noted, however, that their position identifies being meaningful with being empirically meaningful, empirical verifiability that is, inductive provability through observational statements being regarded as the criterion. This characterization leads not only to insuperable difficulties in the demarcation of what is to be regarded as scientific, but, combined with the view that this exclusively determines the meaning of knowledge in general, it also makes its adherents regard all other statements as meaningless or as pseudo-statements. 79 F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

95 80 HANS JUERGEN WENDEL In his critical analysis of the inadequacy of the empirical criterion of meaning, Karl Popper emphasized the importance of an adequate criterion of demarcation: No significant part of science must be excluded; It is problematic to use such a distinction not only to demarcate science from non-science (metaphysics) but also to identify knowledge in general with scientific knowledge. In his Logic of Scientific Discovery 1 Popper therefore proposed falsifiability as the suitable criterion of demarcation that enables us to distinguish between scientific and non-scientific knowledge within the realm of knowledge without imposing a verdict of meaninglessness on the realm of philosophy in the way logical positivists of the Vienna Circle 2 had wanted to. Falsifiability, Popper says, separates two kinds of perfectly meaningful statements: the falsifiable and the non-falsifiable. It draws a line within meaningful language, not around it. 3. From this point of view it looks as if this demarcation is indeed of a conceptual nature and, in addition, factually important because inadequate demarcation concerns everything belonging to empirical science, as the critique of the empiricist criterion of meaning has shown. When Popper says that it is not important to define the concept of science but to give a criterion that factually distinguishes scientific statements in order to delimit them from other statements, he is not opposing the indication of linguistic meaning to the indication of a factual criterion, rather is he opposing definition in the sense of explicit definitory stipulation to analysis in the sense of an explication and clarification of actual linguistic usage. On closer examination, Popper s critique therefore does not replace a criterion of meaning by a criterion of demarcation, rather it is a critique of an inadequate explication of what constitutes the nature of scientific method and a proposal for a better understanding of the peculiar characteristics of science. Nonetheless, Popper rejects all opinions which assume that one could possibly gain insight into the nature of things by conceptual or Wesensanalysis. 4 He considers all such positions to be forms of essentialism in his view an untenable doctrine. 5 Such a position assumes that conceptual clarification or definition enables us to make a statement of the inherent essence or nature of a thing. 6 A definition in this sense states the meaning of a word of the name that designates the essence. (ibid.). But to Popper s mind, questions of conceptual clarification or definition have nothing to do with knowledge. Definitions are never factual. If the meaning of terms depends on conventions 7 alone, then all arguments about concepts insofar as they are concerned with cognitive claims turn out to be fruitless because according to Popper definitions can never add anything to our factual knowledge about the nature of things, 8 since the descriptive content connected with linguistic expressions is of a purely conventional nature. For this reason, the argument about concepts or their definitions is, according to Popper,

96 BETWEEN MEANING AND DEMARCATION 81 an epistemologically fruitless undertaking. Which is why he says: [...] we should altogether avoid, like the plague, discussing the meaning of words. Discussing the meaning of words is a favourite game of philosophy, past and present: philosophers seem to be addicted to the idea that words and their meaning are important, and are the special concern of philosophy. 9 Instead, one should always keep to assertions, to theories, and the question of their truth. One should never get involved in verbal questions or questions of meaning, and never get interested in words. If challenged by the question of whether a word one uses really means this or perhaps that, then one should say I don t know, and I am not interested in meanings; and if you wish, I will gladly accept your terminology. This never does any harm. 10 What we are really interested in, in Popper s view, are our real problems, are factual problems, or in other words, problems of theories and their truth. 11 Thus, Popper wants to have questions of knowledge strictly separated from questions of linguistic meaning. On the other hand the criterion of demarcation should in any event be factual if it is to be useful. 2. WHETHER WE CONSIDER SOMETHING TO BE KNOWLEDGE OR A MATTER OF CONCEPTUAL CLARIFICATION DEPENDS ON THE CONTEXT OF KNOWLEDGE. Popper rather casually admits that useful conceptual stipulations in particular have to be factually well-founded when he says that it is not important to overthrow metaphysics but, to formulate a suitable characterization of empirical science, or to define concepts such as empirical science and metaphysics (LD 37). Giving a suitable criterion of demarcation can therefore be regarded as a factually justified proposal for an agreement or convention (LD 37). What Popper does not see is the connection between questions of linguistic meaning and questions of knowledge. With his explication of the distinction between analytic and synthetic judgements, Kant wanted to emphasize that especially in philosophy, many of our judgements that allegedly extend our knowledge are basically not ampliative judgements but merely hidden clarifications of given concepts in the disguise of ampliative knowledge, a fact we typically tend to overlook. The mere analysis of concepts would therefore often wrongly be regarded as new knowledge. But the extension of knowledge is very different from the clarification of concepts and, therefore, the two have to be kept apart. Kant s motive for the introduction of the distinction between analytic and synthetic judgements, that is, between conceptual clarifications and factual judgements, was essentially epistemological. He was aiming at a strict distinction between judgements that can rightfully claim to be extensions of knowledge and those that only clarify given knowledge. In making this distinction of cognitive value between different kinds of judgements Kant was concerned with the cognitive aspect of our route to knowl-

97 82 HANS JUERGEN WENDEL edge, which is why he distinguished (i) acts of judgement with respect to their cognitive value (for the subject). In Kant s opinion there exists (ii) a fundamental connection between cognitive judgements and the formation of concepts in such a way that knowledge simultaneously influences concepts so that cognition always implies at the same time formation of concepts; he therefore demanded that the appraisal of the cognitive value of a judgement always be referred to a particular level of knowledge (iii). Taking into account the dynamic aspect of knowledge which leads us to look at the progress of knowledge as a succession of levels of knowledge, the cognitive value of acts of judgements in Kant s view changes accordingly: what was informative at a former level of knowledge (and therefore of the formation of concepts) may now be a mere clarification (iv). According to Kant, the distinction between analytic and synthetic judgements refers to the different ways differing with regard to their cognitive value in which subject and predicate can be connected with one another in an act of judgement with respect to their content: If the predicate is already inherent in the subject in the notion I already have l2 then our knowledge about (the objects subsumed under) the subject cannot be extended by relating it to this predicate; in the act of judgement we have nothing more than a clarification of the subject in question, an explication of what I already know of it. Whereas if subject and predicate differ with respect to their content inasmuch as the given subject does not already contain the predicate, then both become connected only in the act of judgement so that in the process the subject comes to be extended through additional determination: thus we have a synthesis in the act of judgement. In distinguishing between analytic and synthetic judgements, Kant was therefore mainly concerned with different types of connection as differing cognitive acts and not with the sources of the validity of such connections. For this reason, Kant regards the difference in the possible relation between the content of subject and predicate, that is, the distinction between analytic and synthetic judgements, as fundamental in terms of its cognitive value. 13 The point for Kant was to distinguish judgements that only clarify already given knowledge (the concept that one already has of the object to be cognized) from those that give some new insight (that extend our concept of the thing), and not to assign them to totally different spheres according to their nature so that the one type has to do with semantics and the other with knowledge. If knowledge including the process of forming concepts progresses then newly gained knowledge necessarily becomes given knowledge, which is in turn capable of being further extended and so on. Consequently, Kant explicitly stressed the primary character of synthesis, of the extension of knowledge, as compared with analysis when he says in the Transcendental Deduction : The reader will easily enough perceive that the possibility of conjunction must be grounded in the very nature of this act, and that it must be equally valid for all conjunction; and that dissection, analysis, which appears to be its contrary, must, nevertheless, always presuppose it; for where the understanding has not previ-

98 BETWEEN MEANING AND DEMARCATION 83 ously conjoined, it cannot dissect! 14. New, newly made knowledge, as we might say, leads to a change in the historical level of synthesis, in what is now to be regarded as a given concept with respect to which other (additional) formerly new judgements now become analytic judgements; for this concept is capable of a more extensive analysis at cognitive level than it used to be at cognitive level The relativity of the analytic-synthetic distinction is thus due to the change in our level of knowledge. Because for Kant, knowledge is essentially the same as the formation of concepts empirical concepts always reflect a certain level of knowledge. The distinction between analytic and synthetic judgements is therefore always related to a respective level of our given knowledge and is fixed with regard to this. Thus, the classification of judgements as analytic or synthetic has the character of being relative to a level of knowledge. So we have the following consequence: The difference regarding the cognitive value of judgements results from a) the assumption of already given connections in the case of analytic judgements and b) the creation of new knowledge by a previously not given new connection (synthesis) that goes beyond what was already connected and conceptually sedimented (knowledge). The reason for the peculiarity of the cognitive value of empirical judgments (as synthetic) is therefore that they impart new knowledge to us (as opposed to a mere clarification of what was given already) and not that the one type bears a relation to knowledge whereas the other does not. Moritz Schlick also emphasized this reciprocal connection between concepts and judgements, according to which our concepts change with knowledge. Thus in his Allgemeine Erkenntnislehre 15, he explicitly writes: Judgments and concepts stand in a peculiar relationship. Concepts are linked together by means of judgments, since every judgment designates a joining of two concepts. And thus a concept constitutes, as it were, appoint at which a series of judgment meet, namely, all those in which the concept occurs. (p. 46) And concerning definitions, Schlick holds definitions of a concept are those judgments that, so to speak, put it in touch with the concepts nearest it. The concept can be locked upon as a brief expression of these connections. (ibid.) In the progress of scientific knowledge because this is never strictly selfcontained we constantly become acquainted with new properties, so that the concepts of these objects acquire in time an even richer content. (p. 47). Conceptual contents, to Schlick as to Kant, are relative to knowledge. And for this reason it is not ipso facto clear whether a judgement is a definition or knowledge. Schlick says: Every judgment places a concept in relation to other concepts [...] If the concept in question is already familiar and defined, then we have an ordinary judgment. If this is not the case, then the concept is to be regarded as having been created by the judgment. The latter thus becomes the definition [...] (p. 47). For Schlick, definitions are basically also factual judgements: theoretically, definitions do not occupy a special position (ibid.). With this, the difference between definitions and judgements also changes as knowledge progresses and is

99 84 HANS JUERGEN WENDEL proven to be relative to knowledge. Insofar as the same linguistic term can stand for different conceptual contents in the course of the progression of knowledge, one and the same sentence may, depending on the particular state of the inquiry, serve either as a definition or as an instance of knowledge (ibid.) 16. In a similar way, Willard Van Orman Quine later pointed out that it is basically the same whether we speak of the linguistic knowledge of concepts or of empirically founded knowledge universally shared by a language community. 17 In this respect the distinction made by Quine between universally shared semantic knowledge and empirically attained additional knowledge roughly corresponds to the relativity of the distinction between given and constructed concepts in Kant s philosophy, and to Schlick s considerations on the cognitive relativity of definition and knowledge. 18 Quine, like the other two, assumed that universally shared knowledge, knowledge about concepts, originally was additional knowledge. But after its acquisition (by the members of a language community) it is no longer a question of knowledge to be capable of evaluating statements based on such knowledge. In their view, this could also be interpreted as changing the meaning of the terms involved. 19 Thus the distinction between agreement to statements on the basis of knowledge universally shared by the language community and agreement on the strength of additional knowledge corresponds approximately to the understanding of the distinction between analytic and synthetic knowledge according to Kant, as expounded above. 20 Once it has become part of general background knowledge (formerly) new knowledge has nothing more than a clarificatory character. On closer inspection, Quine does not give up the distinction between analytic and synthetic; rather he places it in an epistemological context with regard to the cognitive value it has in a language community (knowledge universally shared by the members of a language community versus additional information), and in doing so he is building upon Kant s original intention in making this distinction according to which there exist judgements of fundamentally different cognitive value with reference to a given level of knowledge. 21 When Quine said that no real distinction can be made between the two types of judgement he took no notice of the fact that, with respect to a given level of knowledge, the distinction is already perfectly clear. It would thus appear that we can interpret linguistic meaning of descriptive terms as something that is relative to knowledge: It is only with respect to given knowledge that something is informative (synthetic) or not. That is to say, we always have to take into account the given cognitive horizon (of a certain person or language community) before we are able to say something about the merely clarificatory or truly knowledge-extending character of certain cognitive elements. Therefore we can conclude: concepts demarcate, and factual criteria of demarcation can become components of the meaning of concepts. One might even say that it is especially the most important factual insights into something that are later essential to the defining characterization of a thing. If we see this

100 BETWEEN MEANING AND DEMARCATION 85 connection, then our position towards questions concerning conceptual features gets to be quite important factually, as they enter into factual questions. Therefore, linguistic meaning and factual demarcation cannot be separated and, with this, questions of meaning obtain a far greater factual importance than Popper was willing to concede to them. One might just as well say that in the discussion of conceptual questions we touch on factual questions too and, therefore, they are not fruitless. In the case of empirical concepts this may be relatively unproblematic, but it turns out to be rather far-reaching in the case of epistemological and thus philosophical concepts, because this bears on the question of the possibility of metaphysical knowledge. 3. THE CRITERION OF FALSIFIABILITY ONLY GIVES A POSITIVE CHARACTERIZATION OF EMPIRICAL SCIENCE AND THEREFORE DEMARCATES METAPHYSICS ONLY NEGATIVELY. Popper s criterion of demarcation was intended to avoid problems in connection with inductive logic. It should enable us to admit a system as empirical or scientific only if it is capable of being tested by experience 22. Popper therefore proposed as a criterion of demarcation not the verifiability but the falsifiability of the system concerned. To Popper s mind it is not important that the system shall be capable of being singled out, once and for all, in a positive sense, but one has to demand that its logical form shall be such that it can be singled out, by empirical tests, in a negative sense: it must be possible for an empirical scientific system to be refuted by experience. 23 For such statements, or systems of statements, convey information about the empirical world only if they are capable of clashing with experience; or more precisely, only if they can be systematically tested, that is to say, if they can be subjected 24. For Popper, the problem of demarcation consists primarily in delimiting scientific complexes of statements from other non-scientific complexes through a characterization of empirical science 25. At the same time, the delimitation from metaphysics is also important to Popper. But he only clarified positively what the characteristic feature of scientific statements is. Thus he gives us nothing more than a negative demarcation from everything else, which is non-science. Whether this is a demarcation from metaphysics or from something else is so far not clear. So it remains to be clarified from what Popper actually wanted to demarcate the scientific kind of knowledge.

101 86 HANS JUERGEN WENDEL 4. A CLOSER ANALYSIS OF THE CRITERION OF DEMARCATION SHOWS THAT IT IMPLICITLY CONSISTS OF TWO INDEPENDENT DIMENSIONS OF DEMARCATION. A closer examination of the criterion of demarcation reveals that one can make a distinction that Popper has not made explicit, but that seems to form the implicit basis of his considerations and that allows us to make a multiple demarcation and thus to say something about the peculiarity of the other non-empirical kinds of judgements. There are two main characteristics of statements of empirical science: (1) first of all their logical form of refutability and (2) moreover refutability by experience. That is to say, empirical scientific (systems of) statements must not only be refutable in general but refutable empirically in the sense of methodical testability. Therefore the criterion of demarcation has, strictly speaking, two independent dimensions which Popper does not keep apart, thus incurring a risk of confusion. Popper s criterion of falsifiability is usually referred to as the criterion of demarcation of scientific statements, although falsifiability as a logical property does not delimit science from metaphysics. For Popper unites two aspects in the criterion of falsifiability: one logical, the other having to do with highlighting a particular basis of validity, i.e. experience. That is to say, falsifiability, on the one hand, deals with the logical structure of statements and the logical relation of classes of statements and, on the other, with methodologically characterized statements of these classes that refer to experience in the sense of empirical testability. 26 It is, however, extremely important to keep apart both aspects, the logical one and the one that characterises the basis of validity, because these are two different criteria. It is therefore called for that we take a closer look at the demarcatory capacities of the two dimensions: (1) On the logical form of refutabilty (falsifiability as a logical relation between classes of statements): Scientific statements and systems of statements, in particular are essentially universal statements (universal if-then statements); from them one can deduce certain negative existential statements (existential bans); hence they can conflict with certain statements by virtue of their logical form. 27 It was exactly this feature that made the criterion of verifiability problematic. Another consequence is the logical asymmetry between verifiability and falsifiability 28 that allows us by logical means of the modus tollens to draw inferences from particular statements accepted as true to the falsity of a universal statement. 29 This logical relationship between kinds of classes of statements initially has nothing to do with the fact that empirical statements have this feature. It is only in the ensuing examination which is guided by the interest in scientific statements that the investigation is confined to this subclass of statements, which is

102 BETWEEN MEANING AND DEMARCATION 87 characterized by this relationship. It is typical of statements in empirical science that they belong to the classes of statements that stand in this logical relationship. If by falsifiability in general we understand this logical relationship, then the same fact might also be expressed as follows. Falsifiability in general and empirical falsifiability do not coincide. The first characteristic of scientific statements is therefore that by asserting something they simultaneously exclude something. Hence informativity always entails the possibility of refutability in a case where what was excluded by the assertion is for whatever reason admitted, the assertion must be rejected. One might call this the logical aspect of falsifiability. 30 But this is not an exclusive property of empirical statements alone. (2) On empirical falsifiability (reference to experience): Popper demands not only falsifiability in the sense of the mere logical possibility of refutability, but also in the sence of refutability by experience, empirical falsifiability. That is, scientific statements have to be such that they exclude not just any statements, but certain statements about our experience basic statements whose truth would imply the falsity of the respective scientific statement. 31 The aspect of refutability by experience gives the statements empirical informative content and makes them empirically scientific. Demarcation is thus based on the logical property of refutability falsifiability (or non-falsifiability) in general, on the one hand, and on the particular kind of refutability by experience, the method of empirical testability, on the other. For only the two taken together define the realm of empirical science. 32 Hence what makes any (informative) theoretical system empirically scientific is this reference to our methodical experience. Testability by experience appears as a distinctive method whereby one theoretical system may be distinguished from another 33 Given this positive characterization of empirical scientific (systems of) statements as empirically falsifiable, we might be able to say something about the other systems of statements from which scientific statements are delimited respectively by the two dimensions of the criterion of demarcation. 1. The logical property of falsifiability delimits analytic from synthetic judgements. Synthetic judgements metaphysical as well as empirical are formally delimited by the common logical property of falsifiability (the logical possibility of being refuted). Therefore, the characteristic of falsifiability (testability) has to be regarded as a feature of all informative (synthetic) judgements. We also find this relationship in the case of metaphysical judgements, as in the case of empirical statements, where negations of metaphysical judgements also define contents that are excluded by them in the case of non-empirically testable (metaphysical) statements. Thus, for instance, metaphysical determinism (every event has a cause) excludes something (there is no event that has no cause); an uncaused free event of volition even though it is not empirically testable, would at least be something that is subject to this ban. That is to say, the fundamental logical asymmetry between verifiability and falsifiability, between

103 88 HANS JUERGEN WENDEL logical provability and logical refutability, also exists in the case of nonanalytic metaphysical statements. Consequently, falsifiability distinguishes informative judgements from contradictions and tautologies and thus determines their synthetic character. 2. In opposition to this sole criterion of demarcation between empirical and metaphysical statements (which are falsifiable as well) is empirical testability (i.e. refutability though the property of being related to experience) which, however, does not coincide with falsifiability as a logical property. 34 The difference between metaphysical and empirical statements is therefore solely due to the methodical difference in the type of empirical and non-empirical testability, it being an open question as to what might be regarded as testability in various cases. Since empirical testing is determined by methodology, there can also be criteria for the testing of metaphysical statements. 35 If content is determined by the degree of falsifiability, then content and empirical content do not coincide, i.e. there are non-empirical but by virtue of their logical form nonetheless falsifiable (synthetic) statements that have to be considered as rich in content even though they are empirically empty. Thus we have to distinguish between testability and scientific testability. Which criteria of testability would have to be applied for metaphysical as well as for scientific considerations, respectively, is another question. 5. ANALYSED IN ITS DIMENSIONS, THE CRITERION OF DEMARCATION LENDS GREATER PRECISION TO KANT S STATEMENT WITH REGARD TO THEIR COGNITIVE VALUE AND THE FOUNDATION OF KNOWLEDGE. If we consider both dimensions of demarcation separately, it follows that (a) with the logical aspect of falsifiability or, minimally, the logical possibility of refutation, Popper is basically giving us a reformulation (and in some respects a reformation) of the Kantian distinction between analytic and synthetic judgements, for the criterion of falsifiability in the general sense of refutability delimits empirical from non-empirical judgements but only from those nonempirical judgements that are, in addition, analytic. So on closer examination it turns out to be rather a criterion of demarcation between analytic and synthetic than between metaphysical and empirical judgements. 36 (b) Furthermore we find that the explication of experience as scientific, methodically obtained knowledge (objective testability) is a reformulation of Kant s distinction of the foundations of the validity of knowledge, the distinction between validity a priori and validity a posteriori though (this being how it differs from Kant) not of experience in general but of scientific experience. However, only validity a priori is positively characterized in this way. Popper reconstructs validity by virtue of scientific experience with the rules of the game empirical science as their internal logic 37.

104 BETWEEN MEANING AND DEMARCATION 89 But even if falsifiability must be regarded as a logical property of synthetic judgements that results from their informative character and Popper in the main reformulates a criterion for synthetic, i.e. ampliative judgements something else becomes apparent that Kant had considered to be negligible. In making his distinction Kant was only looking at true judgements, whereas Popper takes into account that when making judgements we never know whether they are true. By making falsifiability the criterion of demarcation, Popper therefore connects two insights: (1) the distinction between informative and non-informative judgements and: (2) the independence of informativity and truth. Therefore, all synthetic judgements as cognitive claims can also be false without our being able to know it. Syntheticity informational content of judgements and their truth are thus independent properties. That is to say, informativity and possible falsity must be able to coincide. Hence, synthetic judgements are not ipso facto extensions of knowledge as Kant assumes, but are only possible extensions of knowledge. A synthetic judgement is an extension of knowledge only in so far as it is true. This property of being merely a possible extension of knowledge finds expression in the logical structure of synthetic judgements: Syntheticity or informativity always means: there exist logically possible classes of events that are incompatible with the assertion in question. With the insight that synthetic judgements are always possible extensions of knowledge but that this is not guaranteed, synthetic character (informational content) and falsifiability coincide both for empirical and metaphysical statements. As a result of the foregoing considerations, we can conclude that: 1. Meaning and factual demarcation do not contradict each other.

105 90 HANS JUERGEN WENDEL The criterion of demarcation can well be interpreted as a criterion of meaning. There is no conflict here if one takes into account the connection between meaning and cognition. Clarifications of concepts originate in factual judgements (in cognitive claims). Philosophical and especially epistemological concepts are therefore such that they are also based on cognitive (epistemic) claims, albeit not to empirical knowledge. 6. INASMUCH AS EPISTEMOLOGY ELUCIDATES CONCEPTS, THE CONNECTION BETWEEN THE CLARIFICATION OF CONCEPTS AND JUDGEMENTS OF KNOWLEDGE MEANS THAT THERE MUST ALSO BE A DIMENSION OF COGNITIVE VALUE IN EPISTEMOLOGY. In conclusion, let me return to the appreciation of epistemological considerations within the framework of a science-oriented philosophy as professed by Schlick. Summing up our deliberations so far we arrive at the question of what this means for the question as to what extent epistemology (or philosophy as a whole) possesses a cognitive claim. The modification of Popper s considerations on the criterion of demarcation allows a more precise determination of the peculiar characteristics of metaphysical judgements. To my mind, they also permit a critical investigation of the farther-reaching assertion that philosophical reflection is nothing more than a mere analysis of concepts. If we consider the epistemological problems that occupied Schlick from the beginning, we realize that they concern the philosophical elucidation and justification of the empirical sciences cognitive claim, especially the understanding of the conceptual apparatus we need for the description, elucidation and explanation of all that can be termed knowledge or scientific cognition. 38 Along with the basic syntactic and semantic concepts of meaning, reference, fulfilment and truth, these would include the conceptual instruments of epistemology like existence, cognition, experience and reality or the question as to the aim of knowledge. In his Allgemeine Erkenntnislehre Schlick still regarded the elucidation of problems related to epistemological realism as factually meaningful and important, questions which under the influence of Wittgenstein he later came to consider as questions of merely conceptual clarification to which no synthetic cognitive value is attached. The connection between the extension and the mere clarification of knowledge as described above should have made it clear that it does not help if, in the treatment of philosophical problems, one withdraws to the position that the elucidation of epistemological questions is nothing more than an analysis of concepts. For analysis, as we were able to see, refers to previous synthetic acts of

106 BETWEEN MEANING AND DEMARCATION 91 judgements except in the case of abbreviations introduced for the sake of economy of language If what Schlick says about concepts in general is also true for epistemological concepts, then definitions and judgements must in principle refer to each other in this field, too. This can only be avoided if we are willing to go as far as Wittgenstein and deny that epistemological questions have any cognitive value. Then, as a consequence, there can no longer be philosophical concepts or definitions thereof. Cognition must then really be replaced by pointing. Beyond verificationism there is, however, no longer any need to do this, especially since this assumption about the nature of philosophical considerations is a presupposed metaphysical thesis. Hence those who are unwilling to banish philosophy completely to the silent world of pointing, and who concede to philosophy at least the status of concepts we must presuppose for the investigation of knowledge, we cannot avoid conceding a status of knowledge to them as well. To my mind, the proposed improvement or modification of the basic concept of falsificationism gives us a good starting-point for the treatment of epistemological considerations, especially since it is also an alternative to verificationism. The latter should be acceptable to a scientific philosophy if one gives up Popper s pretence to a methodology that aims to standardize scientific method in a way that is probably incompatible with what we know about scientific progress. Epistemological judgements and concepts can thus also be regarded as rich in content (factual), even though they are not empirically factual because they are not empirically falsifiable. This, of course, raises the question as to whether we are not reanimating a synthetic apriori though this approach. As far as I can see, this cannot be claimed in the Kantian sense because, for him, a synthetic apriori would have to be an apodictic judgement that must necessarily be decidable by pure reason. Metaphysical judgements as judgements which are not justifiable by experience but formulated with a cognitive claim can only be tested in view of their explanatory capacity as conditions of empirical knowledge in the light of alternatives and in view of their consistency with other philosophical considerations. Therefore, they are to be regarded as metaphysical hypotheses. An acceptable system of philosophical concepts can only be one that is as complete and coherent as possible and that also integrates all philosophical solutions to problems in a consistent manner.

107 92 HANS JUERGEN WENDEL NOTES Karl Popper, Logic of Scientific Discovery (LD) (orig. Logik der Forschung (1935)), New York 1965, Ch. I. Paradigmatic for this are especially Rudolf Carnap s expositions in: Scheinprobleme in der Philosophie (1928), repr. Frankfurt am Main Logic of Scientific Discovery, p. 40 fn. *3. Cf. K.R. Popper, Conjectures and Refutations. The Growth of Scientific Knowledge (1963), especially part XII of the introduction and passim, as well as Objective Knowledge. An Evolutionary Approach, rev. ed. Oxford (Clarendon) 1979, p. 309ff. und p. 122ff. Cf. (together with the works quoted in fn. 4 above) Three Views Concerning Human Knowledge, in his Conjectures and Refutations, p. 97ff. Conjectures, p. 20. Conventions need not be the result of explicit stipulations as claimed by Carnap for postulates of meaning in artificial languages. For natural languages this will, rather, be the exception. Here we are dealing with conventions in the sense of historically crystallized habits in the use of language that were never defined explicitly. David Lewis tries to interpret conventions as the result of problems of coordination (cf. D. Lewis, Konventionen). But the basic idea in this is also: It is a platitude that language is ruled by convention. Words might be used to mean almost anything; and we who use them have made them mean what they do because somehow, gradually and informally, we have come to an understanding that this is what we shall use them to mean. We could perfectly well use these words otherwise. (Lewis, Konventionen, p. 1) Conventions are regularities of behaviour (cf. Lewis, Konventionen p. 59). Conjectures, p. 20. Popper, Objective Knowledge, p Popper regards the position of essentialism as an erroneous view commonly held in the history of philosophy. Cf. Popper, Objective Knowledge, p. 122f. Objective Knowledge, p. 309 f. Objective Knowledge, p Popper also devotes an entire chapter of his autobiography to this question. Cf. Popper, Autobiography in: Paul Artur Schilpp (ed.) The Philosophy of Karl Popper, Book I, La Salle, Illinois (Open Court), Ch. 7: A Long Digression Concerning Essentialism: What Still Divides Me from Most Contemporary Philosophers, p Kant, Critique of Pure Reason, A 8 ( Begriff, den ich schon habe ). Kant says that this distinction is in Ansehung der Kritik des menschlichen Verstandes unentbehrlich and must be regarded as klassisch. (Kant, Prolegomena, 3). Critique of Pure Reason (1781/1787), B 130 (my italics): Man wird hier leicht gewahr, daß diese Handlung ursprünglich einig, und für alle Verbindung gleich geltend sein müsse, und daß die Auflösung, Analysis, die ihr Gegenteil zu sein scheint, sie doch jederzeit voraussetze; denn wo der Verstand vorher nichts verbunden hat, da kann er auch nichts auflösen. Moritz Schlick, General Theory of Kowledge, Wien-New York (Springer) (orig. Allgemeine Erkenntnislehre, Wien (Springer) 1918), p. 45. The dynamic element in Schlick which changes the content of concepts was also pointed out by Haller in: Erkenntnisprobleme bei Moritz Schlick, in: Zurück zu Schlick. Eine Neubewertung von Werk und Wirkung. (ed.) Brian McGuiness, Vienna (Hölder-Pichler-Tempsky) 1985, p. 35. Word and Object, Cambridge, Mass. 1960, ch. II. Cf. my analysis in: Moderner Relativismus, Tübingen (Mohr Siebeck) 1990, Ch. II. Cf. ch. IV: Wissen und Sinn of my book: Benennung, Sinn, Notwendigkeit. Eine Untersuchung über die Grundlagen kausaler Theorien des Gegenstandsbezugs, Frankfurt am Main (Athenäum) 1987, p Quine (Word and Object, 9, p. 38) himself says that one might interpret the situation after the aquisition of some additional knowledge C as change in meaning of a sentence S. We could ask: Now couldn t we just as well have said, instead, that on acquiring C, men have found it convenientimplicitly to change the very meaning of S?

108 BETWEEN MEANING AND DEMARCATION With the difference that this is referred to subjectivity according to Kant, which was replaced by Quine with the respective language community. Interesting in this context are remarks made by Dudley Shapere about changes in scientific language. According to Shapere, changes in science also involve the building of well-founded information into our descriptive vocabulary (Shapere, Reason, Reference, and the Quest for Knowledge, Philosophy of Science 49, 1982, p. 14). We have a process of gradual reformulation, in the light of what we have learned (p. 13). This is a process of internalizing of relevant considerations (ibid.). Logic of Scientific Discovery, p. 40. Logic of Scientific Discovery, p. 40. Logic of Scientific Discovery, p. 313f. The problem of demarcation is what I call the problem of finding a criterion by which we can distinguish the statements of empirical science from non-empirical statements. Popper, Con- jectural Knowledge; My Solution of the Problem of Induction (1979) in ch. 1 of: Objective Knowledge, p. 12fn. 19. That we are dealing with two aspects here becomes clear when Popper explicitly talks about empirical falsifiability, something that would only be analytic if falsifiability were ipso facto empirical falsifiability. Just because Popper is always concerned with empirical falsifiability he mostly uses both terms interchangeably. That is why nomological statements as we find them in the sciences not only assert something but simultaneously exclude something. So, for instance from the universal if-then-statement All planets describe orbits around the sun, there follows the negative existential statement There exists no planet that does not describe an orbit around the sun ; i.e., such nomological statements exclude the occurrence of certain events. Logic of Scientific Discovery, p. 41. By taking recourse to the Aristotelian apagoge, Schopenhauer had already pointed out that one can reap the benefits of this asymmetry in order to oppose potential falsifiers to statements. Schopenhauer said that against an universal statement one needs nur einen einzigen Fall aufzustellen, zu dem der Satz nicht paßt, und derselbe ist umgeworfen: ein solcher Fall heißt Instanz [...] Z.B. der Satz: Alle Wiederkäuer sind gehörnt wird umgestoßen durch die einzige Instanz der Kameele. Die Instanz ist ein Fall der Anwendung der allgemeinen Wahrheit, etwas unter den Hauptbegriff derselben zu Subsumirendes, davon aber jene Wahrheit nicht gilt und dadurch ganz umgestoßen wird. Schopenhauer, Eristik, in: Aus Arthur Schopenhauer s handschriftlichem Nachlaß. Abhandlungen, Anmerkungen, Aphorismen und Fragmente. Herausgegeben von Julius Frauenstädt. Leipzig 1864, p. 26. There we also find the remark that even potential falsifiers are not certain. In the quoted passage Schopenhauer continues: Allein dabei können Täuschungen vorgehen. This aspect of refutability (the existence of the logical possibility of refutation) coincides with the informative content ofa statement. That is to say, informativity always involves falsifiability. Experience in this context always means methodically acquired experience (cf. Logic of Scientific Discovery, p. 53ff.). By empirical we therefore understand was durch die empiristische Methodologie, durch die Theorie der Erfahrung, erst genauer zu präzisieren ist (Popper, Die beiden Grundprobleme der Erkenntnistheorie, Tübingen (Mohr Siebeck), 1979, p. 360). Therefore, there is a need for an independent elucidation of the problems concerning the empirical character of singular statements (the basic statements) Logic of Scientific Discovery, p. 43. For their scientific character cannot be grounded in their empirical falsifiability alone. This difference seems to be what Popper is emphasizing when he says that empirical science seems to be characterized not only by its logical form but, in addition, by its distinctive method (Logic of Scientific Discovery, p. 39), the reference to experience through methodical tests. Logic of Scientific Discovery, p. 39. That the logical characteristic of falsifiability alone the feature of refutability is in general not sufficient as a criterion for the characterization ofempirical science is pointed out by Popper in connection with the difficulty, that there must be many theoretical systems with a logical structure very similar to the one which at any particular time is the accepted system of empirical science (Logic of Scientific Discovery, p. 39). That is to say, we are able to give a factual

109 94 HANS JUERGEN WENDEL description of distinct possible worlds which as descriptions really exclude something. But it is not sufficient that statements are informative (and hence exclude something); they must moreover be empirically testable (exclude something empirical), for only then are they of the kind that describes our world. Not the synthetic character the informative content alone is thus decisive, but also the types of possible instances of testability. That is to say that a scientific system that intends to describe our world should not only be informative but also true; at least it should claim to be true. Therefore, what distinguishes empirical science as a form of knowledge from other forms of knowledge is methodically obtained experience, objective empirical testing. Later, Popper also thought about possible ways of testing metaphysical considerations. On this cf. On the Status of Science and of Metaphysics (1958) in: Conjectures and Refutations, especially p For only with analytic statements can one show in a way which can be decided by logical analysis that for such statements there exist no possible falsifiers. When Popper, for instance says: the statement, It will rain tomorrow will not be regarded as empirical, simply because it cannot be refuted (Logic of Scientific Discovery, p. 41), the demarcation is only due to the fact that the logical form of a system shall be such that it can be singled out by means of empirical tests in a negative sense (Logic of Scientific Discovery, p. 41; my italics). The aspect of testability by experience does not play any role here at all. Another example would be a statement like: Entrepreneurs have productive factors at their disposal. If one takes into account that an entrepreneur is by definition someone who has the productive factor capital at his disposal, then it follows that this statement is analytically true and insofar not empirically scientific. But in the same manner, one could delimit informative metaphysical from merely metaphysical elucidations of concepts. As is well known, even Kant pointed out that many supposed instances of metaphysical knowledge are nothing but concealed analyses of concepts. Logic of Scientific Discovery, p. 53. Cf. also Haller, Erkenntnisprobleme bei Moritz Schlick, p. 34. Institut für Philosophie Universität Rostock Rostock Germany hans-juergen.wendel@t-online.de

110 DAGMAR BORCHERS LET S TALK ABOUT FLOURISHING! MORITZ SCHLICK AND THE NON-COGNITIVE FOUNDATION OF VIRTUE ETHICS 1. LET S TALK ABOUT VIRTUES! My intention here is to make a case for the moral philosophical work of Moritz Schlick which, in my opinion, has not been adequately appreciated in the discussion between two different groups within analytic philosophy, i.e., those who prefer virtue ethics and those who do not. One aspect of their controversy is the rating of the virtues within analytic moral philosophy. Protagonists of virtue ethics like Philippa Foot claim that the analysis of virtues has been neglected for a long time, mainly for methodological reasons, and that the necessary revival must be seen as starting something totally new within the tradition of analytic ethics. But, keeping Schlick in mind, it seems as if there is no need for a revival a promising start has already been made. The justification of my claim proceeds by the following steps: The first step presents Foot s argument and my own. The second step concentrates on Foot s historical error. Referring to the ethical studies of Moritz Schlick, I want to show that contrary to a widespread assumption there was a profound interest in the relationship between happiness and moral practice within the Vienna Circle. The third step will reveal that Foot s attemption is itself flawed and, therefore, does not offer a promising route for those who want to do virtue ethics. The fourth step finally takes us in the right direction, i.e., to Moritz Schlick s inspiring influence concerning the perspectives of virtue ethics within analytic philosophy. This is our program. Let s start with Foot s argument. Foot s Argument There is an increasing number of moral philosophers who have noted a lamentable neglect of virtues within analytic moral philosophy. One of the most vehement pioneers of this development within analytic ethics has been the English 95 F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

111 96 DAGMAR BORCHERS philosopher Philippa Foot. In her well-known paper Virtues and Vices she makes the following claim: For many years the subject of the virtues and vices was strangely neglected by moralists working within the school of analytic philosophy. The tacitly accepted opinion was that a study of the topic would form no part of the fundamental work of ethics; and since this opinion was apparently shared by philosophers such as Hume, Kant, Mill, G. E. Moore, W. D. Ross, and H. A. Prichard, from whom contemporary moral philosophy has mostly been derived, perhaps the neglect was not so surprising after all. 1 My Argument In referring to the moral philosophical work of Moritz Schlick, I want to show the following: (1) Foot s interpretation is historically false. It can be shown that there is a deep interest of philosophers of the Vienna Circle in the virtues and especially in questions concerning possible ways of flourishing. One of the most prominent example is Moritz Schlick. In his ethical work we can find a kind of Neo- Humean virtue ethics. It is based on (a) noncognitive meta-ethics and (b) a descriptive theory of action. (2) This analysis is methodologically false. The background of this misinterpretation is a fixed opinion on how a virtue ethics conception should look. When elaborating her own position, Foot assumes that without criticizing (a) and replacing it by an cognitive view and criticizing (b) and replacing it by a unified conception of practical reason, there will be no working virtue ethics. I will show that her arguments in favour of this project are not convincing. In fact, both the cognitive meta-ethical basis (a) and the non-humean theory of action (b) turn out to be flawed. Furthermore we will see, that with regard to the content of the theory of the virtues, there are great similarities between Schlick and Foot, despite their differences in (a) and (b). The result of this will be the following. Whoever wants to convince other moral philosophers of the outstanding potential of virtue ethics has to present an elaborated theory of the virtues. As we see, a practicable start has already been outlined by Moritz Schlick. The consequence will be the quest for an elaborated noncognitive virtue ethics build upon the foundations laid by Hume and Schlick. The prospects for such a project are promising. I will conclude with a number of proposals on how such a theory could look. But the next step must be to correct the historical assumptions made by Foot.

112 LET S TALK ABOUT FLOURISHING! THE HISTORICAL CORRECTION OF FOOT S THESIS: SCHLICK AND THE VIENNA CIRCLE Ancient Ethics and the Vienna Circle Anyone familiar with the work of the members of the Vienna Circle knows that, although they rejected the possibility of talking scientifically about normative questions, they were nonetheless very interested in moral problems. Even radical noncognitivists like Carnap or Wittgenstein were deeply convinced that ethical considerations formulated as questions How should I live? What sort of person should I intend to be? are the most important ones in our life. In answering them, we do not have to put up a catalogue of moral duties or complicated abstract considerations. These are not questions that call for a theory. Instead, we are asking for practical advice, looking for examples. As Manfred Geier has pointed out, this anti-theoretical approach to ethical problems, together with the deep conviction that moral obligations must be compatible with our inherent search for happiness clearly prove that members of the Vienna Circle could identify with ancient moral philosophy:

113 98 DAGMAR BORCHERS When Karl Popper spoke about morality and ethics, he referred to Socrates. Confronted with a threatening fate, Ludwig Wittgenstein turned to Stoicism for orientation in his intellectual efforts. Schlick responded to ethical questions as an eudemonist whose primary interest was pleasure, taking recourse to Socrates, Aristippus and Marcus Aurelius. Otto Neurath s holistic social ethos revealed shades of an Epicurean happiness where a certain joie de vivre figured centrally.2 So we have to stick to the fact that there is at least, in principle, no gap between ancient virtue ethics and the outlook on morals of the first generation of analytic philosophers on the contrary, there seems to be a certain affinity. Moritz Schlick and the Question for Morals and Happiness One prominent example is Moritz Schlick. Although Schlick was not concerned with a foundation of morality in the quasi-scientific, theoretical way ethics is done today, we should now be permitted to take an analytical look at the formal structure and the metatheoretical implications of his ethical theory. Accordingly, his conception, reconstructed as it is put forward in his two studies in ethics 3, is based on two columns: noncognitive meta-ethical assumptions and a descriptive theory of action. (a) A noncognitive meta-ethics: The background of his conception is an empiricist epistemology that acknowledges only two forms of cognition. Sensible statements are either analytic a priori or synthetic a posteriori. In the first case, cognition may be seen as undoubtable, but it will tell us nothing about the real world. In the second case, we will have a cognition about the world, but it will have only hypothetical character. Statements that are neither analytic a priori nor synthetic a posteriori must be regarded as senseless. Senseless statements are pseudo-statements; they can t even be true or false. Normative judgements belong to the class of statements that are neither analytic priori nor synthetic a posteriori. Being senseless in principle, they do have two characteristics: First, they do not depict a cognition but an emotion, a desire or an interest. Second, they are not descriptive but prescriptive. This means that a scientifically respectable moral philosophy can only be done as a descriptive ethics that refrains totally from normative statements. (b) A descriptive theory of action: According to Schlick, the main task of ethics is to deliver a causal explanation of moral behaviour. Before it is possible to answer the question Why does man act morally? it must be clear why he generally acts to begin with, how human action works. The premise of the analysis is that each act is constituted by an inner disposition ( natural or artificial desires ) and an outer effect ( motive ). The result of Schlick s analysis is the so-called law of motivation, which states that the motive that turns out to be the strongest one will determine the action. Taking the action of drinking water as an example, thirst will be a desire; the sight of water the motive. If the thought of drinking is the most powerful one among other competing ideas, it will there-

114 LET S TALK ABOUT FLOURISHING! 99 fore be decisive for the action itself. Reasoning has no direct influence on action. Only insofar as it creates a desire that turns out to be the most craving one does thought direct action. Principally, action is based on the pursuit of pleasure. The latter is synonymous with the pursuit of happiness, even if we can t explain in detail what happiness means. All we can say is that it does not only include actual pleasure, but also the hope for future enjoyment. The art of life consists in the maximazation and harmonization of both components. What are the implications for morality and moral action? (c) Schlick s virtue ethics: Contrary to Foot s assumption, we will discover a profound virtue ethics-conception within the work of Schlick, focused on the interrelation of morals and happiness. The first thing we have to understand is: How is morality possible in general? Morality will only have a chance to work if it is directly rooted in the structure of human needs. Moral motivation will be solid, if refering to our law of motivation the pleasures of the imagination of moral action correspond to the pleasures of the realization of moral action itself. In other words: moral action must be a real pleasure. Fortunately, we are able to say: Virtue and happiness are compatible. That we are capable of maintaining a certain harmony between virtue and happiness is due to a widespread experience. The common roots of virtue and happiness are the social desires (as part of the natural desires) we find in every individual. Without the company of other people, happiness is hardly imaginable. But happiness has its price. We are urged to shape our desires in a way that makes us sociable. Experience shows that being virtuous is the best way to happiness. An unfortunate constellation of desires will prevent the enjoyment of life. This is why it proves necessary to tame some of them and to strengthen others. This kind of education and practice aims at the harmonization of the desires and the development of our character. To be virtuous means to act morally without reasoning, beyond the pressure of sanctions or conscience. The more virtuous one is, the easier it is to do the right thing. The most important moral virtue is justice; other important moral virtues are benevolence and friendship. All these virtues will make happiness more probable, but not certain. Experience shows that chance and fate, as well as the amorality of others, may thwart the pursuit of happiness and limit the radius of the virtues. But, all things considered, they are the best recipe for flourishing. What about the epistemological status of virtues? Moral virtues represent the experience an individual or a society (as a group of individuals) has made with different kinds of human behaviour. To say that something is good or should be done expresses the positive experience society has had with this special kind of action. Based on social experience, moral norms and the relevant virtues are principally objects of modifications and change over years and centuries. Generally, acting morally is defined as being the opposite of acting thoughtlessly. For Schlick, there is only one fundamental moral insight Be ready for happiness or, more elegantly, Be inspired : Whoever is able and ready to share in the joys of life is invited to them. 4

115 100 DAGMAR BORCHERS Characteristics of Schlick s Conception Now, what are the results of this investigation of Schlick s virtue ethics? We can say that his conception is... Naturalistic; as it is willing to integrate the explanation of human action and the general function of morality into a biological description of human behaviour and the world; Empiristic; as it works on the basis of empirical theories and refers to common social experience for its explanations of human behaviour and the content of certain moral principles; Consequentialistic; as it is embedded in an neo-humean consequentialism criterion of moral actions and principles are their (good) consequences for agents and their surroundings; Noncognitive; as it is based on a noncognitive meta-ethics and as it refuses to impose moral norms or a binding set of virtues; although it is at the same time able to provide a general moral principle morality has to improve our lives. And it fulfills two central claims of a real theory of virtue ethics. Schlick puts the virtues in the centre of his conception and formulates the decisive moral questions the way virtue ethics does: How should I be? and How should I live? Its fundamental question is not What is demanded of me? but How must I live to be happy? It has its source in the desires of the individual, of the agent himself, and thus bears the character not of demand but of desire. 5 Conclusion For this reason, it is absolutely puzzling to hear Foot say that the topic of the virtues has been neglected ever since within analytic philosophy and, moreover, that this neglect has programmatic causes. By putting the stress on happiness and pleasure, Schlick resolutely turned against modern justifications of a pure, categorical ethics of obligation which excludes the tendencies of empirical subjects to seek happiness and to make them responsible for the tender death (euthanasia) of all morals. Instead of adhering to the rigorous Kant and other advocates of an abstract morality based on imperatives, Schlick recalls the ancient art of life. 6 Hence it seems as if, according to Foot, only the fulfillment of very special criteria guarantees that a virtue-ethical theory is real. So let s have a look at what she herself proposes.

116 LET S TALK ABOUT FLOURISHING! THE METHODOLOGICAL CORRECTION OF FOOT S THESIS: FOOT S CONCEPTION AND ITS PROBLEMS Foot s Conception So why are Hume and Schlick not considered moral philosophers doing virtue ethics? The reason is not the disapproval of (c), but the disapproval of (a) and (b). Foot claims that without abolishing (a) and (b) there will be no satisfying virtue ethics. Criticizing (a) means submitting a cognitive meta-ethical basis. Its two decisive arguments are, first, a semantic argument and, second, the conception of Aristotelean necessities (which originally had been developed by E. Anscombe). Moral judgements can qualify as objective insofar as the semantic features of the word good and the word man do provide the decisive criteria for a good person or a good action. Generally, the relevant criteria of what it means to be good are not up to the speaker, but are determined by the function of the object itself the descriptive meaning of good is variable, not a matter of choice. As the purpose of man is in principle explicable, there exists a chance to determine criteria for a good action or a good person. The characteristics of a flourishing human life determine the purpose of man and apply a strict standard to the good which, in principle, could be recognized by any intelligent agent. For Foot, it is obvious that a behaviour that undermines the fulfillment of these needs can never be called a good one. Moral procedures are part of the pursuit of pleasure, the fulfillment of human desires. Without these procedures, leading a good life would be impossible. All in all, someone who does not understand that being moral is as necessary for human beings as hunting in packs is for wolves, is suffering from a cognitive deficit. Criticizing (b) means submitting an anti-humean theory of action. Foot accepts the fact that moral judgements are practical they aim to direct our actions. Nevertheless, in her eyes the noncognitive reconstruction of moral judgements is flawed: in fact, a moral judgement does not depict a desire or a certain interest, but a cognition. The foundation of a moral statement is an insight into the necessities of human life or other causal relations. The recognition of the brute facts of human nature provides reasons for moral acting. The statement Torturing children is bad is based on the cognition that such a behaviour hurts the child mentally and physically and will in no way contribute to a good human life. It is the cognition that is basic, so there needn t be added a special interest in the child s not being beaten. Men act for reasons. Much of what has to be said here will depend on a feature of the operation of human beings to which Aquinas drew attention, when he said that while animals perceive things that are good and go for them, human beings go for what they see as good. [ ] In fact the

117 102 DAGMAR BORCHERS truth and great importance, of what Aquinas said comes out in the recognition of the whole conceptual structure belonging to the idea of human action and desire; as for instance the fact that they can say why they want something, can conduct practical reasoning, and defend a choice. 7 (c) Foot s virtue ethics: Both components together motivate her special conception of the virtues (c). Acting virtuously means doing the right thing for adequate reasons. The function of the virtues is to correct our motives - supporting those that are too weak and limiting those that are too strong. They adjust motivational deficits and enable us to do the right thing each one standing at a point at which there is some temptation to be resisted or deficiency of motivation to be made good. 8 A life without virtues will not be a good one; there will be no flourishing without them: Nobody can get on well if he lacks courage, and does not have some measure of temperance and wisdom, while communities where justice and charity are lacking are apt to be wretched places to live, as Russia was under the Stalinist terror or Sicily under the Mafia. 9 Foot s Problems Foot s own conception claims to be the one that is based on adequate metaethical considerations and a well-founded theory of action. I argue that it is not. Let me briefly explain why. First, there are objections to Foot s cognitive meta-ethics. Foots s way of making the nature of human beings the basis of morality seems to be a methodologically daring enterprise. To be successful, one must be able to determinate the relevant characteristic features of human beings precisely and free of doubt in order to be able to say, in a second step, what kind of behaviour we should perform. But there is no reason to be too optimistic. Indeed, the facts about human nature and the necessities of the species do not provide the purpose of man and, furthermore, no standard of the good. First, we do have a problem of choice: It is not obvious which features will characterize our species. Without doubt it will only be our genes, but they do not give us any information as to how to act and what to do. Other features like the ability to speak, to think or to construct things will also apply to some mammals. Second, we have a problem of relevance; Even if it were possible to determine the relevant characteristics, their normative relevance would be questionable. Facts about our species do not lead to the consequence that we should act morally. They may also suggest that we follow an immoral strategy. In sum, then, since knowledge about human nature is not capable of providing definite normative insights, the claim that moral judgements provide cognition seems to be untenable.

118 LET S TALK ABOUT FLOURISHING! 103 Second, we also have some objections to Foot s theory of action, for nearly the same can be said about her conception of practical reasoning. First, reasonable thinking is not necessarily moral thinking. McDowell has pointed out 10 that a reasonable wolf will immediately see that hunting in packs may be alright for the others. He himself would be better off if he would not join the hunting but only the eating pack. Therefore, we do have the problem of immoral action often being a much better strategy for a happy life. Even if we concede that, all in all, life will be better in harmony with others, this insight will not prevent numerous minor immoral activities. Second, it is not knowledge that leads to action, but an interest: A cognition alone will not help us, as we can easily see in cases of giving up smoking or a diet. It is not knowledge that smoking will be destructive to our health, but feelings like fear or ambition. So we have to realize: Neither facts about human beings nor rational thinking prove that we should act morally. Although it is possible to support a moral standard through reasonable arguments or certain facts about humans, it must be clear that no one is forced to accept the standard by simply being human: [ ] it is one thing to say that by calling a creature a man we imply that he belongs to a species having certain capacities, and quite another thing to say that by so calling him we imply that he belongs to a species whose specific good is of a certain kind. [...] Similarly, if horse is used as a functional word, meaning charger, a horse that throws his rider becomes eo ipso a bad one; but the horse might say to himself I am not trying to be a horse in that sense; I am only a solid-hoofed perissodactyl quadruped (equus caballus), having a flowing mane and tail, and proceed to throw his rider without offence to anything but the rider s standards. [...] The horsebreaker s art would be easy, if one could turn horses into chargers by definition. 11 Thus we must conclude that, contrary to Foot s intentions, moral action still is an explanandum, not an explanans. Nevertheless, Foot s conception of the virtues is a very instructive one and there are no direct contradictions to Foot s virtue ethics as such. Although we will say that (a) and (b) are both defective, we can discover a lot of similarities between Foot and Schlick concerning their conceptions of morals and virtues. Both see morality as a source of happiness and virtues as promising tools for flourishing. Moral action must be a source of happiness, for otherwise it will not be possible. They both insist that the virtues are needed to shape our motives and enable us to react in a competent manner when confronted with a moral problem. That means, that conerning (c), the core of the virtue ethics-conception, they are very close to each other. My suggestion is to follow the path they sketch, but with the meta-ethical background provided by Schlick. So finally, let s have a look at the perspectives of virtue ethics.

119 104 DAGMAR BORCHERS 4. LET S TALK ABOUT VIRTUES THE WAY SCHLICK DOES! After all, it seems to be the most promising strategy to follow Schlick when choosing the foundations of virtue ethics. So I will conclude by presenting some metatheoretical suggestions. An adequate virtue ethics-conception should be Naturalistic: It turns out that the most fruitful background of this examination would be a naturalistic conception of the world. What does this mean? It means seeing the world as a continuous system that exists independently from human perception in time and space. The world as a whole and all its subsystems are the object of an universal evolution. In principle, it is possible to gain insight into the structure of world, how certain things in nature interrelate and how they work. For this purpose, the empirical sciences have proven to be the best tool. Morality has to be seen in the context of human evolution, too. It is part of the history of human evolution and human culture. As such, morality is a legitimate object of scientific research. For me, this is not a point of disagreement between Foot and Schlick as well as the following two items. Empiristic: Our moral life contains a vast amount of heterogeneous phenomena, of which the virtues form a subset. Thus, the examination of the virtues is embedded into a scientific project seeking to improve the understanding of human development. As a descriptive enterprise, ethics is well-advised to bear in mind the methodological standards of the empirical sciences. As an empiristic enterprise, it should be grounded in our social and moral experience and depict it. In addition, its results should at least not be incompatible with such widely accepted theories in anthropology, sociology and biology as, for example, the theory of evolution. Consequentialistic: Morality has to be seen as part of the human pursuit of happiness. An action can qualify as morally good if it is the one with the best consequences. The best consequences are those that will support the pursuit of happiness and minimize suffering. A character trait or a disposition can count as a virtue if it enables flourishing and supports personal or social welfare. Virtues allow for cooperation, shorten moral decision procedures and, all in all, make an indispensible contribution to the stability of morals. Their instrumental function makes them useful. At this point, Foot s inspiring remarks on the function of virtues should be taken into account. Noncognitive: Many moral philosophers have tried to develop a cognitive foundation of virtue ethics, for example Aristotle, Aquinas and, nowadays, not only Philippa Foot but also Martha Nussbaum, John McDowell and Alasdair MacIntyre. As I have shown elsewhere 12, none of these attempts is really convincing. This is why I vote for a noncognitive meta-ethics (though it s absolutely clear that this topic needs additional discussion), and it seems

120 LET S TALK ABOUT FLOURISHING! 105 to me that Schlick s version is a fitting one. A consequentialistic background will suffice to promote certain relevant virtues. There is no need for a cognitive foundation. For many reasons the concentration on the moral agent, especially his character, seems to be reasonable. Character plays an important role concerning the perception of a situation as morally problematic with certain relevant features, as well as concerning the reaction of the agent and, finally, his will to make the best of this situation for all participants. It is important to keep in mind that moral questions are embedded in general considerations of one s identity, personal development and life plans. Obviously there are degrees of moral competence and moral knowledge, and it seems promising to ask what that precisely means. Within such an analysis, virtues and character will play a key role. Conclusion There might be a need for intensive study of virtues. We are even willing to concede that this enterprise will be a real improvement for analytic moral philosophy as it reintroduces a variety of neglected topics into modern ethical theorizing. Thinking about virtues might lead to a deeper understanding of our moral life and moral practice: In ancient Greece, moral philosophy started as a study of ethical and social excellence. But over the centuries, this enterprise has become trivial: The search for moral excellence degenerated into an analysis of the question how to prevent moral faults [...] 13 The various studies of virtues could be regarded as an attempt to halt this process of trivialization. But it is not necessary to demand a new start or even a revival. All we need for a good working virtue ethics already exists in nuce in the work of Moritz Schlick. Therefore, prospering in virtue ethics does not mean quitting analytic ethics, but going back to its beginnings and having a look at its roots. NOTES Foot, Philippa, Virtues and Vices. Oxford: Basil Blackwell 1978, p. 1. Manfred Geier, The Quest for Happiness: Traces of Ancient Life Wisdom within the Moral Philosophical Context of the Vienna Circle, in: Herlinde Pauer-Studer (Ed.), Norms, Values, and Society, Vienna Circle Institute Yearbook. Dordrecht: Kluwer Academic Publishers. p. 13. I am referring to: Moritz Schlick, Problems of Ethics, translated by David Rynin; New York: Dover 1961,-and: Moritz Schlick, Lebensweisheit. Versuch einer Glückseligkeitslehre. Munich: C.H.Beck Moritz Schlick: Problems of Ethics, loc. cit. p Ibid., p. 80.

