Revisiting Discovery and Justification: Historical and philosophical perspectives on the context distinction (Archimedes, 14) 1402042507, 9781402042508

Citation preview

ARCHIMEDES New Studies in the History and Philosophy of Science and Technology

Revisiting Discovery and Justification Historical and philosophical perspectives on the context distinction

Edited by Jutta Schickore and Friedrich Steinle

Archimedes Volume 14

Archimedes NEW STUDIES IN THE HISTORY AND PHILOSOPHY OF SCIENCE AND TECHNOLOGY VOLUME 14

EDITOR JED Z. BUCHWALD, Dreyfuss Professor of History, California Institute of Technology, Pasadena, CA, USA. ADVISORY BOARD HENK BOS, University of Utrecht MORDECHAI FEINGOLD, Virginia Polytechnic Institute ALLAN D. FRANKLIN, University of Colorado at Boulder KOSTAS GAVROGLU, National Technical University of Athens ANTHONY GRAFTON, Princeton University FREDERIC L. HOLMES, Yale University PAUL HOYNINGEN-HUENE, University of Hannover EVELYN FOX KELLER, MIT TREVOR LEVERE, University of Toronto JESPER LÜTZEN, Copenhagen University WILLIAM NEWMAN, Harvard University JÜRGEN RENN, Max-Planck-Institut für Wissenschaftsgeschichte ALEX ROLAND, Duke University ALAN SHAPIRO, University of Minnesota NANCY SIRAISI, Hunter College of the City University of New York NOEL SWERDLOW, University of Chicago Archimedes has three fundamental goals; to further the integration of the histories of science and technology with one another: to investigate the technical, social and practical histories of specific developments in science and technology; and finally, where possible and desirable, to bring the histories of science and technology into closer contact with the philosophy of science. To these ends, each volume will have its own theme and title and will be planned by one or more members of the Advisory Board in consultation with the editor. Although the volumes have specific themes, the series itself will not be limited to one or even to a few particular areas. Its subjects include any of the sciences, ranging from biology through physics, all aspects of technology, broadly construed, as well as historically-engaged philosophy of science or technology. Taken as a whole, Archimedes will be of interest to historians, philosophers, and scientists, as well as to those in business and industry who seek to understand how science and industry have come to be so strongly linked.

Revisiting Discovery and Justification Historical and philosophical perspectives on the context distinction

Edited by

JUTTA SCHICKORE Indiana University, Bloomingdales, IN, U.S.A. and

FRIEDRICH STEINLE Bergische Universität Wuppertal, Germany

A C.I.P. Catalogue record for this book is available from the Library of Congress.

ISBN-10 ISBN-13 ISBN-10 ISBN-13

1-4020-4250-7 (HB) 978-1-4020-4250-8 (HB) 1-4020-4251-5 (e-book) 978-1-4020-4251-5 (e-book)

Published by Springer, P.O. Box 17, 3300 AA Dordrecht, The Netherlands.

www.springer.com

Printed on acid-free paper

All Rights Reserved © 2006 Springer No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording or otherwise, without written permission from the Publisher, with the exception of any material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work. Printed in the Netherlands.

CONTENTS

JUTTA SCHICKORE & FRIEDRICH STEINLE /

Introduction:

Revisiting the Context Distinction

vii

PART I: The contexts of the Distinction DON HOWARD /

Lost Wanderers in the Forest of Knowledge: Some Thoughts on the Discovery-Justification Distinction

Inductive Justification and Discovery. On Hans Reichenbach’s Foundation of the Autonomy of the Philosophy of Science

3

GREGOR SCHIEMANN /

Freedom in a Scientific Society: Reading the Context of Reichenbach’s Contexts

23

ALAN RICHARDSON /

41

PART II: Forerunners? JUTTA SCHICKORE /

A Forerunner?—Perhaps, but not to the Context Distinction. William Whewell’s Germano-Cantabrigian History of the Fundamental Ideas

57

¨ LOTHAR SCH AFER /

Autonomy versus Development: Duhem on Progress in Science

Psychologism and the Distinction Between Discovery and Justification

79

VOLKER PECKHAUS /

99

PART III: Revisions and Applications PAUL HOYNINGEN-HUENE /

Context of Discovery versus Context of Justification and Thomas Kuhn

THOMAS STURM AND GERD GIGERENZER /

How Can We Use the Distinction Between Discovery and Justification? On the

v

119

vi

CONTENTS

Weaknesses of the Strong Programme in the Sociology of Science

133

THOMAS NICKLES / Heuristic Appraisal: Context of Discovery or

Justification? Concept Formation and the Limits of Justification. “Discovering” the two Electricities

159

FRIEDRICH STEINLE /

Contexts of Justifying and Discovering the Nature of Ecosystems: From Concepts to Objects and Vice Versa

183

THOMAS POTTHAST /

On the Inextricability of the Context of Discovery and the Context of Justification

197

THEODORE ARABATZIS /

Contributors

215 231

JUTTA SCHICKORE AND FRIEDRICH STEINLE

INTRODUCTION: REVISITING THE CONTEXT DISTINCTION

The distinction between the contexts of discovery and justification has had a turbulent career in philosophy of science. At times celebrated as the hallmark of philosophical approaches to science, at times condemned as ambiguous, distorting, and misleading, the distinction dominated philosophical debates from the early decades of the twentieth century to the 1980s. In recent years, the distinction has vanished from philosophers’ official agenda. However, even though it is rarely explicitly addressed, it still informs our conception of the content, domain, and goals of philosophy of science. The fact that new developments in philosophy of experimentation and history and sociology of science have been marginalized by traditional scholarship in philosophy indicates that the context distinction still pervades philosophical thinking about science. This volume helps clear the grounds for the productive and fruitful integration of these new developments into philosophy of science. We identify several focal points for the re-assessment of the distinction: the original contexts, especially the work of the Logical Empiricists, its alleged forerunners in the nineteenth century, and its evolution and dissemination throughout the twentieth century. The distinction is usually traced back to Hans Reichenbach’s Experience and Prediction (Reichenbach 1938). In this work, Reichenbach claimed that the context of justification is the only part of scientific practice that is epistemologically relevant and open to philosophical—which, in the perspective of Logical Empiricism, means of course: logical—reconstructions. Thus understood, the distinction delineates the scope of philosophy of science—the justification of fully developed theories—and at the same time shapes its method, logical analysis. Since then, this distinction has served two important, related purposes. First, it has been used to demarcate philosophy of science proper from historical, political, sociological and other empirical approaches to science. Within the analytical framework of the context distinction, doing philosophy of science means uncovering the logical structure of scientific theories and the logical relations between theories and evidence. Secondly, the distinction is taken to imply that there is no logic of discovery and that the process of discovery is excluded from philosophical reconstruction. For several decades, the context distinction dictated what philosophy of science should be and how it should proceed. Only in the 1960s, when the influence of Logical Empiricism was beginning to wane, did the context distinction come under attack. Two main, related lines of critique can be distinguished; both of them questioning philosophy’s narrow boundaries. The first critique concerned the exclusion of discovery from the domain of philosophy of science, the second the exclusion of history and sociology. Following Norwood Russell Hanson’s Patterns of Discovery (Hanson vii J. Schickore and F. Steinle (eds.), Revisiting Discovery and Justification, vii–xix.
C 2006 Springer. Printed in the Netherlands.

viii

JUTTA SCHICKORE AND FRIEDRICH STEINLE

1958), several philosophers began to reconsider the possibility of developing a logic of scientific discovery. This task became a major topic of concern in the 1970s and early 1980s. Thomas Nickles’s two-volume collection on Scientific Discovery, Logic, and Rationality (Nickles 1980a) gathers together important contributions by the “Friends of Discovery” (Ronald Giere’s term). Only some of the contributions, however, are concerned with the possibility of a logic of discovery (e.g., Schaffner 1980). Most of them recast the problem of the logic of discovery as a question about rationality. They focus on the question of whether, and to what extent, scientific discoveries are amenable to reconstructions and explanations in rational terms. Nickles’s introduction draws attention to two important outcomes of the critical assessments of the distinction. First, it points out that the context distinction as it appears in philosophical discussions is ambiguous. It is sometimes temporally, sometimes logically construed. According to the first, temporal version, the process of discovery is followed by the process of justification. According to the second, logical version, the process of discovery is separated from the logical reconstruction, explication, and assessment of scientific achievements.1 Secondly, the introduction notes that the distinction between discovery and justification as it is applied in the actual analysis of scientific practice is somewhat arbitrary because we lack criteria for drawing a clear boundary between those elements of scientific practice that belong to “discovery” and those that belong to “justification” (Nickles 1980a, p. 19). Here the collected efforts of the Friends of Discovery indicate some alternatives. These philosophers claim that instead of a dualist scheme, we need a framework that acknowledges a middle phase of “pursuit” or “prior assessment” as part of the process of knowledge generation that already embraces elements of justification (e.g., Kordig 1978; Schaffner 1980; Darden 1998). Concern with the “logic of scientific discovery” has migrated into cognitive science, where scientific reasoning processes have been studied with the help of computational methods. Kenneth Schaffner’s contribution in Nickles’s collection is representative of this approach (Schaffner 1980). In these works, discovery is thought of as the process of finding efficient heuristic methods for the generation of plausible explanations (Buchanan 1985). Scholars working in this field seek to provide supportive evidence for the “logic of scientific discovery” with computer programs that utilize such heuristics to “make discoveries”, i.e. to represent successful discovery procedures on a computer. Because we can reconstruct the discovery processes with a set of heuristic rules, the argument goes, we can consider them as rational.2 Several Friends of Discovery (and others informed by their work) are now active in the field of philosophy of biology. Many of them have bypassed the abstract debates as to whether a “logic of discovery” is possible through concrete detailed investigations and analyses of research on biological mechanisms.3 Based on the assumption that “much of the practice of science can be understood in terms of the discovery and description of mechanisms” they show how traditional philosophical notions of discovery and theory change can be fruitfully recast in terms of “looking for” and “describing” mechanisms (Machamer et al. 2000, p. 2). In accord with the contributions to Nickles’s collection, they offer more finely grained descriptions of how accounts of biological mechanisms are generated, evaluated, and revised, which replace the neat

INTRODUCTION: REVISITING THE CONTEXT DISTINCTION

ix

dualism between discovery and justification philosophers of biology (Darden 1990; Darden 1991). The second line of critique of the context distinction came from a different angle, and it is this critique that most concerns us in the present volume. In the 1970s, a number of critical assessments were published in connection with debates about the relevance of sociological and historical studies for philosophy of science that unfolded in the wake of Kuhn’s Structure of Scientific Revolutions (Kuhn 1962).4 Provoked by Kuhn’s sweeping historical refutation of philosophers’ rational reconstructions of theory change, many philosophers argued about the nature and scope of the context distinction, aiming to explore the possibility of a historically informed philosophy of science. At an early stage of this debate, Wesley Salmon remarked that the distinction between the contexts of discovery and justification was “a major focal point for any fundamental discussion between history of science and philosophy of science” (Salmon 1970, p. 70). In the following decade, much was written about philosophy’s engagement with history and the marriage and divorce of history and philosophy of science.5 But in the end, the discussion about possible relations between history and philosophy remained inconclusive. The possibility of a historically informed philosophy of science was a promise that was never quite fulfilled. By the late 1980s, the context distinction had largely disappeared from philosophers’ official agendas. Remarkably, today the distinction is most explicitly discussed in the sciences themselves. In methodological introductions of science textbooks, it shapes the regulations for scientific research.6 These textbooks employ a particular version of the distinction, namely the context distinction temporally understood in combination with the hypothetico-deductive (H-D) model of scientific research. According to this version of the distinction, the (unsystematic) process of discovery is completed before the (regulated) process of justification and testing can begin. Ironically, scientists and science educators now attack the distinction and accuse philosophers of describing science in a too simplistic manner (e.g., Harwood 2004). They argue that this twostep scheme is frequently violated in actual scientific practice and that it is in fact an impediment for science education if it is taught as the norm of “proper scientific behavior”. This line of critique of the temporal version of the distinction is certainly persuasive—if not new—but it has not led to fruitful exchanges with philosophers of science. In recent years, philosophers of science have rarely directly addressed, let alone attacked the distinction. But this does not mean that the distinction has been rendered irrelevant or that it has been successfully refuted. On the contrary, the legacy of earlier advocates of the distinction is still effective, and the distinction continues to delineate the scope philosophy of science. Exchanges between philosophy of science, history of science and science studies have been rather sparse; in fact, the disciplines have drifted further and further apart. Most scholars interested in the history and sociology of science have simply abandoned philosophical analyses. Take for example the growing field of science studies. During the last fifteen years or so, STS has gained an increasingly strong academic foothold. But this has not led to a productive exchange between science studies and philosophy of science. On the contrary, historians and

x

JUTTA SCHICKORE AND FRIEDRICH STEINLE

sociologists have largely ignored epistemological concepts and debates in their historical studies and thick descriptions of specific scientific episodes. Often they explicitly deny that the sciences can be profitably analyzed in generalized epistemological terms (e.g., Knorr-Cetina 1981; Collins 1985; Morus 1998; Gieryn 1999). On the other hand, philosophers working in the analytic tradition continue to exclude historical as well as sociological and psychological studies of science from philosophical reflection. Some have utilized bite-size episodes of past science for the purpose of illustrating philosophical points but they rarely consider historical developments over a longer period.7 Many others simply presuppose that investigations of the material culture, the historical changes, and the cultural and social environments of science lack epistemological significance. A quick glance at current journals in philosophy of science shows that articles dealing with special philosophical conundrums that arise out of current work in the sciences such as the units of selection problem mostly draw on material from present science rather than on historical developments, even though histories of epistemological concepts, scientific experimentation, and instruments have flourished for quite some time.8 Papers in general philosophy of science that talk about realism, explanation, the status of laws and so on fall squarely into the traditional context of justification. Only very rarely does one encounter explicit attempts to bridge the gap between different metascientific enterprises.9 The persistent if unstated force of the distinction becomes manifest in professional philosophers’ reactions to certain recent developments in history and philosophy of science, such as the New Experimentalism and the history of philosophy of science. These developments have been marginalized just like the works of the Friends of Discovery 25 years ago, a fact that indicates to what extent our metascientific efforts and assessments are still governed by the very distinction that had seemingly faded away by the late 1980s. Take for example the New Experimentalism. There was a brief upsurge of philosophical interest in experimentation in the early 1980s. Just like the Friends of Discovery, philosophers concerned with experimentation study aspects of the generation of knowledge but their work is not a ramification of earlier studies of the logic of discovery. Rather, it has evolved from Ian Hacking’s new and original approach to the ongoing debates about realism and antirealism that he outlined in Representing and Intervening (Hacking 1983) and from Allan Franklin’s The Neglect of Experiment (Franklin 1986), which, inspired by Hacking, attempts to rehabilitate experiment as a topic for philosophical concern against the “theory-dominated” philosophical literature of the 1960s and 1970s.10 Franklin as well as a few other philosophers informed by his and Hacking’s works focus on the epistemic strategies by which experimental results are justified.11 Nevertheless, experimentation is still rarely considered to be a legitimate topic for philosophical analysis.12 The fact that the New Experimentalism in philosophy is of comparatively little importance13 —except for the realism— antirealism debate—can be easily explained by the fact that the New Experimentalism is a violation of the philosophical agenda implied in the context distinction. However, if it is correct that justificatory elements are an integral part of experimental research, then we need to reconsider the context distinction and its implications in this light.

INTRODUCTION: REVISITING THE CONTEXT DISTINCTION

xi

The New Experimentalism has posed challenges to traditional philosophy that have been little heeded. In contrast, new efforts in the area of history of philosophy of science have attracted more attention in the community, particularly the studies on the institutional and socio-political contexts of Logical Empiricism and its successors on the other side of the Atlantic. A number of studies have corrected the received view of Logical Empiricism that we encounter in standard philosophy textbooks.14 But remarkably little attention has been paid to the fact that history of philosophy of science can shed new light on the very context distinction and its “prehistory”. Judged by the drastic changes that novel historical studies have produced in our image of Logical Empiricism, we may expect similarly surprising results if we explore in detail the original settings of the distinction. The same holds for the “prehistory” of that distinction. Inspired by the idea of the content and delineation of philosophy of science that the received context distinction offers, philosophers usually construct the prehistory of philosophy of science as a sequence of “forerunners” that already contain important elements of that distinction. Among others, John Losee and Larry Laudan have suggested that Reichenbach merely brought into a standard form certain ideas that various philosophers had already expressed throughout the nineteenth century.15 But it is very likely that the context distinction has molded the way in which philosophers write the history of philosophy of science before the advent of Logical Empiricism. If we want to acknowledge properly the important recent work in the history of philosophy of science, the New Experimentalism, as well as the historical studies of scientific experimentation, we need to re-open the debate about the nature, development, and significance of the context distinction and about its merits and flaws. The current philosophical enterprise presents us with several major challenges. First, we need to provide close readings and detailed analyses of the original textual sources for the context distinction. What purposes and goals did this distinction have for Reichenbach and his contemporaries? Are these goals still valid and legitimate today? Secondly, we need to revise those chapters in the history of philosophy of science that have been written through the lens of Logical Empiricism. Are the main endeavors in nineteenth-century philosophy of science really just preliminary versions of the distinction and the philosophical conception that comes with it? If not, is it possible to extract elements from these earlier endeavors that can help us deal with the philosophical challenges that the New Experimentalism and historically, socio-politically and economically oriented science studies have placed before us? Thirdly, we need to map, clarify, and analyze the derivations and mutations of the context distinctions as we encounter them in current history and philosophy of science. What are its different manifestations in present discussions? Drawing on these analyses, we need to consider, fourthly, what is left of the distinction. Is there a version of the distinction that prevails and can be fruitfully applied in a historically informed philosophy of science? The essays in the present collection respond to these challenges. The contributions to the first part place the original textual sources for the context distinction in their respective historical and argumentative contexts. The essays consider

xii

JUTTA SCHICKORE AND FRIEDRICH STEINLE

Reichenbach’s work from various angles. They present close readings of the original texts (Schiemann) as well as reconstructions of the political and philosophical debates in which Reichenbach was engaged while formulating his version(s) of the context distinction (Howard, Richardson). Indeed, Reichenbach’s writings provide an excellent touchstone of the fruitfulness of a historically informed philosophy of science. Reichenbach brought up the distinction in the context of specific controversies. In the 1930s, he disagreed with Otto Neurath about whether and how philosophy of science should engage in political debates. In the 1950s, however, Reichenbach wrote against the American pragmatists, in particular John Dewey. In each case, he drew the distinction differently. To assess the nature, significance, and implications of Reichenbach’s approach, a study of the contexts of the distinction is essential (Howard). The authors take into account that philosophical ideas and systems are situated. These ideas are formulated in specific scientific, institutional, and socio-political contexts. It is only through the analysis of the contexts of the distinction that we will be able to understand fully the original intentions behind it and to assess its merits and demerits. The second section is entitled Forerunners? The question mark signals our discomfort with those scholars who have stressed that Reichenbach just brought into a standard form the distinction between discovery and justification that had long been implicitly present. The contributions to this part extend backwards in time the recent efforts in history of philosophy of science. They focus on those nineteenth-century philosophers who are commonly regarded as advocates of the context distinction avant la lettre and consider whether this interpretation really holds water. They offer new vistas of a number of these so-called “forerunners”: William Whewell, the acclaimed early advocate of a hypothetical deductivist philosophy of science (Schickore), the anti-inductivist Pierre Duhem (Sch¨afer), and Gottlob Frege and his contemporaries, who allegedly prepared the grounds for radical antipsychologism in philosophy (Peckhaus). The fresh looks at the “forerunners” that these essays offer give rise to an extremely important insight. They highlight the extent to which the reception of nineteenthcentury philosophical writings was itself shaped by the early twentieth-century advocates of the context distinction. The appropriation of the works of these nineteenthcentury thinkers in terms of the context distinction led to severe distortions of their views. Actually, these works contain a great deal of material that many twentiethcentury advocates of the context distinction would deem unphilosophical. If we approach their writings in a conscious attempt at overcoming the disciplinary limits that the early twentieth-century context distinction has forced upon us, we find that these works can contribute a lot to the enterprise of a historically informed philosophy of science. While the second part of the volume puts the emphasis on historical revisions, the third group of essays traces the evolution and ramifications of the context distinction as we encounter them in current philosophy of science. Many of the difficulties surrounding the debates about the distinction have emerged from a conflation of the different versions that have evolved from the original sources. We can remove certain problems and see more clearly the value of the distinction if we carefully distinguish

INTRODUCTION: REVISITING THE CONTEXT DISTINCTION

xiii

between evaluative and descriptive attitudes to science. This “lean” distinction separates two perspectives on scientific knowledge or two ways of looking at science (Hoyningen-Huene). None of the contributors to this volume denies the value of this “lean” distinction. Of course, the act of justification presupposes something as its object—a “happy thought”, as Whewell would have called it—and this “something” must come from “somewhere”. Hence, it appears that we also need to preserve some basic kind of temporal distinction between “generating” a knowledge claim and “securing” it. In addition, for the description and explanation of the actual fixation of scientific claims to knowledge some version of the distinction may be required (Sturm and Gigerenzer). However, the essays in this section imply in one way or another that it is awkward to regard the category of justification as one side of a stable dichotomy between justification and discovery, justification and experiment, or justification and theory construction. Justification is not “the other” or “the opposite” of theory construction, experimentation or indeed discovery but an integral part of the extended process of knowledge generation, and our reconstructions need to take this into account. Discovery, in any meaningful understanding of the concept, is a prolonged activity that involves both the generation and the fixation of knowledge claims (Arabatzis, Potthast). We can go even further and contend that the two aspects of generation and fixation are necessarily inextricably intertwined: The fixation of knowledge claims hinges on and must be formulated in terms of the very concepts that are formed and stabilized in the experimental process whose results are being secured (Steinle). Moreover, the contributions to this part also show that our reconstructions should be sensitive to the fact that the fixation of knowledge claims begins at the bench and does not stop with the published result. Ludwik Fleck’s distinction between journal, vademecum, and textbook science [Zeitschriften-, Handbuch- und Lehrbuchwissenschaft] still provides a valuable and instructive framework for the analysis of the generation and fixation of scientific facts. While the journal article allows at least a glimpse of the experimental production of results (say, in the lab)—if only in the highly regulated and schematic sections on “methods” and “observations”—the handbook, and even more so, the textbook account is completely severed from the initial research processes and indeed from the justificatory procedures that were actually employed in experimentation (Steinle). We may even say that the justificatory procedure begins before the initial research takes place. To decide about what to do next, and to obtain funding, scientists (and grant agencies) need to evaluate research plans from the perspective of future fertility (Nickles). Here lies an important topic for future research on justification. We need to trace the ways in which knowledge is secured vis-`a-vis different audiences. New studies of experimentation in combination with recent work on science and the public16 can serve as starting points for more detailed analyses of justificatory procedures. Taken together, the essays included in this volume demonstrate that two very basic conceptual distinctions need to be preserved. The first is the “lean distinction” between two questions or attitudes to science, the descriptive and the evaluative. The second is the distinction between two processes, the generation of “something”—a claim to

xiv

JUTTA SCHICKORE AND FRIEDRICH STEINLE

knowledge—that is to be justified, and the fixation of that claim. But neither of these distinctions can provide a foundation for the dualistic picture of context of justification versus context of discovery. However, if we give it up, what does this mean for the distinction between philosophy of science and science studies that is so closely tied to that bifurcation? Who is now “in charge” of the generation and fixation of knowledge claims, and how should we deal with these issues? Is the study of the generation of knowledge claims merely a descriptive task? Does the fixation of knowledge claims remain an exclusively philosophical (evaluative) category, or shall we treat it strictly as an actors’ category? Or shall we resort to the approach of the “lean distinction” and say: this is simply a question of perspective? Both the generation and the fixation of knowledge claims are major themes that call for collaborative philosophical and empirical studies of (past) science. First, it requires the combined effort of philosophy, history, and science studies to clarify what that “something” is that we have tentatively termed “claim to knowledge”— whether it is an isolated event or a statement, whether it can be credited to a single individual or to a group, how the materiality of the setting shapes that claim, and so on (Arabatzis). Secondly, the contributions to this volume show that there is no particular, distinctive “context” of justification that merits philosophical efforts— justificatory aspects are ubiquitous in science. No doubt, our perspective on the fixation of knowledge claims can be strictly descriptive. But even in this perspective, the understanding of what justification is is not merely an empirical enterprise, and in this regard the endeavor to understand justification strongly resembles our efforts to understand the generation of knowledge claims. With regard to the evaluative task, an interdisciplinary approach is also required. Philosophers need to clarify different aspects of that process in constant engagement with historical material. To answer our evaluative questions of whether and how claims to knowledge are justified, we need to consider justificatory procedures in (past) science and their evolution. We need to analyze conceptually different types of the fixation of knowledge such as the heuristic appraisal of future fertility, the practical considerations that govern the lab work, the formalized reconstructions in the journal article, and the decontextualizations in textbooks and popular magazines. We need to conduct historical research because justificatory norms and strategies, for example those inherent in guidelines for scientific publications and heuristic appraisals, change over time. We need to bring these together with the concepts of justification that philosophy of science has developed and explicated. And we need to combine history with conceptual analysis to explore how justificatory arguments evolve with the formation of the new concepts that are generated in scientific research (Steinle). We can conceive of this critical assessment as a process of iteration. This emphasizes the fact that it is of no fundamental importance whether our initial engagement is with (past) scientists’ justificatory activities or with philosophers’ epistemological analyses of the concept of justification. Both beginnings are possible, legitimate, and plausible. One may wish to call the first approach an epistemologically informed history and the second approach a historically informed epistemology, but in the end, this difference does not matter. The crucial point is that after a few iterations, the approaches

INTRODUCTION: REVISITING THE CONTEXT DISTINCTION

xv

will merge. Regardless of whether we choose to start with actual justificatory procedures or with received epistemological categories of justification, epistemological analyses of the concept of justification need history of philosophy of science and empirical explorations of justificatory processes in (past) science as their corrective. This book is the outcome of a prolonged collective effort. Many of the contributions to this volume were first presented and discussed at the conference Revisiting Discovery and Justification at the Max Planck Institute for the History of Science (Berlin) in spring 2002. Preliminary versions of almost all the contributions of the present volume were discussed again in an authors’ workshop in spring 2004. The volume has benefited greatly from these exchanges. We wish to take this opportunity to thank all participants of these meetings, especially Alessandra Allegrini, Davis Baird, Dick Burian, Jessica Carter, Lindley Darden, Jan Frercks, Giora Hon, David Hyder, Hans Poser, and Friedrich Stadler, for their instructive contributions and comments.17 Warmest thanks go to Hans-J¨org Rheinberger, who offered helpful advice and critique at all stages of our project and provided generous funding.

NOTES 1. See Nickles 1980b, pp. 8–9; see also Hoyningen-Huene 1987 for a detailed analysis of the different aspects of the context distinction. 2. See, for example, Feigenbaum et al. 1971; Simon 1973; Simon et al. 1981; Langley et al. 1987; Simon and Kulkarni 1988; Gooding 1990; Graßhoff and May 1995; Grasshoff et al. 2000; for an overview of recent work see Darden 1997. See also Nersessian 1995 for recent discussions about discovery in cognitive science. 3. For recent investigations of research on biological mechanisms, see especially Darden 1991; Darden 1996; Glennan 1996; Grasshoff et al. 2000; Craver 2001. See also Bechtel and Richardson 1993 on discovering complexity. 4. An important document of these discussions is the volume Criticism and the Growth of Knowledge (Lakatos and Musgrave 1970). It includes, among other things, Lakatos’s important paper on the methodology of scientific research programmes, which offers a rational reconstruction of Kuhn’s dynamic image of scientific revolutions (Lakatos 1970). See also Laudan 1977 for another influential contribution to this discussion. 5. See, e.g. Salmon 1970, p. 70; Feigl 1970; Giere 1973; Burian 1977; Kr¨uger 1979; McLaughlin 1982. For a revision of these debates, see Nickles 1985. 6. See Steinle’s contribution (this volume). 7. A notable exception is Ian Hacking’s work on the historical development of epistemological concepts (see esp. Hacking 1990; Hacking 1995; Hacking 1999). Perhaps not surprisingly, philosophers have focused their attention on his contributions to debates about scientific realism (Hacking 1983). Hacking’s other works are almost exclusively debated by historians. 8. See, for instance, the collections Gooding et al. 1989 and Buchwald 1995 on experimentation as well as Hankins and Silverman 1995 and Joerges and Shinn 2001 on instruments. 9. There are some notable exceptions, e.g., the recent works by Philip Kitcher and Helen Longino, which seek to bring science studies to bear on philosophical thought (see, for example, Kitcher 2001; Longino 2002). See also the contributions to Steinle and Burian 2003 and (specifically on experimentation) Heidelberger and Steinle 1998 for recent interdisciplinary endeavors.

xvi

JUTTA SCHICKORE AND FRIEDRICH STEINLE

10. In more recent publications, the target of Franklin’s critique has shifted from theory-dominated philosophy of science to social constructivism, see e.g., Franklin 1999; Franklin 1994. 11. See, among others, Rasmussen 1993; Bechtel 1994; Culp 1995; Chalmers 2003. 12. For recent philosophical approaches to instruments and experimentation, see especially Rheinberger 1997; Ankeny 2000; Baird 2004; Weber 2005; Steinle 2005 as well as a small body of work on error statistics, see e.g., Mayo 2000; Rudge 2001; Elliot 2004. Most of the recent work on experimental practice, however, is done in case studies focusing on specific examples, often from contemporary science (see, among others, Dror 1999; Rader 1999; Benschop and Draaisma 2000; Gaudilli`ere 2001; Creager 2002; Bresadola 2003). 13. The recent collection on The Philosophy of Experimentation edited by Hans Radder is a noteworthy exception. However, in his introduction (tellingly titled ‘Toward a More Developed Philosophy of Scientific Experimentation’), Radder also notes that after the ‘rapid start’ with Representing and Intervening, the philosophy of experimentation swiftly lost its momentum (Radder 2003, p. 1). 14. Cf. among others Uebel 1991; Cartwright et al. 1996; Giere and Richardson 1996; Stadler 1997; Friedman 1999. 15. See, e.g., Losee 1979; Laudan 1980; Schaffer 1986; McMullin 1990. 16. See, e.g., Cooter 1984; Golinski 1992; Yeo 1993; Winter 1998. 17. We extend our thanks to Jason Baker, Steve Crowley, and Noretta Koertge, who offered insightful comments on this introduction and made valuable suggestions for improving it.