121 106 DAGMAR BORCHERS Geier, The Quest for Happiness: Traces of Ancient Life Wisdom within the Moral Philosophical Context of the Vienna Circle, loc. cit., p. 17. Philippa Foot, Rationality and Virtue ; in: Herlinde Pauer-Studer, H. (Ed.), Norms, Values and Society, Vienna Circle Institute Yearbook, Dordrecht: Kluwer Academic Publishers 1994, p Foot, Virtues and Vices, loc. cit., p. 8. Ibid., p.2. John McDowell, Two Sorts of Naturalism ; in: John McDowell, Mind, Value, and Reality, Cambridge (Mass.): Harvard University Press 1998, p Richard Hare, Geach: Good and Evil, in: Philippa Foot (Ed.), Theories of Ethics, Oxford: Oxford University Press 1967, p. 82. Dagmar Borchers, Die neue Tugendethik Schritt zurück im Zorn? Paderborn: mentis Jaakko Hintikka in: Dagmar Borchers/Olaf Brill/Uwe Czaniera (Eds.), Einladung zum Denken. Ein Streifzug durch die Analytische Philosophic. Vienna: Hölder, Pichler, Tempski Institut für Philosophie Universität Bayreuth Bayreuth Germany Dagmar.Borchers@uni-bayreuth.de

122 CARSTEN KLEIN COORDINATION AND CONVENTION IN HANS REICHENBACH S PHILOSOPHY OF SPACE The concept of coordination ( Zuordnung ) was central to the writings of some of the early followers of Logical Empiricism. In his Allgemeine Erkenntnislehre (1918), Moritz Schlick characterized the process of cognition as a coordination of concepts with objects and of judgements with facts, while defining truth as uniqueness of coordination. Whereas Schlick s conception was realistic in spirit, Hans Reichenbach used in his Relativitätstheorie und Erkenntnis apriori (1920) the concept of coordination in a framework that was still influenced by Neo- Kantianism. He emphasized the role of coordination with respect to the constitution of objects and introduced the idea of coordinative principles ( Zuordnungsprinzipien ) which are apriori in a relativized sense. 1 Later, he abandoned the Kantian approach and moved on to a conventionalist epistemology, trying to separate factual and conventional kinds of coordination and calling the latter coordinative definitions ( Zuordnungsdefinitionen ). I would like to take a closer look at the concept of coordinative definition as it is employed in Reichenbach s Philosophy of Space and Time (1928). 2 I will argue that he employed two quite different concepts of coordinative definition here without distinguishing properly between them. The first one bears a strong similarity to the conception of definitional coordinations that can be found in Schlick s Allgemeine Erkenntnislehre, regarding them as interpretation rules for the concepts of an axiomatic system. But without proper differentiation, Reichenbach used a different type of coordinative definition in the broader sense of conventional elements of our world descriptions, despite their not being coordinations in a proper sense. I will try to show that the most prominent example of a so-called coordinative definition in Reichenbach s philosophy of space, the definition of congruence, is of such a kind. In his Philosophy of Space and Time, Reichenbach introduced the concept of coordination within the context of his distinction between mathematical and physical geometries. He took the idea that the application of a mathematical system to reality can be interpreted as a coordination of implicitly defined concepts with real objects from Moritz Schlick s book General Theory of Knowledge (Allgemeine Erkenntnislehre, 1918). 3 This way, a mathematical geometry is turned into an empirical theory. But not all coordinations can have a factual content and, hence, be true or false. Rather, there must be some coordinations that are definitional in nature, called coordinative definitions F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

123 110 CARSTEN KLEIN Reichenbach explains the idea of coordinative definitions by presenting a simple example. If a distance is to be measured, the unit of length has to be determined beforehand by definition. But that cannot be done by an ordinary conceptual definition, since such a definition does not say anything about the size of the unit. This can only be done by reference to a physically given length 5, i.e. by a coordinative definition. Before such metrical coordinative definitions are given, statements about distances do not have factual meaning. In this sense, Reichenbach calls them logical presuppositions concerning measurements. 6 It is possible to express this insight in a different way: the adoption of a unit of measurement is not determined by facts, but is rather a matter of stipulation. Thus, coordinative definitions are examples of conventional elements in our world-description. In the given example, the coordinative definition is ostensive in nature, it can only be achieved by reference to a physical object: that thing there is to correspond to such and such a concept. 7 According to Reichenbach, there is no difference in principle when there is an insertion of some further concepts between the concept to be defined and the real object. His example is the coordinative definition a meter is the forty-millionth part of the circumference of the earth. Here too we refer to a physical length, the circumference of the earth, even if the reference here is rather remote by means of the interposition of conceptual relations. 8 And the situation is the same when we define the unit of length by reference to a certain wavelength. It is true that not the wavelength itself is observable but only certain phenomena like interference patterns, which are theoretically related to it. Nevertheless, the wavelength is a piece of reality, and thus it can play its part in a coordinative definition. 9 That the unit of length must be defined before measurements are possible is not a very profound insight. But Reichenbach gives us this rather trivial example only in order to clarify the main characteristics of coordinative definitions. After this is done, he turns to the far more interesting case of the relation of congruence. 10 But it will turn out that, contrary to Reichenbach, the definition of congruence is very different in nature compared to the definition of the unit of length. To determine whether two distances at different locations in space are congruent, we have to measure their respective lengths. The standard procedure is to carry a measuring rod from one place to the other and read off the respective numbers. By taking up insights of Hermann von Helmholtz, Reichenbach saw that the measuring procedure just described is subject to a hidden premise namely, the presupposition that the length of the measuring rod did not change while transported. That this is by no means a matter of course becomes clear once we consider the question of determining such changes of length: it seems obvious that this can only be done by comparison with a different measuring rod. Now imagine a force that has the same effect on all objects regardless of their composition, and let this effect be of such a kind that the lengths of these objects change by the same factor while in transit from one point to another. It is easy to

124 COORDINATION AND CONVENTION IN REICHENBACH S PHILOSOPHY OF SPACE 111 see that such a universal force could not be detected, since all relations of length would remain the same, and it is only such relations that can be measured. 11 Therefore, the assumption that such an effect does not arise cannot not be derived from observable facts. 12 Reichenbach concludes: The problem does not concern a matter of cognition but of definition. There is no way of knowing whether a measuring rod retains its length when it is transported to another place; a statement of this kind can only be introduced by a definition. 13 Thus the relation of congruence that holds between objects that are divided spatially is undetermined unless the concept of congruence has been fixed by definition. In this sense, the definition may again be called a logical presupposition concerning measurements. And since it is achieved by a coordination of the concept of congruence with a real object, here again we have a case of a coordinative definition. 14 At least that is what Reichenbach tells us. But is this really a tenable point of view? When Reichenbach says that it is the function of coordinative definitions to give such statements that express the results of measurements an objective meaning, he seems to have a certain semantic model in mind. Coordinative definitions are regarded as semantic designation rules that determine the reference of geometrical concepts. And after their reference is fixed, they can be used in the context of a physical geometry to make assertions about the real world. This way of looking at coordinative definitions obviously follows the model of defining the unit of length: before the concept of a unit is given a reference, it cannot be used to make statements about the lengths of physical objects. But it would be a mistake to over-emphasize the similarity between this simple case and the definition of congruence. With respect to the coordination between concept and object, there is a principal difference between these two cases. It is plausible to consider the definition of the unit of length to be an ostensive definition. There is a physical object, the standard meter in Paris, that can be identified by an ostensive gesture as reference for the concept in question. And an interposition of other concepts does not change the way in which the coordination works. It does not matter whether one has a measuring rod or a pattern of interference: there is always an observable object that can be identified by an ostensive gesture, even if in the latter case the object of reference is not the observable pattern itself but the non-observable wavelength, which is connected to the observed phenomena by a simple conceptual relation. The case of the definition of congruence is completely different. Since it is a two-place-predicate, the coordinated entity can only be a relation, i.e. the relation which obtains between two spatially separated physical objects if their lengths turn out to be equal when measured by a transported rod. The extension of the congruence-predicate, then, is the class of pairs of congruent objects. This reading is supported by Reichenbach s formulation that the concept equality of length is coordinated to a physical structure. 15 At first sight, this seems to be an acceptable view why should we not regard relations as being real in the

125 112 CARSTEN KLEIN same sense as objects? 16 But since we can hardly point to a relation, the claim that coordinative definitions must be achieved by reference can no longer be maintained. To salvage the ostensive nature of coordinative definitions, we could say as an alternative that the objects taken to be congruent are themselves the entities coordinated with the concept. In Reichenbach s text, a hint for this reading can be found as well: There is no way of knowing whether a measuring rod retains its length when it is transported to another place; a statement of this kind can only be introduced by a definition. For this purpose a coordinative definition is to be used, because two physical objects distant from each other are defined as equal in length. 17 The coordinative nature of the definition is substantiated by the fact that it determines physical objects which will be elements of the extension of the concept of congruence. But even if we regard the objects themselves as the piece of reality coordinated to the concept in question, the ostensive nature of the definition cannot be maintained. If we point to two objects and utter the words these objects shall be equal in length, we have not said anything at all about the congruence of all the other existing objects. The stipulated congruence of these two objects implies that a measuring rod does not change its length when transported from one to the other if the measurements give the same numbers. But it does not imply anything about universal changes of measuring rods between any other two objects. But there is another reading of Reichenbach s definition of congruence which, at first glance, seems to avoid all these difficulties. Cannot the transported rod itself be the real object coordinated to the concept in question? 18 After all, the definition of congruence stipulates a certain behaviour of such rods when they are transported. And since it is sufficient to make such a stipulation with respect to one measuring rod only, it could simply amount to an utterance of the sentence this rod does not change its length when being transported together with an ostensive gesture. Is this not a definition by ostensive reference to a piece of reality? This conclusion, however, would be unjustified. In this definition, the ostensive reference is an accompanying element only. It serves as a means for identifying an object, whereas the actual definition is a statement about the behaviour of this object. This definition does not just consist in an ostensive gesture, accompanied by the utterance of the concept in question, as it is the case in the definition of the unit of length. In the latter case we just utter the words unit of length, pointing to the standard meter, but in order to define the concept of congruence we do not just point to a measuring rod and say congruent. It is not a definition by ostension, but with the help of an ostension. Whereas in the case of the unit of length the function of the ostension is to fix the reference of the concept by a kind of baptism, in the case of the definition of congruence the reference is fixed by a stipulative statement about the behaviour of the object identi-

126 COORDINATION AND CONVENTION IN REICHENBACH S PHILOSOPHY OF SPACE 113 fied by the ostension. In order to make the definition, we have to point to the rod, but the concept defined does not refer to the rod but to a relation obtaining between certain pairs of objects. This fundamental difference is substantiated by the fact that an ostensive gesture is not an essential element of the definition of congruence but only a feature of a certain formulation. In the subsequent parts of his book, Reichenbach goes on to formulate the definition of congruence as a statement about the absence of universal forces. 19 This is a definition of congruence because without universal forces, the change in the lengths of transported rods is no longer a matter of stipulation but an observable fact. Thus, statements about the congruence of spatially separated objects now have factual content, are true or false. The stipulative statement about the absence of universal forces still determines the extension of the congruence-predicate, but there is no longer a direct reference to a physical object. Hence, this formulation of the definition is not ostensive but discursive. It is because of this that one might claim that Reichenbach uses the concept of coordinative definition in two respects in his Philosophy of Space and Time: One the one hand, this concept is used for ostensive definitions in analogy to the definition of the unit, where something real gets coordinated to a concept by way of ostension. Typical for this kind of definition, which can be taken as a coordinative definition in the stricter sense, is the property that what we are dealing with are singular sentences which can be substituted in principle by the mere utterance of the word in question together with an ostensive gesture. On the other hand, there can be no doubt that the term coordinative definition is used in a more general sense for definitions which determine the reference of a concept, even if this method of determination does not proceed in an ostensive but in a discursive way. Besides formulating the definition of congruence by way of stipulative elimination of all universal forces, this becomes clear especially with respect to Reichenbach s claim that the specification of a certain physical geometry is a coordinative definition of congruence. For not only does it hold that, given a concept of congruence F, the physical geometry G is empirically determined in a unique way. Given a certain geometry G, the concept of congruence F is also no longer a matter of free choice but is determined by the results of observation. Reichenbach is of the opinion that this case is of the same kind as the coordinative definition of the unit of length by reference to the circumference of the earth. But while it may be possible (with a grain of salt) to treat the circumference as a physical object that can be identified by an ostensive gesture and one that is connected to the unit to be defined by an interposition of concepts, it is hardly possible to point to a geometry in any particular sense. The coordinative definition that is achieved by specifying a geometry is discursive, the reference to a physical object being nowhere necessary. Moreover, this second, discursive type of a coordinative definition which I shall call here a coordinative definition in the more general sense differs from ostensive definitions with respect to its logical form. Definitions of the first type

127 114 CARSTEN KLEIN are singular sentences and can also be expressed as a metalinguistic designation rule: The unit length is the length of the standard meter in Paris or 1m refers to the length of the standard meter in Paris or 1m refers to this length over there together with an ostensive gesture. 20 Its only function is to supply the definiendum with a reference in order to use it for the formulation of physical statements. By contrast, coordinative definitions in the more general sense are universal statements like rigid rods do not change their lengths when transported, there are no universal forces or the geometry of physical space is Euclidean, sentences, that is, which make a general statement about a certain type of physical objects. It is clear that these statements cannot be (re)formulated as simple rules of designation of the type x refers to y, for what these sentences express is a law-like statement which is part of the theoretical description. 21 It is by virtue of its completely different logical status that the definition of congruence plays an important role within the structure of the theory of physical space in the first place, even if it cannot be denied that it also plays a role in determining the extension of the predicate. For as we have seen, different coordinative definitions lead to different geometries, and even our physics of everyday life would have to undergo radical changes if we were to assume that objects change their lengths when transported. Thus the concept of congruence plays an important role within the theoretical description of our world. Similar things cannot be said with respect to the concept of the unit of length. Changing the definition of unit has no effect whatsoever on the structure of the theory: although it would lead to a change of all metrical statements by a certain factor, the formulation of theoretical laws would not be effected. 22 To use an expression of Carl Gustav Hempel, one could say that the concept of congruence has considerably more systematic import than the concept of the unit of length. 23 This is exactly why the freedom with respect to determining the relation of congruence is so interesting: it reflects certain structural properties of the theoretical description of physical space. And as Reichenbach has correctly recognized, 24 the systematic import of the concept of congruence is even considerably higher. For the concept of the equality of length plays a role in nearly all physical laws, since it occurs in almost every physical quantity of derivative nature. Thus, changing the definition of congruence would not only have an effect on physical geometry but on nearly the whole spectrum of physical theories, while changing the concept of the unit of length would only result in a new scaling of metrical statements. This justifies us in taking the concept of congruence as a law cluster concept; that is, as a concept which occurs in a large number of natural laws and the meaning of which is given by these laws. And it is exactly this property that renders the conventionality of congruence an interesting question. That the reference of a concept like unit of length must first be determined by a definition before it can be used to formulate results of measurement is of no significance. That a law-like statement about the behaviour of rigid bodies or the existence of universal forces must be made in order to determine

128 COORDINATION AND CONVENTION IN REICHENBACH S PHILOSOPHY OF SPACE 115 the reference of congruent, by contrast, is by no means a matter of course but provides an interesting insight into the structure of physical theories. This result, however, should rather be formulated within Reichenbach s terminology of equivalent descriptions: The equivalent descriptions which result from changing the standards of length differ only with respect to a constant factor in the numerical values of their metric statements. That the descriptions are equivalent and that the choice between them is of a conventional nature is easy to see; everybody will agree that this conventionality is hardly of any specific epistemological significance. By contrast, the descriptions which result from changing the definition of congruence differ with respect to the premises about the geometry of space and the occurrence of universal forces. That these descriptions are equally admissible and empirically indistinguishable is a significant property of the theory in question. If we consider the complete description G + F, the nature of the definition of congruence becomes clear: It is a stipulative determination of the conventional element of this description. And it is because one part of the description, the F- component, refers to the relation of congruence that the specification of the conventional element results in a coordination. This holds even if it is formulated as a statement about the geometry G. Hence, it is not completely wrong when Reichenbach states that the specification of a geometry G is just a different way of giving a coordinative definition of congruence. But it is misleading, because it seems to imply the view that conventionality is a property not of the whole description but of particular statements within the description, whereas other statements are factual in nature. As a consequence, there would be a definite and unique separation of conventional and factual elements in our description of physical space. In fact, this seems to have been Reichenbach s position in But it is difficult to follow him on this point. Since there is a symmetry in the two methods of determining the conventional element inherent in the description G + F, viz. the specification of F or the specification of G, there is no reason to ascribe the conventionality to F alone in both cases. Reichenbach points out correctly that a statement about the geometry G has no factual meaning unless the concept of congruence is determined. But the same holds the other way round: a statement about universal changes of transported rods has no factual meaning unless the geometry G is specified. Because of this symmetry, calling the specification of G a definition of congruence seems unjustified. As Quine emphasized a couple of years later, we should rather adopt the view that conventionality and factuality are spread out over the whole description. 25 Interestingly enough, this is exactly the position Reichenbach held in 1923: We have a certain freedom in employing definitions and facts. It is only when a definition is given in one place that another assertion becomes an assertion of fact; conversely, the second may be regarded as a definition, which makes the first into an assertion of fact. 26

129 116 CARSTEN KLEIN If this is true, then conventionality can only be ascribed to the whole description, and the existence of a conventional element is shown by the fact that there are alternative descriptions that are logically incompatible but equally true. Why did Reichenbach change his position between 1923 and 1928? I think that one of the reasons is to be found in his aim of arguing against the geometrical relativism he ascribed to Poincaré. He takes the latter to be of the opinion that it is impossible to make an objective statement about the geometry of physical space, and that we are dealing with subjective arbitrariness only, which for Reichenbach amounts to saying that the concept of the geometry of real space itself is meaningless. 27 By contrast, Reichenbach wants to hold on to the view that the concept of a geometry of physical space is meaningful and that statements about this geometry have objective factual content. In order to defend this position, he feels compelled to ascribe the definitional character solely to the relation of congruence, thus preserving the purely factual character of the geometry: Although the statement about the geometry is based upon certain arbitrary definitions, the statement itself does not become arbitrary: once the definitions have been formulated, it is determined through objective reality alone which is the actual geometry.... it is the significance of coordinative definitions to lend an objective meaning to physical measurements. 28 This argument, however, does not succeed. If the statement of the geometry is based on arbitrary definitions, then it is itself arbitrary. For it is possible to obtain any geometry by choosing an appropriate definition of congruence. Reichenbach s real achievement in his efforts to avoid Poincaré s scepticism is the insight that, though not G alone, the combination of G and F does have factual content. Hence, not the physical geometry alone is rescued from arbitrariness by Reichenbach s argument, but the combination of geometry and the concept of congruence. But according to this reading, there is no reason to ascribe conventionality to the specification of congruence alone. The whole description G + F is conventional insofar as there are different but equally admissible descriptions G + F. And it is factual because the class of admissible descriptions is restricted by physical reality: if one of the two components is given, the choice of the other component is a matter of fact. The decision which component, G or F, is specified by stipulation is itself arbitrary. There is no unique separation of factual and conventional components. 29 But there seems to be another reason for Reichenbach s view that conventionality can be ascribed to the specification of congruence in an absolute way: his belief in the alleged similarity of the two kinds of coordinative definitions I have tried to keep apart. Reichenbach seems to think that the connection between the concept of congruence and the geometrical concepts is of the same kind as the connection between the concept unit of length with the concept forty million times the unit of length. According to Reichenbach, the specification of

130 COORDINATION AND CONVENTION IN REICHENBACH S PHILOSOPHY OF SPACE 117 the physical geometry is a definition of congruence in the same sense in which the specification of the physical object coordinated to forty million times the unit of length is a definition of the unit of length. 30 But this analogy is not as farreaching as Reichenbach thinks. There is a very simple derivative relation between the concepts forty million times the unit of length and unit of length. It makes good sense to call the one concept primitive and the other derivative, and it seems therefore natural to read the specification of the derivative concept as a definition of the primitive concept from which it was derived. But the relation between the geometrical concepts (e.g., straight line ) and the concept of congruence is far more complicated, and we have no good reason to insist that congruent is primitive and straight line is derivative. Thus, it is not at all clear why we should read the specification of the geometry as a definition of congruence. That Reichenbach held a different and, I think, more correct view in his 1923 paper may be explained by the influence of Schlick, which is more prominent in this paper than in The Philosophy of Space and Time. The view that the conventionality is spread over the whole system of coordinations and cannot be located in some particular coordinations in an absolute sense is inherent in Schlick s original conception of the coordination of a system of implicitly defined concepts with real objects. Why should any of these concepts be special in the sense that the procedure of coordination must begin with this particular concept, making its coordination definitional and all the others factual? If we consider Reichenbach s argument for the necessity of coordinative definitions:... certain preliminary coordinations must be determined before the method of coordination can be carried through any further; these first coordinations are therefore definitions... 31, we can see no reason why one particular coordination has to be the first and, therefore, definitional one. So even if we conceded, for the sake of argument, that the definition of congruence would work in the way Reichenbach says, viz. by a simple coordination with a physical object, there is no reason why the status of definition should be ascribed to this coordination in an absolute way. To conclude: I hope to have shown that the conception of coordinative definition in Hans Reichenbach s philosophy of geometry is burdened by the fact that it stretches this concept too far. There are conventional stipulations in a description of physical space that are not coordinative definitions in the strict sense, although they do indeed specify the extension of the concept in question. One of these kinds of definitions is the definition of congruence, erroneously presented by Reichenbach as a paradigmatic example of coordinative definitions. Furthermore, the alleged similarity of the definition of congruence to coordinative definitions proper may have been one reason why Reichenbach changed his position concerning the localizability of conventional elements in our descriptions of the physical world. I want to suggest that we should rather describe the situation from the point of view of the whole description: Every description has conven-

131 118 CARSTEN KLEIN tional elements that have to be specified by stipulations. However, only some of these stipulations are coordinative definitions in the strict sense, and these are not the really interesting ones. NOTES For Reichenbach s relativized notion of the a priori cf. Michael Friedman, Geometry, Convention and the Relativized A Priori: Reichenbach, Schlick, and Carnap, in: Wesley Salmon, Gereon Wolters (Ed.), Logic, Language and the Structure of Scientific Theories, Proceedings of the Carnap-Reichenbach Centennial Konstanz Pittsburgh: Pittsburgh University Press 1994, pp Hans Reichenbach, Philosophie der Raum-Zeit-Lehre. Berlin/Leipzig: Walter de Gruyter All further references will be made to the English translation: Philosophy of Space and Time, trans. Maria Reichenbach and John Freund. New York: Dover Publications Moritz Schlick, Allgemeine Erkenntnislehre. Berlin: Springer 1918, , Translated as General Theory of Knowledge, trans. A.E. Blumberg. New York: Springer An explicit reference to Schlick can be found in the Philosophy of Space and Time, loc.cil., on p. 14, fn. 1. As early as 1923, in his paper Ziele und Wege der physikalischen Erkenntnis, Reichenbach had adopted some essential features of Schlick s epistemology: the idea that knowledge can be understood as coordination of concepts with objects, the idea that truth is uniqueness of coordination and the interpretation of concepts as signs. (M. Thirring (Ed.), Allgemeine Grundlagen der Physik, vol. 4 of Handbuch der Physik, H.Geiger und Karl Scheel (Eds.). Berlin: Springer 1929, pp All further references will be made to the English translation: The Aims and Methods of Physical Knowledge, in: Maria Reichenbach, Robert S.Cohen (Eds.), Hans Reichenbach: Selected Writings, vol. 2. Dordrecht: Reidel Reichenbach, The Philosophy of Space and Time, loc. cit., p. 14. The concept of coordinative definition is already introduced in The Aims and Methods of Physical Knowledge, loc. cit., p. 161: The coordination of concepts with objects that is carried out in natural science is not always such that it yields knowledge. While the matter of which concept should be coordinated is, in general, established as a result of research, the coordination may in particular instances consist of an arbitrary stipulation. These cases constitute physical definition. Their mark of distinction is that, instead of coordinating a concept with a particular combination of other concepts, as done in definitions of concepts, they coordinate a concept with an existing object.... As the coordination between object and concept is peculiar to this type of definition, we shall refer to it as coordinative definition. Reichenbach, The Philosophy of Space and Time, loc. cit., p. 15. Ibid., p Reichenbach, Aims and Methods of Physical Knowledge, loc. cit., p. 33. Reichenbach, The Philosophy of Space and Time, loc. cit., pp. 15, 128. Ibid., p.15. Ibid., pp Reichenbach s notion of a universal force is rather obscure. It seems highly questionable to classify a quantity as a force when it has no physical cause, cannot be tested empirically for principal reasons, and can be defined away by choosing a different definition of congruence. Perhaps one should use the notion of universal effects instead, leaving it an open question whether there are in fact any forces causing these effects. A coherent interpretation of the notion of universal force is made even more diffcult by Reichenbach s claim that gravitation is an example of such a force (ibid., p. 256), although gravitational forces are, without question, empirically detectable. Lionel Shapiro claims to have identified no less than four different notions of universal force in the Philosophy of Space and Time (Lionel Shapiro, Coordinative Definition and Reichenbach s Semantic Framework: a Reassessment, in: Erkenntnis 41, 1994, pp ). And according to Adolf Grünbaum, the notion of universal force has to be under-

132 COORDINATION AND CONVENTION IN REICHENBACH S PHILOSOPHY OF SPACE stood in a metaphorical way, with the consequence that the definition of congruence can be formulated without saying anything as to the existence or non-existence of universal forces (Adolf Grünbaum, Philosophical Problems of Space and Time (Second, enlarged edition) (= Boston Studies in the Philosophy of Science, Vol. 12). Dordrecht/Boston: Reidel 1973, pp ). (What Grünbaum has in mind is Carnap s definition of congruence as put forward in his dissertation Der Raum.) An appeal to different methods of metric measurement does not invalidate the argument since the universal force can always be constructed in such a way that it effects the results of all kinds of measurement in the same way. Reichenbach, The Philosophy of Space and Time, loc. cit., p. 16. Ibid., p. 16. Ibid., p. 16. In The Aims and Methods of Physical Knowledge Reichenbach explicitly argues for accepting relations as something real in the same sense as objects (loc.cit., pp ). Reichenbach, The Philosophy of Space and Time, loc. cit., p. 16. Physical objects is an incorrect translation of wirkliche Dinge ( real objects ) in the German original (Philosophie der Raum-Zeit-Lehre, loc. cit., p. 25). This seems to be Reichenbach s view in In his contribution to the Einstein Schilppvolume, he writes: For instance, the concept equal length is defined by reference to a physical object, a solid rod, whose transport lays down equal distances. ( The Philosophical Significance of the Theory of Relativity, in: Paul Arthur Schilpp (Ed.), Albert Einstein: Philosopher Scientist. La Salle, Ill.: Open Court 1949, p. 294.) It seems that the coordinative definition F =0 is simpler... (Reichenbach, The Philosophy of Space and Time, loc. cit., p. 34). Coordinative definitions of this kind are similar to Carnap s designation rules for semantic systems. Their task is to supply the descriptive constants of a theory with a physical interpretation. In the case of individual constants, they are of the form mond designates the moon. (Rudolf Carnap, Foundations of Logic and Mathematics (= International Encyclopedia of Unified Science, Vol. I, No. 3). Chicago: University of Chicago Press 1939, p.9.) The objection that a definition cannot be a law of nature is not cogent. I will argue at some later point that there is no unique distinction between conventional definitions and factual assertions. Hence, the statement rigid rods do not change their lengths during transport can be regarded as a law of nature, as can the statement the sum of the interior angles of every physically realized triangle is 180. This is only correct if the possibility of a definition of the unit of length by a non-rigid body is not taken into account. If one used a rubber band as a standard of length, every length would depend on the momentary state of extension of the rubber band. This would result in an immense complication of physical laws. The possibility of using a non-rigid body, however, does also exist in the case of the definition of congruence. But in the latter case, even after the exclusion of all non-rigid bodies, there is still room for non-trivial variations, which is not true for the definition of the unit of length. According to Hempel, concepts have systematic import if they open up the possibility for formulating explanatory and predicting principles in the form of general laws or theories. (Carl Gustav Hempel, Fundamentals of Concept Formation in Empirical Sciences (= International Encyclopedia of Unified Science, Vol. II, No. 9). Chicago: University of Chicago Press 1952, p. 46. Reichenbach, Philosophy of Space and Time, loc. cit., pp. 23 f. Willard Van Orman Quine, Truth by Convention, in: The Ways of Paradox and other Essays, Cambridge, MA: Harvard University Press , pp (First published in: O.H.Lee (Ed.), Philosophical Essays for A.N. Whitehead. New York: Longmans 1936). Reichenbach, The Aims and Methods of Physical Knowledge, loc. cit., p Reichenbach, Philosophy of Space and Time, loc. cit., pp It has been pointed out several times that Reichenbach s interpretation does not do justice to Poincarés intentions. For a prominent example, see Grünbaum, op. cit., pp Reichenbach, Philosophy of Space and Time, loc. cit., p. 37.

133 120 CARSTEN KLEIN For a more detailed account of Reichenbach s conception of equivalent descriptions and the role it plays in the refutation of Poincaré s geometrical relativism, cf. Carsten Klein, Conventionalism and Realism in Hans Reichenbach s Philosophy of Geometry, in: International Studies in the Philosophy of Science 15, 3, 2001, pp Reichenbach, The Philosophy of Space and Time, loc. cit., p. 34. Ibid., p. 14. Philosophisches Seminar Universität Bonn Am Hof 1 D Bonn Germany cklein@uni-bonn.de

134 ROBERT RYNASIEWICZ REICHENBACH S OFSIMULTANEITY IN HISTORICAL AND PHILOSOPHICAL PERSPECTIVE 1. It is well-known that in the special theory of relativity the simultaneity of distant events is frame-relative: two events that are simultaneous according to the standard Einstein criterion as applied in one inertial frame are not simultaneous according to the same criterion applied in a relatively moving inertial frame. Not so well known is whether, given a fixed inertial frame, there is a fact of the matter whether a pair of distant events are really simultaneous relative to that frame. Einstein thought not. Reichenbach, at least according to the lore philosophers of science are taught, thought not. And up until roughly a quarter century ago, the thesis that frame-relative simultaneity is a matter of convention was the prevailing view. Not any more. There is now a widely espoused anti-conventionalist view inspired by a result published by David Malament in The standard Einstein criterion is the only candidate for a frame-relative simultaneity criterion definable from the causal structure of Minkowski space-time. 1 Hence, according to the neo-anti-conventionalist, if frame-relative simultaneity is conventional, then it is conventional only in some minimal and not very exciting or robust sense. However, claims as to what is exiting or robust, if they are to be meaningful, must be claims about comparisons. I want to take this opportunity to compare and contrast two possible conventionalist positions on frame-relative simultaneity, one of them decidedly more robust than the other. The more robust of these is implicitly suggested by Einstein s 1905 paper. 2 The other, I shall suggest, is the most plausible reading of Reichenbach s conventionalism in connection with his celebrated of distant simultaneity. This will give me the chance not only to engage in a bit of Reichenbach scholarship, but also to impress upon the neo-anti-conventionalist camp that there is more than a minimal conventionality thesis to take into account. 2. In order to set up the conventionalist position suggested in Einstein s original 1905 paper, recall the procedure followed there. One establishes the fixed points of an inertial frame (presumably by testing for inertial forces), lays out a spatial coordinate grid using rigid rods and the laws of Euclidean geometry, and institutes a natural clock at each fixed point. There remains the further task of synchronizing these clocks if one is to be able to describe trajectories as a function of a system-wide time. Without such one cannot assign a three-velocity tan- 121 F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

135 122 ROBERT RYNASIEWICZ gent vector to a point of the trajectory. Thus, one simply stipulates that the time it takes for light to propagate from point A to point B of the frame equals the time of propagation from B to A. From this follows the familiar Einstein synchronization criterion, that a natural clock at A is in synchrony with natural clock at B if where is the time at A of light emission from A, the time at B of its reflection at B, and the time at A of its return to A. The derivation of the Lorentz transformations proceeds directly from this (and the validity of the light postulate) by assuming that the moving frame uses the same synchronization criterion from its point of view for co-moving clocks. In short, the procedure is this. Adopt a synchronization procedure consistent with the empirically testable fact that the average round trip speed of light in any frame is a fixed constant. Then apply the Principle of Relativity to that procedure in order to derive the coordinate transformations to a relatively moving frame. But if, as Einstein insists, the one-way speed of light assumption on which clock synchronization is based is really just a matter of stipulation, what happens if one replaces it with an alternative stipulation compatible with the constancy of round trip average speed, and then applies the Principle of Relativity to the induced non-standard synchronization procedure? Does this yield a consistent alternative 3+1-dimensional formulation of special relativity? 3. With this question in mind, I began to scour the Reichenbach archives for calculations to see if Reichenbach had indeed thought along these lines in the course of formulating his famous of clock synchronization, from which, the standard Einstein criterion falls out as the special case In 1921 Reichenbach published a preliminary sketch of his plan to develop an axiomatization of relativity theory in such a way that it would have the virtue of separating out the factual from the conventional components of the theory by instituting only directly testable propositions as axioms and introducing the conventional components as coordinative definitions. This Bericht über eine Axiomatik der Einsteinschen Raum-Zeit-Lehre introduces only the standard Einstein synchronization criterion as a coordinative definition [Definition 5]. 3 The first appears in print only in 1924 as Definition 2 of the extensively reworked and vastly expanded culmination of that project, Axiomatik der relativistischen Raum-Zeit-Lehre. 4 Moreover, its introduction there appears to have been an eleventh-hour addition. There are two complete copies of a mono-

136 REICHENBACH S OF SIMULTANEITY 123 graph-length draft, suggesting that Reichenbach thought it ready, or nearly ready for press, but in which still only the standard Einstein definition appears. Only amongst scraps of loose handwritten pages did I find a precursor (see fig. 1) to the in the published version. 5

137 124 ROBERT RYNASIEWICZ The portion of typed text is the whole of Definition 5 cut out from a reprint of the Bericht (see fig. 2) and the emendations make it accord verbatim with Definition 2 of the Axiomatik. The context of its introduction also accords, although not verbatim, with the published version the middle of a proof of Proposition 6, which, in order to grasp its upshot, requires some exposition of Reichenbach s overall strategy. 4. When Hermann Weyl reviewed Reichenbach s Axiomatik, 6 he confessed that he found it less than satisfactory: overly tedious and too obscure [wenig befriedigend, zu umständlich und zu undurchsichtig]. 7 This, Reichenbach later complained, was based on a gross misunderstanding. 8 In the review, Weyl had characterized the Axiomatik as, in the main, not a philosophical, but a purely mathematical investigation, and had registered the above assessment from a mathematical point of view [nach mathematischen Gesichtpunkten]. 9 Reichenbach was dismayed that someone of Weyl s rank could have missed his main thrust. Weyl, though, may have well have appreciated Reichenbach s intentions more fully than Reichenbach realized. The execution of the Axiomatik is in fact more easily grasped from a mathematical than a philosophical point of view. Reichenbach begins with the class of all possible world-lines in space-time and then attempts to find conditions, involving only the conformal structure of the space-time, sufficient to select out the distinguished subsets which correspond to inertial frames. Using only the behavior of light rays, he defines separate temporal and spatial metrics for individual frames, which he calls the light-geometry [Lichtgeometrie], and then derives the Lorentz transformations as the isometry boosts for the light-geometry. (A defect of this procedure, and one certainly not lost on Weyl, is that the class of inertial frames cannot be singled out in this fashion, but only a wider class of frames related by conformal transformations. Reichenbach had sensed the problem, but probably at a stage too late for him to make major revisions. Only in a later section does he alert the reader to the difficulty, and then only in an attempt to mitigate the extent to which it might be viewed as undermining his program.) So far no mention is made of rigid rods and material clocks. These notions are introduced only after the development of the light-geometry, at which point Reichenbach sets out a series of matter axioms which collectively assert that these material structures behave in accordance with the light-geometry and thus the Lorentz transformations. The upshot of the entire work is that, apart from the prohibition on super-luminal causal propagation, it is only these latter assertions that separate relativity theory from classical space-time physics. The physical axioms governing the behavior of light do not depend on a principle of relative motion and are consistent with both the (relativistic) light-geometry and the geometry [or rather, kinematics] of classical optics. As Reichenbach explains:

138 REICHENBACH S OF SIMULTANEITY 125 It is remarkable that the Lorentz transformation, conceived in light-geometrical terms, does not contain a new axiom but depends solely on definitions and axioms of the single system. With respect to light, there does not exist a special axiom of uniform translation. Our derivation of the Lorentz transformation via the Galilean transformation has shown that it is merely a reorganization of the relations of uniformly moving systems contained in the Galilean transformation into a new metrical determination. Physically new is Einstein s idea that material structures do not adjust to the Galilean metrical determination but to the light-geometrical one. 10 Faintly visible under this thumbnail sketch, in the notion of the adjustment [Einstellung] of material structures to the light-geometry, is the specter of Weyl or more precisely, that of Weyl s generalization of Riemannian geometry in his attempted unified field theory which provides a clue as to why Reichenbach pursued the strategy of developing a light-geometry in no way dependent on the use of material metrical standards. Noteworthy is the fact that the first two matter axioms are formulated in tune with the conceptual framework of gauge field theory, asserting the path independence of the behavior of material rods and clocks. Indeed, Reichenbach mentions in a footnote in the Introduction 11 that a hint that the construction of a light-geometry is possible occurs in the Appendix to the fourth edition of Weyl s Raum-Zeit-Materie. 12 The notion of adjustment [Einstellung] is also taken over directly from Weyl. In his later reply to Weyl s review of the Axiomatik, though, Reichenbach clarifies that he does not take this to be a concept with explanatory force so much as a shorthand for the problem of finding a fundamental theory of matter. It is worth mentioning that Reichenbach there indicates that he regards the matter axioms to hold only as a first approximation (even to the extent that they are locally valid according to the general theory of relativity) and even cites the positive result of Dayton C. Miller s repetition of the Michelson-Morley experiment as potential evidence (assuming the result is not spurious, as it turned out to be). In contrast to his attitude toward the matter axioms, Reichenbach regards the light axioms as completely secure and describes the light-geometry as the most natural description (from the point of view of descriptive simplicity) of the intrinsic geometry of the electromagnetic field. 5. What role, then, does Reichenbach s of simultaneity play in the construction of the light-geometry? The sad answer is, disappointingly little, at least from a formal point of view. It is introduced early on, primarily in order to show how little needs to be assumed in order to introduce a global time function in conformity with causality constraints. The method is to choose a central clock in the sense of an arbitrarily parameterized world line and then to export this parameterization to every other world line in space-time via for an arbitrary, but fixed value of Reichenbach quickly specializes to in order to define a time function for each stationary spatial frame [stationäres räumliches Koordinatensystem]. The class of inertial frames is later extracted from the class of all stationary spatial frames.

139 126 ROBERT RYNASIEWICZ There is thus no inkling of the possibility of an invariant non-standard synchronization criterion for inertial frames. Rather, it is more likely that Reichenbach believed that the restriction to is in fact necessary for selecting the inertial frames in the course of constructing the light-geometry. This surmise is reinforced by a comment four years later in his Philosophie der Raum- Zeit-Lehre, 13 that Einstein s definition is in fact essential for the special theory of relativity: Diese Definition ist zwar fuer die spezielle Relativitaetstheorie wesentlich He does go on to say that, nonetheless, this definition is not epistemologically necessary any choice between zero and one for the parameter wouldwork and could not be said to be false. But work in what regard? In the Axiomatic, after Reichenbach introduces the generalized and puts it to limited technical use, there follows what might appropriately be called a philosophical scholium, though not labeled as such. It begins by characterizing the light axioms so far introduced as the topological axioms of time order and proceeds to call attention to the topological problem of simultaneity : in essence, does causal structure pick out at a given space-time point a unique hypersurface as causally neither prior nor posterior, or, as is the case if there is a limiting causal process, does there exist an indeterminate region corresponding to an entire family of distinct hypersurfaces? In the former case it is appropriate to speak of absolute simultaneity; in the latter, simultaneity is relative insofar as the definition of simultaneity is not uniquely determined by causal structure. The word relative here, however, does mean specific to the special theory of relativity. In the Introduction to the Axiomatic, Reichenbach carefully distinguishes between what he calls the epistemological and the physical relativity of simultaneity. The former refers merely to the underdetermination by causal structure. The latter is specific to Einstein s theories of special and general relativity. It says in the special theory that a particular definition of simultaneity for uniformly moving systems gives rise to the complete equivalence of all measurement procedures and in consequence the laws of nature have the same form for all such systems. [Sie besagt in der speziellen Theorie, dass bei einer gewissen Definition der Gleichzeitigkeit (Definition 8) für gleichförmig bewegte Systeme völlige Gleichartigkeit aller Massverhältnisse entsteht und die Naturgesetze dann für solche Systeme die gleiche Form haben.] 15 To summarize in light of the results achieved in the Axiomatik: The epistemological relativity of simultaneity refers only to the fact that numerous distinct light-geometries (e.g., Galilean vs. Lorentzian) are consistent with the light axioms. The physical relativity of simultaneity speaks to the consequences of the further imposition of the matter axioms, to the effect that the standard practices of measurement using material rods and clocks agree with the choice of and the resulting Lorentzian geometry. Thus, as far as the Axiomatik is concerned, any adoption of a non-standard simultaneity criterion would necessitate

140 REICHENBACH S OF SIMULTANEITY 127 the adoption of a different and highly irregular set of matter axioms appealing to compensatory factors and the like. As for the position in his Philosophie der Raum-Zeit-Lehre, Reichenbach sends no signal that he now intends the to pertain to any thesis bolder than the epistemological relativity of simultaneity. The comment that if the special theory of relativity prefers the... definition [that] sets equal to ½, it does so on the ground that this definition leads to simpler relations 16 need not be construed to indicate anything more than a belief that other choices would necessitate the adoption of a different set of more complicated matter axioms. It is true that, in contrast to the Axiomatik, he explicitly mentions directionally dependent choices of but these are in connection with the price of adopting a classical (Galilean) light-geometry. 17 The only portion of text that even remotely suggests otherwise reads: It was believed that the coupling of the space and time axes supplied by the Lorentz transformation, according to which every choice of the time axis determines a corresponding space axis as the conjugate diameter, signifies a more fundamental junction of space and time. This coupling, however, is relatively unimportant because it is based on an arbitrary additional requirement, introduced only for descriptive simplicity, for which there is actually no epistemological need. 18 But the passage goes on: The mistake committed here is the one pointed out on page 146; it springs from the erroneous conception that there is a relation between the relativity of simultaneity and the relativity of motion. 19 And indeed, if one goes back to consult page 146, 20 it is clear that what Reichenbach means is that the epistemological relativity of simultaneity has nothing to do with the relativity of motion. It rests solely on the existence of a finite limiting velocity for causal propagation What I have argued is that Reichenbach nowhere suggests the possibility of using a non-standard simultaneity criterion in conjunction with the principle of relativity. Such a criterion would have the following properties. It would result in the same light-geometry as the standard Einstein criterion in the sense that it would yield the same temporal and spatial measures as the standard within each inertial frame. Only the Lorentz transformations would need to be replaced by a conjugate representation of the same group. But each of the matter axioms would remain satisfied without revision, since these in fact do not explicitly involve the Lorentz transformations. I further suggest that Reichenbach believed that this could not be done. Many a reader may wonder whether indeed it can. Rather than formulating such in terms of directly, I ll simply state a one-way speed of light rule that suffices. Using spherical coordinates, let be the angle from the azimuth. Then stipulate that the speed of light V in the direction satisfies the condition:

141 128 ROBERT RYNASIEWICZ where c is the usual average round-trip speed of light and a is an arbitrary scalar less than c. As an exercise, one can verify that this satisfies the requirement of constancy of average round-trip, derive the corresponding conjugate representation of the Lorentz transformations, and develop a complete, non-standard 3+1- dimensional formulation of special relativity. Moreover, it is not too difficult to show that the most general category of frame invariant non-standard simultaneity criteria is given by letting the magnitude of the cosine term in the denominator represent the projection of an arbitrary irrotational vector field (of norm less than c) onto the azimuth. 22 The intent here is not to denigrate Reichenbach for failing to realize a possibility that in fact exists. Rather it is to come to a clear understanding as to the range of possibilities that he in fact did recognize. Indeed, it is to his credit to have articulated the distinction between what he called the epistemological and the physical relativity of simultaneity. For that distinction, just slightly re-articulated, serves to demarcate two distinct levels of convention potentially inherent in the adoption of a simultaneity criterion. The epistemological level concerns the degree of fixity dictated by causal structure alone, entirely apart from constraints deriving from the standard deployment of material measuring rods and clocks for the determination of spatial distances and proper-time lapses, respectively. The physical level corresponds to the degree of freedom that remains with these latter constraints in place. Much of the debate on the conventionality of simultaneity has tended to suffer from not tracking carefully enough the difference between these two levels. Acknowledgements. I would like to thank W. Gerald Heverly and his staff at the Archives of Scientific Philosophy, University of Pittsburgh Libraries for their kind hospitality and assistance. NOTES David Malament, Causal Theories of Time and the Conventionality of Simultaneity, in: Noûs 11, 1977, pp Albert Einstein, Zur Elektrodynamik bewegter Körper, in: Annalen der Physik 17, 1905, pp Hans Reichenbach, Bericht über eine Axiomatik der Einsteinschen Raum-Zeit-Lehre, in: Physikalische Zeitschrift XXII, 1921, pp Hans Reichenbach, Axiomatik der relativistischen Raum-Zeit-Lehre. Braunschweig: Friedrich Vieweg & Sohn Definition 2 appears on p. 26. This corresponds to p. 35 of Maria

142 REICHENBACH S OF SIMULTANEITY Reichenbach s translation, Axiomatization of the Theory of Relativity. Berkeley: University of California Press Document HR of the Hans Reichenbaoh Collection, Archives of Scientific Philosophy and General Manuscripts. Reproduced by permission of the University of Pittsburgh. All rights reserved. In: Deutsche Literaturzeitung 30, 1924, pp Ibid., p Hans Reichenbach, Über die physikalischen Konsequenzen der relativistischen Axiomatik, in: Zeitschrift für Physik 34, 1925, pp See especially Section II, pp Ibid. Reichenbach, Axiomatik der relativistischen Raum-Zeit-Lehre, loc. cit., p. 58. I have used here Maria Reichenbach s translation, loc. cit., p. 76. Emphasis is in the original. Ibid., p.10. Berlin, Springer There are in fact two appendices. Reichenbach obviously intends to refer to Appendix I. Berlin: Walter de Gruyter Translation by Maria Reichenbach, The Philosophy of Space and Time. New York: Dover Ibid., p Emphasis mine. Reichenbach, Axiomatik der relativistischen Raum-Zeit-Lehre, loc. cit., p. 8. The translation is mine. Maria Reichenbach s translation, loc. cit., p. 11, does not adequately convey the intent of the passage. Maria Reichenbach s translation, loc. cit. p See, for example, p. 204 of Philosophie der Raum-Zeit-Lehre, loc. cit. (p. 176 of Maria Reichenbach s translation, loc. cit.) Note that the example worked at length in 26 is an explicit illustration of this. Maria Reichenbach s translation, loc, cit., p Ibid. This pagination refers to Maria Reichenbach s translation, loc. cit. In the original it is p Maria Reichenbach s translation, loc. cit., p It has recently come to my attention that this result is reported in R. Anderson, I. Vetharaniam, and G. E. Stedman, Conventionality of Synchronization, Gauge Dependence and Test Theories of Relativity, in: Physics Reports 295, 1998, pp , and can be traced back to section 9.16 of C. Møller, The Theory of Relativity, second edition. Oxford: Clarendon Press, Department of Philosophy Johns Hopkins University Baltimore, MD U.S.A. ryno@lorentz.phl.jhu.edu

143 JUHA MANNINEN TOWARDS A PHYSICALISTIC ATTITUDE In a footnote to his paper Experience, Cognition and Metaphysics (1926), Moritz Schlick mentioned the two philosophers he thought to be the most promising to his field, Rudolf Carnap and Ludwig Wittgenstein. According to Schlick, all knowledge had to be communicable and expressed through various kinds of symbols. But everything that had to do with the qualitative contents of our experience remained private:... the content of the acquaintance cannot be conveyed by any knowledge; it is not expressible or communicable. 1 Schlick emphasized: This account is absolutely fundamental, and of immeasurable import for the whole of philosophy.... everything qualitative or related to content in our experience must forever remain private and can in no way be known in common to many individuals. However paradoxical it may sound, it is literally true that all our statements, from the commonest utterances of daily life to the most complex assertions of science, reflect only formal relations of the world, and that nothing of the quality of experience enters into them. 2 It was a confirmation of this epistemology that Schlick was now expecting from the two authors he mentioned in his footnote: Cf. the acute and irrefutable remarks of Rudolf Carnap in his forthcoming work, Der logische Aufbau der Welt, where he shows that all scientific judgements must confine themselves to purely structural assertions this concept corresponds to our formal relations. We may add that this holds good of all meaningful judgements whatsoever, since the arguments remain valid for all statements, even the non-scientific ones. Cf. also Ludwig Wittgenstein, Tractatus Logico-Philosophicus,... 3 THE FORMATION OF A NETWORK Schlick had published a revised edition of his Allgemeine Erkenntnislehre in 1925, but it was not a foundation for the Circle that was gathering around him, with the sole exception of Herbert Feigl. Schlick s encounter with the thought of Wittgenstein and also with that of Carnap made him feel that philosophy should take a completely new path. This was reflected in a letter he wrote to Ernst Cassirer in March Schlick was ashamed of his own epistemology: Ich bin seitdem durch die Schule der Logik Russells und Wittgensteins hindurchgegangen und stelle seitdem an das philosophische Denken so verschärfte Anforderungen, dass ich 133 F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

144 134 JUHA MANNINEN die meisten philosophischen Erzeugnisse nur mil grösster Selbstüberwindung lesen kann. Den Tractatus logico-philosophicus von Wittgenstein halte ich für die genialste und bedeutsamste Leistung der gegenwärtigen Philosophie. Leider ist er so barock geschrieben, dass wir in meinem philosophischen Zirkel (an dem hauptsächlich mathematische Kollegen teilnehmen) drei Semester gemeinsamer Lektüre gebraucht haben, um uns zum Verständnis durchzudringen. Auch die Persönlichkeit Wittgensteins (der wahrscheinlich nie mehr etwas publizieren wird) ist wahrhaft genial. Ich glaube fest, dass die Philosophie durch die von der neuen Logik ausgehenden Impulse an einen Scheideweg gelangt ist und dass wir uns dem Leibnizschen Ideal des Philosophierens nähern. Die Grenze gegen leeres Reden und Fragen wird noch viel schärfer gezogen werden mussen als bisher. Wir dürfen, glaube ich, unerschütterlich an dem Satz festhalten, dass alle richtig gestellten Fragen prinzipiell lösbar sind, und zwar entweder durch logische Analyse oder durch empirische Feststellung, und das die unlösbaren Probleme nur falsch gestellte Fragen sind. Ich hoffe sehr, dass die Wiener Philosophen in der nächsten Zeit auch einige Beispiele der exakten Methode vorlegen werden, die in Vorbereitung sind... Ein wirklich glänzendes und durchaus grundlegendes Werk dieser Art stellt das Buch Der logische Aufbau der Welt von Catnap dar, der in vergangenen Semester hier bereits gelesen hat. Es macht mit der logischen Methode Russells ernst und wendet sie mit wahrhaft erstaunlichem Erfolge auf die Fragen der Philosophie, z.b. das Kategorienproblem an. 4 Schlick had been the first philosopher to understand and defend Einstein s theory of relativity. Now he felt that something as momentous was happening in philosophy. Again, his services were needed. The problem with Carnap s book was that it had been lying around for a year but that it was still without a publisher. The huge manuscript containing 566 pages was rejected again and again. Carnap was prepared to delete about a hundred pages, but Schlick was opposed to this. Thus Schlick continued his letter to Cassirer with a request for a favor, knowing that the philosopher s brother Bruno Cassirer was a publisher in Berlin. The result was that Ernst Cassirer immediately explained the situation to his brother, who in turn wrote to Carnap asking for more information. But in those economically hard times the risks connected with a book of that size were again seen as too high. Schlick had better luck with Wittgenstein who had come back to Vienna to carry out his architectural project. After meeting the man he considered a genius, he was able to organize a series of meetings between Wittgenstein and the younger members of his Circle. Carnap, of course, was present, and so was Herbert Feigl, Friedrich Waismann and Maria Kaspar. 5 Wittgenstein, despite his studies with Russell in Cambridge, was as much a Viennese as anybody can be, and he most certainly became a member of Schlick s network. The idea that Carnap should be a Viennese philosopher was of a much more recent origin than the book that he was writing. Carnap began his studies leading to the book the very same year that Schlick abandoned his earlier post in Rostock and took over the chair in Vienna, in He took the initiative for the meeting in Erlangen in 1923 where the idea of an exact philosophy, using the logic of Russell, was born in Germany. 6 On the very second day of the meeting, Hans

145 TOWARDS A PHYSICALISTIC ATTITUDE 135 Reichenbach wrote to Schlick about the need for a journal dedicated to the new philosophy, and he remained the main spokesman for the idea. 7 The members of the First Vienna Circle that had gathered together before World War I also had a constitutive role in the founding and work of the Second Circle known as Schlick s Circle. 8 The mathematician Hans Hahn, the physicist Philipp Frank and the economist Otto Neurath were all shaping the Viennese intellectual milieu in their own ways. Frank held Einstein s former chair of physics in Prague. Reichenbach had only a minor post as an assistant at the Technical University of Stuttgart, but he was busily planning the new journal together with a monograph series. He would have the main responsibility and the publisher would be the Springer Company operating in Berlin and Vienna. However, Springer had been planning to have Frank and Schlick as the editors of a series of monographs in the natural sciences, and now the plans took a different twist. These two had academic chairs and careers that overshadowed Reichenbach, and they were asked in 1925 to edit a completely new series entitled Schriften zur wissenschaftlichen Weltauffassung. 9 Frank was planning a new chair for the philosophy of nature at the University of Prague, the speciality of the chair being defined in complete accordance with the interests of Reichenbach. However, with the help of the Berlin physicists, including Einstein, Reichenbach was soon appointed to an extraordinary chair for natural philosophy in Berlin. Carnap was now looking to Prague. In a letter to Schlick at the beginning of 1926 he wrote: Ich bekam einen erfreulichen Brief von Herrn Prof. Frank, dass man mich für eine neu zu schaffende Lehrkanzel für Naturphilosophie in Betracht ziehen werde. Frank hat ja anscheinend mit Ihnen darüber gesprochen. Da er angibt, dass die Sache soweit in Zukunft liegt, dass ich auch im günstigen Falle die Wiener Privatdozentur einige Zeit würde ausüben können, so habe ich ihm meine grundsätzliche Bereitwilligkeit ausgesprochen.... Jedenfalls bin ich zunächst eindeutig auf die erstrebte Wiener Tätigkeit eingestellt und bereite mich innerlich auf diese vor. 10 Carnap was still living in Germany, in Buchenbach near Freiburg. The book was growing to about twice the size of the originally planned 300 pages and Vienna had to wait for its completion. It was a precondition for a lecturer to present the manuscript as a Habilitationsschrift. Finally, in 1926 Carnap was able to give the MS to Vienna, but he missed the Circle s reading of Wittgenstein s work. Still he hoped mindestens noch die Nachwirkungen einmal spüren zu können. 11 NEURATH S COMMENTS TO CARNAP After only short visits to Vienna Carnap was able to begin his lectures near the end of By then his MS had already been circulating among the members of Schlick s Circle. However, the reception of the book was not only laudatory in the style of Schlick. One harsh criticism came from Otto Neurath, but Carnap

146 136 JUHA MANNINEN had the patience to listen to criticism and to different philosophical views. Neurath was not one of Schlick s favorites. His loud voice and his straightforward habits annoyed Schlick, but as a member of the First Circle, Neurath was still unavoidably part of the world that Schlick was inhabitating in Vienna. In addition, Neurath was philosophically and politically a good friend of Frank and a representative of the Austrian radical wing Social Democrats. On November 21, 1926, Carnap jotted down some of Neurath s objections: Meine Darstellung sei leider mehr gegen Realismus als gegen Idealismus gewendet. Zu starke Betonung des methodischen Solipsismus. Das klinge zu individualistisch. Mehr den Objektivismus betonen. Gleich zu Anfang sagen, dass das Ziel eine objektivistische Welt, die gleiche für alle Individuen, sei. 12 It is clear that this was a summary of Neurath s wishes. Carnap s second note can be seen as belonging to the same category, but it can also be his attempt to take into account the main emphasis of Neurath s objections: 224 Der Realismus des Physikers bleibt bestehen, nur wird er korrigiert zu einem Objektivismus. Vielleicht so: Die gesetzmässigen Zusammenhängen sind objektiv, d.h. dem Willen des Einzelnen enthoben; es gibt aber keine Materie, der man Realität zusprechen könnte; das ist ein metaphysischer Begriff. 13 Neurath had also made some suggestions about how the text could be condensed. Among them was the proposal to leave out Chapter V. In Carnap s letter to Schlick in May 1926 the content of Chapter V was explained: philosophical problems. At that time, Carnap himself had been prepared to exclude it:... könnte ganz wegfallen, da er für den Zusammenhang der Darstellung der Theorie nicht notwendig ist, sondern nur Anwendungsbeispiele bringt; möglicherweise lässt er sich dann später zu einer selbständigen Veröffentlichung verarbeiten. 14 But now, probably because of Schlick s attitude against the shortening of the MS, Carnap commented: Vielleicht ja, aber dann das wichtigere daraus in die vorderen Abschnitte verarbeiten. 15 Carnap s last note is enigmatic, not because of its content, but because of its origin: S Es wird zuweilen gesagt, dass dem praktischen Verfahren der Realwissenschaften, insbesondere der Physik, ein unausgesprochener Realismus zugrunde liege. Hier muss aber deutlich unterschieden werden zwischen der Verwendung einer gewissen Sprache und der Behauptung einer These. Die realistische Einstellung des Physikers äussert sich zunächst in der Verwendung der realistischen Sprache; sie ist berechtigt und häufig zweckmässig und berechtigt (vergl. 52). Ein darüber hinausgehender Realismus als explizite These ist dagegen unzulässig; er muss zu einem Objektivismus (wenn man so sagen will) korrigiert werden. Die gesetzmassigen Zusammenhänge (die in den Naturgesetzen als Implikationen formuliert werden), sind objektiv, dem Willen des Einzelnen enthoben; dagegen würde die Zuschreibung der Eigenschaft real an irgendeine Substanz (sie sei nun Materie, Energie, elektromagnetisches Feld oder was immer) aus keiner Erfahrung herzuleiten, also metaphysisch sein. 16

147 TOWARDS A PHYSICALISTIC ATTITUDE 137 The reference to page 566 alluded to the last page of Carnap s own MS, but Carnap was certainly not quoting from his own MS. So how much of the passage quoted belongs to Neurath and how much to Carnap? Carnap included the passage without any alteration of content in his final printed Der logische Aufbau der Welt, where it is to be found as the last paragraph of Part V, on the metaphysical reality problem. The passage was probably Carnap s own formulation to which Neurath s objections gave rise. It was a denial of metaphysical theses, but an acknowledgement of the justified use of realistic language in physical science. But was this something new with respect to the original MS? The question must be left unanswered at this stage. In any case, it can be observed that there was also a fundamental agreement between Neurath and Carnap. In fact, it was the very same feature that attracted Schlick. Carnap recorded Neurath as commenting: Gut ist: dass der gesamte Raum auf einmal konstituiert wird; nicht ein ganzes aus Teilen zusammengesetzt! Sondern die Teile bekommen erst ihre Bedeutung aus dem Ganzen. (Das ist ja überhaupt eine grundsätzliche Einstellung der Relationstheorie: Die Glieder bekommen ihren Charakter nur durch ihre Stellung im Ganzen). 17 The shift toward the acknowledgement of the realism implied by the language of the physicists was a significant step. A precondition for it had been the concentration on language that Wittgenstein had prepared and that was adopted by the Vienna Circle. It is not important whether Carnap s remarks were written down immediately after Neurath s comments or only later. The important thing is that Carnap no longer was completely neutral in his conception of science and that he still was not forced to make concessions to ontological metaphysics. Carnap s first approach had been neutrality backed by a belief in the power of topology and Bertrand Russell s logic of relations. That kind of exact or scientific philosophy had been the theme of the meetings in Erlangen where Carnap circulated the first sketch of his forthcoming big book, then entitled Vom Chaos zur Wirklichkeit. There seems to be no indication of an essential change of his programme before his remarks on the comments made by Neurath. In the invitation to Erlangen, Carnap had stressed that the discussions should concentrate on die methodischen Prinzipien zur Konstituierung der Naturwirklichkeit aus dem Gegebenen und die Struktur des Wirklichkeitsbereiches. Wirklichkeit im Sinne der materiellen oder physischen Welt, Aussenwelt ; das Gegebene im Sinne der ungedeuteten Sinnesempfindungen (Machs Elemente, Ziehens Gignomene ). Alle diese Ausdrücke sind ohne ihre metaphysische Nebenbedeutungen zu verstehen; neutral ausgedrückt: Konstituierung des sekundären Bereiches aus dem primären. Die metaphysische Frage nach dem Realitätswert der beiden Bereiche wird nicht behandelt (z.b. ist das Primäre bloslich Gegebene und das Sekundäre eine ökonomische Fiktion?) 18

148 138 JUHA MANNINEN Reichenbach was immediately critical of Carnap s approach when he read the MS of the big book. But he had nothing new to offer, only a plea for a realism of the external world. 19 Later on, Carnap s terminology of a primary and a secondary area or system began to circulate in Vienna, as was evidenced by Wittgenstein s adoption of it. 20 A quite different development became viable when Carnap agreed, in his notes on Neurath, that the physicist s realistic language was practical and justifiable, as opposed to realism as an explicit thesis. The epistemological path was now in need of a complementary treatment, doing justice to the ways of science. A REMARKABLE YEAR OF CRYSTALLIZATION In the fall of 1927, Frank wrote to Schlick about the planned monograph series they were to edit together. He listed of seven authors and their themes, among them Otto Neurath and die Fragen einer exakten Wertrechnung, insbesondere über ihre Möglichkeit. 21 It was not listed among the books that were to appear first. Later, when the Circle had gone public, Neurath s theme was announced in a different form: Der wissenschaftliche Gehalt der Geschichte und Nationalökonomie, vol. 4 of the series. Finally, in 1931 Neurath s Empirische Soziologie really appeared in the series with the title announced earlier as its subtitle and with a changed volume number, number 5. The delay cannot be explained simply by the fact that Neurath had no academic position and no opportunity to be a full-time writer. The theme that Frank mentioned in his letter was one that had been attracting Neurath since before the First World War, but there is no indication that he began to work on it when the planning of the book series started. It is probable that Neurath had talked about it with Frank only to reserve a place for the scientific book that he would like to write. Four of the names on Frank s list were members of the First Vienna Circle: Frank, Hans Hahn, Neurath and Richard von Mises. The logician Paul Bernays and the physicist Paul Hertz, both from Göttingen, were also on the list. Hertz had edited together with Schlick the epistemological writings of Hermann von Helmholtz. Frank mentioned further the contributions of Karnap whom he apparently knew through discussions with Schlick. And in a second letter Frank agreed with Schlick s proposal to include Friedrich Waismann. 22 The first book to appear was in fact that by von Mises, his Wahrscheinlichkeit, Statistik und Wahrheit. It was published as vol. 3 in The second book was Carnap s Abriss der Logistik, published as vol. 2 in As time went on, Bernays, Hahn and Hertz withdrew from the project and new authors appeared on the scene. Waismann s book never actually appeared in the series, but it would be constantly in preparation and despite the publisher s protests Schlick reserved for it the number 1 for all eternity. The summer of 1927 was quite unique in the history of the Vienna Circle. Schlick was sure that philosophy had come to a turning point. Both Wittgenstein

149 TOWARDS A PHYSICALISTIC ATTITUDE 139 and Carnap, in Schlick s eyes the main figures behind the big reform, were in Vienna. Discussions with everybody were possible, both in Schlick s official seminar and outside it. Of course, Neurath had been critical of Wittgenstein and the metaphysics he saw in the Tractatus. Thus the genius could not be brought together with Neurath in the formal seminar, but Schlick s network was not restricted to the meetings on Thursday evenings. Schlick could not foresee that communication between Carnap and Wittgenstein would come to an end after only a few meetings. However, something from the encounter with Wittgenstein s Tractatus was impregnated in everybody within the Circle. A touch of Wittgenstein was the Circle s distinguishing feature, for instance, as compared with Reichenbach s Circle in Berlin. 23 Even Neurath admitted later to a young friend of his: Wir die älteste Gruppe hatten Logisierung, Axiomatisierung geschätzt, was die Publikationen deutlich zeigen, aber jene Betonung des logischen Instruments war uns nicht so wichtig erschienen, das Problem der Sprachanalyse nicht so abgesondert zum Bewusstsein gekommen. Da war die Lektüre Wittgensteins sehr heilsam. Einerseits entwickelte sich so zum ersten Mal die gemeinsame Plattform so richtig, andererseits entstanden so die ersten Keime gewisser Gegensätze. 24 For Carnap, the summer of 1927 also meant a renewed and deeper contact with Neurath. This time Carnap was lecturing in Vienna from the beginning of May and he left the city only after the first week of August. In addition to the meetings of the Circle and the meetings with Wittgenstein Carnap had meetings in Neurath s home where Neurath s wife Olga, Hahn s sister, also was present, together with Friedrich Waismann. The letters sent the next winter contain remembrances of this unofficial Circle. Carnap did not come to lecture in Vienna in the winter 1927/1928. He went to Davos in Swizerland for health reasons, but he continued writing at an unhealthy pace. On the last day of 1927, Neurath sent Carnap a letter displaying both politeness and undoubted genuine feeling: Lieber Carnap! Gute Grüsse, Frohe Weichnachtswünsche von mir, meiner Frau und unserem Symphilosophanten Waismann, ditto Neujahrswünsche! Ich nehme wenige ehrwürdige Philosophen öfter zu mein ungewachsenes Maul, als gerade Sie anerkennend und milde aberkennend. Es wäre mir wesentlich wohler auf Erden, wenn Sie den Ruhm der Wiener Schule vermehren würdennicht in Davos, sondern meinen altmodischenräumen Lassen Sie es sich gut sein, schreiben Sie mir und meiner Frau, die wir Sie lieb haben. Schreiben Sie lang, belebt und belehrend, uns, die wir der Weisheit beflissen, die Verknüpftheiten des Daseins erleben und kaum durchschauen. Ihr O.N.

150 140 JUHA MANNINEN [Here was a picture drawn by Neurath, presenting two dragonheads threatening each other but with one body, J.M.] Alles Schöne zum Neuen Jahr! In den nächsten Tagen schreibe ich ausführlich. Fritz Waismann 25 Carnap began his reply when his stay in Davos was coming to its end. The letter to Lieber Neurath und liebe Neuräthin had the first sentence dated February 25, 1928 in Davos, and the rest March 3, 1928 in Tyrol. Carnap wrote: Auch ich freue mich darauf, im Sommer wieder in Wien sein zu können und dann mit Ihnen wieder unsere Diskussionen aufzunehmen. Oder haben Sie inzwischen schon mit Waismann das geschichtsphilosophische Thema zu einem abschliessenden Ende gebracht? Es wird uns aber an Problemen niemals fehlen, denke ich.... Ja, die Verknüpftheiten des Daseins sind wohl manchmal schwer durchschaubar, aber ich bin trotzdem nicht geneigt, sie als ein dräuendes Drachengewirr [a reference to the picture drawn by Neurath, J.M.] anzusehen; ich denke, dass sie sich schon irgendwie entwirren werden, wenn man sie mit Zuversicht, Ruhe und Geduld ins Auge fasst. 26

151 TOWARDS A PHYSICALISTIC ATTITUDE 141 This was actually what Carnap had done during the winter. First, he finished his Abriss der Logistik, accepted for the Circle s series. The manuscript was not too long. It did not need much editing. Second, he wrote the Aufbau afresh and decided on how to publish it. The manuscript had been rejected by Meiner, Mohr, Cassirer, Markus, de Gruyter and a few others. In Schlick s opinion, the big book should have been volume number one in the Schriften zur wissenschaftlichen Weltauffassung, beginning the whole series, but Schlick did not succeed in persuading the publisher to make such a spectacular prelude. 27 Springer had shown interest in the book, but would make no commitment until seeing how the smaller Abriss sold. Carnap felt that he could not wait. Wilhelm Benary had advised applying for subsidies from the Notgemeinschaft Deutscher Wissenschaft. 28 Benary was running the Weltkreis publishing company which had a small but impressive profile. Weltkreis Verlag in Berlin had grown out of the now defunct Erlangen Academy where the first colloquium of scientific philosophy had been held and Benary had edited the Academy s journal Symposion where Carnap had already published some of his work. Before Carnap left Vienna in the summer of 1927, he learned that he would get from the Notgemeinschaft a sum of 1000 German marks for the publication of the Aufbau. 29 It was then that Carnap sold his house in Jena and so he had some money that he decided to put into the same project with the Weltkreis company as publisher. 30 Thus in Davos, Carnap had devoted his energies to his MS, rewriting substantial parts of it. 31 Thirdly, Carnap also decided to use the Weltkreis also to publish as a small book two of his essays that had been rejected by the journals Kant- Studien and Annalen der Philosophie. This book was to be the Scheinprobleme der Philosophie. Das Fremdpsychische und der Realismusstreit (1928). The health cure in Davos had been effective, and Carnap then spent a couple of weeks in Tyrol. NEURATH S UNPUBLISHED MANUSCRIPTS The discussions in Vienna in the summer of 1927 also seem to have prompted Neurath s activity. Unlike the Abriss and the Aufbau, Carnap probably wrote the two essays that were to form his Scheinprobleme in Vienna. The manuscripts of the small pieces were certainly debated in Neurath s Circle. This can be inferred from a letter that Carnap sent to Neurath in October 1928: Heute erhielt ich eine Karte von Ph. Frank über die Scheinprobleme. Ich zitiere daraus, weil es Sie und besonders Ihre Frau sicher interessieren wird, die ja an dem MS so tätigen Anteil genommen hat It is said that Olga Neurath never spoke in Schlick s Circle, but the opposite seems to have been the case in a smaller and more familiar circle. In the same letter, Carnap sent his regards to Liebe Frau Neurath! concerning their recent discussion of the views of Eino Kaila with whom Carnap was then corresponding about the structural conception of knowledge. Olga Neurath ought to call

152 142 JUHA MANNINEN Waismann and ask him to read from Kaila s letters the passages about the reality problem, obviously confirming Carnap s confidence in Kaila. 33 It was quite natural that Olga Neurath, the blind female logician, was enthusiastic about Carnap s relational, structural conception of knowledge. And again we find Carnap in contact with the Neuraths and in the company of Waismann. Back in the summer of 1927 Neurath was either writing his Marxist book Lebensgestaltung und Klassenkampf (l928), or possibly awaiting its publication. Toward the end of the book Neurath expressed his hopes that the working class would take over the task of carrying on science without metaphysics. 34 What exactly was the theme in the philosophy of history that Carnap had been debating with the Neuraths in the summer of 1927 and to which he wanted to return again the next summer? What did Carnap mean here with his talk of Neurath s and Waismann s joint work that they would continue while he was absent? In the Vienna Circle Archives of the Rijksarchief in Noord-Holland, Haarlem, there are two unpublished manuscripts entitled Der wissenschaftliche Gehalt der Geschichte und Nationalökonomie. One of them is handwritten, contains 65 pages and has the code K.3. The text is written on the right sides of the notebook with a few additions on the blank left sides, sometimes even in the margins. The second one is typewritten. It consists of 155 pages, including some advice for the printer. In the eyes of its author, it is a finished MS. It has the archive sign K.2. The shorter text is the older one. The longer text is definitely from a later time. The older text is very tentative. Let us call it the WG1 and the other one WG2. The earlier MS probably dates from the summer 1927 or from the months immediately following it. The text is by no means finished, but a sketch of interconnected themes, of something only in the planning. The handwriting of the sketch is not Neurath s. A comparison with Waismann s handwriting from manuscripts in the Bodleian library, Oxford, shows that it was Waismann who wrote down Neurath s ideas. 35 I cannot exclude the possibility that WG1 was already written before the summer of 1927, but there are indications that it was written just that summer or a little afterwards, at the same time as and independently of Carnap s Davosian efforts. Actually, Carnap s letters to Neurath contain almost a definite description of the process that led to WG1. The title of the MS suggests that the author was aware of attempts to formulate a criterion of scientific content with the aim of separating science from metaphysics. Carnap s two Viennese papers that Neurath read in the summer of 1927, the later Scheinprobleme, developed this line of thought. According to Carnap, every scientific proposition had to express a state of affairs consisting of elementary experiences. Neurath had no sympathy for such experiences, but he was eager to find ways of separating science from metaphysics. Both the WG1 and Scheinprobleme are concerned with this task, but in different ways. Characteristically, Carnap was in search of a logical criterion, whereas Neurath believed that only by developing a new, structurally oriented science could everyday intuitions be left behind.

153 TOWARDS A PHYSICALISTIC ATTITUDE 143 A PHILOSOPHY OF HISTORY Neurath s WG1 was a the sketch for a study of concept formation in history and the social sciences. It was concerned with the language, conceptual structure and, in the end, with the common tasks of all areas of research, seeing them in a perspective which contains a number of the elements of Neurath s later physicalism, but not all of them. It was completely different from the explorations in Lebensgestaltung und Klassenkampf, and it belongs clearly to the history of the Vienna Circle. Neurath s manuscript was not concerned with Carnap s method of constitution, his quasi-analysis, or with the Russellian part of Carnap s programme. Instead, Neurath s text began directly with the quantitative physical world and the presuppositions for its description. Das physikalische Weltbild ist dasjenige Weltbild, welches unabhängig von der spezifischen Sinneserfahrung ist. (Gegen Mach. Mach stützt den Individualismus!) Objektiv kann nur dasjenige Weltbild sein, das allen Erfahrungen gemeinsam ist daher ist es mathematisch. Blinden und Tauben muss etwas gemeinsam sein. Dass es etwas gemeinsames gibt das physikalische Weltbild ist nicht selbstverständlich. 36 For Neurath, numbers were the basis for intersubjective knowledge. People could agree about them much more easily than about such subjective elements as sensory qualities like colors. This pragmatic argument was in accordance with Neurath s work in visual statistics. In Neurath s opinion, the physical world picture was both intersubjective and intersensual. But is it also universal? In the WG1, Neurath denies the possibility of this universality: Die Welt könnte in Teilwelten zerfallen, die nichts gemeinsam haben ja, jede solche Welt könnte noch für sich mathematisierbar sein, ohne dass es ein gemeinsames Weltbild gäbe.... Es kann Welten geben aus verschwommenen Objekten, nebelgleich ineinander fliessend, ohne scharfe Grenze, ja ohne schaft gesonderte Sinnesqualitäten die sich der Logik und Mathematik entziehen. 37 In this and other passages Neurath argued against the universality of physics. He was skeptical about Carnap s ideas: Prinzipiell lässt sich alles darstellen wie das Physikalische. Aber diese Möglichkeit ist noch weit entfernt von der wirklichen Durchführung. Gefahr der Überschätzung des Bloss-Möglichen (z.b. bei Carnap). 38 In his essays on the history of optics, Neurath had adopted the specific view that Pierre Duhem had about the relations of everyday language and the concept formation of physics. Duhem had thought that at the bottom of our most clearly formulated doctrines we always find a confused collection of common sense truths that are deficient in detail and precision, unable to be decomposed into

154 144 JUHA MANNINEN simpler elements. Still, all scientific clarity and certainty are a reflection of the clarity and an extension of the certainty of these common sense truths. 39 Duhem presupposed a continual circulation and a mixed zone between these two:... physical science is both certain and clear because the streams which spring from these two sources run together and mingle their waters intimately. 40 In the WG1, Neurath coined the term Ballungen or aggregations for these imprecise and yet important elements of everyday experience. He introduced the term with a reference to the holistic outlook that was being developed in Gestalt psychology as a criticism against Mach. Everyday objects escaped the purity of both the phenomenal and the physical. However, there was nothing wrong with them: Das meiste, worüber wir Aussagen machen, ist gar nicht präzisiert. Die Schwierigkeiten beginnen erst mit dem Präzisieren. Unpräzisiert sind manche Aussagen oft ganz richtig und werden erst problematisch, wenn sie präzisiert werden sollen. Wir haben eine lange Erfahrung, mit Begriffen zu hantieren, die nicht ganz präzis sind. 41 Developing the ideas first presented in his essay about the lost wanderers of Descartes (1913), Neurath emphasised the importance of decisions for all of science. A preliminary map can only be made better by adopting it, working with it and then correcting it. Science was just such a map. Research presupposed continuous decision making. Metaphysics was the result when these decisions were made tacitly. Once one became aware of them, metaphysics was obsolete. 42 But this was also true of the philosophy of history. According to Neurath s view, every comprehensive philosophy of history was built on decisions. Consequently, none should ever be upheld by coercive means against dissenters. Ich möchte zeigen, dass das Ergebnis einer geschlossenen Geschichtsauffassung ein Entschluss ist und nicht eine durch Erfahrung gegebene Tatsache. Ich möchte zeigen, dass die Erfahrung stets vieldeutig ist. 43 History and the social sciences were at bottom unified and embedded in a broader context of nature. Only the living conditions of people can be seen as something immediately given, but one should search for structures that were theoretical constructions. In the manuscript, Neurath did not deny the necessity of empathetic understanding in the social sciences. In the course of time such an approach could perhaps be replaced by a more scientific one, but Neurath was still warning about overrating causalistic ways of thought. Neurath was sure that there were objects that should be eliminated altogether from a scientific approach, for instance the racial construction of history. However, Neurath did not find it easy to delineate the objects of history and the social sciences. Actually, the sketch ended with an open question: Welche Objekte gestatten zusammenfassende Aussagen? 44

155 TOWARDS A PHYSICALISTIC ATTITUDE 145 THE SIMULTANEOUS BIRTH OF TWO PHYSICALISMS Neurath s remarks from the end of the year 1926 were not unimportant to Carnap when he finally rewrote his Aufbau. One has to keep in mind that these remarks were followed by intensive discussions in the summer of There was an important passage in the notebook that found its way directly into the Aufbau. Moreover, Neurath suggested that Carnap should say in the very beginning that the aim of the book was an objectivistic world shared by all individuals. Carnap did what he was encouraged to do, saying exactly this at the end of his 2. With the original manuscript still missing, it is difficult to speculate what else there might be along the same lines. All traces have not been lost, however. Special weight must be given to a letter that Carnap wrote while revising and completing the book. Carnap s family had come to Davos for Christmas, but the philosopher still wrote a letter to Schlick. He was wondering whether he should save the name of the book, suggested by Schlick, for a later, quite different study. The title was now misleading, he explained: Der bisher beabsichtigte Titel Der logische Aufbau der Welt. Versuch einer Konstitutionstheorie der Begriffe scheint mir in Konflikt zu geraten mit einer Arbeit, die ich für später plane. Das Konstitutionssystem des Buches hat eine eigenpsychische ( solipsistische ) Basis.... An einer Stelle des Buches deute ich kurz an, dass ein anderes Konstitutionssystem möglich ist; mit physischer ( materialistischer ) Basis. Reihenfolge: Physisches, psychisches (ohne Unterscheidung des Ich), Geistiges. Die Leistung dieses Systems ist eine andere: es dient nicht der Erkenntnistheorie, sondern der Realwissenschaft. Es hat als Basisgebiet dasjenige Gebiet, das als einziges eine durchgehende eindeutige Gesetzmässigkeit seiner Vorgänge besitzt. Die psychischen und geistigen Gegenstände werden hier aus den physischen konstituiert... Das zweite System kann aber vielleicht stärkeren Anspruch auf den Namen geltend machen: es baut die Wirklichkeit auf als das eine gesetzmässige Gesamtgeschehen in Raum und Zeit; und ist es nicht vor allem dieser Geschichtspunkt, den wir in der Realwissenschaft bei dem Wort Wirklichkeit vor Augen haben? 46 However, Carnap did not change the title to Erkenntnislogik or to Der logische Aufbau der Erkenntnis as he was planning. In this, he followed Schlick s advice to keep the old name. When Carnap left Davos, he sent Neurath a copy of the page pondering the title, with the following wishes:... da Sie, wie ich vermute, gerade besonderes Interesse haben werden für das Begriffssystem mit physischer Basis. Ich vermute, dass wir über die Mehrleistung, aber auch über die Wenigerleistung dieses Systems im Vergleich zu dem mit eigenpsychischer Basis einig sein werden. 47 Why did Carnap not suggest anything similar to Schlick? Of course, he had discussed these questions with Neurath, and he knew Neurath s philosophical

156 146 JUHA MANNINEN preferences which were quite different from Schlick s. Moreover, the letter could be read as an invitation to continue the memorable discussions. Neurath s WG1 did not, in fact, go as far as Carnap. What Carnap called eine [sic!] gesetzmässige Gesamtgeschehen as it was characterized physically in time and space would also be an answer to Neurath s concerns about the objects of history and social science. When the Aufbau appeared in August 1928, Neurath could read that fundamentally there is only one object domain and that each scientific statement is about the objects in this domain. 48 Carnap s one domain thesis, a corner stone of the Aufbau, may, of course, be much older than the book, even a continuation of his old interest in monism. What made the thesis appealing to Neurath was that this system could also be construed in a physical, materialistic way. Thus, in his WG2, Neurath was no longer worried about the limits of physical language. Within Schlick s network, criticism against the Aufbau was launched very soon. In January 13, 1929, Kaila said in a letter to Schlick how very much he was interested in the von Ihnen angedeutete, von Wittgenstein gegen die Konstitutionstheorie erhobene Einwand, dass es nämlich unmöglich sei, von den Momentan-Gesamt-Erlebnissen als Basis auszugehen. 49 A few days later, Carnap noted in his diary: (Nachdem ein [fruchtloses] Gespräch vorhergegangen, in dem Waismann vergeblich bemühte, Wittgensteins Meinung deutlich zu formulieren.) And further: Über Extensionen. Veranlasst durch Wittgensteins Kritik der Quasianalyse (über die wir gestern gesprochen haben) wollen wir einige logische Fragen klären. 50 These discussions with Waismann and Feigl continued for a number of days in a stricly logical form, but nothing that could be called physicalism was formulated. Neurath was not present, Wittgenstein had already left for Cambridge. Wittgenstein certainly was a critic of Aufbau s starting point and its method, but somehow he did not succeed in saying clearly enough what he had to say. If he had something to replace Carnap s approach, its meaning remained buried and hidden from the Circle, at least from Neurath and Carnap. At the Circle s first conference in Prague, September 1929, Neurath introduced unified science. In the next winter he wrote the first draft of the book planned to carry the title of WG. It was announced that the new book would appear in the Schriften zur wissenschaftlichen Weltauffassung, but Schlick was horrified with the MS and especially with its exclamatory style. He rejected the book. 51 Carnap saw himself in an embarrassing position and offered Neurath help in rewriting the book. 52 Independently, Neurath produced another draft in an astonishingly short time, in just a few weeks. At the beginning of August 1930, Carnap was in Biberwier, Tyrol, together with Feigl and Kaspar. Two years later he told Schlick that he had written the first draft of the essay Die physikalische Sprache als Universalsprache der Wissenschaft during that summer visit. 53 A card that he sent to Schlick from Biberwier reveals how worried he was about the conflict between Schlick and

157 TOWARDS A PHYSICALISTIC ATTITUDE 147 Neurath. Carnap was not allowed to make improvements to Neurath s MS, but instead wrote a paper of his own on the same theme. Then, at the end of the same month, Carnap met Neurath in Vienna, still in the company of Feigl. He wrote to Schlick: Wir haben ihm ruhigend zugeredet. Er war nämlich immer noch erregt und glaubte, dass nicht nur dem MS, sondern ihm persönlich Ungerechtigkeit widerfahren sei. Und nun hatte er tatsächlich schon ein beinahe fertiges neues MS! Er hat uns verschiedene Kapitel daraus vorgelesen. Und wir waren erstaunt zu sehen, dass er hier wirklich in ganz anderem Tone geschrieben hat, sachlich argumentierend, und ernsthaft in der Formulierung. Der Inhalt ist jetzt ein ziemlich anderer, mit gleichen Grundgedanken. 54 It was important for Carnap to resolve the conflict as soon as possible, since he was at least partially responsible for the whole. The first chapter of the manuscript WG2, which was omitted when the Empirische Soziologie was finally published in 1931, was not uninteresting. It was there that Neurath first formulated all of the main characteristics of his physicalism: the intersubjectivity, intersensuality and universality of physical language with a starting point in the Ballungen of our everyday language. 55 However, Neurath did not yet use the term physicalism. This was a term that he seems to have adopted only later in The WG2 was an openly Marxist text, an attempt to join together Marxism and the Vienna Circle. Neurath s physicalism was an attempt to present some of his main convictions along these lines in a form that would also be worthy of debate in Schlick s Circle. With Carnap s support, the introduction of physicalism succeeded in the spring of 1931, and later on. 56 But soon, in the political upheavals of Europe all of Marxism turned out to be a disappointment for Neurath. He never became apolitical, but his interests shifted after his emigration, first to an encyclopedic form of physicalism, then to a defence of a democratic pluralism which he saw as intimately connected with a denial of all systems built on One World assumptions. 57 Who was the first to discover physicalism? The question does not make much sense. Neurath and Carnap were time and again interacting with each other over the course of years, stimulating and criticising each other, and writing independently of each other. Their personal styles, views and aims were so different that it is not altogether easy to describe them as participating in a joint venture. Nevertheless, they shared a common interest in a uniquely close way. The best picture of the situation was probably Neurath s drawing of the two dragons with one body. When they published their articles on physicalism in the Erkenntnis in 1932 and Carnap gave priority to Neurath, he was speaking only part of the truth. He forgot to mention his own role in the complicated story that had begun in Vienna way back in November 1926.

158 148 JUHA MANNINEN Acknowledgments: I am particularly indebted to Dr. Brigitte Uhlemann (Konstanz) for her transcriptions of Rudolf Carnap s texts written with a shorthand and for the Archives of Scientific Philosophy (Pittsburgh University) and the Vienna Circle Archives (Rijksarchief in Noord-Holland, Haarlem) for their permission to quote unpublished archive sources. NOTES M.Schlick, Philosophical Papers, Vol. II ( ), ed. by H.Mulder and B.F.B. van de Velde-Schlick, D.Reidel, Dordrecht 1979, p.99. Cf. M.Friedman, Reconsidering Logical Positivism, Cambridge University Press, Cambridge Ibid., p Ibid., p M.Schlick to E.Cassirer, March 30, The Vienna Circle Archives in the Rijksarchief in Noord-Holland, Haarlem (=VCA). F. Stadler, Studien zum Wiener Kreis. Ursprung, Entwicklung und Wirkung des Logischen Empirismus im Kontext, Suhrkamp, Frankfurt am Main 1997, pp See C.Thiel, Carnap und die wissenschaftliche Philosophie auf der Erlanger Tagung 1923, in: R. Haller and F. Stadler (eds.), Wien-Berlin-Prag. Der Aufstieg der wissenschaftlichen Philosophie, Hölder-Pichler-Tempsky, Wien H.Reichenbach to M.Schlick, March 8, 1923, VCA. Reichenbach planned that he would edit the journal together with Schlick and Kurt Lewin. Carnap was in his opinion too uncritical. Cf. Ph. Frank s memoirs in the Preface to his Modern Science and Its Philosophy, Cambridge, Mass., Harvard University Press 1949, and especially Th. Uebel, Vernunftkritik und Wissenschaft: Otto Neurath und der Erste Wiener Kreis, Springer, Wien L. Friedländer was representing Springer in Vienna and he made the agreement with Schlick and Frank, see his letters to Schlick, August 3, 1925, VCA. R.Carnap to M.Schlick, January 15, 1926, VCA. Ibid. Neurath über Konstitutionstheorie. ASP RC :1R. Archives of Scientific Philosophy of the Twentieth Century, University of Pittsburgh, Special Collections (=ASP). Quoted with the permission of the University of Pittsburgh. All rights reserved. Ibid. R.Carnap to M.Schlick, May 28, 1926, VCA. ASP RC : 1R. ASP RC : 1V. ASP RC : 1R. R. Carnap, Aufforderung zur Teiinahme an Besprechungen über den Aufbau der Wirklichkeit (Strukturtheorie der Erkenntnis), ASP HR Gewiss können Sie das Realitätsaxiom ein metaphysisches nennen; aber ohne dieses Axiom wäre ihr Konstitutionssystem nur ein Schachspiel, und die ganze Wissenschaft auch. Einen Sinn im Sinne der Naturerkenntnis bekommt die Wissenschaft erst mit dem Realitätsaxiom. Ich glaube also, dass Ihre Neutralität ein schöner Traum ist... H.Reichenbach to R.Carnap, February 20, ASP RC See L. Wittgenstein, Wittgenstein s Nachlass, The Bergen Electronic Edition, Oxford CD-ROM 2000, e.g. Item 105, pp.84-8; Item 106, p.203: Item 107, p.205. The world of data was for Wittgenstein the primary or first system with respect to the world of physics which was the secondary or second system. In the Big Typescript, Wittgenstein wrote: Es gibt nicht wie ich früher glaubte eine primäre Sprache im Gegensatz zu unserer gewöhnlichen, der sekundären. Aber insofern könnte man im Gegensatz zu unserer Sprache von einer primären reden, als

159 TOWARDS A PHYSICALISTIC ATTITUDE in dieser keine Bevorzugung gewisser Phänomene vor anderen ausgedrückt sein dürfte; sie müsste sozusagen absolut sachlich sein. Item 213, p Ph. Frank to M.Schlick, VCA. The letter is undated but the approximation Herbst 1927 written by someone else seems reasonable. Waismann had been writing a book for the series, entitled Raumproblem und Phänomenologie; see F. Waismann to M.Schlick, September 20, 1925, VCA. But the idea of writing a book about Wittgenstein went beyond that plan. Reichenbach wrote to Schlick February 22, 1933, VCA, as follows: Nächstens wollen wir [the Berlin Group, J.M.] auch Wittgenstein s Traktat vornehmen. Soon after that Reichenbach was forced to leave Germany, as is well known. O.Neurath to W. Hollitscher, November 22, 1937, in: P.Goller and G.Oberkofler, Walter Hollitscher. Briefwechsel mit Otto Neurath , Die Alfred Klahr Gesellschaft und ihr Archiv. Beiträge zur österreichischen Geschichte des 20. Jahrhunderts, ed. by H.Hautmann, Wien 2000, p O. Neurath and others to R.Carnap, December 31, 1927, ASP RC R. Carnap to O. and O. Neurath, February 25/March 3, 1928, ASP RC L. Friedländer from Vienna informed Schlick about this in May 25, But Schlick made a new attempt, now trying to persuade the Berlin head of the publisher. In November 16, 1928, F. Springer told Schlick personally that the book could not open the series. Both letters are in the VCA. R. Carnap to M. Schlick, March 20, 1927, VCA. R. Carnap s application, April 11, 1927, VCA, and a leaflet informing Schlick about the result in the beginning of August, VCA. R. Carnap to M. Schlick, August 18, 1927, VCA. R. Carnap to M. Schlick, December 23, 1927, VCA: Das MS ist zu 3/5 druckfertig, der Rest wird wohl auch bald fertig sein. Ich habe alles noch durchgearbeitet und überall lieber im Kleinen gekürzt, keine ganzen Abschnitte gestrichen. Das wird dem MS nur nützen, glaube ich, Freilich macht es mehr Arbeit, viele Teile müssen neu getippt werden. R. Carnap to O. Neurath, October 7, 1928, ASP RC Ibid. In his letter in January 28, 1929, Carnap was still teaching the Finnish philosopher: Sie fügen hinzu: im Sinne der realistischen Physik; es gibt aber doch nur eine zulässige, metaphysikfreie Interpretation der Physik. Carnap s letters to Kaila are in Prof. G.H. von Wright s private archive. O.Neurath, Gesammelte philosophische und methodologische Schriften, Vol. 1, ed. by R. Haller and H. Rutte, Hölder-Pichler-Tempsky, Wien 1981, p See J. Manninen, Wie entstand der Physikalismus?, Nachrichten. Forschungsstelle und Dokumentationszentrum für Österreichische Philosophie, Vol. 10, This paper gives a more detailed picture of WG1. VCA ON K.3, pp Ibid. Ibid., p. 22. P. Duhem, The Aim and Structure of Physical Theory, transl. by P.P. Wiener, Atheneum, New York 1981, p Ibid., p VCA ONK.3, p. 11. Ibid., p. 14. Ibid., p Ibid., p. 65. Cf. R. Carnap to O. Neurath, October 7, 1928, ASP RC R. Carnap to M. Schlick, December 23, 1927, a separate sheet Frage über die Wahl des Buchtitels added to the letter. The letter was the same as mentioned in note 26. R. Carnap, Der logische Aufbau der Welt, Weltkreis-Verlag, Berlin-Schlachtensee 1928, 2, 162 and especially 16. Cf. also Friedman, Reconsidering, pp A critical analysis of Carnap s one-domain assumption is given in J. Hintikka, Carnap s Work in the Foundations of Logic and Mathematics in a Historical Perspective, J. Hintikka, Lingua Universalis vs.

160 150 JUHA MANNINEN Calculus Ratiocinator: An Ultimate Presupposition of Twentieth-Century Philosophy, Kluwer, Dordrecht Contrary to Hintikka, the one-domain assumption seems to have been an integral part of Carnap s overall philosophical views. E. Kaila to M. Schlick, January 13, 1929, VCA. To be published in an edition of Schlick s correspondence by R. Fabian, M. Iven and H. Rutte, Springer, Wien. Gespräch mit Waismann. ASP RC Cf. R. Carnap to M. Schlick, September 28, 1932, VCA: Ja ich weiss bis heute noch nicht, wie Wittgensteins Auffassung hier ist, da Waismann ja nicht deutlich Stellung genommen hat, sondern nur sagte, dass er (und Wittgenstein) gewisse Bedenken haben, ohne sie jedoch formulieren zu können. More about the MS and the participants in J. Manninen, Das verbotene Buch des Wiener Kreises, in: Topos, Vol. 17, 2001, p Schlick s letter to Frank, containing his motivations, is to be found among the unidentified letters of the VCA. R. Carnap to O. Neurath, July 25, 1930, ASP RC Carnap s first suggestion was that Heinrich Neider should help Neurath, but he wrote then: Falls Neider nicht will oder kann, ware ich im äussersten Fall bereit, obwohl ich nicht soviel Zeit daran setzen könnte wie N. R. Carnap to M.Schlick, September 28, 1932, VCA. R. Carnap to M. Schlick, August 28, 1930, VCA. See the paper mentioned in note 51. Cf. also J. Manninen, Otto Neurath oder die Unmöglichkeit einer privaten Sprache und die Möglichkeit der Visualisierung, J. Manninen, Feuer am Pol. Zum Aufbau der Vernunft im europäischen Norden, Peter Lang, Frankfurt am Main 1996, pp Cf. Besprechung über Physikalismus am , VCA ON K.4, and the discussions documented in Stadler, Studien zum Wiener Kreis, p. 275 ff. Neurath formulated this in his last period by quoting Karl Popper s view that empirical science should represent only the one real world, the world of our experience, and then he delivered his criticism: I speak of pluri-statements and therefore I see no way how to reach THE ONE WORLD, or one preferred system of statements.... I think that Philipp Frank fought sufficiently the one real world besides the many possible worlds. I myself and he and Duhem and Poincaré belong together, we are only more consistent than Duhem and Poincaré. Neurath also asked: I should like to learn from you what you say about that. No answer was given. O. Neurath to R. Carnap, April 1, 1944, VCA. History of Science and Ideas University of Oulu P.O. Box 1000 FIN Oulu Finland juha.manninen@oulu.fi

161 WOLFGANG HUEMER LOGICAL EMPIRICISM AND PHENOMENOLOGY: FELIX KAUFMANN * 1. THE RELATION BETWEEN PHENOMENOLOGY AND ANALYTIC PHILOSOPHY The history of philosophy of the twentieth century can be best characterized, it has been argued, by the opposition between two of its main movements, phenomenology and analytic philosophy. This strategy seems appropriate since both movements have their beginnings around the turn of the century, and most philosophical positions after the Second World War have their roots in one of the two movements. In addition, there was not much contact between the two movements, but there were exceptions. In this paper I will discuss one of the earliest attempts to actively combine the results of Husserlian phenomenology with the logical empiricism of the Vienna Circle, which was proposed by Felix Kaufmann in a number of articles he published in the early forties. In the last two or three decades an increasing number of philosophers have pointed out that there are important parallels between Husserlian phenomenology and analytic philosophy. Peter Simons, for example, states that the divide between the two traditions did not exist at the turn of the century and only became clearly apparent after the Second World War. 1 Similarly Michael Dummett, who mentions in his History of Analytical Philosophy that to a German student of philosophy in 1903 i.e., two years after the publication of Husserl s Logical Investigations, where he first developed his phenomenological project Husserl and Frege must have appeared [n]ot, certainly, as two deeply opposed thinkers: rather as remarkably close in orientation, despite some divergence of interests. 2 In fact, there are considerable parallels in the positions of the two philosophers, who had exchanged letters and sent each other their publications as early as the 1890s. In addition, Frege reviewed Husserl s Philosophy of Arithmetic, criticizing his project of developing a psychological foundation of logic which, in turn, was one of the factors that convinced Husserl that this project was doomed 3 and to develop his critique of psychologism in the Logical Investigations, where he distinguished normative from pure logic, the latter being about ideal (logical) objects. Moreover, it was this ontological move of accepting the existence of ideal objects that allowed Husserl to further develop Franz Brentano s method of descriptive psychology into his own phenomenological 151 F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

162 152 WOLFGANG HUEMER method, a method that aims at describing the essential aspects of our mental acts. Hence, we could say that Frege indirectly influenced Husserl in his developing the phenomenological method. But also in Husserl s later philosophy, there are striking parallels to Frege, as the ground-breaking work of Dagfinn Føllesdal 4 shows. The exchange and the parallels between phenomenology and analytic philosophy are not limited to the work of Husserl and Frege, though. We find, for example, that there was a strong interest in the early work of Husserl among Anglo-Saxon analytic philosophers in the first two decades of the twentieth century: Russell, for example, brought Husserl s Logical Investigations with him to prison with the intention of reviewing it for Mind without ever fulfilling this intention. In addition, Moore and Ryle showed a great interest in Husserl s Logical Investigations. 5 I assume that their attention to Husserl stems from an interest in Franz Brentano and Bernard Bolzano, and runs parallel to an interest in Meinong s theory of objects. In short, at the beginning of the century the two emerging traditions were rather close, and their relationship was characterized by a mutual interest in the other position. If we look at the situation some thirty years later, however, we find that things have changed dramatically. Husserl s so-called transcendental turn, which was first proposed in his Ideas of 1913, as well as the work of his students, mainly Martin Heidegger, have caused various analytic philosophers to become suspicious of phenomenology, a trend that is reflected in Ryle s articles on phenomenology as well as in the debate between Schlick and Husserl: Schlick harshly criticized Husserl s phenomenological project in the first draft of his book General Theory of Knowledge, to which Husserl reacted in the foreword to the second edition of the Logical Investigations 6, which caused Schlick to revise the argument, but not the tone of his critique in the second edition. After this exchange Husserl ignores Schlick s work, while Schlick goes on to occasionally criticize the phenomenological movement. 7 This exchange between Husserl and Schlick, however, created a clear line of demarcation between the phenomenological movement and the Vienna Circle that was respected by most members of the two schools even though there were some personal connections between the fronts: Rudolf Carnap, for example, attended Husserl s seminars in 1924/25 8. It has been pointed out repeatedly that there are strong parallels between Husserl s notion of constitution and Carnap s constitutional system as developed in The Logical Structure of the World 9. It is not clear how strong Husserl s influence in this phase of Carnap s work actually was 10, but if there was any, it is quite obvious that it diminished drastically in later years. Carnap s attitude towards the phenomenological movement in the thirties is well documented in his harsh attack on Heidegger in the paper The Elimination of Metaphysics Through Logical Analysis of Language, where Heidegger s phenomenology has become the prototype of bad metaphysics.

163 LOGICAL EMPIRICISM AND PHENOMENOLOGY:FELIX KAUFMANN FELIX KAUFMANN The only member of the Vienna Circle who actively tried to bridge the gap between logical empiricism and Husserlian phenomenology is Felix Kaufmann, who was an associated but very active member of the Vienna Circle. 11 Kaufmann was primarily interested in philosophy of law and the methodological foundations of the social sciences. Originally a student of Kelsen, he soon became interested in Husserl, who had a strong influence on his work. Kaufmann contacted Husserl in the early twenties; they met various times and exchanged numerous letters. 12 In their correspondence they hardly discussed philosophical problems, though. The letters are interesting mainly for historical reasons, since they document Husserl s attempt to emigrate from Nazi-Germany to Czechoslovakia in the thirties. Husserl hoped that Kaufmann could support him in this project. Kaufmann did not always have an easy time within the Vienna Circle. Since he did not hide his interest in Husserl, he was considered a phenomenologist by the other members. Gustav Bergmann writes in his Memories of the Vienna Circle that when they [Kaufmann and Neumann, another participant of the meetings who was interested in Husserl] took the floor, during the classical period Schlick occasionally showed a certain impatience and sometimes even interrupted the discussion outright. l3 Bergmann adds that the situation somewhat improved when the Wittgensteinians gained more influence in the Vienna Circle since they [the phenomenologists] claimed to recognize many of the ideas now advanced [by the Wittgensteinians] as old teaching from their own school l4 which was, however, rejected by Schlick and Waismann. The latter, however, in private recommended reading Husserl. 15 This rapprochement might have somewhat improved the situation for the phenomenologists, but definitely backfired for the Wittgensteinians. It was from this side, moreover, that criticism set in. In the Circle Hahn once asked Waismann directly how he distinguished himself any more from a phenomenologist. 16 These remarks clearly show that being associated with Husserlian phenomenology did not improve Kaufmann s position within the Vienna Circle. In the years when he participated at the meetings of the Vienna Circle, Kaufmann was not a professional philosopher, but worked as a manager for the Anglo-Iranian Oil Company. In 1938 he had to emigrate to the USA. He got a call to the New School for Social Research, where he taught philosophy until his death in 1949.

164 154 WOLFGANG HUEMER 3. KAUFMANN S POINTS: EIDETIC INTUITION AND ANALYSIS OF SENSE-DATA After his emigration, Kaufmann somewhat lost his interest in phenomenology. He became interested in Dewey and turned more and more towards pragmatism. Shortly after arriving in New York, however, he wrote some articles in which he discussed the relationship between phenomenology and logical empiricism. 17 In these articles he concentrated on two points. First, he argued that the critique of phenomenology raised by the logical empiricists of the Vienna Circle was based on a misunderstanding of Husserl s notion of eidetic intuition (Wesensschau), which they criticized as bad metaphysics. Second, he gave a phenomenological critique of the empiricist notion of raw sense data. Let me first turn to Kaufmann s discussion of the positivistic critique of Husserl s notion of eidetic intuition. Kaufmann addresses the standard critique brought up against Husserlian phenomenology in the Vienna Circle, especially by Moritz Schlick without mentioning Schlick s name, though. We can see several main strands in this critique. Even though he advocates an empiricist position, Schlick shares Husserl s concerns about psychologism. 18 He cannot, however, accept Husserl s solution, according to which pure logic is concerned with ideal logical objects. For Schlick this move is unacceptable, as it buys into Platonism. And, in fact, Husserl s Logical Investigations do create the impression of having strong Platonistic tendencies, for his arguments are based on the acceptance of a pure logic, a discipline that is concerned with ideal logical objects. However, Schlick s concerns do not really address Husserl s position. As Kaufmann points out 19, Husserl explicitly rejects Platonism in his later work. Moreover, even though some formulations, esp. the talk about ideal logical objects and essences, invite a Platonistic reading of the Logical Investigations, this interpretation is nonetheless a misreading and a misrepresentation of Husserl s concerns. Husserl never explicitly advocates Platonism in the Logical Investigations. In a later text, Draft to an Introduction to the Logical Investigations of 1913 (which was only published posthumously), he complains that many readers have interpreted his text in a Platonistic way: This accusation is not at all justified; it is, rather, in stark contrast to my account and is based on exactly those historical prejudices from which I first had to liberate myself in an arduous way. 20 The historical prejudices Husserl is speaking about have to do with the identification of an acceptance of pure logic with the acceptance of a Platonistic ontology, in which Husserl believed when he first read Bolzano, i.e., more than ten years before he wrote the Logical Investigations: I mistook, however, his [i.e., Bolzano s] original thoughts on presentations, propositions, truths in them-

165 LOGICAL EMPIRICISM AND PHENOMENOLOGY: FELIX KAUFMANN 155 selves to be metaphysical absurdities 21. Only by reading Lotze, Husserl states, did he realize that he was misreading Bolzano, which influenced the development of his critique of psychologism in the Prolegomena of the Logical Investigations in a decisive way. Thus, one can say that Husserl s terminology, at least in his Logical Investigations, was not very clear, or that he should have added a few clarifying paragraphs, but not that Husserl advocates a form of Platonism in this work. Schlick s point concerning Husserl s alleged acceptance of Platonic essences is, however, connected to a second strand of criticism which might even apply to a non-platonistic reading of the ideal object theory. This critique is directed against the kind of mental act in which we are, according to Husserl, directed towards these ideal logical objects, i.e., against the intuition of essences. Schlick doubts that we are in a position to intuit abstract essences. We ask: How in general can non-real objects concepts or judgments be given to us when all that we are acquainted with as given are real contents of consciousness? 22 Somehow, Schlick continues, these non-real, ideal objects must be real contents of conscious acts otherwise we could not have any knowledge about them. He goes on to criticize Husserl for explaining this problem by distinguishing between real and ideal evidence, the latter of which can account for our grasping ideal objects in eidetic intuition. Schlick counters that this does not solve the problem, since we could ask again how one knows about ideal evidence. This would have to be a real property of our mental acts 23, and we face the same problems as before. The strategy adopted by Husserl, according to which we can acknowledge a fact only if we know it with a high degree of evidence, cannot possibly solve any problem, Schlick argues, since one must be aware of the fact that one knows with a high degree of evidence; i.e., one would need evidence for this second order knowledge, which would lead to an infinite regress. This empiricist critique of Husserlian phenomenology, Kaufmann argues, is based on a misunderstanding of Husserl s notion of evidence. Rather than formulate an argument, Kaufmann quotes Husserl s Formal and Transcendental Logic of 1929, a book with which Schlick was likely unfamiliar. In this quotation, Husserl talks about the sensualistic misunderstanding of the notion of evidence, i.e., the characterization of evidence as a certain form of feeling we experience in conjunction with a certain act. Husserl calls that view an absurdity and explains that evidence is not a feeling that accompanies the act, but rather an intentional accomplishment, something that does not arise from one act alone, say a perception, but rather in connection with recollections. This interplay of various partial intentions gives for the first time original certainty in the full sense of the existence of a subjective object called a psychic datum... something to which one can always return and which one can always recognize in reactivation as the Same. 24 In short, evidence is for Husserl not a feeling that accompanies the act, but an accomplishment that makes perception of objects possible. Schlick s critique,

166 156 WOLFGANG HUEMER thus, stems from a misunderstanding of Husserl s notion of evidence and a Platonistic misinterpretation of Husserl s attack on psychologism. With these remarks, Kaufmann showed that the empiricist critique of the phenomenological notion of eidetic intuition is not justified. Kaufmann s second point consists in his argument that the basic assumption of logical empiricism, according to which raw sense data cannot be further analyzed, is itself a pseudo-explanation. Empiricists still do not recognize with sufficient clarity how close the kinship is between the idea of objectively given sense-data, inaccessible to further analysis, and the metaphysical ontology which they oppose. 25 In this context, Kaufmann discusses Carnap s critique of synthetic a priori statements which is based on his distinction between pure and interpreted calculus, where the former is concerned with the signs on a merely syntactical level, whereas the latter assigns meaning to these signs. According to this critique, the tendency to accept synthetic a priori statements stems from a confusion of syntactical sentences, i.e. sentences about language, on the one hand, and object sentences, i.e. sentences about objects in the narrow sense, on the other. The sentence Every color is at a place, is often quoted as an example of a synthetic a priori statement. According to this critique, this is a mistake that is based on a misunderstanding of what the sentence is actually about: it is misinterpreted as an object-sentence when it actually is a syntactical sentence; it is not about empirical objects in the world, but rather informs us how the words it contains are used in language. This becomes most obvious if we substitute the formal language for the object language: Then, e.g., instead of the sentence Every color is at a place, the sentence A color-expression is always accompanied in a sentence by a place designation, would occur. 26 According to this view, the rules of a pure calculus can be described entirely within the limits of logical syntax. 27 In consequence, there is no need for synthetic a priori statements, it is rather determined by the rules of syntax whether a certain expression can be formulated or not. When we actually use a language we have to add the semantic level; i.e., we have to interpret these signs, which can be done by giving a series of correlative definitions that state the meaning of the signs used. This meaning is determined by indicating experiences that shall constitute the sufficient condition for using the expressions in question, e.g., it is stipulated that one may use the sentence A blue thing is at the place p at time t, on the basis of an optical perception of a certain kind. 28 It is at this point that Kaufmann s critique sets in: this analysis of Carnap s position shows, according to Kaufmann, that the logical empiricists ascribe a role to raw sense-data which they cannot play. Carnap s analysis presupposes that [s]igns and formulae are given as objective, i.e., are regarded as identifiable at

167 LOGICAL EMPIRICISM AND PHENOMENOLOGY: FELIX KAUFMANN 157 different times and by different persons. 29 This creates a tension within the empiricist program that bases everything on raw sense data: The empiricist, however, who defines the world in terms of possible experience and bears in mind that every experience is the experience of a particular man at a particular time, should see clearly that the presupposition of an intertemporal and intersubjective harmony of experiences a presupposition implicit in operating with objectively given signs is accessible to, and urgently requires, further analysis. 30 Kaufmann points out that Carnap s program presupposes that we have objective experiences and that the meaning of the words we use have their source in these experiences. What happens, however, if we ask for the meaning of objective experience? We are then confronted with the following situation: The meaning of objective experience is defined in terms of certain rules of method including presuppositions concerning given fundamental meanings. These meanings in turn point back to experiences as constitutive elements. Obviously the meaning of experience cannot be the same in both cases. 31 Kaufmann argues that here we are facing two different strata of experience. In other words, we are using the word experience in two completely different ways. On the one hand, as we have seen above, we use it to define the meaning of other words (like blue, yellow, etc). If we try to clarify the notion of experience, on the other hand, we find that we need to make important presuppositions to use that concept. We conceive of the physical world that is experienced as being the same for every human being. The notion of human being, thus, is presupposed by the notion of objective experience, etc. On this lower stratum of experience, we ask what presuppositions are implicit in the idea of an objective world of experience and how these experiences enter into every single experience of the individual. 32 In consequence, the differences between phenomenology and logical empiricism may be seen as resulting from these distinct strata of experience. Both see the task of philosophy in analyzing meaning. As far as this analysis remains within the stratum of objective meanings it is logical analysis; as far as it transcends this stratum it may be called transcendental analysis. 33 If the logical empiricist would gain some appreciation for this distinction of various strata of experience and try to analyze the meaning of experience on the lower stratum, he learns to see that even the simplest object-perception contains a manifold of complexly interwoven anticipations of one s own and other persons potential perceptions anticipations that my be fulfilled or disappointed. 34 Kaufmann argues, as this quotation shows, that the logical empiricists base their position on the notion of experience which, however, should itself be analyzed. If they would direct their attention to this basic level of their theory, and transcend the stratum of objective experience, they would appreciate the need for

168 158 WOLFGANG HUEMER a holistic understanding of experiences. In its present form, logical empiricism accepts a metaphysical assumption that is not proven, namely that there are experiences that cannot be further analyzed. Logical empiricists, thus, have to face a critique they have brought up against phenomenology: namely, that they do bad metaphysics. Kaufmann s argument, thus, amounts to abandoning the very project of logical empiricism or at least adding a whole new dimension to it, since it challenges its empiricist basis. This does not mean, however, that Kaufmann s diagnosis stands in contrast to the program of the Vienna Circle, since it is compatible with a scientific approach to philosophy and a positivistic point of view. On the contrary, if Kaufmann s argument is right, not to reform one s position in the way suggested and to continue holding the notion of raw sense data would amount to accepting a piece of bad metaphysics. With these two points, Kaufmann has reached his goal to show first that phenomenology and logical empiricism are not incompatible, but rather complement each other on different levels, and second that if the logical empiricists are consistent in seeking their goal, namely, the analysis of scientific methods, then the problems that form the point of departure for phenomenological reflection must emerge within their field of vision. 35 Thus, if the logical empiricists take their own program seriously, then sooner or later they should, according to Kaufmann, gain appreciation for the phenomenological program and possibly start doing phenomenology themselves. 4. ANALYTIC PHENOMENOLOGY We have seen that one of the decisive reasons for the schism between phenomenology and logical empiricism was apart from the differences in their positions which were not that insuperable after all the debate and the resulting personal antagonism between Schlick and Husserl. This mutual opposition continued after the death of its proponents, when most of the members of the Vienna Circle unlike some of the prominent proponents of the phenomenological movement had to emigrate from Europe. In North-America, Schlick s article Is There a Factual A Priori, 36 where he criticizes the phenomenological notion of the synthetic a priori, was anthologized in Feigl s and Sellars book Readings in Analytic Philosophy, and so created a bad press for phenomenology among a younger generation of Anglo-North-American analytic philosophers. Kaufmann s article, on the other hand, was published in a book that was only interesting to North American phenomenologists, and thus escaped the attention of philosophers interested in logical empiricism. It is noteworthy, however, that with his critique Kaufmann points his finger at that weak point of logical empiricism that was aimed at also by Wilfrid Sellars attack on the Myth of the Given 37. Kaufmann, thus, uses Husserlian phenomenology to criticize an assumption of logical empiricism that also has

169 LOGICAL EMPIRICISM AND PHENOMENOLOGY: FELIX KAUFMANN 159 been criticized from within, and therefore in a way that could be more easily accepted by the logical empiricists. The original and valuable aspect of Kaufmann s critique is, however, that he does give a case study of how one can combine the results of Husserlian phenomenology with the method of analytic philosophy. His intention is not to make a historical point on the question of whether the two movements are compatible or not or what they have in common and where their disagreements are, but rather to develop a new methodological standpoint that makes it possible to address and resolve actual philosophical problems. With this strategy, Kaufmann s position is still today an important example not only of how to deal with competing positions in philosophy, but also of how to approach a field that could best be characterized as analytic phenomenology. NOTES * I would like to thank Johannes Brandl, Tommaso Piazza, and Christian Beyer for their helpful comments on an earlier draft of this paper. Peter Simons, Philosophy and Logic in Central Europe from Bolzano to Tarski. Dordrecht: Kluwer, 1992, p. 2. Michael Dummett, Origins of Analytical Philosophy. Cambridge: Harvard University Press, 1993, p. 26. Husserl was also strongly influenced by Bolzano and Lotze. In a conversation with Boyce Gibson he states that Frege s critique hit the nail on the head. (qtd. in Dagfinn Føllesdal, Response by Dagfin Føllesdal, in: H. Dreyfus and H. Hall (Eds.), Husserl, Intentionality. and Cognitive Science. Cambridge: M.I.T. Press, 1982, 52-56, p. 55). Cf. Dagfinn Føllesdal, Husserl s Notion of Noema In: Journal of Philosophy 66, 1969, For a discussion of the reception of Husserlian phenomenology among British philosophers, cf. Mathieu Marion Les Recherches Logiques et leréalisme britannique (2001, forthcoming). Cf. Edmund Husserl, Logische Untersuchungen. Husserliana XIX/2. Dordrecht, Kluwer, 1984, p.535f. Husserl complains about Schlick s superficiality and lack of precision, but misquotes the German title of Schlick s book Allgemeine Erkenntnislehre: Wie bequem es sich manche Autoren mit wegwerfenden Kritiken machen, mit welcher Gewissenhaftigkeit sie lesen, welchen Unsinn sie mir und der Phänomenologie zuzumuten die Kühnheit haben, das zeigt die Allgemeine Erkenntnistheorie [sic!] von Moritz Schlick. For the relationship between Schlick and Husserl, cf. M.M. Van de Pitte, Schlick s Critique of Phenomenological Propositions, in: Philosophy and Phenomenological Research 45, 1984, ; Jim Shelton, Schlick and Husserl on the Foundations of Phenomenology, in: Philosophy and Phenomenological Research 48, 1988, ; or Roger Schmit, Moritz Schlick und Edmund Husserl. Zur Phänomenologiekritik in der frühen Philosophie Schlicks, in: Grazer Philosophische Studien 58/59, 2000, , who argues that Schlick developed his own position by contrasting it with Husserl s. Cf. Rudolf Bernet, Iso Kern, and Eduard Marbach, An Introduction to Husserlian Phenomenology. Evanston: Northwestern University Press, 1993, p.221 and Karl Schuhmann, Husserl Chronik: Denk- und Lebensweg Edmund Husserls. Den Haag, Nijhoff, 1977, p Cf. Guido Küng, The Phenomenological Reduction as Epoche and as Explication, in: The Monist 52, 63-80; Verena Mayer, Die Konstruktion der Erfahrungswelt: Carnap und Husserl In: Erkenntnis 35, ; Tommaso Piazza, Fenomenologianell Aufbau? Carnap, Husserl e la costituzione del mondo, to appear in: Lanfredini, Roberta (Ed.) Forma e contenuto. Milano:

170 160 WOLFGANG HUEMER LED, 2002; or Wolfgang Huemer, Husserl and Haugeland on Constitution, in: Synthese (forthcoming). Verena Mayer conjectures that Carnap s book might be a revised version of a draft that could have had more references to Husserl in a first draft, which, however, might have been deleted under the influence of Schlick. (Cf. Meyer, op. cit., 301, fn. 11). Cf. Friedrich Stadler, Studien zum Wiener Kreis, Frankfurt: Suhrkamp, 1997, esp. p.712ff as well as chapters 6 and 7. These letters are reprinted in: Edmund Husserl, Briefwechsel. Band IV: Die Freiburger Schüler. K. und E.Schuhmann (Ed.). Dordrecht: Kluwer, Gustav Bergmann, Memories of the Vienna Circle. Letter to Otto Neurath (1938), in: Friedrich Stadler (Ed.), Scientific Philosophy: Origins and Development. Vienna Circle Institute Yearbook 1. Dordrecht: Kluwer, 1993, , p Ibid. Ibid. Ibid. Cf. Felix Kaufmann, Phenomenology and Logical Empiricism, in: Marvin Farber (Ed.), Philosophical Essays in Memory of Edmund Husserl. Cambridge: Harvard University Press, 1940, and his Strata of Experience, in: Philosophy and Phenomenological Research 1, 1941, pp He does, however, express the assumption that psychologism might only be the result of imprecise formulations:...they must surely have had some inkling of the fact that a concept is not a reality of consciousness but, as it were, an unreal fiction. (Moritz Schlick, General Theory of Knowledge. Transl. by Albert Blumberg. Wien/New York: Springer, 1974, p. 134) Cf. Kaufmann Phenomenology and Logical Empiricism, op. cit., p Edmund Husserl, Entwurf einer Vorrede zu den Logischen Untersuchungen, in: Tijdschrift voor filosofie 1, 1939, and , p My translation. Dieser Vorwurf ist völlig unberechtigt, er steht mit dem Inhalt meiner Darstellungen in schärfstem Widerspruch und beruht auf der Übermacht eben der historischen Vorurteile, von denen ich mich einst mühsam losringen musste. Husserl, Entwurf einer Vorrede, op. cit, p My translation. Seine originellen Gedanken über Vorstellungen, Sätze, Wahrheiten an sich missdeutete ich aber als metaphysische Absurditäten. Moritz Schlick, General Theory of Knowledge, op. cit., p Cf. Moritz Schlick, General Theory of Knowledge, op. cit., p. 141: How do we know anything about an ideal self-evidence or about its possibility? Its existence must make itself known realiter in some way in our consciousness, through a feeling of self-evidence or some other phase of mental reality. And then all the earlier objections are revived, and everything remains as it was before: the problem pursues us no matter how often we seek to elude it by some twist or turn. Edmund Husserl, Formal and Transcendental Logic. Husserliana XII. Dordrecht, Kluwer, 1970, p. 166; quoted in: Felix Kaufmann, Phenomenology and Logical Empiricism, op. cit., 136. Felix Kaufmann, Phenomenology and Logical Empiricism, op. cit., p Ibid., p Ibid. Ibid. Ibid, p.131. Ibid. Felix Kaufmann, Strata of Experience, op. cit., p Felix Kaufmann, Phenomenology and Logical Empiricism, op. cit., p Felix Kaufmann, Strata of Experience, op. cit., p Felix Kaufmann, Phenomenology and Logical Empiricism, op. cit., p Ibid, p Moritz Schlick, Is There a Factual a Priori? in: Herbert Feigl and Wilfrid Sellars (Eds.): Readings in Philosophical Analysis. New York : Appleton, 1949,

171 LOGICAL EMPIRICISM AND PHENOMENOLOGY: FELIX KAUFMANN Cf. Wilfrid Sellars, Empiricism and the Philosophy of Mind, in: Herbert Feigl and Michael Scriven (Eds.) Minnesota Studies in the Philosophy of Science, vol. 1. Minneapolis: University of Minnesota Press, 1956, pp Department of Philosophy Universität Erfurt Germany

172 ARTUR KOTERSKI BÉLA VON JUHOS AND THE CONCEPT OF KONSTATIERUNGEN Béla von Juhos took part in the famous protocol sentences debate. He tried to defend and develop Schlick s conception of Konstatierungen. Juhos became one of the staunchest critics of physicalism, arguing against Carnap, Neurath and Hempel. Like Schlick, he thought the so-called left wing of the Circle had abandoned empiricism when they rejected comparison of facts and propositions as Verdopplung. On an another philosophical front he fought against Ayer who, similarly to Juhos, claimed that one can compare propositions with facts, but disagreed with Schlick s and Juhos idea of the incorrigibility of Konstatierungen. Juhos tried to show that the very grammar of those K-statements makes them certain and incorrigible. Here, Juhos ideas are very similar to those of Schlick; because of that he is often ignored as a non-interesting case of another uncritical follower. This is not true, and this paper tries to trace the differences between Schlick s and Juhos conceptions of Konstatierungen. 1. THE ORIGIN OF KONSTATIERUNGEN Béla von Juhos was Moritz Schlick s student and disciple; his papers on the empirical basis problem are deeply embedded in Schlick s theory of Konstatierungen. The story of Konstatierungen starts with the first book Schlick wrote, Allgemeine Erkenntnislehre (1918). Though we will not find this concept itself there, we do find something else, namely fundamental judgments : [...] propositions in the system of judgments by virtue of which the system rests directly on real facts we may call fundamental judgments 1. Schlick divides fundamental judgments into two groups: [...] definitions, in the narrower sense, and historical judgments. This requires some additional explanation. Definitions seem to be deictive definitions. Historical judgments are descriptive sentences about directly perceived states of affairs. All the remaining propositions in science are hypothetical. Fifteen years later he argued that: 163 F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

173 164 ARTUR KOTERSKI What was originally meant by protocol sentences, as the name indicates, are those statements which express the facts with absolute simplicity, without any moulding, alteration or addition [...] 2. If protocols were to be really clear, he would finally have to leave out historical judgments. Why? Because, as soon as you write them down, you cannot separate them from the hypothetical part of knowledge. This is precisely the doctrine of Konstatierungen, even if in a nutshell. Accordingly, let s jump to his later paper to see it in detail. 2. SCHLICK S VERSION OF KONSTATIERUNGEN 2) 3) 4) Moritz Schlick s conception of Konstatierungen was a counter-proposal to the physicalistic conception of protocol or basic statements. He tried to show that Neurath s proposal was throwing empiricism abroad. The particular theses of Neurath s theory so strongly disliked by Schlick were the following: 1) All sentences are hypothetical. To be a genuine sentence is to be hypothetical. The truth of any p consists (if anything) in consistency with the accepted body of statements. The very word truth is a metaphysical anachronism. And if we use it at all, we only chose to do so for our convenience in speaking. Statements are compared only with statements. Anything else is just sinnlos. One should not talk about reality, as it is Verdopplung, i.e. doubling metaphysics. His reasons are following: 1) Even if all scientific statements are hypotheses, there must be a link between the scientific model and the world that is modeled. No such solid link no empiricism. 2) To cultivate science we need an unambiguous criterion of truth, not truth on decision. 3) With physicalism you cannot distinguish real science from fictitious systems. Being able to do so requires having the link that lets us compare sentences with reality. 4) It is natural to talk about the link with reality, and it is constituted by Konstatierungen. But what exactly is a Konstatierung? It is not easy to say. It is something of the form So-and-so here now plus more crucially the appropriate gesture; for example Blue patch here now (when, let us say, the speaker s finger points at that patch). This looks like an observational statement and from time to time Schlick used such a term. What is peculiar to Konstatierungen?

174 BÉLA VON JUHOS AND THE CONCEPT OF KONSTATIERUNGEN 165 We can again put it into four steps: 1) Konstatierungen (K for short) are observational sentences, outside of the system-language (i.e. scientific language), so they do not have to be hypothetical. And they are not. Thus we have a solid base: incorrigible observational statements. 2) As we have this kind of Archimedean point for science, we have the genuine criterion of truth: the sentence is true if it is by the medium of K in agreement with reality. We cannot decide for reality. 3) Again, agreement with reality shows us which self-consistent system is the real science. Thus Konstatierungen enable our actions in the world. 4) We know that K are true, because we directly compare them with reality, because we do verify them we make them true. As one can see, the most important feature of this type of observational sentences is their incorrigibility they are certain. This is quite an expensive feature, because you have to: a) put K outside of science: You have to, because to keep them certain you need to say that they are timeless. So, simply put, you cannot write them down, as during the process of writing they become hypothetical. b) Then you have sentences that cannot be uttered or written down especially since when you state them (this seems a bit ironic!) you have to add a gesture, and this surely cannot be written down. c) In 1933, you had to talk about truth and about comparing sentences with reality without a good theory to support you. Schlick did not solve these problems: a) How can there be any process in the physical world which is timeless? As Neurath all too often noticed, it is a metaphysical idea. b) How is K a synthetic statement when you cannot revise it? It is not empirical. c) How can K be a statement, when you cannot say it? Not only are they outside of science but they are something pre-linguistic. d) How do you compare a sentence with reality? e) If K are of any use for science, they should be intersubjectively testable. But how is intersubjective verification of K possible? Let s have a look at how Juhos deals with them. 3. JUHOS VERSION When you look for differences between Schlick s and Juhos conceptions, they are quite hard to find. What you see prima facie is that when Schlick talks about

175 166 ARTUR KOTERSKI Konstatierungen, Juhos talks about K-sentences. But, of course, that s too simple. Juhos wanted to defend the basic idea of Konstatierungen, but at the same time to avoid the unpleasant consequences we know from Schlick s case. The basic idea of K-Sätze is to repeat that there are some privileged sentences that have empirical content, but are not hypothetical. Are K-sentences genuine sentences? K-sentences are like I feel pain, I see blue here now. This is very reminiscent of Schlick s concept, but what is really important is that there is no gesture! Can we write down K, then? According to Juhos, yes. Then it cannot be timeless. And it isn t. But how can he dismiss the argument that, when you say or write anything, you may forget what was at the very beginning or that, in the blink of an eye, the letters change their positions, so that in the end you say or write something completely different? Alas, there is no really good answer to be found in Juhos papers. He would probably have said that when you forget what was at the beginning, it is an error in action, not the kind you make when you (do not) understand a K-sentence. When you understand K, you truly know whether it is true or false. Note that K may be false, which was not the case with Schlick. So when I say I see a red patch now while I do not see it, I know that it is false. Cases where somebody says as a slip of the tongue it is red instead of it is green he described as trivial. As errors in action. (By the way: if it is not an error in action, i.e. when looking at the white wall you say I see a red wall, you have a false K.) This is another difference with Schlick. He would not say it is an error. For him, that would be another true K! So we can say that Konstatierungen are genuine sentences, that they are not timeless and that they can be written down. And they can be true or false, like any other real statement. But if they are genuine sentences, can we use them in science? Are K-sentences scientific? Juhos Konstatierungen are real, though, for sure, not physicalistic statements, and cannot be translated into thing-language, even if it is true that one can deduce from the system-language a sentence that looks identical to the given K. But they are not the same, as the method of obtaining them is different: in the case of K it is direct in the case of the K-mirror it is indirect. So Juhos says exactly what his teacher taught him: K are outside of the system. Are they useless to science? Neurath would say that, because K are not intersubjectively controllable, they are of no use at all. Not only can Robinson not

176 BÉLA VON JUHOS AND THE CONCEPT OF KONSTATIERUNGEN 167 understand Friday, but he is not able to understand his own diary. To put it more strongly: K is verifiable only for its owner (and only for an infinitely short moment). If K are not verifiable for others, then these others cannot understand them. K are not only useless, but also have no meaning. Juhos answer was and again it is different from Schlick s standpoint that K are unverifiable only in a technical or practical way. According to him, it is in principle possible to verify K. So they cannot be meaningless, as it is the logical and not the technical possibility of verification that decides if a sentences has meaning. Here is an example given by Juhos: There are two persons, S and B (a blind man). Let say that B perceives the sensation of blue any time S sees something blue. So B is able to test K of S in a direct, i.e. non-behaviouristic way. To answer the question on the scientific status of K we may state as follows. They are not in the system language. But physicalism is false, and the system language is not the only one. There are also other types of sentences, and to properly describe some types of phenomena you have to use them. So they are indispensable to science. Schlick also thinks so. But for the founder of the Vienna Circle, they have only a verifying function. For Juhos they describe in an unquestionable way the states of affairs that cannot be described in physical language. Are K really certain? If K describe states of affairs, if they are not timeless, are they really certain? How do we know that K are certain how do we verify them? Schlick wrote that there are two procedures involved in the verification of a hypothesis. First, you have to understand it, i.e. to know its observational consequences or the method of its verification. Then you apply this method to check the truth value of your hypothesis. You can know the former and not know the later. But in the case of K these two processes coincide, so you have only one: when you understand K, you already know that it is true. And there cannot be any doubt about it. According to Schlick, it is meaningless to say: I am not sure if I feel pain now or It is quite probable that I feel pain now. Juhos accepted this view but he put more stress on K s grammar; it is the grammar of Konstatierungen that secures their certainty. When we verify any hypothesis we have to check other sentences. And the truth value of this hypothesis depends on the truth value of those sentences. So, any hypothesis is a truth function of other sentences. When we know the truth value of those sentences, we know whether the hypothesis is true. If K is not hypothetical, then it cannot be a truth function of other sentences. But then there cannot be sentences such that knowledge of their truth value teaches us about the truth value of K. This also means that any two K cannot contradict each other.