REFERENCES Ankeny, R. (2000), “Fashioning Descriptive Models in Biology: Of Worms and Wiring Diagrams,” Philosophy of Science 67 (supplement): S260–S272. Baird, Davis (2004), Thing Knowledge: A Philosophy of Scientific Instruments (Berkeley: University of California Press). Bechtel, William (1994), “Deciding on the Data: Epistemological Problems Surrounding Instruments and Research Techniques in Cell Biology,” PSA: Proceedings of the Biennial Meeting of the Philosophy of Science Association, pp. 167–178. Bechtel, William and Robert Richardson (1993), Discovering Complexity (Princeton: Princeton University Press). Benschop, Ruth and Douwe Draaisma (2000), “In Pursuit of Precision. The Calibration of Minds and Machines in Late Nineteenth-Century Psychology,” Annals of Science 57: 1–25. Bresadola, Marco (2003), “At Play With Nature: Luigi Galvani’s Experimental Approach to Muscular Physiology”, in F. L. Holmes et al. (eds.), Reworking the Bench: Research Notebooks in the History of Science (Dordrecht, Boston and London: Kluwer), pp. 67–92. Buchanan, B. G. (1985), “Steps Toward Mechanizing Discovery”, in K. Schaffner (ed.), Logic of Discovery and Diagnosis in Medicine (Berkeley: University of California Press), pp. 94–114. Buchwald, Jed Z. (ed.) (1995). Scientific Practice. Theories and Stories of Doing Physics (Chicago and London: University of Chicago Press). Burian, Richard M. (1977), “More Than a Marriage of Convenience: On the Inextricability of History and Philosophy of Science,” Philosophy of Science 44: 1–42. Cartwright, Nancy, Jordi Cat et al. (1996), Otto Neurath: Philosophy Between Science and Politics (Cambridge: Cambridge University Press). Chalmers, Alan (2003), “The Theory-Dependence of the Use of Instruments in Science,” Philosophy of Science 70: 493–509. Collins, Harry (1985), Changing Order: Replication and Induction in Scientific Practice (London: Sage). Cooter, Roger (1984), The Cultural Meaning of Popular Science: Phrenology and the Organization of Consent in Nineteenth Century Britain (Cambridge: Cambridge University Press). Craver, Carl F. (2001), “Role Functions, Mechanisms, and Hierarchy,” Philosophy of Science 68: 53–74.

INTRODUCTION: REVISITING THE CONTEXT DISTINCTION

xvii

Creager, Angela (2002), The Life of a Virus: Tobacco Mosaic Virus as an Experimental Model, 1930– 1965 (Chicago: The University of Chicago Press). Culp, Sylvia (1995), “Objectivity in Experimental Inquiry: Breaking the Data-Technique Circles,” Philosophy of Science 62: 430–450. Darden, Lindley (1990), “Diagnosing and Fixing Faults in Theories”, in J. Shrager and P. Langley (eds.), Computational Models of Scientific Discovery and Theory Formation (San Mateo: Morgan Kaufman), pp. 319–346. Darden, Lindley (1991), Theory Change in Science: Strategies from Mendelian Genetics (New York: Oxford University Press). Darden, Lindley (1996), “Essay Review. Generalizations in Biology,” Studies in History and Philosophy of Science 27: 409–419. Darden, Lindley (1997), “Recent Work in Computational Scientific Discovery”, in M. Shafto and P. W. Langley (eds.), Proceedings of the Nineteenth Conference of the Cognitive Science Society (Mahwah, NJ: Lawrence Erlbaum), pp. 161–166. Darden, Lindley (1998). Discovery, Evaluation, Revision: Cycles in Scientific Change (abstract). International Congress on Discovery and Creativity, Ghent. Dror, Otniel E. (1999), “The Scientific Image of Emotion: Experience and Technologies of Inscription,” Configurations 7: 355–401. Elliot, Kevin (2004), “Error as Means to Discovery,” Philosophy of Science 71: 1–24. Feigenbaum, E. A., B. G. Buchanan et al. (1971), “On Generality and Problem Solving: A Case Study Using the DENDRAL Program,” Machine Intelligence 6: 165–190. Feigl, Herbert (1970), “Beyond Peaceful Coexistence”, in R. H. Stuewer (ed.), Historical and Philosophical Perspectives of Science (New York et al.: Gordon and Breach), pp. 3–11. Franklin, Allan (1986), The Neglect of Experiment (Cambridge: Cambridge University Press). Franklin, Allan (1994), “How to Avoid the Experimenters’ Regress,” Studies in History and Philosophy of Science 25: 463–491. Franklin, Allan (1999), Can That Be Right? (Dordrecht, Boston and London: Kluwer). Friedman, Michael (1999), Reconsidering Logical Positivism (Cambridge: Cambridge University Press). Gaudilli`ere, Jean-Paul (2001), “Mapping as Technology: Genes, Mutant Mice, and Biomedical Research, 1910–1965”, in B. Joerges and T. Shinn (eds.), Instrumentation. Between Science, State and Industry, Sociology of the Sciences Yearbook XXII (Dordrecht: Kluwer), pp. 29–48. Giere, Ronald (1973), “History and Philosophy of Science: Intimate Relationship or Marriage of Convenience?,” British Journal for Philosophy of Science 24: 282–297. Giere, Ronald N. and Alan W. Richardson (1996), Origins of Logical Empiricism (Minneapolis: University of Minnesota Press). Gieryn, Thomas F. (1999), Cultural Boundaries of Science: Credibility on the Line (Chicago: The University of Chicago Press). Glennan, Stuart (1996), “Mechanisms and the Nature of Causation,” Erkenntnis 44: 49–71. Golinski, Jan (1992), Science as Public Culture. Chemistry and Enlightenment in Britain, 1760–1820 (Cambridge: Cambridge University Press). Gooding, David (1990), Experiment and the Making of Meaning (Dordrecht: Kluwer Academic Publishers). Gooding, David, Trevor Pinch et al. (eds.) (1989), The Uses of Experiment. Studies in the Natural Sciences (Cambridge: Cambridge University Press). Grasshoff, Gerd, R. Casties et al. (2000), Zur Theorie des Experimentes. Untersuchungen am Beispiel der Entdeckung des Harnstoffzyklus (Bern: Bern Studies for the History and Philosophy of Science). Graßhoff, Gerd and Michael May (1995), “Methodische Analyse wissenschaftlichen Entdeckens,” Kognitionswissenschaft 5: 51–67. Hacking, Ian (1983), Representing and Intervening (Cambridge: Cambridge University Press). Hacking, Ian (1990), The Taming of Chance (Cambridge: Cambridge University Press). Hacking, Ian (1995), Rewriting the Soul. Multiple Personality and the Sciences of Memory (Princeton: Princeton University Press).

xviii

JUTTA SCHICKORE AND FRIEDRICH STEINLE

Hacking, Ian (1999), “Historical Meta-Epistemology”, in W. Carl and L. Daston (eds.), Wahrheit und Geschichte (G¨ottingen: Vandenhoeck & Ruprecht), pp. 53–77. Hankins, Thomas L. and Robert J. Silverman (1995), Instruments and the Imagination (Princeton: Princeton University Press). Hanson, Norwood Russell (1958), Patterns of Discovery (Cambridge: Cambridge University Press). Harwood, William S. (2004), “A New Model for Inquiry. Is the Scientific Method Dead?,” Journal of College Science Teaching 33. Heidelberger, Michael and Friedrich Steinle (eds.) (1998), Experimental Essays—Versuche u¨ ber das Experiment (Baden-Baden: Nomos). Hoyningen-Huene, Paul (1987), “Context of Discovery and Context of Justification,” Studies in History and Philosophy of Science 18: 501–515. Joerges, Bernward and Terry Shinn (eds.) (2001), Instrumentation. Between Science, State and Industry (Dordrecht: Kluwer). Kitcher, Philip (2001), Science, Truth, and Democracy (Oxford and New York: Oxford University Press). Knorr-Cetina, Karin (1981), The Manufacture of Knowledge (Oxford: Pergamon Press). Kordig, Carl (1978), “Discovery and Justification,” Philosophy of Science 45: 110–117. Kr¨uger, Lorenz (1979), History and Philosophy of Science—A Marriage for the Sake of Reason. Abstracts 6: 6th International Congress for Logic, Methodology, and Philosophy of Science (Hannover: Dr. B¨onecke), pp. 108–112. Kuhn, Thomas S. (1962), “The Structure of Scientific Revolutions”, in O. Neurath et al. (eds.), Encyclopedia of Unified Science, vol. I (Chicago: University of Chicago Press). Lakatos, Imre (1970), “Falsification and the Methodology of Scientific Research Programmes”, in I. Lakatos and A. Musgrave (eds.), Criticism and the Growth of Knowledge (London: Cambridge University Press), pp. 91–195. Lakatos, Imre and Alan Musgrave (eds.) (1970), Criticism and the Growth of Knowledge (Cambridge: Cambridge University Press). Langley, P., Herbert A. Simon et al. (1987), Scientific Discovery: Computational Explorations of the Creative Processes (Cambridge, MA: MIT Press). Laudan, Larry (1977), Progress and its Problems (Berkeley: University of California Press). Laudan, Larry (1980), “Why Was the Logic of Discovery Abandoned?”, in T. Nickles (ed.), Scientific Discovery, vol. I (Dordrecht: Reidel), pp. 173–183. Longino, Helen E. (2002), The Fate of Knowledge (Princeton: Princeton University Press). Losee, John (1979), A Historical Introduction to the Philosophy of Science, 2nd edition (Oxford: Oxford University Press). Machamer, Peter, Lindley Darden et al. (2000), “Thinking About Mechanisms,” Philosophy of Science 67: 1–25. Mayo, Deborah G. (2000), “Experimental Practice and an Error Statistical Account of Evidence,” Philosophy of Science. Supplement 67: S193–S207. McLaughlin, Robert (1982), “Invention and Induction. Laudan, Simon and the Logic of Discovery,” Philosophy of Science 49: 198–211. McMullin, Ernan (1990), “The Development of Philosophy of Science 1600–1900”, in R. C. Olby et al. (eds.), Companion to the History of Modern Science (London and New York), pp. 816–837. Morus, Iwan Rhys (1998), Frankenstein’s Children: Electricity, Exhibition, and Experiment in EarlyNineteenth-Century London (Princeton: Princeton University Press). Nersessian, Nancy J. (1995), “Opening the Black Box: Cognitive Science and History of Science,” Osiris: A Research Journal Devoted to the History of Science and its Cultural Influences 10: 194–211. Nickles, Thomas (1980a), “Introductory Essay: Scientific Discovery and the Future of Philosophy of Science”, in T. Nickles (ed.), Scientific Discovery, Logic, and Rationality (Dordrecht: Reidel), pp. 1–59. Nickles, Thomas (ed.) (1980b), Scientific Discovery, Logic, and Rationality (Dordrecht: Reidel). Nickles, Thomas (1985), “Beyond Divorce: Current Status of the Discovery Debate,” Philosophy of Science 52: 177–206.

INTRODUCTION: REVISITING THE CONTEXT DISTINCTION

xix

Radder, Hans (2003), “Toward a More Developed Philosophy of Scientific Experimentation”, in H. Radder (ed.), Philosophy of Scientific Experimentation (Pittsburgh: Pittsburgh University Press), pp. 1–18. Rader, Karen (1999), “Of Mice, Medicine, and Genetics: CC Little’s Creation of the Inbred Laboratory Mouse, 1909–1918,” Studies in History and Philosophy of Biology and Biomedical Sciences 30: 319. Rasmussen, Nicolas (1993), “Facts, Artifacts, and Mesosomes: Practicing Epistemology with the Electron Microscope,” Studies in History and Philosophy of Science 24: 227–265. Reichenbach, Hans (1938), Experience and Prediction. An Analysis of the Foundations and the Structure of Knowledge (Chicago: The University of Chicago Press). Rheinberger, Hans-J¨org (1997), Toward a History of Epistemic Things (Stanford: Stanford University Press). Rudge, David Wyss (2001), “Kettlewell from an Error Statistician’s Point of View,” Perspectives on Science 9: 59–77. Salmon, Wesley C. (1970), “Bayes’s Theorem and the History of Science”, in R. H. Stuewer (ed.), Historical and Philosophical Perspectives of Science (New York et al.: Gordon and Breach), pp. 68–86. Schaffer, Simon (1986), “Scientific Discoveries and the End of Natural Philosophy,” Social Studies of Science 16: 387–420. Schaffner, Kenneth (1980), “Discovery in the Biomedical Sciences: Logic or Irrational Intuition?”, in T. Nickles (ed.), Scientific Discovery: Case Studies, Boston Studies in the Philosophy of Science, vol. 60 (Dordrecht: Reidel), pp. 171–205. Simon, Herbert A and D. Kulkarni (1988), “The Process of Scientific Discovery: The Strategy of Experimentation,” Cognitive Science 12: 139–175. Simon, Herbert A. (1973), “Does Scientific Discovery Have a Logic?,” Philosophy of Science 40: 471– 480. Simon, Herbert A., Patrick W. Langley et al. (1981), “Scientific Discovery as Problem Solving,” Synthese 47: 1–28. Stadler, Friedrich (1997), Studien zum Wiener Kreis (Frankfurt am Main: Suhrkamp). Steinle, Friedrich (2005), Explorative Experimente. Amp`ere, Faraday und die Urspr¨unge der Elektrodynamik. Boethius 50. (Stuttgart: Franz Steiner Verlag). Steinle, Friedrich and Richard M. Burian (2003), “Special Issue: History of Science and Philosophy of Science,” Perspectives on Science 10. Uebel, Thomas E. (ed.) (1991), Rediscovering the Forgotten Vienna Circle. Austrian Studies on Otto Neurath and the Vienna Circle (Dordrecht, Boston and London: Kluwer Academic Publishers). Weber, Marcel (2005), Philosophy of Experimental Biology (Cambridge: Cambridge University Press). Winter, Alison (1998), Mesmerized: Powers of Mind in Victorian Britain (Chicago: The University of Chicago Press). Yeo, Richard (1993), Defining Science. William Whewell, Natural Knowledge, and Public Debate in Early Victorian Britain (Cambridge: Cambridge University Press).

PART I The Contexts of the Distinction

DON HOWARD

LOST WANDERERS IN THE FOREST OF KNOWLEDGE: SOME THOUGHTS ON THE DISCOVERY-JUSTIFICATION DISTINCTION

1. INTRODUCTION: WHAT QUESTIONS SHOULD WE ASK

Neo-positivism is dead. Let that imperfect designation stand for the project that dominated and defined the philosophy of science, especially in its Anglophone form, during the fifty or so years following the end of the Second World War. While its critics were many,1 its death was slow, and some think still to find a pulse.2 But die it did in the cul-de-sac into which it was led by its own faulty compass. The project was this: To provide global, formal explications of central methodological notions—confirmation, explanation, laws—and global, formal answers to global questions about the structure and interpretation of scientific theories, including, most famously, the realism—antirealism debate.3 That the neo-positivist project would lead to a dead end should have been clear, and was clear to some, from very early on. W.V.O. Quine made the point already in 1951 in “Two Dogmas of Empiricism” (Quine 1951) with his critique of the notion of analyticity, and I have always read Nelson Goodman as having tried to make a similar point in his 1953 lecture, “The New Riddle of Induction” (Goodman 1955), with the argument that no merely formal explication of the notion of confirmation can mark the distinction between a law-like generalization such as “All emeralds are green” and its nonsensical cousin, “All emeralds are grue.” Quine and Goodman also pointed out where the error lay, Quine with his explicit endorsement of epistemological naturalism (Quine 1969) and Goodman with his suggestion that the difference between a projectable predicate like green and an unprojectable predicate like grue was to be explained (not explicated) by means of a concept of entrenchment borrowed from anthropological linguistics. The error lay in the philosopher of science’s principled restriction of philosophical attention to the merely formal. Quine and Goodman believed that making sense of empirical science required the tools of empirical science, Quine celebrating such reflexivity (to use the fashionable modern term) as not vicious but virtuous circularity, if, that is, the goal is taken to be not justification, which Hume already taught us was impossible, but, more modestly, the description of how, to borrow a phrase from Quine, sensory input becomes theoretical output. The project whose ultimate demise was thus foretold by Quine and Goodman was born in the 1930s, chiefly through the work of Rudolf Carnap and Hans Reichenbach. Foremost among its central dogmas, foremost at least for the purpose of legitimating a conception of the philosophy of science as a purely formal enterprise, was the distinction between the context of discovery and the context of justification and the 3 J. Schickore and F. Steinle (eds.), Revisiting Discovery and Justification, 3–22.
C 2006 Springer. Printed in the Netherlands.

4

DON HOWARD

associated claim that the philosophy of science confines itself to formal questions within the context of justification, for the perspectives on science thus proscribed as inherently non-philosophical—history, sociology, psychology, biology and kindred studies of the contingent—treat of subjects falling within the realm of what Hume dubbed “matters of fact.” Logic alone survives as the one “science” in a science of science. Many questions cannot be asked and answered by a philosophy of science so constrained. Thus, neo-positivism was as if dumbstruck by questions about the role of science in human affairs, whether they be questions about nuclear weapons, environmental problems, or biotechnology.4 And the relationship between the history of science and the philosophy of science could be theorized in no way other than as an insult to the historian, with the suggestion that history’s only role was to provide rationally reconstructed case studies as vindication for the philosopher’s normative methodological claims. But neo-positivism is dead, and so we now ask questions that were once taboo. Are we, however, asking all of the right questions? The point of the present paper is to suggest that, perhaps, we are not. We might think ourselves like Theseus. Having slain the Minotaur, do we now follow our Ariadne’s thread back out of the cave? Perhaps. But might we not be more like Hansel and Gretel, having found that the crumbs with which we marked the trail into the woods have all been gobbled up? I worry that we are seriously lost. To return from the realm of metaphor, my concern is that unless we understand the proscriptive work really intended by the authors of such neo-positivist dogmas as the DJ distinction, we cannot know what questions we should now be asking. It is good that we now explore the way in which experimental practice, the social structures of science, and the psychology of the researcher play a role in the acceptance of scientific theories. As I have argued elsewhere, however, when, in 1938, Reichenbach beatified the DJ distinction in Experience and Prediction (Reichenbach 1938), his target was not—or not just—epistemological naturalism. His target was, instead, Otto Neurath, Philipp Frank, and the left wing of the Vienna Circle, Neurath especially being then famous for his effort to theorize a positive role for social and political values in theory choice (Howard 2000). The history of that encounter between Reichenbach and Neurath is retold, briefly, in the next section of this paper. For now, I just state the lesson of that history, a lesson itself elaborated later on. The lesson is that, if what was forbidden was the assertion of a positive role for social and political values in theory choice, then the task today is to resume the conversation with Neurath about whether there is and should be such a role for values in the doing of science. 2. THE HISTORICAL BACKGROUND: REICHENBACH AND NEURATH ON VALUES AND THEORY CHOICE

In 1913, Neurath published a lovely little essay in which one finds adumbrated all of the main ingredients of his mature philosophy of science. As well known in its day as it is unknown today, it carries a curious title: “Die Verirrten des Cartesius

LOST WANDERERS IN THE FOREST OF KNOWLEDGE

5

und das Auxiliarmotiv. Zur Psychologie des Entschlusses,” nicely translated as “The Lost Wanderers of Descartes and the Auxiliary Motive (On the Psychology of Decision)” (Neurath 1913). The image of the lost wanderer is taken from a passage in the Discourse on Method, where Descartes contrasts theory and practice by pointing out that, unlike science, practical action requires our making firmly fixed, if perhaps ungrounded and uncertain decisions. In the realm of praxis we are like people lost in a wood. To find a way out, one must simply decide to keep moving in a given direction, though there be no reason for preferring that direction over others. Neurath believes that Descartes is wrong, that science, like practical action, requires ungrounded decisions. That by which we so decide he calls the “auxiliary motive” (note—not “auxiliary reasons”). He mocks the foundationalist illusion of fixed rules of method and fixed criteria of theory choice as “pseudo-rationalism.” Why are there no such foundations? The argument proceeds from an essentially Duhemian view of science (Duhem 1906), wherein theories are interconnected wholes, and theory choice is underdetermined by logic and experience: Whoever wants to create a world-view or a scientific system must operate with doubtful premises. Each attempt to create world-picture by starting from a tabula rasa and making a series of statements which are recognized as definitively true, is necessarily full of trickeries. The phenomena that we encounter are so much interconnected that they cannot be described by a one-dimensional chain of statements. The correctness of each statement is related to that of all the others. It is absolutely impossible to formulate a single statement about the world without making tacit use at the same time of countless others. Also we cannot express any statement without applying all of our preceding concept formation. On the one hand we must state the connection of each statement dealing with the world with all the other statements that deal with it, and on the other hand we must state the connection of each train of thought with all our earlier trains of thought. We can vary the world of concepts present in us, but we cannot discard it. Each attempt to renew it from the bottom up is by its very nature a child of the concepts at hand. (Neurath 1913, p. 3)

Basically this same view of theory choice is reiterated, with occasional refinement and clarification, in Neurath’s writings over the next twenty-five years. On one key point, however, Neurath was much more explicit in later years. It is that prominent among the auxiliary motives are social and political values.5 On Neurath’s view, it is a contingent fact, well supported by historical evidence, that we do choose among empirically equivalent theories on the basis of our estimation of the likelihood of their serving our favored social and political ends, this especially in sciences like economics and sociology. Denial of this fact being symptomatic of pseudo-rationalism, it were better that we be honest with ourselves and others about our so choosing. The cause of scientific objectivity and the cause of human freedom (for Neurath those two causes are one) are better served by open, public debate about the values whereby we choose, for the agents of regressive social and political interests hide their agendas behind the disguise of pseudo-rationalism. It is surely no accident that, unlike Carnap, Reichenbach, and Moritz Schlick, who cut their philosophical teeth on general relativity, Neurath was an economist and sociologist. Briefly imprisoned after the First World War for his role as President of the Central Economic Office in the short-lived Bavarian Soviet Republic, Neurath

6

DON HOWARD

went on to become a prominent figure is the Austrian Social Democratic Party, for many years Director of the Social and Economic Museum in Vienna, an institution with deep roots in the worker education movement. An Austro-Marxist, Neurath’s far-left politics infused all of his philosophical work. His ambition for the Vienna Circle, the Verein Ernst Mach, the International Encyclopedia of Unified Science, and the logical empiricism that was their philosophical doctrine was that they would promote an image of science and scientific philosophy as engines of progressive social change.6 Though not Marxists of any stripe, Reichenbach and Carnap were also socialists, Reichenbach having been a prominent leader of the student socialist movement in Berlin at the end of the First World War. But Reichenbach and Carnap disagreed profoundly with Neurath over the role of social and political values in theory choice. It is not widely enough known that their famous dispute over the role of conventions in science, a dispute that we recall, if at all, in the form it took in the protocol sentence debate of the early 1930s, was, in important measure, an argument over this very question (see Neurath 1932; Carnap 1933; Schlick 1934; see also Zhai 1990). Even as a dispute over the role of conventions, however, the issue is not always clearly enough understood. Here it is in outline. In an effort to craft an empiricism capable of defending the empirical integrity of general relativity against mainly its neo-Kantian critics, Schlick and Reichenbach had in the 1920s elaborated a view of conventions according to which only analytic coordinating definitions are conventional. They held that once the coordinating definitions are fixed by convention, each remaining synthetic proposition in a scientific theory is thereby endowed with a determinate empirical content of such a kind that the proposition’s truth or falsity is unambiguously determinable on the basis of the corresponding experience. Alternative conventions are possible, and simplicity considerations might lead us to prefer one set over another, but the resulting differences in theory are held to be inconsequential, being no different in kind than the difference between English units and metric units. The very fact that two alternative theories are empirically equivalent guarantees that they are just two different ways of saying the same thing, for empirical content is the only content of theory. This is verificationism, the view later attacked by Quine in “Two Dogmas.” The view worked against neo-Kantian critics of general relativity by securing the synthetic, empirical status of claims about the metrical structure of spacetime.7 But the verificationism of Schlick and Reichenbach does other work as well. For if it works, then it also blocks Neurath’s assertion of a role for social and political values in theory choice. If the difference between alternative conventions is no more than the difference between two languages, and if every synthetic proposition has its own, determinate, empirical content, then theory choice is, in principle, univocally determined by considerations of logic and evidence, leaving no room for the operation of values of any kind. Neurath’s theoretical and epistemological holism inclined him to a different view of conventions. Such holism being inhospitable to the analytic synthetic distinction upon which the Schlick—Reichenbach view is based, Neurath tended like Duhem before

LOST WANDERERS IN THE FOREST OF KNOWLEDGE

7

him and Quine after him to view all of the propositions composing a theory as equally conventional or equally empirical in character. Such holism being incompatible with the verificationist semantics of Schlick and Reichenbach, Neurath tended to view the differences between alternative empirically equivalent theories as being often quite significant, especially from the point of view of the differential capacities of alternative social and economic theories to promote specified social ends. It was this argument with Neurath that Reichenbach resumed when, in 1938, in Experience and Prediction, he premiered the distinction between the context of discovery and the context of justification.8 Reichenbach wrote Experience and Prediction in English during his exile from Nazism in Turkey. He did so for the explicit purpose of introducing himself and his philosophical movement to the North American audience he hoped soon to be addressing directly. Neurath is not named as the target of the DJ distinction, and that he was the target was not evident then, and is not evident now, to Anglophone readers ignorant of the German philosophical literature from which Reichenbach was emerging. But it would have been crystal clear to Reichenbach’s former Viennese and Berlin colleagues. The distinction is introduced in section one of Experience and Prediction, “The Three Tasks of Epistemology.” These three tasks are, respectively, the descriptive, the critical, and the advisory task. The descriptive task involves only the rational reconstruction of historical episodes for the purpose of bringing to the fore those crucial logical aspects of the episode that reside in the context of justification. The critical task involves the direct analysis of those logical features of the structure and interpretation of theories. It, too, resides wholly in the context of justification and forms the heart of scientific philosophy. The advisory task is most interesting. One might think that the philosopher of science should offer advice about the ends served by the different choices we make when doing science, as Neurath surely would, but Reichenbach recommends a more modest advisory role, in which the advisory task collapses into the critical task. The philosopher of science should advise not about ends, but only about means for the attainment of such ends: We may therefore reduce the advisory task of epistemology to its critical task by using the following systematic procedure: we renounce making a proposal but instead construe a list of all possible decisions, each one accompanied by its entailed decisions. So we leave the choice to our reader after showing him all factual connections to which he is bound. It is a kind of logical signpost which we erect; for each path we give its direction together with all connected directions and leave the decision as to his route to the wanderer in the forest of knowledge. And perhaps the wanderer will be more thankful for such a signpost than he would be for suggestive advice directing him into a certain path. Within the frame of the modern philosophy of science there is a movement bearing the name of conventionalism; it tries to show that most of the epistemological questions contain no questions of truth-character but are to be settled by arbitrary decisions. This tendency, and above all, in its founder Poincar´e, had historical merits, as it led philosophy to stress the volitional elements of the system of knowledge which had been previously neglected. In its further development, however, the tendency has largely trespassed beyond its proper boundaries by highly exaggerating the part occupied by decisions in knowledge. The relations between different decisions were overlooked, and the task of reducing arbitrariness to a minimum by showing the logical interconnections between the arbitrary decisions was forgotten. The concept of entailed decisions, therefore, may be regarded as a dam against extreme conventionalism; it allows us to separate the arbitrary part of the system of

8

DON HOWARD knowledge from its substantial content, to distinguish the subjective and the objective part of science. The relations between decisions do not depend on our choice but are prescribed by the rules of logic, or by the laws of nature (Reichenbach 1938, pp. 14–15)

As Reichenbach had earlier explained, the concept of entailed decisions was one of the more important discoveries already made under the heading of epistemology’s critical task. The suggestion seems to be that too much attention to the context of discovery leads us to exaggerate the significance of logically and empirically ungrounded choices that lead to a genuine parting of the ways, choices that Reichenbach here terms “volitional bifurcations.” Put on the blinders, focus on the context of justification, note the central role of entailed decisions—choices that are implied by prior choices— and the genuinely volitional aspect of science nearly disappears. Narrowing the focus to the context of justification, declaring unphilosophical all questions in the context of discovery, means, then, our declaring unphilosophical all of those questions about the role of social and political values in theory choice that Neurath addressed under the heading of the auxiliary motive, all of those questions about social and political values that, for Neurath, mattered most.9 Scientific philosophy as Reichenbach characterizes it is but another species of pseudo-rationalism. Neurath died late in 1945 and so did not survive the war to speak to a new Anglophone audience on behalf of a philosophy of science that claimed a positive role for social and political values in theory choice. Philipp Frank was the one friend of Neurath’s view to survive the emigration. Through the 1950s he promoted the Neurath line in many papers and books (see especially Frank 1953 and Frank 1957), but he was doing so from the comparatively ineffective position of a part-time teacher of physics undergraduates at Harvard. It was Reichenbach at UCLA and Carnap at, first, Chicago, and then as the late Reichenbach’s successor at UCLA who were training a new generation of Ph.D.s in the philosophy of science and setting the agenda for the discipline in its new North American home. Reichenbach continued to advocate a politically disengaged philosophy of science in works such as The Rise of Scientific Philosophy (Reichenbach 1951). Carnap had come around partially to Neurath’s view in the protocol sentence debate, but he never wavered in his defense of analyticity, and while his personal political activities, his favorable reception of Kuhn’s Structure of Scientific Revolutions as co-editor of the International Encyclopedia of Unified Science, and his characterization of external questions respecting the choice of a linguistic framework as “pragmatic” (Carnap 1950) might suggest some subtlety in his mature view, his public philosophical example, at least, never even hinted at a sympathy for Neurath’s view of the place of social and political values in theory choice.10 The fact is that when neo-positivism emerged in the 1950s as the “received view” in the newly institutionalized and rapidly expanding discipline of the philosophy of science, no one was left to speak for Neurath. Interlocutors might have been found among the followers of John Dewey’s theory of science, as witness the example of Charles Morris, co-instigator with Neurath and later co-editor with Carnap of the International Encyclopedia of Unified Science. But the old socialist Dewey was dead

LOST WANDERERS IN THE FOREST OF KNOWLEDGE

9

by 1952 and the ranks of his followers were thinning rapidly, Dewey’s conception of a theory of science being deliberately driven out of the academy by thinkers as diverse in their sympathies as Sidney Hook and Ernest Nagel. Science might have won the war in defense of freedom and democracy, but not even those values were to be accorded a place in the postwar project of the philosophy of science. 3. RESUMING THE CONVERSATION I: LETTING NEURATH SPEAK

If the real target of Reichenbach’s introduction of the distinction between the context of discovery and the context of justification was Neurath’s philosophy of science, with its assertion of a role for social and political values in theory choice, then questioning the distinction requires, first and foremost, our resuming the conversation with Neurath. In fact, several scholars have recently sought to do just that, most notably Nancy Cartwright.11 But while much has lately been written about Neurath’s politics, a specifically philosophical engagement with his claims about the place of value in science has, thus far, not been a central theme in the conversation. Cartwright, for example, is drawn more to Neurath’s celebration of theoretical pluralism. I want us to take up the conversation about values in science. The potential benefits to the discipline of our renewing the conversation about values are many. Most importantly it might open up the prospect of the discipline’s once again finding something helpful to say about the place of science in human affairs, this in a philosophical voice, and not just as citizen participants in the political process who merely happen to be philosophers of science. It might also open up the possibility of a rapprochement between the philosophy of science and those historians and sociologists of science who feel themselves estranged from a purely formal philosophy of science mainstream. It might make possible our welcoming our feminist colleagues into the mainstream. And it might make possible our reintegrating the philosophy of science with philosophy more generally, not now primarily on the epistemological ground where once we met when general methodology defined the philosophy of science mainstream, but in the domains of moral philosophy and, to the extent that contextual values are accessible only historically, the history of philosophy.12 Moreover, these benefits can accrue without threat of a loss of the philosophical rigor upon which we philosophers of science have always prided ourselves. If we are to resume the conversation with Neurath about values in science, let us start by giving him the chance to speak first. He did, after all, have a lot to say on this score, much of it quite insightful. I turn then, first, to some central points in the further elaboration of Neurath’s socially and politically engaged philosophy of science. A complete account of Neurath’s philosophy of science is not to be given here,13 but rather my own rather selective list of points important for the subsequent conversation. a. Values and Objectivity Begin with the all-important question about values and objectivity. Is scientific objectivity so obviously threatened by according values a central place? Two points are

10

DON HOWARD

to be made under this heading. First, Neurath’s general epistemological framework is that of Duhemian underdeterminationism. The place of values in science is secured by the fact that, on Neurath’s view, logic and experience underdetermine theory choice. But turn that argument around and it implies that values come into play only within what I like to call the domain of underdetermination. That is, logic and experience are first allowed to do all of the work they can do. Only then do we ask which of several empirically equivalent theories is most conducive to the achievement of our social and political ends. Though values have an essential, unavoidable role to play, science as thus portrayed is as rigorously empirical and logical, hence, in this sense, as objective as it can possibly be. Value considerations are not intended to trump considerations of logic and experience; they are intended to respect them. Second, given the fact that social and political values play a role in theory choice, the cause of objectivity is best served by openness and honesty about that role. This is a point often stressed in our day by feminist philosophers of science.14 But androcentrism is not the only interest served by the pretense of a science that lives above the fray. Neurath thought the interests of capital equally well served by that pretense. Openness about the role of values in science is not always easy to achieve, however, especially for those—typically those in power—whose agendas are thus served by the pose of value neutrality. Think only about the absurd pretense of scientific objectivity among the practitioners of neo-classical economics. Neurath believed that clarity and honesty about the role of values in science is more easily achieved, though of course not necessarily so, by the dispossessed and the downtrodden. Here is how he expresses his version of what we would call standpoint theory: Marxism makes it understandable why the bourgeoisie, conditioned by its class position, becomes ever more unscientific in the field of social theory. To many bourgeois it may seem degrading, and an infringement of the dignity which is conceded to science, if one looks at it from the point of view of the class struggle. The proletariat appreciates science properly only as a means of struggle and propaganda in the service of socialist humanity. Many who came from the bourgeoisie are worried whether the proletariat will have some feeling for science; but what does history teach us? It is precisely the proletariat that is the bearer of science without metaphysics (Neurath 1928, p. 297).