177 168 ARTUR KOTERSKI But what happens if we negate K? According to the logic we use every day, when we know that p is true, we also know that not-p is false and vice versa. So not-k is a truth function and is hypothetical. It seems that the negation of K was a problem for Juhos theory. First, he wrote that negated K is not K anymore. Then, when he tried to show the privileged status of K, he wrote that they are not negated in the way the hypotheses are negated. According to him, when you make a conjunction of a hypothetical sentence and its negation, you obtain something called a complete contradiction. The only way to solve it is to cancel one of them in a purely arbitrary way. But when you have a conjunction of K and not-k you have a contradiction that is incomplete. Why is it incomplete? To know that not-k (e.g. This is not green ) is true you need some additional sentence (like This is blue ). So the negation of K is not a truth function of only one variable. Thus when you know only that K is true, you do not know that not-p is false at the same time. After a short discussion with Ayer he abandoned both theses. Juhos agreed that K can be negated like any other sentence and he stated that the negated K is still K. What he defended until the very end was the incorrigibility of K. Are K really incorrigible? This question may look redundant, as we have already talked about K being certain. But remember Neurath s example of Kalon. Kalon is a gifted scientist and he is able to write with two hands at the same time. So with his left hand he writes that Kalon states that there is a table in the room and Kalon can see it, and with his right one that Kalon states that the room is empty. If they are incorrigible we have to accept both of them, which is not too comfortable. First, Juhos played a trick with negation. Negated K is not a function of only one variable. So to cancel the contradiction, we have to fill the gaps. There will be no contradiction if we say that with his left hand Kalon described what he saw in a room A, and with his right hand what he saw in the other one, room B. Even if he had not dropped this idea of intuitionistic negation, he would have had serious problems, as Kalon can make his statements precise. (Nota bene, he could have just said that one of Kalon s K is true while the second is false but it seems he did not notice this possibility.) So, finally, can we can cancel K? No. To defend this expensive feature Juhos explicitly wrote that you cannot correct your K, because it is not the outcome of your mistake, so there is nothing to correct. We make corrections only when we make mistakes. 3

178 BÉLA VON JUHOS AND THE CONCEPT OF KONSTATIERUNGEN CONCLUSION To avoid pitfalls, Juhos rejected the idea that some processes are timeless, so the act of obtaining K must take time; it s clear then that one can utter or write down any K they are not designed as pre-linguistic entities. Because they have many features of ordinary hypotheses they are intersubjectively testable (i.e. verifiable), they are accepted on empirical grounds (and let us assume that Juhos explained it appropriately) there is still one question pending: how it is possible that one cannot revise them? Juhos thought that they cannot be subject to correction, and this thesis is the last (though the most important) link with Schlick s doctrine. In the end he repeats what we already know from his mentor: Konstatierungen do not appear as the effect of mistakes, so there is nothing to correct in them. They are incorrigible because of their nature. Then we have again the old question about the possibility of such sentences. Having thus gone in a big circle just to avoid pitfalls Juhos came back to the point of departure. That is why we may say that it seems Juhos was not successful in rescuing Konstatierungen as descriptions of reality that are absolutely certain. NOTES M. Schlick, The General Theory of Knowledge, Library of Exact Philosophy, Vol. 11, Wien New York: Springer-Verlag, 1974, p. 78. M. Schlick, Über das Fundament der Erkenntnis, Erkenntnis, Bd. IV (1934), p. 79. Carnap was very disappointed when he read Juhos manuscript of Empirische Sätze und Logische Konstanten (later published in The Journal of Unified Science (Erkenntnis), Vol. VIII (1939/1940), pp ). In a letter to Neurath he expressed his highly negative opinion of Juhos work: Juhos makes [...] formal mistakes; after I pointed to errors in an earlier MS and earlier conception, I regret now I was his teacher (Carnap to Neurath, [ASP(RC) ]). Quoted by permission of the University of Pittsburgh. All Rights reserved. Department of Logic and Methodology Faculty of Philosophy and Sociology Maria Curie-Sklodowska University Pl. MCS 4, Lublin Poland Artur.Koterski@umcs.lublin.pl

179 PAOLO MANCOSU MATHIEU MARION WITTGENSTEIN S CONSTRUCTIVIZATION OF EULER S PROOF OF THE INFINITY OF PRIMES Es war eben auch hier die Hand eines Schleifers notwendig, um den Glanz der Edelsteine Eulers voll herauszuarbeiten. L. Kronecker 1. INTRODUCTION Ever since Georg Kreisel ended his review of Wittgenstein s Remarks on the Foundations of Mathematics (1956) by saying that it seems to me to be a surprisingly insignificant product of a sparkling mind (1959, 158) there has been a strong presumption that Wittgenstein did not know much about mathematics in general and about issues pertaining to the foundations of mathematics in particular, and that his remarks on these topics were marred by this lack of proficiency in mathematical matters. One may indeed ask how much mathematics did Wittgenstein know or, in other words, how much mathematics was he able to do? For a long time, it has been generally believed that Wittgenstein held a peculiar foundational stance, strict finitism. This is certainly not the case today. Indeed, one would nowadays tend to approve of Gordon Baker and Peter Hacker, when they write that his philosophy of mathematics does not defend a form of strict finitism, depsychologized intuitionism or constructivism and his purpose was not to take sides in the debates between rival schools of mathematicians, but rather to question the presuppositions which provided the framework of their debates (1985, 345). We will discuss a mathematical proof found in Wittgenstein s Nachlaß, a constructive version of Euler s proof of the infinity of prime numbers. Although it does not amount to much, this proof allows us to see that Wittgenstein had at least some mathematical skills. At the very least, the proof shows that Wittgenstein was concerned with mathematical practice and it also gives further evidence in support of the claim that, after all, he held a constructivist stance, at least during the transitional period of his thought ( ). 171 F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

180 172 PAOLO MANCOSU / MATHIEU MARION 2. BEHMANN, KAUFMANN, AND WITTGENSTEIN On October 24, 1930, Heinrich Behmann, a student of Hilbert at Göttingen then teaching mathematics at the University of Wittenberg-Halle, 1 submitted a paper to Mathematische Annalen entitled Zur Frage der Konstruktivität von Beweisen (1930). Behmann visited Vienna from the 9th to the 20th of September, 1930, to deliver a talk for Schlick s circle. During his stay he learned about a conjecture by the Viennese philosopher, Felix Kaufmann, according to which, roughly, all existence theorems of classical mathematics which do not rely on non-constructive existential axioms, such as the axiom of choice, are in effect constructive existence theorems. In other words, it was conjectured by Kaufmann that one should be able to show that all such existence theorems rely on an implicit exhibition of an instance. Kaufmann had made the conjecture in a book that had appeared in the spring of 1930, Das Unendliche in der Mathematik und seine Ausschaltung (1930, 66-7; 1978, 57-8). 2 In his paper, Behmann proposed a proof of this conjecture. Criticisms from Paul Bernays, who was at the time an assistant-editor for the Annalen, and from Kurt Gödel forced him to withdraw his paper, which remained unpublished. In its first section, entitled A Conjecture by Kaufmann, Behmann presents three constructivist restrictions on methods of proofs (and, consequently, on admissible theorems): (1) Elimination ofnon-constructive existential axioms (in particular the axiom of choice) (2) Elimination of non-predicative existential proofs (in so far as they are not already excluded by (1)) (3) Elimination of concepts of higher type (that is, restriction to the so-called restricted functional calculus ) 3 (The use of the expression non-predicative in (2) was an oversight on Behmann s part; it should read non-constructive.) There is no point in discussing here restriction (3), as it does not bear on the topic of this paper. Behmann clearly was of the opinion that restriction (1) is a genuine one, i.e., there are indeed theorems of classical mathematics that cannot be proved without use of the axiom of choice. Now, restriction (2) would also be a genuine restriction if there were indirect existential proofs (i.e., proofs by contradiction or reductio ad absurdum) that cannot be turned into direct ones or if there were proofs of existential claims which, although not relying on non-constructive existential axioms, can only be proved by contradiction. Kaufmann s conjecture is to the effect that this is not the case, in other words that all indirect existential proofs satisfying (1) are also constructive. As Behmann put it in his paper: Mr. Kaufmann has recently formulated an interesting conjecture [...] He is of the opinion that an indirect proof, in so far as it satisfies condition (1), is also in essence constructive, more precisely, that it relies on a hidden instance, which through an appropriate analysis of the demonstration can automatically be made explicit. 4

181 W ITTGENSTEIN S CONSTRUCTIVIZATION 173 The truth of Kaufmann s conjecture would have implied that existential proofs of classical mathematics that do not rely on non-constructive existential axioms, would implicitly satisfy the requirement of constructivity. This, in turn, would imply that the requirement of constructivity would not entail a restriction of current mathematics. Although Kaufmann shared Brouwer s rejection of (non-denumerable parts of) set theory, he believed that classical mathematics was already constructive and therefore that the elimination of the non-denumerable infinite would leave mathematics as it is, so to speak. Gödel himself provided within a few days a counterexample that refuted Behmann s claim that he had proved Kaufmann s conjecture. 5 (However, Gödel s objection to Behmann s proof does not use his incompleteness theorems.) Nevertheless, Behmann s unpublished typescript is interesting for a number of reasons. In it, he presented a general method by means of which one could turn, allegedly, a non-constructive existence proof into a constructive one. The techniques used for the proof involve one of the earliest uses of graph-theoretical and topological considerations in the study of proofs. 6 Behmann s method can also be seen as related to proof-theoretical methods developed later by Hilbert & Bernays, Herbrand, and Kreisel. Of peculiar interest to us here, however, is the fact that Behmann presented, in the last section of his paper, an application of his method to the constructivization of a well-known existence proof presented by Leonhard Euler in his Introductio in analysin infinitorum (1748, I, 235). There exist many proofs of the infinity of prime numbers, 7 starting with Euclid s famous proof. 8 Euler proved the theorem by reductio ad absurdum relying on the equation: where ranges over primes. 9 The left-hand side is the harmonic series. Euler had proved that it diverges, i.e., one can say that the sum is infinite. Euler s proof works as follows: assume..., are all the prime numbers. From the equation one obtains Thus, is finite, which is a contradiction. 10 However, as Leopold Kronecker pointed out in his Vorlesungen über Zahlentheorie (1901, 270), equation (A) is actually invalid; he rectified Euler s proof

182 174 PAOLO MANCOSU / MATHIEU MARION using: for z > 1. If there were only a finite number of primes the product would remain finite as z approaches 1, while the sum increases indefinitely. Kronecker also showed how to define an interval from to in the left-hand sum, of which it is shown that it contains at least one prime, however great is taken (1901, ), thus providing a constructive addition to Euler s proof. In this connection he made a very important remark: This proof, as Euclid s proof, can also be cast in a form which yields an interval that contains a new prime number no matter how big is. Only then is the last and highest requirement satisfied which has to be imposed on a rigorous mathematical proof. 11 Euler s proof is of great importance for results about the infinity of primes in general arithmetical progressions, e.g., for Dirichlet s theorem for with and relatively prime. 12 The constructivization of Euler s proof is thus far from being a mere exercise for idle minds. At all events, what is of specific interest to us, however, is the fact that, in a footnote right at the beginning of the section of his paper dealing with the constructivization of Euler s proof, Behmann wrote: I simultaneously learned about a transformation [Überführung] of this proof into a constructive form by Mr. Wittgenstein and about Kaufmann s conjecture. 13 The attribution of a proof to Wittgenstein is quite an astounding claim and we should like to look closely into it. As it turns out, even this application of Behmann s method to Euler s proof is not satisfactory, but it remains of interest to us because of the light it sheds on Wittgenstein s thinking about mathematics. On October 6, before submitting his paper, Behmann had therefore written to Kaufmann: Since I have no memory of the details of Wittgenstein s completion of Euler s proof that Mr. Waismann had presented, I would very much like to find out to which degree my transformation [Umwandlung] of the proof, which I worked out independently on the basis of the general scheme that I developed, corresponds to Wittgenstein s. At any rate, it would be necessary to reach an agreement with Mr. Wittgenstein, when I shall have to mention this point, which, for reasons of fairness, can hardly be ignored; it seems best to do it through the mediation of Mr. Waismann. 14 Answering Behmann s request, Kaufmann wrote reassuringly on the 12th that: Concerning the proof that you have given as an example of your general theorem and that of Wittgenstein, Mr. Waismann considers that they are indeed related but that your footnote appropriately deals with the relationship between them, all the more so since this result is of fundamental importance to your general considerations. He does not consider himself, however, to be in a position to reach a final decision on this point and will, in accordance to your request, forward your proof to Wittgenstein and inform him about its circumstances. At all events, there is no reason to delay publication. Perhaps I should ask you to modify your

183 WlTTGENSTEIN S CONSTRUCTIVIZATION 175 note to the effect that you had learned about Wittgenstein s transformation [Überführung] into a constructive form of Euler s proof about prime numbers prior to devising your own constructivization of the proof, while not mentioning in this context that you had previously learned about my conjecture, since the latter is already in my book. 15 As a result of Kaufmann s request, Behmann reworded the footnote as follows in the final version of his paper: The fact that I recently learned about a transformation into a constructive form of this proof by Mr. Wittgenstein motivated me to devise the general procedure that I described above. Kronecker had already given in his Vorlesungen über Zahlentheorie a constructive completion (not, however, a transformation [Umformung]) of Euler s proof. 16 One may infer from these quotations that Behmann learned about Kaufmann s conjecture and Wittgenstein s proof during his short stay in Vienna and that he learned about the latter talking to Waismann. It does not seem that Wittgenstein actually took part in any discussions involving Behmann, Kaufmann and Waismann; the latter seems to have acted as a go-between. It is also certainly clearly implied that Wittgenstein had his own constructive version of Euler s proof, which closely resembled Behmann s. This evidence prompted a search for Wittgenstein s proof. The section on Euler s Proof which is printed in Philosophical Grammar (1974, ) does not contain a proof but it ends with some calculations that are hard to interpret at first blush. The book Philosophical Grammar was edited by one of Wittgenstein s literary executors, Rush Rhees, who used as a basis the so-called Big Typescript, which is a voluminous typescript of 768 pages produced by Wittgenstein in Anthony Kenny (1984) has shown the extent of Rhees editorial intervention but a comparison of the section on Euler s proof with the corresponding section in the recently published Big Typescript (2000, 427-9) shows that they are virtually similar. However, most of the remarks that form the Big Typescript originated in manuscripts for the years that are by now published as the first five volumes of the Wiener Ausgabe. It turns out that the remarks that form the section on Euler s proof in the Big Typescript and the Philosophical Grammar originated in one of these manuscripts, namely MS 108, which is entitled IV. Philosophische Betrachtungen, 17 and can be dated from the summer and autumn of 1930 therefore around the time of Behmann s visit to Vienna. 18 In this manuscript, now published as part of volume 2 of the Wiener Ausgabe (1994), 19 the strange calculations that end the section on Euler s proof are continued and one can actually see the point of these, as it is clear that Wittgenstein is calculating a bound below which one can find a new prime number greater than all the elements of any given set of primes,..., thus providing a constructivization of Euler s proof. It appears that both Wittgenstein and Behmann used as a point of departure Euler s original proof, based on the invalid equation (A) and not the corrected version provided by Kronecker, although Behmann learned about it at one stage, 20 since he mentions it in the footnote that we already quoted. It is unlikely that either of them used Euler s original version but it is impossible to tell from which

184 176 PAOLO MANCOSU / MATHIEU MARION presentation of the proof they worked. We shall now look at both proofs. 3. THE TRANSFORMATION OF EULER S PROOF IN BEHMANN Behmann starts from Euler s invalid equation: Assuming that (B) the number of primes is finite, say,...,, one obtains that This implies (D) that is finite, which is a contradiction, since Euler has already shown that the harmonic series diverges. One should notice that the last equality of (C) is justified by the unique factorization theorem. That is every must be obtained from the previous product. By uniqueness each will only appear once. The above proof is non-constructive since it does not tell us how to find the next prime beyond Behmann s project was to show that, by a transformation of the proof into a direct one, one could extract a bound for finding another prime number not contained in,...,. Although we will not present Behmann s general strategy in detail, the following outline should suffice to get the gist of the transformation. According to Behmann s own conventions the proof could be schematised thus: One can, for ease, read the arrows as indicating implications. According to Behmann s strategy we obtain a direct proof by using the negation of D and reversing the arrows of implication from D to C and also negating C and reversing the arrow of implication from C to B. In short the direct proof would look as follows:

185 WITTGENSTEIN S CONSTRUCTIVIZATION 177 One should note here that states that there are infinitely many primes. Thus, the direct proof of the result of the infinitude of prime numbers would proceed in three steps: from the fact that the harmonic series diverges and equation (A) one would obtain And, therefore, there exist infinitely many prime numbers However, this single transformation does still not give us the required instance for the existential. For this reason, Behmann claims that this information must be hidden in the statement that diverges. Indeed, the statement can be re-expressed as saying that This, says Behmann can be shown easily by rewriting the harmonic series as and noticing that each expression in parenthesis is > The fact that every real number can be majorized through the addition of sufficiently many sums of which the first is 1, the second and all the others greater than, can easily be shown by choosing the number of summands to be This will certainly suffice for (Here, gives the greatest natural number smaller or equal to.) We thus need to sum up members of the transformed harmonic series to go beyond or, what amounts to the same thing, elements of the original harmonic series, (i.e. choose 21 In this way one obtains a general proof of In turn, by substituting the product (denoted by for one obtains The proof is concluded by noticing that the right-hand side

186 178 PAOLO MANCOSU / MATHIEU MARION of the inequality with finitely many summands must contain at least one summand that cannot be obtained by multiplying out the factors on the left hand side, i.e., a summand whose denominator must be divisible through a prime greater than if one decomposes into prime factors all numbers between and Two things should be pointed out about the above transformation. First, although Behmann wants to use this theorem as an example of how the general strategy for turning indirect existential proofs into direct proofs is supposed to work, it should be remarked that the constructivization of Euler s proof does not follow from syntactic manipulations but rather by refining one of the initial assumptions. However, in general we have no guarantee that for every assumption of the form we would be able to provide such refinement. On the contrary, we know that in general this is not possible. Secondly, unlike Wittgenstein s approach to the problem, Behmann seems to be already knowledgeable about which bounds are going to work for the situation at hand. Indeed, later on he points out that he could have transformed the proof by using the better inequality which gives a bound for the next prime under where is the natural logarithm. 4. BACK TO WITTGENSTEIN The section on Euler s proof in Philosophical Grammar begins with a straightforward reflection on Euler s proof: From the inequality can we derive a number which is still missing from the combinations on the right-hand side? Euler s proof that there are infinitely many prime numbers is meant to be an existence proof, but how is such a proof possible without a construction? The argument goes like this: The product on the right is a series of fractions in whose denominators all multiples of the form occur; if there were no numbers besides these, then this series would necessarily be the same as the series and in that case the sums also would necessarily be the same. But the left-hand side is and the right-hand side only a finite number so there are infinitely many fractions missing in the right-hand series, that is, there are on the left-hand side fractions that do not occur on the right. (1974, 383; 1994, 321-2; 2000, 427) Again, it is clear that Wittgenstein discusses Euler s equation (A). As he points out a bit further, if one extends the sum on the left-hand side until it is greater than any

187 WITTGENSTEIN S CONSTRUCTIVIZATION 179 number which would be the result of the right hand product for prime numbers up to any given one, this part of the sum must contain a term that doesn t occur in the right hand series, for if the right hand series contained all those terms it would yield a larger and not a smaller sum (1974, 385; 1994, 324; 2000, 429). As was pointed out, there is a passage in Wittgenstein s manuscripts for the year 1930 where we find the following computations, which were neither reproduced in the Big Typescript, nor in Philosophical Grammar. This passage should be read in conjunction with the calculations at the end of the section on Euler s proof in both of these: Thus, a sufficient condition for is that If I represent the beginning of the series of the ordered segments that are then the first of these segments goes from the second from the third from the 1 to 3 4 to to 63 to The sum of the series developed up to thus surely exceeds Thus, we have It is thus the case that in the first whole numbers, there must be one which is not divisible by any of the first numbers. (1994, 325) We will reproduce Wittgenstein s calculations in the appendix and limit ourselves here to a synthesis of the results. First of all, Wittgenstein solves the following general problem: if one starts from for arbitrary how far does one have to go for the harmonic series to sum up to a number i.e., to determine such that He works out a bound through algebraic manipulations (see appendix) whereby the condition is And since he thus concludes that from any point of the harmonic series one only needs to go to in order for that segment of the harmonic series to sum up to a number greater than 1. Applying the result to the harmonic series from its start (i.e., ) Wittgenstein determines that This is not as good a bound as the one used by Behmann although it is closely related since Wittgenstein s evaluation, requires always two times as many steps as Behmann s. This is due to the fact that Wittgenstein s condition on gives him a very slow start when trying to majorize any number We have to assume that the full passage from MS 108 gives us a good idea of the proof by Wittgenstein that was referred to by Behmann, Kaufmann and Waismann in their exchange of letters. Therefore, Waismann s claim (quoted above in Kaufmann s letter dated 12.X.1930) that the two proofs are related seems to be justified.

188 180 PAOLO MANCOSU / MATHIEU MARION 5. WITTGENSTEIN ON MATHEMATICAL EXISTENCE Wittgenstein s critical discussion of Euler s original proof and his constructive version of it are, if not of much mathematical interest, of great significance to the interpretation of his philosophy: this is the sole instance of an extended discussion of a specific existential proof and the sole instance of a proof in the whole of his writings and, as such, they cannot be bypassed. Wittgenstein s comments on Euler s proof are extremely critical: Here again we have that remarkable phenomenon that we might call proof by circumstantial evidence in mathematics something that is absolutely never permitted. It might also be called a proof by symptoms. The result of the summation is (or is regarded as) a symptom that there are terms on the left that are missing on the right. The connection between the symptom and what we would like to have proved is a loose connection. That is, no bridge has been built, but we rest content on seeing the other bank. All the terms on the right-hand side occur on the left, but the sum on the left-hand side is and the sum of the right-hand side is only a finite number, so there must... but in mathematics nothing must be except what is. The bridge has to be built. (1974, 384; 1994, 322; 2000, 427) These critical comments amount to a straightforward rejection of the validity of existential proofs: he calls them proofs by circumstantial evidence and says that they are absolutely never permitted ; alternatively, he calls them proofs by symptoms but claims that there are no symptoms in mathematics. He uses the metaphor of the bridge and likens existential proofs to resting content with seeing the other bank, while he requests that the bridge has to be built. A further critical remark in the same section can help us to understand Wittgenstein s more general thoughts about existential proofs: We might also put the question thus: if you had only this proof, what would you bet on it? If we discovered the primes up to N, could we later go on for ever looking for a further prime number since the proof guarantees that we will find one? Surely that is nonsense. For if we only search long enough has no meaning. (That goes for existence proofs in general). (1974, 384; 1994, 323; 2000, 428) Not only is this reasoning perfectly in line with Wittgenstein s proof, which provides an interval within which one will find a new prime number, but it also squares perfectly well with the very reasoning of Brouwer and Weyl which is at the basis of their rejection of the Law of Excluded Middle. 22 Moreover, it bears an obvious relation to Kronecker s demand that an existential proof must be accompanied by a bound for the existential if it has to satisfy the demands of mathematical rigor. Wittgenstein s critical discussion of Euler s original proof and his constructive version of it are also extremely useful because they allow us to infuse meaning into his more general pronouncements about mathematics. (From an exegetical point of view, to ignore or to bypass these remarks and the proof would simply amount to plain dishonesty.) For example, one finds in the very passage of MS 108 that we

189 WITTGENSTEIN S CONSTRUCTIVIZATION 181 have been referring to, the following remark, which encapsulates Wittgenstein s central thought about foundations of mathematics: In mathematics everything is algorithm and nothing is meaning; even when it doesn t look like that because we seem to be using words to talk about mathematical things. Even these words are used to construct an algorithm. (1974, 468; 1994, 321; 2000, 494) A similar remark was uttered by Wittgenstein, during a meeting with Schlick and Waismann in June 1930: I believe that mathematics, once the conflict about its foundations has come to an end, will look just as it does in elementary school where the abacus is used. The way of doing mathematics in elementary school is absolutely strict and exact. It need not be improved upon in any way. Mathematics is always a machine, a calculus. The calculus does not describe anything [ ] A calculus is an abacus, a calculator, a calculating machine; it works by means of strokes, numerals, etc. (1979, 106) And remarks to the same effect made their way into the typescript of 1931 which was edited under the title Philosophical Remarks: Let s remember that in mathematics, the signs themselves do mathematics, they don t describe it. The mathematical signs are like the beads of an abacus. (1975, 157) we can t describe mathematics, we can only do it. (And that of itself abolishes every set theory.) (1975, 159) In mathematics everything is algorithm : it is obvious that the rejection of (nonconstructive) existential proofs is entailed by this very Kroneckerian thesis. Perhaps it is worth pointing out here another element contained in these remarks, i.e. the idea that there are no descriptions in mathematics. According to Wittgenstein it is a mistake to think that a foundational theory (set theory or Principia Mathematica) could be a description, hence a theory. As he would put it Calculation with letters is not a theory (1979, 136); what is calculus must be separated off from what attempts to be (and of course cannot be) theory (1974, 468; 2000, 494). This is the true ground of Wittgenstein s limited anti-theoretical and anti-foundational stance, a stance that has been entirely misunderstood by those who claim that he held no theses at all VIENNESE FINITISM What about Wittgenstein s reaction to Kaufmann s conjecture? If there is any discussion of it in Wittgenstein s Nachlaß, it remains to be found. A few remarks are nevertheless in order. To begin with, although it seems from the available evidence that Wittgenstein never met with Behmann, it is also clear that he was not kept out of the loop. A copy of the last version of Behmann s paper was sent to Wittgenstein, who must have known about Kaufmann s conjecture. There are at any rate traces left of contacts between Wittgenstein and Kaufmann. For example, in a letter to Kaufmann dated April 6, 1930, Wittgenstein wrote:

190 182 PAOLO MANCOSU / MATHIEU MARION Thank you very much for sending me your book. Sadly, I am too busy to be able to read it. I hear from Mr. Waismann that our views coincide on many points. 24 There is also the incredible story of a suitcase bearing the name Kaufmann that was apparently found in Wittgenstein s lodgings in Cambridge after his death and which Rush Rhees mistakenly sent to Walter Kaufmann, only later realising that the owner must have been Felix Kaufmann. Later, Rhees unsuccessfully tried to recover the suitcase; one will never know its contents. 25 It is most probable that Wittgenstein would have found Kaufmann s attempt at grounding arithmetic with help of Husserl s phenomenology uncongenial. However, Kaufmann s finitism is not very far from Wittgenstein s own Standpunkt, which also should be seen as a form of finitism. 26 It should be recalled here that, just prior to Behmann s visit to Vienna, Waismann had just participated in the famous meeting at Königsberg organized by the Wiener Kreis and Hans Reichenbach s Gesellschaft für empirische Philosophie, by giving a paper at a symposium on foundations of mathematics which included presentations by Rudolf Carnap on logicism, Arend Heyting on intuitionism, and Johann von Neumann on formalism. 27 Waismann s paper, entitled Über das Wesen der Mathematik. Der Standpunkt Wittgensteins (1982), was not ready to be included alongside the others in the 1931 issue of Erkenntnis devoted to the Königsberg meeting; it appeared only in It is clear from Wittgenstein s attitude prior to the meeting that he took the opportunity to have his views publicized very seriously. There was, therefore, such a thing as Wittgenstein s Standpunkt. Waismann s presentation of his ideas, although rather faithful, was not formal. As Carnap pointed out in the discussion that was to be published in the same issue of Erkenntnis, the Standpunkt Wittgensteins had not yet reached a stage of development at which it could be evaluated (Hahn et alii 1931, 141). It never reached that stage, either in Waismann s or in Wittgenstein s own work. We are now left with the task of reconstructing this Standpunkt Wittgensteins and the above discussion of his remarks on the constructivization of Euler s proof should be seen as a contribution to this task. That Kaufmann s finitism is not very far from Wittgenstein s own Standpunkt is supported by the above-quoted letter, where Wittgenstein writes: I hear from Mr. Waismann that our views coincide on many points. Therefore, something like Kaufmann s conjecture could not have been very far from Wittgenstein s mind. This would go a long way to explain his interest in constructivizing Euler s proof. But, before one might be tempted to speak of an indigenous, Viennese form of finitism, it should be pointed out that Kaufmann s conjecture implied that classical mathematics was already constructive and that Brouwer s critique was to leave it essentially untouched. Wittgenstein did not believe this. In his manuscripts for the years , we find claims to the contrary, e.g.: What will distinguish mathematicians of the future from those of today will really be a greater sensitivity, and that will as it were prune mathematics;... (1974, 381; 1996, 23; 2000, 426) It is clear that Wittgenstein had no illusions about the effects of philosophical

191 WITTGENSTEIN S CONSTRUCTIVIZATION 183 clarity on mathematics or about the consequences of his Standpunkt, parts of which was the rejection of pure existential proofs. 7. APPENDIX Wittgenstein s goal is to determine what bound is necessary for He begins by transforming the left-hand side into such that This, in turn, can be written as At this point, in order to work out a bound, Wittgenstein s divides the above sum into a first part up to and a second part. Moreover, since we have We now consider the expression in square brackets. The 1 s sum up to difference and the is such that we can bound every one of its terms by a multiple of Indeed. And in general,

192 184 PAOLO MANCOSU / MATHIEU MARION Thus, As for the second part of the sum Wittgenstein looks at the sum obtained from it by replacing each summand by Since there are many summands, this new sum amounts to Also, it is clearly less than the sum from which it is obtained. Now the question becomes: Knowing a bound on how big must we choose so that The left hand side is equal to Thus, iff iff iff iff iff

193 WITTGENSTEIN S CONSTRUCTIVIZATION 185 iff iff The final observation consists in noticing that majorizes for all Remark: in Wittgenstein (1994), p. 325, the symbol should be read as In Wittgenstein (1974), p. 386 and (2000), p. 429 some of the equality signs should be NOTES For details about Behmann s work in the foundations of mathematics, see Mancosu (1999). See Mancosu (2002), 1, for a discussion of the meaning of that passage. The paper also reconstructs the debate surrounding the discussion of Behmann s proof. 1. Ausschluß nicht-konstruktiver Existenzbeweise (insbesondere des Auswahlaxioms). 2. Ausschluß nicht-prädikativer Existenzbeweise (soweit nicht bereits durch (1) betroffen). 3. Ausschluß von Begriffen höherer Stufe (Beschränkung auf den sogennanten engeren Funktionenkalkül). Behmann(1930), p.l. Hierzu hat neuerdings Herr Kaufmann eine interessante Vermutung ausgesprochen [...] Er ist der Ansicht, daß auch ein indirekter Existenzbeweis, sofern er der Anforderung (1) genügt, seiner Natur nach konstruktiv sei, genauer gesagt, auf einer verstecken Aufweisung beruhe, die sich durch eine geeignete Analyse des Beweises zwangläufig zur Evidenz bringen lassen müsse. Behmann (1930), p.l. On Gödel s objection to Behmann s paper, see Mancosu (2002), 5. This aspect of Behmann s paper has been studied by Mancosu (2002). For a list of various proofs of the infinity of prime numbers, see Dickson (1952), pp Euclid s proof, a typical case of reductio ad absurdum, is as follows: suppose that there is a greatest prime number, One defines from the primes the number which must be either prime or composite. If it is prime, then we have a contradiction, since it would be bigger than all the prime numbers smaller or equal to If it is composite, it must be divisible exactly by a prime number. But this prime divisor cannot be any prime smaller or equal to number, bigger than Notice that this is a special form ofriemann s zeta function because they would all leave a remainder of 1. Therefore there must be another prime for real z appears in a work by Euler dated 1744, where he introduced where the s are prime numbers. The zeta function The previous equation involves the summation formula for the geometric series, i.e. for any prime we have Note that on account of the unique factorization theorem (which is independent of the infinitude of the prime numbers) every natural number appears only once as denominator in the expanded product. Moreover, the series obtained by multiplying absolutely convergent series is also absolutely convergent and its sum is the product of the sums in the series factors. Auch diesem Beweise kann man ebenso wie dem Euklidischen eine solche Form geben, dass sich aus ihm ein Intervall ergiebt, innerhalb dessen sicher eine neue Primzahl sich befindet, wie groß auch angenommen worden ist; erst dann ist ja auch der letzten und höchsten Anforderung genügt, die man an einen strengen mathematischen Beweis zu stellen hat. (1901, 270)

194 186 PAOLO MANCOSU / MATHIEU MARION As pointed out in Hasse (1950, 183), while proofs of specific cases rely on a generalization of Euclid s proof, Dirichlet s general result relies on Euler s proof. Kronecker had already pointed this out in his Vorlesungen (1901, 273). Eine Überführung dieses Beweises in die konstruktive Form durch Herrn Wittgenstein ist mir gleichzeitig mit der Kaufmann schen Vermutung bekannt geworden (Kaufmann archive, Konstanz, ). Da ich die Ausführungen Herrn Waismanns wegen der Wittgensteinschen Ergänzung des Eulerschen Beweises in den Einzelheiten nicht mehr in Erinnerung habe, würde es mich sehr interessieren, wie weit die von mir ziemlich unabhängig auf Grund des von mir entwickelten allgemeinen Schemas durchgeführte Beweisumwandlung mit der Wittgensteinschen übereinstimmt. Jedenfalls wäre es auch wohl notwendig, sich mit Herrn Wittgenstein wegen der Erwähnung dieses Punktes, die aus Billigkeitsgründen nicht gut zu umgehen ist, zu verständigen, was wohl am besten durch Herrn Waismann geschehen kann (Kaufmann archive, Konstanz, ). Was den Zusammenhang Ihres als Beispiel für den allgemeinen Satz gegebenen Beweises mit demjenigen von Wittgenstein betrifft, so meint Herr Waismann, daß wohl eine Verwandtschaft bestehe, daß aber Ihre Anmerkung diesem Sachverhalte hinreichend Rechnung trage, umsomehr als das Ergebnis von zentraler Wichtigkeit in Ihren allgemeinen Ueberlegungen liege. Er maßt sich aber eine endgültige Entscheidung hierüber nicht an und wird daher, ganz in Einklang mit Ihrem Wunsche, Wittgenstein Ihren Beweis schicken und ihn über die Begleitumstände informieren. Fur keinen Fall liegt irgendein Grund zur Verzögerung der Publikation vor. Vielleicht darf ich Sie noch bitten, die in Frage kommende Anmerkung in der Weise zu ändern, daß Sie schreiben, Sie hätten die Ueberführung des Euler schen Primzahlbeweises in die konstruktive Form durch Wittgenstein vor Ihrer Ausarbeitung dieses Beweises erfahren, ohne in diesem Zusammenhange die Kenntnisnahme von meiner Vermutung zu erwähnen, da ja diese schon in meinem Buche (S. 66 f.) enthalten ist (Kaufmann archive, Konstanz, ). Eine mir kürzlich bekannt gewordene Überführung dieses Beweises in die konstruktive Form durch Herrn Wittgenstein hat die Aufstellung meines obigen allgemeinen Verfahrens mit veranlaßt. Eine konstruktive Vervollständigung (nicht Umformung) des Eulerschen Beweises is bereits von Kronecker (Vorlesungen über Zahlentheorie, S. 270ff.) angegeben worden Behmann (1930), p.6, note 1. For details about Wittgenstein s Nachlaß, see von Wright (1993). The notes of conversations with Wittgenstein and Schlick taken by Waismann also contain a reference to Euler s proof during the meeting of September 25, 1930: Euler s proof is immediately in error, as soon as prime numbers are written down in the form For if the next index is to mean an arbitrary number, then this already presupposes a law of progression, and this law can be given only in terms of an induction. Thus the proof presupposes what it is supposed to prove (1979, ). This remark is rather unhelpful but at least it indicates that Euler s proof was at the forefront of Wittgenstein s preoccupations in late September 1930, just a few days after Behmann had left Vienna. For the remarks on Euler s proof, see pp Behmann wrote to Kaufmann on October 11: Außerdem will ich u.a. noch ergänzend bemerken, daß eine Vervollständigung des Eulerschen Beweises im Sinne der Konstruktivität bereits von Kronecker (Vorlesungen Über Zahlentheorie, S ) angegeben worden ist. Doch betrachtet Kronecker dies durchaus als Zusatz zu dem gegebenen nicht-konstruktiven, nicht als Umwandlung in die konstruktive Form (Kaufmann archive, Konstanz, ). One should notice that if one does not use real but only natural numbers, this simply means that for any in it is enough to sum up elements of the harmonic series to go beyond For a discussion of the numerous passages where Wittgenstein says things that are in basic agreement with this reasoning, see Marion (1998), chaps. 4 & 6. On these matters, see Marion (1998), chap. 1. The full letter reads: Sehr geehrter Herr Kaufmann! Besten Dank für die Zusendung Ihres Buches. Ich bin leider zu sehr beschäftigt, es lesen zu können. Ich höre von Herrn Waismann, daß unsere Ansichten in mancher Hinsicht übereinstimmen. Hochachtungsvoll, Ihr ergebener L. Wittgenstein (Kaufmann archive, Konstanz, ). This story is not to be found in Ray Monk biography of Wittgenstein (1990a) but he told it in a newspaper article that appeared at the same time as his biography (1990b).

195 WITTGENSTEIN S CONSTRUCTIVIZATION See Marion (1998). These papers were published in Erkenntnis, vol. 2, 1931, pp For English translations, see Benacerraf & Putnam (1983, 41-65). REFERENCES Baker, G. & P. M. S. Hacker, 1985, Wittgenstein. Rules, Grammar and Necessity, Oxford, Blackwell. Behmann, H., 1930, Zur Frage der Konstructivität von Beweisen, unpublished manuscript, Wissenschaftliche Sammlung, ETH, Zurich, Hs 914: 18. A first draft is quoted from the copy preserved in the Kaufmann archive in Konstanz. Benacerraf, P. & H. Putnam (eds.), 1983, Philosophy of Mathematics. Selected Readings, sec. ed., Cambridge, Cambridge University Press. Dickson, L. E., 1952, History of the Theory of Numbers, volume I, New York, Chelsea. Euler, L., 1748, Introductio in analysin infinitorum, Lausanne; reprint: Brussels, Cultures & Civilisations, Hahn, H. & R. Carnap, K. Gödel, A. Heyting, K. Reidemeister, H. Scholz, J. von Neumann, 1931, Diskussion zur Grundlegung der Mathematik, Erkenntnis, 2, English translation in John W. Dawson, Discussion on the foundations of mathematics, History and Philosophy of Logic, 5, 1984, Hasse, H., 1950, Vorlesungen über Zahlentheorie, Berlin, Springer. Kaufmann, F., 1930, Das Unendliche in der Mathematik und seine Ausschaltung, Vienna, Deuticke. Kaufmann, F., 1978, The Infinite in Mathematics, Dordrecht, D. Reidel. English translation of Kaufmann (1930). Kenny, A., 1984, From the Big Typescript to the Philosophical Grammar, in The Legacy of Wittgenstein, Blackwell, Oxford. Kreisel, G., 1959, Wittgenstein s Remarks on the Foundations of Mathematics, British Journal for the Philosophy of Science, 9, Kronecker, L., 1901, Vorlesungen über Zahlentheorie, Leipzig, Teubner. Mancosu, P., 1999, Between Russell and Hilbert: Behmann on the Foundations of Mathematics, The Bulletin of Symbolic Logic, 5, , Mancosu, P., 2002, On the Constructivity of Proofs. A Debate among Behmann, Bernays, Gödel, and Kaufmann, in Reflections on the Foundations of Mathematics. Essays in Honor of Solomon Feferman, edited by W. Sieg, R. Sommer, & C. Talcott, Association for Symbolic Logic, Lecture Notes in Logic, vol. 15, Marion, M., 1998, Wittgenstein, Finitism, and the Foundations of Mathematics, Oxford, Oxford University Press. Monk, R., 1990a, Ludwig Wittgenstein. The Duty of Genius, London, Jonathan Cape. Monk, R., 1990b, Unphilosophical Investigations, The Independent, London, Saturday 6 October, p. 31. von Wright, G. H., 1993, The Wittgenstein Papers, in L. Wittgenstein, Philosophical Occasions , J. Klagge & A. Nordmann (eds.), Indianapolis, Hackett, Waismann, F., 1982, Über das Wesen der Mathematik. Der Standpunkt Wittgensteins, in Lectures on the Foundations of Mathematics, Amsterdam, Rodopi, ; English translation: The Nature of Mathematics: Wittgenstein s Standpoint, in S.G. Shanker (ed.), Ludwig Wittgenstein: Critical Assessments, vol. 3, London, Croom Helm, Wittgenstein, L., 1956, Remarks on the Foundations ofmathematics, Blackwell, Oxford. Wittgenstein, L., 1974, Philosophical Grammar, Oxford, Blackwell. Wittgenstein, L., 1975, Philosophical Remarks, Oxford, Blackwell. Wittgenstein, L., 1979, Ludwig Wittgenstein and the Vienna Circle, Oxford, Blackwell.

196 188 PAOLO MANCOSU / MATHIEU MARION Wittgenstein, L., 1994, Wiener Ausgabe, vol. 2, Philosophische Betrachtungen-Philosophische Bemerkungen, Vienna, Springer. Wittgenstein, L., 1996, Wiener Ausgabe, vol. 5, Philosophische Grammatik, Vienna, Springer. Wittgenstein, L., 2000, Wiener Ausgabe, vol.11, The Big Typescript, Vienna, Springer. Paolo Mancosu Department of Philosophy 314 Moses Hall University of California Berkeley, CA U.S.A. mancosu@socrates.berkeley.edu Mathieu Marion Department of Philosophy University of Ottawa 70 Laurier PO Box 450, STN.A Ottawa, Ontario Canada, K1N 6N5 mmarion@nottawa.ca

197 GRACIELA DEPIERRIS QUINE S HISTORICAL ARGUMENT FOR EPISTEMOLOGY NATURALIZED Quine s proposal of a naturalized epistemology has exerted enormous influence in contemporary analytic epistemology. A great deal of attention has been paid, in particular, to Epistemology Naturalized (EN) 1 the article in which the proposal is most explicitly made. It has not been sufficiently emphasized, however, that Quine s argument for a new epistemology is historical in character. Quine attempts to show that a naturalized epistemology is the only escape from a sequence of failures of traditional epistemology. I will argue that Quine s historical argument in EN is based on a very one-sided and misleading history of traditional epistemology. Therefore, it does not offer sufficient reasons to accept Quine s naturalistic program. I It is important to follow Quine s moves in his presentation of the epistemology he claims himself to have inherited in order to understand how he intends to persuade the reader to favor a naturalized epistemology. Quine begins EN by claiming that epistemology occupies itself with the foundations of science. Immediately after this claim, Quine proceeds to illuminate the character of epistemological studies by means of a parallel between the reductivist efforts in the foundations of mathematics and the rest of epistemology. Quine points out that there have been two kinds of studies in the foundations of mathematics: those concerning the meaning of mathematical concepts (the conceptual side of the epistemology of mathematics), and those concerning the truth of mathematical propositions (the doctrinal side). Conceptual studies were supposed to provide definitions that would generate all the mathematical concepts from clear and distinct ideas, whereas doctrinal studies were supposed to provide logical proofs that would generate all the theorems from selfevident truths. Quine writes: Ideally, the definitions would generate all the concepts from clear and distinct ideas, and the proofs would generate all the theorems from self-evident truths. (EN, p. 70) 189 F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

198 190 GRACIELA DEPIERRIS Quine adds that because of Gödel s results, and because mathematics can be reduced only to set theory, the reduction does not achieve the goal envisioned by the traditional epistemologist: Reduction in the foundations of mathematics remains mathematically and philosophically fascinating, but it does not do what the epistemologist would like of it: it does not reveal the ground of mathematical knowledge, it does not show how mathematical certainty is possible. (EN, p. 70) In this way, Quine injects the characteristics standardly attributed to Descartes s epistemological project into the foundations of mathematics. Such a superposition of the Cartesian project and logicism is revealed by the use of the terminology of clear and distinct ideas and certainty. Later, the reference to Descartes is explicit: The Cartesian quest for certainty had been the remote motivation of epistemology, both on its conceptual and its doctrinal side. (EN, p. 74) Thus, Quine repeatedly associates the logicist program initiated by Frege with an epistemological goal of achieving certainty by means of clear and distinct ideas. 2 Quine proceeds to establish a parallel between studies in the foundations of mathematics and studies in the foundations of knowledge of nature. In the same way in which it was hoped that mathematics could be reduced to logic; it was also hoped, on Quine s reading, that knowledge of nature could be reduced to sensory experience. This involved explaining the notion of body in sensory terms (on the conceptual side of the project) and justifying our knowledge claims about nature in sensory terms (on the doctrinal side) (EN, p. 71). The topic of the foundations of knowledge of nature prompts Quine to refer to Hume s views. According to Quine, Hume is concerned both with conceptual and doctrinal problems in epistemology. On the conceptual side, Quine thinks that Hume attempts to reduce by definition the language of bodies to a language of sensory impressions. 3 What is interesting about Quine s remarks here is that they reveal Quine s tendency to view traditional empiricism as a direct antecedent to his own concerns. In particular, in attributing a project of conceptual reduction to Hume, Quine s linguistic concerns appear to be the last link in a series of important transformations in the conceptual studies starting with traditional empiricism and going through, as we will see below, Russell and Carnap which studies remain within empiricism despite the transformations. On the doctrinal side, according to Quine, Hume despairs of the attempt to reduce the justification of our knowledge of the truths about nature to statements about impressions: What then of the doctrinal side, the justification of our knowledge of truths about nature? Here, Hume despaired... general statements, also singular statements about the future,

199 QUINE S HISTORICAL ARGUMENT FOR EPISTEMOLOGY NATURALIZED 191 gained no increment of certainty by being construed as about impressions. (EN, pp , emphasis added) Quine is right in attributing to Hume a radical skeptical argument regarding our knowledge of the future an argument regarding our knowledge of the causal laws of nature 4 and Quine accepts this causal skepticism explicitly using one of his most famous alliterative phrases: On the doctrinal side, I do not see that we are farther along today than where Hume left us. The Humean predicament is the human predicament. (EN, p. 72) Quine explains the kind of progress that has been achieved on the conceptual side of epistemology by noting that the translations Hume supposedly attempted can be successfully replaced by contextual definitions. In this regard, Quine mentions Russell s contextual definitions and culminates his summary of the progress attained on the conceptual side of epistemology by identifying Rudolf Carnap s project in Der logische Aufbau der Welt 5 with Russell s project in Our Knowledge of the External World 6 of reducing the external world to a logical construct of sense data. The close association of the Aufbau with Russell s program suggests that the goal of the Aufbau is to translate all the sentences about the external world into sentences about sense data (with the help of logical terms and set theory): To account for the external world as a logical construct of sense data such, in Russell s terms, was the program. It was Carnap, in Der logische Aufbau der Welt of 1928, who came nearest to executing it. (EN, p. 74) Quine then turns to the doctrinal side of epistemology and repeats that here Hume s negative result concerning causal inference remains in force. Quine illustrates this point by showing how Hume s skepticism undermines any attempt to give a doctrinal import to Carnap s constructions in the Aufbau. According to Quine, if Carnap s constructions had been successfully completed, they would have permitted the translation of all sentences about the external world into terms of sense data, logic, and set theory, but the mere fact that a sentence can be rendered in those terms does not imply that it is possible to prove it from observation sentences by means of logic and set theory. And the reason this is not possible is the lesson we have learned from Hume: even the most modest of generalizations about observable traits always covers more cases than those observed so far. Thus, for Quine, Hume s result is equivalent to the failure of the attempt to prove natural science to ground or legitimize it on the basis of immediate experience by means of logical deduction: It was sad for epistemologists, Hume and others, to have to acquiesce in the impossibility of strictly deriving the science of the external world from sensory evidence. (EN, p. 75)

200 192 GRACIELA DEPIERRIS And Quine explicitly identifies the recognition of this failure with the insight into the hopelessness of the Cartesian quest for certainty, which, according to Quine, had been the remote motivation of epistemology on both of its sides: The hopelessness of grounding natural science upon immediate experience in a firmly logical way was acknowledged. The Cartesian quest for certainty had been the remote motivation of epistemology, both on its conceptual and its doctrinal side; but that quest was seen as a lost cause. To endow the truths of nature with the full authority of immediate experience was as forlorn a hope as hoping to endow the truths of mathematics with the potential obviousness of elementary logic. (EN, p. 74) Quine then raises the question of what could have been the motivation of Carnap s heroic effort on the conceptual side of epistemology, when the above-mentioned hopes on the doctrinal side had been abandoned. Here the suggestion is that Carnap in the Aufbau focuses on the conceptual side of epistemology after recognizing the failure of the traditional doctrinal studies as Quine conceives them. Thus, Quine implies that Carnap has paid attention to the failure of a deduction of natural science from immediate experience, and has accepted Hume s negative result. Moreover, in referring to Carnap s heroic efforts, Quine suggests that Carnap s conceptual studies in the Aufbau are still guided by the Cartesian quest for certainty. In Quine s view, there were two good reasons to undertake the conceptual studies of the Aufbau, even after the hope of certainty on the doctrinal side of epistemology was abandoned. In the first place, the rational reconstructions of the Aufbau, if they had succeeded, would have elicited and clarified the sensory evidence for science, even if the inferential relation between sensory evidence and theory is not certain. In the second place, independently of any question regarding the nature of the evidence, the success of the Aufbau would have brought about a deeper understanding of our language about the external world. For, according to Quine, Carnap s rational reconstructions would have endowed this language with the degree of clarity enjoyed by observational terms, logic, and set theory. On Quine s reading, if Carnap s rational reconstructions had been successful in translating science into observational terms, logic, and set theory, they would have shown all the rest of the concepts of science to be theoretically superfluous. Thus, this translational reduction would have legitimized the theoretical language of science to whatever degree the concepts of set theory, logic, and observation are themselves legitimate. As it turns out, the legitimization Quine has in mind is a linguified empiricist and foundationalist legitimization: the translational reduction would have established the essential innocence of physical concepts by showing that they are theoretically dispensable (EN, p. 76). These claims reveal that for Quine the sole route that he sees as possibly achieving the legitimization of the language of science, if it were not doomed to failure, is the foundationalist empiricist reduction of the language of science to a supposedly clear and certain language of sensory data.

201 QUINE S HISTORICAL ARGUMENT FOR EPISTEMOLOGY NATURALIZED 193 Quine claims, still in the context of presenting the motivations and goals of the Aufbau, that two cardinal tenets of empiricism remain unassailable to this day. One is that whatever evidence there is for science is sensory evidence; the other is that all inculcation of meaning of words must rest ultimately on sensory evidence. These are precisely the reasons that explain the continuing attractiveness of the idea of a logischer Aufbau in which the sensory content of discourse would stand forth explicitly (EN, p. 75). Thus, Quine reveals here that he believes that Carnap is mainly concerned, as Quine still is, with illuminating these two aspects of empiricism and that the enduring value of the Aufbau is precisely Carnap s retention of these empiricist tenets. We can now appreciate that, until this point in the development of EN, the history of epistemology Quine recounts is a succinct history of empiricist foundationalism, and the suggestion is that this epistemology alone sets the stage for Quine s own epistemological concerns. Precisely in the context of the history of failures of empiricist foundationalism, Quine introduces his proposal for an epistemology naturalized which keeps the empiricism and throws away the foundationalism. Quine makes this introduction in a peculiar manner. He first suggests that Carnap s rational reconstructions are creative but fantastic (Quine calls them make-believe ). He then states that the stimulation of the sensory receptors is all the evidence anyone has ultimately available in arriving at a picture of the world, thus suggesting again that on this point he agrees with a view that allegedly is Carnap s. Finally, Quine introduces the idea of an epistemology naturalized by means of two rhetorical questions that single out empirical psychology as the best means to study the very topic with which both Quine and allegedly Carnap are concerned: But why all this creative reconstruction, all this make-believe? The stimulation of his sensory receptors is all the evidence anybody has had to go on, ultimately, in arriving at his picture of the world. Why not just see how this construction really proceeds? Why not settle for psychology? (EN, p. 75) Quine points out that this resort to empirical psychology would have been regarded by traditional epistemology as leading to vicious circularity. However, such scruples cease to be relevant when we abandon the project of deducing science from observations: Such a surrender of the epistemological burden to psychology is a move that was disallowed in earlier times as circular reasoning. If the epistemologist s goal is validation of the grounds of empirical science, he defeats his purpose by using psychology or other empirical science in the validation. However, such scruples against circularity have little point once we have stopped dreaming of deducing science from observations. If we are out simply to understand the link between observation and science, we are well advised to use any available information, including that provided by the very science whose link with observation we are seeking to understand. (EN, pp )

202 194 GRACIELA DEPIERRIS In what follows I sketch an alternative view of the epistemological developments Quine uses in his historical argument. My aim is to show that the central difficulties of the epistemological projects Quine discusses are very different from what Quine takes them to be. Quine s positive epistemological program as presented in EN does not address the most challenging problems of past epistemology and, for precisely this reason, it also obscures the most difficult challenges faced by Quine s own proposal. II The philosophers who define the history of traditional epistemology, for Quine Descartes, Hume, and Carnap do not understand the problems that concern Quine in the manner he imagines. In the brief history I relate below, I concentrate on key challenges these philosophers bequeath to us which are absent in Quine s historical argument. Kant s critical philosophy in particular (a philosophy Quine completely ignores) plays a crucial role in shaping the fate of the Cartesian and Humean challenges, and in providing a frame for Carnap s philosophy. If we present the development of traditional epistemology in a more accurate way, we can then appreciate that the proposal for an epistemology naturalized cannot be based on Quine s historical argument. 7 As we have seen, Quine separates the project of attaining certainty in the foundations of mathematics from the project of attaining certainty in the knowledge of nature. Quine conceives traditional epistemology as maintaining the independence of these two foundational enterprises, even though they share the same goal of certainty. This separation in Quine s conception of traditional epistemology is again due to the fact that Quine exclusively considers an alleged project of logically deducing our knowledge of nature from immediate sensory data, and he links this project to the goal of attaining Cartesian certainty. Quine does not make the horrendous mistake of believing that Descartes attempts logically to deduce our knowledge of nature from sensory data, 8 but, blinded by a crude form of empiricism, Quine ignores the real context in which the goal of certainty arises in Descartes s philosophy. In particular, Quine does not appreciate the way in which this goal is a consequence of Descartes s rationalism. In both first philosophy and physics, Descartes attempts to establish the a priori foundation of our knowledge. In his metaphysical project Descartes reveals the process of discovery and validation of the a priori foundation of any knowledge (including mathematics as well as sensory knowledge), and in his physics he opposes the obscurity and confusion of the sensory basis of Aristotelian physics by grounding physics on geometry. The certainty that we humans can achieve through the contemplation of clear and distinct ideas is modeled on the certainty we do achieve in geometry. Descartes s pursuit of certainty is an integral part of a conception according to which geometry is independently upheld as the most perfect form of human knowledge, and of Descartes s central

203 QUINE S HISTORICAL ARGUMENT FOR EPISTEMOLOGY NATURALIZED 195 interest in the subordination of our empirical knowledge of nature to a priori knowledge in a single unified system. Certainty is not for Descartes, therefore, an end in itself, prior to and independent from his rationalistic approach to the whole of knowledge. 9 There is in Descartes an ambiguity between what we might call a quasiperceptual and a quasi-platonic conception of the justificatory power of clear and distinct ideas. On the quasi-platonic conception the warrant for clear and distinct ideas is not psychologistic. Rather, they represent objective eternal truths independent of the mind of any individual subject. The distinctive Cartesian version of this conception then views the sensory and empirical world as completely and exactly describable in terms of the clear and distinct ideas of geometry, and, in this way, our empirical knowledge is completely subordinated to a priori knowledge. However, there is also in Descartes the suggestion that quasi-perceptual, direct presentation of ideas prior to understanding is itself an independent source for the legitimization of any knowledge. This conception of ultimate evidence derived from Descartes s theory of ideas is suitable for a theory of knowledge modeled on empirical perception. Hume, through Locke and Berkeley, takes up this conception, although not without modification. Hume draws the distinction between a priori and empirical knowledge between relations of ideas and matters of fact but the source of justification for both a priori and empirical knowledge for Hume is the direct inspection in a quasi-perceptual way of items present before the mind. However, the source of ultimate justification in Hume begins and ends with the individual subject. In Descartes, as we saw above, clear and distinct contemplation with the mind s eye does not reduce completely to the observation of mental contents of an individual subject, but rather consists in the contemplation of mental contents representing objective truths. It is precisely the independent legitimacy and objectivity of the a priori that allows Descartes to attempt the subordination of sensory knowledge to a priori knowledge which is, of course, quite absent in Hume. It is only by conceiving certainty in a Humean, psychologistic guise that Quine can then slide from the goal of Cartesian certainty to the goal of certainty in empiricist foundationalism. This lacuna in Quine s story becomes even clearer when we turn to Kant. For, Kant s project continues but also transforms the Cartesian project of subordinating empirical knowledge to a priori knowledge. Thus, one of the central objectives of Kant s Critique of Pure Reason is to demonstrate how it is possible that our synthetic a priori knowledge provides its formal structure to the empirical part of our knowledge and thus secures its objectivity. This is precisely the content of Kant s Copernican revolution in epistemology. In Kant the relationship between the a priori ingredients and the rest of the knowledge which the a priori legitimizes is not deductive but transcendentally constitutive. According to Kant, there exist a priori components of our knowledge without which our empirical knowledge would not be possible. These a

204 196 GRACIELA DEPIERRIS priori components constitute the very object of knowledge, since without them our experience would be chaotic and would be confined to individual empirical subjects. We would not, therefore, attain the coherence and objectivity that it is possible to attain in our judgments. Kant s constitutive relationship of the a priori with respect to the empirical knowledge of nature is a very different relationship than the deductive relationship favored by foundationalism. 10 Quine does not appreciate that Carnap in the Aufbau continues with the rationalist and Kantian tradition that bestows on a priori knowledge in the case of the Aufbau, the new mathematical logic of Frege and Russell an objectifying role with respect to empirical knowledge. 11 In the Aufbau Carnap reconstructs, by means of the logic of Principia Mathematica, 12 the process through which a priori and empirical knowledge are combined and integrated in the total system of science. Carnap offers definitions of the concepts of natural science in a single system on the basis of some initial concepts. Although Carnap chooses elementary experiences of an individual human subject as the initial concepts of the construction, the goal of the Aufbau is to exhibit the logical order and relations that lead to the construction of the object of knowledge of the science of nature. In the Aufbau, the object of scientific knowledge is best represented by the logical form or structure, not by the content of the subjective experiences from which the construction begins. Indeed, the transformation of subjective elementary experiences into objective concepts takes place by means of logical relations. Thus, Carnap intends, in a Kantian vein, to exhibit how certain logical constructions make possible ( constitute ) the very possibility of intersubjective communication of purely subjective contents and, thereby, make possible the adjudication of both truth and falsity to the sentences of natural science. To suggest, as Quine does, that Carnap in the Aufbau engages in the conceptual side of epistemology because he recognizes Hume s skeptical result, is to ignore that Carnap is completely oblivious to Hume s causal skepticism. There is no reference at all in the Aufbau to Hume s skepticism. 13 More importantly, the Kantian tradition that has influenced Carnap faces problems quite different from Hume s. On the one hand, Kant s conception of a priori knowledge, as part of Kant s transcendental idealism, allows for an answer to Hume s causal skepticism. On the other hand, however, the post-humean problem faced by philosophers working in the Kantian tradition concerns doubts regarding central features of the Kantian synthetic a priori in particular, its relation to an a priori intuition. In this way, the universal validity and unrevisability of the Kantian constitutive a priori has also been called into question. After the Aufbau, the Kantian influences on Carnap persist. In Empiricism, Semantics, and Ontology, 14 for example, Carnap works with a version of the Kantian constitutive relation according to which linguistic rules laid down in advance make possible and determine the cognitive content of our statements. Carnap relativizes the constitutive role of the a priori by making these rules revisable. In Carnap s conventional, relativized, and revisable version of the Kantian constitutive relation, the cognitive content of sentences is determined by

205 QUINE S HISTORICAL ARGUMENT FOR EPISTEMOLOGY NATURALIZED 197 the rules that first erect a given linguistic framework. Questions of truth and falsity are internal questions to be raised within a framework, that is, after certain rules for forming statements and for testing, accepting, or rejecting them 15 have been laid down. Even in the linguistic framework of everyday life, in which we refer to the spatio-temporally ordered observable things and events in the physical world, there are implicit rules that define the framework. These rules (which can be made explicit by rational reconstruction) determine the methods of empirically testing our statements about things in the world. It is always possible to choose to operate within a different linguistic framework constituted by different rules. Quine holds that the traditional project is one of logically deducing our theories of nature from a privileged starting point endowed with certainty. However, in the tradition developing from Descartes s rationalistic conception of certainty, the goal is the subordination of empirical knowledge to a priori knowledge. This subordination is explicitly articulated by Kant into one in which the relationship between our own a priori contribution and sensory data is not deductive. Rather, in the Kantian conception, the a priori provides the framework within which experience is made to fit. And Carnap s project in the Aufbau, as we have seen, has been decisively influenced by this Kantian conception. In suggesting that all Carnap has inherited from traditional epistemology is the failure of the attempt to achieve certainty by deducing the truths of natural science from immediate experience, Quine gives a one-sided and extremely impoverished characterization of the development of the epistemological tradition that culminates with Carnap. Quine completely neglects to consider that the epistemological concerns he himself inherits do not reduce to a simple-minded obsession with certainty or the logical deduction of knowledge of nature from some privileged basis. They centrally consist, instead, in establishing a relationship between a priori and empirical knowledge according to which the former makes possible the objectivity of the latter. III Kant is the first epistemologist to articulate explicitly the distinction between questions of fact (quid facti), and questions of right (quidjuris) the distinction between empirical factors involved in the formation of our beliefs, and global normative questions regarding their in principle justification, legitimacy, or objectivity. 16 Hume also operates with his own version of this distinction, although not so explicitly. 17 Hume is concerned, in his role as a radical skeptic, with the justification in principle of our most fundamental beliefs. In particular, Hume adopts the global normative point of view of radical skepticism precisely when he asks about the ultimate grounding of any causal inference we can produce leading to the formulation of causal laws of nature. On the other hand, Hume allows our natural tendencies to rescue us from skepticism, and adopts the

206 198 GRACIELA DEPIERRIS non-skeptical point of view of common life and science. From this standpoint, Hume develops a model of a naturalized epistemology that is independent of his radical skepticism. In Kant, as in Hume, the point of view from which the philosopher raises questions regarding the legitimacy and objectivity of our knowledge is also independent of the point of view of common life and science. Kant takes issue with Hume s skepticism from what Kant calls a transcendental standpoint. There is, for Kant, a fundamental distinction between the transcendental and the empirical standpoints: Not every kind of knowledge a priori should be called transcendental, but that only by which we know that and how certain representations (intuitions or concepts) can be employed or are possible purely a priori. The term transcendental, that is to say, signifies such knowledge as concerns the a priori possibility of knowledge, or its a priori employment. Neither space nor any a priori geometrical determination of it is a transcendental representation; what can alone be entitled transcendental is the knowledge that these representations are not of empirical origin, and the possibility that they can yet relate a priori to objects of experience. The application of space to objects in general would likewise be transcendental, but, if restricted solely to objects of sense, it is empirical. The distinction between the transcendental and the empirical belongs therefore only to the critique of knowledge; it does not concern the relation of that knowledge to its objects. (Critique of Pure Reason, A56-7/B80-81) Therefore, in order either to obtain Hume s radical causal skepticism or to give a Kantian transcendental answer to it, we must accept the distinction between a scientific explanation or even a local question of justification of our knowledge claims, on the one hand, and their in principle philosophical legitimization, on the other. In the traditional epistemology of Descartes, Hume, and Kant, the adoption of a characteristically philosophical standpoint external to science and everyday beliefs always involves global normative questions concerns the justification in principle, legitimacy, or objectivity of our knowledge as a whole. Similarly, after the Aufbau, Carnap proposes in Empiricism, Semantics, and Ontology the distinction between external and internal questions regarding linguistic frameworks. But there Carnap neither adheres to Hume s skeptical arguments nor argues in a transcendental Kantian style about the legitimacy and objectivity of our knowledge. External questions, for Carnap, are instead deflated into pragmatic questions regarding which form of language is more convenient to adopt. Nonetheless, whereas internal questions always concern questions of local justification within one or another particular framework, external questions concern the global question of which form of language for the total system of science is most convenient. Quine s historical argument for turning epistemology into a chapter of empirical psychology relies on the failure of traditional epistemology to deduce science from sensory evidence, in particular, on Hume s causal skepticism. However, the recognition of Hume s general skeptical result requires the external

207 QUINE S HISTORICAL ARGUMENT FOR EPISTEMOLOGY NATURALIZED 199 standpoint from which the traditional philosopher asks global normative questions. Without the normative external standpoint of traditional epistemology, there simply is no global philosophical skepticism. IV These last considerations allow me to reply to an obvious objection to my procedure. Even if Quine gives a completely fictitious history of the epistemology that culminates with Carnap, one might object, a distorted history does not need to affect the strength of Quine s positive proposal of an epistemology naturalized. We can forget the history of failures the negative motivation and keep the positive proposal. In reply, I argue that there are important ambiguities in Quine s exposition concerning whether or not Quine can avoid occupying the external standpoint of the traditional philosopher. These ambiguities result from the extent to which Quine endorses, perhaps unwittingly, traditional empiricism, and from the way in which Quine formulates his own positive proposal. The general character of Quine s epistemology naturalized is very much like the approach Hume adopts as a scientist of human nature. Hume s naturalistic point of view can be characterized in the words Quine uses in EN to characterize his own project: Epistemology, or something like it, simply falls into place as a chapter of psychology and hence of natural science. It studies a natural phenomenon, viz., a physical human subject. This human subject is accorded a certain experimentally controlled input certain patterns of irradiation in assorted frequencies, for instance and in the fullness of time the subject delivers as output a description of the three-dimensional external world and its history. (EN, pp ) But Quine takes it as unproblematic to hold within his naturalistic approach that: The relation between the meager input and the torrential output is a relation that we are prompted to study for somewhat the same reasons that always prompted epistemology; namely, in order to see how evidence relates to theory, and in what ways one s theory of nature transcends any available evidence. (EN, p. 83) These words contain, in my view, an important ambiguity. Quine is not merely referring here to the psychological study of the empirical origin of our beliefs. Nor is Quine referring to the evidential relation between certain data and a specific scientific theory. Instead, Quine talks of the attempt to discover how evidence in general relates to theory in general, and how any available evidence goes beyond any theory of nature as such. Moreover, to reinforce the suggestion that he is concerned with a global epistemological question, Quine also claims that to attempt such a discovery has been the motivation of traditional epistemology. This description of the problem of the relation between evidence and theory fits rather well with the approach of traditional philosophical studies that attempt

208 200 GRACIELA DEPIERRIS to show the grounding in principle of our knowledge of nature, not with the approach of an empirical psychologist whose concern would typically be the specific evidence relevant to a specific psychological generalization. In Word and Object, and other writings, Quine explicitly claims that the point of view external to the body of scientific and common-sense knowledge, from which the traditional philosopher intends to evaluate the legitimacy in principle of the body of knowledge, cannot be adopted any longer. In Quine s words, it is not possible to opt for a cosmic exile. 18 Thus Quine rejects the distinction between the external and internal standpoints of Hume, Kant, and Carnap. In light of this, it is peculiar indeed that throughout his historical argument in EN Quine repeats that the Humean predicament is the human predicament, and uses the Humean predicament to explain, for example, why the rational reconstructions of the Aufbau could not be reasonably taken as trying to justify the science of nature, but rather as part of a project of conceptual reduction. How are we possibly to understand the Humean predicament if not as the result of a manner of investigating epistemological questions characteristic of an autonomous philosophical standpoint? It appears that if Quine wishes to accept Hume s inductive skepticism, Quine should acknowledge the generality of Hume s results, and thus the autonomy of Hume s skeptical standpoint with respect to empirical science. Quine never raises the question of how his acceptance of Hume s inductive skepticism, which as we have seen Quine takes to be the permanent legacy, on his reading, of traditional epistemology, is possibly consistent with his rejection of the possibility of any general normative epistemological standpoint. NOTES In a note at the beginning of the article, Quine says that it was first presented in Vienna in September In what follows I cite from the article as it appears in W. V. Quine, Ontological Relativity and Other Essays. New York: Columbia University Press, 1969, pp shall abbreviate the title as EN. It is difficult to attribute to Frege s program the goal of certainty as understood by Quine. Thus, it is possible that here Quine is referring only to Bertrand Russell s logicist program as it relates to Russell s epistemology. Nonetheless the remarks as they stand are misleading. On the controversy whether Frege has an epistemological projects, and if so, of what kind, see, for example, Paul Benacerraf, The Last Logicist, in: P. French et al. (Eds.), Midwest Studies in Philosophy VI. Minneapolis: University of Minnesota Press, 1981; Thomas Ricketts, Frege, The Tractatus, and the Logocentric Predicament, in: Noûs, Vol. XIX, No. 1, March 1985; Thomas Ricketts, Objectivity and Objecthood: Frege s Metaphysics of Judgment, in: L. Haaparanta and J. Hintikka (Eds.), Synthesizing Frege. Dordrecht: Kluver, 1986; Graciela De Pierris, Frege and Kant on A Priori Knowledge, in: Synthese 77, 1988; Joan Weiner, Frege in Perspective. Ithaca: Cornell University Press, 1990; Tyler Burge, Frege on Knowing the Third Realm, in: Mind, Vol. 101, No. 404, 1992; Joan Weiner, Realism bei Frege: Reply to Burge, in: Synthese 102, Quine writes: By [Hume s] identification of bodies with impressions he did succeed in construing some singular statements about bodies as indubitable truths, yes; as truths about impres-

209 QUINE S HISTORICAL ARGUMENT FOR EPISTEMOLOGY NATURALIZED sions, directly known (EN, pp ). Later, Quine adds: Hume s and [Alexander Bryan] Johnson s desperate measure of identifying bodies with impressions ceased to be the only conceivable way of making sense of talk of bodies, even granted that impressions were the only reality (EN, p. 72). In Hume s Pyrrhonian Skepticism and the Belief in Causal Laws, Journal of the History of Philosophy, Vol. XXXIX, No. 3, July 2001, pp , I argue against what I call the vehicle view of the relationship of Hume s skepticism and his naturalism a view initially suggested by Norman Kemp Smith and further developed by more recent naturalistic interpretations. In my view, Pyrrhonian radical skepticism constitutes for Hume a permanently available frame of mind for those who engage in philosophical studies. Rudolf Carnap, Der logische Aufbau der Welt. Berlin: Weltkreis, 1928; second edition, Hamburg: Meiner, 1961; translated by Rolf A. George, The Logical Structure of the World. Berkeley and Los Angeles: University of California Press, Bertrand Russell, Our Knowledge of the External World as a Field for Scientific Method in Philosophy. London: Allen & Unwin, For lack of space, the alternative history I present below is only a rough sketch. However, Quine does oversimplify Hume s views in suggesting that Hume attempts such a deduction. In Causation as a Philosophical Relation in Hume, Philosophy and Phenomenological Research, Vol. LXIV, No.3, May 2002, I argue that Hume s skeptical argument concerning causal inference does not reduce to an argument against the possibility of deducing the conclusions of our causal inferences from sensory data, but, rather, it centrally concerns the impossibility in principle of legitimizing the causal inference by means of probable arguments that is, by means of arguments based on past and present experience. On the relationship between Descartes s method in metaphysics and the method of geometry, see the interpretation of Descartes by Martial Gueroult, Descartes and the Order of Reasons, translated from the second French edition of 1968 by Roger Ariew. Minneapolis: University of Minnesota Press, I discuss the Kantian relation of constitutivity in The Constitutive A Priori, Canadian Journal of Philosophy, Supplementary Volume 18, December 1992, pp I base my interpretation of the Aufbau on the detailed analysis of this work by Michael Friedman, in his book Reconsidering Logical Positivism. Cambridge, UK: Cambridge University Press, 1999, especially Part Two. Alfred North Whitehead and Bertrand Russell, Principia Mathematica. Cambridge, UK: Cambridge University Press, In the Aufbau Carnap refers only to what he takes to be a positive theory in Hume, namely, a regularity view of causation (Aufbau, section 165). Rudolf Carnap, Empiricism, Semantics, and Ontology, Revue Internationale de Philosophie 11, 1950, reprinted in Meaning and Necessity, second edition. Chicago and London: University of Chicago Press, Rudolf Carnap, Meaning and Necessity, op. cit., p For Kant s explicit reference to the distinction between quid facti and quid juris, see Critique of Pure Reason, A84-92/B In Hume s Pyrrhonian Skepticism and the Belief in Causal Laws, op. cit., I argue that Hume non-simultaneously adopts two independent standpoints the standpoint of radical Pyrrhonian skepticism, on the one hand, and the standpoint of our natural non-skeptical beliefs, on the other. W. V. Quine, Word and Object. Cambridge, Massachusetts: The MIT Press, 1960, pp Dept. of Philosophy Indiana University Bloomington IN U.S.A. gdepierr@indiana.edu

210 ELLIOTT SOBER TWO USES OF UNIFICATION 1. INTRODUCTION Carl Hempel 1 set the tone for subsequent philosophical work on scientific explanation by resolutely locating the problem he wanted to address outside of epistemology. Hempel s problem, as I will call it, was not to say what counts as evidence that X is the explanation of Y. Rather, the question was what it means for X to explain Y. Hempel s theory of explanation and its successors don t tell you what to believe; instead, they tell you which of your beliefs (if any) can be said to explain a given target proposition. At the same time that Hempel s problem developed outside of epistemology, another project gained a footing within that discipline, one that also laid claim to the concept of explanation. This is the Peircean idea of abduction of inference to the best explanation which was put forward as a solution to the problem of understanding how nondeductive inference works. 2 To evaluate competing theories in the light of a body of observations, we are supposed to ask which of those theories is most explanatory of the observations. Or, more generally, we need to ascertain which of the competing theories, if added to our stock of beliefs, would lead to a total system of beliefs with the most explanatory power. 3 Peirce s problem, as I will call it, is the epistemological problem of specifying the rules that govern inference to the best explanation. In addition to the fact that these two discussions of the concept of explanation are located on opposite sides of the fence that separates epistemology from the rest of philosophy, the two projects differ in further respects. Hempel s question is yes or no. Does X explain Y, or not? However, for the Peircean proponent of abduction, the fundamental question is comparative does X explain Y better than Z does? An additional contrast derives from the fact that Hempel wanted to explicate the concept of an ideally complete scientific explanation. Although he recognized that the discoveries of science can be explanatory without satisfying his requirements he termed these fragments explanation sketches he once again set the problem that most of the subsequent literature addressed. The task is not to understand what allows one proposition to help explain another; 4 rather, the goal is to understand what it means for one proposition to provide a complete explanation of the other. Students of abductive inference, on the other hand, 205 F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers, Printed in the Netherlands.