Objective science is, for Neurath, therefore, precisely—if ironically—a value-laden science that is just honest with itself about its value-ladenness. b. Values All the Way Down A politically engaged philosophy of science secures the objectivity of science by restricting the operation of value considerations to the domain of underdetermination. But how large is that domain? A famous and crucial claim of Neurath’s in the protocol sentence debate was that the holism and the consequent underdetermination go all the way down to the protocol sentences, the observation sentences wherein theory meets experience. This was a principal premise in Neurath’s argument for a physicalist protocol language. Since phenomenalist protocols lack the veridicality required for them to perform their intended foundationalist work, that lack of veridicality a consequence

LOST WANDERERS IN THE FOREST OF KNOWLEDGE

11

of the unavoidably propositional character of our protocol sentences, better to adopt a physicalist protocol language, the putative referents of which—medium-sized physical objects—have an advantage over the subjective contents of momentary, private experience, the advantage of being public objects. But we pay a price for the public character of physicalist protocols, for like all discourse about physical objects, physicalist protocols are entangled in the web of belief: Science is ambiguous—and is so on each level. When we have removed the contradictory groups of statements, there still remain several groups of statements with differing protocol statements that are equally applicable; that are without contradictions in themselves but exclude each other. Poincar´e, Duhem and others have adequately shown that even if we have agreed on the protocol statements, there is an unlimited number of equally applicable, possible systems of hypotheses. We have extended this tenet of the uncertainty of systems of hypotheses to all statements, including protocol statements that are alterable in principle. (Neurath 1934, p. 105 [translation corrected]; see also Neurath 1932, pp. 94–95).

Neurath had the courage of his convictions in thus acknowledging what some might see as a fatal implication of his view. What happens to scientific objectivity if the domain of underdetermination, the domain in which value considerations play a proper role, extends to the whole of science? More is to be said later about the values all the way down problem. For now, just a mitigating observation. In principle all physicalist protocols in all scientific fields are underdetermined. In practice, however, this phenomenon will be far more pronounced in the social and economic sciences that were Neurath’s chief concern. But this is also the realm in which most of us never doubted the theory-ladenness of observation. More so than in physics, social data need an interpretation before they become evidence.15 From this perspective, what first appeared to be a threatening implication of Neurath’s view turns out to be little more than a platitude. c. Praxis and Theory Choice A recurring theme in Neurath’s philosophical works, especially when he was writing, as he did regularly, for an audience of his fellow socialists, is that, at least in the social and economic sphere, theory choice is driven by the need for practical action, a need often so compelling and often felt in similar enough ways by many members of relevant communities as to obscure the fact that a genuine array of empirically equivalent theoretical alternatives is always available. “Action” here means both direct political action and the more mundane activities of everyday life, and typically, for Neurath, “action” means not the private actions of atomic individuals but collective social action. I quote a longish passage to illustrate the feel of Neurath’s thinking on this point: This is how matters stand in every “layer” of scientific work, not only in the narrower sphere of systems of hypotheses, as Poincar´e and Duhem have pointed out with such intensity. But these initiatives in multiplicity are constricted by life. A whole human lifetime is hardly long enough to immerse oneself in even a single view and to give full thought to its consequences. And how soon one senses the weakening effect of isolation. Thus one deserts the lonely, though perhaps auspicious, notions of an outsider to join in the work in a way of thought that enjoys more support and has therefore better

12

DON HOWARD chances of greater scientific achievement. In such ways it happens that not even too many possibilities are treated by several groups at the same time: through adaptation and selection a kind of assimilation of whole generations takes place—not to speak of the cases in which certain trains of thought are anathema, persecuted and suppressed. This insight that a logically tenable multiplicity is reduced by life has little hope of response because it contradicts the usual view of a connection between achievement and “success.” The representatives of a victorious doctrine are too much inclined to believe that their victory could be justified as it were by closer logical investigation. Many see the course of the history of science like that. (Neurath 1935, p. 117)

Neurath adds: “If, in spite of these comments on multiplicity and uncertainty, one sets unswervingly to the work that is seen as a common one, one can do so only because one knows how much the historical situation reduces the manifoldness via facti” (Neurath 1935, p. 119). When the action in question is direct political action, there might be more conscious awareness of choices being made: The Marxist, as strict scientist, must admit that the course of history allows of various interpretations. But successful collaboration is possible only when those who act fix on one possibility, whether by agreement or propaganda. This choice is itself a matter of action and resolution, but that does not mean that such action has no scientific basis. (Neurath 1928, p. 293)

It is not as though Neurath was the first or last philosopher of science to think about the relation between theory and practice, but what is more commonly intended under that heading are questions addressed via a distinction between pure and applied science, where it is assumed that we first choose our theories on more narrowly epistemic grounds and then act by applying those theories. That is not what Neurath intends. On Neurath’s view, practical considerations intrude at the start as well as the finish. Far more can and should be said about Neurath’s philosophy of science. Would that time permitted, for example, our thinking about his promotion—this with almost missionary zeal—of a unity of science not via theoretical unification but via the adoption of a universal physicalist protocol language. Neurath intended this, too, to do political work, for the physicalist protocol language was recommended in part for its being inhospitable to the obscurantist rhetoric favored by apologists for capital and other socially regressive interests. But it was Neurath’s employment of the Duhemian view of empirically underdetermined theory choice for the purpose of securing a place for social and political values in science that Reichenbach aimed to block with the DJ distinction. Let us keep the focus there for now, and let us now venture a response to Neurath. 4. A DIGRESSION: THE EPISTEMIC—NON-EPISTEMIC DISTINCTION

Before we can begin our response to Neurath, however, we must do some philosophical housecleaning. One cobweb, in particular, must be removed, namely, the distinction between epistemic and non-epistemic values, a distinction introduced in reaction to Kuhn. Replying to various early critical responses to The Structure of Scientific Revolutions, Kuhn sought to defend his claims about incommensurability while

LOST WANDERERS IN THE FOREST OF KNOWLEDGE

13

denying the relativist implications that some readers found in his story about paradigm conflict. According to Kuhn, all parties to a paradigm dispute typically share a common ground in their commitment to all manner of values, including simplicity, explanatory power, fertility, and empirical accuracy. Where they disagree is in their weighing of these shared values in assessing the merits of competing paradigms (Kuhn 1969, pp. 184–186; Kuhn 1973). To this perhaps not wholly convincing argument Ernan McMullin responded, not unsympathetically, by recommending a distinction between what he termed epistemic and non-epistemic values. The former, the truthinducing values, were to be accorded a respectable place in science, the latter not, for a science that chooses according to the former is less open to the charge of relativism (McMullin 1982). A theory’s being conducive to one’s favored social and political ends would be regarded as a non-epistemic value, and choosing among theories on such grounds invites the charge of relativism. That empirical adequacy is an epistemic virtue few would doubt. Things get a bit murkier when we turn to explanatory power, fertility, and simplicity. Consider the case of Einstein on simplicity. One would be hard pressed to find a stronger advocate of simplicity as a criterion of theory choice, especially in frontier physics far removed from the realm of direct empirical testability. But near the end of his life, after decades of reflecting on the question, Einstein gave up the attempt to define this elusive virtue, concluding that while scientists tended to agree in their judgments of simplicity, the comparison of theories with respect to simplicity amounted to “a reciprocal weighing of incommensurable qualities” (Einstein 1946, pp. 21, 23; see also Howard 1998). Fertility is no more likely to be captured in a formula, and for all that our intuitions might incline us to expect otherwise, even explanatory power proves hard to characterize in a sufficiently general manner.16 Does the frustration we feel in seeking global definitions of simplicity, fecundity, and explanatory power tell us something about those virtues, or does it tell us something about the underlying distinction between epistemic and non-epistemic virtues? I suspect the latter. Who would doubt the prima facie reasonableness of choosing among empirically equivalent theories on the basis of putatively epistemic virtues? But why do we think that in so choosing we choose the theories more likely to be true? It depends partly on how we think about truth. If we think that there is only one truth about nature and that in choosing among empirically equivalent theories we take a step, fallible though it might be, toward the establishment of that one truth, then the appeal and the cogency of the epistemic—non-epistemic distinction is evident. But at this point I find Quine helpful. This most thoroughgoing Duhemian was famous for arguing that, if theory choice is underdetermined by considerations of logic and evidence, then there can be no global, extra-theoretical talk of truth or the approach to the ultimate truth, for one can put through a Tarski truth definition only theory by theory: “Truth is immanent, and there is no higher. We must speak from within a theory, albeit any of various” (Quine 1981, pp. 21–22; see also Quine 1960, chapter 1; Quine 1969). If Quine is right, then the ground falls out from under the epistemic—non-epistemic distinction, for the distinction requires access to an extratheoretical semantic perspective whose existence Quine denies. The reasonableness of preferring simple to

14

DON HOWARD

complex theories remains, but so too does the reasonableness of choosing theories that will make the world a better place. 5. RESUMING THE CONVERSATION II: DOING NEURATH THE COURTESY OF A REPLY

Now we are ready to begin replying to Neurath about the place of social and political values in science. Where to begin? I want to begin with what I think is the most serious challenge posed to us by Neurath, namely, the argument that underdetermination and hence a role for value considerations, goes all the way down to the level of the protocol sentences. Note that while Quine has not expressed interest in promoting a role for social and political values in theory choice, he agrees with Neurath that even our observation sentences are entangled in the web of belief (see Quine 1951, p. 43; Quine 1960, pp. 42–45). Why do we find the claim that it is values all the way down so worrisome? It is surely worrisome if one’s ambitions for epistemology are of the foundationalist, justificatory sort that Neurath and Quine disavow. If we can choose from among our protocols only those flattering to our political agendas, then protocols cannot exert upon theory choice the kind of univocal empirical control for which foundationalist empiricists like Schlick hoped. But if the aim is simply to describe how science is done, why be distressed by the claim that there are values all the way down? If that’s the way the world is, then that’s the way the world is. Wishing it were otherwise will not make it so. The real worry occasioned by the claim that it is values all the way down is that radical, anything-goes relativism then threatens. Such worries are, however, misplaced, being partly consequent upon a frequent misunderstanding about the underdetermination thesis. Consider the latter in its most extreme, Quinean form: Theory choice in all domains is empirically underdetermined, this underdetermination persisting even if one were in possession of the infinity of all possible evidence. It follows, says Quine, that any favored hypothesis or, if Neurath is right, any cherished protocol can be saved as long as one is prepared to make sufficiently radical alterations elsewhere in the web of belief. But from the assertion that any one proposition can be, thus, insulated from refutation, it does not follow that anything goes, and this precisely because of the theoretical and epistemological holism that is the flip side of underdetermination. The web of belief is a deeply interconnected whole. To save a cherished hypothesis or protocol, one must make changes elsewhere in the web. The freedom of choice is not a freedom simply to deny the manifest evidence of the senses. One has to interpret, one has to tell a coherent story, and one has to tell a story that works. Yet another reason why many see the threat of radical relativism in the claim that it’s values all the way down is that we tend, wrongly, to model scientific decision making as if it took place in a social and historical vacuum. In fact, such choices are made by communities of inquirers whose members typically share, in large measure, a biology, a history, a language, an education, a paradigm, perhaps, and much else besides, the effect of which is to incline them to similar choices. Do we wonder,

LOST WANDERERS IN THE FOREST OF KNOWLEDGE

15

with Einstein, how it can be that the experts agree in assessments of simplicity even though simplicity proves hard to define? No deep mystery here. A virtual necessary condition on the experts’ being recognized as experts is their being similarly socialized into the communities’ shared ways of regarding nature. The surprise would be if they disagreed in their assessments of simplicity. And the same holds, though perhaps to a lesser degree, when the values are not aesthetic but social and political. There are no a priori principles guaranteeing the possibility of consensus, just contingent, empirical facts about communities of inquirers making more or less likely the possibility of consensus. Be the ground for its possibility a necessary or a contingent one, consensus is still consensus.17 Mention of socialization brings us to another important point well understood by Neurath, which is that theory choice, including the choices that constitute the empirical basis of science, is a collective enterprise. Theory choice does not take place in a vacuum and it is not the work of disembodied, isolated, individual knowers; it is the work of social groups. But if we regard even the experience upon which science rests as a social achievement,18 yet another perspective avails itself from which to regard the question of radical relativism and the claim of values all the way down. For it shows us that, even as philosophers, we can and should be asking questions about the social structures through which such experience is achieved. An older literature frequently dramatized the threat of radical, anything-goes relativism by the example of the Lysenko affair. In this rightly notorious case, Soviet work in evolution and genetics was set back by over a generation thanks to obvious perversions of scientific practice in the name of a political agenda (Joravsky 1970). But the problem here was not that Soviet geneticists had a faulty philosophy of science, and the threat was not to be met by the fantasy of a value-neutral science, though that was the standard prescription. The problem was that the social institutions of science were not working as they should. The passing whims of Joe Stalin were no good substitute for peer review. There once was a time when Robert Merton’s work on the social norms of science was deemed a subject matter fit for the philosophy of science journals.19 But that was a long time ago. Today we might fault Merton’s naive philosophical understanding of science, his overly simple assumptions about truth and objectivity, and his reluctance to adopt a sociological perspective on the content of science, as opposed to its institutional structures, but that should not blind us to the value of his example. For the recollection of the older philosophical literature on Merton’s norms should remind us that philosophers qua philosophers can and should concern themselves with questions about the social structure of science. Our doing so would pay dividends. I would be delighted, for example, if one could turn to the philosophy of science journals for tough-minded discussions of proprietary research. If philosophers of science were once so exercised over the Lysenko affair, why are they not even more exercised by this new and distressing trend, where research is done under contract to private corporations with the stipulation that the corporation, not the scientist, owns the results of the research and can block their publication in the standard scientific venues. Why should we trust Merck Pharmaceuticals any more than we trust Joe Stalin?20

16

DON HOWARD

If philosophers qua philosophers can take the naturalistic turn by way of asking questions about the social structures of science, then yet another perspective opens up on the question of objectivity, a perspective famously associated with Helen Longino (Longino 1990; Longino 2002). Many worry that viewing science as a social enterprise entails, inevitably, some compromise with objectivity, for it opens the door to interests other than an interest in the truth. Longino argues, with deliberate irony, that, far from subverting the cause of objectivity, the socializing of our view of science promotes objectivity (or at least intersubjectivity) by making public debate about those interests an integral part of science. Neurath’s comments on pseudo-rationalism are directed toward a similar end. Bad science is done when we lie to ourselves about the place of social interest in science. Good science, objective science, is science that is honest about social interest, about the auxiliary motive, for that is the way to reveal those interests for public scrutiny and debate. In addition to the claim that it is values all the way down, Neurath, as we heard, insisted that the need to act played a major role in resolving underdetermination. If act we must, then we cannot pause like Buridan’s ass before empirically equivalent piles of straw. If act we must, then often our choices will have been made for no other reason, if a “reason” it be, than that some choice must be made. And for Neurath that’s a good thing. More commonly, however, we choose with a practical aim in view. As mentioned above, this is not the way mainstream philosophers of science have usually theorized the relation between theory and practice. What is usual is to distinguish pure science from applied science and then argue that the epistemic merits of the former are unaffected by the latter. Neurath’s way of conceiving the relation makes practice an essential part of the epistemological story. What are the consequences of such a reorientation? One consequence is that this reorientation of the question about theory and practice gives us a more hospitable framework in which to pose questions such as those that interest contemporary philosophers of experiment. Many students react at first with puzzlement to the claim that there is significant cognitive content in experimental practice. How can that be? Experiments are things that I do with my hands. But any good pianist understands what it means to say that there is knowledge in one’s fingertips, and anyone who learned to ride a bicycle at age five should understand what it means to say that one has knowledge in the seat of one’s pants. The problem with the pure—applied distinction is that it makes doing discontinuous with knowing, or rather “knowing how” discontinuous with “knowing that.” On Neurath’s view, all knowing is just a form of doing. Making a choice between empirically equivalent theories in the course of acting is how knowledge is made, and this at every level, right down to the making of protocol statements.21 Let me forestall misunderstanding by noting that such making is not what is intended when the phrase “social construction” is spoken with a Scottish accent. What is made on Neurath’s view is knowledge as genuine as knowledge can be, the making being constrained by all manner of factors other than just social interest, factors such as evidence. As that other Neurathian, Quine, would remind us, we can put through a Tarski truth definition, albeit not a single one for all theories.

LOST WANDERERS IN THE FOREST OF KNOWLEDGE

17

Another consequence of our reorienting our thinking about theory and practice along Neurath’s lines is a moral one. The pure—applied distinction makes possible a most annoying moral dodge: As a scientist I am responsible only to the truth and not responsible for the consequences of applying the theories I discover and prove. But if the application is a part of the proof, then responsibility to the truth entails responsibility for the applications. I think that we used to talk about this under the name “pragmatism.” A final consequence that I would mention here of our thus reorienting our thinking about theory and practice is its suggesting that we are overlooking a lot if we think that the only epistemological questions worth asking are questions about the acceptance of theories. If theory choice is so thoroughly practical in character, then a finer vocabulary of epistemic attitudes is needed to do justice to scientific practice. Merely acting on a theory need entail nothing about one’s having accepted the theory as true or as possessing a high degree of verisimilitude. As noted above, sometimes I choose just because I have to. Sometimes I choose because my choice is permitted given the relevant constraints. Sometimes I choose because my choice flatters my prejudices. Sometimes I choose because my choice represents the way I wish things were even if I have no clue that they are so. And sometimes I choose because my choice seems the most prudent basis for action after weighing such evidence as might be at hand and the risks of acting on the basis of other choices. We are not helped by being told that attitudes other than grounded acceptance all fall below the divided line, in the realm of illusion and mere opinion.22

6. CONCLUSION

It is no accident that, as we have neared the end of this paper, the density of references to Dewey and to feminist philosophy of science has steadily grown. Deweyan pragmatism and feminist philosophy of science are just about the only two philosophical projects of the twentieth century capable of carrying on a respectful conversation with Neurath. It is also no accident that neither occupies a place in the philosophical mainstream. Continuing the conversation with Neurath would require our bringing in those voices far more than was possible here. Important questions would be raised. Thus, an interesting difference between Neurath and Dewey concerned the question of whether science has a role to play in the choice of ends, Dewey saying yes, Neurath no. Save that question for another occasion.

NOTES 1. A list of important early critics from within the philosophy of science mainstream would have to include Stephen Toulmin (Toulmin 1953, 1961, 1972), Norwood Russell Hanson (Hanson 1958), Paul Feyerabend (Feyerabend 1975, 1978), and Thomas Kuhn (Kuhn 1962). It can be argued that the effect of later critiques by, for example, social constructivists and feminist philosophers of science, has been felt more strongly outside of the philosophy of science mainstream than within, for better or worse. For more on why that has been the case, see Howard 2003.

18

DON HOWARD

2. Those with the best stethoscopes work in the domain of inductive logic and probability theory. Perhaps the most interesting case, however, is that of Philip Kitcher. His recent work seeks an opening for questions about the way science lives in a social and political context, but his anxiety about the threat of relativism draws him to assumptions about truth and objectivity so strong as to make problematic any systematic integration of value considerations in the doing of science (see, for example, Kitcher 1993; Kitcher 1998a; Kitcher 1998b; Kitcher 2002; Kitcher and Cartwright 1996). 3. Here I rely heavily on Arthur Fine’s way of characterizing the problem (Fine 1984a; Fine 1984b). 4. Philip Kitcher is, of course, the notable exception among mainstream philosophers of science (see Kitcher 1985; Kitcher 1996). Still more exceptional is Kristin Shrader-Frechette, whose work as a philosopher of science takes her deep into the heart of policy debates (see Shrader-Frechette 1985; Shrader-Frechette 1991; Shrader-Frechette 1993). 5. The idiom of “values” was not that in which Neurath, himself, expressed the point, explicit talk of “value” carrying the wrong connotations in an early twentieth century Germanophone environment where the Werturteilsstreit was a recent memory (Albert and Topitsch 1971; Ciaffa 1998). But it was the idiom employed in the Anglophone literature, especially in North America, whether in Reichenbach’s public disagreements with Dewey over values in science or in Frank’s later promotion of Neurath’s position. For more on the North American setting, see Howard 2003. 6. There is now an abundance of good secondary literature on Neurath. See, for example, Cartwright et al. 1996; Nemeth 1981; Nemeth and Stadler 1996; Reisch 1995; Stadler 1997; Uebel 1991; Uebel 1992. 7. Classic statements of this view of the role of convention in science are Reichenbach 1928 and Schlick 1935. For a discussion of the development of this view and the background of debates over the empirical integrity of general relativity, see Howard 1994. 8. The idea behind the DJ distinction was certainly not original with Reichenbach, it having long been central to debates about psychologism. Thus Popper deploys much the same idea in his critique of induction in his Logik der Forschung (Popper 1934). For more on the history of late-nineteenth and early-twentieth century debates about psychologism see Coffa 1991; Kusch 1995; and Peckhaus, this volume. What is original with Reichenbach and influential in the further development of logical empiricism is the specific formulation of the distinction and the use to which it is put. 9. Promoting a politically disengaged philosophy of science was more than prudent in those parts of Europe under fascist control in the 1930s. It might have been prudent and was surely expedient in North America in the 1930s, 1940s, and 1950s (see Howard 2003 and Reisch 2005). 10. See, however, the not entirely unsympathetic discussion of his disagreement with Neurath on this point in his “Intellectual Autobiography” (Carnap 1963, pp. 22–23). For more on Carnap’s reaction to Kuhn’s The Structure of Scientific Revolutions when it was submitted for publication as part of the Encyclopedia, see Reisch 1991. 11. See, especially, Cartwright 1999; see also the authors cited above in note 6. 12. Major chapters of the history of philosophy are, after all, really about the history of science and the history of the philosophy of science. This is especially true, as Gerd Buchdahl (Buchdahl 1969) argued, in the modern period. Sch¨afer (this volume) discusses Duhem’s arguments for integrating history and philosophy of science. 13. Those wishing a more complete rehearsal of Neuarth’s views should consult first Uebel 1992 and Cartwright et al. 1996. 14. See, for example, Longino 1990; Longino 2002. 15. But even in the physical sciences there can be a significant interpretive moment in the selection of one’s protocol sentences. See, for example, Sch¨afer’s discussion (this volume) of Duhem’s views on the “moral” element in theory choice associated with the assessment of the reliability of the sources of one’s evidence, his example being Kepler on the 8 of arc accuracy in Tycho’s data. 16. And recall from the discussion of Goodman’s “New Riddle of Induction” that even the notion of empirical adequacy might prove hard to pin down. For more on recent skepticism about global, unitary characterizations of methodological norms, see Stump and Galison 1996.

LOST WANDERERS IN THE FOREST OF KNOWLEDGE

19

17. Sch¨afer (this volume) discusses Duhem’s view of the manner in which tradition constrains theory choice and, thereby, makes more likely a measure of continuity in the historical development of science. 18. Dewey, a famous critic of the spectator theory of knowledge, stressed the social nature of experience (Dewey 1929), as do a growing number of contemporary philosophers of science (see, for example, Longino 1990; 2002; Solomon 2001; and Kusch 2002). 19. The classic sources are Merton 1938; Merton 1942. Note that the former was published in Philosophy of Science. For more on the place of Mertonian sociology of science in pre-1950s North American philosophy of science, see Howard 2003. See also Wang 1999; Douglas 2004. 20. Kuhn was interested in social structures, but he tended not to be interested in using sociological tools in what one might term a knowledge-critical manner. He taught us to think of a paradigm as the shared commitments of the members of a scientific community, but he did not bother to ask questions about what kinds of community structures were conducive to good science, once, that is, one gets beyond claims about socialization and normal science as a precondition for the doing of science. 21. Here is another place where there are interesting and provocative parallels in the work of Dewey; see again Dewey 1929. See also Nickles, this volume. 22. The need for a more varied vocabulary of epistemic attitudes is a theme in Bas on van Fraassen’s The Empirical Stance (van Fraassen 2003).

REFERENCES Albert, Hans and Topitsch, Ernst (eds.) (1971), Werturteilsstreit. Darmstadt: Wissenschaftliche Buchgesellschaft. Buchdahl, Gerd (1969), Metaphysics and the Philosophy of Science; The Classical Origins: Descartes to Kant. Cambridge, MA: MIT Press. Carnap, Rudolf (1933), “Ueber Protokolls¨atze.” Erkenntnis 3: 215–228. Carnap, Rudolf (1950), “Empiricism, Semantics, and Ontology.” Revue internationale de philosophie 4(11): 20–40. Carnap, Rudolf (1963), “Intellectual Autobiography.” In The Philosophy of Rudolf Carnap. Paul Arthur Schilpp (ed.) The Library of Living Philosophers, Vol. 11. La Salle, IL: Open Court, pp. 3–84. Cartwright, Nancy (1999), The Dappled World: A Study of the Boundaries of Science. Cambridge: Cambridge University Press. Cartwright, Nancy et al. (1996), Otto Neurath: Philosophy between Science and Politics. Ideas in Context, no. 38. Quentin Skinner et al. (eds.) Cambridge: Cambridge University Press. Ciaffa, Jay A. (1998), Max Weber and the Problems of Value-free Social Science: A Critical Examination of the Werturteilsstreit. Lewisburg, NJ: Bucknell University Press. Coffa, J. Alberto (1991), The Semantic Tradition from Kant to Carnap: To the Vienna Station. Linda Wessels (ed.) Cambridge: Cambridge University Press. Douglas, Heather (2004), “Robert Merton and the Ethos of Science.” Paper presented at the Fifth International Conference of HOPOS, the International Society for the History of Philosophy of Science, San Francisco, pp. 24–27 June 2004. Duhem, Pierre (1906), La Th´eorie physique. Son objet et sa structure. Paris: Chevalier & Rivi`ere. Einstein, Albert (1946), “Autobiographical Notes.” In Schilpp 1949, pp. 1–94. Quotations are taken from the corrected English translation in: Autobiographical Notes: A Centennial Edition. Paul Arthur Schilpp, trans. and ed. La Salle, Illinois: Open Court, 1979. Feyerabend, Paul (1975), Against Method: Outline of an Anarchistic Theory of Knowledge. London: NLB. Feyerabend, Paul (1978), Science in a Free Society. London: NLB. Fine, Arthur (1984a), “The Natural Ontological Attitude.” In Jarret Leplin (ed.) Scientific Realism. Berkeley: University of California Press, pp. 83–107. Reprinted in Fine 1986, pp. 112–135.

20

DON HOWARD

Fine, Arthur (1984b), “And Not Anti-Realism Either.” Noˆus 18:51–65. Reprinted in Fine 1986, pp. 136–150. Fine, Arthur (1986), The Shaky Game: Einstein, Realism, and the Quantum Theory. Chicago: University of Chicago Press. Frank, Philipp (1953), “The Variety of Reasons for the Acceptance of Scientific Theories.” In Frank (ed.) 1956, pp. 13–26. Frank, Philipp (1957), Philosophy of Science: The Link Between Science and Philosophy. Englewood Cliffs, NJ: Prentice-Hall. Goodman, Nelson (1955), “The New Riddle of Induction.” In Fact, Fiction, and Forecast. Cambridge, MA: Harvard University Press, pp. 59–83. First delivered as a lecture at the University of London, May 26, 1953. Hanson, Norwood Russell (1958), Patterns of Discovery: An Inquiry into the Conceptual Foundations of Science. Cambridge: Cambridge University Press. Howard, Don (1994), “Einstein, Kant, and the Origins of Logical Empiricism.” In Language, Logic, and the Structure of Scientific Theories: The Carnap-Reichenbach Centennial. Wesley Salmon and Gereon Wolters (eds.) Pittsburgh: University of Pittsburgh Press; Konstanz: Universit¨atsverlag, pp. 45–105. Howard, Don (1998), “Astride the Divided Line: Platonism, Empiricism, and Einstein’s Epistemological Opportunism.” In Idealization in Contemporary Physics. Niall Shanks (ed.) Poznan Studies in the Philosophy of the Sciences and the Humanities, Vol. 63. Amsterdam and Atlanta: Rodopi, pp. 143– 163. Howard, Don (2000), “Vertrieben und Verirrt: Politics and the Philosophy of Science in Exile.” Paper delivered at the Third International Conference on the History of the Philosophy of Science, Vienna, Austria, 6–9 July 2000. Howard, Don (2003), “Two Left Turns Make a Right: On the Curious Political Career of North American Philosophy of Science at Mid-century.” In Logical Empiricism in North America. Alan Richardson and Gary Hardcastle (eds.) Minneapolis: University of Minnesota Press, pp. 25–93. Joravsky, David (1970), The Lysenko Affair. Cambridge, MA: Harvard University Press. Kitcher, Philip (1985), Vaulting Ambition: Sociobiology and the Quest for Human Nature. Cambridge, MA: MIT Press. Kitcher, Philip (1993), The Advancement of Science: Science without Legend, Objectivity without Illusions. New York: Oxford University Press. Kitcher, Philip (1996), The Lives to Come: Tthe Genetic Revolution and Human Possibilities. New York: Simon & Schuster. Kitcher, Philip (1998a), “Reviving the Sociology of Science.” PSA 98. Proceedings of the 1998 Biennial Meeting of the Philosophy of Science Association. Part II, Symposia Papers. Don Howard (ed.) Philosophy of Science S67: pp. S33–S44. Kitcher, Philip (1998b), “A Plea for Science Studies.” In Koertge 1998, pp. 32–56. Kitcher, Philip (2002), Science, Truth, and Democracy. New York: Oxford University Press. Kitcher, Philip and Cartwright, Nancy (1996), “Science and Ethics: Reclaiming Some Neglected Questions.” Perspectives on Science: Historical, Philosophical, Social 4: 145–153. Koertge, Noretta (ed.) (1998), A House Built on Sand: Exposing Postmodernist Myths about Science. New York: Oxford University Press. Kuhn, Thomas S. (1962), The Structure of Scientific Revolutions. Chicago: University of Chicago Press. Kuhn, Thomas S. (1969), “Postscript–1969.” In The Structure of Scientific Revolutions, 2nd ed. Chicago: University of Chicago Press, pp. 174–210. Kuhn, Thomas S. (1973), “Objectivity, Value Judgment, and Theory Choice.” Matchette Lecture delivered at Furman University, 30 November 1973. First published in: The Essential Tension: Selected Studies in Scientific Tradition and Change. Chicago: University of Chicago Press, 1977, pp. 320–339. Kusch, Martin (1995), Psychologism: A Case Study in the Sociology of Philosophical Knowledge. London and New York: Routledge. Kusch, Martin (2002), Knowledge by Agreement: The Programme of Communitarian Epistemology. Oxford: Clarendon Press; New York: Oxford University Press.

LOST WANDERERS IN THE FOREST OF KNOWLEDGE

21

Longino, Helen (1990), Science as Social Knowledge: Values and Objectivity in Scientific Inquiry. Princeton, NJ: Princeton University Press. Longino, Helen (2002), The Fate of Knowledge. Princeton, NJ: Princeton University Press. McMullin, Ernan (1982), “Values in Science.” In PSA 1982, Vol. 2. P. D. Asquith and T. Nickles (eds.) East Lansing, Michigan: Philosophy of Science Association, pp. 3–28. Merton, Robert K. (1938), “Science and the Social Order.” Philosophy of Science 5: 321–337. Merton, Robert K. (1942), “Science and Technology in a Democratic Social Order.” Journal of Legal and Political Sociology 1: 115–126. Nemeth, Elizabeth (1981), Otto Neurath und der Wiener Kreis. Revolution¨are Wissenschaftlichkeit als Anspruch. Frankfurt and New York: Campus. Nemeth, Elizabeth and Stadler, Friedrich (eds.) (1996), Encylopedia and Utopia: the Life and Work of Otto Neurath (1882–1945), Vienna Circle Institute Yearbook, no. 4. Dordrecht, Boston, and London: Kluwer. Neurath, Otto (1913), “Die Verirrten des Cartesius und das Auxiliarmotiv. Zur Psychologie des Entschlusses.” Jahrbuch der Philosophischen Gesellschaft an der Universit¨at Wien. Leipzig: Johann Ambrosius Barth. Page numbers and quotations from the English translation: “The Lost Wanderers of Descartes and the Auxiliary Motive (On the Psychology of Decision),” In Otto Neurath. Philosophical Papers, 1913–1946. Robert S. Cohen and Marie Neurath (ed.) and trans. Vienna Circle Collection, Vol. 16. Henk L. Mulder, Robert S. Cohen, and Brian McGuinness (eds.) Dordrecht, Boston, and Lancaster: D. Reidel, 1983, pp. 1–12. Neurath, Otto (1928), Lebensgestaltung und Klassenkampf. Berlin. E. Laub. Page numbers and translations from the excerpt translated as “Personal Life and Class Struggle” in Neurath 1973, pp. 249–298. Neurath, Otto (1932), “Protokolls¨atze.” Erkenntnis 3:204–214. Page numbers and translations from the excerpt translated as “Protocol Sentences” in Neurath 1983, pp. 91–99. Neurath, Otto (1934), “Radikaler Physikalismus und ,wirkliche Welt’.” Erkenntnis 4: 346–362. Page numbers and translations from the excerpt translated as “Radical Physicalism and the ‘Real World”’ in Neurath 1983, pp. 100–114. Neurath, Otto (1935), “Einheit der Wissenschaft als Aufgabe.” Erkenntnis 5:16–22. Page numbers and translations from the excerpt translated as “The Unity of Science as a Task” in Neurath 1983, pp. 115–120. Neurath, Otto (1973), Empiricism and Sociology. Marie Neurath and Robert S. Cohen (eds.) Vienna Circle Collection, Vol. 1. Dordrecht and Boston: D. Reidel. Neurath, Otto (1983), Philosophical Papers 1913–1946. Robert S. Cohen and Marie Neurath (eds.) and trans. Vienna Circle Collection, Vol. 16. Dordrecht, Boston, and Lancaster: D. Reidel. Popper, Karl Raimund. (1934), Logik der Forschung. Zur Erkenntnistheorie der modernen Naturwissenschaft. Schriften zur wissenschaftlichen Weltauffassung, Philipp Frank and Moritz Schlick (eds.) Vienna: Julius Spring. Quine, W. V. O. (1951), “Two Dogmas of Empiricism’.” Philosophical Review 60:20–43. Reprinted in: From a Logical Point of View. Cambridge, MA: Harvard University Press, 1953, pp. 20–46. Quine, W. V. O. (1960), Word & Object. Cambridge: MIT Press. Quine, W. V. O. (1969), “Epistemology Naturalized.” In Ontological Relativity and Other Essays. New York and London: Columbia University Press, pp. 69–90. Quine, W. V. O. (1981), “Things and Their Place in Theories.” In Theories and Things. Cambridge, Massachusetts and London: Harvard University Press, pp. 1–23. Reichenbach, Hans (1928), Philosophie der Raum-Zeit-Lehre. Berlin: Julius Springer. Reichenbach, Hans (1938), Experience and Prediction: An Analysis of the Foundations and the Structure of Knowledge. Chicago: University of Chicago Press. Reichenbach, Hans (1951), The Rise of Scientific Philosophy. Berkeley and Los Angeles: University of California Press. Reisch, George (1991), “Did Kuhn Kill Logical Empiricism?” Philosophy of Science 58: 264–277. Reisch, George (1995), “A History of the International Encyclopedia of Unified Science.” Ph.D. Dissertation. University of Chicago.

22

DON HOWARD

Reisch, George (2005), How the Cold War Transformed Philosophy of Science: To the Icy Slopes of Logic. Cambridge: Cambridge University Press. Schlick, Moritz (1934), “Ueber das Fundament der Erkenntnis.” Erkenntnis 4: 79–99. Schlick, Moritz (1935), “Sind die Naturgesetze Konventionen?” In Actes du Congr`es International de Philosophie Scientifique, Paris 1935. Vol. 4, Induction et Probabilit´e. Actualit´es Scientifique et Industrielles, no. 391. Paris: Hermann, 1936, pp. 8–17. Shrader-Frechette, Kristin (1985), Risk Analysis and Scientific Method: Methodological and Ethical Problems with Evaluating Societal Hazards. Dordrecht and Boston: D. Reidel. Shrader-Frechette, Kristin (1991), Risk and Rationality: Philosophical Foundations for Populist Reforms. Berkeley: University of California Press. Shrader-Frechette, Kristin (1993), Burying Uncertainty: Risk and the Case against Geological Disposal of Nuclear Waste. Berkeley: University of California Press. Solomon, Miriam (2001), Social Empiricism. Cambridge, MA: MIT Press. Stadler, Friedrich (1997), Studien zum Wiener Kreis. Ursprung, Entwicklung und Wirkung des Logischen Empirismus im Kontext. Frankfurt: Suhrkamp. English translation: The Vienna Circle: Studies in the Origins, Development, and Influence of Logical Empiricism. Vienna and New York: Springer-Verlag, 2001. Stump, David and Galison, Peter (eds.) (1996), The Disunity of Science: Boundaries, Contexts, and Power. Stanford, CA: Stanford University Press. Toulmin, Stephen (1953), The Philosophy of Science: An Introduction. London: Hutchinson & Co. Toulmin, Stephen (1961), Foresight and Understanding: An Enquiry into the Aims of Science. Bloomington, IN: Indiana University Press. Toulmin, Stephen (1972) Human Understanding, Vol. 1. Princeton, NJ: Princeton University Press. Uebel, Thomas E. (1992), Overcoming Logical Positivism from Within: The Emergence of Neurath’s Naturalism in the Vienna Circle’s Protocol Sentence Debate. Amsterdam and Atlanta: Rodopi. Uebel, Thomas E. (ed.) (1991), Rediscovering the Forgotten Vienna Circle: Austrian Studies on Otto Neurath and the Vienna Circle. Boston Studies in the Philosophy of Science, Vol. 133. Dordrecht, Boston, and Lancaster: Kluwer. van Fraassen, Bas (2002), The Empirical Stance. New Haven: Yale University Press. Wang, Jessica (1999), “Merton’s Shadow: Perspectives on Science and Democracy since 1940.” Historical Studies in the Physical and Biological Sciences 30: 279–306. Zhai, Zhenming (1990), “The Problem of Protocol Statements and Schlick’s Concept of ‘Konstatierungen.”’ PSA 1990: Proceedings of the Biennial Meeting of the Philosophy of Science Association, Vol. 1. East Lansing, MI: Philosophy of Science Association, pp. 15–23.

GREGOR SCHIEMANN

INDUCTIVE JUSTIFICATION AND DISCOVERY. ON HANS REICHENBACH’S FOUNDATION OF THE AUTONOMY OF THE PHILOSOPHY OF SCIENCE

Hans Reichenbach’s distinction between a context of discovery and a context of justification continues to be relevant all the way up to the present. This can be seen clearly in the tense relationship between the history and the philosophy of science. In the current debates about the relationships between these two disciplines one encounters arguments that Reichenbach used to defend this distinction, as well as arguments brought forth by his critics.1 Sometimes the discussions even refer directly to the influence of Reichenbach’s distinction (Giere 1999, pp. 11–18 and 217–230). Historically, this influence can be understood to have gone hand in hand with the significance of logical empiricism in the twentieth century for the development of philosophy of science and science studies, i.e., history, sociology and psychology of science. Reichenbach used the distinction in 1938 in his “Experience and Prediction”, which played a key role in the new beginning of logical empiricism in the US. Among the many motives that might have led Reichenbach to formulate this distinction, his intention to contribute to the foundation of the autonomy of a “scientific philosophy” presumably had central importance.2 In this context, its function was to clarify Reichenbach’s stance towards other philosophical trends, to prove the homogeneity of the methodology and content of philosophy of science, and to distance philosophy of science from rival disciplines. Reichenbach’s remarks suggest—and I shall return to this point—that “context of discovery” means above all a part of the research conducted in the natural sciences. One of the messages that Reichenbach wanted to communicate with his distinction was: the “scientific philosophy” of logical empiricism can provide a justification for the theories brought forth in the natural sciences, whereas the natural sciences themselves are not in a position to do so. Since the “historical turn” accompanying Thomas S. Kuhn’s “The Structure of Scientific Revolutions”, the distinction of contexts has been increasingly influential in distinguishing among the philosophy of science and the disciplines of science studies— in particular to the relationship between the philosophy and the history of science. To put it simply, from the perspective of analytical philosophy of science—into which the tradition of logical empiricism passed over—the historical presentation of the natural sciences now stands alongside the natural sciences, which are themselves the province of the context of discovery. I do not want to go into the details of the debates about the justification of disciplinary boundaries and relationships between the history and the philosophy of science. Rather, I would like to assume that Reichenbach’s distinction lives on, and 23 J. Schickore and F. Steinle (eds.), Revisiting Discovery and Justification, 23–39.
C 2006 Springer. Printed in the Netherlands.

24

GREGOR SCHIEMANN AND WUPPERTAL

to seek arguments in his texts that would justify their relevance in this field. The persuasive force of these arguments transcends the contingent circumstances apart from which their genesis and local transmission cannot be made understandable. These arguments have not yet been sufficiently acknowledged in the expansive secondary literature dealing with the context distinction—which might not be a small matter, considering their current influence. My thesis is that, for Reichenbach, the context distinction unfolds its relevance for the formation of disciplines in relation to the theory of induction. This connection has until now only seldom been mentioned.3 This is all the more astounding considering that most of the few passages where Reichenbach employed the distinction subsequent to1935 make direct reference to induction (Reichenbach 1935a, pp. 172–173; Reichenbach 1938, pp. 6–7 and 381– 382; Reichenbach 1944, p. 80; Reichenbach 1947, p. 2; Reichenbach 1949a, pp. 433– 434; Reichenbach 1949b, pp. 291–293; Reichenbach 1951, p. 231). Since induction plays a key role for Reichenbach in the natural sciences as well in epistemology, he is confronted with the question of the autonomy of epistemology. How does the application of and the thinking about induction differ between epistemology and the natural sciences? Are the differences gradual or categorical in nature? Reichenbach’s conclusive answer goes back to the pragmatic justification of induction that he developed in 1933. This justification emphasizes the normative character of epistemology in contrast to the disciplines that study science in a more descriptive manner. I am not interested in attempting a new defence of this justification—which is still controversial today—but in highlighting its role in establishing a discipline’s specific claim to rationality.4 The plausible elements of this function come to light if one looks beyond Reichenbach’s exaggerated fixation on logical analysis. They provide positive points of reference that can be used today in understanding the relationship between the philosophy and the history of science. My presentation of the connection between the context distinction and induction takes into account only the version of the distinction—among the many in Reichenbach’s work (Schiemann 2003; Hoyningen-Huene, this volume)—that relates to its function of contributing to the distinction between the tasks and practices of epistemology and those of other disciplines—which is probably its historically relevant function.5 I can only mention here that this version—as well as the others—goes back to a traditional distinction between genesis and validity that reaches back to Kant (Schiemann 2003; Stadler 2002). I shall begin by characterizing the context distinction as employed by Reichenbach in “Experience and Prediction” to differentiate between epistemology and science (1). Following Thomas Nickles and Kevin T. Kelly, one can distinguish two meanings of the context distinction in Reichenbach’s work. One meaning, which is primarily to be found in the earlier writings, conceives of scientific discoveries as potential objects of epistemological justification. The other meaning, typical for the later writings, removes scientific discoveries from the possible domain of epistemology. The genesis of both meanings, which demonstrates the complexity of the relationships obtaining between epistemology and science, can be made understandable by appealing to the historical context (2). Both meanings present Reichenbach with the task

DISCOVERY ON HANS REICHENBACH’S FOUNDATION

25

of establishing the autonomy of epistemology through the justification of induction. Finally, I shall expound this justification and address some of its elements of rationality characterizing philosophy of science (3). 1.

Since the connection between the context distinction and induction is already made clear in his presentation in section 1 of “Experience and Prediction”, and since Reichenbach provides here a relevant basis for subsequent reflections, I will go into some detail about this passage. Reichenbach speaks in section 1 of three tasks that are typical for epistemology in contrast to other disciplines: the “descriptive”, the “critical” and the “advisory” task. The Descriptive Task The descriptive task consists in the “rational reconstruction” of scientific “thought processes”. (By “science” Reichenbach means here and indeed for the most part natural science.) The concept of “rational reconstruction” is borrowed from Rudolf Carnap (Carnap 1928). It characterizes the normative task of epistemology: “to construct thinking processes in a way in which they ought to occur if they are to be ranged in a consistent system” (Reichenbach 1938, p. 5). Any logical connections missing between the “starting point and issue” of a real thought process are produced in such a way that the postulate of greatest possible correspondence is fulfilled (Reichenbach 1938, p. 5). This postulate has to presuppose that real scientific processes have something at least approximating a logical content. In order to characterize the relative difference between the reconstruction and its object, Reichenbach introduces his context distinction. He does justice to the relativity of this difference by defining rational reconstruction by analogy to scientific practice: “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 1938, pp. 6–5—emphasis in original).

Thus, the distinction between contexts of discovery and justification is intended initially to mark a distinction within scientific practice. Scientific knowledge, according to Reichenbach, is often presented in a form different from that in which it was found. (Following Reichenbach, one could speak of an “inner-scientific” distinction of contexts.) The demarcation of the descriptive task of epistemology from science is a further step beyond this initial distinction. Epistemology differs from science in that its domain is limited to the context of justification, which exists to some extent also within science.

26

GREGOR SCHIEMANN AND WUPPERTAL

Reichenbach does not offer any reason why scientists could not in principle reconstruct their theories themselves. He notes only two factors that prevent scientific presentations from meeting the demands of logic. First, there are “the traces of subjective motivation from which [. . . the scientific expositions] started” (Reichenbach 1938, p. 7). Secondly, “scientific language, being destined like the language of daily life for practical purposes, contains so many abbreviations and silently tolerated inexactitudes” (Reichenbach 1938, p. 7). But these explanations raise more questions than they answer. Why should “subjective motives” and “practical goals” lead to non-logical elements in the presentation of theories? Can these factors also influence logic and epistemology? What stands in the way of eliminating non-logical elements in the sciences—given that they are present—to a degree that satisfies logicians?6 Critical Task Until the initially vague demarcation is sharpened, there can be no foundation of the autonomy of epistemology. This is achieved in part by reference to the following two tasks. In its critical task, epistemology is no longer bound to the postulate of greatest possible correspondence. Reichenbach assumes now that the logical content of scientific thinking is so defective that a rational reconstruction cannot achieve a logically consistent structure. The “analysis of science” must therefore replace reconstruction. We “replace actual thinking by such operations as are justifiable, that is, as can be demonstrated as valid” (Reichenbach 1938, p. 7).

But the analysis of science remains bound to the “starting point and issue” of real thought processes. Within this general condition the inductivist conception of science emerges, which Reichenbach was to develop further in the course of the book. This conception regards relations among data as the starting point and scientific theories—which reproduce and inductively generalize these relations by applying mathematical functions to them—as the issue of scientific thought processes. The goal of epistemology—as set out in chapter 5 of “Experience and Prediction”—is a formal demonstration of the validity of the inductive relations between data and theories. In pursuing this goal, the analysis of science can rely on structures that already exist—however incompletely formulated—in scientific presentations. Indeed, Reichenbach propounds the “thesis that all inferences occurring [in science] are of the inductive type” (Reichenbach 1938, pp. 370–371). The analysis of science endeavors to explicate the logical form of this inductive structure by resolving it into a structure of deductive relations, leaving the rule of induction as the only non-deductive relation (Reichenbach 1949a, p. 471). He understands the rule of induction—to choose one of his many formulations—as “the assumption that an event which occurred n times will occur at all following times” (Reichenbach 1938, p. 341). I should like merely to note at this stage that the justification of this rule would exceed the scope of an

DISCOVERY ON HANS REICHENBACH’S FOUNDATION

27

epistemology characterized by the two tasks I have addresses thus far. As David Hume showed, the rule cannot be logically proven. Its justification—if at all possible – demands that one abandon the field of logical analysis, which is the third task of epistemology licenses. Advisory Task In performing its advisory task, epistemology meets science as an independent discipline with positive suggestions. This is necessary, as Reichenbach sees it, because the analysis of science may be an indispensable tool for making correct decisions between different modes of presentation and different directions in which science might be developed. With respect to different modes of presentation, Reichenbach only speaks of the choice between “equivalent conceptions” (Reichenbach 1938, p. 9), i.e., theories that he considers logically equivalent because they agree in all observable facts (Reichenbach 1938, p. 374; Reichenbach 1951, p. 180). One could mention at this point also the case of the non-equivalent theories Reichenbach discusses in chapter 5, which differ in their prognoses and therefore present a problem of making a choice too (Reichenbach 1938, pp. 375–376). Unfortunately, Reichenbach offers no precise characterization of these two types of relationship between theories. It remains unclear just how broadly he construes the concept of logical equivalence. Strictly speaking, logical equivalence is only derived from empirical equivalence when all theoretical concepts are defined by observational concepts (Friedman 1983, p. 280; Klein 2000, p. 85). In any case, it is clear that for Reichenbach only logical analysis can determine whether theories are equivalent or non-equivalent. In the case of equivalent theories, science can be relied upon to choose between them on the basis of mere expediency. As for non-equivalent theories, on the other hand, Reichenbach believes that his theory of induction provides the framework in which a decision can be made (Reichenbach 1938, pp. 375–379). Hence, epistemology could exercise an important influence in the presentation of scientific theories. Apparently Reichenbach does not take into account (at least not in the context of “Experience and Prediction”) the empirical underdetermination of scientific theories, as formulated by Pierre Duhem. Underdetermined theories may have the property of being logically incompatible and thus elude the competency of an analysis of science. If Reichenbach had taken this property into consideration, he could only have maintained the role of epistemology in examining and choosing between theories by abandoning his limitation to logical analysis and—like Neurath, for example—admitting values as a philosophical topic.7 That this would not have been wholly out of the question is demonstrated by his ideas about “volitional bifurcations” which concern the direction in which science might be developed. Choices about the aims of research count as such bifurcations (Reichenbach 1938, pp. 10–11). Although Reichenbach does not mention it at this stage, justification of the rule of induction would also have to be included here, since it depends on the specific aim of predicting the future. Moreover, fundamental concepts

28

GREGOR SCHIEMANN AND WUPPERTAL

of a philosophy of science—i.e., causality and probability—can be traced back to volitional bifurcations (Reichenbach 1938, p. 11). Reichenbach does make the claim that the advisory task could be reduced to the critical task if epistemology is restricted to “constru[ing] a list of all possible decisions, each one accompanied by its entailed decisions [. . . and] leav[ing] the choice to our reader” (Reichenbach 1938, p. 14). But in the course of the book he does not justify his assertion that epistemology itself is free from volitional bifurcations. In subsequent chapters he makes decisions about the fundamental concepts of philosophy of science and logical presentation without displaying the alternatives neutrally to the reader.8 If one traces the relationship between epistemology and science step-by-step through the delineation of the three tasks, one arrives at a paradoxical conclusion. On the one hand, the distance between the two disciplines increases successively from the initial postulate of maximal correspondence to the merely formal relationship of inductive structure, and finally to the stage at which epistemology fulfils its advisory task as an independent discipline. But there is also a sense in which the distinction between the two disciplines becomes increasingly problematic. When Reichenbach asserts that the formal structure of scientific reasoning is essentially inductive and knowledge thereof is decisive in examining and choosing theories, the question presents itself to the reader: why should science have to yield the task of analyzing this structure to another discipline at all? Epistemology, in Reichenbach’s view, differs from science in that it is limited to the context of justification and focuses on its logical presentation. But the inductive structure of logical presentation refers to the question of discovery. Reichenbach conceives of “the aim of induction” as finding “the laws of nature in their most general form, including both statistical and so-called causal laws” (Reichenbach 1938, p. 350). If real processes of discovery until now have not followed the inductive method, could epistemology develop an alternative logic of discovery? Reichenbach seems however to suspect that there are also rational elements already in real processes of discovery. In his critique of Karl Poppers’s “Logik der Forschung” in 1935, he states that “the process of constructing scientific hypotheses ought to be rationalized” (Reichenbach 1935b, p. 281). Can the thought processes involved in discovery be at least partially grasped through inductive logic? These questions show that the theory of induction plays a decisive role in the characterization of the context distinction as a criterion for distinguishing the two disciplines. 2.

Reichenbach gives no clear answer to these questions. In fact, his views on the context distinction were subject to vacillations over the years that cannot neatly be assigned to specific time periods. One can however tentatively distinguish two phases in which he emphasizes different aspects of its meaning (Nickles 1990, p. 158; Kelly 1991, pp. 137–139). In an early phase he places weight on the rationalization of processes of discovery as a matter for epistemology, whereas in a later phase he is inclined to

DISCOVERY ON HANS REICHENBACH’S FOUNDATION

29

doubt that this rationalization is possible and eliminates it from the epistemological context of justification. The Early Phase: Rationalizing Discovery The last two paragraphs of “Experience and Prediction” belong to the early phase. Here the distinction of contexts does not prevent scientific discoveries from following the inductive method too. “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 [, . . . ] that sometimes those men will be best in making inductions who believe they posses other guides” (Reichenbach 1938, p. 403) (Reichenbach uses the term “principle of induction” here and elsewhere synonymously with “rule of induction”.)

The context distinction assumes the shape of a distinction formulated in Marxist terminology between super- and supporting structure.9 The inductive supporting structure is susceptible of justification, whereas the mythology of the superstructure—i.e., “instinctive presentiments” and “belief in a mystic harmony between nature and reason” (Reichenbach 1938, pp. 403–404), but also “belief in induction, belief in a uniformity of the world” (Reichenbach 1938, p. 403)—is not. These notions, which for whatever inexplicable reasons are necessary for the application of induction in the sciences, could only be left to psychology to investigate. Within the framework provided by this conception, the history of science—which Reichenbach does not address—would deal with the superstructure as well as the supporting structure (cf. Salmon 1970). As much as these and other formulations in the last two paragraphs of “Experience and Prediction” seem to place Reichenbach in proximity to a “logic of discovery” (Nickles 1980; Curd 1980; Laudan 1980), a close reading reveals that he does not go beyond general statements like those in the passage quoted above. In none of the examples he mentions does he claim that natural scientists “find” theories by applying an inductive method.10 It always remains an open question how the scientists arrive at their theories. It may be that they are guided by inductive considerations, but as for this question, Reichenbach says merely: “we do not maintain anything” (Reichenbach 1938, p. 382). He does not question the distinction between supporting structure and superstructure. At the same time, it stands firm in “Experience and Prediction” that theory generation in the sciences may in principle proceed inductively, and that the context of discovery belongs among the possible objects of an analysis of science.11 Against this background, Reichenbach must have been uncertain whether it might have been better to advise scientists to employ the inductive method in their pursuit of discovery instead of abandoning themselves to mythology. His ambivalent stance is reflected in the rhetoric with which he invests the distinction of contexts in the penultimate paragraph. On the one hand, he disparages scientists by suggesting that they

30

GREGOR SCHIEMANN AND WUPPERTAL

misunderstand the nature of their own method and are trapped in the false consciousness of the superstructure.12 On the other hand, he himself yields to mythology by invoking the unfathomable genius who is not guided by logic but “follows other ways” (Reichenbach 1938, p. 382). Directly after establishing that the principle of induction “is the only rule the physicist has at hand” (Reichenbach 1938, p. 380), Reichenbach broaches the subject of the context distinction. The distinction posits a gap precisely at the point where epistemology and science are at their closest to each other. If induction is the decisive methodological principle not only in justification but also in discovery, then the innerscientific validity of the context distinction—and along with it its function as a criterion for distinguishing between science and epistemology—is relativized significantly. Within this framework the context distinction takes on an arbitrary—and indeed dogmatic—character.13 That this dogmatic character does not exhaustively mark the distinction is clear in view of the fact that Reichenbach still has two options for justifying the distinction. First, the nature of the non-rational factors in the sciences remains unexplained (i.e., the influence of the “subjective motivations”, “practical purposes” and the mythological superstructure). If these factors unavoidably exist and hinder scientists from taking over the logical analysis of their own discoveries and justifications, the independence of epistemological analysis of science would be justified. But if this cannot be established—and Reichenbach does not establish it—then epistemology would nevertheless retain one last task, which is in fact central in Reichenbach’s book although he does not mention it as such: namely, the justification of the inductive procedure in the sciences. This task is not only beyond the scope of a logical analysis but also inaccessible to empirical science (see part 3 below). The Later Phase: Discoveries Do Not Belong to the Context of Justification By 1949—when the English edition of the “Theory of Probability” appeared—the later phase had already begun. Here—after emphasizing the significance of induction for the empirical sciences again—Reichenbach sets epistemology in sharp opposition to the context of discovery. He states clearly: “I simply refuse the challenge of developing rules for a logic of discovery. Such rules do not exist. The logic of induction has to do with the critical process of checking over solutions that already exist” (Reichenbach 1994, pp. 439–448).14

Hence, the logical distinction gives way to a temporal one, according to which the epistemological reconstruction of a new theory cannot begin until scientists have formulated the theory.15 This qualification need not have any effect upon the conception of the inner-scientific relationship between discovery and justification. True, Reichenbach now stresses that discoveries can “only be analyzed with psychological, not with logical means” (Reichenbach 1951, p. 231; Reichenbach 1994, p. 439). But he still believes for example that the “logical schema of the theory of relativity corresponds surprisingly with the program which controlled its discovery” (Reichenbach

DISCOVERY ON HANS REICHENBACH’S FOUNDATION

31

1949b, p. 293). Though the relationship between discovery and justification in the sciences may be complex, the tasks of epistemology remain henceforth limited to the context of justification. Epistemology has no pretensions about reconstructing the context of discovery.

The Historical Context of the Discussion of Induction Although I do not want to over-emphasize the vacillations in Reichenbach’s conception of the context distinction, I would like to have a closer look at its historical context— including its establishment through the justification of induction. The vacillations can be attributed to uncertainties that are understandable in light of the considerable difficulties faced by the logical empiricists as they re-settled to the US in the 1930s. According to Ronald N. Giere, philosophy in German in the opening decades of the twentieth century was shaped by a “Kantian framework”, whereas the “philosophical climate” of the US was dominated by the traditions of Empiricism and Pragmatism (Giere 1999, p. 225). Giere invokes this difference to explain the fact that probability and induction, which had “not played a significant role” for the logical empiricists before their relocation, became “major topics in the philosophy of science.” Indeed, although Reichenbach had occupied himself as early as 1916—in his dissertation— with the theory of probability, it was not until he left Berlin that he began actively publishing work on the induction problem.16 From that point on, induction was to remain in the foreground—not only in “Experience and Prediction” but also in the English edition of the “Theory of Probability.” In the US, induction was an ideal area for Reichenbach to establish himself with a contribution.17 With his justification of the rule of induction, he succeeded. He himself believed that he had no less than “finally arrived at a solution to the induction problem”, which had stood “as the unsolved riddle before all empiricist philosophy since the time of Hume” (Letter of June 3, 1935 to Ernst von Aster, quoted in Kamlah 1994, p. 533). In “Experience and Prediction” and in the English edition of the “Theory of Probability”, the justification of the rule of induction assumes central importance. Reichenbach intended to show with the context distinction that the logical analysis of induction and the justification of induction he had proposed are among the tasks that belong exclusively within the province of epistemology. At the same time, only after justifying the rule of induction could Reichenbach demonstrate that logical analysis avails itself exclusively of justified laws and thereby establish the distinction between contexts of discovery and justification. Justification of the laws of deduction is unproblematic, since they “always lead to true sentences if the premises are true. [. . . But for] the rule of induction such a proof is not possible” (Reichenbach 1949a, p. 471). In the absence of a justification of the rule of induction, there can be no response to the Humean argument that one assumes the validity of induction out of habit. Induction would belong alongside the non-rational elements of discovery among the objects that can only be analyzed by psychology. Thus neither the earlier

32

GREGOR SCHIEMANN AND WUPPERTAL

interpretation of the context distinction nor the later one can be maintained without a justification of the rule of induction. 3.

In order to distinguish Reichenbach’s justification of the rule of induction from a “validation”, Herbert Feigl characterized it as a “vindication”. Rules are validated when they are derived from a fundamental principle. When this is impossible—as in the case of the rule of induction—their use can be vindicated by defending its appropriateness for a given purpose (Feigl 1950). I would like to add that a rule which has been justified in this manner can be regarded as rational in an instrumental sense, since the appropriateness of its application is established only as a means to a given goal. A more broadly construed concept of rationality could surely encompass the well-founded choice of goals. But this would not accurately represent Reichenbach’s procedure, since he gives no reasons for his choice of the aim of the scientific method, namely “of predicting the future” (Reichenbach 1949a, p. 474, or Reichenbach 1938, pp. 349–350). Moreover, he does not address alternatives to this resolution—like Goethe’s decision to limit his scientific research to the description of nature, or the atomistic program of explaining nature, to mention just two of the historically most significant examples—although such a discussion would be possible within the framework of the advisory task of epistemology (see part 2 above). The decision in favor of prognosis determines Reichenbach’s formulation of the rule of induction, i.e., “that an event which occurred n times will occur at all following times” (Reichenbach 1938, p. 341; see above). He believes that a defense of induction can only succeed within the framework of his theory of probability. The core of his argumentation can however be understood independently of the specific probabilistic interpretation with which he invests it. It corresponds to the structure of Blaise Pascal’s famous wager, according to which one must decide in favor of belief in the existence of God, since this is the only way one might—if God indeed exists—attain truth and salvation.18 Like Pascal, Reichenbach assumes two possible states of the world and two possible courses of action. For Reichenbach, the world can be either uniform or non-uniform. Insofar as he does not think it possible to know the actual state of the world, he shares Hume’s scepticism. The possible courses of action—given independently of the state of the world—can be divided disjunctively into the inductive and the non-inductive. This scheme yields four possible states: 1. The application of the inductive method leads in the long run to certain success in the uniform world. 2. In the non-uniform world its successes are merely coincidental, thus its success is on the whole uncertain. 3. Uncertainty of prognostic success is also characteristic of the state of the nonuniform world when the inductive method is not employed, i.e., when either no methodology or non-inductive methods such as clairvoyance are used.