211 206 ELLIOTT SOBER usually take it as a given that the competing hypotheses one needs to assess rarely provide complete explanations in themselves. I have distinguished Hempel s problem and Peirce s problem because I want in this essay to describe how the concept of unification pertains to each. I ll do this by evaluating two principles that address the problem of how one should explain two similar observations. Giving a full description of what similarity amounts to in this context would be difficult, but for our purposes we can restrict ourselves to the following sufficient condition: two observational propositions are similar when they apply the same predicates to different individuals. 5 The observation that a is green and the observation that b is green are similar in this sense, as are the observation that c runs faster than d and the observation that e runs faster than f. Here, then, are two principles about how the similar observations and should be explained: (P) If you want to judge which of the hypotheses U and D is more plausible in the light of the observations and the fact that U provides a unified explanation of and while D gives a disunified account, is evidence in favor of U. (H) If U and D are both true, and U provides a unified description of phenomena and while D provides a disunified description of those phenomena, then U may explain and but D does not. Principle (P) is an epistemological claim that addresses Peirce s problem; it says that unification is confirmatory. Principle (P) does not rule out the possibility that the only true explanations of the observations are disunified; rather, it says that in deciding what to believe, it is a point in favor of U that it provides a unified account. Principle (H), on the other hand, is not epistemological; it addresses Hempel s problem, asserting that similar phenomena must receive a unified explanation. The point of (P) is to help you figure out which of U and D to believe; in contrast, the point of (H) is to help you decide which of the propositions you already believe provides an explanation of two target observations. My thesis will be that these two uses of the concept of unification should receive very different assessments. I ll argue that unification has no objective connection with Hempel s problem. Whether we embrace or reject (H) depends on what sort of explanation we seek; there are perfectly legitimate scientific goals that make unified explanation a virtue while other equally worthy scientific goals have the opposite effect. I ll argue that the status of (P) is different; there is an objective reason why unification is relevant to Peirce s problem. 2. ANTIREDUCTIONISM AND HEMPEL S PROBLEM Hilary Putnam has had a decisive influence on how philosophers think about reductionism. In fact, he has had that influence twice. The first time was his essay with Paul Oppenheim 6 in which he defended reductionism as an empirical

212 TWO USES OF UNIFICATION 207 thesis about the relationships that connect the different sciences. The second occasion was a pair of articles in which he argued against his earlier position. 7 The anti-reductionism that this more recent Putnam-time-slice espoused continues to be a powerful presence in philosophy; anti-reductionism is now more or less the received view among philosophers of science and philosophers of mind, and the standard argument for this standard view derives from Putnam. Putnam s argument appeals to the value of unification in explanation, and criticizes reductionism for failing to deliver the requisite sort of unification. It is ironic that he and Oppenheim earlier had seen reductionism as the vehicle for achieving unification among the sciences; reductionism changes from hero to villain in Putnam s thinking, but the value of unity remains constant. The reductionism that Putnam aims to refute is micro-reductionism. This is the idea that the properties of wholes can be explained solely in terms of the properties (including relational properties) of parts. The social sciences reduce to psychology, psychology to biology, biology to chemistry, and chemistry to physics, or so the micro-reductionist claims, because the objects described at higher levels turn out to be composed exclusively of the objects described at lower levels. Putnam did not gainsay the thesis that societies are made of individuals, that brains are made of cells, and so on. Rather, he rejected the reductionist s claim about explanation that every event explainable by a higher-level science can also be explained by a lower-level science. Putnam challenges this reductionist thesis about explanation by way of a delightfully simple example. He asks the reader to consider a plank of wood that has two holes a square hole that is one inch on a side and a circular hole that is one inch in diameter. He points out that a square peg that is 15/16th of an inch on a side will fit through the square hole, but not through the round one. Why is this so? Putnam claims that the right explanation for this fact is given by the macroproperties just cited. It is the geometric shapes and dimensions of the peg and the holes that do the explaining. Putnam then claims that it would be entirely unexplanatory to cite the positions and other properties of the molecules that compose the peg and the board. This welter of micro-details would miss the wood for the trees. The micro-story, says Putnam, is either a non-explanation, or a vastly inferior explanation. Since the macro-story explains something that the microstory cannot explain (or cannot explain nearly as well), reductionism is false. Putnam claims that there are objective standards that distinguish good explanations from bad ones (and from non-explanations); it is on this objective basis that he praises the macro-story and blames the micro-story. Good explanations should advert to details that are explanatorily relevant, and omit details that are explanatorily irrelevant, where explanatory relevance is supposed to be an objective notion, not just a function of our interests. Putnam says that one feature of explanatory relevance is the idea that good explanations are general they apply not just to the specific system under study, but to others. This is why the macro-story is superior to the micro-story; there are lots of other peg and board systems that have the dimensions specified, while there are relatively few (per-

213 208 ELLIOTT SOBER haps none) that have the exact molecular configuration that the micro-story describes. The imperative to find explanations that are general is equivalent to the imperative to find explanations that are unified. To see this, consider two peg and board systems, which differ in their micro-configurations but have in common the macro-dimensions that Putnam mentions. A unified explanation of why the pegs in both cases fit through one hole but not the other will appeal to the macroproperties that the two systems have in common. In contrast, the micro-details do not provide a unified explanation; from the micro point of view, the two systems are different. This is where principle (H) makes its appearance. In the search for explanation, macro-accounts are preferable to micro-accounts because they are more general, and generality is just unification by another name. Putnam never says much about what he means by explanatory relevance. In the example he describes, it may seem obvious that a detailed description of all the many molecules does not enhance our understanding. But surely it isn t always true that micro-details are explanatorily irrelevant. For example, suppose we observe that Smith and Jones both have lung cancer. A unified explanation of this pair of observations might be that (U) Smith and Jones both smoked cigarettes. However, suppose that Smith and Jones smoked different brands of cigarette, and that these brands contain different carcinogenic ingredients. This allows us to formulate the following disunified explanation: (D) Smith smoked brand X, which contained carcinogenic x-particles, while Jones smoked brand Z, which contained carcinogenic z-particles. Is it really so obvious that (U) is the uniquely correct explanation that (D) is either no explanation at all or a terrible explanation? On the contrary, I think it is obvious that (D) may have considerable explanatory interest. Perhaps the different carcinogenic ingredients give rise to different types of tumor, and maybe these different tumors are best treated by different therapies. Putnam values generality but accords no explanatory importance to detail. It would be equally one-sided to value detail only, with no recognition of the virtue of generality. Rather, I suggest that which type of explanation is better depends on your interests; (U) is better if your goal is generality and unification, but (D) is better if your goal is detail. Both goals are important in science. I suggest that the same is true of Putnam s peg and board example. Perhaps the molecular details don t interest Putnam or most of the rest of us, but that doesn t mean that they can t have any scientific interest at all.

214 TWO USES OF UNIFICATION 209 Is it possible that an account of explanatory relevance might be offered that shows that there is an objective basis for ruling out the micro-explanation in the case of Putnam s peg and board, but allows that the disunified explanation (D) is explanatorily relevant in the case of Smith and Jones? I am skeptical. The cases appear to be structurally the same, as the following causal diagrams indicate: micro-constitution of the peg and board at time macro-properties of the peg and board at the peg goes through the square hole but not the round one at micro-constitution of the smoke and of Jones lungs at cigarette smoke at Jones gets lung cancer at Jones inhales Here the relation of synchronic determination that connects lower-level to upperlevel events is represented by the single arrow and the diachronic relation of causation is represented by the double arrow This is not to deny that there is (or may be) a difference between the two examples. Let us suppose that the macro-configuration of the board and peg deterministically ensures that the peg will fit through one hole but not the other, but that Smith s smoking merely confers a probability (that is less than unity) on his getting lung cancer. This difference entails a further difference that the macro-configuration screens-off the micro-configuration from the effect in the case of the board and peg, but need not do so in the case of smoking and cancer. That is, the first probabilistic equality will be true, but the second need not be: Pr(the peg fits through one hole but not the other the macro-dimensions of the board and peg) = Pr(the peg fits through one hole but not the other the macro-dimensions of the board and peg & the micro-configuration of the molecules in the board and peg) = 1.0. Pr(Smith gets lung cancer Smith smokes cigarettes) = Pr(Smith gets lung cancer Smith smokes cigarettes & Smith inhales x- particles). The screening-off relation will fail in the smoking example, if there are different types of cigarette that differ in how carcinogenic their ingredients are. The reason that screening-off cannot fail in the deterministic case, but can do so in the probabilistic case, is that in both cases the probability of the effect, given the macro-configuration, is an average. When there are n micro-configurations (i = l,2,...,n) that have positive probability, given the macro-configuration described,

215 210 ELLIOTT SOBER Notice that if the left-hand side of this equality has a value of unity, each of the n probabilities of the form Pr[Effect Macro & Micro(i)] must also have a value of unity; if a set of probabilities averages to one, all must have a value of one. However, if Pr(Effect Macro) < 1.0, Pr[Effect Macro & Micro(i)] and Pr[Effect Macro & Micro(j)] may have the same value, but they need not. Even if screening-off holds in the case of Putnam s example, but is violated in the smoking example, it would be a mistake to embrace screening-off as the key to understanding explanatory relevance. For one thing, in a Markov chain that links a distal cause to a proximate and this to an effect (E), the proximate cause screens off the distal cause from its effect, but this should not lead us to conclude that the distal cause is not explanatorily relevant to the effect. 8 I propose that precisely the same point holds when we consider the relationship between the macro-cause its supervenience base (S), and the effect term (E): Even if the macro cause screens off its supervenience base (S) from the macro-effect (E), it does not follow that S is explanatorily irrelevant to E. 9 Suppose the different brands of cigarette are equally carcinogenic (and so screening-off holds); it still may be explanatorily relevant to say that Smith s lung cancer traces back to his inhaling x-particles; going beyond the weaker claim that he smoked cigarettes is not irrelevant babble. Although Putnam does not provide an account of what the relation of explanatory relevance amounts to, there is a natural candidate that is worth mentioning. If an explanation must describe the causes of the event one wishes to explain, 10 and if the difference between cause and non-cause is objective, then some proposed explanations are objectively false. It is false to say that the peg fit through one hole but not the other because the board was painted red, and it is equally false to say that Smith got lung cancer because his fingers had nicotine stains on them. However, I know of no plausible reading of the concept of cause that entails that the board s micro-constitution was like its color, or that Jones inhaling x-particles was like his having yellow fingers. Putnam, like Jaegwon Kim, thinks one has to choose between the micro- and macro-accounts. 11 Kim argues that the causal action is to be found solely at the micro-level; Putnam contends, as we have seen, that the explanatory action is at the macro-level alone. In fact, both of these monolithic positions are mistaken; there is no need to choose. Both micro and macro provide true descriptions of the causal facts, and both thereby provide true causal explanations. 12 The two types of description have complementary virtues; one offers detail while the other offers generality and unification. 13

216 TWO USES OF UNIFICATION MODEL SELECTION AND PEIRCE S PROBLEM Model selection is the name for a family of statistical inference problems in which the hypotheses considered contain adjustable parameters (a concept I ll explain below). There are a variety of criteria now on the market for choosing among such models. Since my goal here is to show how unification is relevant to Peirce s problem, I won t attempt to provide a complete survey and critical assessment of the options. It will suffice to describe briefly one of the most influential and interesting approaches that due to Hirotugu Akaike; 14 like other proposals, it applies to a context in which unification is epistemically significant and thereby provides an explication of principle (P). As an example, let s consider the X- and Z-cigarettes that Smith and Jones smoked. Suppose you want to infer how the number of cigarettes smoked affects an individual s chance of getting lung cancer. Although there are many different mathematical models that might be considered, for the sake of simplicity I ll restrict my attention to linear models; these models maintain that each additional X-cigarette smoked increases one s chance of getting lung cancer by a constant amount, and that the same is true of each additional Z-cigarette. However, I will not assume that an X-cigarette and a Z-cigarette are equally risky. Deciding whether this is so is precisely the model selection problem I want to describe. Suppose we examine a large number of smokers and ascertain, of each individual, how many X- and Z-cigarettes he or she smoked, and also whether he or she has lung cancer. With enough individuals in our sample, we can describe the frequency of lung cancer within different categories of smokers, ranging from those who have smoked only a few cigarettes in their lifetimes to those who have smoked many. We also may be able to compare the incidence of lung cancer among individuals who have smoked the same number of cigarettes, but who have smoked different brands. There will be an upward trend in this data set the more cigarettes smoked, the higher the frequency of lung cancer. There are two linear models that I want to consider. The first says that it is merely the total number of cigarettes smoked that determines one s risk of lung cancer: (U) This model is unified because it assumes that the effect of smoking an additional X-cigarette is precisely the same as the effect of smoking an additional Z- cigarette. Smoking is smoking, this model says; it doesn t matter which brand you smoke, as far as the probability of getting lung cancer is concerned. An alternative model, also linear in form, allows for the possibility that the different brands may pose different risks:

217 212 ELLIOTT SOBER (D) This model is disunified, in that it provides separate representations of the effects of the two types of cigarette. It won t matter to our subsequent discussion if we assume that and must have different values; the point is that (D) does not constrain them to be the same. Notice that the unified model is simpler, in that it contains two adjustable parameters and whereas the disunified model contains three and Notice also that each model can be viewed as an infinite disjunction, each disjunct of which is formed by assigning values to the model s adjustable parameters. How are these two models related to the data we have gathered? In each case, we can find the best-fitting member of the model by finding the maximum likelihood values of the adjustable parameters i.e., the values of these parameters that maximize the probability of the data. These likeliest members of the two models, L(U) and L(D), contain no adjustable parameters; rather, their values have been adjusted. It is pretty much inevitable that L(D) will fit the data better than L(U). 15 Does this mean that we should automatically prefer the more complex model? All of the model selection criteria now used in science answer this question with an emphatic no. Although the reasons they give are different, their bottom line is the same there can be good reasons for preferring (U) over (D) even though L(U) fits the data worse than L(D) does. Akaike s approach to this problem is especially instructive. Akaike suggested that the problem of model selection be conceived in terms of the goal of finding models that will be predictively accurate. This conception of the goal of model selection is what I mean by Akaike s framework. Akaike also proposed a means for achieving that goal; he proved a theorem that describes how one can obtain an unbiased estimate of a model s predictive accuracy. This theorem is the basis for what has come to be called the Akaike information criterion (AIC). This separation of Akaike s framework from his criterion is important; there may be circumstances in which AIC is not the best criterion to use in model selection, even granting the goal of maximizing predictive accuracy. Akaike s idea of predictive accuracy has to be understood in terms of a twostep process. Models that contain adjustable parameters make predictions in the following sense: first one draws a set of data from the true underlying distribution and uses those data to estimate the values of the model s parameters (by maximum likelihood estimation). One then uses that fitted model to predict a new data set drawn from the same distribution. In terms of our previous notation, we use a model M to make a prediction about new data by first using the old data to find L(M) it is L(M) that makes a definite prediction. The predicted values may be close to the new data, or far away (as measured by the Kulback-Leibler distance measure). Imagine using a model repeatedly in this two-step process; there will doubtless be some variation among these repetitions in terms of how

218 TWO USES OF UNIFICATION 213 well the fitted model predicts new data. The average performance of the model is what defines its predictive accuracy. The predictive accuracy of M is the expected likelihood of L(M). Having models that are predictively accurate may be a desirable goal, but how is one to ascertain how predictively accurate a model is apt to be? Akaike s (1973) remarkable theorem provides an answer: An unbiased estimate of the predictive accuracy of model One takes the logarithm of the likelihood of the fitted model and subtracts k, the number of adjustable parameters. 16 Complex models, when fitted to the data, tend to have higher log-likelihoods than simpler ones, but they also incur a larger penalty because of their complexity. For a complex model to have a higher AIC value than a simpler one, it isn t enough that the complex model fit the data better; it must fit the data better by a sufficient margin to overcome the fact that it is more complex. In terms of our problem of comparing the unified and disunified models (U) and (D), the Akaike criterion entails that if L(U) and L(D) fit the data about equally well, we should prefer the unified model. Unified models aren t always preferable to disunified models, but unification nonetheless is epistemically relevant. Unification is evidence of superior predictive accuracy when the unified model contains fewer adjustable parameters. As mentioned earlier, Akaike s criterion for model selection is not the only game in town. However, AIC and the other criteria now in use agree that high likelihood of L(M) is good news for a model, whereas a large number of parameters is bad. The criteria differ in how they weight the importance of these two factors. However, for present purposes, their points of agreement are more important than the things they disagree about. AIC is typical of these other criteria in that it embodies a defeasible presumption in favor of unification. 4. CONCLUDING COMMENTS I began this paper by asserting that unification has no objective significance in the context of Hempel s problem, but that it is objectively relevant to solving Peirce s problem. I expressed this contrast by saying that the relevance of unification depends on our interests in the former case, but not in the latter. However, in explaining Akaike s ideas, I first described the goal of predictive accuracy and then outlined how AIC provides a means for achieving that goal. Doesn t this mean that unification is equally subjective and interest-relative in the two contexts that principle (P) has no more claim on our allegiance than principle (H)? No the contrast between subjective and objective survives our recognizing that the difference between good and bad inference depends on what our goals are in

219 214 ELLIOTT SOBER drawing inferences. If you want to figure out which of several models will be most predictively accurate, then it is an objective fact that unification is relevant. However, if you want to figure out which of the propositions you already believe should be used to explain a body of observations, then whether you choose a unified or a disunified account depends on your interests. Disunified explanation is not a contradiction in terms, nor are disunified explanations inherently inferior to unified explanations. If we discover that smoking cigarettes increases one s risk of lung cancer, and that X-cigarettes and Z-cigarettes differ in their carcinogenic properties, we face a choice. We can explain Smith and Jones cancer by describing what they have in common, or by describing a way in which they differ. They got cancer because they both smoked cigarettes, but it also is true that inhaling x-particles caused Smith s cancer while inhaling z-particles caused Jones. 17 NOTES In Carl G. Hempel, Aspects of Scientific Explanation and Other Essays in the Philosophy of Science. New York: Free Press, Gilbert Harman, Inference to the Best Explanation. Philosophical Review 74, 1965, pp ; Gilbert Harman, Change in View. Cambridge: MIT Press, 1986; William Lycan, Judgment and Justification. Cambridge: Cambridge University Press, 1988; Peter Lipton, Inference to the Best Explanation. London: Routledge, This more general formulation is needed if future events don t explain the occurrence of past ones, or if one effect of a common cause does not explain the other. An exception to this pattern is Paul Humphreys, The Chances of Explanation. Princeton University Press, 1989, which provides a sustained discussion of the idea of an explanatory factor. Or more precisely one proposition can be obtained from the other by uniform substitution of names. Paul Oppenheim and Hilary Putnam, Unity of Science as a Working Hypothesis, in H.Feigl, G.Maxwell, and M.Scriven (eds.), Minnesota Studies in the Philosophy of Science, Minneapolis: University of Minnesota Press, 1958, pp Hilary Putnam, Psychological Predicates, In W.Capitan and D.Merrill (eds.), Art, Mind, and Religion. Pittsburgh: University of Pittsburgh Press, Reprinted as The Nature of Mental States in Mind, Language, and Reality, Cambridge, England: Cambridge University Press, 1975; Hilary Putnam, Philosophy and our Mental Life, In Mind, Language, and Reality. Cambridge, England: Cambridge University Press, Wesley Salmon, Scientific Explanation and the Causal Structure of the World. Princeton: Princeton University Press, There is another reason to reject screening-off as an explication of the concept of explanatory relevance. Suppose Moriarty s death is caused by Holmes shooting him through the heart. Presumably, the shooting is explanatorily relevant. However, the fact that Holmes shot Moriarty is screened off from Moriarty s death by a weaker, disjunctive specification of all the many events that would have killed Moriarty if they had occurred. For example, given that either Holmes shot Moriarty through the heart or Watson poisoned him with cyanide, Moriarty was bound to die, regardless of whether Holmes in fact poisoned him. David Lewis, Causal Explanation, In Philosophical Papers, vol. 2, Oxford University Press, 1986, pp Jaegwon Kim (in Supervenience and Mind, Cambridge: Cambridge University Press, 1993) argues that countenancing both micro- and macro-causal explanations of the same event com-

220 TWO USES OF UNIFICATION mits one to an objectionable form of overdetermination. However, overdetermination is usually understood as the occurrence of two causes that are independent of each other, with each sufficient for the effect. In this sense, Moriarty s death is overdetermined, if Holmes and Watson each simultaneously fire a bullet through his heart. Applying this point to the question of mental and physical causation that Kim considers, we should conclude that there is no genuine overdetermination here; if the macro-facts supervene on the micro, then micro and macro are not independent. See Elliott Sober, Physicalism from a Probabilistic Point of View, Philosophical Studies 95, 1999b, , for discussion. Elliott Sober, The Multiple Realizability Argument against Reductionism. Philosophy of Science 66, 1999, Unlike Putnam, Philip Kitcher (in Explanatory Unification and the Causal Structure of the World. In P. Kitcher and W. Salmon (eds.), Scientific Explanation. Minneapolis: University of Minnesota Press, 1989, pp ) does provide an account of explanation, rather than simply relying on an intuitive understanding of what explanatory relevance is. Kitcher rejects the idea that explanations must be causal and uses the idea of unification as his central idea. I won t discuss Kitcher s proposal here except to note that it seems not to deal adequately with the problem of explanatory asymmetry; on this, see Eric Barnes Explanatory Unification and Scientific Understanding, Proceedings of the Biennial Meetings of the Philosophy of Science Association, 1, 1992, pp. 3-12, and Todd Jones, How the Unification Theory of Explanation Escapes the Asymmetry Problem, Erkenntnis, 43, 1995, pp In an earlier publication, Kitcher (in 1953 and All That A Tale of Two Sciences. Philosophical Review, 93, 1984, pp ) used Putnam s peg and board style of reasoning to argue that Mendelian genetics does not reduce to molecular biology, but without the apparatus of the unification theory of explanation that Kitcher later developed. Hirotugu Akaike, Information Theory as an Extension of the Maximum Likelihood Principle. In B. Petrov and F. Csaki (eds.), Second International Symposium on Information Theory. Budapest: Akademiai Kiado, 1973, pp See also Kenneth Burnham and David Anderson, Model Selection and Inference a Practical Information-Theoretic Approach. New York: Springer, 1998; Malcolm Forster and Elliott Sober, How to Tell when Simpler, More Unified, or Less Ad Hoc Theories will Provide More Accurate Predictions. British Journal for the Philosophy of Science 45, 1994, pp. 1-36; Elliott Sober, Parsimony, Instrumentalism, and the Akaike Framework, Philosophy of Science, 2002, forthcoming. To see why, consider a simpler problem: what are the maximum likelihood values for the parameters in (U) and (D) that pertain to a single category of smoker? For example, suppose that the frequency of lung cancer among those who have smoked two packs a day for thirty years is for those who have smoked X-cigarettes exclusively and is for those who have smoked Z-cigarettes exclusively. If these two types of smoker are equally represented in the sample, L(U) will say that the probability of cancer for such individuals is , while L(D) will say that the probability of cancer for the X-smokers is and the probability for the Z- smokers is The data are more probable according to the latter hypothesis. The formulation of Akaike s result that Forster and Sober (1994) recommend is that an unbiased estimate of the model s predictive accuracy per datum is (1/N){Log-likelihood[L(M)] - k}, where N is the number of data. Here I m glossing over the problem of how facts about type- and token-causation are related, on which see Elliott Sober, Two Concepts of Cause, in P. Asquith and P. Kitcher (eds.), PSA 1984 Proceedings of the Philosophy of Science Association, E. Lansing, Michigan, vol. 2, pp The fact that Smith s smoking increased his risk of getting lung cancer does not mean that his smoking token-caused the cancer; perhaps a micro-examination of Smith s lungs would reveal that his cancer was due to asbestos exposure, not smoking. The statistical inference problem described above concerns claims about type causation, whereas the explanation of Smith and Jones lung cancer presumably calls for facts about token causation.

221 216 ELLIOTT SOBER Department of Philosophy University of Wisconsin 5185 Helen C. White Hall Madison, Wisconsin U.S.A. and London School of Economics and Political Science Houghton Street WC2A 2AE London UK

222 CHRISTOPHER HITCHCOCK UNITY AND PLURALITY IN THE CONCEPT OF CAUSATION It has become common in the scientific realism literature to distinguish between the empirical and super-empirical virtues of scientific theories. 1 Empirical virtues include predictive accuracy, fit with data, and so forth. The super-empirical virtues include unity, simplicity, mathematical elegance, explanatory power, and so forth. Both types of virtue are believed to play a role in theory choice. In this essay, I wish to focus on the virtue of unity. Conceptions of unity do not themselves form a unity. When Otto Neurath championed the unity of science movement, 2 and when Putnam and Oppenheim championed the unity of science as a working hypothesis, 3 they hardly meant the same thing by the unity of science. The kind of unity I have in mind is the comprehending of a maximum number of facts and regularities in terms of a minimum of theoretical conceptions and assumptions. 4 It is in this sense that Newton s gravitational theory unifies such diverse phenomena as the ebb and flow of the tides, the acceleration of balls down inclined planes, the periods of pendula, the orbits of the planets, and the results of the Cavendish torsion experiments. Perhaps the best theoretical account of this notion of unity is provided in Kitcher s unificationist theory of explanation. 5 This type of unity is often taken to underwrite theory choice. The story so far is not without its difficulties. Elliott Sober s contribution to this volume 6 provides a powerful argument that unity is not a super-empirical virtue after all, but an empirical one, since unity can enhance predictive accuracy. In her recent book, Unifying Scientific Theories: Physical Concepts and Mathematical Structures, 7 Margaret Morrison argues that unity rarely does play the sort of role in theory choice that it is often supposed to play. I will let this pass, for my concern here is not with the role of unity in scientific theory choice, but in philosophical theory choice. When evaluating a philosophical theory, such as act utilitarianism as a theory of right and wrong actions, it is also fairly natural to make a distinction akin to that between empirical and super-empirical virtues. Philosophical theories are often evaluated by comparing them with intuitions : the intuition that it is wrong to kill an innocent person in order to save the lives of two others is often taken to provide evidence against act utilitarianism. These intuitions play a role analogous to observations in scientific theory appraisal. On the other hand, we also evaluate philosophical theories on the basis of other non-intuitive factors, such as clarity, simplicity, and patent relevance to the topic in question. On these scores, act utilitarianism arguably scores well: it is simple, clear (in theory, if not in practice), and the promotion of happiness and prevention of unhappiness 217 F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

223 218 CHRISTOPHER HITCHCOCK seems directly connected to the ends of ethical theorizing. In evaluating philosophical theories, we trade off these different virtues in an attempt to reach what Rawls has called a reflective equilibium. 8 The status of the intuitions that serve as a foundation of philosophical theorizing is unclear. What are intuitions and where do they come from? Are they really analogous to sensory perceptions? Are they modular, or are they deeply theory-laden? Without providing detailed answers to these questions, it seems clear that reliance on intuitions in philosophical theory evaluation is much more problematic than the reliance on empirical observations in scientific theory evaluation. When it comes to the evaluation of philosophical theories, then, the non-intuitive virtues can be expected to play a substantial role. In particular, it seems that unity should play a prominent role in philosophical theory choice. In what follows, I will consider the possibility of relying on unity as a criterion for evaluating philosophical theories of the concept of causation. I strongly suspect that the morals to be drawn here carry over to other philosophical subjects. The sort of unity described by Feigl and Kitcher, the sort of unity that is often thought to play a role in scientific theory choice, really involves both a unity and a plurality: a maximum number of facts and regularities are to be accommodated by a minimum of theoretical conceptions and assumptions. Newton s theory is unified because it is able to bring a plurality of diverse phenomena under one theoretical treatment. Without this plurality tides, pendula, planetary orbits, and so on there could be no unification; there would be nothing to unify. We have a unity at the level of theoretical treatment, but a plurality at the level of phenomena treated. The situation we currently find in the literature on causation exhibits the opposite pattern. There is a plurality of theoretical perspectives on the nature of causation: regularity theories, 9 probabilistic theories, 10 counterfactual theories, 11 mark transmission theories, 12 and others. At the same time, there is a unity at the level of the phenomena to be comprehended. The different theoretical perspectives are all vying for the privilege of analyzing one specific phenomenon: causation. Of course, this sort of situation is not unknown in science. In the years following 1915, General Relativity vied with modified classical theories for the privilege of explaining gravitational phenomena. This sort of pattern is typical of periods of scientific revolution, 13 when there is both a lack of agreement upon a scientific paradigm, and a disproportionate amount of attention devoted to the small range of phenomena that threw the prevailing paradigm into a state of crisis. But scientific revolutions, unlike many parallel philosophical debates, do resolve themselves, and theoretical unity is thought to be one of the criteria that determines the outcomes of such revolutions. The situation in the current literature on causation is no doubt familiar to many readers. Indeed, some might wonder how it could be otherwise. I hope to suggest that it could and should be otherwise by reviewing some examples of unification in the theory of causation. Two of the prominent theoretical

224 UNITY AND PLURALITY IN THE CONCEPT OF CAUSATION 219 approaches to causation mentioned above the probabilistic and mark transmission theories are given their first clear exposition in Reichenbach s The Direction of Time. 14 This seems odd, since these two theories are typically treated as competitors. Salmon 15 considered it necessary to present a systematic critique of probabilistic approaches to causation in order to motivate the need for his own process theory. Yet Reichenbach champions both theories at once. How is this possible? The answer is that Reichenbach attempts a unification of the two theories. The central definition in his mark transmission theory is the following: Definition 1. If a mark made in an event shows in an event then is 16 causally relevant to The central definition of the probabilistic theory is: Definition 2. An event is causally relevant to a later event if and there exists no set of events which are earlier than or simultaneous with such that this set screens off from 17 (The set of events screens off from just in case, for each Note that these are both definitions of the same relation: causal relevance. Reichenbach proves that these two definitions are equivalent, given a certain set of assumptions that he takes to be empirically true. These assumptions are: Assumption 1. If a mark made in shows in then Assumption 2. If a mark is made in then either or Assumption 3. If screens off from and if a mark made in shows in then it also shows in Assumption 4. If a set screens off from and if a mark made in shows in then it also shows in at least one of the events 18 is the event that results from applying a mark to I think that there are good reasons to doubt that these assumptions are correct; Assumption 2 strikes me as especially dubious. The idea seems to be this: either is causally relevant to or it is not. If it is not, then marking will make no difference to the probability that a mark is present at so we will have On the other hand, suppose that is causally relevant to then, there is some fixed probability (presumably high) that whatever structural features are present at will also be present at That is, there is some fixed probability that structure will be preserved. It is this assumption that strikes me

225 220 CHRISTOPHER HITCHCOCK as implausible. Let and be states of a chalkboard, separated by some period of time. If the chalkboard is not marked with chalk at the earlier time it is very unlikely that a chalk mark will spontaneously appear during the intervening period. Thus will very close to one. On the other hand, if the chalkboard is marked at the earlier time, there may will be some intermediate probability (neither very close to zero nor very close to one) that the chalk will dissipate during the intervening period. Thus while in violation of Assumption 2. My concern, however, is not with the details of Reichenbach s argument, but with his overall strategy. Instead of treating the probabilistic and mark transmission theories as philosophical rivals, he sought to unify them. He provided suitably formalized versions of the two theories, and a set of sufficient conditions for their equivalence. There are other examples of this strategy. Hume famously defined a cause to be both an object precedent and contiguous to another, and where all the objects resembling the former are plac d in a like relation of priority and contiguity to those objects, that resemble the latter and an object precedent and contiguous to another, and so united with it in the imagination, that the idea of the one determines the mind to form the idea of the other, and the impression of the one to form a more lively idea of the other. l9 These two seemingly disparate definitions are brought together by Hume s associationist psychological theory. By observing the constant conjunction of the two objects, the mind acquires the habit of making the transition from the impression of the first to a lively idea of the second; hence those pairs of objects that participate in the sort of constant conjunction described in the first definition will give rise to the sort of conditioned expectation described in the second. Of course, these two definitions need not be extensionally equivalent: there could be constant conjunctions that go unobserved, or conjunctions that remain constant in our experience but variable outside of our experience. Brian Skyrms 20 offers a partial unification of probabilistic and counterfactual theories of causation. According to recent probabilistic theories of causation, 21 C counts as a cause of E just in case for all 22 where is a partition of the outcome space into cells representing causally homogeneous background conditions. According to Lewis s counterfactual theory, a sufficient (but not necessary) condition for C to cause E is that the latter counterfactually depend upon the former: if C had not occurred, E would not have occurred. 23 (Lewis identifies causation with the transitive closure of counterfactual dependence.) One standard semantics for counterfactual conditionals is provided by Robert Stalnaker. 24 This semantics includes a set of possible worlds, and a selection function S that takes as arguments one possible world and one proposition. Intuitively, S(w, F) is the closest possible world to w in which F is true. The counterfactual if C had not occurred, then E would not have occurred is thus true just in case is true in is the actual world, O(C) is the proposition that C occurs, and O(E) is the propo-

226 UNITY AND PLURALITY IN THE CONCEPT OF CAUSATION 221 sition that E occurs. Skyrms demonstrates that a selection function induces a family of partitions on the set of possible worlds precisely the sorts of partition that figure in probabilistic theories of causation and (subject to certain further constraints) vice versa. This result provides confirmation for the sneaking suspicion that at root, the two approaches amount to much the same thing. Daniel Hausman, in his Causal Asymmetries 25 provides careful re-formulations of a variety of theories of causation, with particular attention to their treatment of the directionality of causation. He also formulates a number of principles, most importantly the Independence Theory of Causal Priority, or CP. Hausman shows that the various theories of causation either presuppose CP, or that their (partial) success can be explained in terms of CP. In this way, Hausman argues that CP provides the common core of diverse theories of causal asymmetry. This brief list four examples may convey the impression that I could continue ad nauseam. I cannot. I provide these examples because they remain very much the exception rather than the rule. Despite the importance and interest of exploring the systematic relationships between superficially competing theories of causation, very few have chosen to undertake this sort of project. One reason may be that this sort of project tends to require formalization (Hume being the exception here), and many philosophers lack the formal facility to carry it out. Another may be a general shift in philosophical focus since the heyday of logical empiricism. The logical empiricists, especially Carnap, placed great value on exploring the logical interrelationships among various concepts and theoretical posits. Since then, there has been a shift in philosophical theorizing toward straight intuition-fitting. I have made the case that the plurality of theories of causation cries out for unification. But there is another side to this coin. Most philosophers view the various theories of causation as rivals vying for the privilege of analyzing one concept: causation. Of course, in trying to accommodate our causal intuitions about a diverse range of cases, theories of causation are trying to unify a plurality: they are trying to find the common element that makes all of these cases instances of causation. But I wish to argue that there is a greater plurality here, and hence a greater opportunity for unification, than philosophers have typically recognized. 26 This is a conclusion for which I have argued at length elsewhere, 27 so I will let one illustration suffice. The example, due to Germund Hesslow, 28 concerns the effect of oral contraceptives on the chances of thrombosis. Thrombosis is one of the most dangerous potential side effects of birth control pills. Yet pregnancy itself is an even greater risk factor for thrombosis; since birth control pills are very effective in preventing pregnancy, use of oral contraceptives may lower the overall probability of thrombosis. 29 Hesslow presents this as a counterexample to probabilistic theories of causation: birth control pills cause thrombosis, even though they lower the probability of thrombosis.

227 222 CHRISTOPHER HITCHCOCK I suggest a different diagnosis. There is a legitimate sense in which birth control pills both cause and prevent thrombosis. Birth control pills prevent pregnancy, which in turn is a risk factor for thrombosis. This constitutes a causal route along which the effect of birth control pills is to prevent thrombosis. But there is an independent route whereby birth control pills cause thrombosis. This route can be detected statistically by controlling for pregnancy: among those women who do become pregnant, the chance of thrombosis is higher for those women who have taken birth control pills; and among those women who do not become pregnant, the chance of thrombosis is also higher for those women who have taken birth control pills. It is because of the existence of such a route that we are justified in saying that oral contraceptives cause thrombosis. Nonetheless, the net effect of birth control pills, taking into consideration all of the causal routes, is to prevent thrombosis. More generally, when we say that C causes E, we may be saying that there exists a causal route between C and E whereby the effect of C is to promote E's occurrence; or we may be saying that the net effect of C is to promote E's occurrence. This plurality of causal notions is different from the sort of plurality of causal notions described by Elliott Sober 30 and Ned Hall. 31 These authors argue that there are different concepts of causation that are captured by different theories of causation. The different notions of causation described by these authors do not stand ready to be unified; they rather embody distinct and incompatible conceptions of the nature of causation. (Of course, even if these causal concepts require different causal theories, they may become unified through the unification of those theories themselves.) By contrast, the two causal concepts described in the previous paragraph, causation along a particular route and net causal effect, can readily be accommodated within the same theoretical framework. Consider, for example, the probabilistic approach to causation. As we noted above, according to most versions of this theory, C counts as a cause of E when for all in the partition Cells of this partition correspond to causally homogeneous conditions in which other causally relevant factors are held fixed. For instance, in evaluating the causal relevance of birth control pills for thrombosis, we must hold fixed factors such as a woman s age, and whether or not she smokes. Should we also hold fixed whether or not she becomes pregnant? Yes and no. If we do not hold this further factor fixed, then we are evaluating the net effect of birth control pills on thrombosis. The conditional probabilities that we are comparing, in this case, reflect the influence of birth control pills on thrombosis along all of the available causal routes. If we do hold fixed whether or not a woman becomes pregnant, we evaluate only the contribution of birth control pills to thrombosis along the route that bypasses pregnancy. By holding fixed whether or not pregnancy occurs, we screen off any effect that birth control pills might have by preventing pregnancy. This suggests a further strategy for unification. Instead of defending a theory of causation by showing how it matches our intuitions about causation simpliciter, we could identify different causal concepts, and show how they can all

228 UNITY AND PLURALITY IN THE CONCEPT OF CAUSATION 223 be accommodated within the same theoretical framework. To the extent that a theory can informatively characterize the similarities and differences among our various causal concepts, more credit should redound to the theory. NOTES See, for example, Paul Churchland, The Ontological Status of Observables: In Praise of the Superempirical Virtues, in Paul Churchland/Clifford Hooker (eds.), Images of Science. Chciago: University of Chicago Press 1985, pp See, for example, Otto Neurath, Unified Science as Encyclopedic Integration, in: Otto Neurath/Charles Morris (eds.), Foundations of the Unity of Science, Toward an International Encyclopedia of Unified Science, vol. 1. Chicago; University ofchicago Press 1938, pp Hilary Putnam and Paul Oppenheim, Unity of Science as a Working Hypothesis, in: Michael Scriven/Herbert Feigl/Grover Maxwell (eds.), Concepts, Theories, and the Mind-Body Problem. Minneapolis: University of Minnesota Press 1958, pp Herbert Feigl, The Orthodox View of Theories: Remarks in Defense as Well as Critique, in: M. Radner/S. Winokur (eds.), Minnesota Studies in the Philosophy of Science, Volume IV. Minneapolis: University of Minnesota Press 1970, p. 12. Philip Kitcher, Explanatory Unification, in: Philosophy of Science 48, 1981, pp Elliott Sober, Two Uses of Unification, in: Friedrich Stadler (ed.), Vienna Circle Institute Yearbook Dordrecht/London/Boston: Kluwer Academic Publishers Cambridge/New York: Cambridge University Press John Rawls, A Theory of Justice. Cambridge: Harvard University Press 1971, p. 20. More precisely, reflective equilibrium refers to the process wherein we allow our intuitions to be modified in light of theoretical considerations, then adjust our theory to accommodate the new intuitions, and so on until a fixed point is reached. E.g. John Mackie, The Cement of the Universe. Oxford: Oxford University Press 1974, chapter 3. E.g. Patrick Suppes, A Probabilistic Theory of Causality. Amsterdam: North-Holland, 1970; Ellery Eells, Probabilistic Causality. Cambridge: Cambridge University Press E.g. David Lewis, Causation, in: Journal of Philosophy 70: E.g. Wesley Salmon, Scientific Explanation and the Causal Structure of the World. Princeton: Princeton University Press The terminology for this paragraph is of course taken from Thomas Kuhn, The Structure of Scientific Revolutions. Chicago: University of Chicago Press Berkeley and Los Angeles: University of California Press Salmon, op. cit., chapter 7. Reichenbach, op. cit., p Ibid., p. 204, with minor changes in notation. Ibid., pp , with minor changes in notation. David Hume, A Treatise of Human Nature. Second edition, Oxford: Oxford University Press, 1978, p Pragmatics and Empiricism. New Haven: Yale University Press 1984, pp E.g., Eells op. cit. Although see my Causal Generalizations and Good Advice, in: Monist 84, 2001, pp , for discussion of this contextual unanimity condition. Two further requirements must be added: C and E must be events that actually occurred, and they must be distinct from one another. A Theory of Conditionals, in: Nicholas Rescher (ed.), Studies in Logical Theory. Oxford: Blackwell 1968, pp Lewis s preferred semantics is somewhat more complex than Stalnaker's.

229 224 CHRISTOPHER HITCHCOCK Cambridge: Cambridge University Press This has been less of a problem for researchers in areas outside of philosophy. In his important new book Causality: Models, Reasoning, and Inference (Cambridge: Cambridge University Press 2000), the computer scientist Judea Pearl presents definitions of a wide variety of causal concepts: causal effects, total effects, direct effects, actual causes, contributing causes, and so on. Christopher Hitchcock, A Tale of Two Effects, in: The Philosophical Review 110, 2001, pp ; Of Humean Bondage, in: The British Journal for the Philosophy of Science, forthcoming. Discussion:Two Notes on the Probabilistic Approach to Causality, in: Philosophy of Science 43, 1976, Whether it does or not will depend upon a variety of other factors: the age of the woman in question, whether or not she smokes, how likely she is to become pregnant in the absence of birth control use, and so on. Two Concepts of Cause, in: Peter Asquith/Philip Kitcher (eds.), PSA 1984, Vol. II. East Lansing: Philosophy of Science Association 1985, pp Causation and the Price of Transitivity, in: Journal of Philosophy 97, 2000, California Institute of Technology Pasadena CA U.S.A. cricky@caltech.edu

230 DIEDERICK RAVEN EDGAR ZILSEL S RESEARCH PROGRAMME: UNITY OF SCIENCE AS AN EMPIRICAL PROBLEM INTRODUCTION The unity of science movement was itself far from unified. 1 There may have been unity on the rallying call for a unity of science but that is as far as it went. Not only was there disagreement among the main protagonists on what was meant by the unity of science, but also on how to achieve it. In this paper I shall deal with Edgar Zilsel s ( ) conception. It represents an interesting break with the more programmatic approaches of Carnap, Neurath, c.s.; Zilsel emphasizes the need for an empirical demonstration of the unity of science, which has interesting meta-philosophical implications. 2 The unity of science is not just central to his research programme; it is what constitutes the latter s unifying principle, although this is far from evident if one takes a look at the historical essays he published in exile in the USA during the 1940s. (These important essays on the emergence of science, well known among historians of early modern science, have given rise to the so-called Zilsel Thesis, which holds that modern science came into being when, between 1300 and 1600, the social barriers between those who labored with there minds and tongues, i.e. the university scholars and the humanists, and those who worked with their hands, i.e the superior artisans, eroded because of the rise of free-enterprise capitalism. 3 ) But simultaneously, he also published a couple of smaller and far less known essays, directed against Southwest-German Neo-Kantianism (Rickert, Windelband), Dilthey s philosophy of life, and interpretative sociology (Max Weber, W. Sombart, G. Simmel, R. Stammler, Alfred Weber, etc.). His main argument was that philosophers of cultural science and the humanities had a false understanding of natural science. Because of this false understanding, they erroneously postulated a fundamental methodological difference between the natural sciences and the human sciences. According to Zilsel, this difference does not exist. In his view the historian is basically confronted with the same problems as the natural scientist. If the science of history is not compared with classical mechanics but with, for example, geophysics that is, the physics of earthquakes, sea-currents, volcanology, and meteorology one will arrive at the conclusion that historical phenomena are hardly more difficult to predict than the weather, and certainly no more difficult than earthquakes and volcanic 225 F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

231 226 DIEDERICK RAVEN eruptions. What would scientists think of a geophysicist who abandoned the search for geophysical laws because of their inexactness? (SOMS: 202) At first glance, these two sets of essays do not seem to have much in common. Closer investigation of Zilsel s life and work, however, reveals an inner connection at least in Zilsel s intention. This inner connection is what I call Zilsel s empirical problem of the unity of science: showing that law-like statements in the socio-historical sciences are indeed possible. In order to show this, I shall discuss in some detail Zilsel s earlier work and the theoretical development through which Zilsel went. This will allow me to explain what the idea of the unity of science meant to Zilsel. THE APPLICATION OF NATURAL SCIENTIFIC TO THE HUMANITIES METHODS Zilsel s first book, The Application Problem (1916), attempts to solve the philosophical problem of applying statistics to an irrational nature. The main topic of the book is a riddle posed by the so-called law of large numbers. The law states that with a large number of repeated throws of a chance game... the relative frequency almost equals the mathematical probability. 4 Nature, however, could be rather different. She could produce frequencies quite different from the expected result. It is therefore only at first glance trivial to ask why the law of large numbers is applicable at all. Zilsel construed this problem as being part of a wider one: how can rational mathematical constructions be applied to a vague and irrational nature? This is what Zilsel termed the application problem. His analysis of the epistemological and ontological status of the law of large numbers led to the conviction that there are general philosophical problems that are related to all sciences but not solvable within any one specific science. In opposition to the proponents of logical positivism within the Vienna Circle, Zilsel believed that the discussion of such problems was fruitful and should not be denounced as consisting of metaphysical Scheinprobleme. At the same time, however, he did not believe in the capacity of philosophy to solve fundamental problems independently of empirical research and distrusted philosophy as an independent discipline. Zilsel was outspoken in his desire to unite these fundamental philosophical problems with the contemporary problems presented in empirical research and despised all attempts by schoolmasters... who would separate... philosophy from the empirical disciplines. 5 His position is that at present [it is] only possible to fruitfully discuss philosophical problems in... intimate relation to living science. 6 The interaction between the sciences and classical philosophy of the century is his prime example. Accordingly, Zilsel began to study the application problem not as a methodological problem but in the context of the statistical analysis of history and culture. How is the nature of society to be perceived if it is subordinate to the law of large numbers? Zilsel s second major book, On the Development of the

232 EDGAR ZILSEL S RESEARCH PROGRAMME 227 Concept of Genius (1926), presents a sociological analysis of the genius personality cult which dominated Renaissance humanism and, in Zilsel s eyes, continued to pervade the ideals of his time. Though it focussed on the Renaissance, the book aimed to find and explain the empirical laws which govern the development of the concept of genius in different periods and cultures. Zilsel s commitment to this project and its methodology were at stake when the University of Vienna denied him the right to teach at the professorial level by refusing this work. I shall not discuss here in any detail why Zilsel s Habilitation was rejected and his academic career torpedoed. Though much ideological politics was involved here, it still is possible to say that the philosophers Reininger and Meister the most outspoken critics of the Zilsel MS did have a point when they complained that the MS they were presented with has nothing to do with philosophy 7, for it was to a large extent a philological and historical analysis of the genealogy of the concept of genius supplemented by a set of law-like statements specifying under which socio-economic and institutional conditions the concept of genius could develop. Even today in these post-foucaultian times the MS would not easily qualify as a thesis in philosophy. Zilsel s fight for the acceptance of his Habilitationsschrift was lost in a dispute over what constitutes the subject matter of philosophy proper. Zilsel had outspoken ideas about precisely this point, and he was by no means willing to compromise. When Schlick at one point asked Zilsel to withdraw the MS and submit a new, more philosophically oriented Habilitationsschrift, Zilsel firmly refused: my interests, in recent years, have turned toward the application of natural scientific methods to the humanities as well as toward the disclosure of fairly exact laws concerning the events in these fields. This area of research shall be occupying my mind for a considerable period of time, yet I have already collected rather extensive material, especially regarding the history of the concept of genius. The results of the finished parts are formulated with respect to this material. How long the final completion of the other parts will occupy me, I, of course, cannot say today. I could not, however, justify to myself that the direction of my scientific work be influenced by any considerations other than by the problems themselves, my interest in them, and my previous work. 8 However, Zilsel withdrew his application for the venia legendi in a letter to the Dean on the third of November 1924 and refused to present another manuscript which belonged to the most narrow definition of philosophy, as had been requested by the evaluating committee. At whatever cost to his career, Zilsel would stubbornly champion his understanding of philosophy. In his letter to the Dean, he writes: I have rather tried to develop my philosophy of nature and history with the help of physical and historical factual material in the hope of serving philosophy better than I would were I to cut her off from the fruitful ground of the individual sciences. It appears very improbable that I will fundamentally change my method in completing my planned philosophical work... I thus withdraw my application. 9

233 228 DIEDERICK RAVEN The substance of the controversial work was published in 1926 as The Development of the Concept of Genius. At the end of the book, a final result is formulated: Laws on the concept of genius. The structure of these laws is a thesis-like summary in which all information on spatial and temporal locations is omitted so as to attain the character of a general if-then proposition. The scientific validity of these laws rested upon their testability through cross-cultural comparisons. They were meant as provisional scientific hypotheses, to be confirmed, rejected, or modified through research. Zilsel considered the analysis of renaissances in different cultures and times as a suitable field for testing the laws. It is exactly this point that offended his enemies. Zilsel s method carried a dramatic shift of perspective in its wake: People, professional work, and cultural periods would lose their uniqueness and come to be seen as mere variables in a temporal development. Although the resulting laws only barely met the minimum theoretical requirements, they were of paradigmatic significance to Zilsel for establishing a socio-historical science that would be fueled by cooperation and apply comparative cross-cultural methods. In his essay Physics and the Problem of Historico-Sociological Laws of , he pursued this topic and provided examples of the kinds of laws he had been looking for in his book on The Development of the Concept of Genius. UNITY AS AN EMPIRICAL PROBLEM Zilsel s attempt to combine philosophical analysis with detailed historical research caused not only his Habilitation to falter but, to a certain extent, it also distanced him from the central figures of the Vienna Circle. In a letter to Reichenbach, Zilsel objected to the content-free methodology and logic of science as practiced today. 11 His target was explicitly, though not exclusively, the Vienna Circle members practice of writing abstract philosophy. For Zilsel, the Vienna Circle could be ironically characterized as an empiricist school without empirical research. This critical evaluation of the methodology of the Vienna Circle is particularly clear in two book reviews: Max Adler s Textbook of the Materialist Conception of History (1930) is criticized by Zilsel for having presented only three concrete examples from real history in the entire book. 12 In a review of Otto Neurath s book Empirical Sociology, Zilsel noted that the book has no intrinsic interest in the living content of sociology in any case, a lot less than in the promotion of the basic logical ideas of the Vienna school of philosophy. Thus, in this empirical sociology, fertile empiricism withdraws behind logic. 13 Zilsel considered himself an exception, in terms of epistemological discourse as well as due to his question as to whether history is a potential field for studying laws. His dislike of abstract philosophical argument and his belief in accurate case studies do not imply any serious doubts about the validity of the basic ideas of logical empiricism. Zilsel most certainly did agree with them. But in his own

234 EDGAR ZILSEL S RESEARCH PROGRAMME 229 historical research, Zilsel had come to realize how difficult the task of linking the social sciences with the natural sciences was. This made him sceptical about the usefulness of formulating a programme of unity as opposed to generating unity by means of empirical research. Zilsel did not believe that a programme based on logical analysis and language construction as elaborated for example by Carnap could help in uniting the social and natural sciences. He did not believe that much was to be gained by arguing in principle, i.e. by utilizing solely philosophical arguments, that in the socio-historical sciences law-like statements are not only feasible but also should be the main goal of research. He clearly felt that simply arguing for the emptiness of the anti-naturalist stance of the followers of Dilthey, Rickert, c.s would not and could not convince. 14 What if they were right that law-like explanations could not be found in the socio-historical sciences? Zilsel believed the anti-naturalists rebellion to be a major threat to the unity of science. But he also believed that the burden of proof lay upon the naturalists. The anti-naturalists could only be defeated by producing the equivalent to what in mathematics is known as a existence proof, i.e. by coming up with examples of law-like explanations in the socio-historical sciences. This is what I call Zilsel s empirical problem of the unity of science. STATISTICAL LAWS But there is a problem looming large here. Without an elucidation of the concept of law, a few examples of law-like statements in the socio-historical sciences would not prove very much. Zilsel s empirical problem of the unity of science does bring with it the need for a theoretical study of the concept of law. And this is exactly what he works on beginning in the late twenties. 15 I have not come across a statement by Zilsel in which he explains where he is looking for the needed theoretical clarification of the concept of law. Still, I believe a reasonable guess can be made as to what his line of thinking is here. As my starting point I take an essay he published in 1932 in Der Atheist. 16 In that essay, Zilsel discusses the developments within the physical sciences such as the work of Ludwig Boltzman on the application of statistical methods to physics as well as the introduction of relativity theory and quantum mechanics. These innovations suggested to him that the central ideas of the mechanical world view as formulated in the seventeenth century were no longer plausible. Because the concept of law was central to the emergence of the mechanical world view, its tenability was implicated as well. That the concept of law was as much in need of re-thinking as the rest of the foundations of physics was due to the revolutionary developments going on at the time. Zilsel s position here is that just as the macro-laws of physics are statistical in character, so are the socio-historical laws. 17 Confirmation of this line of interpretation can be found in Zilsel s 1942 essay Problems of Empiricism, which he wrote as his contribution to the Interna-

235 230 DIEDERICK RAVEN tional Encyclopedia of Unified Science. 18 This essay is, in my reading, the most articulate statement on the confluence of his two projects. It is a comprehensive analysis of the development and meaning of the concept of law, parallelled by an investigation into the social origins and development of modern science. Summarizing the results of his earlier essays, Zilsel describes how the experimental method and the concept of law formed the mechanistic metaphysics of a necessary and complete inner-connectedness of all natural phenomena. Three characteristics are central to mechanistic metaphysics: (a) differentiation of reality into the real world of mechanical laws and the world of appearance; (b) reduction of all empirical laws to mechanical laws; (c) an introspective psychology of knowledge. The concluding section ( The Decline of the Mechanical Conception of Nature ) describes how these characteristics fell apart and gave way to a new concept of law, a non-reductionist view of reality, and an epistemology no longer based on psychology. (199). The essay also deals with the issue of why the human sciences followed a separate path of development. Zilsel discusses the various origins of psychology, political science, history, economics and sociology. He connects the practical needs to which they responded to their research ideals. None of them, at least in their early stages, were seriously interested in the experimental method and in causal laws, but were oriented toward rational norms, the progress of humanity, and class interests. Zilsel also touches upon the sociological schools which deny the possibility of any sociological laws. They maintain that in social research, causes and laws have to be replaced by types, by understanding that is, empathy, by wholeness, entelechies and values (194). He calls this position a rebellion against causality and summarizes all of the objections he had developed in his earlier essays. The differences between the origins of science and those of the humanities provide a causal explanation for their different methods and ideals. His project on the social origins of modern science explains the origins of science and the non-participation of the humanities in its development. Still, the anti-naturalists, defenders of hermeneutics, and introspectionists could be right in denying a common epistemology to all sciences grounded on the concept of law. That is the reason why the very idea of the unity of science is at stake if the concept of law is not applicable to the humanities. As long as the mechanistic world view dominated the concept of law, the perspectives for theory development in the humanities on the basis of empirical laws would have been rather ridiculous one could know in advance that the gulf between the fundamental concepts of mechanics and the theoretical terms of culture, history, and society could not be bridged by any sort of reductionism. Zilsel s opponents thus did have a point in defending themselves against reductionism.

236 EDGAR ZILSEL S RESEARCH PROGRAMME 231 TO BE INVESTIGATED SOCIOLOGICALLY AND CAUSALLY Seen against this background, the relationship between the two projects that constitute the core of Zilsel s empirical problem of unity is that the project on the social origins of modern science is the case study for the project on the concept of law. The kind of laws that Zilsel is looking for in his historiographical project are of a statistical nature which explains why he is presenting his material a number of times in the form of long lists of sociological characteristics of the people he is studying. This interpretation squares with the only early statement on his project on the social roots of modern science I have been able to find. When Reichenbach asked Zilsel to contribute an essay to the new journal Erkenntnis, Zilsel replied in a positive way: I would most like to work out for print a talk I recently gave at the Wiener kulturwissenschaftliche Gesellschaft (Society for Cultural Studies, Vienna). The work is directed toward exact natural science in a twofold sense: firstly, because it considers historical processes themselves as natural processes and attempts to connect them by statistical laws; and secondly, because it treats the origins of the exact sciences, while it understands the so called humanities as they are practiced today as remnants of a prescientific age. 19 In this early statement from 1930, both projects are intimately bound together. The quotation, furthermore, indicates how he planned to relate statistical analysis with an epistemological claim and a broader sociological explanation. To show that Zilsel still held on to these ideas later in life, I shall quote two longer passages from essays published while he was in the USA. The first concerns the use of the comparative method in historical sociology. Especially noteworthy is how Zilsel links this comparative method to finding and verifying causal explanations. The rise of science is usually studied by historians who are primarily interested in the temporal succession of the scientific discoveries. Yet the genesis of science can be studied as a sociological phenomenon, too. The occupations of the scientific authors and of their predecessors can be ascertained. The sociological function of these occupations and their professional ideals can be analyzed. The temporal succession can be interrupted and relevant sociological groups can be compared to analogous groups in other periods and other civilizations the medieval scholastics with Indian priest-scholars, the Renaissance humanists with Chinese mandarins, the Renaissance artisans and artists with their colleagues in classical antiquity. Since, in the sociology of culture, experiments are not feasible, comparison of analogous phenomena is virtually the only way of finding and verifying causal explanations. It is strange how rarely investigations of this kind are made. As the complex intellectual constructs are usually studied historically only, so sociological research for the most part restricts itself to comparatively elementary phenomena. Yet there is no reason why the most important and interesting intellectual phenomena should not be investigated sociologically and causally. 20

237 232 DIEDERICK RAVEN The second comment concerns the manner in which a causally oriented sociohistorical science needs to be carried out: Among sociologists there are today various schools and many controversies; some schools even disregard the investigations of most of the others. It might be generally agreed that sociology is an empirical science. It is based on observation and comparison, if not yet on experiment. As it does not deal with individuals but investigates groups and mass phenomena, the general sociological statements which appear in still rather uncritical forms in Hegel and Comte must be based on careful and complete collection of material if reliable results are to be achieved. It is here that statistical and sometimes even quantitative methods were successfully introduced in sociology. They were, however, largely applied to quite elementary phenomena and their use frequently resulted in mere collections of material. Causal and comprehensive sociological theories, based on statistics, are still lacking. 21 These three comments characterize the relation between Zilsel s two projects. His goal was to demonstrate the feasibility of laws in the humanities. Given his philosophical position, he could not just argue the case in general. He had to show, through detailed socio-historical analysis, that causal historical and comparative research was possible. His project on the social origins of modern science was an attempt to search for the laws that would causally explain how the concept of law came to dominate our understanding of nature and the reasons for its lack of use in the humanities. His project on the concept of law intended to demonstrate that there are no meaningful restrictions preventing one from doing so. The former project, in turn, demonstrates the fruitfulness of doing so. The focal point of both projects is the question Zilsel kept asking for 30 years: the empirical as well as theoretical clarification of what it means to say that there are also laws in history and sociology. 22 This question is at the centre of Zilsel s conception of the unity of science. NOTES Cf. F. Stadler, Studien zum Wiener Kreis: Ursprung, Entwicklung und Wirkung des Logischen Empirismus im Kontext. (Frankfurt: Suhrkamp, 1997). All of Zilsel swartime essays, together with previously non-published material, are collected in E. Zilsel The Social Origins of Modern Science, edited by D. Raven, W. Krohn and R.S.Cohen, with a foreword by J.Needham (Dordrecht, etc: Kluwer Academic Publishers, 2000) (Boston Studies in the Philosophy of Science Vol 200); the acronym SOMS will be used for references to this volume. Appendix IV of SOMS contains a complete bibliography of Zilsel s work. Cf. SOMS, part I, pp E.Zilsel, Das Anwendungsproblem: Ein philosophischer Versuch über das Gesetz der grossen Zahlen und die Induktion (The Application Problem, A Philosophical Investigation of the Law of Large Numbers and its Induction) (Leipzig: Barth, 1916) p. 3. Apparently Zilsel s analysis made a deep impression in mathematical circles, cf. Stadler, Studien zum Wiener Kreis, p. 802.