DISCOVERY ON HANS REICHENBACH’S FOUNDATION

33

4. But it in fact makes a difference in the uniform world which of these non-inductive variants one chooses. Only the absence of any methodology at all would in general be unsuccessful. The possible success of a non-inductive method, however, could only be assessed in comparison to the results of the inductive method. There can be no guarantee of prognostic success in the uniform world when induction is not employed.19 Given Reichenbach’s suppositions, induction is the surest procedure for attaining successful prognoses—when success is possible. The first state is the only rational one. The vindication has a pragmatic character, since—and herein lies another resemblance to Pascal’s wager—it does not defend belief in the success of actions but the rationality of a procedure for choosing actions.20 “Actions directed by the rule of induction are legitimate attempts at success; no form of belief is required for the proof ” (Reichenbach 1949a, p. 481). Hence, belief in the success of induction is just as mythical in nature as belief in the success of non-inductive procedures in the sciences. The validity of Reichenbach’s vindication remains controversial.21 As I have already stated, I do not want to enter into a discussion of the criticism. Rather, I would like simply to concede the argument—with one qualification—in order to show how a claim to rationality can be utilized in establishing the autonomy of a discipline. The qualification concerns Reichenbach’s exclusive identification of the goal of science with prognosis. Although one may concede the plausibility of his argument that the choice of induction in pursuance of this goal is rational, one ought not forget that other goals might be incompatible with induction and other means necessary to achieve them. To stick with the aforementioned examples, Goethe’s research of nature led him to a critique of induction, and the atomistic program of explaining nature led to the assumption of non-inductive hypotheses. I would like to emphasize three of the things that can be learned from Reichenbach’s pragmatic justification of induction and its relation to the context distinction. First, Reichenbach is right to emphasize the tremendous significance of induction in the natural sciences. Since the Aristotelian beginnings of science and of theoretical reflection about the methods of science, induction has played a key role in both areas. The inductive procedure is among the methods of contemporary science the application of which transcend epochal and disciplinary boundaries. The focus on such non-local structural patterns in the development of science is more characteristic of the philosophy of science than of the history of science (Radder 1996). Induction being among these structural patterns (others would be mathematization and technization), the context of justification being the name of an appropriate domain for theoretical reflection about these patterns, and the context of discovery being the name of an appropriate domain for the presentation of their historical reality, one may say that the context distinction reflects a plausible relation between philosophy and history of science. Secondly, the discussion of induction is not only thematically, but also methodologically, exemplary of specific characteristics of philosophy of science. This is made

34

GREGOR SCHIEMANN AND WUPPERTAL

clear by the argumentative character of the vindication of induction, upon which Reichenbach bases epistemology’s claim to justification. According to Reichenbach “the justification of a theory in terms of observational data is the subject of the theory of induction,” which—apart from the rule of induction—avails itself exclusively of deductive laws (Reichenbach 1951, p. 231). This claim of justification holds for every application of the theory of induction, whether it be with approximately logical content as in the sciences, or with exclusively logical form as in Reichenbach’s epistemology. But the vindication—in contrast to its object, namely induction—is categorically distinct from the scientific justification, since it proceeds not empirically but argumentatively. Thus the vindication of induction establishes the autonomy of epistemology even if an epistemological analysis of science cannot be distinguished sharply from the inductive procedure of the sciences, as in the earlier of the aforementioned phases of the context distinction. Argumentative justifications are typical of philosophy and still constitute a criterion for distinguishing philosophy from science studies. Thirdly, in addition to distinguishing philosophy form other disciplines, the vindication yields an inner-philosophical distinction as well. Its object differs for example from all non-inductive procedures, such as Popper’s deductive falsificationism. Its pragmatic character distinguishes it from Carnap’s attempt to meet Hume’s challenge with a new concept of theoretical rationality (Schramm 1993, pp. 548–553). 4. CONCLUDING REMARKS

I set out to look for arguments to account for the relevance of Reichenbach’s distinction between a context of discovery and a context of justification upon the establishment of the autonomy of philosophy of science (which he calls epistemology). The autonomy of this discipline is problematic for Reichenbach because of its close relationship to the practice and theory of science. Justification is conceived of in his writings as essentially inductive and as a topic for the sciences and epistemology. The epistemological justification differs from the scientific one only in that it is limited to logical form. It remains unclear why the sciences should not also be competent to carry out a purely logical analysis of their inductive methods of inference. According to Reichenbach, it is more typical for the sciences than for epistemology to aim for discoveries. In his writings it remains an open question to what extent the processes of scientific discovery are inductive and hence a potential object of inquiry within the context of (scientific or epistemological) justification. In part, he seems not to want to exclude the possibility of future processes of discovery employing a strictly inductive logic, which would then dissolve the context distinction. But as for discoveries made up to the present, he assumes for the most part that they arise from circumstances as yet not uniformly characterized. Induction, on the other hand, is for Reichenbach a method generally characteristic of the sciences and of the justification of theories. In Reichenbach’s view, epistemology stands in danger of being dissolved into the sciences. But his context distinction is no merely verbal, last-ditch attempt at distinguishing epistemology. Reichenbach characterizes epistemology by appealing to

DISCOVERY ON HANS REICHENBACH’S FOUNDATION

35

the fact that it thematizes general characteristics of science. Although philosophy of science has long since stopped limiting itself to the logical analysis of scientific methods, this is still an aspect—alongside its focus on contemporary problems— which distinguishes it from most research in the history of science. Reichenbach’s justification of induction demonstrates paradigmatically how philosophy employs its own method—essentially argumentative in nature—outside the domain in which it seeks to justify the knowledge produced in other disciplines. This vindication belongs alongside Pascal’s wager in the tradition of pragmatic arguments. Analyzing it reveals that Reichenbach’s one-sided focus on the prognostic goals of science and on the inductive method does not entail the exclusion of discoveries from the context of justification. Rather, it impugns the claim to validity of non-inductive procedures taken to be pre-requisites to the formulation of theories. If one drops the claim to exclusivity with which Reichenbach invests his vindication, one can learn from his procedure that the foundation of the autonomy of a discipline can emerge from a justification that ascribes rationality to the methods of that discipline. Reichenbach was right in viewing methodological rationality as dependant upon the goals attributed to a given domain of research, yet he overlooked the plurality of such goals. ACKNOWLEDGMENTS

I am grateful to the participants of the Workshops “Revisiting Discovery and Justification” at the Max-Planck-Institut f¨ur Wissenschaftsgeschichte for their helpful comments. Thanks also to Professor Kamlah for his stimulating suggestions, and to John Michael for his translation. NOTES 1. The literature concerning the debate between philosophy and history of science is extensive. For an overview see Laudan 1990; Hull et al. (eds.), 1992; Nickles 1995. For more recent publications, see for example Radder 1997; Pinnick and Gale 2000; Burian and Steinle 2002. 2. Reichenbach’s motivations in formulating the context distinction are discussed in Zittlau 1981, 44; Nickles 2002; Stadler 2002; Howard 2004; Howard, this volume. 3. Concerning the reception of the context distinction generally, i.e. insofar as it cannot be attributed exclusively to Reichenbach, see Hoyningen-Huene 1987; for the reception of Reichenbach’s version of it Schiemann 2003, p. 237. In connection with the treatment of Reichenbach’s theory of induction, there is some discussion of the context distinction in Salmon 1991; Kelly 1991. As for Reichenbach’s theory of induction, see also Clendinnen 1979; Schramm 1993. 4. Concerning the debate about Reichenbach’s justification of the theory of induction, see Clendinnen 1979; Schramm 1993; Kamlah 1994, 545–549; Gerner 1997, 165 et seq. More recently Piller 1987; Schurz 1988. 5. This version corresponds most nearly to the fourth version characterized by Hoyningen-Huene (this volume), although Reichenbach counts the natural sciences among those empirical disciplines from which epistemology is distinct. 6. In another passage, Reichenbach explains “the division of labour between the physicist and the philosopher” by referring to the “limitation of human capacities”: “It appears to be practically impossible that the man who is looking for new physical laws should also concentrate on the

36

7. 8. 9. 10.

11.

12.

13.

14.

15. 16.

17.

18.

19. 20.

GREGOR SCHIEMANN AND WUPPERTAL analysis of his method”. Moreover, there is a difference in the mentalities of the two groups of people: “The discovery of general relations [. . . ] requires a mentality different from that of the philosopher, whose methods are analytic and critical rather than predictive” (Reichenbach 1949b, p. 292). Cf. Howard’s and Richardson’s discussions in this volume. The definition of probability based on frequency is, for example, by no means the only alternative; and the reduction of causality to inductive relations of probability is also controversial. Maria Reichenbach and Hermann Vetter translate “super- and supporting structure” with “Unter¨ und Uberbau” (Reichenbach 1983, p. 253). Referring to Galileo he constrains himself to the formulation “he found that the quantities measured fit into the formula s = gt2 /2, and inferred, by means of the inductive principle, that the same law holds for similar cases” (Reichenbach 1938, p. 371). Of Darwin he only writes: when “Darwin formulated [his] theory, it was based on facts” (Reichenbach 1938, p. 390). And as for Einstein, he remarks merely that he “saw—as his predecessors had not seen—that the known facts indicate such a theory; i.e., that an inductive expansion of known facts leads to the new theory” (Reichenbach 1938, p. 382). Reichenbach does not consistently assert the rationalizability of discoveries throughout his early writings: “the procedure of discovery is however hardly rationalizable” (Reichenbach 1935a, p. 172). On other occasions Reichenbach states more explicitly that he does not even believe scientists who claim to have arrived at their theories by non-inductive means: (Reichenbach 1935b, pp. 281–282). Some interpreters claim that Reichenbach does not justify the distinction in “Experience and Prediction” (Giere 1999, p. 228; Kelly 1991, pp. 137–140). Since later discussions of the distinction do not repeat the justification, Kelly speaks of a “dogma” (Kelly 1991, p. 139). This characterization is inaccurate, though, considering that Reichenbach attempted to justify the distinction not only directly upon introducing it but also in the context of his work. In Reichenbach 1949b, the “challenge of developing rules for a logic of discovery” is not sharply distinguished from logical analysis: the “philosopher of science is not much interested in the thought processes which lead to scientific discoveries” (Reichenbach 1949b, p. 292). This meaning matches up with Hoyningen-Huene’s Version 1 (this volume). He made the first announcement of his “illumination of the induction problem” in a letter to Moritz Schlick dated February 22, 1933 (Kamlah 1994, p. 546). There is an anecdote that Reichenbach proclaimed as the Nazis closed the Berlin University: “Now I understand the induction problem” (Giere 1999, p. 226). Giere 1996 assumes that Reichenbach applies the context distinction in order to establish induction as a topic for Logical Empiricism. However, Howard 2004b, p. 23, points out that the context distinction received critically among American Pragmatists. Pascal 1984, Fragm. 233. On the structure of Pascal’s wager, cf. Hacking 1975, pp. 63–72; Jordan 1994. On the similarity between Reichenbach’s pragmatic justification and Pascal’s wager see Creed 1939; Salmon 1991; Gerner 1997; Schramm 1999; Kamlah 1977, p. 479 draws attention to a difference. Reichenbach 1933; Hertz 1936; Reichenbach 1936; Reichenbach 1949a, pp. 469–482; Reichenbach 1938, pp. 348–357; Salmon 1991, p. 100; see also literature of Footnote 4. In 1939, Reichenbach referred explicitly to the close systematic relationship between the pragmatic character of his justification of induction and the work of American pragmatists Dewey and Peirce on the induction problem (Reichenbach 1939, pp. 187–190). It cannot be ruled out that he may have been chosen to work on the issue of induction, and also proposed a pragmatic justification of the rule of induction, with a mind to improving his chances for a professorship in the US. In doing so, Reichenbach distanced himself from the typically Kantian transcendental justification, which he himself developed in his dissertation on the related problem of probability. The pragmatic and the transcendental justifications share only the general demand for “a necessary

DISCOVERY ON HANS REICHENBACH’S FOUNDATION

37

condition” of knowledge in common (Reichenbach 1936. cf. Kamlah 1977, pp. 476–480; Kamlah 1989, pp. 443–447; Richardson, this volume). 21. Cf. Footnote 4.

REFERENCES Carnap, R. (1928), Der logische Aufbau der Welt (Berlin: Weltkreis). Clendinnen, J. F. (1979), “Inference, Practice and Theory”, in Salmon (ed.), pp. 85–128. Creed, I. P. (1940), “The Justification of the Habit of Induction” The Journal of Philosophy 37: 85–97. Curd, M. (1980), “The Logic of Discovery: An Analysis of Three Approaches”, in T. Nickles (ed.), Scientific discovery, Logic and Rationality (Dordrecht: Reidel), pp. 201–219. Feigl, H. (1950), “De Principiis Non Disputandum . . . ?”, in M. Black (ed.), Philosophical Analysis (Ithaca), pp. 119–56. Friedman, M. (1983), Foundation of Space-Time Theories (Princeton: Princeton University Press). Gerner, K. (1997), Hans Reichenbach—sein Leben und Wirken, eine wissenschaftliche Biographie (Osnabr¨uck: Phoebe-Autorenpress). Giere, R. N. (1999), Science without Laws (Chicago: University of Chicago Press). Hacking, I. (1968), “One Problem about Induction”, in L. Lakatos (ed.), The Problem of Inductic Logic (Amsterdam : North-Holland Publishing Company), pp. 44–59. Hacking, I. (1975), The Emergence of Probability (Cambridge: Cambridge University Press). Haller, R. und F. Stadler (eds.), (1993), Wien–Berlin–Prag: Der Aufstieg der wissenschaftlichen Philosophie. Zentenarien Rudolf Carnap—Hans Reichenbach—Edgar Zilsel (Wien: H¨older-PichlerTempsky). Hertz, P. (1936), “Kritische Bemerkungen zu Reichenbachs Behandlung des Humeschen Problems”, Erkenntnis 6: 25–31 Howard, D. (2004), “Two Left Turns Make a Right: On the Curious Political Career of North American Philosophy of Science at Mid-Century”, manuscript. Hoyningen-Huene, P. (1987), “Context of Discovery and Context of Justification”, Studies in History and Philosophy of Science 18: 501–515. Hull, D. et al. (eds.), (1992), “Symposium: What has the History of Science to say to the Philosophy of Science?”, in D. Hull et al. (eds.), (1992): Proceedings of the 1992 Biennial Meeting of the Philosophy of Science Association, pp. 467–496. Jordan, J. (ed.) (1994), Gambling on God. Essays on Pascal’s Wager (Lanham: Rowman & Littlefield). Kamlah, A. (1977), “Erl¨auterungen, Bemerkungen und Verweise zum Vortrag ‘Rationalismus und Empirismus’ und zum Buch ‘Der Aufstieg der wissenschaftlichen Philosophie”’, in Reichenbach 1971– 1999 Vol. 1, pp. 466–480. Kamlah, A. (1985), “The Neokantian Origin of Hans Reichenbach’s Principle of Induction”, in N. Rescher (ed.), The Heritage of Logical Positivism (Lanham: University Press of America). Kamlah, A. (1989), “Erl¨auterungen, Bemerkungen und Verweise zu den Schriften dieses Bandes”, in Reichenbach 1971–1999 Vol. 5, pp. 371–454. Kamlah, A. (1994), “Erl¨auterungen zu: H. Reichenbach, Wahrscheinlichkeitslehre”, in Reichenbach 1971–1999 Vol. 7, pp. 519–549. Kamlah, A. (1998), “Hans Reichenbach”, in E. Craig (ed.), Routledge Encyclopedia of Philosophy (London/New York: Routledge). Kelly, K. T. (1991), “Reichenbach, Induction, and Discovery”, Erkenntnis 35: 123–149. Klein, C. (2000), Konventionalismus und Realismus. Zur Erkenntnistheoretischen Relevanz der empirischen Unterbestimmtheit von Theorien (Paderborn: Mentis-Verlag). Laudan, L. (1980), “Why Was the Logic of Discovery Abandoned?”, in Nickles (ed.), pp. 173–183. Laudan, L. (1990), “The History of Science and the Philosophy of Science”, in R. Olby (ed.), Companion to the History of Science (London/New York: Routledge), pp. 47–59.

38

GREGOR SCHIEMANN AND WUPPERTAL

Nickles, T. (1980), “Introductory Essay: Scientific Discovery and the Future of Philosophy of Science”, in Nickles (ed.), pp. 1–59. Nickles, T. (1990), “Discovery”, in R. Olby (ed.), Companion to the History of Science (London New York: Routledge), pp. 148–165. Nickles, T. (1995), “Philosophy of Science and History of Science”, in Osiris: A Research Journal Devoted to the History of Science and its Cultural Influences Vol. 10, pp. 139–163. Nickles, T. (2002), “The Discovery-Justification-Distinction and Professional Philosopy of Science”, in Schickore and Steinle (eds.), pp. 67–78. ¨ hg. und eingel. von E. Wasmuth (Stuttgart: Reclam). Pascal, B. (1984), Gedanken. Ub., Piller, C. (1987), “Das Vindizierungsargument—seine Wichtigkeit, seine Wirksamkeit seine Widerlegung”, Grazer Philosophische Schriften 29: 35–58. Pinnick, C., and G. Gale (2000), “Philosophy of Science and History of Sciene: A troubling Interaction”, Journal for General Philosophy of Science 31: 109–125. Radder, H. (1996), In and About the World. Philosophical Studies of Science and Technology (New York: State University of New York Press). Radder, H. (1997), “Philosophy and History of Science: Beyond the Kuhnian Paradigm”, Studies in History and Philosophy of Science 28: 633–655. Reichenbach, H. (1929), “Ziele und Wege der physikalischen Erkenntnis”, in H. Geiger und K. Scheel (eds.), Handbuch der Physik. Vol. 4. (Berlin: Springer). Reichenbach, H. (1933), “Die logischen Grundlagen des Wahrscheinlichkeitsbegriffes”, Erkenntnis 3: 401–425. Reichenbach, H. (1935a), “Zur Induktionsmaschine”, Erkenntnis 5: 172–173. ¨ Reichenbach, H. (1935b), “Uber Induktion und Wahrscheinlichkeit. Bemerkungen zu Karl Poppers ‘Logik der Forschung”’, Erkenntnis 5: 267–284. Reichenbach, H. (1935c), Wahrscheinlichkeitslehre (Leiden: A.W. Sijthoff’s Uitgeversmaatschappij). Reichenbach, H. (1936), “Warum ist die Anwendung der Induktionsregel f¨ur uns notwendige Bedingung von Voraussagen?”, Erkenntnis 6: 32–40. Reichenbach, H. (1938), Experience and Prediction. An Analysis of the Foundation and Structure of Knowledge (Chicago/Illinois: University of Chicago Press). Reichenbach, H. (1939), “Dewey’s Theory of Science”, in P. Schilpp (ed.), The Philosophy of John Dewey (Evanston: North Western University). Reichenbach, H. (1944), “Philosophische Grundlagen der Quantenmechanik”, in Reichenbach 1971– 1999 Vol. 5. Reichenbach, H. (1947), “Grundz¨uge der symbolischen Logik”, in Reichenbach 1971–1999 Vol. 6. Reichenbach, H. (1949a), The Theory of Probability. An Inquiry into the Logical and Mathematical Foundation of the Calculus of Probability (Berkeley/Los Angeles: University of California Press). Reichenbach, H. (1949b), “The Philosophical Significance of the Theory of Relativity”, in P. A. Schilpp (ed.), Albert Einstein: Philosopher—Scientist (Evanston: The Library of Living Philosophers, pp. 287–311). Reichenbach, H. (1951), The Rise of Scientific Philosophy (Berkeley etc.: University of California Press). Reichenbach, H. (1968), Der Aufstieg der wissenschaftlichen Philosophie (Braunschweig: Vieweg). Reichenbach, H. (1971–1999), Gesammelte Werke in 9 B¨anden (Braunschweig/Wiesbaden: Vieweg). Reichenbach, H. (1983), Erfahrung und Prognose. Eine Analyse der Grundlagen und der Struktur der Erkenntnis, in Reichenbach 1971–1999 Vol. 4. Reichenbach, H. (1994), “Wahrscheinlichkeitslehre”. 2. Auflage auf Grundlage der erweiterten amerikanischen Ausgabe bearbeitet und herausgegeben von G. Link. Mit Erl¨auterungen von A. Kamlah, in Reichenbach 1971–1999 Vol. 7. Salmon, W. C. (1968), “The Justification of Inductive Rules of Inference”, in L. Lakatos (ed.), The Problem of Inductic Logic (Amsterdam: North-Holland Publishing Company), pp. 24–43. Salmon, W. C. (1970), “Baye’s Theorem and the History of Science”, in H. Stuewer (ed.) Historical and Philosophical Perspectives of Science (Minnesota: University of Minnesota Press). pp. 68–86. Salmon, W. C. (1991), “Hans Reichenbach’s Vindication of Induction”, Erkenntnis 35: 99–122.

DISCOVERY ON HANS REICHENBACH’S FOUNDATION

39

Salmon, W. C. (ed.), (1979), Hans Reichenbach: Logical Empiricist (Dordrecht etc.: Reidel). Schickore, J. and F. Steinle (eds.), (2002), Revisiting Discovery and Justification. Preprint 211. MaxPlanck-Institut f¨ur Wissenschaftsgeschichte (Berlin). Schiemann, G. (2003), “Criticizing a difference of contexts. On Reichenbach’s Distinction between “Context of Discovery” and “Context of Justification””, in F. Stadler (ed.), The Vienna Circle and Logical Empiricism. Re-evaluation and Future Perspectives (Dordrecht), pp. 237–252. Schramm, A. (1993), “Zwei Theorien der Induktion—Reichenbach und Carnap”, in Haller und Stadler (eds.), pp. 538–554. Schurz, G. (1988), “Das Vindizierungsargument funktioniert doch: Eine Erwiderung auf Christian Piller”, Grazer Philosophische Studien 32: 187–195. Stadler, F. (2002), “Challenging the Dogma of the Ahistorical Philosophy of Science: The Case of Relativism”, in Schickore and Steinle (eds.), pp. 27–40. Steinle, F., and R. M. Burian (2002), “Introduction: History of Science and Philosophy of Science”, Perspectives and Science 10: 391–397. Zittlau, D. (1981), Die Philosophie von Hans Reichenbach (M¨unchen: Minerva-Publikation).

ALAN RICHARDSON

FREEDOM IN A SCIENTIFIC SOCIETY: READING THE CONTEXT OF REICHENBACH’S CONTEXTS

The distinction between the contexts of discovery and justification, this distinction dear to the projects of logical empiricism, was, as is well known, introduced in precisely those terms by Hans Reichenbach in his Experience and Prediction (Reichenbach 1938). Thus, while the idea behind the distinction has a long history before Reichenbach, this text from 1938 plays a salient role in how the distinction became canonical in the work of philosophers of science in the mid twentieth century. The new contextualist history of philosophy that has arisen in recent years invites us into an investigation of the nuances of philosophical distinctions and their roles in shaping the development of disciplines. Logical empiricism played a key role in the historical development of philosophy of science and this contextualist history has revealed a much richer set of projects in logical empiricism than the potted histories had allowed. Many stories have been told about the contexts of justification and discovery; few of those stories have paid more than passing attention to the larger projects in epistemology and meta-epistemology that Reichenbach was pursuing when he drew the distinction. This brief essay will seek partially to rectify that lack in, I hope, a somewhat surprising way. I shall stress the connection between this canonical distinction and some other epistemological and social terms that loom large in Reichenbach’s text, arguing that the social relevance of scientific philosophy for Reichenbach cannot be set aside in understanding his use of the DJ distinction. My point is, therefore, historical and reflexive. If we attend to the larger significance of the project in scientific philosophy that Reichenbach was advancing, we can see more clearly why the DJ distinction was introduced and rethink the significance of questioning the distinction. I do not mean to defend the distinction here, but I do hope that my discussion of Reichenbach’s project reveals its attractiveness to him and, potentially, to us. First, some familiar stage-setting: Reichenbach’s distinction was meant, as is clear from the first few pages of Experience and Prediction, to solve the key meta-epistemological question of his time, clearly to demarcate epistemology from psychology.1 In this project, Reichenbach found aid and comfort in Rudolf Carnap’s claim in the Aufbau that epistemology provided rational reconstructions of the process of cognition, writing: “For this logical substitute the term rational reconstruction has been introduced; it seems an appropriate phrase to indicate the task of epistemology in its specific difference from the task of psychology” (Reichenbach 1938, pp. 5–6). This

41 J. Schickore and F. Steinle (eds.), Revisiting Discovery and Justification, 41–54.
C 2006 Springer. Printed in the Netherlands.

42

ALAN RICHARDSON

demarcation between psychology and epistemology frames the terms within which Reichenbach then proceeds to introduce the distinction between the contexts of justification and discovery—discovery is a psychological process internal to an individual scientist and at least potentially touched by genius; justification is a communal and communicative process. The famous words with which Reichenbach introduces the distinction explicitly analogize the context of justification to the public presentation of scientific knowledge (Reichenbach 1938, pp. 6–7): 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 way, for instance, in which a mathematician publishes a new demonstration, or a physicist his logical reasoning in the foundation of a new theory, would almost correspond to our concept of rational reconstruction; and the well-known difference between the thinker’s way of finding his 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 concerned with constructing the context of justification.

There are many places that one might with profit pause over in this passage, but one aspect of the passage that has received little attention is the last phrase of the last sentence: why does Reichenbach say not that epistemology is concerned only with the context of justification but, rather, that epistemology is “only concerned with constructing the context of justification”? The simplest answer, the one most clearly expressed in passages following this one in Reichenbach’s text, is simply that even scientific language is inexact and imprecise and only philosophers hold the tools needed to eliminate such imprecision. Thus, in the actual practice of science, scientific publication is the closest analogue to a fully rationalized discourse of science, but even that requires a bit of rational reconstruction. This simple story is, for the overall significance of the context distinction, only the less interesting half of the story, however. The introduction of logically precise terms is, in the envisioned process of fully rationalizing science, itself a social process that first creates the conditions of rationality even within science itself. There must be a “public,” a community of agents all speaking the same language, before the context of justification is even possible. By drawing attention to both the need for clear meanings and the role of social decision in making scientific knowledge possible, Reichenbach’s account of epistemology gives scientific philosophy a social function and, ultimately, a political significance. Scientific philosophy is meant to help create the conditions for a rationalized society within which justification finally makes clear and explicit sense.2 Reichenbach’s book was, I shall argue, an intervention in the social order in virtue of being an intervention in the epistemic order, given his account of the proper tasks of epistemology. I will argue for this through attention to some of the ways in which Reichenbach uses motivating terms in the book. Ultimately, Reichenbach is engaged in an effort common among the scientific philosophers we associate with logical empiricism—the effort to construct on scientific grounds the epistemic space in which beliefs can be freely adopted but rationally controlled. This is a delicate balancing act. On the one side are irrationalists and even unbridled conventionalists who threaten

FREEDOM IN A SCIENTIFIC SOCIETY

43

to make all knowledge-making into free flights of imagination; on the other side are inductivists and epistemic determinists of various stripes who threaten to make our epistemic life a matter of machine-like rule-following or the iron laws of history. The problem of knowledge for the logical empiricists is the problem of how to allow for the freedom of thought and yet enforce proper concern for epistemic responsibility. This problem becomes almost identical with the problem of liberal democratic tolerance, of course—as well it might since the need for a rationalized international socialist democratic society was, the logical empiricists agreed, the only real option to fascism in Europe in the 1930s.3 TEXTUAL EVIDENCE IN EXPERIENCE AND PREDICTION

Some aspects of my thesis are fairly obvious in the text. Reichenbach’s book is very much dedicated to stressing the uncertainty of all claims to knowledge while holding out the possibility of knowledge. Indeed, he suggests in the Preface that the term “logical empiricism” should give way to “probabilistic empiricism” (Reichenbach 1938, viii). The key to Reichenbach’s probabilistic empiricism is the idea that knowledge is for the prediction of the future course of experience and for the possible control of the future course of experience. It is for this reason that the problems of the interpretation of probability and the justification of induction are central to the work. Moreover, the “pragmatic justification of induction” is quite explicitly to be contrasted with any “inductivism” that finds an algorithmic inductive logic as the key to the establishment of certain scientific truth. Indeed, the pragmatic justification of induction leaves open the possibility that inductive inference does not succeed in predicting the future course of experience—the world might turn out that way. The fragility of knowledge and the need for a creative freedom of the scientist that is consistent with epistemic responsibility in an uncertain world are main themes of the work. While all of this may be clear, the claim that Reichenbach is trying to make a point at the interface of epistemology and social philosophy might still be seen as dubious. I admit that the textual evidence in the book is more suggestive than probative. Nevertheless, there are several aspects of Reichenbach’s view that suggest lessons that he elsewhere takes to be social and political lessons. The key to the connection is the various ways Reichenbach invokes the freedom of the scientist while demanding rational control as a necessary condition of knowledge. The first place in the text where this theme is rehearsed is right at the front, when Reichenbach stresses the ineliminable role of decision in science even as he limits the scope of it beyond what he finds in “extreme conventionalism” (Reichenbach 1938, p. 15). The first important distinction Reichenbach makes within epistemology is between free decisions within the knowledge system and true-or-false claims made once the appropriate decisions have been made. This is, of course, where Reichenbach imports his work regarding conventions in space-time theories and broadens it into an account of conventions throughout the structure of knowledge. But Reichenbach is at pains even in these early sections to both deepen and limit the significance of free decisions in knowledge. He performs both of these tasks by distinguishing between

44

ALAN RICHARDSON

conventions and “volitional bifurcations” in science. The latter are introduced with these words (Reichenbach 1938, p. 10): There are decisions of another character which do not lead to equivalent conceptions but to divergent systems; they may be called volitional bifurcations. Whereas a convention may be compared to a choice between different ways leading to the same place, the volitional bifurcation resembles a bifurcation of ways which will never meet again.

A volitional bifurcation marks a place where the decisions of scientists or the societies in which they live have permanent consequences for what can be known. Significantly, Reichenbach’s first example of such a volitional decision in science is one regarding the very aim of science itself. Moreover, he defends a non-standard decision as to a goal of science with a claim about practical rights (Reichenbach 1938, p. 10): What is the purpose of scientific inquiry? This is, logically speaking, a question not of truth-character but of volitional decision, and the decision determined by the answer to this question belongs to the bifurcation type. If anyone tells us that he studies science for his pleasure and to fill his hours of leisure, we cannot raise the objection that this is “a false statement”—it is no statement at all but a decision, and everyone has the right to do what he wants.

The very goal of the pursuit of knowledge can only be grounded in epistemic freedom, guaranteed by personal rights. Volitional bifurcations, on the other hand, according to Reichenbach, limit the freedom within science through the notion of “entailed decisions” (Reichenbach 1939, p. 13): The system of knowledge is interconnected in such a way that some decisions are bound together; one decision, then, involves another, and, though we are free in choosing the first one, we are no longer free with respect to those following. We shall call the group of decisions involved by one decision its entailed decisions.