238 EDGAR ZILSEL S RESEARCH PROGRAMME Zilsel, Philosophische Bemerkungen, in Der Kampf, 1929, 22, pp ; republished in E. Zilsel Wissenschaft und Weltanschauung: Aufsätze , K.Acham and G.Mozeti (eds.), 1992, pp , p. 40. Ibid, p. 39. Quote from the Zilsel file at the University of Vienna; we will use the acronym ZF/UV for references to this file. The complete text of this letter is given by Edgar Zilsel und die Einheit der Erkenntnis. (Wien: Löcker Verlag, 1981), p , n. 11. ZF/UV. SOMS, pp Zilsel to Reichenbach, HR , May 2, In quoting from the Reichenbach papers, which are located at the Archives of Scientific Philosophy in the Twentieth Century, Pittsburgh, Penn. we will use the acronym HR followed by the inventory control number of that collection, followed by the date of the letter. Quoted by permission of the University of Pittsburgh. All rights reserved. E. Zilsel, Partei, Marxismus, Materialismus, Neukantianismus, Der Kampf, 1931, 24, pp , p. 21; republished in Wissenschaft und Weltanschauung, op. cit., 1992, pp E.Zilsel Review of O.Neurath Empirische Soziologie, Der Kampf, 1932, 25, pp , p. 93, republished in Wissenschaft und Weltanschauung, op. cit., pp Apart from his methodological queries, Zilsel did not have the highest opinion of Neurath s book. Reichenbach asked him to review it in Erkenntnis (HR , November 18, 1931) and Zilsel replied: By the way, Neurath s book does not seem to me to be worked out very well and I don t believe Erkenntnis would lose a lot if a review were omitted. (HR , December 5, Quoted by permission of the University of Pittsburgh. All rights reserved.) For a thorough analysis of the background of this debate see K.C. Köhnke, The Rise of Neo- Kantianism: German Academic Philosophy between Idealism and Positivism. (Cambridge: Cambridge University Press, 1991). This debate was hard to lay to rest cf. H.-J.Dahms, Positivismusstreit: Die Auseinandersetzungen der Frankfurter Schule mit dem Logischen Positivismus, dem Amerikanischen Pragmatismus und dem Kritischen Rationalismus. (Frankfurt am Main: Suhrkamp, 1994). This is confirmed in correspondence with George Sarton in June 1939 shortly after Zilsel arrived in the USA. Zilsel lets Sarton know that he has arrived in America with two copious manuscripts, on Natural and Historical Laws the one, on the Social Roots of Science the other, [which] in spite of preparations lasting years are not quite finished yet and therefore are not published. Zilsel to Sarton, June 3, 1939, quoted from the Sarton papers held at the Houghton Library of Harvard University. See especially Zilsel s popular essay Das mechanistische Weltbild und seine Überwindung, Der Atheist, 1932, 6(9), pp This the central theme of Physics and the Problem of Historico-Sociological Laws, SOMS, pp See volume II, number 8 of the Encyclopedia. This issue is called The Development of Rationalism and Empiricism and consists of two independent but complementary essays one written by Zilsel and the other, titled Aspects of Scientific Rationalism in the Nineteenth Century, written by George de Santillana. HR , May 2, Quoted by permission of the University of Pittsburgh. All rights reserved. The Sociological Origins of Modem Science, SOMS, pp. 19, emphasis added. Problems of Empiricism, SOMS, p. 194; unless otherwise indicated all subsequent quotations are from this essay, and page references will be given in the main text. The quotation is from the Problemstellung of Zilsel s project, Laws of nature and historical laws, SOMS, p. 233.

239 234 DIEDERICK RAVEN Department of Anthropology Utrecht University PO Box NL-3508 TC Utrecht The Netherlands

240 GREGOR SCHIEMANN CRITICIZING A DIFFERENCE OF CONTEXTS ON REICHENBACH S DISTINCTION BETWEEN CONTEXT OF DISCOVERY AND CONTEXT OF JUSTIFICATION With his distinction between the context of discovery and the context of justification, Hans Reichenbach gave the traditional difference between genesis and validity a modern standard formulation. Reichenbach s distinction is one of the well-known ways in which the expression context is used in the theory of science. The extensive criticism of Reichenbach s distinction in the last century can be understood as criticism of a context distinction. This criticism could be summed up by saying that Reichenbach s view was very one-sided; it concentrated on particular aspects of the difference between discovery and justification and thereby underrated their common feature of being a part of scientific knowledge. Critics proposed other concepts of context, or they questioned the existence of Reichenbach s context distinction, but they did not question the use of the concept of context. My argument is that Reichenbach s concept is unsuitable and leads to contradictions in the semantic fields of genesis and validity. I would like to demonstrate this by examining the different meanings of Reichenbach s context distinction. My investigation also shows how the difference between genesis and validity precedes Reichenbach s context distinction and indicates approaches for meaningful applications of the concept of context to the phenomena designated by Reichenbach. Considering the extensive critical reception of Reichenbach s distinction, it is truly surprising that his argumentation has received virtually no analysis so far. 1 This circumstance is all the more remarkable considering that an analysis would only need concentrate on relatively few aspects of Reichenbach s work. Reichenbach introduces his distinction in passing and hardly explains it. He refers to it briefly in his Zur Induktions-Maschine (Reichenbach 1935). In the first paragraph of Experience and Prediction, he uses the distinction to explain the tasks of epistemology, and touches on it again in the next-to-last paragraph of this work (Reichenbach 1938, in the German translation Reichenbach 1983). He also briefly mentions it in the introduction to Elements of Symbolic Logic (Reichenbach 1948, in the German translation Reichenbach 1999) and in a passage in his book The Rise of Scientific Philosophy (Reichenbach 1951, in the German translation Reichenbach 1968). I will reconstruct Reichenbach s argumentation only insofar as it is required for a criticism of the use of the concept of context. The expression context is 237 F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

241 238 GREGOR SCHIEMANN an amply vague term that is similar in meaning to the German expression Zusammenhang, but is found more commonly than the latter in scientific usage. Reichenbach probably took the expression Zusammenhang from everyday language, as a matter of course. I would like to assume that there are sufficient similarities between the spectrum of meaning which this expression had in those days and which the context expression has today. In order to discuss a specific application, it is necessary to limit the diverse meanings of the term context. My definition aims to summarize features of its everyday usage, as documented in relevant dictionaries, and to combine these with the meaning found in Reichenbach s texts. Accordingly, a context designates a non-singular class of phenomena or a field of reference which is distinguished so clearly from other comparable fields of reference that it is reasonable to give it a summarizing concept (e.g., the context of the meaning of a text in contrast to other contexts of meaning in different texts by the same author, the context of a specific situation contrasted with other contexts of action, historical contexts as parts of a more comprehensive history). Within these rough guidelines, I will begin by reconstructing the way in which Reichenbach introduces the distinction between discovery and justification as a difference of contexts. The common features of the distinction are given with the concept of knowledge, and its specific differences are given with Carnap s method of rational reconstruction. Rational reconstruction identifies conditions of validity and can be contrasted with the genesis of its objects. Reichenbach also uses the method to characterize justification and discovery as contexts. Using the concept of context in this way, he achieves neither an intensional nor an unambiguous extensional definition of the two proposed fields of reference (1). Drawing on the numerous meanings of the term context, I will then emphasize some chief characteristics and review, through exemplification, the usage of this term. First of all, I turn to the context of discovery as the nonrational part of all scientific knowledge and show that this meaning cannot be defined consistently (1a). For the context of justification, one can distinguish two main cases: the context of justification is either contrasted with the context of discovery, or it forms a unit therewith. In the first case, the use of the context term becomes paradoxical, insofar as justification separated from scientific practice does not represent a field of reference which could be specifically contrasted with another field of reference (1b). In the second case, the unifying definitions contradict the contextual meaning of discovery and justification (1c). In the last section, I point to a useful application of the concept of context which can be found in Reichenbach s argumentation and which refers to the practical conditions of justification (2).

242 CRITICIZING A DIFFERENCE OF CONTEXTS REICHENBACH S DISTINCTION BETWEEN CONTEXTS Reichenbach himself does not designate his distinction as normative. Rather, he introduces it in the first paragraph of Experience and Prediction in order to explain a task of epistemology, which he calls the descriptive one because its results are supposed to correspond to real thinking. 2 To characterize this task, he first distinguishes between the internal and external relations between those human utterances the whole of which is called knowledge (Reichenbach ). 3 Reichenbach seems to understand scientific disciplines as closed units which may be combined with utterances of another kind (ibid.). 4 The social sciences are responsible for the analysis of the external relationships. A sociologist, for instance, might report that astronomers construct huge observatories containing telescopes in order to watch the stars, and in such a way the internal relation between telescopes and stars enters into a sociological description. The report on contemporary astronomy begun in the preceding sentence might be continued by the statement that astronomers are frequently musical men, or that they belong in general to the bourgeois class of society; if these relations do not interest epistemology, it is because they do not enter into the context of science (ibid.). 5 The concept of knowledge encompasses not only scientific notions and theories but also the entire scientific, i.e. non-epistemological, practice. Reichenbach advocates a comprehensive concept of knowledge, the historically changeable criteria of which are determined by the social sciences. The inside and outside of knowledge is not separated by a sharp line of demarcation, as Reichenbach himself admits. Nevertheless, Reichenbach takes this separation as a basis for further differentiation. This differentiation does not yet lead to the distinction between discovery and justification, but it does result in the preceding separation of the internal structure of knowledge into a system of logical interconnections of thought and the actual way in which thinking processes are performed (Reichenbach f.). 6 Reichenbach gives this separation a universal validity and also allocates responsibility to different subject disciplines: the analysis of the logical connections is to be the task of epistemology, and the analysis of the real thought processes is to be the task of psychology. 7 This assignment of competencies underlines the considerable range covered by the concept of knowledge, a concept which includes thinking that follows no logical rules and can at best be understood psychologically. Moreover it prepares the use of a concept of context which can be applied to the reference fields of academic disciplines or to their corresponding methods, respectively. In order to introduce the as yet missing link between real thinking and the fictitious field of reference of epistemology, Reichenbach resorts to Rudolf Carnap s concept of rational reconstruction. 8 Epistemology is connected with

243 240 GREGOR SCHIEMANN the starting and ending points of the real scientific thought processes and tries to rationally reconstruct logically structured links, which are in greatest possible agreement with the real thought processes, between these points. 9 The purpose of this procedure is to investigate the conditions of validity: [The] fictive set of operations [...] is chosen from the point of view of justifiability; we replace actual thinking by such operations as are justifiable, that is, as can be demonstrated as valid (Reichenbach ). 10 The distinction between actual thinking and its rational reconstruction belongs to the tradition of the difference of genesis and validity. While genesis of knowledge generally means its origin and evolution, its validity designates the intersubjective, definite and objective basis of its recognition. 11 The distinction between genesis and validity expresses that the legitimacy of validity claims of knowledge is independent of a report on their genesis. The distinction does not rule out answering validity questions by referring to the conditions of origin or development. Rather, the question of validity not only presupposes the genesis, but must refer to its results, whose validity conditions are in question. Moreover, the distinction is not characterized by a temporal relation of succession. At every stage of a genesis, one can inquire about validity. Because genesis and validity do not designate separate fields of reference, but rather describe two properties which are constitutive of every object of knowledge, the concept of context should not be used. Nonetheless, Reichenbach introduces the context distinction to explain how epistemology is responsible for determining validity conditions. The ambiguities arising from this are linked to an analogy which he draws between epistemological and scientific justifications: If a more convenient determination of this concept of rational reconstruction is wanted, we might say that it corresponds to the form in which thinking processes are communicated to other persons instead of the form in which they are subjectively performed. [... The] well-known difference between the thinker s way of finding [... a] theorem and his way of presenting it before a public may illustrate the difference in question. I shall introduce the terms context of discovery and context of justification to mark this distinction. Then we have to say that epistemology is only occupied in constructing the context of justification. But even the way of presenting scientific theories is only an approximation to what we mean by the context of justification (Reichenbach f. emphasis in original). 12 To what does Reichenbach relate the expression of context (expressed notably in the singular)? One can distinguish two main cases. In the first case, he admits with his analogy that rational reconstructions are similar to the normal representation of theories in scientific practice. This would suggest understanding the context of justification as a part of scientific practice and its reconstruction as an epistemological activity. 13 This interpretation finds support in Reichenbach s 1935 remark on the context difference. In this remark, the procedure of justifica-

244 CRITICIZING A DIFFERENCE OF CONTEXTS 241 tion refers to a method by which the researcher makes his theories public. 14 The formal criteria of justification which must be fulfilled in the communication of theories are analogous in structure to the criteria of epistemological reconstruction. The transition between scientific justification and epistemological reconstruction is fluid. Reichenbach leaves room for justifying activities in science too, when he strictly limits the extension of the context of discovery. This context contains only those procedures which the individual researcher uses during the discovery of new theories. 15 In The Rise of Scientific Philosophy, Reichenbach also assumes an extension that only partially covers scientific knowledge when he relates the context of discovery explicitly to the act of discovery (Reichenbach ). 16 On the other hand however, in the last sentence of the aforementioned quotation from Prediction and Experience Reichenbach clearly contrasts the expression context of justification with scientific practice. This gives the impression that the context of justification might not be the subject but only the result of rational reconstruction and might therefore not appertain to scientific work. 17 This interpretation finds support in the competence Reichenbach allocates to subject disciplines, along the lines of his context distinction: We emphasized that epistemology cannot be concerned with the [... context of discovery] but only with the [... context of justification]; we showed that the analysis of science is not directed toward actual thinking processes but toward the rational reconstruction of knowledge (Reichenbach ). 18 The mutually exclusive and universally conceived fields of reference of epistemology and psychology meet in this passage with those of the context distinction. An expansion of the context of discovery to all actual thought processes is not excluded and would only be the reverse side of limiting the context of justification to an exclusively epistemological field of reference. Several authors followed this interpretation in equating Reichenbach s context distinction with the fields of reference or methods of epistemology and psychology. 19 It is this understanding more than any other, which refers back to the traditional difference between genesis and validity, which connects with Reichenbach s division of subject disciplines. It transforms the difference of aspects of knowledge into a difference of contexts and thus, mistakenly, allows object properties to be contrasted as if they were fields of reference. However, my reconstruction proceeds on the assumption that this case has but little relevance in Reichenbach s work. He mostly relates the context of discovery to just one vaguely defined aspect of scientific work. But even less well-defined is his extension of the context of justification. However, before I come to discuss this in more detail, I will begin as proposed with a closer consideration of the context of discovery.

245 242 GREGOR SCHIEMANN a) Context of Discovery According to Reichenbach, the sciences have the task of finding logical interconnections between divergent ideas about newly observed facts (Reichenbach ). 20 In order to fulfil this task the scientific genius, paradoxically, never finds himself committed to the narrow steps [...] of logical reasoning (ibid.): 21 The act of discovery escapes logical analysis; there are no logical rules in terms of which a discovery machine could be constructed that would take over the creative function of the genius. But it is not the logician s task to account for scientific discoveries; all he can do is to analyze the relation between given facts and a theory presented to him with the claim that it explains these facts (Reichenbach ). 22 As in this quotation, Reichenbach often applies the expression discovery to laws and theories. Examples are not only Boyle s law, 23 Newton s law of gravitation 24 or quantum mechanics, 25 but also formal theories like the non-euclidean geometries. 26 In addition, he uses the expression for the discovery of phenomena, including blood circulation 27 or electric current, 28 as well as for technical inventions such as the telescope, the air-pump, 29 the railroad or the radio. 30 The full scope of the concept corresponds to the comprehensive sociological concept of knowledge, which is not based on criteria of rationality, but on historical features. The semantics of the expression discovery has a realistic connotation and assumes that knowledge is not so much produced, but is rather, like facts, found. Accordingly, the discoverer only has a function. He is guided, as Reichenbach writes, by an attitude toward knowledge, by the desire to come to know something about the secrets of nature. 31 His action takes place under the compulsion of a drive, but is nonetheless determined by his will. 32 Basically, the discoverer only removes alien circumstances that conceal his view of the essential parts of reality. Therefore, the search for a discovery must be directed towards objects that have no inner connection with the discovery itself. This explanation of the difference claimed to exist between the process of discovery and its result, is characterized by the duality of will and knowledge. On this basis, Reichenbach characterises discoveries as non-rational. Accordingly, the discovery of theories, he claims, is guided by unjustified presumptions, follows no exact methods (Reichenbach ), and resembles an irrational guessing (ibid. 231). 33 Only this negative characterization of lacking rationality or, respectively, logical structure, constitutes the particular difference from the context of justification. This property not only excludes every rational reconstruction, but also leads to psychology, the subject matter of which includes discoveries that have but a limited capacity of explanation:

246 CRITICIZING A DIFFERENCE OF CONTEXTS 243 Let me say that I should be the last to discredit the work of the great men of science. [...] The obscurity of the birth of great ideas will never be satisfactorily cleared up by psychological investigation (Reichenbach ). 34 Not even in retrospect, with background knowledge of the personalities and the historical circumstances involved, does Reichenbach believe that the occurrence of discoveries is to be understood. 35 He sees discoveries as following an irreducible voluntary process, the accidental character of which is most clearly evident when contrasted with the logical structure of the process s own result the discovery. It is questionable whether this characterization applies to any of the objects in Reichenbach s context of discovery. The main criticism concerns the inclusion of inventions. The history of technology in the latter half of the last century has destroyed the idealised image of ingenious personal achievements of technicians and engineers. 36 New technical constructions can result from a complex net of coincidences just as well as from a systematic research process. The origins of the technical systems mentioned by Reichenbach (railroad, broadcasting) go back to countless conditions and practical goals that are quite accessible to rational analyses. This might also be the case for the emergence of new empirical laws or theories. Moreover, the very concepts of law and theory are linked to testable validity conditions, which contradicts the supposed inclusion in the nonrational context. Empirical laws and theories refer to data according to proved rules. 37 Critics of Reichenbach s context distinction have pointed out that the non-rational aspects of discoveries are only dominant in an initial stage in which new intuitive ideas, hypothetical presumptions and so forth are important. 38 Their characterization as a context would nonetheless still be problematic, insofar as this would assume a division of genesis and validity. It seems to be more suitable to suppose a minimisation of validity for specific stages of the genesis. One can also formulate this criticism by using the terminology of context. Not the inner aspects of knowledge, not its genesis or validity, but the external influencing factors can be arranged into several contexts, i.e., into particular fields of reference. Discoveries could be inextricably entangled in the most diverse psychological, social, historical etc. contexts. b) The Epistemological Context of Justification I now turn to both extremes of the meaning of justification. The first is the assumption that justification is an exclusively epistemological activity and therefore situated beyond scientific work. Epistemology enjoys far-reaching freedoms in fulfilling the task of reconstructing logical structures. Only the starting and the endpoints of a rational reconstruction must match empirical data according to rules of correspondence. However, rational reconstruction represents only the first stage in the epistemological procedure of justification. Rational reconstruc-

247 244 GREGOR SCHIEMANN tion can fail in its aim to replace the real thinking process with a logically structured system, because it may be impossible to find connections between the real starting and endpoints. In its (second) critical task, which has priority over the descriptive one, epistemology is no longer committed to the demand for correspondence with the real processes, but rather to achieving valid thinking (Reichenbach ). 39 Reichenbach allocates this analysis of science to the logic which he separates fundamentally from experience. 40 Deductive logic is empty, consisting of tautologies, that is, it does not express properties of physical objects. 41 According to Reichenbach, the "manifold forms of [logical] induction [...] are expressible in terms of deductive methods" that need only to be to supplemented by one non-analytical principle induction by means of enumeration (Reichenbach ). 42 This restriction of the purely analytical character might not, he says, prevent one from being allowed to ascribe absolute validity to logic, even though this quality is "unknowable, since we never know whether we have it" (Reichenbach ). One could describe the analytical character of logic as also being relatively context-independent. The absolute has neither boundaries nor is it specifically distinguished from something else. If the concept of the context of justification were applied exclusively to logic, it would receive the paradoxical semantics of a non-contextual context. Reichenbach, as I would like to maintain, uses the concept of context to defend the context-independence of his own concept of logic. The semantics of demarcation, which is combined with the context term, served him as a means for contrasting logical investigation with scientific practise. He did not notice the resulting inconsistent definition of the concept of context. This shortcoming is not only a consequence of an inadmissible division of genesis and validity, but also of a conception of logic that is no longer maintainable after Quine s criticism of the distinction between the analytical and the synthetical. c) Scientific Contexts of Justification and of Discovery However, Reichenbach does not only contrast justification and discovery of knowledge. In a twofold manner, he also understands them as an unit. Following a terminology used by Lutz Danneberg, I would like to distinguish between a model of succession and a model of levels. 43 The succession model has a horizontal and excluding structure. It divides up the two aspects of knowledge into two contexts following each other within a given period of time, being parts of the same process. 44 The model of levels, on the other hand, has a vertical order. It abolishes the exclusive non-rational characterization of discoveries and, instead, assumes that they partially satisfy a logic of induction. It views the practical process of science as the surface of a hidden logical structure. Both models point towards a cancellation of the concept of context. Once again, the succession model reveals the distortion of the relation between genesis and validity caused by the concept of context. In this model, the

248 CRITICIZING A DIFFERENCE OF CONTEXTS 245 scientist arrives at a new finding, without having previously been occupied with the validity that he needs to present his findings in his scientific community: [The] same scientist who discovered his theory through guessing presents it to others only after he sees that his guess is justified by the facts. It is this claim of justification in which the scientist performs an inductive inference, since he wishes to say not only that the facts are derivable from his theory, but also that the facts make his theory probable and recommend it for the prediction of further observational facts (Reichenbach ). 45 The assumption that the effort of justification does not start before the discovery is complete underlines the small extension of the context of discovery (cf., 1a). Logic does not determine the emergence of a finding, but follows immediately after its establishment. With this, Reichenbach himself reaches the limit of his concept of discovery and justification contexts. The contact of the two contexts already suggests their overlap and their cancellation. If inductive considerations determine the first communication about a new finding, why should they not also already influence the intuitively guided process of discovery? 46 With the model of levels, Reichenbach takes a different course. The forms of justification mentioned so far are based on findings which have already been advanced. This situation corresponds to Reichenbach s dictum that epistemology does not maintain anything about the question of how [... a discovery] is performed (Reichenbach ). 47 In the model of levels, by contrast, he relates the object of justification to the search for new knowledge which has not yet been successfully completed. Induction turns from a method of justification into a method of searching. 48 In 1951, he notes generally: Induction is the instrument of a scientific method that is intended to discover something new, something going beyond a summary of previous observations (Reichenbach ). 49 Reichenbach argues for using induction in processes such as the extrapolation and interpolation of data, but he does not discuss to what degree scientists use induction in order to find new laws and theories. The assumption that they do use it is, however, strongly suggested by his reconstructions; Galileo s law of falling bodies and Kepler s law of planetary motion result inevitably, in his view, from observed bodily positions. 50 He sees the simplest combination of the two laws as being represented in Newton s law. 51 From this perspective, historical progress comes close to a succession of solutions of mathematical probability problems. I will not repeat last century s well-known debate within the history of science on the inductive or accumulative view of the evolution of knowledge. 52 With regard to Reichenbach s use of the concept of context, the more important question is in what way the contexts of justification and discovery are related, if the former has an effect on the latter that is not found in other variants of their meaning. For Reichenbach, the inductive view of the progress of knowledge is not only a possible, but also a hidden, already given, reality. He is so convinced

249 246 GREGOR SCHIEMANN of his reconstructions that it seems probable to him that they also influence and even in fact control the actual discovery process: If we were to analyze the discoveries of [... scientists], we would find that their way of proceeding corresponds in a surprisingly high degree to the rules of the principle of induction [...]. The mysticism of scientific discovery is nothing but a superstructure of images and wishes; the supporting structure below is determined by the inductive principle. [... It] seems to be a psychological law that discoveries need a kind of mythology (Reichenbach ). 53 Accordingly, discoveries could have been following inductive logic all along without science having noticed it. Science would have the wrong, logic the only correct consciousness of the real process. With this view, Reichenbach gives his dual conception of knowledge an ontological meaning. The two new contexts of the upperstructure and the substructure are separated by the unchanged criterion of rational reconstruction. The mythological upper level is as non-rational as the discoveries in the model of succession; the substructure has a structure analogous to epistemological justification. The model of levels puts the structure of justifications under scientific practice and transforms only the non-rational elements of discoveries into insignificant surface phenomena. Situated now between both levels are discoveries influenced by inductive logic. At the price of the introduction of two new fields of reference, the inevitable interaction of justification and discovery has at least led to a dissolution of the separate context of discovery. 2. CONTEXTS OF SCIENTIFIC JUSTIFICATION I assumed the expression context to be a concept that means a field of reference that is specifically distinguished from other fields. With this meaning, the concept is suitable for the characterization not only of the external influences on science (cf., 1a), but also for the inner conditions of scientific knowledge. Reichenbach himself offers an example when he derives the practical necessity of scientific justifications from the requirements of academic communication. In this sense, scientific justifications have a context, the context of their verbal presentation and written publication independent of whether or not they themselves form a context. Reichenbach takes the reconstruction and analysis of epistemology as criteria for the examination of justification in science. Because inaccuracies inevitably occur under the practical conditions of science, the epistemological examination of scientific justifications is necessary. Where logical shortcomings exceed a certain measure, epistemology has the (third) task of advising the researcher. 54 But the criteria of epistemology are not sufficient to examine the validity conditions of normal science. For instance, logical inaccuracies can be necessary in the pursuit of research tasks. The communication of most scientific findings

250 CRITICIZING A DIFFERENCE OF CONTEXTS 247 would be impossible if one were to insist on the proof of the countless accepted alleged logical connections. In this respect, there is a strained relationship between the interests of communication and justification. On the other hand, Reichenbach refers to an equivalence between communication and justification, so that successful scientific communication also requires a minimum of justification. Reichenbach hereby offers a surprising common ground with modern contextualism, which also claims that the legitimate requirements of justification find their measure in the particular argumentative context within science. 55 Reichenbach would agree with contextualists that science does not need to worry about the justification of its statements independent of its practice this is indeed Reichenbach s reason for separating scientific and epistemological tasks. Reichenbach would furthermore accept that only the conditions of communication constitute that part of practice in which the practical necessity for the justification can be determined positively. Finally, he would even be able to agree with contextualism that science must only justify knowledge to the extent that the specific context of communication requires. 3. CONCLUDING REMARKS Assuming a concept of context that was probably meant by Reichenbach and is commonplace today, I have examined different meanings of his distinction between discovery and justification. For this distinction, the difference between genesis and validity is fundamental, primarily because it affects the preceding separation of knowledge into actual thought processes and the system of the logical connections. If one understands discoveries as a subset of genesis and their justification as a subset of validity, it is of course evident that the presupposed concept of context cannot be reasonably applied to them. It follows, then, that discoveries cannot be separated from validity questions any more than justifications can be separated from questions of the origin and evolution of their objects. Because of the comprehensive definition of knowledge, the exclusion of the connection between validity and genesis caused by the application of the concept of context is especially obvious in Reichenbach. In the large spectrum of meanings of knowledge that Reichenbach puts into the extension of his concept of discovery, one can easily find counter-examples to refute his division of the conditions of origin and validity. Conversely, as a method that comprises the whole variety of scientific knowledge, rational reconstruction must have a general definition that is situated at such a distance from the definition of its specific objects that designating justification as context becomes questionable. Reichenbach s attempts to combine the separation of discovery and justification in the models of succession or levels adhere to the one-sided difference of genesis and validity and are not convincing.

251 248 GREGOR SCHIEMANN It is not the conceptual properties of knowledge, but its practical conditions which offer opportunities for a reasonable application of the concept of context. Historical, cultural, economic, social, communicative etc. conditions, in which knowledge is developing and valid at the same time, offer opportunities to employ the concept. While Reichenbach s unifying understanding of these manifold conditions prevents just this, his well-informed description of the practical scientific justification processes permits a reasonable use of concepts of context. NOTES Approaches appear in Nickles 1980, Curd 1980, Siegel 1980 and Danneberg Literature references dealing with the reception can be found in Nickles 1980, Hoyningen-Huene 1987 and Danneberg Reichenbach Der Unterschied der inneren und äußeren Beziehungen zwischen den rnenschlichen Äußerungen, deren Ganzes Erkenntnis genannt wird (Reichenbach ). Äußerungen anderer Art (Reichenbach ). So könnte ein Sozialwissenschaftler berichten, daß die Astronomen große Observatorien bauen, die Fernrohre zur Beobachtung der Sterne beherbergen; auf diese Weise ginge die innere Beziehung zwischen Fernrohren und Sternen in eine soziologische Beschreibung ein. Der Bericht über die heutige Astronomie, der im vorhergehenden Satz begann, könnte mit der Aussage fortgesetzt werden, Astronomen seien oft musikalisch oder gehörten meistens der bürgerlichen Klasse an. Daß diese Beziehungen die Erkenntnistheorie nicht interessieren, rührt daher, daß sie nichts mit dem Inhalt der Wissenschaft zu tun haben (Reichenbach f). Es besteht ein großer Unterschied zwischen dem System logischer Verknüpfungen im Denken und der tatsächlichen Art und Weise, wie die Denkprozesse ablaufen (Reichenbach ). Reichenbach f. Catnap Reichenbach Die fiktiven Operationen [...] werden unter dem Gesichtspunkt der Rechtfertigung gewählt; das wirkliche Denken wird durch legitimierbare Operationen ersetzt, das heißt, durch solche, deren Gültigkeit erwiesen werden kann (Reichenbach ). SeeTiehl 1980ff.; Mühle 1971 ff. Um den Begriff der rationalen Nachkonstruktion auf einfachere Weise zu kennzeichnen, könnte man sagen, er entspräche der Art, wie Denkvorgänge anderen Menschen mitgeteilt werden, als der Art, wie sie sich subjektiv vollziehen. [... Der] wohlbekannte Unterschied, wie jemand einen Lehrsatz findet und wie er ihn einem Publikum vorführt, ist wohl ein gutes Beispiel. Ich führe dafür die Ausdrücke Entdeckungszusammenhang und Rechtfertigungszusammenhang ein. Dann können wir sagen, daß sich die Erkenntnistheorie nur mit der Konstruktion des Rechtfertigungszusammenhangs beschäftigt. Aber selbst die Art und Weise, wie wissenschaftliche Theorien dargestellt werden, ist nur eine Annäherung an das, was wir mit Rechtfertigungszusammenhang meinen (Reichenbach ). In theory of science the distinction is mostly understood as a conceptual instrument for better understanding the process of scientific knowledge acquisition from the emergence to the recognition of a finding. See Nickles 1980 and Hoyningen-Huene Verfahren, in welchem [... der einzelne Forscher] seine Theorien vor der Öffentlichkeit darlegt (Reichenbach ). Verfahren, welche[...] der einzelne Forscher bei der Auffindung neuer Theorien benutzt (Reichenbach ). Der Entdeckungsakt selbst (Reichenbach ).

252 CRITICIZING A DIFFERENCE OF CONTEXTS This interpretation is supported by Danneberg : The philosopher doesn t reconstruct given procedures of justification or explanations; it is he himself who creates the explanation ( Der Philosoph rekotistruiert nicht vorliegende Rechtfertigungsverfahren oder Begründungen; er selbst ist es, der Begründung schafft ). Naturally, scientists are also acting as philosophers if justifying theories is philosophy. Wir betonten, daß sich die Erkenntnistheorie nicht mit dem [... Entdeckungszusammenhang] beschäftigen kann, sondern nur mit dem [... Rechtfertigungszusammenhang]; wir zeigten, daß sich die Analyse der Wissenschaft nicht auf die tatsächlichen Denkvorgänge richtet, sondern auf die rationale Nachkonstruktion der Erkenntnis (Reichenbach ). See also Reichenbach (where Reichenbach uses explanation and justification synonymously). Siegel speaks of two parallel distinctions. Nickles 1980 claims that Reichenbach only wanted to logically differentiate between the psychological processes which occur when a scientist thinks of new ideas and the logical argument which exhibits the degree to which those ideas are supported by the facts and other evidential considerations. [l]ntimately connected with the [...] distinction between the process of discovery and the methods of justification is according to Hoyningen-Huene the distinction between academic disciplines. See also Footnote 53. Hoyningen-Huene f. also offers literature references supporting this claim. [L]ogische Beziehungen zwischen unterschiedlichen Hypothesen über neue Beobachtungsdaten aufzufinden (Reichenbach ). [N]ie an die pedantischen Schritte [...] des logischen Denkens gebunden gefühlt (Reichenbach ). Der Entdeckungsakt selbst ist logischer Analyse unzugänglich; es gibt keine logischen Regeln, auf deren Grundlage eine Entdeckungsmaschine gebaut werden könnte, die die schöpferische Funktion des Genies übernehmen würde. Es ist jedoch auch gar nicht die Aufgabe des Logikers, wissenschaftliche Entdeckungen zu machen, er kann nur die Beziehungen zwischen gegebenen Tatsachen und einer Theorie analysieren, die mit dem Anspruch aufgestellt wird, daß sie diese Tatsachen erklärt (Reichenbach ). Reichenbach Reichenbach Reichenbach Reichenbach f. Reichenbach Reichenbach Reichenbach Reichenbach Die Einstellung auf ein Wissen, de[n] Wunsch etwas zu erfahren von den Geheimnissen der Natur (Reichenbach ). Reichenbach [I]rrationale[s] Raten[...] (Reichenbach ). Ich möchte betonen, daß ich der letzte wäre, der das Werk der großen Männer der Wissenschaft herabsetzen wollte. [... Das] Geheimnis großer Schöpfungen wird nie zufriedenstellend durch psychologische Untersuchungen aufgeklärt werden können (Reichenbach ). For Reichenbach, the baffling emergence of Newton's theory of gravity and Einstein's theory of relativity are paradigms of this misunderstanding. See Reichenbach See e.g. Staudenmaier Kordig correctly emphasized that quite generally Real discoveries are well established. What is well established is justified. E.g., Laudan 1977, Kordig 1978, see. Nickles ff. [G]ültige[s] Denken (Reichenbach ). See also Reichenbach (where Reichenbach uses Rekonstruktion synonomously with Nachkonstruktion ). Reichenbach On Reichenbach s Kantianism, in which the categorical separation lives on, see e.g. Hecht (1994). Reichenbach Die verschiedenen Formen der Induktion [...] können durch deduktive Methoden dargestellt werden, zu denen lediglich die Induktion durch Aufzählung hinzutritt (Reichenbach ).

253 250 GREGOR SCHIEMANN Danneberg ff. allocates a variant of Popper s distinction between discovery and justification to his succession model, which converges with my definition. He ascribes the model of levels (which diverges with my definition) to the class of meaning from Reichenbach s distinction which in general concerns validity and genesis (see p. 231). The textual basis for the claim that there are no textual grounds for thinking that Reichenbach s distinction is a temporal one (Nickles ) is lacking. Derselbe Wissenschaftler, der seine Theorie durch raten entdeckte, [teilt] sie seinen Kollegen erst mit[...], nachdem er gesehen hat, daß die Tatsachen sein Raten gerechtfertigt haben. Die induktive Schlußweise kommt gerade in diesem Rechtfertigungsanspruch zur Geltung, denn der Wissenschaftler will nicht nur behaupten, daß die Tatsachen aus seiner Theorie ableitbar sind, sondern auch, daß die Tatsachen seine Theorie wahrscheinlich machen und man die Theorie darum zur Voraussage zukünftiger Ereignisse verwenden darf (Reichenbach ). In line with the reception of Reichenbach s distinction as a criteria for the analysis of the process of scientific knowledge acquisition (see footnote 13), the critics mostly presuppose a succession model and contest the time separation. See Hoyningen-Huene [I]ch sage nichts über die Frage der Theoriefindung (Reichenbach ). The fact that Reichenbach considered discoveries to be on the one hand philosophically meaningless and on the other to be inductively controllable is for Laudan an example of hardly surpassable nonsense and confusion in the philosophy of discovery. The model of levels is above all related to those interpretations of Reichenbach s distinction made by theoretical scientists interested in a logic of discovery. Induktion wird in der Wissenschaft benutzt, wenn es sich darum handelt, etwas Neues zu entdecken, d.h. zu einer Erkenntnis zu kommen, die über die Summe der bisherigen Beobachtungen hinausgeht (Reichenbach ). Reichenbach Reichenbach ff. See e.g. Diederich Hg Würde man die Entdeckungen [... von Wissenschaftlern] analysieren, so fände man, daß ihre Vorgehensweise in überraschend hohem Maße den Regeln des Induktionsprinzips entspricht [...]. Das mystische Gerede über die wissenschaftliche Entdeckung ist nur ein Überbau von Bildern und Wünschen; der stützende Unterbau wird vom Induktionsprinzip bestimmt. [... Es] scheint ein psychologisches Gesetz zu sein, daß Entdeckungen eine Art Mythologie brauchen (Reichenbach f.). Reichenbach ff. More generally, in the sense of Analytic Philosophy, contextualism designates an epistemological theory that standards of knowledge and justification vary with context (Brower 1998). See Brower 1998 and Williams 2001 for a survey, and introductory literature, and also Jutta Schickore s contribution in this volume. REFERENCES Brower, B.W. (1998); Contextualism, Epistemological, in: Routledge Encyclopedia of Philosophy. London. Carnap, R. (1928): Der logische Aufbau der Welt. Berlin. Chmielecka, W. (1982): The Context of Discovery and Justification: A Reappraisal, in: W. Krajewski (Ed.), Polish Essays in Philosophy of the Natural Sciences. Dordrecht 68, 63ff. Danneberg, L. (1994): Die philosophische Analyse im Logischen Empirismus. Explikation und Rekonstruktion, in: Danneberg, Kamlah und Schäfer (Hg.) Danneberg, L., A. Kamlah und L. Schäfer (Hg.) (1994): Hans Reichenbach und die Berliner Gruppe. Braunschweig/Wiesbaden. Diederich, W. (Hg.) (1974): Theorien der Wissenschaftsgeschichte. Beiträge zur diachronischen Wissenschaftstheorie. Frankfurt am Main. Hecht, H. (1994): Hans Reichenbach zwischen transzendentaler und wissenschaftsanalytischer Methode, in: Danneberg, Kamlah und Schäfer (Hg.)

254 CRITICIZING A DIFFERENCE OF CONTEXTS 251 Hoyningen-Huene, P. (1987): Context of Discovery and Context of Justification, in: Stud. Hist. Phil.Sci. 18, 501ff. Kamlah, A. (1991): H. Reichenbach: Prinzipien, Konventionen, Wahrscheinlichkeit, in: J.Speck (Hg.), Grundprobleme der großen Philosophen, Philosophie der Neuzeit VI. Göttingen, 67ff. Kamlah, A. (1993): Hans Reichenbach Leben, Werk und Wirkung, in: R.Haller und F.Stadler (Hg.), Wien-Berlin-Prag. Der Aufstieg der wissenschaftlichen Philosophie. Centenarien Rudolf Carnap, Hans Reichenbach, Edgar Zilsel. Wien, 238ff. Kantorovich, A. (1988): Philosophy of Science: From Justification to Explanation. British Journal far the Philosophy of Science 39, 469ff. Klein, C. (1998): Konventionalismus und Realismus, Zur erkenntnistheoretischen Relevanz der empirischen Unterbestimmtheit von Theorien. Paderborn. Kordig, C.R. (1978): Discovery and Justification, in: Philosophy of Science 45, 110ff. Nickles, T. (1980): Introductory Essay: Scientific Discovery and the Future of Philosophy of Science, in: Nickles (Ed.) Nickles, T. (1990): Discovery, in: R. Olby (Ed.), Companion to the History of Science. London / New York. Nickles, T. (1995): Philosophy of Science and History of Science, in: Osiris 10, 139ff. Nickles, T. (Ed.) (1980): Scientific Discovery, Logic and Rationality. Dordrecht etc. Mühle, G. (1971 ff.): Art. Genese, genetisch, in: J.Ritter und K.Gründer (Hg.), Historisches Wörterbuch der Philosophie. Darmstadt. Poser, H., und U. Dirks (Hg.) (1998): Hans Reichenbach, Philosophie im Umkreis der Physik. Berlin Reichenbach, H. (1929): Ziele und Wege der physikalischen Erkenntnis, in: H.Geiger und K. Scheel (Hg.), Handbuch der Physik. Bd. 4. Berlin, S. 1 ff. Reichenbach, H. (1935): Zur Induktionsmaschine, in: Erkenntnis 5, 172f. Reichenbach, H. (1938): Experience and Prediction. An Analysis of the Foundation and Structure of Knowledge. Chicago/Illinois. Reichenbach, H. (1940): On the Justification of Induction, in: The Journal of Philosophy 37 (4): 97ff. Reichenbach, H. (1948): Elements of Symbolic Logic. New York. Reichenbach, H. (1951): The Rise of Scientific Philosophy. Berkeley etc. Reichenbach, H. (1968): Der Aufstieg der wissenschaftlichen Philosophie. Braunschweig. Reichenbach, H. (1971ff): Gesammelte Werkein 9 Bänden. Braunschweig/Wiesbaden Reichenbach, H. (1983): Erfahrung und Prognose. Eine Analyse der Grundlagen und der Struktur der Erkenntnis, in: Reichenbach 1971 ff., Band 4. Reichenbach,H.(1999):Grundzügedersymbolischen Logik, in:reichenbach 1971ff.,Band 6. Salmon, W.C. (1979): The Philosophy of Hans Reichenbach, in: Salmon Ed. (1979), 1ff. Salmon, W.C. (Ed.) (1979): Hans Reichenbach, logical empiricist. Dordrecht etc. Siegel, H. (1980): Justification, Discovery, and the Naturalizing of Epistemology, in: Philosophy of Science 47, 297ff. Staudenmaier, M. (1994): Rationality versus Contingency in the History of Technology, in: M. R. Smith and L. Marx (Hg.), Does Technology Drive History? Cambridge usw. Thiel, C. (1980ff.): Art. Geltung, in: J.Mittelstrass (Hg.): Enzyklopädie Philosophie und Wissenschaftstheorie. Mannheim usw. Williams, M. (2001): Problems of Knowlwedge. A Critical Introduction to Epistemology. Oxford. Zittlau, D. (1981): Die Philosophie von Hans Reichenbach. München. Philosophisches Seminar Universität Tübingen D Tübingen Germany gregor.schiemann@uni-tuebingen.de

255 GIORA HON CONTEXTUALIZING AN EPISTEMOLOGICAL ISSUE: THE CASE OF ERROR IN EXPERIMENT The force of a philosophical position is its generality: the more general the position, the more forceful it is. Thus, knowledge claims within a general philosophical system are context-invariant, or at least one strives to make them so. This means that one seeks to make knowledge propositions, their meaning and their grounding free of context-sensitive distinctions. To attain generality, knowledge claims should be prized as much as it is possible out of specific relations, which could only delimit philosophy and restrict it to a certain context. Context expresses a defined structure consisting of either abstract or concrete elements, which are linked together in a way that a coherent relation emerges between each element and the structure as a whole. While the elements are constitutive of the structure, it is the structure that provides the coherence that links in turn the elements consistently. The melody, the tune, is the structure that provides the constitutive notes with coherence. The structure, that is, the constitutive elements together with their emerging coherent relations to the set as a whole and to one another, determines a reference frame. Within its limited, defined scope, this reference frame provides knowledge claims with coherence that helps facilitate the grounding of the claims, but at the cost of generality. The experimental context the central issue of this paper is both of material and theoretical nature. One conducts an experiment always within some material context and its theoretical understanding. I proceed from the assumption that philosophy strives to make its knowledge claims context-free and seeks generality when questions of validity and meaning arise. In view of this presupposition, I put forward the claim that philosophy of science is fundamentally different from philosophy at large: knowledge claims in philosophy of science, that is, propositions of science, are by rule context-dependent. In principle, so the claim goes, one cannot prize knowledge claims in philosophy of science out of their context. The occurrences of error in experiment are a case in point. What is error? Error is a multifarious epistemological phenomenon which consists essentially in a separation between elements that may be either concrete or abstract. In the former case material objects that have been intended to coincide are in fact at a distance; in the latter, incoherence produces a breach between propositions that are assumed to concur one with another. 1 The Latin erro means A. In general, to wander or stray about, to wander up and down, to rove... B. 253 F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

256 254 GIORA HON In particular, to miss the right way, to lose one s self, go astray... (Lewis and Short 1966, 657). Error is an expression of divergence whose mark is discrepancy a discrepancy which emerges from a procedure of evaluation against a chosen standard. The nature of this discrepancy, the reason for its occurrence, how to treat it and what can be learnt from it once it has been perceived and comprehended, constitute the vast subject of the problem of error (Hon 1998, 466). This characterization of error is general. It does not depend on any specific philosophical view, that is, it is not committed metaphysically. The definition is rather functional as it emerges from the practice of detecting error and rectifying it. How do we detect, identify, characterize and eventually rectify an error? Error is recognized as such due to a certain procedure of evaluation. It is essentially a comparative procedure, based on a chosen standard. We observe here a two-tier system: rendering an intended objective which has not materialized failure or a proposition which was considered true to be in fact false, presupposes evaluative criteria at another level than the elements in question. The application of these evaluative criteria at a different level from an objective or a proposition at stake reveals principally the chief characteristic of error, namely, divergence (for details see Hon 1995b). Any additional element to this characterization, for example, a theory that explains the divergence, would make this analysis less general but then more informative. The idea now is to examine the relevance of a purely philosophical analysis of error, in its general formulation, to an actual, concrete example of error in the experimental sciences. Put differently, I am intent on probing the relation between general philosophical claims on error and analyses of error in philosophy of science. Prima facie the issue should be a straightforward procedure of applying a theory to its case: philosophy provides the theory and science the example. However, as it will emerge shortly, the issue is problematic. I shall argue that we may see here the difference between philosophy at large and philosophy of science: in contrast to philosophy, we do need in philosophy of science to address the concept in context, that is, we need to contextualize the concept in order to establish a meaningful linkage between the theory and its concrete cases. This is the reason why my examples for a general theory of error do not come from the history of scientific error theory (for an overview see Sheynin 1995), e.g., Gauss s mathematical theory of error, but rather from purely philosophical systems such as that of Descartes and Bacon. The immediate consequence of this probing is the demonstration that in philosophy of science history and its material situation which form the context here become essential ingredients for the claim to know the experimental errors that vitiated a certain result. The objective then is to demonstrate via the notion of experimental error the essential role that context plays in philosophy of science as distinct from philosophy at large. I seek to show that in the case of error in science, and more specifically in experiment, analysis cannot remain on an abstract, general level; it has to address also the material situation and its current knowledge, which is in a

257 THE CASE OF ERROR IN EXPERIMENT 255 word history. Put differently, I argue that knowledge claims concerning errors in experiments would have to be related to make them meaningful and valid to a certain context: a structure, a web of elements that cohere into a discipline with definite acknowledged methodologies and certain philosophical presuppositions. To be sure, general philosophical positions are also committed to certain presuppositions and thus are not purely general and do refer to some basic, lean context. However, as I shall try to show, the purely philosophical analysis is not rich enough to be linked insightfully to the problem of error in experimentation. Let us consider two drastically different metaphysically committed philosophical theories of error, namely, the theory of Descartes and that of Bacon. I have chosen these theories since each exhibits an explicit and definite account of error. The objective is to examine the relevance of the theories to documented errors in the history of science. The fact that these are seventeenth century philosophical theories should not pose a problem, as we are interested in examining the relation between the general and the particular. Descartes reported that upon examining closely his own errors, he discovered their causes and moreover the mechanism that bring them about. Errors, Descartes claimed, depend on a combination of two causes, to wit, on the faculty of knowledge that rests in me, and on the power of choice or of free will that is to say, of the understanding and at the same time of the will (Descartes 1973, 174). According to Descartes, the reason for the occurrence of errors is thus due to the sole fact that, since the will is much wider in its range and compass than the understanding, I do not restrain it within the same bounds, but extend it also to things which I do not understand; and as the will is of itself indifferent to these, it easily falls into error... (Descartes 1973, ; cf., ). The crucial philosophical innovation of this theory of error is the introduction of free will into the process of attaining knowledge. This is, of course, much in line with Descartes general epistemological outlook in which the will plays an essential role. The Fourth Meditation from which these quotations are taken, is characteristically devoid of any example or illustration of error. Descartes is interested solely in the general mechanism that brings about error. Clearly, this theory is much richer than the functional definition of error I presented. Descartes theory provides a mechanism on which the general functional definition is silent. It is plain, says Descartes, that abstaining from passing judgment on any thing that one does not perceive with sufficient clearness and distinctness, would not result in error. If however one were to proceed and affirm or deny under these circumstances, then one would misuse one s God given faculty of free will, and one could fall into error. Error, according to Descartes, is an

258 256 GIORA HON expression of privation, the result of the misuse of free will which cajoles the partial understanding to proceed to unclear and indistinct regions (ibid., ). Consider now the following historical fact. In 1906 the experimenter Walter Kaufmann ( ) published in the Annalen der Physik his definitive paper on the constitution of the electron. He detailed his experiment and stated the following conclusion: The... results speak decisively against the correctness of Lorentz s theory and consequently also of that of Einstein s theory. If one were to consider these theories as thereby refuted, then the attempt to base the whole of physics, including electrodynamics and optics, upon the principle of relative motion would have to be regarded at present as also 2 unsuccessful (Kaufmann 1906, 534). According to current physical theories this experimental result is in error. At present, physics the whole of physics is based on the principle of relative motion. Consider further the reactions to this result of three contemporaneous, prominent physicists who were engaged directly with the result of this experiment, namely, Poincaré, Einstein and Lorentz. Poincaré accepted Kaufmann s experimental result though he warned against a possible error. He suggested that an error might have occurred which could pertain to the working of the apparatus. He, however, did not suspend his judgment and assented to the conclusion that classical theories of the electron had been confirmed; hence, the relativity principle could not serve as a solid foundation for physics. Einstein did not find any error; he suspected a systematic error but could not find its source. Indeed, he praised Kaufmann s expertise but all the same rejected the result outright. Einstein was not prepared to accept classical theories, although, as he himself admitted, the calculated curves that these theories had yielded fitted the observed curve considerably better than the curve obtained from the relativity theory. In his opinion, these classical theories should have been ascribed a rather small probability because their basic assumptions about the mass of moving electrons are not made plausible by theoretical systems that encompass a wider complex of phenomena (Einstein 1907, 439). Einstein could not have given up the relativity principle on methodological and aesthetic grounds. Lorentz, for his part, vacillated until some fifteen years later he finally rejected Kaufmann s experimental result. 3 Initially, he had thought that the interpretation was wrong; then he found the experimental result correct and relinquished his own hypothesis. He, however, later argued that the apparatus had not functioned as had been assumed. Moreover, in his opinion, mistakes were made in the reduction of the data (Hon 1995a, ). In sum, the error associated with Kaufmann s experiment assumed at the time different characteristics: error of interpretation, mistakes in reducing the data, and malfunctioning of the apparatus (i.e., the apparatus did not function as it had been assumed it would). Finally, it was claimed that there had been no error at all

259 THE CASE OF ERROR IN EXPERIMENT 257 and equally maintained that a systematic error had occurred whose source was unknown. A consensus eventually emerged that Kaufmann s experimental conclusion was false, and the result of the experiment was pronounced erroneous, but then there was no agreement as to the cause of the failure. In such erroneous cases the alleged empirical findings simply fade away, and there appears to be no need for the concerned scientists to reach an agreement as to the characteristics of the error. However, it is precisely with such cases that the historian and philosopher of science may gain an insight into the practice of science. I have argued elsewhere that these different characterizations of the error in Kaufmann s experiment reflect the specific philosophical outlook and methodological disposition held by each of the protagonists (Hon 1995a). I generalized the Kaufmann case and claimed that the failure of a particular experiment may be attributed to different kinds of error, depending on one s philosophical perspective and methodological disposition. The different sets of epistemological and methodological elements constitute the different contexts in which an error, that is, an experimental error, is determined and characterized. Thus, the identification of an experimental error and its characterization may depend contextually on the philosophical outlook and methodological disposition by which the experiment is examined. My intention was neither to render the concept of error relative nor to explain the occurrence of error in sociological terms; rather, I argued for a close connection between, on the one hand, epistemological framework and methodological approach and, on the other hand, detection and characterization of error. In a word, I argued for contextualization. As I have indicated, I wish in this presentation to take a different direction and to establish a linkage between the case of the erroneous result of Kaufmann and a general philosophical analysis of the concept of error such as that of Descartes. Reflect therefore on the possibility of applying this theory of error to an experimental failure, say, Kaufmann s experiment. In what way could this undoubtedly innovative way of analysing errors help to understand the error in Kaufmann s experiment? It should be stressed that the Kaufmann experiment was regarded in the physics community of the early years of the last century as one of the most important experiments of the time. Planck dedicated much time to the analysis of the result of this experiment and Lorentz stated that Kaufmann s finding is certainly one of the most important results of modern physics (quoted by Hon 1995a, 170; cf., ). So any illuminating observation about error and experimental error in particular would have been most helpful, but it appears that such observation could not have originated in a theory like that of Descartes. I confess that apart from a few trivial remarks about error in general, I fail to see any insightful bearing of this theory on the understanding of error in experimental science let alone on the error in Kaufmann s experiment. One may respond by arguing that Descartes philosophy is not designed to address experimental issues and indeed has no bearing on empirical matters. In any event, how

260 258 GIORA HON could it bear on twentieth century experimental science? According to Descartes theory, knowledge is to be based on the logical, the mathematical and the geometrical. If this, however, were the case then Descartes is not discussing error but rather fallacy. As we know from his other essays (most notably, Principles of Philosophy), Descartes is not concerned solely with fallacies. He speaks for example of the prejudices of our youth and how our mind fatigues itself as sources of error (Descartes 1973, 237, ). I would therefore press on and claim that we have here in the Kaufmann case an attempt at knowledge of the physical world, established albeit erroneously, and yet there is no linkage, insightful linkage that is, between this failure and the analysis of error in Descartes system of philosophy. Are we prepared to be satisfied with the claim that Kaufmann did not restrain his will? What about a faulty evacuation technique that resulted in a poor vacuum in the set-up? Were the plates of the condenser precisely parallel to create a homogenous electric field or did the condenser exert an inhomogeneous field? One may associate these possible sources of experimental error with the overpowering of the will, but I find that not illuminating. It should be stressed however that Descartes theory of error is highly relevant to his own philosophical position and it is in this connection that this theory is insightful and indeed crucial. The theory of error reflects in a negative mode the way Descartes understood the attainment of knowledge, thus the theory may facilitate a clear view of Descartes own metaphysical position (for details see Hon 1995b). In this sense, Descartes system of philosophy provides a context for his theory of error. Let us now turn to another example, to a theory of error which is closely associated with the empirical approach, namely, the idols of Bacon s Novum organum. Bacon was philosophically aware of the problem of error and explicitly addressed it. Indeed, he deployed the notion of error as a lever with which he hoisted his new program for the sciences. As expounded in the Novum organum (Bacon [1620] 1859; 1960; 2000), his programmatic philosophy consists of two principal moves: first, the recognition of error and its rebuke if not elimination, and then the commencing anew of the true science based on experiment and induction, what he called the true way of interpreting Nature (Bacon 1859, 51 (I, 1xix)). I argue that, however innovative, Bacon s approach is found wanting especially when experiment, the very instrument of his research, is in question. Once again a philosophical theory of error presents in the context of science a view of error that is not insightful. Bacon expounds in considerable detail the subject of the obstacles to the true interpretation of nature, before proceeding to unfold his positive program: the method of inductive inquiry based on experimentation. He devotes nearly the whole of the first book of Novum organum to the examination of these obstacles which he calls idols: idols of the tribe, the cave, the market place and the theatre.

261 THE CASE OF ERROR IN EXPERIMENT 259 Let me go briefly through the features of each group of idols. Note the generality of the analysis, but at the same time the inductively driven scheme. Idols of the tribe is errors incidental to human nature in general. The most prominent of these errors are the tendency to support a preconceived opinion by affirmative instances, whilst neglecting all counter-examples; the tendency to generalize from a few observations, and to consider mere abstractions as reality. Errors of this type may also originate in the weakness of the senses, which affords scope for mere conjectures (Bacon 1859, 21, (I, xli, xlv-lii)). Idols of the cave are errors that are incidental to the peculiar mental and bodily constitution of each individual (the cave is a direct reference to Plato s simile in the Republic). These errors may be either of internal origin, arising from the peculiar physiology of the individual, or of external origin, arising from the social circumstances in which one is placed by education, custom and society in general (ibid., 22, 29-30, (I, xlii, liii, lviii)). The third class of idols comprises idols of the market place, that is, errors arising from the nature of language the vehicle, as Bacon puts it, for the association of men, their commerce and consort (ibid., 22-23, (I, xliii, lix, lx)). Language, according to Bacon, introduces two fallacious modes of observing the world. First, there are some words that are merely the names of things which have no existence. Secondly, there are names of things which do exist, but are confused and ill defined (ibid., 34 (I, lx)). Bacon is aware then that by being opaque to nature, language may lead the researcher astray. Finally, idols of the theatre are errors which arise from received dogmas of philosophical systems, and even from perverted laws of demonstrations (ibid., 23 (I, xliv); cf., (I, liv, lxi lxvii)). Here Bacon refers mainly to three kinds of error: sophistical, empirical and superstitious. The first error corresponds to Aristotle who has, according to Bacon, made his Natural Philosophy so completely subservient to his Logic as to render it nearly useless, and a mere vehicle for controversy (ibid., 30 (I, liv; cf., lxiii)). The second error, the empirical, refers to leaping from narrow and obscure experiments to general conclusions. Bacon has in mind particularly the chemists of his time and Gilbert and his experiments on the magnet (ibid., (I, liv, lxiv; cf., lxx)). The third error, the superstitious, represents the corruption of philosophy by the introduction of poetical and theological notions, as is the case with the Pythagorean system (ibid., (I, lxv)). Bacon demands that all the idols be renounced and abjured with a constant and solemn determination (ibid., 49 (I, lxviii)). He insists upon purging and freeing the intellect from them, so that the approach to the Kingdom of Man, which, as Bacon conceived of his quest, is founded on the Sciences, may be like that to the Kingdom of Heaven (ibid.). Thus, having performed, as he writes, these expiations and purgation of the mind, one may come to set forth the true way of interpreting Nature (ibid., 51 (I, lxix)). The religious connotation is explicit and should be underlined.

262 260 GIORA HON Bacon designed the typology to shed light on the nature of sources of error. The scheme of idols presents a systematic and methodical view of the elements involved in the obstruction of knowledge: the interplay of sources of error pertaining to the nature of the mind in general, to individuals and their community, to language and doctrines. Bacon neatly classifies the idols as either adventitious or innate (Bacon 1960, 22 (The Plan of the Great Instauration)). They proceed progressively from the innate to the adventitious, from the most persistent to the easiest to discard. The scheme may appear somewhat artificial and contrived, but it constitutes an essential element of Bacon s comprehensive conception of the emergence of new knowledge and its impediments. In many respects the scheme of idols was innovative and anticipated new disciplines, namely, the study of anthropology, ethnology, psychology, linguistic and cultural, political and religious ideologies (Coquillette 1992, ; for references see 300 footnote 24). To return to my main question: could this all-embracing typology of errors be applicable to the analysis of experimental error such as that of Kaufmann s experiment? Let us be more specific and ask whether the idols can address, within the confinement of Bacon s own system, the very method of research that Bacon advocates for use, namely, experimentation? It will doubtless occur to some, Bacon acknowledges the question, that there is in the Experiments themselves some uncertainty or error; and it will therefore, perhaps, be thought that our discoveries rest on false and doubtful principles for their foundation (Bacon 1859, (I, cxviii)). This appears to be a surprising remark. As befits an empirically based philosophy, this cautious remark suggests a direct link to experimentation, something that cannot be found in Descartes system. Is there then a linkage between Bacon s theory of error and experimental error? The answer is No! Bacon dismisses the threat right away; this is nothing, he exclaims, for it is necessary that such should be the case in the beginning. By way of an analogy he explains that it is just as if, in writing or printing, one or two letters should be wrongly separated or combined, which does not usually hinder the reader much, since the errors are easily corrected from the sense itself. And so men should reflect that many Experiments may erroneously be believed and received in Natural History, which are soon afterwards easily expunged and rejected by the discovery of Causes and Axioms (ibid., 112 (I, cxviii)). Bacon assures us that we should not be disturbed by these objections and he reiterates this confidence in his outline for experimental history (Bacon 1960, 280 (viii)). However, he admits that it is true, that if the mistakes made in Natural History and in Experiments be important, frequent, and continuous, no felicity of wit or Art can avail to correct or amend them (Bacon 1859, 112 (I, cxviii)).

263 THE CASE OF ERROR IN EXPERIMENT 261 Thus, if there lurked at times something false or erroneous in Bacon s Natural History which have been proved with so great diligence, strictness, and, Bacon adds, religious care, what then must be said, he asks rhetorically, of the ordinary Natural History, which, compared with ours, is so careless and slipshod? or of the Philosophy and Sciences built on quicksands (ibid.)? Notwithstanding Bacon s resolute assurance and his confidence in his method, the objections are disturbing. Bacon does not provide convincing arguments in defense of his position. He would have us believe that the analogy between a printer s error and an experimental error is faithful. However, it is precisely the sense of the context the meaning, which according to Bacon s analogy is given that the experimental sciences lack and in fact seek to discover. The blackening of photographic plates in close vicinity to cathode-ray tubes had been a well known nuisance before Röntgen perceived the right context in which he transformed this nuisance into a great discovery that revolutionized experimental physics. The two types of error, namely, the printer s and the experimental, are categorically different. (I distinguish elsewhere between these two possible faults, which I call mistake and error respectively (Hon 1995b).) Further, would Bacon claim that Kaufmann was subject to an idol of the market place, or was it of the theatre. In any event, the realization of the error to which Kaufmann had fallen did not emerge from repeating the experiments and reading the text more carefully in the right context. Rather, new experiments with new designs and better evacuating techniques became available that confirmed Einstein s theory of special relativity and thereby rendered Kaufmann s results erroneous (Hon 1995a). Surprisingly, it appears that Bacon did not apply consistently his critical scheme of errors to the very instrument of his inquiry experiment. Admittedly, he was concerned with errors that beset the mind: once one had purged one s mind from the idols, one was ready to embark on the true way of interpreting nature. At issue here is not whether this instruction to cleanse one s mind of idols is practicable or not, but rather can the instrument of one s inquiry be itself an object of critical scrutiny. Indeed, it had taken some time before the question: What exactly is an experiment in physics? was explicitly raised and addressed in a philosophical context. But this context would be already philosophy of science Duhem s 1906 The Aim and Structure of Physical Theory (Duhem 1974, 144). Bacon s typology was innovative: it addresses sources of error and locates their context in a systematic way. We see here a philosophical theory that declares its aim to establish a linkage to the brass tacks of science. Nevertheless, the persistent impediment, which the occurrence of errors poses the attainment of knowledge from experiment, cannot be completely covered by Bacon s scheme of idols of the mind. While elements that have to do with the background theory of experiment and its interpretation may be addressed by Bacon s system of idols, there are other elements that cannot be covered by the theory such as the mal-functioning of the instrument or taking inaccurate observations and meas-

264 262 GIORA Hoeresy urements. Once again the question arises as to how the theory acknowledges and addresses, for example, problems with instrumentation: the inhomogeneous field that probably vitiated Kaufmann s calculations. In conclusion, we see here two philosophical theories of error that cannot be linked in an insightful way to the particulars of error in the experimental sciences. Generalizing this observation, I suspect that one would not find philosophical theories, and I stress philosophical theories as distinct from scientificmathematical theories of error, that have a bearing on the impediments of error in experiments. I leave it to the reader to take up the challenge and to come up with a counter-example. What I hope to have established is an indication of a lack of an insightful linkage between philosophy at large and the occurrence of error in experimental sciences, between the general theory and its presumed case. Error is an epistemological phenomenon, but it appears that general epistemological theories that do address the issue of error cannot provide insights into error in science and especially experimental science. To establish such linkage, one needs a theory of error that brings into consideration the material context and its knowledge that is, history. I submit that here, from this perspective of experimental error, we can point to a definite difference between philosophy at large and philosophy of science. While philosophy seeks to liberate itself from context in search for generality, philosophy of science renders context a necessity, and turns it to its advantage. ACKNOWLEDGEMENTS I thank Jutta Schickore for organizing the Context session and for inviting me to take part in it. I am indebted to her and to Gregor Schiemann for instructive comments on an earlier version of this paper. NOTES Notice that lies too consist of separation but there the gap is intentional. Die Ergebnisse sprechen entschieden gegen die Richtigkeit der Lorentzschen und somit auch der Einsteinschen Theorie; betrachtet man diese aber als widerlegt, so wäre damit auch der Versuch, die ganze Physik einschließlich der Elektrodynamik und der Optik auf das Prinzip der Relativbewegung zu gründen, einstweilen als missglückt zu bezeichnen. It is worth noting from a sociological point of view that Einstein s reputation was at that time newly minted and the relativity theory was one of its central pillars. By contrast Lorentz s reputation had already been secured whatever the outcome of Kaufmann s experiment. He had after all founded particle-based electrodynamics.