It is this notion of entailed decisions—the way decisions ramify and interact within the system of scientific knowledge—that leads Reichenbach away from the dangers of unbridled conventionalism (Reichenbach 1938, p. 15): The concept of entailed decisions, therefore, may be regarded as a dam erected against extreme conventionalism; it allows us to separate the arbitrary part of the system of knowledge from its substantial part, to distinguish the subjective and the objective part of science.

It is important to note that the notions of the “arbitrary” and the “subjective” here do not signal that these elements are harmful to or eliminable from science. Reichenbach introduces “entailed decisions” in an effort to locate and to minimize the arbitrary and subjective, but the role of subjective volition, as we have already seen, cannot be reduced to zero within science. Science is based on decisions. Unbridled conventionalism, according to Reichenbach, misunderstands the way decisions ramify, but there is a necessary element of the volitional within science. This understanding of the interconnection of the volitional is the first step not in eliminating, but in disciplining the arbitrary and subjective. Science, as a given social enterprise4 , could not proceed on the basis of each individual making his or

FREEDOM IN A SCIENTIFIC SOCIETY

45

her own choices, unconstrained by the need for co-ordination and control. It is of interest that this sort of intersubjective control is, in fact, the language within which Reichenbach demarcates his scientific philosophy from the concerns of traditional metaphysics. For example, in section 26 and following, Reichenbach discusses the scientifically acceptable (by his lights) understanding of the problem of other minds. He argues that there is no introspective access to the operations of “psychical” phenomena. He notes that this claim is disputed in metaphysics, which claims each of us has privileged access to his or her own psychical phenomena (Reichenbach 1938, pp. 227–228): It is a current opinion among philosophers that what we have said is valid only for our observations of other persons, as we cannot share their psychical life, but that for our own person there is another means of observation, a direct view into our internal life. This distinction is one of the profound misunderstandings on which the traditional metaphysics is based. To clarify this question, let us enter into an analysis of the difference between our own personality and other personalities. There is, of course, a specific difference; but it is not of the type assumed by traditional philosophy.

What is notable for our purposes in this discussion is how he goes on to speak of behavioristic psychology as involving the “control” of behaviors. Thus, he considers the warrant for the claim that someone else sees (properly, i.e., as three-dimensional) a stereoscopic image. The pertinent question, according to Reichenbach is (1938, pp. 229–230): Now let us see how we control the statement that another person has the stereoscopic image. That the person is looking through the stereoscope is not a sufficient reason to believe that he has the impression. We control it by his reactions . . . . When the stereoscopic effect occurs, almost every person, especially if untrained, shows a sudden expression of joy and surprise, by an exclamation or a smile. This reaction, in combination with the other ones, is a very good indicator.

Throughout his entire discussion of the virtues of behavioristic psychology, Reichenbach returns again and again to the notion of control. Behaviorism, in sticking to a physicalist language of stimulus and reaction, succeeds where the vague, metaphysical, introspective psychology fails (1938, p. 241): It is the advantage of behaviorism that an objective language is obtained which can be controlled by everybody; reports of the person are not needed, and the method is applicable to animals as well as men.

Where he objects to behaviorism, it is in the confusion of this point about language with a point about method: the objective, physicalist language of behaviorism does not obviate the usefulness of self-observation as a method in psychology; here, too, the language is the language of control (1938, p. 243): “The method of self-observation is, I think, a necessary element of psychology; it is to be controlled but not to be dropped.” The point in stressing the use of the notion of “control” here is to stress the other side of the problem of dependence of knowledge upon the will: the will must be disciplined. “Control,” for Reichenbach, involves the intersubjective co-ordination of the scientific will and the checking of claims against the world and with one another. The point of

46

ALAN RICHARDSON

the context of justification and the intersubjectivity of scientific knowledge is that they allow such checking. “Control” seems to have been chosen because of its relations to the word “kontrollieren” of Reichenbach’s native German; this word suggests less domination by a superior power (putting someone under one’s control) and more the checking of equipment or of procedure (as in “control group”). Indeed, Reichenbach uses “control” in this way quite explicitly in discussing unanticipated experimental results in physics—one checks one’s accounts of abstractions and inferred concrete objects (his illata) by checking them against independently achievable alterations in observable and manipulable concrete objects, such as wires and batteries (1938, pp. 274–275): Imagine an engineer who discovers a new effect in a vacuum tube, say, a sudden rise of anodic current when a certain pressure of a specific gas is poured into the tube. At first he will not believe in this physical interpretation of his experience. He will look over his wires, batteries, and screws to ascertain whether the concreta basis of his inferences is unchanged. He will then control his instruments and his set by replacing his tube by another tube of known effects; he thus determines whether his concreta basis leads to usual concrete effects if it is used in a normal way. He connects in this way the observed fact with a wider concreta basis. Whoever takes part in practical work with abstracta and illata—and almost every branch of higher engineering is occupied with such things—will know that this return to the concreta basis is used as the only decisive method of control.

So, I am not suggesting a coercive political agenda in Reichenbach’s work; something like “science inevitably seeks political control.” Rather, I am suggesting an interesting connection that is quite necessary given Reichenbach’s claims about scientific freedom: science will achieve objectivity, despite the necessary expression of scientific freedom, through a demand for epistemic control; science seeks claims that can be checked against the world and which epistemic agents can agree upon (so we can check one another). This politically tinged language is the natural language of expression in Reichenbach’s scientific epistemology. Reichenbach’s logical empiricism stressed both a human right to epistemic freedom and the need for exercise of the will in the pursuit of knowledge; this had to be counter-balanced with a source of epistemic responsibility in the control of claims to know. The construction of the context of justification, then, is the source of epistemic responsibility of science and the social responsibility of the scientific philosopher.

METAPHYSICS AND SCIENCE IN NEURATH AND REICHENBACH

“Control” in the sense at issue is obviously close kin to “verifiability” and, even more so, to “testability.” This is not a sense of “control” unique among logical empiricists to Reichenbach, moreover. Otto Neurath also used the word in this way (Neurath [1931] 1983, p. 48): What do I mean by a positive statement, and how can I test it? A statement that cannot be controlled is a thesis devoid of sense. Those who thus succeed in formulating a system of laws which they apply in predicting events were best regarded as ‘representatives of a scientific conception of the world’.

FREEDOM IN A SCIENTIFIC SOCIETY

47

Indeed, Neurath was as likely to use the notion of “rationally uncontrolled assertion” as “unverifiable sentence” is discussing the nature of metaphysics.5 This allows Neurath to avoid the pseudo-rationalism (another of his favored terms) of a belief any general method of verification in favor of a fallible procedure of check and control. More than this, however, the term “uncontrolled” as used by Neurath and Reichenbach suggests a danger in metaphysics. Metaphysical claims lack all connection to empirical claims. Belief in such claims, thus, does not help and may actually hinder ability to act in the world of physical things. Metaphysicians, therefore, consciously or unconsciously, offer stories that might decrease agents’ abilities constructively to act in the world. Thus, for example, one of the main lessons of Neurath’s Anti-Spengler is exactly that belief in the Spenglerian story of the decline of western civilization decreases one’s ability to act in the world: dedicated to “youth and the future they shape,” the work ends with a plea to “young people who take life seriously” to “advance to strong constructive activity” (Neurath [1921] 1973, pp. 158, 213). Similarly, pseudo-rationalist metaphysics, which offers reasons why certain procedures must lead to knowledge, decreases the number of options one seriously considers through a misleading story of rational inevitability. It is unusual to place Neurath and Reichenbach side-by-side when discussing logical empiricism; Neurath is the leader of the “left Vienna Circle” whereas Reichenbach, while not a Circle member at all, is often firmly associated with views of the “right” Circle.6 I suggest that there is more to the connection between Neurath and Reichenbach than has sometimes been allowed. In 1938, Reichenbach is firmly within the physicalist wing of logical empiricism and his views on “control” illuminate the twin resources of epistemic control available within the physicalist phase of logical empiricism: On the one hand, on Reichenbach’s view individual concrete material objects simply presented themselves to us in our experience; he calls this the “peremptory character of immediate things” (1938, p. 275). These things are beyond our willful control and present the material conditions of knowledge; they are the things in experience whose brute existence nothing worthy of being called “knowledge” could ignore. On the other hand, communal volitional decisions as to the meanings of words provide the semantic control over the languages we speak. Our joint decision to speak the same language and to hold one another to its requirements provided the grounds upon which to object to deviant, deceptive, or meaningless speech. As it did for Neurath, Reichenbach’s interest in “control” finds expression also in the tight connection that he makes—if only by decision—between knowledge and prediction. Prediction is all about control of experience aided by being able reliably to foretell the consequences of natural and social acts. Indeed, Reichenbach’s pragmatic justification of induction—perhaps the most famous portion of the book—makes explicit the way in which Reichenbach sees reliable action as the point of knowledgeseeking (1938, p. 346): Inductive inference cannot be dispensed with because we need it for the purpose of action. To deem the inductive assumption unworthy of the assent of a philosopher, to keep a distinguished reserve, and to meet with a condescending smile the attempts of other people to bridge the gap between experience and prediction is cheap self-deceit; at the very moment when the apostles of such a higher philosophy

48

ALAN RICHARDSON leave the field of theoretical discussion and pass to the simplest actions of daily life, they follow the inductive principle just as surely as does every earth-bound mind.

Indeed, the inductive principle is a principle of action: inferring in accordance with the assumption that there is a limit of frequency is acting and, if Reichenbach’s justification of induction works, is a way of acting epistemically that leads to reliable action generally, if there is reliable action at all.

EPISTEMIC CONTROL AND THE REJECTION OF A PRIORI KNOWLEDGE IN REICHENBACH

It is worth recalling the form of the pragmatic justification of induction in Reichenbach’s text: The argument says that assuming that there is a determinate limit of frequency and inferring from observed frequencies to the limit frequency, adjusting as needed, will lead (eventually) to the correct limit of frequency if there is a limit at all. Thus, one is warranted in so inferring, regardless of whether there are frequency limits (laws of nature, one might say), since the point of knowing is reliable prediction and this process leads to reliable prediction if prediction can be reliable at all. Now back in his dissertation in 1915, Reichenbach had provided what he thought of as a transcendental deduction of the need for probability judgments if the world was to be objectively represented at all. The pragmatic justification of induction seems, on the face of it, to be a similarly transcendental argument, with a somewhat weaker conclusion. Reichenbach’s pragmatic argument has a tint of Kantianism about it; it sounds almost as if he is saying that the inductive principle is a necessary precondition of knowledge. Reichenbach himself noted this tone and sought to argue against reading too much Kantianism into his position in 1938. For our purposes here, what is most interesting to note are the terms in which Reichenbach distances himself from a Kantian reading of the argument (1938, p. 360): There might be raised, instinctively, an objection against our theory of induction: that there appears some thing like “a necessary condition of knowledge”—a concept which is accompanied since Kant’s theory of knowledge with an unpleasant flavor. In our theory, however, this quality of the inductive principle does not spring from any a priori qualities of human reason but has its origin in other sources. He who wants something must say what he wants; he who wants to predict must say what he understands by predicting.

This is quite informal talk but it is suggestive of the key transformation between the Kantian and the probabilistic empiricist position, as Reichenbach saw it. On the one hand, we have the familiar move to meaning: what is at stake is the meaning of “prediction.” More importantly, however, in my view is the phrase “he who wants something must say what he wants.” The key metaepistemological move in Reichenbach’s rejection of Kantianism is the movement from reason to will. The traditional a priori is the constraints on knowledge offered by the very nature of reason; the necessary conditions for knowledge for Reichenbach, on the other hand, are the changing desires and interests that form the decisions needed for science.

FREEDOM IN A SCIENTIFIC SOCIETY

49

The key lies in the metaepistemological point of Reichenbach’s epistemic voluntarism. We no longer have an a priori guarantee in transcendental philosophy that knowledge is possible at all. We do need to determine what we want of knowledge and, given our choices regarding the goal of knowledge, we can determine whether knowledge in that sense is possible at all and under what conditions. We have here a sort of voluntary a priori—a willful, existential Kantianism. Of course, Reichenbach holds no truck with romantic individualism, so this existential Kantianism must, if it is to help construct the context of justification, become a matter of communal, not individual, choice.7 These are the terms in which Reichenbach expresses the import of the “formalistic conception of logic” (1938, p. 334f). The formalistic conception contrasts with a traditional “a prioristic interpretation” of logic, that sought to subject us to the commands of the mind (1938, p. 334): For the first interpretation, which we may call the aprioristic interpretation, logic is a science with its own authority, whether it is founded in the a priori nature of reason, or in the psychological nature of thought, or in intellectual intuition or evidence—philosophers have provided us with many such phrases, the task of which is to express that we simply have to submit to logic as to a kind of superior command.

The formalistic interpretation of logic frees us from this superior command by recognizing the necessities of logic as the entailed necessities of our choices of linguistic forms: we subject ourselves to the consequences of our own choices, nothing more and nothing else. As Reichenbach says (1938, p. 336) “logical necessity . . . is nothing but a relation between symbols due to the rules of language.” These rules are, moreover, our rules; within a rationalized context of justification, we will have chosen these rules quite explicitly. This epistemic voluntarism is Reichenbach’s general weapon against a priorism in all its forms. In late work, such as his (1951) Rise of Scientific Philosophy, Reichenbach deploys this voluntarism in his discussion of the nature of ethics, insisting that if we properly see the logical type of the ethical imperative, we shall see that such imperatives simply express volitions. On this basis, he sets up a second-order volition that he dubs “the democratic principle”: “Everybody is entitled to set up his own moral imperatives and to demand that everyone follow these imperatives” (1951, p. 295). What is the nature of this principle? It is not a dictate of reason; it is a freely adopted principle of rationalized social life (1951, p. 296): I do not derive my principle from pure reason. I do not present it as the result of a philosophy. I merely formulate a principle which is at the basis of all political life in democratic countries, knowing that in adhering to it I reveal myself as a product of my time. But I have found that the principle offers me opportunity to propagate and, in large measure, to follow my volitions: therefore I make it my moral imperative.

The language to this point in Reichenbach’s discussion has been robustly individualistic but Reichenbach does not wish to be seen as a radical individualist—indeed, his democratic principle is opposed to an “anarchist principle” that says simply “everyone has a right to do what he wants” (1951, p. 294). Part of the argument against this

50

ALAN RICHARDSON

anarchic individualism is an obscure argument about the difference between rights to act and rights to demand (1951, p. 295f). A more promising line of argument, given Reichenbach’s metaethical stance is provided by a sort of pragmatic argument based on a psychological principle of harmonizing effects of living within groups whose individuals have opposing volitions (1951, p. 297): Whoever wants to study ethics, therefore, should not go to the philosopher; he should go where moral issues are fought out. He should live in the community of a group where life is made vivid by competing volitions . . . . There he will experience what it means to set his volition against that of other persons and what it means to adjust oneself to group will . . . . The exponent of individualism is shortsighted when he overlooks the volitional satisfaction which accrues to belonging to a group.

Lest there be any doubt in the minds of the reader of his text that this argument is meant to recall the pragmatic justification of induction, Reichenbach ends the argument is this way (1951, p. 301): We try to pursue our own volitional ends, not with the fanaticism of the prophet of an absolute truth, but with the firmness of the man who trusts his own will. We do not know that we shall reach our aim. Like the problem of a prediction of the future, the problem of moral action cannot be solved by the construction of rules that guarantee success. There are no such rules.

Lest it seem that I am making too much of this metaepistemological reorientation from reason’s immutable structure to the framing role of changeable volitions, it is interesting to see that in 1928, in a popular work, Reichenbach drew this inference explicitly ([1928] 1978b, p. 244): Rational knowledge in our sense is not tantamount to categorization within the pre-established cubbyholes of a reason that governs a priori, but simply amounts to unconditional faith in the power of the human capacity for knowledge—within the framework of a critique of its own goals. Thus the rational element is itself subject to change; and it emerges with increasing clarity that the basic stance of science is a faith more akin to an instinct than to rational insight, to will than to knowledge. Thus the will, the tenacious, malleable, indefatigable, and yet, eternally modifiable will is probably the basic element that truly represents the world view standing behind the scientific investigation of nature.

SUMMARY AND CODA

I have argued that the need within a scientific culture to balance the freedom necessary for scientific thought against the need for epistemic responsibility is the larger agenda behind Reichenbach’s famous distinction between the contexts of justification and of discovery. In constructing the context of justification, scientific epistemology serves a social function; it exposes and institutes the conditions of epistemic responsibility. I have argued for this through an effort to excavate the terms within which Reichenbach expresses his epistemological project, especially the notions of volition and control. This project has much to recommend it beyond the narrow confines of interpreting Experience and Prediction, however. It helps illuminate two issues in the Reichenbach literature precisely by bringing them together. The first issue is a standing issue within the interpretation of Reichenbach’s philosophy of the significance of his acceptance

FREEDOM IN A SCIENTIFIC SOCIETY

51

of Schlick’s term “convention” over his own earlier use of a neo-Kantian “a priori.” Some, like Michael Friedman (Friedman 2001), have downplayed this move as primarily terminological: the key to understanding convention as used by Reichenbach from the mid-1920s is to see it as playing the same role as the variable or relativized a priori of his earl work on the theory of relativity (Reichenbach [1920] 1965). Others, like Don Howard (1994), have argued that we miss a significant moment of anti-Kantianism if we downplay this shift in Reichenbach’s vocabulary. I have stressed the metaepistemological level at which Reichenbach rejects the a priori—in essence, he rejects an account of the a priori that makes the a priori the realm of the inviolable demands of reason. He so ties the a priori to this account of it, that in rejecting the account, he rejects the very notion of the a priori. Thus, I split the difference between Friedman and Howard. I agree with Friedman that if we, as philosophical interpreters, have a sensible notion of the variable a priori in hand, then we can see Reichenbach as accepting such a notion, even when he uses the term “convention” to denote it. I agree with Howard that we miss something if we do not take seriously Reichenbach’s commitment to rejecting the term “a priori” because of association between that notion and the notion of reason’s demands. The conventional is the realm of the freedom of the will, not the necessary demands of reason. Reichenbach’s metaepistemology leads him to reject the associations of the a priori with “ways we must think.” This does not solve all remaining issues but it does raise issues at the proper interpretative level, I believe: the issue is not “is there an a priori element in Reichenbach’s epistemology circa 1938?”—a question that depends on our ability to make sense of a variable a priori—but, rather, “why did Reichenbach come to associate apriorism with the necessary and inviolable demands of unchanging reason between 1921 and 1938?” (or, to put the issue in a different way, we can ask “what would Reichenbach have objected to in my phrase ‘existential Kantianism’?)8 This leads to the second issue, which is rarely if ever considered in the arguments over the first issue, but, in fact, on my view, holds the key to it. This issue is the relation of Reichenbach’s scientific epistemology to his political point of view and activities. As I have argued, the metaepistemology of Reichenbach’s Experience and Prediction can be read as suggesting a social responsibility for scientific philosophy: exposing and creating the conditions of transparent rationality of discourse. This required an acknowledgment of an ineliminable role for volition in the construction of knowledge and the concomitant need for co-ordination and control. This connects quite explicitly with the terms Reichenbach used as a student radical in the 1910s in his writings on university reform. The Free Student Movement, according to Reichenbach (Reichenbach [1913] 1978a, p. 110), “reject[ed] every authoritarian morality that wants to replace the autonomy of the individual with principles of action set forth by some external authority.” This individualism did not stand opposed to community; indeed, it formed, according to Reichenbach a proper sort of socialism (1978a, p. 110): It is incorrect to speak of a contradiction between individualism and socialism . . . . When we demand the autonomy of the individual and require at the same time that the individual grant to everyone else the same right to self-determination, we are really presenting one and the same thought from two different aspects.

52

ALAN RICHARDSON

The reason why it is the same thought is because the enemy of both is solely and wholly an external and binding constraint on the will of anyone. Reichenbach argued there were no such constraints and came increasingly to see a commitment to the a priori as expressing the view he so forcefully rejected. Just as the metaepistemology of Experience and Prediction reflects his social and ethical agenda, so too does his rejection of the a priori reflect his metaethical view that moral imperatives are volitions, not offerings of reason. Reichenbach introduces the discussion of the nature of ethics in chapter 17 of Rise of Scientific Philosophy precisely by tying his views to the rejection of an a priori element in knowledge (Reichenbach 1951, p. 276): The exposition of the second part of this book has so far been concerned with questions of knowledge; it was shown, in particular, how the synthetic a priori was eliminated in the cognitive field. The present chapter will be concerned with a similar analysis of the field of ethics. The idea of a synthetic a priori has been applied not only to knowledge but also to ethics . . . . It is the problem of the present chapter to replace the cognitive and aprioristic conception of ethics by a conception compatible with the results of scientific philosophy.

Reichenbach’s rejection of the a priori is a systematic commitment, based on the view that the a priori is irredeemably tainted with the mark of the necessary demands of reason. The rejection of this view is the primary lesson of his account of the role of volition in knowledge and in ethics. Reichenbach’s metaperspective uses the language of the faculties of the mind to argue for the primacy of will over reason. In using decision and choice as the weapon against the a priori, Reichenbach explicitly re-introduced a notion of the will back into theories of knowledge—a theory of the subjective element in knowledge becomes a theory of free and constrained choice of structures within which to represent the world.9 We best read Reichenbach not as attempting to give a theory of knowledge as simply a theory of accurate representation, but as also a theory of the conditions of choice that induce a genuine human and social responsibility for knowledge claims. Reflection on the social and epistemic situation surrounding the theorists of knowledge on the 1910s through the 1950s will help reveal why such responsibility was a key, if now forgotten, theme of epistemology at the time.

ACKNOWLEDGMENTS

Portions of this essay were presented at the conference celebrating Michael Friedman’s work at the University of South Carolina in October 2004 and as the 2005 Hans Reichenbach Lecture at UCLA. I thank my audiences at each event, especially Mary Domski, Michael Dickson, Michael Friedman, David Kaplan, and Maria Reichenbach, for comments and encouragement. I wish to thank the other contributors to this volume, and especially its editors, for useful comments on an earlier draft. My thinking about these issues has been stimulated over the years by conversations with many people, especially Michael Friedman, Gary Hardcastle, Don Howard, George Reisch, and Thomas Uebel. John Beatty served as the material precondition

FREEDOM IN A SCIENTIFIC SOCIETY

53

of the possibility of this essay. As I was writing the essay my good friend and valued colleague, Stephen Straker, passed away; I dedicate the essay to his memory.

NOTES 1. I am interested in the distinction as it appears in Reichenbach’s text and I do not wish to say much more about it than he does. If called upon to place Reichenbach into Hoyningen-Huene’s (this volume) typology, I’d say that Reichenbach was committed to versions one, three, and four, but that the key issue was how to demarcate the proper task and tools of philosophy of science, which I think is the import mainly of version four. This is why the distinction is not, in the first instance, between discovery and justification, but between the contexts of discovery and justification. 2. Thomas Uebel (Uebel 2005) has recently stressed the social aspects of Neurath’s philosophy of science, contrasting it with a less completely social picture in Reichenbach’s philosophy of science. Uebel is certainly right that the social aspect of Neurath’s philosophy is much more explicit and self-conscious; it is nonetheless there also in Reichenbach. 3. The politics of logical empiricism are now no longer news. See, for example, Stadler, 2001; Uebel 2000; Howard, 2003; Reisch, 2005, and the essays in Heidelberger and Stadler, 2003. 4. “Every theory of knowledge must start from knowledge as a given sociological fact” (Reichenbach 1938, p. 3)—this is the first sentence of Experience and Prediction. 5. Thus, according to Uebel 2004, p. 255, Neurath wrote a letter to Rudolf Carnap in which he says “I like to use the word ‘metaphysics’ when I am confronted with a view that is supported by the tendency to formulate uncontrollable assertions” (Neurath to Carnap, 29 February 1935; ASP/RC 029-09-80). Cartwright et al. 1996 offers a view (or more than one view perhaps) of what Neurath meant by “rationally uncontrolled assertions” in metaphysics. 6. On the terminology of “left” and “right” here, see Uebel 2004. For Reichenbach as “right wing,” see Howard 2003. 7. A recent commentator who has made much of Reichenbach’s voluntarism and has also sought to use resources from existentialism to rehabilitate empiricism is Bas van Fraassen (van Fraassen 2002). 8. I have attempted to say more about Reichenbach’s rejection of the a priori recently in Richardson 2005. 9. Interestingly, one contemporary of Reichenbach, C.I. Lewis (Lewis 1929) also argued for the primacy of will over reason and then, rather than reject the a priori, provided on this basis a “pragmatic conception of the a priori.”

REFERENCES Cartwright, Nancy, Cat, Jordi, Fleck, Lola, and Uebel, Thomas E. (1996), Otto Neurath: Philosophy between Science and Politics (Cambridge: Cambridge University Press). Friedman, Michael (2001), Dynamics of Reason (Stanford: CSLI Publications). Heidelberger. Michael and Stadler, Friedrich (2003), Wissenschaftsphilosophie und Politik (Vienna and New York: Springer). Howard, Don (1994), “Einstein, Kant, and the Origins of Logical Empiricism,” in W. Salmon and G. Wolters (eds.), Language, Logic, and the Structure of Scientific Theories (Pittsburgh: University of Pittsburgh Press; Konstanz: Universit¨at Konstanz), pp. 145–205. Howard, Don (2003), “Two Left Turns Make a Right: On the Curious Political Career of North American Philosophy of Science at Midcentury,” in G. L. Hardcastle and A. W. Richardson (eds.), Logical Empiricism in North America (Minneapolis: University of Minnesota Press), pp. 25–93. Lewis, C. I. (1929), Mind and the World Order (New York: Scribner). Neurath, Otto (1973), “Anti-Spengler,” in M. Neurath and R. S. Cohen (eds.), Otto Neurath: Empiricism and Sociology (Dordrecht: Reidel), pp. 158–213.

54

ALAN RICHARDSON

Neurath, Otto (1983), “Physicalism: The Philosophy of the Viennese Circle,” in R. S. Cohen and M. Neurath (eds.), Otto Neurath: Philosophical Papers, 1913–1946 (Dordrecht: Reidel), pp. 48–51. Reichenbach, Hans (1938), Experience and Prediction (Chicago: University of Chicago Press). Reichenbach, Hans (1951), The Rise of Scientific Philosophy (Berkeley: University of California Press). Reichenbach, Hans (1965), Theory of Relativity and A Priori Knowledge (Berkeley: University of California Press). Reichenbach, Hans (1978a), “The Free Student Idea: Its Unified Contents,” in M. Reichenbach and R. S. Cohen (eds.), Hans Reichenbach: Selected Writings, 1909–1953 (Dordrecht: Reidel), vol. 1, pp. 108–123. Reichenbach, Hans (1978b), “The World View of the Exact Sciences,” in M. Reichenbach and R. S. Cohen (eds.), Hans Reichenbach: Selected Writings, 1909–1953 (Dordrecht: Reidel), vol. 1, pp. 241–244. Reisch, George (2005), How the Cold War Transformed Philosophy of Science. (Cambridge: Cambridge University Press). Richardson, Alan (2005), “‘The Tenacious, Malleable, Indefatigable, and Yet, Eternally Modifiable Will’: Hans Reichenbach’s Knowing Subject,” Proceedings of the Aristotelian Society, supplementary volume 78, pp. 73–87. Stadler, Friedrich (2001), The Vienna Circle: Studies in the Origins, Development, and Influence of Logical Empiricism (Vienna and New York: Springer). Uebel, Thomas E. (2000), Vernunftkritik und Wissenschaft: Otto Neurath und der erste Wiener Kreis (Vienna and New York: Springer). Uebel, Thomas E. (2004), “Carnap, the Left Vienna Circle, and Neopositivist Antimetaphysics,” in S. Awodey and C. Klein (eds.), Carnap Brought Home: The View from Jena (Chicago: The Open Court), pp. 247–277. Uebel, Thomas E. (2005), “Epistemic Agency Naturalized: The Protocol of Testimony Acceptance,” Proceedings of the Aristotelian Society, supplementary volume 78, pp. 89–105. Van Fraassen, Bas C. (2002), The Empirical Stance (New Haven, CT: Yale University Press).

PART II Forerunners?

JUTTA SCHICKORE

A FORERUNNER?—PERHAPS, BUT NOT TO THE CONTEXT DISTINCTION. WILLIAM WHEWELL’S GERMANO-CANTABRIGIAN HISTORY OF THE FUNDAMENTAL IDEAS

INTRODUCTION

William Whewell’s philosophical work has often been considered as a “forerunner” to the distinction between the context of discovery and the context of justification, and sometimes Whewell is presented as an “early advocate” of that distinction (Losee 1979; Laudan 1980; Hoyningen-Huene 1987; Schaffer 1994; Yeo 1993). In contrast to other nineteenth-century “forerunners”, notably Duhem and the anti-psychologists (see Sch¨afer and Peckhaus, this volume), Whewell does not owe this dubious honor to the advocates of early twentieth-century Logical Empiricism. Rather, he was made a forerunner by those philosophers who have been concerned with hypothetico-deductivist approaches to science. Larry Laudan, for example, has claimed Whewell for his study of the emergence of epistemological fallibilism. According to Laudan, the link between the logic of discovery and the justification of theories was abandoned in the early nineteenth century, and it was then, that criteria for justification were found to be independent of the generation of theories. Whewell appears as one of the central figures in this development, because he held that “(1) theories can be appraised (“verified”) independently of the circumstances of their generation, and (2) such modes of appraisal, even if fallible, are more germane to the process of justification than any fallible rules of discovery would be” (Laudan 1980, p. 181). In the following, I examine to what extent Whewell can be claimed as a forerunner for the context distinction. I proceed in two steps. In the first section of my essay, I revisit the recent debates about Whewell’s alleged hypothetico-deductivism. I argue that Whewell’s position is in agreement only with a very specific aspect of the context distinction—the separation of the initial moment of generation and the process of validation—but certainly not with the additional division between scholarly disciplines that is commonly derived from that distinction. In the second section of the paper, I show that the reconstruction of Whewell’s philosophical program from the perspective of hypothetico-deductivist philosophy produces a distorted image of his project. Actually, Whewell advocated a hermeneutic and critical approach to knowledge, which centered on the history of the “fundamental ideas” of science. If we approach Whewell from this new angle, we can extract from his writings a number of inspiring contributions to a historically informed philosophy.

57 J. Schickore and F. Steinle (eds.), Revisiting Discovery and Justification, 57–77.
C 2006 Springer. Printed in the Netherlands.

58

JUTTA SCHICKORE

The Orthodox Context Distinction Those contributions to this volume that are concerned with Reichenbach’s version of the context distinction show that Reichenbach’s approach is, in fact, quite different— and considerably more sophisticated—than the formulations of the distinction that one encounters in philosophy textbooks (see Richardson and Schiemann, this volume). Much closer to the textbook versions comes Herbert Feigl’s rendering of the distinction, which is included in a paper tellingly titled The Orthodox View of Theories. This paper, published in 1970, defends the main claims of Logical Empiricism. Feigl acknowledges the numerous different renderings of the distinction. However, he maintains that if not the wording, then at least the intent of the distinction is “quite clear: It is one thing to retrace the historical origins, the psychological genesis and development, the social-political-economic conditions for the acceptance or rejection of scientific theories; and it is quite another thing to provide a logical reconstruction of the conceptual structures and of the testing of scientific theories” (Feigl 1970, p. 4). The point of Feigl’s distinction is to separate the scientists’ actual reasoning from logical reconstruction and the theory’s origin, its development and acceptance from its conceptual structure and testing. Feigl’s distinction thus comes closest to version 3 in Hoyningen-Huene’s classification (Hoyningen-Huene, this volume) As it is common among advocates of Logical Empiricism, the distinction is then combined with a distinction between scholarly disciplines (version 4 in Hoyningen-Huene’s classification). Through the separation of the two contexts, the scope of the logical reconstruction is delineated. It is only the end product, the established theory, and its relation to the facts that are considered in the reconstruction. It will be convenient to call Feigl’s version of the distinction the “orthodox” one. It is a good starting point for a discussion about potential “forerunners” because it was this type of distinction rather than the much richer and more ambiguous version proposed by Reichenbach that has been in the center of the debates in philosophy of science. This distinction, and the image of philosophy that is implied by it, has shaped our understanding of the tasks of philosophical analysis of science ever since. The principal task of philosophy of science is the reconstruction and evaluation of given theories, evidence, and the relations between the two. From the perspective of historically inclined philosophers, one of the particularly dire consequences of the orthodox distinction is that it has been instrumental in separating history and philosophy of science. The distinction implies, first, that no aspect relevant to the development of a specific theory bears on the assessment of its final state. The testing of the product alone determines its merits. Secondly, the distinction implies that the rules that govern the logical reconstruction and testing cannot be derived from (past) science. Ronald Giere made this point explicit: “If one grants that epistemology is normative, it follows that one cannot get an epistemology out of the history of science—unless one provides a philosophical account which explains how norms are based on facts” (Giere 1973, p. 290). In other words, the distinction would only be undermined if the philosopher could show just how the study of the actual acceptance of theories contributes to the normative criteria for reconstruction. An impossible task (says Giere). There can be no epistemologically significant role for history, and hence, no

A FORERUNNER?—PERHAPS, BUT NOT TO THE CONTEXT DISTINCTION

59

intimate relationship between history and philosophy of science. The best the two can get is a marriage of convenience, a more or less peaceful coexistence within the walls of a few HPS departments. One may wish to object that even fervent anti-historians like Norwood R. Hanson allowed for a more intimate relationship. Even though Hanson was convinced that “the logical relevance of history of science to philosophy of science is nil” (Hanson 1962, p. 585), he conceded that history does have a bearing on philosophy: “[T]he risk of inferring that there is thus no connection at all between the two is the risk that philosophers of science may not know what they are talking about, a verdict none of us can accept silently” (Hanson 1962, p. 586). In other words, Hanson suggested that history of science delimit the scope of philosophy of science. Only very few philosophers would want to dispute Hanson’s contention, especially if history is construed broadly, so as to include present science. Still, the suggested connection is not an “intimate relationship”. History provides only the theme, but it does not contribute to philosophy in any significant way. Hence, Giere’s challenge is not met. Hanson’s connection between history and philosophy is a somewhat lopsided companionship rather than a truly intimate relation: Philosophy’s precedence over history is left virtually untouched. To forge a truly intimate relationship, we are required to show just how history contributes to our understanding of the honorific aspects of the concept of knowledge. Or so it seems. Because if we take into account that the idea of philosophy on which we are currently operating is itself a product of the context distinction as it was introduced in the early twentieth century, we may perhaps want to consider another option. We may want to be more radical and question the limits that early twentiethcentury philosophy of science posed to philosophical analysis. It is in this spirit that Don Howard has taken up the dialogue with Neurath (see Howard, this volume). It is in this spirit also that I would like to take up the conversation with Whewell. After all, Whewell himself maintained a close relation between history and philosophy of science. His major books, the History of the Inductive Sciences from the Earliest to the Present Time and the Philosophy of the Inductive Sciences, Founded upon their History, were published in quick succession, 1837 and 1840. Whewell stressed that these projects mutually informed each other, reporting in a letter to a friend: “when I find myself, in the course of my historical researches, becoming metaphysical and transcendental, I open Book two, in which all these things fall into their places and will in the end make the most beautiful system that can be imagined” (Todhunter 1876a, p. 193). In what follows, I examine Whewell’s work with special attention to the possibility of a historically informed philosophy of science.

Happy Guesses William Whewell, born in 1794, spent his whole academic life of altogether fifty-four years at Trinity College, Cambridge, twenty-six of them as Master of Trinity. He read mathematics but pursued various additional topics on the side, occupying himself with the study of the classics, mineralogy, poetry, as well as German language and

60

JUTTA SCHICKORE

literature. After graduating second wrangler in 1816 he became first a fellow of his college. Subsequently, he was appointed professorships in mineralogy and, later, moral philosophy. Many commentators have been astounded by the bewilderingly wide range of Whewell’s writings. His first publication was a textbook on mechanics, followed by articles related to his research on the tides and crystallography. He published, among other things, tracts on liberal education, several sermons, lectures on moral philosophy, the Bridgewater Treatise on astronomy, and translations of contemporary German poetry. He contributed a number of reviews to review journals, and his History and Philosophy of the Inductive Sciences comprise several volumes. The interpretation of Whewell’s philosophy as a “forerunner” to the orthodox context distinction draws on a number of statements from the Philosophy that seem to suggest a crucial role for scientists’ bold ventures beyond the bounds of the known and familiar. Whewell wrote, for example, that “we must acknowledge [. . . ] that, speaking with strictness, an Art of discovery is not possible;—that we can give no Rules for the pursuit of truth which shall be universally and peremptorily applicable;—and that the helps which we can offer to the inquirer in such cases are limited and precarious” (Whewell [1840] 1996b, p. 483). He also claimed that “[s]cientific discovery must ever depend upon some happy thought, of which we cannot trace the origin, some fortunate cast of intellect, rising above all rules. No maxims can be given which inevitably lead to discovery” (Whewell [1840] 1996b, p. 186). Elsewhere, he stressed that “advances in knowledge are not commonly made without the previous exercise of some boldness and license in guessing. The discovery of new truths requires, undoubtedly, minds careful and scrupulous in examining what is suggested; but it requires, no less, such as are quick and fertile in suggesting” (Whewell [1840] 1996b, p. 221). But these statements must be read in the context of Whewell’s entire methodology. Whewell outlined the main methodological aspects of his philosophy of discovery in the second part of the Philosophy.1 This book provides an analysis of the nature of scientific discoveries. Roughly speaking, a scientific discovery binds together, or colligates, a set of facts by bringing them under a general conception. The colligation produces something new. It even shows the previously known facts in a new light, as it is “never the mere sum of the facts” (Whewell [1840] 1996b, p. 250). A successful colligation requires “clear and exact” facts and “distinct” ideas. These elements are not simply “given”, they are not there to be brought together. Rather, they are themselves products of scientific activity. According to Whewell, a successful colligation requires the specification of facts—for example through systematic observation, measurements and experimental manipulations of matter—and the clarification of ideas, i.e., the exposition of the definitions and axioms which are tacitly implied in a concept. Therefore, scientific activity consists of three main elements, namely, “the Decomposition and Observation of Complex Facts; the Explication of our Ideal Conceptions; and the Colligation of Elementary Facts by means of those Conceptions” (Whewell [1840] 1996b, p. 483). These steps are by no means successive. Rather, the scientists go back and forth between binding together the facts, clarifying the idea, rendering the facts more exact, and so forth.

A FORERUNNER?—PERHAPS, BUT NOT TO THE CONTEXT DISTINCTION

61

The “happy thought” or “happy guess” is one aspect of the colligation, namely, the act of conceiving the appropriate idea that binds together a set of facts. It is a central claim of Whewell’s “antithetical philosophy”2 that facts cannot be observed as such, but that observation always involves the supply of conceptions. Facts can never be combined to form new truths “except by means of some new conceptions, clear and appropriate”, and it is these new conceptions, which become manifest as happy thoughts. This “previous condition of the intellect, and not the single fact, is really the main and peculiar cause of the success. The fact is merely the occasion by which the engine of discovery is brought into play sooner or later. It is, as I have elsewhere said, only the spark which discharges a gun already loaded and pointed; and these is little propriety in speaking of such an accident as the cause why the bullet hits its mark” (Whewell [1840] 1996b, p. 189). Such an idea is a “happy thought” because there is no way of forcing anybody’s mind to produce it. There is no way of prescribing a method for making “happy” guesses. In this sense, happy guesses are accidental; in other words, the advancement of science depends on serendipity. Moreover, only very fortunate creatures can conceive happy thoughts. “Such accidents never happen to common men” (Whewell [1840] 1996b, p. 190). It is only geniuses, men like Kepler or Fresnel, who conceive of the fundamental ideas of the sciences.3 However, making a happy guess is not yet a discovery. The happy guess is a necessary but not a sufficient condition for discoveries. The main point about the (natural) philosopher’s activity is “not that he never conjectures hazardously, but that his conjectures are clearly conceived, and brought into rigid contact with the facts. He sees and compares distinctly the Ideas and the Things;—the relations of his notions to each other and to phenomena” (Whewell [1840] 1996b, p. 220). In an important sense, scientific discoveries are not accidental. Whewell stressed: “No scientific discovery can, with any justice, be considered due to accident. In whatever manner facts may be presented to the notice of a discoverer, they can never become the materials of exact knowledge, except they find his mind already provided with precise and suitable conceptions by which they may be analyzed and connected” (Whewell [1840] 1996b, p. 189). This is, in a way, already suggested by the image of the loaded gun: the happy thought is not a “wild guess” but only he whose mind is prepared to see something very specific will actually notice it.4 Moreover, the colligation based on the happy thought must be shown likely to be true on the data at hand. The criteria of testing are predictive power, simplicity, and, in addition, “consilience”, that is, a higher range of generality (broader applicability) of the theory that the actual colligation produced.5 Whewell’s philosophical methodology of discovery contains two non-anticipatable elements. First, the “happy guess”, the spell of originality that engenders a hypothesis, and secondly, the ingenuity that enables the individual to be original. The underlying idea is that the individual’s creative act of generating a novel thought is not amenable to rational reconstruction. While this specific variant of the distinction between the generation and validation of ideas—which, by the way, does not feature in Hoyningen-Huene’s classification—is perhaps not as undisputed as the “lean” distinction6 between factual and normative perspectives,7 it is certainly widely

62

JUTTA SCHICKORE

acknowledged amongst philosophers. Still, it has become clear that Whewell’s philosophy of science cannot be rendered in terms of the orthodox distinction, which opposes the logical reconstruction and the testing of theories—the realm of philosophy—with the “historical origins, the psychological genesis and development, and the sociopolitical-economic conditions for acceptance and rejection”. Whewell’s approach draws the line differently. On one side of the divide, we have “happy thoughts”, which are necessary but not sufficient for the generation of new knowledge. They manifest the rather narrowly construed moment at which new ideas first emerge in somebody’s mind. On the other side of the divide, we have collective debates by which the ideas and facts are clarified, bound together and tested. Because the colligation must be shown likely to be true on the data at hand, aspects of the theory’s “historical origin” are admitted into the domain of philosophical assessment. History is also admitted on another level. The application of the criteria of consilience requires assessment of consecutive theories in view of their increasing generality. One might want to describe this as a part of the “testing” of the happy guess. But again, this testing is very different from the models of testing that are implied in the orthodox context distinction. The context distinction opposes the justification with the “historical origins, psychological genesis and development, and the social-political-economic conditions”, while the Whewellian notion of testing relies on the historical origins and development of a theory as part of its test. Whewell’s methodology of discovery prescribes how the happy thoughts are to be integrated into the given system of facts and ideas. The testing of a theory is not merely consequential and not detached from the theory’s origins, on the contrary, it involves integrating the happy guess into a framework of known facts and evolving fundamental ideas. A colligation, if properly done, has as such justificatory force. All things considered, Whewell’s “happy thoughts” are a somewhat unhappy option for the advocates of the orthodox context distinction. Fundamental Ideas In fact, Whewell’s Philosophy challenged the disciplinary divide that emerges from the orthodox context distinction in an even more fundamental sense. The history of the sciences is an integral part of Whewell’s philosophical approach to science. The complete title of the Philosophy is Philosophy of the Inductive Sciences, Founded upon their History. This is usually taken to be an indication for the close relation between the Philosophy and Whewell’s other principal text, the History of the Inductive Sciences, which was written at roughly the same time. According to the standard interpretation, Whewell regarded the history of science as the quarry, as it were, that the philosopher was to exploit. He “saw history as a source of lessons for the best means of making further discoveries” (Yeo 1993, p. 163); he “proposed to examine the actual process of discovery in the various sciences in order to see if any patterns are displayed therein” (Losee 1979, p. 120).8 According to this interpretation, the methodology that Whewell laid out in Part II of the Philosophy is simply the generalized result of the examination of the actual progress in the sciences that the History displayed. If Whewell had really held this view, it would be very hard indeed to say in what way his project was more than a sophisticated history of science. Moreover, this

A FORERUNNER?—PERHAPS, BUT NOT TO THE CONTEXT DISTINCTION

63

interpretation is at odds with the fact that Whewell himself regarded the first part of the Philosophy—comprising more than 700 pages devoted to the history of the fundamental ideas of science—as the essence of his philosophy. In a letter to a friend, he wrote: “the Philosophy of Science, in my interpretation of the phrase, is the discussion of the Fundamental Ideas which [science] involves” (Todhunter 1876a, p. 284), and in the introduction to the Philosophy, Whewell claimed that “an exposition and discussion of the fundamental ideas of each Science may, with great propriety, be termed the PHILOSOPHY of such science” (Whewell [1840] 1996a, p. 76). I therefore suggest that the second part of the Philosophy’s title refers not so much to the History of the Inductive Sciences but rather to the history of the fundamental ideas. The project of a history of the fundamental ideas is the pivot of Whewell’s entire work. If we bring the fundamental ideas to the center of Whewell’s writings, all these things fall into their places and his seemingly scattered works become integral parts of a larger whole. To appreciate fully the nature of the fundamental ideas, we need to contextualize Whewell’s Philosophy and read it against the background of the intellectual and institutional environment in which it was conceived. This approach as such is not new. Geoffrey Cantor for instance interpreted Whewell’s History in the context of the Liberal Anglican historiography that dominated the interests of the Trinity dons in Whewell’s immediate vicinity (Cantor 1991).9 Cantor’s reading of the History can be fruitfully extended to the Philosophy. Redirecting attention from Liberal Anglican historical scholarship to hermeneutics, I argue that the Philosophy offers a hermeneutic and critical theory of the nature of knowledge that resonates with principal themes in the works of the Trinity Fellows. Against this background, it becomes clear that Whewell’s philosophy is by no means a mere “generalization” of actual patterns of past scientific activity but a critical analysis of the advancement of science, which aims to understand the past better than the historical actors themselves. To expose the hermeneutic character and critical thrust of Whewell’s Philosophy, I first examine Whewell’s analysis of the fundamental idea of the so-called “secondary mechanical sciences”. These are acoustics, optics, and “thermotics”, the sciences that deal with sound, light, and heat. I then show how the account of the secondary mechanical sciences fits into the entirety of Whewell’s endeavors and indicate the extent to which his writings had absorbed his intellectual environment. Fundamental ideas are what the mind contributes to knowledge. I have already noted that Whewell regarded the fundamental ideas as basic organizing concepts that govern the way in which the facts relevant to specific scientific fields can be ordered and structured. There is at least one fundamental idea that is particular to each science. The fundamental ideas include Space, Time (including Number), Cause (including Force and Matter), Media, Polarity, Chemical Composition and Affinity, Substance, Likeness, Means and Ends, Symmetry, and Vital Power.10 The secondary mechanical sciences are governed by the idea of the medium. Like every other science, the secondary mechanical sciences are considered twice, once in the History, once in the Philosophy. The History narrates the gradual establishment of the laws of these sciences, such as reflection, refraction, and polarization.

64

JUTTA SCHICKORE

The relevant chapter of Part I of the Philosophy exposes and discusses the fundamental idea. It begins with an explication of this idea, stating that the sciences of sound, light, and heat depended on the “Fundamental Idea of Media by means of which we perceive those qualities. Like the idea of cause, this idea of a medium is unavoidably employed, more or less distinctly, in the common, unscientific operations of the understanding; and is recognized as an express principle in the earliest speculative essays of man” (Whewell [1840] 1996a, p. 266). Starting from this explication, the chapter goes on to expose this idea, or, as Whewell put it, “to illustrate the principles on which we necessarily reason respecting the secondary qualities of bodies” (Whewell [1840] 1996a, p. 274). This sounds like a Kantian operation, and to a certain extent, it was. Whewell aimed to bring to light how the idea of the medium serves as the condition of the possibility of diverse theories and practices. However, he departed from Kant when he strongly emphasized that fundamental ideas unfold over time. They begin as vague, latent notions, which have to be made clear and distinct. The clarification is a collective enterprise, which is maintained by the scientists’ debates and practices. The chapter outlines the nature of the fundamental idea by presenting a wide range of concepts and practices that are united by the fundamental ideas and at the same time illustrate its unfolding. It traces the steps of this development, proceeding from Greek beginnings to Locke’s, Reid’s, and other philosophers’ discussions of primary and secondary qualities. The philosophical distinction between primary and secondary qualities is shown to hinge on the presupposition that colors, sounds, and tastes are effects that objects produce on us by means of the medium. According to Whewell, the next step in the unfolding of the idea was the introduction of measuring instruments and conventional scales, through which the relevant facts could be made more exact. Secondary qualities require specific kinds of devices. While we can measure length directly, by addition of extension, there is no direct measure of changes in quality, in intensity of sensation, such as the intensity of shades of red. Whewell praised Johann Heinrich Lambert for having introduced measurements of that kind. His photometry, pyrometry, and hygrometry are early examples of the reduction of secondary qualities to quantities (see Whewell [1840] 1996a, p. 309). Fraunhofer’s black lines of the solar spectrum served Whewell as another, more recent example. Here it is the sensation of color that is rendered precise: Through Fraunhofer’s discovery, “the prismatic spectrum of sunlight became, for certain purposes, an exact Chromatometer” (Whewell [1840] 1996a, p. 315). In this way, the history of the fundamental ideas establishes a two-way relation between instruments and measurements and the ideas. The introduction of instruments like photometers, chromatometers and thermometers is informed by the idea of the medium and at the same time a major contribution to its unfolding. Several commentators have noted that Whewell deployed his History and Philosophy for the purpose of advancing particular scientific views.11 This is certainly correct. In fact, Whewell’s philosophical theory of knowledge openly encourages philosophers to do so. Along with the application of novel instruments, the scientific controversy is the principal mode of advancing fundamental ideas. “[S]truggles and

A FORERUNNER?—PERHAPS, BUT NOT TO THE CONTEXT DISTINCTION

65

conflicts”, Whewell maintained, “are always requisite before the conceptions acquire that clearness which makes them fit to appear in the enunciation of scientific truth” (Whewell [1840] 1996b, p. 176). It is for this reason that the chapter on the secondary mechanical sciences traces the debates about the secondary qualities. It is for this reason also that Whewell reported the struggle over emission and wave theories. This battle was of course still ongoing when he wrote. In the 1830s, supporters of the wave theory were pushing forward their views against the corpuscular theory, which had been foremost since the days of Newton.12 Whewell regarded this controversy as vital for bringing the fundamental idea of the medium to maturity, and both in the History and in the Philosophy, he sought to advance it. He took a stance in favor of the wave theory. Taking sides with Airy, Herschel, and other proponents of the wave theory against the advocates of the emission theory (most prominently, Brewster, one of the most fervent opponents of Whewell’s Bridgewater Treatise of 1834), Whewell showed that the wave theory was the most adequate clarification of the fundamental idea of the secondary mechanical sciences.13 In Part II of the Philosophy, the message is emphasized with graphic means through the “inductive table of optics”. The table is displayed on a folding chart, which illustrates very effectively the consilience that the wave theory affords.14 In many ways, Whewell’s book delineates and exemplifies how the philosopher qua exponent of the fundamental ideas played an active part in their development. This brief outline of the fundamental idea of the secondary mechanical sciences indicates that Whewell’s Philosophy goes well beyond the recovery of actual patterns of past science in at least three respects. First, it aims to understand the past better than the historical actors themselves understood it. Secondly, it is a history of the present in the sense that the hermeneutic exposition is to show how the fundamental ideas have become entrenched in the everyday worldview. The claim is that only through the reworking of these debates, we can fully grasp our very own epistemic notions, in this case, our ideas about perceiving light, heat, and sound. Thirdly, Whewell’s Philosophy is critical in the sense that the philosopher partakes in the process of clarifying those ideas which are still imprecise. All three aspects, the hermeneutic, the presentist, and the critical, resonate strongly with the work that was pursued by the members of the so-called “Cambridge Network”.15 This circle was a loosely connected group of historians, theologians, mathematicians, and other scholars, most of them connected to Trinity College. The group members were united by a common aim, the modernization of university education and the promotion of mathematics and the sciences.16 Another outstanding feature of the group’s intellectual life is that many of its members were inspired by contemporaneous German thought. The Trinity Fellows offered, as Walter Cannon put it, “a heady combination of historical scholarship, German Idealism, and modern poetry, along with the best of modern science, and all this in a Christian context” (Cannon 1964, p. 78). This Germano-Cantabrigian scholarship17 was, to a large extent, hermeneutic and historical learning, modeled on novel German conceptions of historical study and hermeneutic analysis. Important inspiration for the circle came from the historian Barthold Georg Niebuhr and the philosopher-theologian Friedrich

66

JUTTA SCHICKORE

Schleiermacher. Two of Whewell’s fellow dons, Connop Thirlwall and Julius Hare, promulgated Niebuhr’s and Schleiermacher’s views at Cambridge. Many Trinity dons took part in the debates about what Hare later characterized as the “German notion of history”,18 viz., the endeavor “to estimate all prior systems according to their historical position in the progressive development of human thought, to show what truths it was the especial province of each to bring out, and how each fulfilled its appointed work” (Hare and Hare 1871, p. 470). The historian Niebuhr and his famous multi-volume R¨omische Geschichte played a prominent role in these debates. Niebuhr’s book, first published in 1811–1812, offered a new method of source criticism to support the author’s rather startling claim that an indigenous poetic oral tradition existed, which reflected Rome’s earliest history.19 Niebuhr aimed to devise interpretative techniques through which myths could be distinguished from those elements of folklore that related to historical reality. He placed great emphasis on the faculty of “guessing and divining” as the mental faculty that helped sort source material into evidence and useless remains. It was the close connection between poetry and history as well as the element of “divining” that many contemporaneous scholars found extremely appealing in Niebuhr’s work. Friedrich Schleiermacher faced similar methodological problems in his own field of scholarship, Evangelical theology. Starting out from New Testament exegesis, Schleiermacher aimed to develop a “general hermeneutics”, so as to provide the theoretical grounding of the technique of understanding. The counterpart of hermeneutics, which aimed at recovering a text’s meaning, is “criticism”, the art of judging the authenticity of a text. Schleiermacher stressed that both arts presupposed each other.20 Like Niebuhr, Schleiermacher introduced an element of “divination” into his system. One of the key parts of Schleiermacher’s concept of understanding was “divinatory” understanding, the “surmising” of a meaning.21 He explained that the act of understanding involves the understanding of a subject as something universal and finds the individual aspect by comparison with other subjects included under that universal. But to posit something under a universal in the first place, divination is required (cf. Schleiermacher 1998, p. 93). Schleiermacher regarded this act of divinatory understanding also as “guessing”, adding that this type of immediate understanding required “divinatory boldness” (Schn¨adelbach 1984, pp. 117–118). “We arrive nearly everywhere via a leap” (Schleiermacher 1998, p. 240). Niebuhr’s and Schleiermacher’s writings became well known in England through the multifarious activities of Julius Hare. Fluent in German and deeply immersed in German literature, philosophy, and philology from an early age,22 Hare very effectively propagated the study of things German among the Trinity students and fellows. Together with another Trinity don, Connop Thirlwall, Hare undertook the translation of Niebuhr’s work into English. His own book, which he initially wrote together with his brother,23 also did a lot to gain hermeneutic-historical thought a wider English audience. First published in 1827, the two-volume work was immensely successful. It went through many printings and editions and stayed in print until the early twentieth century. Leafing through the text, the reader finds something like a commonplace

A FORERUNNER?—PERHAPS, BUT NOT TO THE CONTEXT DISTINCTION

67

book, cluttered with aphorisms and useful observations on the conduct of a good Christian, the nature of art, the relationship between man and woman, and so forth. On closer scrutiny, she finds numerous essays on historical, literary, and philosophical topics, which reverberate most all the themes that made feelings run high at the Trinity dinner table. Turning to the cover, she catches sight of the book’s title: Guesses at Truth. A longish essay is devoted to the relation between poetry and history. Hare celebrated “that union of the poet with the philosopher, which is essential to form a perfect historian: he has the imaginative and plastic power of the first, the reflexion and discretion of the latter; and all his other faculties are, as they ought to be, under the dominion of the most penetrative practical understanding” (Hare and Hare 1827b, p. 223). The notion that in an important sense, the retrospective interpretation of the past supersedes past actors’ judgments of their own affairs, is the theme of several pieces. Following Schleiermacher’s diction that the hermeneutic task was to understand an utterance “first just as well and then better than its author” (Schleiermacher 1998, p. 23), Hare emphasized that we can understand past actors better than they could understand themselves, and that we can understand our own epistemic position only through a recovery of our history. He presented this position in graphic terms: “The last and fullest theory on any subject enables us better to fix both the positive and the relative value of all previous treatises concerning it. Only after the sun has mounted above the horizon, do we perceive the cause and nature of twilight. [. . . ]There is nothing more amusing, it is true, and little more instructive, than to follow the march of the human mind through any particular region of knowledge: but in such investigations it is well to have the map of the country according to the latest and correctest survey lying open before us, to understand the difficulties which were to be, which have been, and which still remain to be overcome, and then to examine the manner of accomplishing what has already been done. This is far better than creeping at the heels of successive discoverers, borrowing their eye-glasses, and throwing aside the improved telescopes of the present day, through fear of seeing further than they did: for in this way we shall rarely see so far; since few men have ever emptied our all the contents of their minds, at least if there was much in them, on paper, or communicated all their knowledge, still less their power and art of applying it. The civilized man may be better off than the savage, but not as a savage” (Hare and Hare 1827a, pp. 240–242).

This is worth quoting in full, because the passage contradicts very effectively the common view that the Germano-Cantabrigians proposed some kind of historical relativism, according to which the historical actors were to be understood in their own terms. Theirs was a presentist perspective, inspired and permeated by the idea of progress. Thirlwall made the connection explicit: “The progress of society is something in which the student of history has a deep personal concern. He is himself a part of that which he sees. He is carried along by the movement which he scans, and contributes in some measure to modify it by his presence. His actual position and prospects have been determined by the past, and it is only by the light of the past that he can discern their real nature and bearings” (quoted after Forbes 1952, p. 168). Hare and Thirlwall’s views match perfectly the presentist position of

68

JUTTA SCHICKORE

Whewell’s Philosophy, which traced the unfolding of the fundamental ideas from the perspective of the actual inductive sciences. It was Hare who introduced Whewell to German thought. Hare coached Whewell in the German language when both were undergraduates at Cambridge. The two were intimate friends24 and stayed in close contact even after Hare had left Cambridge to take the living of Herstmonceux in Sussex. As early as 1819, after Whewell had published his textbook of mechanics, Hare asked Whewell if he intended “to keep your history of the science for the next volume, or to publish it separately, or to keep it back to form a part of the history of mathematics from the creation of the world down to the year of the Lord 1836?” (Todhunter 1876a, p. 101) This rather prophetic remark not only shows that Whewell entertained the idea of writing a history of the sciences as early as 1819 but also that Hare was one of the people with whom he discussed his projects. Much later, Hare noted that in England it had been “Dr. Whewell”, who had applied the German notion of history to the history of science (Hare and Hare 1871, p. 470). Whewell was well acquainted with the “German metaphysicians”, as he frequently called contemporaneous German scholars, and he appreciated his friend Hare’s attempts to promulgate Schleiermacher’s ideas in England. In a letter to Hare, written in 1834, he noted that “Schleiermacher and the best of the Germans” required a specific intellectual discipline, and that the truth to be found in the writings of these men “must be taken up in the mind of some genuine Englishman and given out in a suitable form, before they will take a national hold upon us. Do this, if you can and will, and you will be an immense benefactor to England” (Todhunter 1876b, p. 196). Whewell was certainly sympathetic to Hare’s Guesses. He dedicated his four sermons on the foundations of morals to Hare with the note: “Yet to you I may say, without any doubt of receiving your assent, that this employment of Guessing at Truth, is both in itself praiseworthy and if carried on with a humble trust in the Divine Giver of all Truths, is full of deep and wide sources of practical blessing” (Whewell 1837a, iv). So one can surmise that Niebuhr and Schleiermacher’s conception of “guessing”, reinforced through Hare’s book, was a major source for Whewell’s notion of the “happy guess”, all the more so because Whewell himself drew a close connection between Germano-Cantabrigian scholarship and his conception of inductive hunches, those “instantaneous hypotheses which metaphysical writers tell us are so useful in discovery” (Todhunter 1876b, p. 268). At least the author of Whewell’s obituary in Macmillan’s Magazine found the connection between Hare and Whewell rather obvious. He observed that none of the readers of Whewell’s Philosophy “could fail to find in it much that was suggestive if not convincing, and many brilliant guesses at truth, if not clear discoveries of it” (Clark 1866, p. 547). Whewell certainly fully embraced the hermeneutic-historical approach in his Philosophy and made use of the conceptions of exposing, enunciating, and divining in several ways. After all, he explicitly noted that we have “to divine the general form of the relation by which phenomena are connected” (Whewell [1840] 1996b, p. 542,

A FORERUNNER?—PERHAPS, BUT NOT TO THE CONTEXT DISTINCTION

69

emphasis added). He stressed that there was more to the fundamental idea than meets the eye; “the Idea is disclosed but not fully revealed, imparted but not transfused, by the use we make of it in science” (Whewell [1840] 1996a, p. 70). It was the task of the history of the fundamental ideas to tease out the deep structures conditioning the stock of scientific concepts and practices, to “detect and enunciate in words the principles which are thus, perhaps silently and unconsciously, taken for granted by those who have a share in the establishment of scientific truth”, which, as Whewell noted, was “a task of some difficulty” (Whewell [1840] 1996b, pp. 182–183). And in a sense, the structure of the whole chapter on the secondary mechanical sciences can be understood as a manifestation of the hermeneutic circle. The chapter presents a wide range of concepts and practices as united by the very idea whose unfolding they illustrated in their succession. It is an integral part of hermeneutic philosophy that the business of understanding is “an infinite task” (Schleiermacher 1998, p. 23). Schleiermacher portrayed this business as a collective process, in the course of which “we more and more give each other reciprocal support, in that each provides the others with points of comparison and analogies” (quoted after Schn¨adelbach 1984, p. 118). The collective enterprise of clarification, conceived as controversy, takes center stage in Whewell’s philosophical program. Whewell repeatedly stressed that ideas were clarified by controversies. Therefore, the history of the fundamental ideas of science traced not only these ideas “as they have successively come into notice in the progress of science; the gradual developement [sic] by which they have arrived at their due purity and clearness” but also, “as a necessary part of such a history,” it provided “a view of some of the principal controversies which have taken place with regard to each portion of knowledge” (Whewell [1840] 1996a, p. 76). An essential part of such controversies is the clarification that is achieved through communication, that is, through “the desire of men to impress their opinions on others [. . . ] In trying to make others understand them, they learnt to understand themselves. Their speculations were begun in twilight, and ended in the full brilliance of day” (Whewell [1840] 1996b, p. 173). Whewell believed that the gradual clarification of an idea was at the same time a gradual entrenchment in our everyday worldview. The clearer the idea, the more common it became. According to Whewell, the idea first appears in the genius’s bold guess, which the lesser scientists take up and help unfold and develop. It then becomes the property of well-educated men before it finally gets entrenched in our everyday worldview. As such, the fundamental idea becomes a natural element of the education of children. For Whewell, all this had grave implications for the selection of appropriate topics for intellectual education. The chapter in the Philosophy that deals with intellectual education demands that only the clear ideas and the sciences that are informed by them—arithmetic, elementary geometry and mechanics—ought to be taught to the student. Those sciences that grounded in ideas that were still imprecise, such as chemistry, ought not to be part of the school or university curriculum (see Whewell [1840] 1996b, pp. 513–520). Whewell’s tract on university education, which was