265 THE CASE OF ERROR IN EXPERIMENT 263 REFERENCES Bacon, F., [1620] 1859, Novum organum, translated by A.Johnson (London: Bell & Daldy Fleet Street). Bacon, F., [1620] 1960, The New Organon, edited with an Introduction F.H.Anderson (Englewood Cliffs, New Jersey: Prentice Hall). Bacon, F., [1620] 2000, The New Organon, edited by L.Jardine and M. Silverthorne (Cambridge: Cambridge University Press). Coquillette, D. R., 1992, Francis Bacon (Stanford, California: Stanford University Press). Descartes, R., 1973, Philosophical Works, rendered into English by Elizabeth S.Haldane and G. R.T. Ross in two volumes, Vol. 1 (Cambridge: Cambridge University Press). Duhem, P., [1906] The Aim and Structure of Physical Theory (New York: Atheneum). Einstein, A., 1907, Über das Relativitätsprinzipund die aus demselben gezogenen Folgerungen, Jahrb. Radio, u. Elek. 4, Hon, G., 1989, Towards a Typology of Experimental Errors: an Epistemological View, Studies in History and Philosophy of Science 20, Hon, G., 1995a, Is the Identification of Experimental Error Contextually Dependent? The Case of Kaufmann s Experiment and Its Varied Reception, in Buchwald, J. (ed.), Scientific Practice: Theories and Stories of Doing Physics (Chicago: Chicago University Press), pp Hon, G., 1995b, Going Wrong: to Make a Mistake, to Fall into an Error, Review ofmetaphysics 49, Kaufmann, W., 1906, Über die Konstitution des Elektrons, Annalen der Physik 19, Lewis, C.T. and Short, C., 1966, A Latin Dictionary (Oxford: Clarendon Press). Sheynin, O., 1995, Density Curves in the Theory of Errors, Archive for History of Exact Sciences 49, Department of Philosophy University of Haifa Haifa Israel hon@research.haifa.ac.il

266 JUTTA SCHICKORE THE CONTEXTS OF SCIENTIFIC JUSTIFICATION. SOME REFLECTIONS ON THE RELATION BETWEEN EPISTEMOLOGICAL CONTEXTUALISM AND PHILOSOPHY OF SCIENCE Problems of knowledge and justification are central both to philosophy of science and to analytic epistemology. By analytic epistemology I mean here the North American tradition of epistemology that has developed out of the Viennese logical empiricist philosophy of science after the Vienna Circle s emigration to America and Canada. Analytic epistemology has since become a philosophical enterprise separate from philosophy of science. In fact, these fields of study have drifted so far apart that it is sometimes hard to think that the two have common origins. Particularly the debates about knowledge in these fields seem to have not much in common: While philosophers of science have been concerned with refining and revising models for the reconstruction and assessment of scientific theories, analytic epistemologists have almost exclusively debated the necessary and sufficient conditions of knowledge as justified true belief. 1 Gettier s problem 2 has dominated the discussion. Each time a new set of conditions is suggested, a novel Gettier-type counter-example emerges. Nowadays, Gettier is a label for epistemologists activities rather than a proper name. 3 In recent years, significant re-orientations have occurred in both analytic epistemology and philosophy of science. In analytic epistemology, the focus has shifted from analyses of the necessary and sufficient conditions of knowledge to the issue of justification itself, and analytic philosophers have begun to consider the structure of epistemic justification, its internalist or externalist nature, 4 its sources and limits. 5 In philosophy of science, formal models for the reconstruction and assessment of theories have lost their vigour. Instead, philosophers of science have begun to acknowledge that scientific rationality has to do with giving and asking for reasons, making value-laden decisions, and so on. In both fields, however, a new topic has emerged, one which might help to bring the two traditions closer together again. This is the topic of context: scholars in both analytic epistemology and philosophy of science have begun to pay attention to contextual factors. They concede the context-relativity of epistemic concepts and argumentation. More specifically, they argue that the validity of justification procedures depends on various features of context rather than on absolute epistemic criteria F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

267 266 JUTTA SCHICKORE In the following paper, I examine whether the contextualist account of justification that analytic epistemologists have developed can be employed in philosophy of science. I begin with a brief synopsis of the debates about justification in philosophy of science, and I describe the attempts at a well-tempered, contextualised epistemology that have recently emerged. I then focus on the notion of contextualised justification that has taken shape in some branches of analytic epistemology. Finally, I consider whether this kind of epistemological contextualism could be fruitfully applied to accounts of theories of justification in philosophy of science. TRANSFORMATIONS IN -CENTURY THEORIES OF SCIENTIFIC JUSTIFICATION In the first half of the century, formal models for the reconstruction of scientific theories dominated philosophical theories of scientific justification. According to the so-called Received View, justification was a matter of logical argument and inference. 7 The criteria of justification were taken to be reducible to explicit sets of formal rules of inference, be it inductive or deductive. Philosophy of science had a distinctly normative and critical function: It was the task of philosophy of science to justify scientific theories by exhibiting the formal relationships between a set of theoretical statements and the available evidence. The set of criteria for the rational reconstruction of theories, it was assumed, was universally valid. That is, the criteria governing the justified acceptance of scientific claims were not open to abandonment, modification, or replacement in the light of new knowledge. As is well known, formal models of scientific justification came under severe attack in the 1960s and 1970s when Quine, Putnam, Kuhn, Feyerabend, Polanyi & Co. naturalised and historicised epistemology and formulated alternative models of scientific rationality and change. These alternative views of the structure and development of science also implied new accounts of scientific justification. Notably, the critics attacked quite different features of the by-then-dominant, received model of justification. It will be convenient to distinguish between four strands of argument: 1. Arguments targeting the internal consistence of the reconstructions provided by the Received View; 2. arguments targeting the logicism inherent in it; 3. arguments attacking the methodological apriorism implied in the Received View; and 4. arguments questioning its universalism. 1. As even orthodox logical empiricists like Herbert Feigl admitted, there were several problems with the Received View, such as the various problems of axiomatization or the difficulties in developing a sound notion of confirmation (see Feigl 1970), some of which proved unsolvable. 2. More fundamentally, Norwood Hanson, Michael Polanyi, Thomas Kuhn, and others rejected the virtual identification of justification, rationality, and logical inference implied in the Received View. 8 They denied that justification could

268 THE CONTEXTS OF SCIENTIFIC JUSTIFICATION 267 be completely reduced to any explicit set of formal rules of inference. Kuhn, for example, suggested that scientific theories or rather, paradigms could be assessed against a background of cognitive, epistemic values, such as simplicity and fruitfulness. The application of these values, however, cannot be formalised: they must be interpreted and weighed against each other whenever a choice is required Yet other philosophers, among them Imre Lakatos and Larry Laudan, 10 expressed critique from a different, more general angle. They doubted the possibility of successfully constructing an ultimate a priori proof to show that any set of standards for the justification of knowledge claims is universally valid. But if there is no such a priori proof of epistemology s evaluative principles, then the question arises as to how the standards of justification can themselves be justified. These scholars have sought to show how historical studies of science could serve as constraints for our epistemological standards of theory assessment. 4. In emphasising the impossibility of a priori epistemology and the undeniable importance of socio-cultural factors for the advancement of science, some scholars have taken the argument further and have denied that there are any universal standards of justification. Notoriously, Feyerabend advocates the view that there are no valid standards of justification besides the ones which scientists themselves use in a given scientific episode, and that these locally valid criteria have changed again and again over the course of time. 11 According to this view, epistemology must be confined to the study of science, past and present, in its cultural and social settings. Studying scientific justification would then mean reconstructing the standards of assessing knowledge claims from the actions and utterances of scientists in a given period. However, the debates about justification have taken yet another turn. Particularly the critics of formal models of justification who took up the last three of the arguments against the Received View have themselves been criticised for advocating positions that completely abandon the priority of epistemology to the sciences. Several philosophers, among them Harvey Siegel, Dudley Shapere, and Richard Burian, have claimed that historicised and naturalised epistemologies eventually lead to the dissolution of epistemology proper. 12 A study of what scientists take to be justified or unjustified knowledge claims, the argument goes, can tell us nothing as to whether the claims are indeed justified, that is, whether the reasons the scientists give for their claims are good ones. 13 Taken together, the transformations of the philosophical discussions about justification indicate that to arrive at a sound account of scientific justification, one needs to steer clear of both the extremes of the Received View and its criticisms and indeed, Burian and others have aimed at an epistemology that occupies a position between the poles. 14 As Burian and Shapere have argued, such an epistemological middle ground can be found by contextualising philosophical accounts. Taking up the second strand of argument against a purely formal account of justification, Burian shows that the assessment of knowledge claims is dependent on historical information. Building on Lakatos s methodology of

269 268 JUTTA SCHICKORE scientific research programmes, Burian argues that the formal reconstruction of justification procedures is not inappropriate but incomplete. The criteria for evaluating theories must include tools for understanding the plasticity of a theory; otherwise, they cannot evaluate its explanatory and predictive power and fitness. To understand these, historical information regarding the theoretical background; the status of neighboring and auxiliary theories; the temporal order among experimental discoveries, theoretical derivations and predictions by competing theories; and so on 15 is required. Historical information regarding the continuity of different versions of a theory helps evaluate its fruitfulness and predictive success. Proper evaluation of a theory is, as Burian puts it, contextual l6. It is its specific, concrete history that bears on its epistemic status, and thus provides the rich and multifaceted context of justification. Taking up the third and fourth strand of argument against the Received View, Shapere claims that the epistemological criteria of assessing knowledge claims are affected by the growth of scientific knowledge and change accordingly. He emphasises that it is not only the factual scientific beliefs that undergo alterations, but also epistemological conceptions such as the criteria of what counts as an explanation and the criteria of application of scientific terms. These alterations are intimately connected with the content of scientific belief. 17 Acknowledging that science has a bearing on justification, both projects thus point in the same direction, namely, towards a well-tempered epistemology. However, as there are several different strands of argument against formal methods of justification which imply different alternative models, it is not quite clear between which poles the position could be located. Indeed, well-tempered epistemologists position themselves between various extremes. Burian, for example, calls for an epistemology that takes a middle ground between historicism and logicism. Pure historicism conceived as a descriptivist position, he writes, is no more viable than pure logicism. We must be concerned, therefore, to locate the soundest position between these poles and to occupy it. 18 In contrast, Dudley Shapere advocates a view of science which, while attempting to avoid the absolutes of Platonic-Kantian-positivist ways of thinking, also endeavours to escape a relativism in which there is no such thing as knowledge and progress. l9 Shapere thus attempts to find a middle ground between apriorism and relativism. On closer examination, however, the two projects are actually concerned with two different levels of justification. While Burian seeks to develop a new model of justification that acknowledges the context-dependence of theory assessment, Shapere focuses on the possibility of justifying the justification criteria, that is, the framework for reconstructing theories. In other words, Shapere contextualises the philosophical account on a new level: In his account, the scientific enterprise provides the context of justification in that it has a bearing on the criteria of justification. Taken together, the criticisms put forward against the Received View of scientific justification indicate that the development of a convincing model of

270 THE CONTEXTS OF SCIENTIFIC JUSTIFICATION 269 justification is not just a question of finding a middle way between two extremes, such as logicism vs historism or apriorism vs relativism. Rather, the philosophical account of justification must be positioned in a multidimensional space: First, it must offer a plausible and coherent model of scientific justification and justified knowledge that is located within the triangle of logicism, historicism, and universalism: It must offer a general framework that does not reduce justification to mere logical inference, and that is sensitive to the scientific context itself without totally naturalising or historicising epistemology. That is, the study of science should not be completely reduced to a merely descriptive narration of scientists activities. But secondly, it is also essential that the approach provide arguments for the justificatory force of the framework of reconstruction. These arguments must avoid both apriorism and relativism: The justification of the framework should abandon all attempts at ultimate proofs of epistemology s evaluative principles. Yet to retain the normative thrust of epistemology, the approach must explain why a certain framework for reconstruction is appropriately considered justificatory. It is particularly this last requirement that is at stake in all attempts at a well-tempered epistemology. BEYOND THE EXTREMES: ANALYTIC CONTEXTUALISM Analytic epistemology has been dominated by attempts at coming to grips with the skeptical challenge, that is, with the skeptic s claim that there is no knowledge about the external world. There is a striking parallel between the welltempered approaches in philosophy of science and contextualism in analytic epistemology: The contextualist notion of justification that has taken shape in recent years has also been introduced as a well-tempered alternative to two extremes, namely, foundationalism and coherentism. Contextualism in analytic epistemology is meant to provide an analysis of our day-to-day epistemic practices and strategies of justification beyond these extremes. Both foundationalism and coherentism have been the target of the skeptic: foundationalism holds that every empirical statement to be justified must ultimately derive at least some of its justification from a class of basic statements which are justified independent of the support such statement may derive from other statements. According to coherentism, a statement is justified if and only if it coheres with a certain kind of system of statements. Both positions are vulnerable to skeptical arguments. Skeptical anti-foundationalists maintain that there is no such class of basic statements, whereas skeptical anti-coherentists hold that coherentism cannot distinguish a consistent fairy-tale from knowledge. The contextualist weakens the requirement that a statement is justified if it coheres with the whole belief system. 20 Justification does not involve the whole belief system but only a significant array of knowledge claims. Contextualism seeks to bypass the skeptical challenge: The contextualist concedes that there is no class of basic statements that are justified independent of the support such

271 270 JUTTA SCHICKORE statements may derive from other statements. Rather, individual statements do not have a fixed epistemic status: they can serve as basic statements in one case, while being in need of justification in another. Several analytic philosophers have argued that the context in which a knowledge claim is assessed plays a crucial role in the assessment of whether or not this claim can be regarded as justified. Contextualism in contemporary analytic epistemology defends the position that the context of a discourse affects the epistemic standards that a speaker must meet (or fails to meet, respectively) in order for a statement to be true. Contextualism about justification implies that the standards for justified belief that a subject must meet in order to render true a sentence describing a belief of hers as justified vary with context. 21 This reference to context serves as an argument to accommodate skepticism about our knowledge of the external world. As Michael Williams has argued, there is no such thing as our knowledge of the world in the sense of a body of knowledge claims that has to be assessed as a whole. 22 Rather, we need to distinguish between two kinds of discourse: philosophical and ordinary ones. It is assumed that highly specialised philosophical inquiries demand stricter standards for knowledge than ordinary discourse. In epistemological Contextualism, the notion of context thus refers to the level of discourse. The level of discourse is characterised by the strictness of standards that are applied in the justification of claims to knowledge. The idea that skeptical arguments do not threaten ordinary claims to knowledge is a common feature of the various brands of Contextualism. To what extent can analytic Contextualism illuminate scientific justification? One might object that this approach does not lend itself to the specific problems of philosophy of science because epistemic contextualism merely emphasises a pragmatic distinction that philosophers of science have tacitly presupposed all along the implication being that philosophers of science and of scientific justification in particular need not concern themselves with skeptical arguments. These fundamental doubts about our knowledge of the external world are bracketed in philosophy of science one might even say that philosophical theories about science are generally neutral vis-a-vis these possibilities. However, a few contextualists take contextualism a step further. They argue that the body of knowledge of the world must be further divided. The standards of justification for ordinary claims to knowledge are also sensitive to context but they are standards of justification, nonetheless. It is this additional step that makes a contextualist theory of justification appear to be a good candidate for a sound position between the poles. RULING OUT RELEVANT ALTERNATIVES In the following, I concentrate on contextualism of the relevant alternatives kind, which Fred Dretske and others have developed. 23 This kind of contextualism seeks to formulate lower, more adequate standards for knowledge than

272 THE CONTEXTS OF SCIENTIFIC JUSTIFICATION 271 traditional analytic epistemology by allowing for various kinds of contextual factors that determine what need not be justified. It is assumed that a statement is justified if all relevant alternatives can be ruled out; whereby relevant alternatives are all relevant ways in which the statement might turn out to be wrong. 24 To cast doubt on a particular knowledge claim, we can formulate numerous contradicting hypotheses that invalidate the knowledge claim in question, because we can always think of numerous things that could be the case, thereby rendering this knowledge claim not justified. 25 These alternative possibilities are all relevant to the truth of a knowledge claim in the sense that, if they were realised, one would not know what one claims to know. Yet these are not possibilities that one need be in an evidential position to exclude. 26 These possibilities are simply irrelevant. A knowledge claim is justified if we can rule out all relevant ways in which we might end up being wrong. The justification of knowledge depends on the context in such a way that the context determines which of our doubts are relevant. How does all this apply to scientific cases? Contextualists rarely talk about scientific investigations, and if they do, they often consider only very trivial examples. 27 If we want to assess the merits of analytic contextualism, we need to extend its scope and consider its main claims with respect to more complex episodes. One of the initial considerations would be this: On which level should we place the epistemic practice of science? Only if it is placed on the level of everyday epistemic practice, instead of on the level of specialised philosophical inquiry, the contextualist approach offers a promising alternative to recent philosophical models. One could then say that to justify her claim, a scientist need not rule out skeptical possibilities. For example, in a basic experimental situation where a scientist accumulates pointer readings, she is not required to rule out the possibilities that she is led astray by a malicious demon, that the instrumentation might not continually exist, or that perceptions are merely confused judgments and thus not an adequate means of acquiring knowledge although if they were, or if the demon did interfere, the scientist s claim to knowledge would not be true. Possibilities like these can be considered irrelevant alternatives. If we present the epistemic practice of science as a special kind of ordinary practice rather than of rigid philosophical argumentation, we unburden philosophy of science from skeptical concerns. 28 Yet to most philosophers of science, this move is so obvious that it verges on the trivial. If at all, the merits of the contextualist account must thus lie in the second step of contextualisation, where the context is introduced as the argumentative environment of a particular knowledge claim. Contextualists hold that justification does not even require ruling out any and all possible alternatives besides the skeptical possibility. Only some counterclaims actually challenge the reliability of our knowledge claims, namely, those counterclaims that are embedded in a set of reasons why these particular alternatives should be recognised as significant. As Michael Williams puts it, all inquiries, including scientific ones, have a certain direction that is determined by what counts as relevant possibilities of error. Contextual factors

273 272 JUTTA SCHICKORE determine not only the level but also the angle of scrutiny. 29 This, then, is another respect in which context is invoked: With respect to ordinary discourse, context is introduced as the appropriately motivated challenges that determine the direction of scientific inquiry. The existence of a specific, appropriately motivated challenge provides the context of justification in that it not only determines whether evidential justification is required at all but also what exactly it is that is to be justified. The point is that the justification of a claim to knowledge depends on what exactly is at stake in a certain research environment. MERITS AND SHORTCOMINGS OF A CONTEXTUALIST THEORY OF SCIENTIFIC JUSTIFICATION Is this a promising perspective for philosophers of science? As I have noted above, the question is actually twofold: Can we formulate a convincing contextualist account of scientific justification, and does the model convincingly explain the justificatory force of justification procedures? In the following, I can offer only a few considerations as to whether epistemological contextualism can help solve some of the problems of a theory of scientific justification. To fully analyse the merits of the contextualist framework, one would need to examine specific cases in greater detail. Contextualism provides a separate framework for the reconstruction of a justification procedure that allows for a systematic account of how it is that the context bears on the justification of a claim to knowledge: The contextualist approach requires us to reconstruct how a particular challenge arises and how the sets of available evidence and background assumptions divide up into candidates for relevant and irrelevant alternatives. One might wish to object that contextualism is yet another version of falsificationism, because it reconstructs justification procedures as procedures of fending off alternatives. But I think that contextualism allows for a sophisticated falsificationism that avoids the difficulties of both Popper s and Lakatos s falsificationist accounts. First of all, as many critics, including Lakatos, have pointed out, Popper s strong falsificationist programme does not leave room for the exploration and development of the various aspects of a proposed theory. Precisely to acknowledge the constructive aspects of theory generation, Lakatos introduced his method of research programmes The contextualist model also emphasises the constructive side of research. Contextualists hold that the practice of justifying is only activated by finding oneself in the context of a properly motivated challenge. 30 Yet while Lakatos claims that to be significant, attempts at falsification must be embedded in a competing research programme, 31 contextualism operates on a smaller scale. To be relevant, challenges need not be embedded in a comprehensive research programme. Scientific problems are justified by piecemeal strategies that address only parts of a large-scale set of statements. For example, to be relevant, doubts

274 THE CONTEXTS OF SCIENTIFIC JUSTIFICATION 273 can be couched in an argument challenging the means of obtaining empirical data. 32 The contextualist framework thus seems particularly apt to deal with the intricacies of experimental practice. But how well does contextualism fare with respect to the justificatory force of its framework? To what extent can the relevant alternative account of justification itself be justified? The contextualist account gains its thrust by reconstructing scientific justification practice from the perspective of everyday practice. It thus hinges on the very concept of knowledge and justification we commonly apply. Our practice of epistemic evaluation both in everyday and scientific situations ascribes knowledge to justified (true) knowledge claims, that is, to knowledge claims which can resist reasonable doubts, doubts which raise a relevant possibility of error. One might say that contextualism is both externalist and internalist. It is externalist because it is based on an explication of knowing that is not extracted from a study of science, and it is internalist because it is based on a reflexive analysis of our own practice of epistemic evaluation manifested in our usage of the terms knowledge and justification. In this sense, contextualism lays claim to normativity. It is normative in that it seeks to specify the general features of acceptable scientific justification procedures by elucidating the normative elements of our everyday epistemic practices. Another merit of contextualism is that it does not require us to begin our analysis with the identification of the context that bears on scientific justification. It is the reconstruction of the properly motivated challenges that comes first. Contextualism therefore bypasses a critique which some historians have put forward namely, that context is a virtually meaningless notion: We can forever expand the list of factors bearing upon epistemology from the immediate surroundings of certain individual philosophers to remote traditions and cultural resources in which philosophical arguments are embedded the relevant community of discourse may include all of Western civilization. And more, the historian David Harlan writes in his critique of historical contextualism. 33 In contextualism however, it is the identification of the sum of alternatives, of the challenges which are considered relevant in a specific situation, that outline the context of justification. However, the plausibility and success of the contextualist analysis hinges on the contextualist s ability to provide an explication of the notion of relevance. Only if the contextualist can formulate criteria to single out particular alternatives, will she provide more than a mere analytic description of normative practices. As yet, the explication of relevance is wanting. It is tempting to say that relevance is a matter of probability, in the sense that only probable alternatives must be ruled out. But this alone does not solve the problem, because one need to know just how probable an alternative must be, and how degrees of probability could be determined. 34 It might be more promising to analyse the notion of relevance by confronting ordinary and scientific practice with each other. Everyday and scientific practice are not identical. While it is in principle convincing to maintain that scientific justification requires the refutation of the

275 274 JUTTA SCHICKORE contextually relevant rather than of all possible doubts, the epistemic standards of science are, after all, somewhat stricter than everyday ones. A reflexive reconstruction of the standards of scientific justification from the perspective of those of ordinary discourse might help illuminate our criteria of relevance. Without doubt, the lack of an explication of relevance is one of the major desiderata of epistemological contextualism, if applied to philosophy of science. Nevertheless, I think that the relevant alternatives account of justification has some potential particularly as a framework for reconstructing scientific justification and even more so experimental reasoning, and I hope to have offered some ideas as to how such a framework might look like. Last but not least, the framework of epistemological contextualism offers a promising opportunity to once again unite analytic epistemology and philosophy of science. ACKNOWLEDGEMENTS I thank Gregor Schiemann and Giora Hon for helpful comments on earlier versions of this paper. NOTES Wolfgang Spohn has even claimed that the analytic conception of knowledge is completely irrelevant to the formal models of justification. (See Spohn 2001, 36.) See Gettier For more on the discussions about the Gettier problem, see Fogelin Among the most recent Gettiers are Cheyne 1997, Engel Jr 2000, Hetherington In analytic epistemology, internalism means the position that the epistemic subject must have privileged access to the reasons for beliefs, while externalism denies this condition. See Alston In contrast, in philosophy of science, externalism refers to a reconstruction of science that is based on a general theory of knowing or a logic of confirmation or demonstration. Internalist philosophy of science is based on the presupposition that the very success of science represents a sufficient reason for taking it as an object of study. On externalist and internalist philosophy of science, see McMullin 1970, section 4. For an overview over recent developments in analytic epistemology, see Grundmann For more on context and justification in philosophy of science, see, among others, Hacking 1999, Laudan 1996, esp. 132, Echeverrìa 1995, Peursen On contextualism in recent analytic philosophy, see, among others, Willaschek 2000, Stalnaker 1999, Longino 1999, Taschek I use Putnam s 1962 term Received View as an umbrella term for the by then dominant view that scientific theories are to be construed as axiomatic calculi and must be interpreted by means of correspondence rules. (For a comprehensive account of the Received View, see Suppe 1977, ) See, e.g., Hanson 1965, Polanyi 1966, Kuhn For more on cognitive values, see Kuhn 1970, esp. Postscript, sect.2. See, e.g., Lakatos 1974, Laudan See Feyerabend 1978.

276 THE CONTEXTS OF SCIENTIFIC JUSTIFICATION See Siegel 1980, Shapere 1980, Shapere 1977, Burian See Siegel 1980, 319. Burian 1977, 1 1 ; see also Nickles 1980, 49, and Leplin 1987, 806f. Burian 1977, 39. Burian 1977, 34. Shapere 1980, 68. Burian 1977, 11. Shapere 1980, 95. See Annis 1978, 213. (De Rose 1999, 189) Other versions of analytic contextualism include the semantic analysis of the truth conditions of knowledge attributions. In the following, 1 do not consider these versions but concentrate on epistemological analysis. Williams 1991, xx. For more on the relevant alternatives account of contextualism, see, among others, Dretske 1970, Dretske 1972, Dretske 1981, Willaschek See Williams 1991, 185. The whole of these alternatives forms the set of statements that are necessarily eliminated by what is known to be the case. (See Dretske 1981, 371) See Dretske 1981, 374. See, e.g., Annis 1978, 215, Sanford 1981, 381f., and Willaschek 2000, 59f. But the question of how scientific discourse could be distinguished from ordinary practice is still open. Williams 1999, 59; see also Williams 1988, 428f. Williams 1999, 53. Cf. Lakatos Dretske maintains that [t]he way we come to know, the channel (so to speak) over which we obtain our information, is, 1 submit, always the locus of irrelevant alternatives. Others can challenge the reliability of this channel (our visual system, our auditory system, the newspapers, the pocket calculator), and if it turns out to be unreliable in some way they will thereby have discredited our claim to knowledge. But others cannot discredit our claim to knowledge merely by pointing out that the channel over which we received our information could be unreliable or that we do not know it to be reliable. (Dretske 1981, 374) It is unclear to me whether Dretske holds that such challenges will never be relevant. If so, then he is plainly mistaken, especially in the case of scientific practice, where the ever-increasing complexity of the channels the scientists design to acquire information makes them susceptible to disturbances and noise. The way in which we acquire knowledge is one major locus of potentially relevant alternatives. Admittedly, Dretske does not discuss science but merely everyday experience. See also David Sanford s comment on Dretske s paper (Sanford 1981, 381). Harlan 1989, 594. For more on relevance and probability, see Sosa 1988, 143ff. REFERENCES Alston, W. 1989, Internalism and Externalism in Epistemology, in: Epistemic Justification, Alston, W. (Ed.), Ithaca, Annis, D. 1978, A Contextualist Theory of Epistemic Justification, in: American Philosophical Quarterly 15 (3): Burian, R.M. 1977, More Than a Marriage of Convenience: On the Inextricability of History and Philosophy of Science, in: Philosophy of Science 44: Cheyne, C. 1997, Epistemic Value and Fortuitous Truth, in: Principia 1 ( ). De Rose, K. 1999, Contextualism, in: Blackwell's Guide to Epistemology, Sosa, E. and J.Greco (Eds.), Malden and Oxford: Blackwell. Dretske, F.I. 1970, Epistemic Operators, in: The Journal of Philosophy 67 (24):

277 276 JUTTA SCHICKORE Dretske, F.I. 1972, Contrastive Statements, in: The Philosophical Review: Dretske, F.I. 1981, The Pragmatic Dimension of Knowledge, in: Philosophical Studies 40: Echeverrìa, J. 1995, The Four Contexts of Scientific Activity, in: Theories and Models in Scientific Processes, Poznan Studies in the Philosophy of Science and the Humanities, vol. 44, Herfel, W.E. et al. (Eds.), Amsterdam: Rodopi, Engel Jr, M. 2000, Internalism, the Gettier Problem, and Meta-epistemological Skepticism, in: Grazer Philosophische Studien 60: Feigl, H. 1970, The 'Orthodox View' of Theories: Remarks in Defense as Well as Critique, in: Analyses of Theories and Methods of Physics and Psychology, Minnesota Studies in the Philosophy of Science, vol. IV, Radner, M. and S. Winokur (Eds.), Minneapolis: University of Minnesota Press. Feyerabend, P. 1978, Against Method: Outline of an Anarchistic Theory of Knowledge, London: Verso. Fogelin, R. 1994, Pyrrhonian Reflections on Knowledge and Justification, New York and Oxford: Oxford University Press. Gettier, E. 1963, Is Justified True Belief Knowledge?, in: Analysis 23: Grundmann, T. 2001, Erkenntnistheorie. Positionen zwischen Tradition und Gegenwart, Paderborn: mentis. Hacking, I. 1999, Historical Meta-Epistemology, in: Wahrheit und Geschichte, Carl, W. and L.Daston (Eds.), Göttingen: Vandenhoek und Ruprecht, Hanson, N. R. 1965, Patterns of Discovery, Cambridge: Cambridge University Press. Harlan, D. 1989, Intellectual History and the Return of Literature, in: The American Historical Review 94 (3): Hetherington, S. 2001, A Fallibilist and Wholly Internalist Solution to the Gettier Problem, in: Journal of Philosophical Research 26: Kuhn, T. S. 1970, The Structure of Scientific Revolutions, 2., Chicago: The University of Chicago Press. Lakatos, I. 1970, Falsification and the Methodology of Scientific Research Programmes, in: Criticism and the Growth of Knowledge, Lakatos, I. and A.Musgrave (Eds.), London: Cambridge University Press, Lakatos, I. 1974, History of Science and its Rational Reconstructions, in: The Interaction Between Science and Philosophy, Elkana, Y. (Ed.), Atlantic Highlands, NY: Humanities Press, Laudan, L. 1984, Science and Values. The Aims of Science and Their Role in Scientific Debate, Berkeley: University of California Press. Laudan, L. 1996, Beyond Positivism and Relativism: Theory, Method, and Evidence, Boulder, Colorado: Westview Press. Leplin, J. 1987, The Bearing of Discovery on Justification, in: Canadian Journal of Philosophy 17 (4): Longino, H. E. 1999, Feminist Epistemology, in: The Blackwell Guide to Epistemology, Greco, J. and E.Sosa (Eds.), Malden and Oxford: Blackwell, McMullin, E. 1970, The History and Philosophy of Science: A Taxonomy, in: Historical and Philosophical Perspectives of Science, Stuewer, R. H. (Ed.), New York et al.: Gordon and Breach, Nickles, T. 1980, Introductory Essay: Scientific Discovery and the Future of Philosophy of Science, in: Scientific Discovery, Logic, and Rationality, Nickles, T. (Ed.), Dordrecht: Reidel, Peursen, C.A.v. 1989, Discovery as the Context of any Scientific Justification, in: Man and World 22: Polanyi, M. 1966, The Tacit Dimension, New York: Doubleday and Company. Sanford, D.H. 1981, Knowledge and Relevant Alternatives: Comments on Dretske, in: Philosophical Studies 40: Shapere, D. 1977, What Can the Theory of Knowledge Learn From the History of Knowledge?, in: The Monist 60: Shapere, D. 1980, The Character of Scientific Change, in: Scientific Discovery, Logic, and Rationality, vol. I, Nickles, T. (Ed.), Dordrecht: Reidel, Siegel, H. 1980, Justification, Discovery and the Naturalizing of Epistemology, in: Philosophy of Science 47:

278 THE CONTEXTS OF SCIENTIFIC JUSTIFICATION 277 Sosa, E. 1988, Knowledge in Context, Skepticism in Doubt. The Virtue of our Faculties, in: Philosophical Perspectives 2 (Epistemology): Spohn, W. 2001, Vier Begründungsbegriffe, in: Erkenntnistheorie. Positionen zwischen Tradition und Gegenwart, Grundmann, T. (Ed.), Paderborn: mentis, Stalnaker, R.C. 1999, Context and Content, Oxford: Oxford University Press. Suppe, F. (Ed.) 1977, The Structure of Scientific Theories, 2., Urbana, Chicago und London. Taschek, W.W. 1998, On Ascribing Beliefs: Content in Context, in: The Journal of Philosophy 95 (7): Willaschek, M. 2000, Wissen, Zweifel, Kontext. Eine kontextualistische Zurückweisung des Skeptizismus, in: Zeitschrift für philosophische Forschung 54 (2): Williams, M. 1988, Epistemological Realism and the Basis of Scepticism, in: Mind 97 (387): Williams, M. 1991, Unnatural Doubts. Epistemological Realism and the Basis of Scepticism, Cambridge, MA and Oxford: Blackwell. Williams, M. 1999, Skepticism, in: The Blackwell Guide to Epistemology, Greco, J. and E. Sosa (Eds.), Malden and Oxford:Blackwell, Department of History and Philosophy of Science University of Cambridge Free School Lane Cambridge CB2 3RH U.K. js427@cam.ac.uk

279 DANIEL COHNITZ * MODAL SKEPTICISM. PHILOSOPHICAL THOUGHT EXPERIMENTS AND MODAL EPISTEMOLOGY INTRODUCTION In this paper I examine some of the criticism the method of thought experimentation was confronted with during the last decades, and shall try to defend it at least for certain problems, restricted by certain constraints. First of all, in section (1) I give a motivation to defend thought experimentation against the skeptic. 1 Confronted with serious criticism, it might be easier to give up that method and to do philosophy without those scenarios. I shall suggest (although not argue for it here) that it would not be so easy to give up this method, and that without it we could no longer answer certain interesting questions. To clarify what I mean by thought experiment, I shall give a brief regimentation of the subclass of philosophical arguments on which I want to concentrate. In (2) I shall discuss one line of attack against thought experiments: modal skepticism. Against this attack I shall present in (3) an informational account of possibility that shows how increase in knowledge of actualia will result in increase in knowledge of possibilia. In (4) I shall briefly address the sources of modal errors. In the concluding part (5) of this paper, I shall give an assessment of what else would have to be done to provide a satisfying methodological account of thought experimentation. 1. THIS WORLD IS NOT ENOUGH One of the most basic methods of philosophy is and has always been the consideration of counterfactual cases and imaginary scenarios. One purpose of doing so, obviously, is to test our theories against such counterfactual cases. 2 Although this method is widespread, it is far from being commonly accepted. Especially during the last two decades, it has been confronted with criticism ranging from complete dismissal 3 to denying only its critical powers 4 to a cautious defense 5 of the use of thought experiments as counter-examples. 6 Of course, one can find innumerable thought experiments which do not constitute clear-cut counter-examples against the theory they are meant to criticize. 281 F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

280 282 DANIEL COHNITZ They quite often appear to be question-begging or indecisive, or it turns out that the intuitions the thought experimenter intends to invoke are not universally shared by his audience. Some might find a certain scenario perfectly conceivable, whereas others are convinced that the case described is definitely impossible, and still others might claim that they do not know what to say at all about such a counterfactual situation. Diagnosing such disagreement leaves us with two possible reactions: we could simply abandon the method and do philosophy without it, or we could try to find out whether the method can be trusted (at least) under certain circumstances, meeting certain constraints. The latter reaction would involve an epistemological inquiry. What I suggest (although I cannot argue for it here) is that, in the case of thought experimentation we lack an alternative method that serves the same purpose. Roy Sorensen, Sören Häggqvist, Frank Jackson 7, and others have argued that we could no longer answer certain interesting questions without it. 8 So what we are left with (if Sorensen, Häggqvist and Jackson are right) is the task of finding the sources of disagreement about certain thought experiments, and proving to the skeptic that an inquiry concerning modal epistemology is a promising enterprise. Necessity Refuters a Description But let me first characterize in a semi-formal way what I have in mind when I speak of philosophical thought experiments or modal arguments. As Sorensen and Häggqvist have argued, there is an important subclass of arguments usually labeled thought experiments 9 (Sorensen calls them necessity refuters ) which consist of the following ingredients 10 : (I) The target claim T: This can be any statement which implies a necessity claim. It might be e.g. some reductionist thesis (in the sense that to reduce either means to claim identity between two rigid designators or involves some strong supervenience), a conceptual analysis (stating necessary and/or sufficient conditions for the application of a certain concept, claimed to be conceptually necessary), a law statement and so forth. In fact, every deductive argument as a whole could do, since it implies an entailment claim (that the premises logically entail the consequent). T itself or its implied consequence is of the form (II) The accommodation of refuter R: The partial description of a possible world that (i) belongs to the set of possible worlds (I) is talking about, and (ii) in which the refuter R, is true. (III) The possibility claim following from (II). (IV) The conclusion of the argument of the form or some inference leading to the negation of T, the refutation of the target claim.

281 MODAL SKEPTICISM 283 If one wants to find examples of such a necessity refuter, one could turn to the philosophy of mind: Let a version of physicalism be the target claim T. Physicalism is usually taken to hold that mental truths are reducible to physical truths and, hence, that the so described mental and physical events are identical. This reduction thesis implies that identity is a metaphysical necessity in the Kripkean sense. It follows that if physicalism is true, it is metaphysically necessary that if all physical facts are given, all mental truths are fixed as well. This establishes step (I). Now consider the metaphysical possibility of a zombie, who is physically indiscernible from us, but lacks all qualia, i.e., phenomenal states. And let a zombie world be a world physically indiscernible from ours but inhabited by zombies instead of human beings. Let s assume that one could accommodate such a situation as a possible world in the relevant metaphysical sense 11, and that the phenomenal states the zombie world is lacking are mental facts in our world. This would establish (II) and hence (III), namely that there is a metaphysically possible world in which all statements about physical events are true as well, but the relevant mental statements are not. From here we could go on to the refutation of the target claim and conclude that physicalism is false. 2. MODAL SKEPTICISM Arguments of this sort are certainly formally valid. What is doubtful about such arguments is the justification of the modal premise (II): can we establish that a certain state of affairs is possible in the relevant sense? Skepticism with respect to thought experiments often amounts to giving a general negative answer to this question. One of the strongest criticisms of the method of thought experimentation, especially of arguments involving modal premises (like instantiations of the schema given above), is modal skepticism as explicated and defended by Peter van Inwagen. 12 There are, of course, other valid modal arguments which similarly involve an assertion of possibility as their crucial premise. In fact, Peter van Inwagen characterizes a subclass of such thought experiments which prima facie differs in structure from our regimentation of a necessity refuter. 13 He is concerned with arguments of the following form: It is possible that I exist and nothing material exists Whatever is material is essentially material hence, I am not an immaterial thing. Given the example of an argument against physicalism above, it is easy to see that Inwagen s modal arguments could be structured in the form of necessity refuters if that would matter. However, the epistemological problem that

282 284 DANIEL COHNITZ concerns the justification of premise (II) is exactly the same problem in all arguments of concern to Inwagen. Inwagen starts with the observation that one can always turn a modal argument into a contrapositive argument which is to say that whenever somebody argues from a refuter R to the negation of T, the strategy to argue from T to the negation of R is always an open choice. It is not entirely clear what the argumentative status of Inwagen s observation is. If he intended it to reveal a problem specific to thought experiments, it must be noted that this is not a specialty of modal arguments: in fact every deductive argument can be countered by a contrapositive argument. This triviality instantiates the logical structure of the so-called Duhem-Quine Thesis : if an empirical theory and a set of background assumptions are inconsistent with certain empirical data, logic alone can t decide between a refutation of the theory, the background assumptions or the data. This correct logical observation applies to non-modal arguments as well. The only difference with non-modal arguments is that we possess a philosophical theory about ways of judging the reliability of our premises relative to the negation of our conclusion in their case. Hence we can explain why not every contrapositive argument appears to be as convincing as the original. This philosophical theory consists in an elaborate epistemology which tells us under what conditions we can trust the empirical data more than our empirical theory. 14 In the modal case, we are lacking such a straightforward epistemology. Peter van Inwagen gives an outline of a rather weak modal epistemology (based on an idea of Stephen Yablo) and concludes that this epistemology should lead us to a general skepticism concerning all modal claims which are not bound to our everyday life. Whoever claims to have a special insight into possibilities beyond that is, according to Inwagen, simply fooling himself: The illusory character of their conviction [to have a capacity to determine the truth value of such possibilities] is sometimes disguised by talk of logical possibility, for it is often supposed that there is a species of possibility that goes by this name and that one can determine a priori whether a concept or state of affairs is logically possible. But there is no such thing as logical possibility not, at least, if it is really supposed to be a species of 15 possibility. Nevertheless, and this is even admitted by Inwagen, we know of certain modal truths. We know that certain propositions are possible, although we do not know that they are true (maybe because they are false, or because we don t know whether they are true or false). This knowledge is of possibilities which are bound to our everyday life. Unfortunately, Inwagen doesn t provide any better definition. Maybe we can circumscribe basic modal knowledge as knowledge of possibilities we are most likely to consider to rank our everyday preferences. Inwagen claims that these modal claims of ours are not as doubtful and indeed may represent modal knowledge. But if can not be inferred from the fact that we could not prove p to be impossible, and is not inferred from p s being true,

283 MODAL SKEPTICISM 285 we run into an epistemological problem concerning such basic modal knowledge a problem which, according to Inwagen, we obviously have: how do we acquire basic modal knowledge? Although Inwagen claims not to know how to answer questions like that 16, he suggests that Stephen Yablo s modal epistemology 17 is a step towards a solution. Yablo defends a weaker version of Hume s maxim, iff p is conceivable. Yablo s weaker version restricts p is conceivable to I can imagine a possible world I take to verify p, and defends the thesis that If I am able to imagine a world I take to verify p, I am prima facie justified in believing that Yablo s thesis is weaker than Hume s, since not all possible ways of conceiving something are said to be reliable guides to possibilities. Only this certain sort, which involves the objectual (instead of a mere propositional ) imagination of a world I take to verify the proposition in question, can be defended as the methodologically useful and epistemically successful way of conceiving something. Inwagen accepts Yablo s thesis, since it seems that this constraint is strong enough to rule out all apparent possibilities which go far beyond our everyday life. Any possible world to verify such a p would be either too difficult to specify in sufficient detail, or simply no world that could be taken to verify p (for apparently all cases which are usually considered in necessity refuters). 18 For example, to specify a possible world which could be taken to verify (P) There is a naturally purple cow. would be to spell out in detail that there is a possible purple pigment such that the coding for the structures that would be responsible for its production and its proper placement in a cow s coat could be coherently inserted into any DNA that was really cow DNA or even cow-like-thing-but-for-color DNA. 19 Whereas imagining the headline Mutant purple cows cloned in New Jersey lab on the cover of a conceivable issue of Scientific American, would surely not suffice to verify (P), since this world could only be taken to verify a disjunctive proposition with It s the April 1 issue of Scientific American and it's a hoax, as one of its disjuncts besides (P). On the other hand, Yablo s thesis seems to justify our unquestioned basic modal knowledge, since for these cases it seems relatively easy to imagine in considerable detail a world verifying p. For example, a proposition like (Q) The desk could have been two feet to the left.

284 286 DANIEL COHNITZ is surely basic modal knowledge in Inwagen s sense. For (Q), it wouldn t be too hard to commit oneself to the possibility of a whole, coherent reality of which the truth of (Q*) The desk is two feet to the left. is an integral part. We simply would alter the history of the actual world insofar that, e.g., I decided differently when furnishing my office, and placed the desk two feet to the left. This account would rule out weird possibilities, whereas it would explain our basic modal knowledge. Hence, Yablo s modal epistemology supports modal skepticism. [N]othing we imagine is absolutely impossible. David Hume 3. BASIC VS. REMOTE MODAL KNOWLEDGE Well, actually it does not. (P) and (Q*) would both be ruled out. Why should we think that a world with a different decision made by me can be specified coherently? Because we did that? Not quite! One of the differences between (P) and (Q*) is not that (P) is so far away from the actual world that we would have to fill in too many unknown details in the design of a possible world. The truth is that in (P) s case, we know too well what details there are to be filled in (about pigments and cow DNA). It is more likely that the amount of biochemical knowledge we already have simply rules out (P). 20 Whereas in (Q*) s case it is not as likely that our background knowledge might rule (Q*) out. So again, why should we think that a world with a different decision made by me can be particularized coherently? We can think so since there is no available information ruling it out. For all we know (better: for all we justifiably hold to be true), there is no inconsistency in altering the history of the world in that way. Of course, we don t know too many of the natural laws which govern my mind, and which determined my choice when I furnished my office. Maybe there are no determining laws. And if there are such laws, maybe they are compatible with a different choice by me. We simply don t know. So even worse for (Q*)? It s not only that details are missing (as in (P) s case) we even don t know what kind of details are to be filled in. But as long as we don t know that, why should there be any reason not to be justified in believing that it is possible that (Q*) or that (P)? What is established so far is that we do not know these statements with certainty. We might have made an error, we might have overlooked an impossibility. Is this a reason not to be justified in believing p, if you are not certain that p?

285 MODAL SKEPTICISM 287 So far, we have established that the distinction modal skepticism wants to draw between basic and remote modal knowledge is absolutely unmotivated by Yablo s epistemology. But Inwagen s claim that anybody who accepts [Yablo s] account should be a modal skeptic is not fully refuted yet. What we now have to show is that modal skepticism does not follow from Yablo s thesis. This modal skepticism would now be general in kind, since we have just seen that a basic/remote distinction is definitely not supported. To argue that we are sometimes justified in believing something to be possible just because we don t know it to be impossible, is to reject Inwagen s main claim that what is not known to be impossible isn t therefore possible. So the task now is to show that such inferences can indeed yield justified beliefs. One way to do that is with an informational account of possibility. In what follows, I shall sketch such an account, which is influenced by Jon Barwise s paper Information and Impossibilities. 21 The purpose of this excursion is to give a partial explanation of some of the notions just used. Most other accounts of thought experiments fall short of providing such a further explanation of what they actually mean when they demand that the setting of a thought experiment should be relevant, or how modal knowledge can increase. I hope that the following will be somewhat clarifying here. When one tries to use the notion of possibility to explicate information content (like, e.g., Dretske s information-theoretical account of knowledge 22 ), one easily arrives at the inverse relationship principle which is the center of informationalism: The Inverse Relationship Principle (IRP): Whenever there is an increase in available information there is a corresponding decrease in possibilities, and vice versa. 23 A problem of such accounts is that they usually are not able to provide a sufficiently fine-grained analysis. Consider some exercise in a math textbook: you are asked to compute a number, say the product of two large numbers. After reading the instruction, you simply do not know what the solution (the product) is. You start computing and come to the solution. What is this new knowledge you have when you now know the value of the product? What is the content of this new information? According to the IRP and a standard interpretation of impossibilities, the information content would be analyzed in terms of possibilities which are now ruled out by the computation. But since the solution of this mathematical equation is mathematically necessary, there simply are no such possibilities. The set of possibilities that are ruled out by the computation is empty. Hence, the new information can t be modeled within such a framework. One way to overcome this problem is to use impossible worlds as the members of the set of worlds now ruled out by the computation. The set would not be empty anymore and would allow for differentiation between knowledge of

286 288 DANIEL COHNITZ different necessary truths. The idea behind this version of informationalism is that impossibilities are those states of the system under investigation that are ruled out by (i.e., compatible with) the currently available information about the system. States not so ruled out are possibilities. 24 This obviously supports a relativistic and pragmatic notion of possibility. The set of possible worlds (or possible states, if one wants to emphasize the link between informationalism and moderate modal realism) is (i) relative to a knowledge situation ( currently ) and (ii) relative to the availability of information about the relevant issues. Condition (i) makes perfect sense when one wants to talk about possibilities and their inverse relationship to information. Relative to newly available information, certain states of the system under investigation can turn out to be impossible states during the inquiry. But before this information was available, they were possibilities that is why we now can describe the different knowledge situations. In fact, our modal knowledge did increase. This positive account of how we can increase our knowledge of possibilities can be accompanied by another argument against modal skepticism. In this paragraph I shall show that modal skepticism is ill-motivated if it is supposed to be a local skepticism, a skepticism that is not supposed to shed equal doubt on all of our beliefs. Consider Peter van Inwagen s example of the infallible Standard Atlas: Suppose that the infallible Standard Atlas marks many islands as uninhabitable, none as inhabitable, and makes no claim to completeness in that matter. We could, if we liked, say that the islands marked uninhabitable in the Standard Atlas were cartographically uninhabitable. In doing this, we should be calling attention to the fact that our knowledge that these islands were uninhabitable had a certain source. But would there then be any sense in saying that an island was cartographically inhabitable just in case it was not cartographically uninhabitable? Very little; I should think. 25 Inwagen suggests that if we only know that some islands are uninhabitable, we aren t therefore justified in believing that it is possible to live on the others. The argument runs like this: (P1) We have some negative modal knowledge, but do not know whether it is complete. (It is possible that there are some impossibilities we don t know to be impossibilities.) (P2) If (P1) is true, we are not justified in deeming something possible just because we don t know it to be impossible. (K1) Therefore we aren t justified in deeming anything possible (which we don t know to be true). (K2) Therefore, talk of possibilities in general is nonsense. There are additional premises to be added if this argument is supposed to turn out valid. But whatever they might be, I suppose either one of them or (P2) must be false. To see this, compare this line of reasoning concerning our knowledge of

287 MODAL SKEPTICISM 289 possibilia with our knowledge of actualia: on a fallibilistic account (however subtle it is) there is only negative knowledge. Are we ever (in Inwagen s sense) justified in deeming All ravens are black true? Does this affect the fallibilist s notion of objective truth? Is a Critical Rationalist ever justified in believing anything positively? I guess so. But then the case for possibilia is as good or bad as for actualia, at least on the grounds of Inwagen s argument. It would look like this: (P1) It is possible that there are some falsities we don t know to be falsities. (P2) If (P1) is true, we aren t justified in deeming something true just because we don t know that it is false. (K1) Therefore we aren t justified in believing any positive actualia (e.g., we aren t justified in believing All ravens are black ). 26 (K2) Therefore, talk of positive actualia (truth in general) is nonsense. In Inwagen s example, nothing excludes the possibility of one s being able to live on those islands. This latter claim is relative to our currently available information. Since informationalism tries to explain changes in information, it allows for changes in the set of possibilities. Inquiry will bring us closer to the truth, just as well as it does in a realist s epistemology about actualities. At no particular epistemic juncture are we entitled to claim certainty of our beliefs. Even an ideal epistemic situation could not guarantee that our beliefs were not completely mistaken. The same holds for the case of possibilia: new information puts constraints on what we deem possible, and inquiry could converge, just as in the case of actualia there doesn t have to be a final destination where we would reach certainty. Inwagen s argument again does not show that there is a specific problem with modal knowledge. 4. MODAL ERRORS We can now take all of Inwagen s arguments in support of modal skepticism to be rejected. Hence, we should try doing some constructive work and see how our proposed explication of modal inquiry given above can yield a classification of errors which, of course, can occur and which should be detectable and correctable if we hope to increase our modal knowledge. Consider the Hesperus- Phosphorus example. Before the information was available that these heavenly bodies are one and the same object (which, it is often said, was a belief held by the ancient Babylonians), it was imaginable that Hesperus would outlast Phosphorus and vice versa. Something we now would describe as metaphysically impossible. That it is now known to be metaphysically impossible is the case because we now correlate this statement to another set of information about the relevant issues. There was a time when the identity of Hesperus and Phosphorus wasn t known, at which it was deemed possible that the brightest heavenly body

288 290 DANIEL COHNITZ in the morning sky would outlast the brightest heavenly body in the evening sky. Since we now have information about this issue different from the ancients, it is no longer an epistemic possibility for us. Barwise describes this modal error as an error stemming from ignorance. Another source of modal errors stems from what one counts as relevant in a certain context. The existence of different notions of possibility suggests that for different problems, different issues are relevant and accordingly, different information is made available for the very context in which one claims something to be possible. Hence, another modal error stems from the fact that one can mistakenly overlook available relevant information if one does not realize that an issue is indeed relevant for this very context. I shall discuss one such example in the next part of this paper. The instinctive is just as fallible as the distinctly conscious. Its only value is in provinces with which we are very familiar. Ernst Mach Intuition is an extremely frail reed upon which to build a philosophy. Robert Nozick 5. CRITICISM AND CONCLUSION We will now see how our proposed account would model a sample thought experiment and the criticism raised against it. I shall also emphasize what the obvious shortcomings of the proposed account are, and suggest what else would have to be done to provide a satisfying methodology for thought experimentation. After introducing necessity refuters, I gave an example for a modal argument from the physicalism debate. 27 I said above that physicalism would be rejected by the thought experiment mentioned, if the metaphysical possibility of a zombie world could be accommodated. Various authors seem to hold that the task involves entertaining as possible only a world in which all qualia are absent. Sören Häggqvist s treatment of thought experiments, for example, seems to support this view. 28 But according to Häggqvist, if we want to establish a possibility premise for such a modal argument, we stipulate the scenario S and accommodate it in a consistent environment C, where C should be as conservative (as close to the actual world) as possible. This last requirement resembles Inwagen s claim that our modal beliefs are of significance only if bound to our everyday life. In fact, a lot of philosophers hold that the crucial difference between good and bad thought experiments is to be found in their closeness or remoteness from the actual world. I think this misses the point entirely; such a requirement is too inclusive and too exclusive at the same time.

289 MODAL SKEPTICISM 291 It is too strong since there are mathematical possibilities which I can use in a philosophical or mathematical argument to reject an apparent mathematical necessity, one that can be sufficiently accommodated in a consistent environment that has nothing to do with the actual world. 29 What is crucial is their consistency with mathematical truths. A closeness requirement is, on the other hand, too weak, since however far we move from the actual world, it is not the degree of difference that matters, but whether the difference involves overlooking relevant available information. 30 It is for this reason that Barwise and Perry reject the zombie argument against physicalism. 31 Imagining a world that lacks qualia means designing a world in which phenomena are missing that are among the physical phenomena of our world (e.g., the physical effects of mental causation). But since at the same time a world is required which is physically indiscernible from ours, this bit of information about our world is relevant! Overlooking it means imagining a possible world with no argumentative power against the target claim. 5.1 A Dilemma for Informationalism Doesn t this seem all too easy? Well, actually it seems to be begging the question. As far as we have studied the informational account, our modal knowledge seems to be derived knowledge. We infer the possibilities by constructing models which solve all the issues of certain parts of our background knowledge. Now we can ask whether physicalism itself is part of that background knowledge we take to decide whether zombies are possible. If it is not (and if it is not implied by the relevant background knowledge either), it is trivial to find a model that solves all the issues but is nevertheless incompatible with physicalism (unless physicalism was implied, according to the definition of logical consequence in modeltheoretic terms). If, on the other hand, physicalism was implied, it is no wonder we cannot find a possible world contradicting it. Conclusion: thought experiments are either trivial or question begging. Are we back to square one? Machianism One 33 way to escape the dilemma posed in 4.1 is to put evidential weight on intuitions as non-inferential beliefs. 34 We would rather leave physicalism out of the relevant background knowledge and activate our intuitions to bridge the gap. Are the (trivially existing) counter-models intuitively satisfying? Following our epistemological program, we would then have to account for the reliability of intuitions. What are they? Where do they come from? What are the normal conditions for intuiting something reliably? This, I guess, addresses Inwagen s original problem: why should we have a faculty to non-inferentially judge possibilities at all, and why should this faculty be reliable? Platonism is

290 292 DANIEL COHNITZ one way to answer all these questions in an incredible way. 35 Machianism answers all questions convincingly, but not necessarily satisfyingly. Machianism is a conscious misnomer 36, due to the fact that the better label intuitionism is already in use for another philosophical position. Machianism explains our intuitions as dispositions which evolved during our evolutionary history. Some of them were shaped by the circumstances our ancestors had to struggle with, some were not so shaped. The former can be assumed to be commonly shared within our species. They might include intuitions concerning logical laws 37 as well as fairness intuitions 38 and grammaticality intuitions 39, or intuitions about how to behave in an iterated prisoner s dilemma. 40 When it comes to grammaticality (as in Noam Chomsky s theory), intuitions are all there is to know. If they are commonly shared, evolutionary history explains why this is so, and in which circumstances they are triggered under normal conditions. The biggest plus is that they are reliable in an objective sense: to be commonly shared just is to be true. For all other intuitions this is dubious. Imagine a moral realist who tests his theory against evolutionary hard-wired fairness intuitions. Wouldn t these intuitions be a bit too contingent to form the basis for a moral realist s ethics? Is, to take another example, ontology a theory about what there is, or a theory about what evolution made us intuit what there is? As much as this sounds unsatisfying to the realist, as much does it dissolve his object. Robert Nozick, who certainly is a Machian in the sense introduced here, claims the following: The strength and depth of our intuitions about statements cannot be used as powerful evidence for their necessity when those statements are of a kind such that selection would lead to strong intuitions of their self-evidence even if they were (only) contingently true. [...] If our intuitions of necessity, and our inability to imagine otherwise, can be accounted for without assuming the existence of necessities, should we believe there are any such necessities at all? Machianism as we merely sketched it here explains all epistemological questions concerning intuitions and thought experiments, but it leaves another one untouched: why should we care? Nozick, again, hits the nail on the head: Philosophers who give great weight to intuitions need to offer some account of why such intuitions are reliable and are to be trusted; at least, they need to sketch how we would have acquired a reliable capacity of this sort. Descartes based his confidence in thought processes that involve clear and distinct ideas upon the existence of a good God who would not deceive him. Upon what do contemporary philosophers of intuition base their claims? Of course, if the purpose of such philosophy is merely to codify and systematize the intuitions that (for whatever reason) are held, then a philosophy built upon intuitions will need no further basis. And it will have no further validity.42 Is this any better than Modal Skepticism? Wouldn t this lead to the same rejection of modal arguments? I don t think so. Even if one ascribed to Nozickian

291 MODAL SKEPTICISM 293 Skepticism, thought experiments would still be relevant methodological tools. Let s conclude with what we ve found out and see what consequences follow. We have seen that an important subclass of thought experiments is used as a device to choose between competing philosophical theories. Their methodological place is within theory dynamics. This method crucially involves talk of possibilities and necessities. The notion of possibilities and their relevance for theory dynamics is not the methodological problem of thought experiments, contrary to what Peter van Inwagen suggested. Philosophical theories are modal in nature, they quantify possible worlds which can count as counterexamples. A broad notion of possibility that conforms to our not-impossible-hence-possible inferences is explicable, e.g. in the terms of informationalism. Modal Skepticism as a local form of skepticism is not supported by any of Inwagen s arguments. The explication of the notion of possibility given here advocates a certain methodological position: what counts as a counterexample is not an issue of closeness or remoteness to the actual nature, it is an issue of relevance. What is relevant is determined by the modal kind of the target theory itself. Although this seems all too convincing, there still remains the problem of how to deal with philosophical positions (targets and refuters) that cannot simply be inferred from other knowledge we accepted as relevant. So it seems like we need a faculty of non-inferential insight into possibilities. One way to deal with this, Machianism, doesn t support rock-bottom relevance for the possibilities we might find intuitively plausible. Nevertheless, it says for some of them what they are, where they come from, and when they are a reliable source of evidence for what. Imagine a possible world in which philosophers aware of all this would want to keep doing scientifically respectable philosophy. Is there any special 43 reason for them to abandon the method of thought experimentation? As the arguments have shown, there are indeed no special reasons to give up thought experimentation. So maybe there is a lesson as to how to deal with them? Well, according to our results, there is definitely room to ascribe different roles to the reactions triggered by imaginary scenarios: (A) Your theory is about intuitions with respect to X, not because intuitions are reliable evidence of what X is, but because intuitions are the data. 44 You always have to take them seriously, anyhow (if they are in conflict with your theory, you have to find a causal explanation for why this is a deviant case). (B) Your theory is not about intuitions, but about X. 45 Intuitions conflict with your theory about X. You should try a Machian method of explaining them away. If you are not successful, then that is evidence that these intuitions are on to something. (C) Your theory is not about intuitions, but about some X that really couldn t interfere with the evolution of intuitions. Additionally, they seem to converge against you. Intuitions are clearly irrelevant, but you might want to find a compromise. This strategy might be necessary if your theory is

292 294 DANIEL COHNITZ ethical in nature and has such counterintuitive consequences that you cannot hope to get them realized in laws (e.g., in a democracy). NOTES * I would like to thank Dieter Birnbacher, Axel Bühler, Jason Hagen, and John Perry for helpful comments on earlier versions of this paper Like Sören Häggqvist and Peter van Inwagen, I shall refer to philosophers who criticize the method of thought experimentation as skeptics. I cannot argue for that position here, but one can find arguments in Sören Häggqvist, Thought Experiments in Philosophy. Stockholm: Acta Universitas Stockholmiensis 1996 and Roy Sorensen, Thought Experiments. Oxford: Oxford University Press Kathy Wilkes, Real People. Personal Identity without Thought Experiments. Oxford: Clarendon Press Daniel C. Dennett, Consciousness Explained. London: Penguin Peter Unger, Identity, Consciousness & Value. Oxford: Oxford University Press 1990, Sorensen, lbid., Häggqvist, Ibid., Tamar Szabó Gendler, Thought Experiment. On the Power and Limits of Imaginary Cases. New York: Garland Publishing Inc The tradition of this criticism was, of course, present throughout the whole history of philosophy. Nevertheless, it was intensely discussed during the 90 s. These methodological worries were probably due to the more and more excessive use of far fetched counterexamples in the discussion concerning Personal Identity in the last century. Frank Jackson, From Metaphysics to Ethics, Oxford: Oxford University Press Sorensen, Ibid., p. 19. The term thought experiment is certainly used in a far broader sense in common (scientific) language. I do not wish to construct a taxonomy of thought experiments or to give a conceptual analysis of this term. What I am dealing with is a methodological problem concerning, e.g., this very subclass I characterize. My regimentation differs in several respects from Sorensen s and Häggqvist s. Regimentations are always more or less adequate relative to one s interests. However, the three of us seem to have the same subclass of thought experiments in mind, although we semi-formalize them in different ways. The sentence operators are dummies for necessity and possibility respectively, in any sense of necessity or possibility whatsoever, as well as in their interpretation in deontic logic or in any other interpretation that is of philosophical interest and validates the argument. I shall say more on this later. Peter van Inwagen, Modal Epistemology, in: Philosophical Studies, 92, 1988, pp Inwagen, Ibid., p. 81: Inwagen claims that there are valid modal arguments that cannot be so described, and refers to the Gettier cases which are, in fact, necessity refuters (Sorensen, Ibid.,p. 137). This might be a very complex task to judge sometimes. Coherentism tries to model our reasoning in such situations. See Thomas Bartelborth, Begründungsstrategien. Ein Weg durch die analytische Erkenntnistheorie. Berlin: Akademie Verlag Inwagen, Ibid., p. 71. Note that this denial of the existence of logical possibilities is independent of the notions of logical necessity or impossibility. The latter two are, according to Inwagen, meaningful epistemological notions: the logically impossible is that which can be seen to be impossible on the basis of logical considerations alone, which is also true of logical necessity. Hence the qualifier logical indicates the source of knowledge for the modal claim. The problem with logical possibility is, again according to Inwagen, on the other hand, that the fact that a certain thing cannot be proved to be impossible by a certain method doesn t make this thing possible in any sense of possible whatsoever.

293 MODAL SKEPTICISM Another question like that would be the question of how we can know that a proposition which we know to be true is also necessary. Stephen Yablo, Is Conceivability a Guide to Possibility?, in Philosophical and Phenomenological Research, 53, 1993, pp At first it seems as if Inwagen wanted to treat basic modal knowledge as non-inferential knowledge (Ibid., p. 70), but Yablo s maxim clearly involves inferential processes. I shall come back to this point in part 5. Inwagen, Ibid, p. 78. Is not consistent with (P). Jon Barwise, Information and Impossibilities, in: Notre Dame Journal of Formal Logic, 38, 1997, pp Actually Dretske defines information with the use of possibilities. Informationalism doesn t do that. Informationalism explicates the relationship between possibilities and information. Fred Dretske, Knowledge and the Flow of Information. Cambridge, Mass.: MIT Press Barwise Ibid., p Barwise Ibid., p.491. Inwagen Ibid., p Than the ones we know to be necessary on logical considerations alone. Actually, there is more to be said about it, but there is not enough space available to do it here; one might want to turn to John Perry, Knowledge Possibility and Consciousness, Cambridge, Mass.: MIT Press 2001 or Daniel Cohnitz, Two-Dimensionalism and the Metaphysical Possibility of Zombies, in: Foundations of the Formal Sciences II. Applications of Mathematical Logic in Philosophy and Linguistics, [Trends in Logic]. Dordrecht: Kluwer Academic Publishers 2002 for more on this problem. Häggqvist, Ibid. For examples of the sort of thought experiments I have in mind, one could turn to Lakatos mathematical counterexamples in Imre Lakatos, Proofs and Refutations, Cambridge: Cambridge University Press This is a point at which we might have a look at physics. Take Einstein s Clock in the box thought experiment and his Travelling at the Speed of Light. The Clock in the box was not a successful thought experiment, whereas the latter was. How could you describe the difference between them in terms of closeness? Perry Ibid, argues in a way different from Barwise Ibid., but the following summary is neutral with respect to that difference. At this point, the difference between Barwise and Perry matters. Barwise s argument really seems to be question-begging, whereas Perry s is actually not. For more on this see Perry Ibid. and Cohnitz Ibid. Another way to escape from this dilemma is to develop a semantical theory that connects only a fragment of our knowledge (our pure semantical knowledge) with relevant possibilities. This is what Modal Rationalism tries to do. With the help of this bridge, we would then be in a position to inquire the realm of the relevantly possible without considering e.g. physicalism at issue. Positions of this type were recently defended by David Chalmers, Frank Jackson, and Christopher Peacock. The idea is, roughly, that we have a priori access to certain semantical properties of the statements in question. The so called A-Intension or Primary-Intension is knowable a priori based just on the fact that we understand the statement in question. Some of these a priori accessible properties (e.g., this statement has a necessary primary intension ) are supposed to imply relevant possibilities if certain conditions are met (e.g., If this statement has a necessary primary intension, then it is either not a posteriori or not metaphysically necessary. ). But it is highly questionable whether this account can be spelled out in sufficient detail. One worry is that the a priori accessible, purely semantical properties are not sharply defined in these accounts. If there is no clear boundary between the contribution of pure semantical knowledge and the contribution of the context to our evaluation of certain statements, there won t be a very content-heavy epistemological point for modal rationalism. Another worry is that there is good reason to believe that the main claim of modal rationalism is false. There are statements that seem to have the relevant a priori accessible semantic properties, but whose other properties do

294 296 DANIEL COHNITZ not correspondent to to the prediction of the Modal Rationalist s thesis. For more on this see Daniel Cohnitz, The Science of Fiction, forthcoming. I owe most insight contained here to Johannes Haag. For technical reasons, I take intuitions to be beliefs. They might not be beliefs as defined by George Bealer, but they will remain noninferential -which is what matters here. For more on Platonism and thought experiments see James Brown s writings on thought experiments. Roy Sorensen linked the following ideas back to Ernst Mach, who certainly wouldn t have liked this too much. Since I could not think of any better name and since this is the Vienna Circle Institute Yearbook, I coined the term Machianism for the following position, although my reading of Mach doesn t always coincide with Sorensen s interpretation. Robert Nozick, Invariances: The Structure of the Objective World, Cambridge, Mass.: Harvard University Press Brian Skyrms, Evolution of the Social Contract. Cambridge: Cambridge University Press Noam Chomsky, New Horizons in the Study of Language and Mind. Cambridge: Cambridge University Press Robert Axelrod, The Evolution of Cooperation. New York: Basic Nozick Ibid., p Nozick Ibid., p Special would be a reason that goes beyond general skepticism. It should be special insofar as it applies to thought experiments especially. The question of grammaticality in Chomsky and the fairness intuitions of Skyrms are such cases. Consider this example: your theory is about Personal Identity, and there are intuitions according to which Personal Identity, matters to us in a special, irreducible way. John Perry s theory, like most reductionist theories, cannot account for these intuitions, so he has to explain them away through a Machian explanation, giving an evolutionary explanation for why these intuitions evolved. See John Perry, Identity, Personal Identity and the Self. Cambridge, Mass.: MIT Press forthcoming. Philosophy Department University of Düsseldorf Germany cohnitz@phil-fak.uni-duesseldorf.de

295 F.O. ENGLER STRUCTURE AND HEURISTIC: IN PRAISE OF STRUCTURAL REALISM IN THE CASE OF NIELS BOHR 1. INTRODUCTION In the present paper I will try to shed some new light on Niels Bohr s philosophy of physics. Here my primary intention is to reconstruct Bohr s philosophy of physics by demonstrating its important influence on the overall heuristic of his research programme. I shall focus especially on the development of the older quantum theory ( ) to which Bohr made almost all of his influential contributions as a nuclear physicist. 1 In this respect, it is well-known that Bohr s major contribution to the older quantum theory consisted in a coherent and successful explanation of the hydrogen and hydrogen-like atoms on the basis of his celebrated planetary model of the atom. This model, first put forward by Bohr in a comprehensive way in Part I of his famous Trilogy in , was central to atomic research for a long time, making Bohr for many years the foremost authority in the physical community with respect to problems concerning the structure and the constitution of the atom. Regarding Bohr s philosophy of physics, my main argument is that Bohr first of all was a kind of a realist, namely a structural realist. I will argue that this kind of philosophy above all led Bohr to his important theory of the atom, marking the core assumption of the period of the older quantum theory. By showing the importance of structural realism in the case of Niels Bohr and his major contribution to quantum physics, I strongly underline from a general point of view the claim made by Henri Poincaré that physicists believe that their highly corroborated scientific theories bring to light through their mathematical structure true relationships between things, while they are in a sense incapable of uncovering the true nature or essence of the things in themselves. When considering Bohr s philosophy of physics it seems to me that Poincaré s structural realism concurs much better with the overall philosophical position Bohr held as a physicist, eagerly trying to work out his atomic programme, than with the pragmatic or instrumentalistic view with respect to the abstract mathematical structure of the atomic theory that is attributed to Bohr in keeping with the position of empirical realism. Before I outline my claim concerning Niels Bohr s structural realism in connection with its heuristic influence on his research programme, let me first discuss the position of empirical realism that has hitherto 297 F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

296 298 F.O. ENGLER been widely regarded together with the well known and familiar notion of complementarity as the main characteristic of Bohr s philosophy of physics. I will show that the prevailing viewpoint fails to account for a central feature of Bohr s philosophy of physics, and that this can be remedied by the position of structural realism. 2. THE SHORTCOMINGS OF THE PREVAILING VIEWPOINT OF NIELS BOHR S PHILOSOPHY OF PHYSICS: COMPLEMENTARITY AND EMPIRICAL REALISM In the recent past, philosophers of science have predominantly identified Bohr s philosophy of physics with the notion of complementarity essentially combined with an overall epistemological lesson one might call empirical realism. 3 According to the latter it is rational for a physicist to believe in the existence of an objective, mind-independent world but to deny that we have cognitive access to the real nature of objects and properties of this world independently of possible empirical knowledge. In accordance with this point of view, the mathematical structures of our mature scientific theories are merely pragmatic tools or instruments that guarantee the prediction and coherent organization of the empirical results of our experiments without having any physical content of their own. Mathematics is thus reduced to the handmaiden of physics. Because of this claim, we can have knowledge only of perceptible phenomena. For the description of these phenomena, classical concepts are indispensable. They are indispensable since their application for the description of our experience makes that description objective in the sense of being unambiguously communicable. As a consequence, the notion of truth is an epistemic concept. Following Jan Faye s most comprehensive characterisation of Bohr s philosophy of physics we find that he in general allocates to Bohr the viewpoint of an empirical realist. Faye writes: But since Bohr maintained that all we can acquire in the way of scientific knowledge is what is accessible to us through experience as being subsumable under classical concepts, truth, according to him, becomes nothing but a feature of our cognitive capacities. 4 By stressing the pragmatic strain as its crucial for an understanding of Bohr s philosophy of physics that source is generally attributed to Bohr s teacher Harald Høffding Dugald Murdoch argues in a very similar way as Faye: For Bohr, however, the primary aim of physics is to help us make sense of our perceptual experience. He does not doubt that our experience is of an independently existing physical world, but he denies that the main aim of physics is the comprehension of the imperceptible structure of that world for its own sake. We construct theories of the microstructure of the physical world in order better to comprehend the world in which we move and have our being the macroscopic world. 5

297 STRUCTURE AND HEURISTIC 299 This characterization of Bohr s general viewpoint regarding his philosophy of physics as empirical realism corresponds very well with his epistemological view of the notion of complementarity in connection with quantum phenomena. Concerning Bohr s notion of complementarity it is generally thought that it presupposes two basic assumptions: (1) the indispensability of classical concepts in the sense that we feel compelled to use the visual pictures of the language of classical physics in order to describe all possible experience we have, including those in quantum mechanics, and (2) the indivisibility of the finite magnitude of the Planck elementary quantum of action or the quantum postulate that implies a natural limit for the application of classical descriptions within quantum mechanics resulting from the impossibility of neglecting the interaction between a quantum phenomenon and the agency of observation. 6 Bohr summarizes these two general assumptions in the following passage from the Introductory Survey of Atomic Theory and the Description of Nature as follows: According to the author, it would be a misconception to believe that the difficulties of atomic theory may be evaded by eventually replacing the concepts of classical physics by new conceptual forms. Indeed, as already emphasized, the recognition of the limitation of our forms of perception by no means implies that we can dispense with our customary ideas or their direct verbal expressions when reducing our sense impressions to order. Nor is it likely that the fundamental concepts of the classical theories will ever become superfluous for the description of physical experience. The recognition of the indivisibility of the quantum of action, and the determination of its magnitude, not only depends on an analysis of measurements based on classical concepts, but it continues to be the application of these concepts alone that makes it possible to relate the symbolism of the quantum theory to the data of experience. At the same time, however, we must bear in mind that the possibility of an unambiguous use of these fundamental concepts depends solely upon the self-consistency of the classical theories from which they are derived and that, therefore, the limits imposed upon the application of these concepts are naturally determined by the extent to which we may, in our account of the phenomena, disregard the element which is foreign to classical theories and symbolized by the quantum of action. 7 A closer look at Bohr s writings shows, however, that the acceptance of the quantum postulate (2) is more fundamental because, for him, it naturally conditioned the notion of complementarity and therefore the applicability of classical concepts in a new form of description within quantum physics. Bohr argues:... that the fundamental postulate of the indivisibility of the quantum of action is itself, from the classical point of view, an irrational element which inevitably requires us to forego a causal mode of description and which, because of the coupling of phenomena and their observation, forces us to adopt a new mode of description designated as complementary in the sense that any given application of classical concepts precludes the simultaneous use of other classical concepts which in a different connection are equally necessary for the elucidation of the phenomena. It is pointed out that we immediately encounter this feature when considering the questions of the nature of light and of matter. 8

298 300 F.O. ENGLER As a consequence, we find in the above passage the well-known meaning of the notion of complementarity as an epistemological approach to quantum phenomena, namely that two quantum phenomena are to be called complementary if they are mutually exclusive while at the same time they are both necessary for a complete description of the quantum object concerned. In this respect Bohr repeatedly stated in further writings that, in order to describe an individual and clearcut quantum phenomenon, we have to consider the interaction between the quantum object under question and the specific experimental arrangements and conditions surrounding the investigation of this quantum object. Therefore according to Bohr the resulting quantum phenomenon is characterized by a feature of individuality, or as he later calls it, wholeness. 9 As a result of the indispensable interaction that is actually conditioned by the magnitude of the quantum of action, it is no longer possible to acquire knowledge of a quantum object in the way we usually grasp and characterize a classical object in isolation as a comprehensive cluster of physical properties such as position, momentum, time, energy and so on a cluster that can be broken down in all its parts. Instead all we can know is the individual manifestation of an quantum phenomenon under certain experimental arrangements. In Bohr s own words: Now the quantum postulate implies that any observation of atomic phenomena will involve an interaction with the agency of observation which is not to be neglected. Accordingly, an independent reality in the ordinary physical sense can neither be ascribed to the phenomena nor the agencies of observation. 10 Therefore one can draw the epistemic conclusion from the notion of complementarity that the knowledge we can gain of a quantum phenomenon under certain experimental conditions reveals just a certain aspect but not the whole story. For example: If we want to get information about the position of an atom, it is impossible to disclose its momentum. In order to get a complete picture we have to arrange other experimental conditions and obtain results on other quantum phenomena which, though incompatible with each other, are nonetheless each described in classical terms and together form a complete picture. As a consequence, all we can know about the quantum world that, nevertheless, exists independently of our minds are the individual quantum phenomena that are in principle comprehensible to us through experience as being subsumable under classical concepts according to the specific experimental arrangements. Consequently it seems rational that we end up with the prevailing position of empirical realism regarding Bohr s philosophy of physics. But as already mentioned before, it is very important for Bohr to examine and disclose the natural or general conditions underlying all human understanding. 11 For him, therefore, in order to establish the limits of the applicability of our classical concepts within quantum mechanics and as a consequence to embed classical theory in a more general theory, we have to come to some understanding of these more general conditions and their connections to classical theory. Consequently there is not simply the epistemic lesson of empirical real-

299 STRUCTURE AND HEURISTIC 301 ism Bohr wanted to teach us by introducing his notion of complementarity concerning the description of individual quantum phenomena. Bohr s conception of complementarity essentially also requires knowledge of the more general conditions that reasonably explain the possible description of quantum phenomena and make it understandable on natural grounds why the applicability of classical concepts is limited within the quantum domain. The later Bohr states the following: The term complementarity, which is already coming into use, may perhaps be more suited to remind us also of the fact that it is the combination of features which are united in the classical mode of description but appear separated in the quantum theory that ultimately allows us to consider the latter as a natural generalization of the classical physical theories. 12 To my mind, therefore, the merely pragmatic or instrumentalistic interpretation of Bohr s position concerning the abstract mathematical structure of the atomic theory explicitly expressed by Murdoch in the quotation mentioned above fails entirely to take into account Bohr s attempt to grasp at the lawful structure of the microworld as a more general condition for our empirical knowledge, which is generally seen as the description of nature. 13 In that respect Bohr goes well behind and beyond experience by assuming that the atomic theory reveals a more general and lawful structure of nature that cannot be subsumed under classical concepts but constitutes, rather, the possibility of empirical knowledge within quantum mechanics. In an addendum to the introductory survey of Atomic Theory and the Description of Nature, Bohr summarizes his view concerning the place of atomic theory in the description of nature as follows: In particular, it was my desire to emphasize that, despite the great success attending the discovery of the building blocks of atoms a discovery depending on the application of classical concepts the development of atomic theory has, nevertheless, first of all given us a recognition of laws which cannot be included within the frame formed by our accustomed modes of perception. 14 That the lawful relations of atomic theory are fundamental for the description of nature, and the recognition thereof the primary aim, is expressed by Bohr in the following: In fact, the indivisibility of the quantum of action demands that, when any individual result of measurement is interpreted in terms of classical conceptions, a certain amount of latitude be allowed in our account of the mutual action between the object and the means of observation. This implies that a subsequent measurement to a certain degree deprives the information given by a previous measurement of its significance for predicting the future course of the phenomena. Obviously, these facts not only set a limit to the extent of information obtainable by measurement, but they also set a limit to the meaning which we may attribute to such information. We meet here in new light the old truth that, in our description of nature, the purpose is not to disclose the real essence of the phenomena but only to track down, so far as is it is possible, relationships between the manifold aspects of our experiences. 15

300 302 F.O. ENGLER To my mind Bohr states in this passage a position that has so far been attributed most prominently to Henri Poincaré and is called structural realism. In order to compare Bohr s position with Poincaré s point of view, let me therefore quote the following passage from Poincarés famous Science and Hypothesis that has usually been interpreted as the main evidence of his structural realism. Poincaré wrote: And let no one say that thus we reduce physical theories to the rôle of mere practical recipes; these equations express relationships, and if the equations remains true it is because these relationships preserve their reality. They teach us, now as then, that there is such and such a relationship between some thing and some other thing; only this something we formerly called motion, we now call electric current. But these appellations were only images substituted for the real objects which nature will eternally hide from us. The true relationships between these real objects are the only reality we can attain, and the only condition is that the same relationships exist between these objects as between the images by which we are forced to replace them. If these relationships are known to us, what does it matter if we deem it convenient to replace one image by another. 16 The most striking similarity between the positions of Niels Bohr and Henri Poincaré is that they both deny that we can possibly gain from science any knowledge of the nature of the things as they are in themselves. And furthermore, they both think that the relationships exhibited to us through the lawful structure of our general theories as conditions for possible human knowledge are the only remaining features of reality we are able to grasp. To the extent that both Poincaré and Bohr are of one mind in considering these points, it is in my opinion reasonable to claim that Bohr held the position of structural realism. However, there is no explicit evidence in Bohr s writings in favour of structural realism. The same is true for the position of empirical realism. Thus, even in spite of lacking evidence we should, however, rely on the bold thesis of Albert Einstein: that in order to consider a scientist s work, one ought to give attention to what he does and not to what he says. In the next section I therefore intend to further underline my claim that it is rational to assume that Bohr was a structural realist by reconstructing his atomic research programme. This way it can be shown that in his work as a physicist, Bohr was guided by the overall philosophical idea of structural realism. 3. BOHR S STRUCTURAL REALISM AND ITS INFLUENCE ON THE HEURISTIC OF HIS ATOMIC RESEARCH PROGRAMME It seems to me that the most appropriate way of starting my reconstruction of the influence of Niels Bohr s structural realism on the heuristic of his atomic research programme is to quote the following letter by Bohr from June 19, 1912 to his brother Harald. Here Niels Bohr expresses his wish for immediate publication of an essay later known as the Rutherford Memorandum 17, in which Bohr

301 STRUCTURE AND HEURISTIC 303 claims for the first time to reveal something of the real structure of the atoms. Niels Bohr writes: Dear Harald, Perhaps I have found out a little about the structure of the atoms. Don t talk about it to anybody, for otherwise I couldn t write to you about it so soon. If I should be right it wouldn t be a suggestion of the nature of a possibility (i.e., an impossibility, as J.J.Thomson s theory) but perhaps a little bit of reality. It has grown out of a little information I got from the absorption of (the little theory I wrote about last time). You understand that I may yet be wrong, for it hasn t been worked out fully yet (but I don t think so); also, I do not believe that Rutherford thinks that it is completely wild; he is a man of the right sort, and he would never say that he was convinced of something that was not fully worked out. Believe me, I am eager to finish it in a hurry, and to do that I have taken off a couple of days from the laboratory (this is also a secret). This was intended only as a little greeting from Your Niels 18 Back in the summer of 1912 when Bohr s scientific breakthrough with respect to the constitution and the structure of the atom took place, Bohr was in Manchester and especially concerned with the unsolved stability problem of the Rutherford atomic model that, however, successfully explained some experimental results like the large angle scattering of This model, as it is well known and expressed by Bohr in the following words, consists of a positive charge concentrated in a point (in an extension that is very small compared with the dimensions of the atoms) surrounded by a system of electrons, the total charge of which is equal to that of the positive kernel ; the kernel is also assumed to be the seat of the mass of the atom. 19 Using mechanical and electro-dynamical concepts alone, Bohr was facing the problem that the stability of the Rutherford atom could not be explained. Based on the constant motion of electrons around the atomic kernel, the classical theory predicts a radiative dissipation of energy coupled by a steady contraction of the atom. Obviously, such an apparent instability of the system of electrons does not occur in nature. Hence the alleged stability of the atom could not be explained by the assumption of the classical theory alone. Following Bohr s account, it was in particular his attempt to determine a quantitative relation that allows one to limit the permissible number of radii of the electron-orbits or -rings around the kernel to a certain amount and thus to avoid the stability problem. 20 In this respect Bohr defines the concept of the socalled permanent or stationary states of electrons where for each of them there exist definite values for the radius and the spectral frequencies. The latter were in some way related to mechanical modes of vibration within the atom. That there must be definite values for these physical magnitudes was clear to Bohr because of the existence of a vast number of sharp and discrete spectral lines and patterns charted for many of the chemical elements. Furthermore, Bohr had some numerical evidence at hand that only a number of electronic configu-

302 304 F.O. ENGLER rations in the atom are stable. After exceeding the number of seven electrons the instability of a ring begins where this seems to allow for an explanation of the periodic properties of the chemical elements. 21 Bohr summarizes his consideration concerning the stability problem of the atom in the following passage of the Rutherford Memorandum: In the investigation of the configuration of the electrons in the atoms we immediately meet with the difficulty (connected with the mentioned instability), that a ring, if only the strength of the central charge and the number of electrons in the ring are given, can rotate with an infinitely great number of different times of rotation, according to the assumed different radius of the ring; and there seems to be nothing (on account of the instability) to allow from mechanical considerations to discriminate between the different radii and times of vibration. 22 And Bohr goes on to express the aim of his further research: In the further investigation we shall therefore introduce and make use of a hypothesis, from which we can determinate the quantities in question. This hypothesis is: that for any stable ring (any occurring in the natural atoms) there will be a definite ratio between the kinetic energy of an electron in the ring and the time of rotation. This hypothesis, for which there will be given no attempt of a mechanical foundation (as it seems hopeless), is chosen as the only one which seems to offer a possibility of an explanation of the whole group of experimental results, which gather about and seem to confirm conceptions of the mechanism of radiation as the ones proposed by Planck and Einstein. At that point, Bohr had not yet identified the exact relation required in order to define stationary states. But it should be emphasized, however, that at this stage of his research Bohr was already from a general point of view aware of the fact that he could possibly gain some knowledge of the structure of the atom as indicated by his letter mentioned above. For this reason it seems plausible that Bohr was looking for a distinct relation so as to reveal something of the structural condition of the atom. With this lawful relation, Bohr was also hopeful that he could explain the process of radiation. The problem of the explanation of the process of radiation did not appear on Bohrs agenda until February In a letter to Rutherford on 31 January 1913, Bohr, reflecting on his research, wrote: I do not at all deal with the question of calculation of the frequencies corresponding to the lines in the visible spectrum. I have only tried, on the basis of the simple hypothesis which I used from the beginning, to discuss the constitution of the atoms and molecules in their permanent state. 23 But the situation immediately changed when H. M. Hanson, an expert on spectroscopy, arrived in Copenhagen a few days later and asked Bohr for an explanation of the Balmer formula of the frequencies of the lines of the spectrum of the hydrogen atom. As soon as I saw Balmer s formula, Bohr from time to time stated in later years, the whole thing was immediately clear to me. 24 From this point it was just a few weeks until Bohr had finished the trilogy that would win him immortality in physics.

303 STRUCTURE AND HEURISTIC 305 Right at the beginning of Part I of the trilogy, Bohr has raised the stability problem again but during the course of his treatise he established the required quantitative relation characterising the permanent states that structures the atom explicit, using Planck s elementary quantum of action, which is symbolized by h. As Max Jammer has already pointed out, by adopting Planck s universal constant, Bohr did not intend to show the physical significance of the elementary quantum of action but rather to account quantitatively for the stability of the atom. 25 By the introduction of this quantity, Bohr writes underlining this assumption, the question of the stable configuration of the electrons in the atoms is essentially changed, as this constant is of such dimension and magnitude that it, together with the mass and charge of the particles, can be deemed a length of the order of magnitude required. 26 Thus, with the magnitude at his disposal that allows one to consider atomic dimensions, Bohr was now able to present the quantitative relation defining the permanent states of the atom. Furthermore Bohr accounted for hitherto unexplained empirical facts by deducing the law of the line spectrum of hydrogen (the Balmer formula) from the relation that structures the atom in the correct way. Without going here into greater detail regarding Bohr s formal derivation which can be found elsewhere in the literature 27 I will discuss Bohr s general result as it relates to structural realism and the consequences for the heuristic of his research programme. Bohr writes: We shall now return to the main object of this paper the discussion of the permanent state of a system consisting of nuclei and bound electrons. For a system consisting of a nucleus and an electron rotating round it, this state is, according to the above, determined by the condition that the angular momentum of the electron round the nucleus is equal to The connection between Bohr s major result of the trilogy and his intention stated in the Rutherford Memorandum can be seen as follows. Bohr explains this major assumption in the following manner: If we therefore assume that the orbit of the electron in the stationary states is circular, the calculation on p. 5 can be expressed by this simple condition: that the angular momentum of the electron round the nucleus in a stationary state of the system is equal to an entire multiple of a universal value, independent of the charge on the nucleus. 29 The angular momentum M is defined by Bohr as where W is the mean value of the kinetic energy and is the frequency of the revolution of the electron around the nucleus (angular velocity). Furthermore, we have the simple condition that where is the entire multiple of the universal value that is equal to. After some calculation we get the relation that can be found on page 5 in the original paper and that in an appropriate manner allows us to define the required definite ratio between the kinetic energy of an electron in the ring and

304 306 F.O. ENGLER the time of rotation expressed by Bohr above in the passage from the Rutherford Memorandum. From this expression Bohr proceeded to successfully deduce the Balmer formula. We can thus conclude that Bohr s invariant condition in the form of the constancy of the angular momentum has to be regarded as the underlying quantitative relation for the structure of the atom. Finally, it is this invariant condition that enables Bohr to discover from the continuum of the manifold of allowable states the appropriate stationary states for the quantification of the atom. The general heuristic device Bohr has applied to investigate the atom thus reflects his philosophical idea that physicists should look for quantitative relations in order to reveal something of the structure of the reality. Subsequently it was Arnold Sommerfeld who generalized Bohr s theory of the hydrogen atom to the extent that the theory accounts for the fine structure of the hydrogen spectral lines. 30 Here Sommerfeld was decisively relying on the overall philosophy of Bohr s atomic programme. In the following Sommerfeld describes his view that is attributed to the idea that micro nature is structured as a network of permanent states: FÜr die allgemeine Auffassung des Naturgeschehens lernen wir aus der Schärfe der Spektrallinien oder aus der ihr Rechnung tragenden Quantentheorie der Spektrallinien: Die stationären Bahnen der Elektronen im Atom (und weiterhin im Molekül) bilden kein Kontinuum, sondern ein Netzwerk. Der Phasenraum, als Mannigfaltigkeit aller denkbaren, auch der nicht stationären Zustände, ist von den Bildkurven der stationären Bahnen maschenartig durchzogen. Die Größe der Maschen ist durch das Plancksche h bestimmt. 31 Sommerfeld s search for a more general condition than Bohr s invariant one that would enable him to know something of the supposed structure of the atom has led him in the following to the well-known generalized quantum condition that the phase integral for every coordinate is an integral multiple of the quantum of action. 32 Subsequently Epstein and Schwarzschild showed that Sommerfeld s theory, further generalized by more universal coordinates, successfully explains the Stark effect. 33 Therefore, from a more general point of view, the underlying idea of the successful development of the older quantum theory was to uncover in a more and more comprehensive manner the structural conditions of the atom that reasonably allow the definition of stationary states, thus solving the stability problem and explaining the process of atomic radiation. In this way, the general idea of structural realism influenced the heuristic of the older quantum theory. 4. CONCLUSION It is the overall philosophical idea of structural realism that all physical phenomena are governed by a certain structure of reality. Through the mathematical relations of our successful theories, this structure is accessible to us. This phi-

305 STRUCTURE AND HEURISTIC 307 losophical position essentially guided Bohr and later Sommerfeld in the construction of their theories. Both were searching for distinct quantitative relationships in order to structure the atom, and they both believed that these relationships, as they were improved further and further and successfully explained more and more empirical facts, had, to some extent, to reflect the structure of reality. Bohr never said that we have direct access to the things in themselves. We certainly cannot reveal the true nature of the referents of various theoretical terms. But we can, nevertheless, reasonably assert to apprehend the structure of reality that allows to account for hitherto unexplained empirical facts. Later, Bohr stressed as expressed in section 2 that the lawful structure of quantum mechanics discloses the natural conditions of our human knowledge in general and makes it therefore reasonable that our classical concepts are limited. Bohr wrote: We are aware even of phenomena which with certainty may be assumed to arise from the action of a single atom. However, at the same time as every doubt regarding the reality of atoms has been removed and as we have gained a detailed knowledge of the inner structure of atoms, we have been reminded in an instructive manner of the natural limitations of our forms of perception. 34 Bohr s epistemic lesson of complementarity expresses these limits, which are, however, determined by the natural conditions that were discovered over the course of his atomic research programme. Therefore, it seems rational to me to suppose that Bohr s philosophy of physics above all should be described by a position that is analogous to the one held by Henri Poincaré, namely structural realism. This position guided Bohr and later Sommerfeld throughout their atomic research programme. Furthermore, it contains the basic assumption that, because of the ascertainable underlying structure of the microworld, it is indeed possible to have knowledge about complementary quantum phenomena. NOTES Jammer (1966),Chap.3. Bohr (1913). Honner (1987), Chap. 5; Murdoch (1987), Chap. 4 and 10 ; Faye (1990), Chap. 6 and 8; Favrholdt (1994). Faye (1990), 206. Murdoch(1987), 221. Scheibe (1972), 22-27; Murdoch (1987), 57-61; Faye (1990), Bohr (1929a), 16. Bohr (1929a), 10. Bohr (1927), 53; (1929a), 4. Bohr (1927), 53. Bohr(1929a), 21.

306 308 F.O. ENGLER Bohr (1929a), 19. See also Faye (1990), 221f. Bohr (1929a), 21f. Bohr (l929a), 18. Poincaré (1913), 140f. The importance of the passage quoted above has been expressed by Elie Zahar in that it encapsulated the whole of Poincaré s position on the status of scientific theories (See Zahar (1994), 57.). For a similar appraisal regarding Poincaré s structural realism which, moreover, has recently been rediscovered as a most arguable position in the battle between scientific realism and antirealism, I refer in addition to Worrall (1989) and Psillos (1999), Chapt. 7. The term,,rutherford Memorandum was coined by J. L. Heilbron and T. S. Kuhn (1969), 244. Bohr (1972), 559. Bohr (1912), 136. Jammer (1966), 74f.; Hoyer (1981), ; Rigden (2001), Bohr (1912), 136. Bohr (1912), 137. Bohr (1972), 579. Bohr (1963), XXXIX. Jammer (1966), 75. Bohr (1913), 162. Jammer (1966), Bohr (1913), 180. Bohr (1913), 175. Sommerfeld (1916). Sommerfeld (1916), 4f. Sommerfeld (1916), 9. Jammer (1966), Bohr (1929b), 103. REFERENCES Niels Bohr (1912): The Rutherford Memorandum, in: Niels Bohr: Collected Works, Léon Rosenfeld, General Editor, Volume 2, Work on Atomic Physics ( ), Ulrich Hoyer, Editor, (North-Holland, Amsterdam), 1981, Niels Bohr (1913): On the Constitution of Atoms and Molecules, Papers of 1913 reprinted from the Philosophical Magazine with an Introduction by Léon Rosenfeld, (Munksgaard Ltd., Copenhagen, W.A. Benjamin, New York), Reprinted further in: Niels Bohr: Collected Works, Léon Rosenfeld, General Editor, Volume 2, Work on Atomic Physics ( ), Ulrich Hoyer, Editor, (North-Holland, Amsterdam), 1981, Niels Bohr (1927): The Quantum Postulate and the Recent Development of Atomic Theory, in: Bohr (1934), Niels Bohr (1929a): Introductory Survey, in: Bohr (1934), Niels Bohr (1929b): The Atomic Theory and the Fundamental Principles underlying the Description of Nature, in: Bohr (1934), Niels Bohr (1934): Atomic Theory and the Description of Nature. Four Essays with an Introductory Survey, (Cambridge University Press, Cambridge). Niels Bohr (1972): Niels Bohr: Collected Works, Léon Rosenfeld, General Editor, Volume 1, Early Work ( ), J. Rud Nielsen, Editor, (North-Holland, Amsterdam). David Favrholdt(1994): Niels Bohr and Realism, in: Faye, Folse (eds.) (1994), Jan Faye (1990): Niels Bohr: His Heritage and Legacy. An Anti-Realist View of Quantum Mechanics, (Kluwer Academic Publishers, Dordrecht).

307 STRUCTURE AND HEURISTIC 309 Jan Faye and Henry J.Folse (1994) (eds.): Niels Bohr and Contemporary Philosophy, (Kluwer, Dordrecht). John L. Heilbron, Thomas S. Kuhn (1969): The Genesis of the Bohr Atom, in: Historical Studies in the Physical Sciences, Volume 1, (Philadelphia), Ulrich Hoyer (1981): Introduction to Part II: Constitution of Atoms and Molecules, in: Niels Bohr: Collected Works, Léon Rosenfeld, General Editor, Volume 2, Work on Atomic Physics ( ), Ulrich Hoyer, Editor, (North-Holland, Amsterdam), Max Jammer (1966): The Conceptual Development of Quantum Mechanics, (Tomash Publishers, American Institute of Physics), Dugald Murdoch (1987): Niels Bohr s philosophy of physics, (Cambridge University Press, Cambridge). Henri Poincaré (1913): The Foundations of Science: Science and Hypothesis, The Value of Science, Science and Method, (The New Science Press, New York), Stathis Psillos (1999): Scientific Realism. How science tracks truth, (Routledge, London). John S. Rigden (2001): The quantization of the atom in three acts, in: Annalen der Physik, Volume 10, Number 1-2, (Wiley-VCH, Berlin), Erhard Scheibe (1973): The Logical Analysis of Quantum Mechanics, (Pergamon Press, Oxford). Arnold Sommerfeld (1916): Zur Quantentheorie der Spektrallinien, in: Annalen der Physik, Volume 51, 1-94, John Worrall (1989): Structural Realism. The Best of Both Worlds?, in: David Papineau (ed.): The Philosophy of Science, (Oxford University Press, Oxford) 1996, Elie G.Zahar (1994): Poincaré s Structural Realism and his Logic of Discovery, in: Jean-Louis Greffe, Gerhard Heinzmann, Kuno Lorenz (eds.): Henri Poincaré. Science and Philosophy, (Akademie Verlag, Berlin und Albert Blanchard, Paris), 1996, Institut für Philosophie Universität Rostock Augartenstraße 34 D Germany olag.engler@philfak.uni-rostock.de

308 UWE CZANIERA THE NEUTRALITY OF META-ETHICS REVISITED HOW TO DRAW ON EINSTEIN AND THE VIENNA CIRCLE IN DEVELOPING AN ADEQUATE ACCOUNT OF MORALS 1. NONCOGNITIVIST META-ETHICS AND THE IMMORALITY REPROACH Moral philosophy has not been the central issue on the agenda of the Vienna Circle and Logical Empiricism. At the very least, we might say that its contributions to moral philosophy do not constitute the work the Vienna Circle is famous for. Even so, this philosophical movement has become notorious for its theory of morals. This theory is the meta-ethical theory called noncognitivism. Noncognitivism has been put forward in several forms, but here is the formulation which I think captures the spirit of the theory most comprehensively: Noncognitivism: Moral judgements or principles do not express knowledge. Moral judgements and principles are therefore not capable of being true or false. Those instruments which have been proposed in the project of assessing moral quality (revelation, reason, intuitions, emotions, experience) are either not cognitive instruments, or they deliver no assessments of moral quality. We find this position expressed in several famous quotations: When we read the description of a murder with all its physical and psychological details, the bare description of these facts will contain nothing that we could call an ethical sentence. The murder will be located on the same level as every other incident. It may happen that the description calls forth our grief or our anger, or it may be possible that we read something about the grief or anger called forth by this murder in others, but those are only facts, facts, and once more facts, but not ethics. 1 Either one provides empirical criteria for the predicates used in the normative sciences or one does not. In the first case, a sentence containing such a predicate becomes an empirical judgement of fact, but no value judgement. In the second case, it becomes a meaningless sentence [a Scheinsatz, as Carnap would have it]; a sentence expressing a value judgement would be impossible to form. 2 We discover that moral philosophy simply consists in the statement that ethical concepts are pseudo-concepts and therefore not fit for analysis F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers, Printed in the Netherlands.

309 314 UWE CZANIERA Noncognitivist meta-ethics managed to stir up much controversy. This was largely due to the suspicion that noncognitivism might have destructive effects on practical morality study ethics and morals decline! I would like to call this suspicion noncognitivism is dangerous to the public the immorality reproach. We find its typical formulation in a letter written by Rudolf Carnap to Otto Neurath, where Carnap talks about Oskar Kraus: As it was told me, he has said in his seminar that my view, according to which ethical norms are not scientifically justifiable, is dangerous to the public. He has also asked himself whether it was not his duty to bring me to trial. But he would not do it, since I myself would certainly not believe what I say. 4 Of course, even if there were such effects, this would not show the falsity of the incriminated ideas. In any case, it remains to be shown whether the immorality reproach is valid or not. And when this is done, we have to see whether the result sheds some interesting light on moral philosophy in general. So let us at first ask: Is the immorality reproach sound or is it defective? 2. EVALUATION OF THE REPROACH 2.1 Input vs. Output Neutrality of Meta-ethics My answer to this first question runs as follows: Yes; under certain circumstances there may be an important grain of truth to be found in the suspicion. Most members of the Vienna Circle and Logical Empiricism would have denied this (a member who would have assented to my answer was Otto Neurath). They would have maintained that the results of philosophical analysis on a metaethical level could not affect moral practice on the ethical level ( neutrality of meta-ethics (= NOM)). But if we distinguish between two meanings of NOM, we see that in one decisive respect they were not right: In an input sense, NOM seems to be true: From a meta-ethical point of view, moral principles provide the input for an investigation that does not necessarily miss its aim if it fails to justify these principles. The leading questions of meta-ethics are Herbert Feigl s famous two questions tailored appropriately: What do we mean if we talk about a moral principle? and How do we know it? The question is not: How can we make sure that every moral principle survives scrutiny? We might propose the following methodological standard for meta-ethical theories: that they regard norms and value judgements merely as their subject of inquiry. If there is a true meta-ethical theory, then it should be derivable by separate philosophers even when they disagree on certain ethical issues. Put in this input sense, I assume that the

310 THE NEUTRALITY OF META-ETHICS REVISITED 315 logical empiricists and at least many moral philosophers would agree to NOM. In an output sense, NOM seems to be false: Output neutrality of meta-ethics would mean that meta-ethical views do not influence ethical views. But it seems conceivable that the moral decisions of a person are, at least to a certain degree, influenced by her assumptions concerning the nature of moral values. I think that there is nothing contradictory in the idea that noncognitivism may knock the bottom out of a given moral position. Consider the following example. The leader of a rough gang of youths might say: What I order is right. I order you to kick the dog; therefore, it is right to kick the dog. A weak-minded member of the gang might obey his leader at first, but might come to another conclusion after reading Ayer s Language, Truth, and Logic : Well, the noncognitivist analysis of our leader s orders yields the following result: Our leader s commands express her feelings. Therefore, if she orders us to kick the dog, she wants me to kick the dog. Reformulated this way, the order of the leader loses authority, and our member might say: Hey, you are only telling us what you want us to do. There are no binding reasons behind your orders, and I will not obey them any longer, since I myself like dogs! It is not even necessary that our member be confronted with the theory of noncognitivism if it is the right and complete metaethical theory, a rather minor occasion that initiates reflection upon values should yield a similar result. This might be generalized. It is conceivable that some people think that certain acts should be committed or omitted because certain propositions about values or duties are right. And it is certainly logically possible that these people will not think likewise about those acts if they come to think that the relevant propositions about values and duties are, in fact, meaningless propositions. If a moral judgement MJ is asserted, and either a noncognitivist replies or one s own reflection shows that there are no reasons to believe in the truth of MJ, then this might be relevant for the adherents of MJ even if the judgement about MJ is not conceived of as a moral judgement itself, but as a meta-ethical judgement. It is then not neutral with regard to the ethical level. 2.2 Incompleteness of Noncognitivism The obvious reaction to meta-ethics lack of output neutrality is to reject noncognitivism. How should it otherwise be possible to explain the relative stability and uniformity of moral systems? If noncognitivism is right, should we not expect a much more variable and much more scattered pattern of morals to prevail in the world? I, for one, do not think that this would be the right approach to easing the tension between

311 316 UWE CZANIERA (1) (2) (3) the falsity of output NOM, the stability and uniformity of moral evaluation systems, and noncognitivism. It is not the right approach because there seems to be no possibility of building up a satisfactory cognitivist meta-ethics. Cognitivism, although proposed in a great variety of theories, either runs into relativism or cannot explain why moral values should be of any relevance to our practical life. 5 The right approach seems to me to be a move that may be called a Humean move, adding a fourth element to the scene a theory that explains (2) in light of the fact that (3) is correct. Noncognitivism is not false, but incomplete. After all, this also seems to have been the view of the logical empiricists. They did not blindly accept output NOM, and I even think that a major part of logical empiricist moral philosophy can be read as being devoted to the project of combining (3) and (2). They saw that noncognitivism and output neutrality cannot both be asserted simulteanously and coherently. In the light of (1), one has to provide reasons that show how (3) and (2) might be combined. But whatever they saw, the logical empiricists did not offer a satisfactory, comprehensive and systematic treatment for these problems. It is Schlick who comes closest to such a treatment. 6 How could such a systematic treatment look, and what would be the consequences? 3. CONSEQUENCES ON THE B ACKGROUND OF EINSTEIN S D ISTINCTION BETWEEN CONSTRUCTIVE THEORIES AND PRINCIPLE THEORIES 3.1 Einstein s Distinction In answering the question posed above, we may profit from a distinction drawn by Einstein, who distinguished between constructive theories and principle theories. The aim of constructive theories is to provide a comprehensive understanding of a certain class of phenomena. To achieve this aim, constructive theories state hypothetical, conjectural principles providing unifying explanations. You take a small quantity of laws or principles and a small quantity of different entities, and build up a rich variety of phenomena which at first glance do not seem to be theoretically connectable. Einstein classified the kinetic theory of gases as a constructive theory it reduces mechanical, thermal, and diffusional processes to movements of molecules. It uses the hypothesis of molecular motions to build up this variety of phenomena. Another example would be the theory of evolution. According to Einstein, the advantages of constructive theories are completeness, adaptability, and clearness.

312 THE NEUTRALITY OF META-ETHICS REVISITED 317 Principle theories, on the other hand, do not start with hypothetically constructed principles. They start with empirically discovered characteristics of natural processes and search for a mathematically specified theoretical representation of them. This representation can then be used to stipulate precise conditions for the development of the processes in question. We may therefore say that the aim of principle theories is not explanatory, but rather predictive. If a principle theory succeeds, it succeeds in predicting and governing certain phenomena. Einstein classified the theory of relativity as a principle theory. According to Einstein, the advantages of principle theories are logical perfection and the security of the foundation. 3.2 Einstein s Distinction Applied to Ethics What happens if we apply Einstein s distinction to Vienna Circle and logical empiricist moral philosophy? (1) I think that one can see immediately that the meta-ethical theory of noncognitivism has to be classified as a constructive theory of morals. It is a theory that states one principle we should rely on when trying to provide a comprehensive understanding of morals. This principle is to be conceived of as hypothetical and conjectural, and it states that there are no objective values at the basis of our moral behaviour. We may use this principle to construct a unifying explanation of the phenomenology of morals. (2) The enterprise of normative ethics would lead to a principle theory of morals. This theory would be aimed at developing a theoretical representation of morals on the basis of the observable moral life. The principles governing this representation could then be used to classify real-life phenomena into morally acceptable and morally unacceptable ones. If someone operating on the meta-ethical level asserts the doctrine of output NOM, we may reconstruct this false assumption as being due to the premises that a) practical implications may only result from a principle theory of morals, and b) constructive theories have no implications for principle theories. After having classified noncognitivism and normative ethics in the described way, we face a requirement and a prospect: 3.3 The Requirement Vienna Circle and Logical Empiricist moral philosophy amounts largely to the proposition that there can be no principle theory of morals (that is, that there can be no normative ethics). This proposition is justified by giving a certain constructive theory of morals, namely noncognitivism. But this is not enough. There is no sharp boundary to be drawn between constructive theories and principle

313 318 UWE CZANIERA ones. The doctrine of output NOM is presumably false. It is therefore not sufficient to merely give a constructive theory that says why there can be no principle theory. The constructive theory must also say why it does not matter if there is no principle theory. Vienna Circle and Logical Empiricist moral philosophy imply the question: What is the cement of morality if it does not consist of values? Philosophers like Schlick, Carnap and Menger seem to have supposed this, but they did not deal with the question in an explicit and systematic way. I think that we find the most promising theories concerning the cement of morality along the line of a tradition that leads from David Hume to John Leslie Mackie. These theories try to show very broadly speaking why it is in our own self-interest to behave morally. They see widespread preferences of the members of a certain species inhabiting a certain world as underlying morality. If such theories succeed, they have provided a theory of allegedly objective values that is itself devoid of the highly problematical assumption of objective values. 3.4 The Prospect If we find a satisfactory noncognitivist theory of the foundations of morality, we have a constructive theory of morals that provides the main principles underlying our moral behaviour. There is no apparent reason why we should not make use of them when facing moral problems. It seems not to be a bad idea to consult the principles of our usual moral thinking when working out strategies to deal with new moral challenges. And since it is the task of a constructive theory of morals to spell out these principles in detail, we may say that the constructive theory of morals can provide reasonable moral advice. The constructive theory does inform us about the commands that would apply in a certain situation if we were willing to stick to our ordinary morality. This is, of course, a highly pragmatic approach. There is no compelling reason to proceed this way. Given an interest in the welfare of people, it might be a rational route to take, but there is no force of whatever nature ethical, physical, or epistemological that compels us to take this route. Nevertheless, if we do, there may result improvements in working out moral principles and perhaps even a principle theory of morals on the basis of the constructive theory. In that case, we would have overcome a ban on normative ethics without giving up its basic premise noncognitivism. 4. CONCLUSION Vienna Circle and Logical Empiricist moral philosophy should not be conceived of as being merely aimed at destroying the whole enterprise of ethics. Its critical examination may lead to a systematic, comprehensive, and even practically relevant understanding of morals.

314 THE NEUTRALITY OF META-ETHICS REVISITED 319 NOTES Ludwig Wittgenstein, Vortrag über Ethik, in: Ludwig Wittgenstein, Vortrag über Ethik und andere kleine Schriften. Frankfurt/M.: Suhrkamp Verlag 1989, pp. 9 19; p. 12f. Rudolf Carnap, Die Überwindung der Metaphysik durch logische Analyse der Sprache, in: Erkenntnis 2, 1931, pp ; p Alfred Jules Ayer, Sprache, Wahrheit und Logik. Stuttgart: Reclam Verlag 1987, p Rudolf Carnap on Oskar Kraus in a letter to Otto Neurath in See, for example, Uwe Czaniera, Gibt es moralisches Wissen? Paderborn: mentis Verlag See Moritz Schlick, Fragen der Ethik. Frankfurt/M.: Suhrkamp Verlag Department of Philosophy University of Bayreuth D Bayreuth Germany Uwe.Czaniera@uni bayreuth.de

315 DAGMAR BORCHERS NO WOMAN, NO TRY? ELSE FRENKEL-BRUNSWIK AND THE PROJECT OF INTEGRATING PSYCHOANALYSIS INTO THE UNITY OF SCIENCE AIM AND THESIS Objective Today, the Vienna Circle, once founded for the discussion of scientific advancements, has itself become an object of research. Up to now, a lot of previously unknown details about its members and their work have been presented and discussed on various occasions. Nevertheless, much remains to be done concerning the female members of the Vienna Circle and their contributions to a project called wissenschaftliche Weltauffassung. One of them is the psychologist Else Frenkel-Brunswik. 1954, in her American exile, Frenkel-Brunswik published the paper Psychoanalysis and the Unity of Science 1, presenting arguments in favour of the integration of psychoanalysis into the unity of science. In her view, the work of Freud could count as a serious scientific theory primarily for two reasons: First, he himself introduced his disputed theoretical concepts like the unconscious or instinct in full cognizance of their scientific doubtfulness. But according to Frenkel-Brunswik, his conceptual methodology is by all means compatible with those methods found, for example, in modern physics. As she sees it, the controversial concepts are necessary to intermediate between surface phenomena and the underlying deep-seated mental structures. Second, modern empirical research in psychology has proved quite a few of those hypotheses put up by Freud and other early psychoanalysists. So if it is justified to regard psychoanalysis as a scientific theory, it will at the same time be clear that this discipline is an unproblematic part of the unity of science. Claim How do we assess these questions today? A present-day re-evaluation is not a simple enterprise. On the contrary, it would seem to be a rather complex inquiry for at least three reasons: 323 F. Stadler (ed.), The Vienna Circle and Logical Empiricism: Re-evaluation and Future Perspectives, Kluwer Academic Publishers. Printed in the Netherlands.

316 324 DAGMAR BORCHERS The dynamic development of psychoanalysis itself; The dynamic development of our assessment of what can count as scientific ; The changes concerning the conception of the unity of science. The result of this inquiry will be different according to the dimension we examine. Is it possible to find a point in this three-dimensional space confirming Frenkel-Brunswik s thesis about the unproblematical integration of psychoanalysis into the unity of science? Whatever the answer might be, taking up Frenkel- Brunswik s question is worthwhile, and will hopefully open some new perspectives for psychoanalysis itself. It should be clear that this article can only provide some very brief sketches of these three different developments. I want to show the following: Although there are new ways describing scientific theories that seem to comply with Frenkel-Brunswik s intentions, the development within psychoanalysis itself does not, so that, altogether, the prospects for her project have become even worse. Let s begin by looking at how the enterprise of a unified science developed. 1. UNITY OF SCIENCE PROSPECTS AND PROBLEMS Unity of Science within the Vienna Circle The unity of science was a central aim within the Vienna Circle. Two basic epistemological assumptions formed the background of this project 2 : 1. All cognition stems from experience. (Basis Theorem) 2. There are only two classes of meaningful sentences: Analytical statements a priori and synthetical statements a posteriori. (Sense Theorem) The consequence of this two theorems is a verdict on all kinds of metaphysics. Thus, it makes sense interpreting the project of the unity of science as an attempt to determine all sentences permitted by the sense and basis theorems. The aims of the project are the following: To provide, first, a clear distinction between science and pseudoscience, and, second, a proof of the inner compatibility of the different sciences. For us, it is important to realize that a dualism of science and humanities, of explaining and understanding has been rejected within the Vienna Circle. The central demand (following the two theorems) of this project is that all synthetical sciences be formulated in a language whose descriptive elements are either related directly to observable objects; or belong to an unproblematic class of simple predicates; or are reducible to the simple basic vocabulary by explicit definitions.

317 NO WOMAN, NO TRY? 325 A language that seems to fulfill these conditions and, therefore, serves the aims of the unity of science is the physicalistic language (including qualitative as well as quantitative expressions, as long as the former are refering to observable relations and features), which seems to have two necessary features: intersubjectivity, since its terminology can basically be understood by everybody and, second, universality, the ability to express each and every fact. But how should the necessary reduction proceed, especially in the case of psychology? The idea was to reduce the terminology of psychology to the terminology of biology which, in turn, should be translatable into the terminology of physics. According to Frenkel-Brunswik, Freud himself shared this methodological vision: While in the beginning Freud was intensively dominated by neurophysiological thinking, decisive progress in psychoanalysis did not occur until after he freed himself from the search for such analogies and turned to more openly psychological models. However, Freud always considered this step to be temporary and necessitated by the present imperfect state of the biological sciences. 3 It is an interesting fact that protagonists of the Vienna Circle like Schlick, Neurath and others 4 have been very optimistic concerning the possible integration of psychology thinking mainly of a special kind of behavioristic psychology. Nevertheless, they were also interested in psychoanalysis and its development, thinking that it would be able to overcome those methodological difficulties it shared with other totally new disciplines, finally turning out to be a real science. Very soon, behaviorists, gestalt psychologists, reflexologists, individual psychologists, psychoanalysists will couch their tenets in a physicalistic language able to successfully compare them. Thus, there will develop the exhilarating scientific atmosphere we already know from physics. And all this will be achieved without sacrificing anything of the wealth that psychology offers so far. 5 Unity of Science Today Today, the unity of science is a controversial subject. Generally, the interest in this project has decreased. For many scientists research on complex systems seems to be more interesting than thinking about the unification of research. Instead of searching for the unity of all scientific disciplines, the autonomy of the sciences is stressed: There is a widespread demand for pluralism instead of methodological monoculture. Furthermore, it is not clear how such an enterprise could be successful at all. For example, there are controversies about possible and sensible forms of unity: Unity of method? Of terminology? Of structure? As all sciences run through a development of further differentiation, it turned out that questions like this were much more complicated than they seemed to be at the beginning of this project. According to Manfred Stöckler 6 in his inventory,

318 326 DAGMAR BORCHERS the prospects for the unity of science are ambivalent: Success and failure of those reductions that already took place do not offer an homogeneous picture stressing the successful operations implies the danger of a one-sided diet; stressing the failures does not prove the impossibility of the whole project. Stöckler himself votes for a compromise between radical reductionism and total pluralism he sees the project of the unity of science as a regulative idea, a standard for the seriousness of the different disciplines a methodological corrective and interdisciplinary vision. Not giving up the whole idea might be reasonable in light of the necessity of intersubjectivity and the testability of scientific hypotheses on the one hand as well as the necessity of interdisciplinary research and cooperation on the other both require a common terminology and mutual understanding among the various disciplines. Conclusion Although there are many scientists who postulate the death of the whole project, for the moment we will assume that the search for unity still is a sensible and interesting challenge to modern science. Even if integration into the unity of science might not be the main interest of psychoanalysists today, the prove for the status of psychoanalysis as a serious scientific enterprise has, fortunately, not been totally given up. Let us look, now, at what kind of definitions of scientific theories are offered within the philosophy of science. 2. PHILOSOPHY OF SCIENCE AND THE STRUCTURE OF THEORIES Within the Vienna Circle as well as within the philosophy of science in general the description and definition of scientific theories turned out to be a difficult undertaking. What I want to show in this section is that, beyond the statement view as a first proposal, there have been promising developments within the philosophy of science in the analysis of the structure of theories. Two proposals in particular seem to offer new perspectives to psychoanalysis Lakatos remarks on the non-statement-view and Einstein s distinction between constructive and principle theories. The Statement View Essential parts of Rudolf Carnap s philosophical work are dedicated to the analysis of the structure of empirical cognition. In his book Der logische Aufbau der Welt 7 he made his first attempt to show how the concepts of empirical theories have to be interrelated to secure their empirical significance.

319 NO WOMAN, NO TRY? 327 (Carnap later developed different systematizations, never being satisfied with his own proposals.) Generally, a theory is defined as a class of statements of different kinds, formulated in two different languages: the theoretical language T which contains theoretical terms (like Energy E, Electron etc.) and logical symbols (needed to formulate the relations among objects) and the observational language O which contains observables (like colours, lengths etc.) and logical symbols (needed to formulate direct observations). Within the theoretical language T, the theory itself is formulated. To become an empirical theory, T has to be empirically interpreted. This interpretation works with correspondence rules that guarantee the empirical significance of the theory, otherwise in danger of being mere metaphysical speculation. They connect certain statements of the theoretical language T with certain statements from the observational language O. Some theoretical terms will gain empirical content by this procedure. All other theoretical terms gain their empirical meaning by way of axioms and theorems of the theoretical and definitional chains that connect the uninterpreted terms with the interpreted ones. It follows that only those hypotheses that contain interpreted terms are directly testable, all others only indirectly. To ensure that the latter are not mere speculation, it is necessary to put up an empirical criterion for the uninterpreted theoretical terms within those hypotheses. Carnap demands that a hypothesis containing an interpreted theoretical term possess prognostical relevance to count as empirically meaningful. There has to be at least one statement of T containing this term that leads to forecasts not possible without it. The elimination of the theoretical term would render the forecast no longer possible. So any scientist will have to provide an interpretation by making forecasts on the basis of the uninterpreted theoretical terms. Problems for Psychoanalysis This analysis of Carnap forms the background for the optimistic evaluation of psychoanalysis as a scientific theory made by Frenkel-Brunswik. What items must a psychoanalytical theory fulfil with regard to the statement-view? First, there must be a clear conceptual scheme, a clear distinction between theoretical and observational terms, where the theoretical ones must have empirical significance. 8 In her paper, Frenkel-Brunswik argues for the legitimacy and the scientific status of psychoanalytical concepts. These concepts are either theoretical ones like the unconscious or instinct, carefully introduced by postulates, or concepts that can be empirically defined by reductions. In fact, the controversial concepts of psychoanalysis are necessary to intermediate between surface phenomena and the underlying deep-seated mental structures. In her view, Freud s conceptual methodology is by all means compatible with methods found, for example, in modern physics. Freud obviously was aware of the necessity of providing a clear conceptual scheme, for he always stressed the hypothetical character and the explanatory function of his terminology. Nevertheless,

320 328 DAGMAR BORCHERS Frenkel-Brunswik admits that a strict distinction between these two classes of terms is not possible within psychoanalysis. For example, the term the unconscious could be classified as an abstract theoretical term as well as an observational one, since there are links to observable facts such as the behaviour of the patient during the therapy, his dreams and associations. Although she argued passionately for her view, it has always been controversial. Today, this assessment would no longer be supported, as diverse studies show that the real difficulties are enormous: Freud did not always use his own terminology in a consistent way and in many cases it is not possible to provide the necessary empirical interpretation. 9 Second, the possibility of the systematization or axiomatization of the theory must be given. Even Frenkel-Brunswik concedes that this might be a problem: she agrees that improvements concerning its axiomatization would be useful, especially a more systematic differentiation between the basic assumptions and their derivations. The aim of this effort should be to uncover logical contradictions and empirical gaps as well as insufficient evidence. D. Rapaport repeated this assessment in 1970 with regard to Freud. 10 A third demand would be the testability and applicability of the theory. Just where this test actually should take place has always been a controversial question: within psychoanalysis or independently of its own methods within other disciplines as, for example, empirical psychology. 11 Frenkel-Brunswik claimed that a lot of psychoanalytical hypotheses had been tested successfully within empirical psychology. Today, there are different opinions concerning the success of empirical tests within psychology: Most of the various inventories done by different psychologists do not support Frenkel-Brunswik s optimistic assessment: As P. Kline 12, H.-J. Eysenck and G. Wilson 13, and R. Sears 14 showed, many psychoanalytical hypotheses failed empirical tests or proved to be totally untestable. Conclusion The fact that there are weighty problems for psychoanalysis concerning the statement-view tells less about its scientific credibility than it would seem to. The first reason for this surprising assessment is that the statement-view itself got into difficulties it has been critizised a lot within philosophy of science. The main critical point is the fact that a strict and clear distinction between theoretical and observational concepts has never been successfully carried out in any discipline and seems nearly impossible with regard to real theories. Furthermore, even the protagonists of Logical Empiricism did not think that the demands of the statement view should be all that count. There might be other reasons to hold on to psychoanalytical theories: Obviously it would be difficult to prove that the theories of Freud or Jung could satisfy the criteria of acceptance advanced by Logical Positivism and shaped according to the

321 NO WOMAN, NO TRY? 329 model of mathematical physics. [...] Hence, the truth of Freudian or similar theories may not be understood otherwise than pragmatically. It may or may not be convenient to accept them. But the convenience of a theory like psychoanalysis depends on a great many factors, among which agreement with observations is only one. We have always to consider its agreement with the experience of everyday life, with the general philosophy of the period, its ability to support some ways of life, some political, moral, and religious creeds. [...] It depends upon whether one believes that a theory like that of psychoanalysis provides important practical help in life or not. 15 Getting away from the statement view obviously could have some advantages for psychoanalysis and indeed, there is at least one alternative that picked up the insight quoted above and therefore seems to be a promising new option: the nonstatement view. Advantages of the Non-statement-view After a long period of discussing the central epistemological premises of Logical Empiricism (especially sense and basis theorem) it became clear that even empirical theories would have massive problems with respect to their secure foundation indeed, security would in no way be available. All data and facts are themselves theory-laden, depending on the assumptions that have been made during their acquisition. This means that if there arises a clash between the prognosis made by a theory and an observation, then we do in fact have a collision between two inconsistent statements that must not compellingly be dissolved in favour of the observation. The non-statement view, developed by philosophers such as J. Sneed16, W. Stegmüller17 and U. Gähde18, describes the structure of theories in a totally different way than the statement view. According to Imre Lakatos 19, a theory consists of a hard core and a soft belt: The core includes the central thesis of the theory. The soft belt includes the paradigmatic field of applications and further intended applications as well as various supplement hypotheses which help to perform these applications. When under attack, a scientist will always try to protect the core of his theory and if necessary make modifications within the soft belt, where it is in principle possible to introduce supplementary hypotheses or make some alterations of less important assumptions. For us, the following thesis of Lakatos concerning the collisions of two statements is most interesting: Even if there are such abnormities, a (wrong) theory can function as a problem-solving tool as long as it is methodologically valuable. It is legitimate to keep a theory if the following conditions are fulfilled: 1. There is no competitive theory which is confronted to a lesser extent with such abnormities.

322 330 DAGMAR BORCHERS 2. The theory is able to develop supplementary hypotheses that make fruitful new prognoses and insights possible can themselves be empirically tested in a successful way. A theory that fulfills these demands is a progressive one, while its failing counterpart is referred to as degenerative. For psychoanalysis, this view obviously has interesting consequences: It implies a liberalization concerning the demands on scientific theories. For example, even if the thesis about the Oedipus Complex should not be true in a strict sense, it will be accepted as long as it produces interesting explanations of behavioral patterns and forecasts of future behavior. With regard to legitimate methodologies, strategies and hypotheses, the non-statement-view implies heterogenity as long as theories are fruitful, they will be welcomed. Finally, it seems to be a position that does not see the plurality of approaches within one discipline as being a defect: renewal and changes within a discipline are seen as principally progressive forces. This means that the aim of pinning psychoanalysis down to one kind of method, one object of research and one terminology will no longer be acceptable. Therefore, it seems to be a view that is made for the defence of outsiders and minorities as long as they are open to criticism and renewal. Perspectives & Problems Even within this description of scientific theories, there are difficulties lying in wait for psychoanalysis. In fact, it has to fulfill new and, as it seems, equally difficult demands. First, it has to be structured in the way described making clear the thesis within the hard core, as well as the assumptions and hypotheses within the soft belt. It is promising that this step indeed seems to be possible: As Wolfgang Balzer and Phillio Marcou showed, a reconstruction of Freud s early theory of the unconscious will be possible within a structuralist framework. It [...] lends itself to a completely precise logical treatment and has all the features of empirical theories in the social sciences in general. 20 Furthermore, in case of a collision, it has to develop hypotheses with progressive impact and demonstrate that the progressive ones really do have some fruitful and innovative impact. It has to overcome its tendency to react with immunisation strategies, as happened frequently when parts of Freud s work were under attack. 21 Even for the statement-view, psychoanalysis has to take care for the testability and applicability of its hypotheses. As we have seen before, this is at least a highly problematic, if not a lethal demand even today. Neverthelesss, if Balzer and Marcou are right, then the advancements within philosophy of science turn out to be advancements for the scientific claims of psychoanalysis as well.

323 NO WOMAN, NO TRY? 331 Einstein s Proposal Finally, I will come to a second proposal made by Albert Einstein which corresponds to Frenkel-Brunswik s assessment of psychoanalytical theories. Stressing the methodological advantages of psychoanalysis, Frenkel-Brunswik herself gives us an inspiring lead as to what kind of scientific theory psychoanalysis is: Freud considers the assumption of the unconsciousness as necessary because the data of the consciousness are exeedingly defective. Conscious acts alone do not enable us to account for the parapraxes and dreams of healthy persons and of the mental symptoms or obsessions of the sick. Our most intimate daily experience introduces us to sudden ideas, the source of which we are ignorant, and to results (of mentation) arrived at in ways uncertain. Thus, Freud introduced the concept of the unconscious as an abstract, hypothetical construct [...] 22 Thus, for Frenkel-Brunswik, psychoanalysis reduces complex phenomena to a relatively simple scheme: See, for example, the concept of the unconscious and how it helps to explain a wide variety of phenomena;... construes its elements hypothetically: See, for example, Freud s introduction of new concepts and their remoteness from the observable;