70

JUTTA SCHICKORE

published in 1837 while Whewell was working on the Philosophy, outlines the same educational principles. This similarity shows that the writings on philosophy, history, and education were in fact parts of the same overall project.25 The educational tract integrates moral philosophy into the canon, which indicates that Whewell’s sermons and lectures on moral philosophy were also part and parcel of his endeavor.26 Whewell envisaged “a future period of mental culture” where not only the ideas of chemistry and geology, but even “the ideas which are the basis of sound criticism, morals, and politics,” would become “equally distinct and equally diffused” as those of arithmetic and mechanics. This, then, was “the grand and boundless vista of possible and probable intellectual refinement and civilization which the future offers” (Whewell 1837b, p. 29).27 Whewell alluded to this grand vista in the Philosophy. Concluding Part I, he stated that the analysis of the sciences leads us “to a point of view from which we have a prospect of other provinces of knowledge, in which other faculties of man are concerned besides his intellectual, other interests involved besides those of speculation. On these it does not belong to our present plan to dwell: but even such a brief glance as we have taken of the connexion of material with moral speculations may not be useless, since it may serve to show that the principles of truth which we are now laboriously collecting among the results of the physical sciences, may possibly find some application in those parts of knowledge towards which men most naturally look with deeper interest and more serious reverence” (Whewell [1840] 1996b, p. 165). The “fair and lofty temple of Truth” that Whewell had set out to erect (Whewell [1840] 1996b, p. 165) was still unfinished. Many fundamental ideas, those of politics, language, and moral science, were still insufficiently unfolded. Nevertheless, they were part of the edifice of science.28 Time and again, Whewell linked the history of the fundamental ideas to moral philosophy in his sermons and lectures on the foundation of morals and on moral philosophy.29 The preface to the sermons emphasizes that the “fundamental idea of a moral law” and the “system of Ethics” that could be constructed on it, needed to be developed (Whewell 1837a, viii, ix). The second sermon repeats the main features of the history of the fundamental ideas—the successive appearance of ideas and their clarification and general establishment through controversies—before it discusses the Greeks, who “with so acute and subtle a spirit seized all the elementary truths of the science of space, entirely erred and mistook when they would have proceeded to other portions of knowledge, which now appear to us no less clear and plain” (Whewell 1837a, p. 41). The introductory lectures on moral philosophy, which flank the year of publication of the Philosophy, convey a similar message. A the end of the first lecture of 1839, Whewell insisted that the principles of the inductive sciences had to inform moral science, too: “Inquiries into the nature of truth, the means and methods of its discovery, and the philosophy of science, even though they set out from the study of physical science, if they are at all successful, cannot fail to exercise a strong and favourable influence upon our studies with regard to moral truth, moral science, and the true philosophy of human life” (Whewell 1841, p. 28). He conceded that “ethics is

A FORERUNNER?—PERHAPS, BUT NOT TO THE CONTEXT DISTINCTION

71

practical and physics is speculative—that physics offers much certainty and ethics much perplexity” but insisted that they were two extremes on the same scale rather than an antithesis: “In physical science, the speculative element predominates conspicuously over the practical—yet still in the different sciences this predominance is less and less. In geometry it is complete; in mechanics nearly complete; in chemistry less so, for how many chemical arts are there which men practice better than they understand! In the physiological sciences the empire of speculation is less complete still; for how far are medicine, and agriculture, and rural economics, from being sciences, in any proper sense of the term! Ethics therefore, the most practical of the sciences, may still be a speculative science, in the same manner in which those occupations are. And the certainty, and the general theoretical insight, which obtain in the subjects just mentioned, are not so great, but that we may hope to attain to the same degree of theoretical truth, even in a subject in its outset so practical as Ethics” (Whewell 1841, p. 42). The quotation impressively illustrates that the notion of the fundamental idea is the key concept that integrates Whewell’s seemingly scattered writings into one single enterprise. Whewell ventured even beyond this civilizing educational project. He carved out an elite position for the critical philosopher.30 In his vista, the collective business of clarifying vague fundamental ideas becomes an integral part of the task of philosophy. The Whewellian philosopher is not only concerned with the hermeneutic recovery of the past but also assumes the role of the constructive critic, who through his participation in the sciences’ universal conversation helps remove vagueness from the fundamental ideas. Whewell’s remark that there were two ways of acquiring clear ideas, education and discussion (see Whewell [1840] 1996b, p. 506), must be understood in this sense. Through education, only those ideas that are already sufficiently clarified can be acquired. Through discussion, one helps make the other ideas clear. Long before the publication of the Philosophy, Whewell had contributed to this task with a number of reviews on science in journals such as the British Critic and the Quarterly Review. His reviews concerned general reflections on the scientific enterprise (for instance, the review of Mary Somerville’s Connexion of the Sciences) as well as newly emerging sciences, especially political economy and geology (e.g. the reviews of Jones’s Essay on the Distribution of Wealth and on Lyell’s Principles of Geology).31 In the Philosophy, these isolated critiques came together to form a comprehensive whole, which fulfilled the function of critique much more effectively.32 Ultimately, the philosopher’s role of the critic extends to philosophy itself. Whewell stressed that “without at all disparaging the value or importance of the labors of those who have previously written respecting the foundations and conditions of human knowledge, it may still be possible to add something to what they have done. The writings of all great philosophers, up to our own time, form a series which is not yet terminated” (Whewell [1840] 1996a, p. 6). Consequently, he included in his Philosophy a “[r]eview of opinions on the nature of knowledge and of the methods of seeking it” (this is the title of Book XII), that is, a critical survey of philosophical writings.33

72

JUTTA SCHICKORE

CONCLUSION

What are the implications for the original problem, Whewell’s relation to the orthodox context distinction? We can now safely conclude that Whewell was not a “forerunner” of that distinction. He did draw a distinction between the generation and validation of knowledge claims, but his “happy guesses” merely refer to the narrowly construed moment when an idea is conceived. The “historical origins” as well as the “sociological conditions” for acceptance and rejection of a theory do feature in his conception of philosophical reconstruction. By reading Whewell’s Philosophy against the background of contemporaneous Germano-Cantabrigian scholarship, I hope to have demonstrated that, and why, the common interpretation of Whewell’s approach as a forerunner to the context distinction provides a distorted image of his project and fails to grasp the essence of Whewell’s philosophy. With the fundamental ideas taking center stage, we can read Whewell’s seemingly scattered writings as integral parts of an all-encompassing project. We can then also appreciate the true nature of this project. Rather than a mere extract of the best of past science, the history of the fundamental ideas is an endeavor to understand better the principal conceptions of the history of present science. The history of the fundamental ideas sets out to show how the fundamental ideas have become entrenched in our everyday worldview. Last but not least, the history of the fundamental ideas aims to participate, via exposition and critical controversial discussion, in the virtually endless task of divining truth. By no means did Whewell’s philosophy foreshadow the disciplinary divide that the orthodox context distinction implies. On the contrary, his philosophical project reaches beyond the boundaries that early twentieth-century philosophy created. To be sure, Whewell’s Philosophy cannot provide a response to Giere’s challenge. It cannot show how norms can be extracted from facts.34 The notion of epistemology that emerges from Whewell’s approach integrates history in a different way. Whewell’s Philosophy suggests that to expose the epistemological concepts and practices that we unavoidably employ, we must rely on historical information. It is ultimately through this reworking of the past that we can fully grasp our very own epistemic notions. In this sense, the Whewellian project points to a notion of a historically informed epistemology which is different from the historical study of the sciences. The Whewellian reworking of the past brings us to a position in which we can, qua philosophers, contribute to the advancement of knowledge. We do so by contributing to the clarification of those basic organizing concepts that underlie current scientific pursuits. I am borrowing the term “organizing concept” from Ian Hacking (Hacking 2002), and my choice is deliberate. If we still wish to see Whewell as a forerunner, we should rather claim him for historicized epistemology. This suggestion may come as a surprise for all those who have dumped Whewell in the traditional philosophy of science camp, but it is much less surprising once we acknowledge the institutional and argumentative contexts of Whewell’s work.

A FORERUNNER?—PERHAPS, BUT NOT TO THE CONTEXT DISTINCTION

73

ACKNOWLEDGMENTS

I am grateful to the other contributors to this volume for their instructive comments on the penultimate draft of my article. Parts of the research for this essay were funded by a grant from the Wellcome Trust. NOTES 1. In the first and second edition of the Philosophy, the second volume presents Whewell’s methodology of science together with the “review of opinions on the nature of knowledge”, i.e., the history of the philosophy of knowledge. In the third edition of the Philosophy, these two topics are taken apart. The methodological writings form the Novum Organum Renovatum (Whewell 1858), and the history of philosophy appears under the title Philosophy of Discovery (Whewell 1860). 2. On Whewell’s antithetical philosophy, see Fisch 1991. 3. On the notion of genius in Whewell’s work and in early Victorian science in general, see Schaffer 1986 and Yeo 1993. It is not quite clear to me whether or not Whewell holds that it also needs genius to conceive of the ideal conceptions, the “lower-level” generalizations that are implied by the fundamental ideas of science. 4. Whewell’s “spark” is thus an interesting counterpart to Kuhn’s “anomalies”: both are phenomena that can be noticed only by a mind that is prepared, but while the “spark” has a synthesizing function, the anomaly will ultimately be destructive and create novelty only by discarding what is established. 5. As Laura Snyder has pointed out, this is the reason why Whewell’s philosophy is not a deductivist system. For it is essential that the happy thought be inferable from the data prior to any testing (see Snyder 1997, p. 597). Snyder thus suggests that Whewell’s position can perhaps best be described as a version of “generative justification” (Snyder 1997, p. 598; on generative justification, see Nickles 1985, esp. 197–202). Whewell himself maintained that induction and deduction are, in fact, two sides of the same coin—or, as he put it, “the two are the operation of the same mind traveling over the same ground. Deduction is a necessary part of Induction. Deduction justifies by calculation what Induction has happily guessed. Induction recognizes the ore of truth by its weight; Deduction confirms the recognition by chemical analysis. Every step of Induction must be confirmed by rigorous deductive reasoning, followed into such detail as the nature and complexity of the relations (whether of quantity or any other) render requisite. If not so justified by the supposed discoverer, it is not Induction” (Whewell [1840] 1996b, p. 258). 6. For the “lean” version of the context distinction, see Hoyningen-Huene (this volume). 7. On the one hand, this distinction leads into conceptual problems, as we would not speak of the “generation” of an idea if the idea did not at least possess a certain value or promise for future research. On the other hand, one might want to dispute the inherent individualist notion of generation and claim that the generation of ideas is a social act. 8. For similar views, see Cannon 1964, p. 85; Elkana 1984, xvi; McMullin 1990, p. 832. 9. See also Williams 1991; Yeo 1993; Sloan 2003 for expositions of Whewell’s social environment and intellectual context. 10. See Whewell [1840] 1996a, xix. Note that the sciences informed by Space and Number are pure sciences, not inductive sciences. The pure sciences “do not infer special Theories from Facts, but deduce the conditions of all theory from Ideas” (ibid., xix). As I shall explain in more detail below, there are more fundamental ideas. 11. See, e.g., Cantor 1983, esp. 3–8. 12. For the British debates on the wave theory, see Cantor 1975 and Cantor 1983. 13. Most of the critical discussion is presented in the History. For an outline, see Cantor 1991. 14. Only two “inductive tables” are included in the Philosophy. The second one is the table of Whewell’s most mature model science, astronomy.

74

JUTTA SCHICKORE

15. See Cannon 1964 on the “Cambridge Network”. 16. On the rise of mathematics at Cambridge, see Warwick 2003. 17. John Stuart Mill’s epithet for the group was “Gemano-Coleridgeans”—but as I shall show, the Trinity Dons did not derive all their knowledge about things German from Coleridge. Cf. Philip Sloan’s essay on Richard Owen’s biology. Sloan shows how much Owen relied on Whewell, who, in turn, owed not only to Coleridge but also to Schelling, Plato and Kant (Sloan 2003, p. 53). 18. On the Liberal Anglican ideas of history, see Forbes 1952. See also Cantor 1991. 19. I am drawing on Robert Preyer’s outline of Niebuhr’s approach, see Preyer 1985, pp. 46–49. Preyer exposes the links between Niebuhr’s work and the writings of Herder and Wolf. For a detailed discussion of Niebuhr’s thesis about the oral tradition of Roman history along with contemporaneous criticism, see Bridenthal 1972. 20. See Schleiermacher 1998, p. 3. 21. I am following Herbert Schn¨adelbach’s concise characterization of Schleiermacher’s hermeneutics (see Schn¨adelbach 1984, pp. 110–118). 22. On Hare, see Distad 1979. E. H. Plumptre’s introductory memoir to the 1871 edition of the Guesses at Truth acknowledges Hare’s dept to Schleiermacher; see Hare and Hare 1871, p. xxvii. 23. After Augustus Hare’s untimely death in 1834, Julius continued to issue numerous additions and revisions to the original text. 24. J. Stair-Douglas, one of Whewell’s biographers, reports “intimacy” between Whewell and Julius Hare in 1817 (Stair-Douglas 1881, p. 33). 25. In a letter to Jones of 1835, Whewell reported that he conceived of “a pamphlet about mathematics as a part of liberal education, which will all be founded on the principles of the true philosophy without my telling people more of them than is requisite to be told for the purpose” (Todhunter 1876b, p. 214). 26. In his article “Passing on the Torch . . . ” Perry Williams exposes the links between Whewell’s history, philosophy, the moral and educational writings. He draws attention to the little vignette that adorns the title page of the History, the Philosophy, and the most important writings on education and moral philosophy: a flaming torch, which is passed on from one hand to another. This image illustrates the close connection and common purpose of these works (Williams 1991, pp. 117–118). Williams argues that Whewell’s educational goal was “conceived in response to the alarming spread of utilitarianism, atheism, and radicalism” (142). 27. With respect to his alleged “deductivism”, Whewell has been portrayed as an early Popperian philosopher (see Wettersten and Agassi 1991). If the first part of my reconstruction indicates why this view might be mistaken, the second part reinforces these doubts on another level. While Popper stressed that the scientific method of falsification is informed by everyday reasoning practices, which are prior, Whewell emphasized how the everyday worldview is continuous with the fundamental ideas of science, which are prior. 28. In a letter written long before the publication of the Philosophy, Whewell intimated that as to “morals and taste” he would have to “take advantage of my own philosophy, which, as it points out that all knowledge comes by induction and that induction is guessing, allows us to publish guesses, acknowledgedly imperfect, as contributions to knowledge” (Todhunter 1876b, p. 187). But in the end, he seemed to have shied away from including them, as only the physiological and palaetiological sciences are treated. 29. The dedication to Hare that precedes the sermons explicitly refers to Whewell’s exchanges with Hare when he “still resided in our much-loved Trinity College;—when I had the delight of constant intercourse with you, and such themes were not unfamiliar to our conversation. In the main purpose of my Sermons, I know that I shall have you for my favourer, for I had you for my forerunner” (Whewell 1837a, iii). 30. Readers of Whewell have commented on his Platonic view of scientific reasoning; see, e.g., Butts 1968, p. 22. However, it seems to me that the most “Platonic” aspect of Whewell’s philosophy is the privileged position that is assigned to the persona of the philosopher-critic.

A FORERUNNER?—PERHAPS, BUT NOT TO THE CONTEXT DISTINCTION

75

31. See Whewell 1834; Whewell 1831a; Whewell 1831b. On Whewell’s early activities as a reviewer, see Yeo 1993, esp. 87–115. 32. Indeed, as Yeo notes, Whewell later abandoned his activity as a reviewer because he doubted whether the review journal was the most effective genre for his critical pursuit; see Yeo 1993, p. 112. 33. This part was later published as a separate volume, the Philosophy of Discovery. 34. At this point, Whewell may well have protested: After all, he believed that that the fundamental ideas are of divine origin. These ideas enable us to have knowledge about the world because they resemble the ideas God used to create the world. So one might say that they are a priori truths that need to be discovered. As they are completely discovered, they become necessary (see Snyder 1994, p. 796 for Whewell’s notion of necessary truths). Whewell’s theory of knowledge is normative because the fundamental ideas which unfold through science are re-discoveries of the God-given, true ones. They can be assessed in terms of their distance from these true ones. The philosophical reworking of the historical record exposes the unfolding of the fundamental ideas. It is thus a completion of their development, and an intimate relation between history and philosophy is forged in the sense that retracing the historical origins and the development of the fundamental ideas is a necessary step for advancing both the sciences and the philosophical enterprise.

REFERENCES Bridenthal, Renate (1972), “Was There a Roman Homer? Niebuhr’s Thesis and Its Crisis,” History and Theory 11: 193–213. Butts, Robert E. (1968), “Introduction”, William Whewell’s Theory of Scientific Method (Pittsburgh), pp. 3–29. Cannon, Walter F. (1964), “Scientists and Broad Churchmen: An Early Victorian Intellectual Network,” Journal of British Studies 4: 65–88. Cantor, Geoffrey N. (1975), “The Reception of the Wave Theory of Light in Britain: A Case Study Illustrating the Role of Methodology in Scientific Debate,” Historical Studies in the Physical Sciences VII: 109–132. Cantor, Geoffrey N. (1983), Optics After Newton. Theories of Light in Britain and Ireland 1704–1840 (Manchester). Cantor, Geoffrey N. (1991), “Between Rationalism and Romanticism: Whewell’s Historiography of the Inductive Sciences”, in M. Fisch and S. Schaffer (eds.), William Whewell. A Composite Portrait (Oxford: Clarendon), pp. 67–86. Clark, W. G. (1866), “William Whewell. In Memoriam,” Macmillan’s Magazine XIII: 545–552. Distad, N. Merrrill (1979), Guessing at Truth. The Life of Julius Charles Hare (1795–1855) (Shepherdstown: The Patmos Press). Elkana, Yehuda (ed.) (1984). William Whewell: Selected Writings on the History of Science, (Chicago and London: The University of Chicago Press). Feigl, Herbert (1970), “The ‘Orthodox View’ of Theories: Remarks in Defense as Well as Critique”, in M. Radner and S. Winokur (eds.), Analyses of Theories and Methods of Physics and Psychology, Minnesota Studies in the Philosophy of Science, Vol. IV (Minneapolis: University of Minnesota Press). Fisch, Menachem (1991), “Antithetical Knowledge”, in M. Fisch and S. Schaffer (eds.), William Whewell. A Composite Portrait (Oxford: Clarendon), pp. 289–309. Forbes, Duncan (1952), The Liberal Anglican Idea of History (Cambridge: Cambridge University Press). Giere, Ronald (1973), “History and Philosophy of Science: Intimate Relationship or Marriage of Convenience?,” British Journal for Philosophy of Science 24: 282–297. Hacking, Ian (2002), “Historical Ontology”, Historical Ontology (Cambridge, Mass.: Harvard University Press), pp. 1–26. Hanson, Norwood Russell (1962), “The Irrelevance of History of Science to Philosophy of Science,” The Journal of Philosophy 59: 574–586.

76

JUTTA SCHICKORE

Hare, Julius Charles and Augustus Hare (1827a), Guesses at Truth, Vol. II (London: John Taylor). Hare, Julius Charles and Augustus Hare (1827b), Guesses at Truth, Vol. I (London: John Taylor). Hare, Julius Charles and Augustus Hare (1871), Guesses at Truth, New edition (London and New York: Macmillan). Hoyningen-Huene, Paul (1987), “Context of Discovery and Context of Justification,” Studies in History and Philosophy of Science 18: 501–515. Laudan, Larry (1980), “Why Was the Logic of Discovery Abandoned?”, in T. Nickles (ed.), Scientific Discovery, Vol. I (Dordrecht: Reidel), pp. 173–183. Losee, John (1979), A Historical Introduction to the Philosophy of Science, 2nd edition (Oxford: Oxford University Press). McMullin, Ernan (1990), “The Development of Philosophy of Science 1600–1900”, in R. C. Olby et al. (eds.), Companion to the History of Modern Science (London and New York), pp. 816–837. Nickles, Thomas (1985), “Beyond Divorce: Current Status of the Discovery Debate,” Philosophy of Science 52: 177–206. Preyer, Robert O. (1985), “The Romantic Tide Reaches Trinity: Notes on the Transmission and Diffusion of New Approaches to Traditional Studies at Cambridge, 1820–1840”, in J. G. Paradis and T. Postlewait (eds.), Victorian Science and Victorian Values: Literary Perspectives (New Brunswick, NJ: Rutgers University Press), pp. 39–68. Schaffer, Simon (1986), “Scientific Discoveries and the End of Natural Philosophy,” Social Studies of Science 16: 387–420. Schaffer, Simon (1994), “Making Up Discovery”, in M. A. Boden (ed.), Dimensions of Creativity (Cambridge/Mass: MIT Press), pp. 13–51. Schleiermacher, Friedrich (1998), Hermeneutics and Criticism And Other Writings (Cambridge: Cambridge University Press). Schn¨adelbach, Herbert (1984), Philosophy in Germany, 1831–1933 (Cambridge: Cambridge University Press). Sloan, Philip (2003), “Whewell’s Philosophy of Discovery and the Archetype of the Vertebrate Skeleton: the Role of German Philosophy of Science in Richard Owen’s Biology,” Annals of Science 60: 39–61. Snyder, Laura J. (1994), “It’s All Necessarily So: William Whewell on Scientific Truth,” Studies in History and Philosophy of Science 25: 785–807. Snyder, Laura J. (1997), “Discoverers’ Induction,” Philosophy of Science 64: 580–604. Stair-Douglas, J. (1881), The Life and Selections from the Correspondence of William Whewell, D. D. (London: C. Kegan Paul & Co.). Todhunter, Isaak (1876a), William Whewell, D. D. Master of Trinity College Cambridge: An Account of his Writings with Selections from his Literary and Scientific Correspondence, Vol. I (London: Macmillan). Todhunter, Isaak (1876b), William Whewell, D. D. Master of Trinity College Cambridge: An Account of his Writings with Selections from his Literary and Scientific Correspondence, Vol. II (London: Macmillan). Warwick, Andrew (2003), Masters of Theory. Cambridge and the Rise of Mathematical Physics (Cambridge: Cambridge University Press). Wettersten, John and Joseph Agassi (1991), “Whewell’s Problematic Heritage”, in M. Fisch and S. Schaffer (eds.), William Whewell. A Composite Portrait (Oxford: Clarendon), pp. 345–369. Whewell, William (1831a), “Jones—On the Distribution of Wealth and the Sources of Taxation,” British Critic 10: 41–61. Whewell, William (1831b), “Lyell’s Principles of Geology,” British Critic 9: 180–206. Whewell, William (1834), “Mrs Somerville on the Connexion of the Sciences,” Quarterly Review 51: 54–68. Whewell, William (1837a), On the Foundations of Morals. Four sermons (Cambridge: J. & J. J. Deighton). Whewell, William (1837b), On the Principles of English University Education (London). Whewell, William (1841), Two Introductory Lectures to Two Courses of Lectures on Moral Philosophy, Delivered in 1839 and 1841 (Cambridge: John W. Parker).

A FORERUNNER?—PERHAPS, BUT NOT TO THE CONTEXT DISTINCTION

77

Whewell, William (1858), Novum Organum Renovatum (London: John W. Parker). Whewell, William (1860), On the Philosophy of Discovery (London: John W. Parker). Whewell, William (1996a), The Philosophy of the Inductive Sciences [1840], Vol. I (London: Routledge/Thoemmes). Whewell, William (1996b), The Philosophy of the Inductive Sciences [1840], Vol. II (London: Routledge/Thoemmes). Williams, Perry (1991), “Passing on the Torch: Whewell’s Philosophy and the Principles of English University Education”, in M. Fisch and S. Schaffer (eds.), William Whewell. A Composite Portrait (Oxford: Clarendon), pp. 117–147. Yeo, Richard (1993), Defining Science. William Whewell, Natural Knowledge, and Public Debate in Early Victorian Britain (Cambridge: Cambridge University Press).

¨ LOTHAR SCHAFER

AUTONOMY VERSUS DEVELOPMENT: DUHEM ON PROGRESS IN SCIENCE

For Pierre Duhem, it was not just a question of personal delight or ambition that in addition to his work as physicist he also engaged in extensive studies in philosophy and history of science.1 The integration of these diverse studies was intended to yield a better understanding of science, which, in turn, should orient the research activity of the scientist toward progress more effectively2 and provide a better method of introducing students to science. In the following, I will show in section (A) how Duhem—in order to demonstrate his conception of physics as an autonomous discipline—had to demarcate the logical structure of physics by a number of distinctions, among them the distinction between discovery and justification (DJ distinction). In sections (B) and (C), I will, then, show how Duhem brought into play two extra-logical means to bridge the gap of abstraction and yield a closer resemblance of the logical form of science to actual science. Duhem is of interest to present-day philosophy of science for two reasons: first, his combination of structural and historical analysis of physical theory exemplifies the orientation in meta-scientific research, which became eminent and paradigmatic since the sixties of the last century (Hanson, Kuhn, Lakatos, Toulmin); second, Duhem’s recourse to extra-logical means was not meant to open the road to a domain of irrationality, but rather to get hold of the rational elements in the creative developments in science. His vast historical studies are inspired by his conviction that the development of science is goal oriented and that the analysis of innovative steps can reveal a “method of discovery,” which the productive scientist is following. In this respect, Duhem is on the way to the recent rejuvenation of a logic of discovery,3 which led to diverse designs of programs for heuristic methodologies (H. A. Simon, P. Langley, G. Graßhoff). A. DEMARCATING THE LOGICAL STRUCTURE OF PHYSICAL THEORY

Duhem’s La Th´eorie Physique: Son Objet, Sa Structure4 had strong influence on the basic approach to science as formulated in the program of the Vienna Circle, in which the explication of the “logic of science” as its rational core formed a primordial tenet.5 Issues of demarcation were en vogue; they should help to get a better understanding of science, even to define science. Especially Duhem’s plea for physics as an autonomous discipline, not depending on metaphysical foundations or extra-scientific influences whatever, converged with the positivistic conception of science in Mach’s tradition.

79 J. Schickore and F. Steinle (eds.), Revisiting Discovery and Justification, 79–97.
C 2006 Springer. Printed in the Netherlands.

80

¨ LOTHAR SCHAFER

Duhem established the logical (mathematical) structure of theory as the rational core of science. His reconstruction of physics required the strict abolition of explanatory ambition (the metaphysical heritage of traditional science) and restriction to the descriptive function of physical theory. According to this, the only appraisal of physical theory that could claim to be rational consisted in the check of empirical adequacy,6 which is restricted to the (purely internal) context of justification. In spite of his positivistic conception of science, Duhem did not share the view of the irrational character of theory development. Prima facie his holistic view on science seems to support this view: neither logic nor experimental results can dictate which hypotheses the scientist should reject and which ones he should adopt to accommodate theory and empirical data. Yet, according to Duhem, the scientist is not free to choose whichever he pleases among the hypotheses available at a time. Experience and common sense will guide his mind in a particular situation, which is always a situation couched in one tradition or another. Theories of the past contain ideas that function as nuclei of the victorious theories of the future, which the “good sense” of the scientist will enable him to grasp. Duhem’s analysis of the structure of physical theory presupposes a distinction similar to the DJ distinction. The autonomy of physical science rests on the logical structure of physical theory,7 which can be explicated in terms and operations intrinsic to physics (Aim 21) and which defines the formal structure of science (context of justification). At the same time, his work is permanently directed against the separation of the two; the very project of understanding the structure and aim of science does not allow to confine the philosopher’s work to the context of justification exclusively. To understand science as a human activity, as an enterprise developing in time, requires transgressions into the context of the history of ideas and invocations of common sense, i.e., reasoning per analogiam across the borders of the domain under investigation. Duhem analyzes the different powers and operations that conjointly produce the complex entity called physical theory in much the same way in which an experimenter has to analyze and calibrate the measuring device, he wants to put at work: “Thus have we gone about the analysis of physical theory [. . . ] We have studied in successive order the mechanism of each one of the operations which go to make up a theory of physics, and have noted how each of them contributes to realizing the aim of the theory” (Aim 3). The first step of the analytical project consists in the demarcation of the analysandum. The peculiar form of reasoning as in physical theorizing has first to be discriminated from ordinary forms of reasoning—where we make use of generalizations and laws, too. Yet, according to Duhem, there is a fundamental difference between scientific way and ordinary way of experience (Aim 144–164; 259–268). In the case of everyday experience one can move without problems from individuals to general notions and vice versa. “Common sense laws” (like “all men are mortal”) can be applied spontaneously and the epistemic operations on which we do so remain unconscious. Duhem speaks of instinctive and unreflective operations that yield everyday experience.

AUTONOMY VERSUS DEVELOPMENT: DUHEM ON PROGRESS IN SCIENCE

81

These operations that are sufficient to generate ordinary experience are by no means adequate to apply scientific laws to individual cases. Because of the symbolic character of a physical law, special competence is needed to relate scientific concepts to singular cases. “A physical law is a symbolic relation whose application to concrete reality requires that a whole group of laws be known and accepted” (Aim 168). Interpretation of physical laws presupposes a special training of the faculty of judgment, otherwise, it is impossible to correlate experimental data (results of measurement) with the numerical values derived from physical laws. Therefore, scientific knowledge according to Duhem cannot be understood just as a refinement of ordinary experience or common sense but is a distinct type of knowledge owing its difference to the symbolic representation (mathematical structure) of the theories. The second demarcation concerns some variant of the DJ distinction, separating two features within physics. Duhem is working with this distinction not as classifying distinct kinds of statements, like descriptive versus normative statements, the former used to describe discoveries, the latter to justify laws on empirical grounds. Because of his pragmatic approach, Duhem takes this distinction, rather, as a distinction between different kinds of operations or activities: between acts involved in (1) the development of theories and (2) the evaluation of theories.8 He wants to discriminate between (1) acts concerning the genesis, formation, development of theories, and (2) acts of “choice of hypotheses,” leading to acceptance or rejection of the generated theories in question. Looking at science as a historian, Duhem recognizes the importance of reasoning per analogiam for theory development. “The history of physics shows us that the search for analogies between two distinct categories of phenomena has perhaps been the surest and most fruitful method of all the procedures put in play in the construction of physical theories” (Aim 95f.). Physical analogy, which Maxwell had understood as “partial resemblance between the laws of a science and the laws of another science which make one of the two sciences serve to illustrate the other,” functions as a methodologic tool in theory construction. And this all the better if high-level theories are involved. This will—given favorable circumstances—yield a more precise form of physical analogy. The analogy between electrostatic charged bodies and warm bodies is a case to illustrate the merits of analogous reasoning. Obviously the two kinds of phenomena (distribution of heat in a good conducting body and distribution of electric charges in good conducting bodies) belong to totally different kinds of physical objects. Yet, the formal structure of the theories that classify and describe the different phenomena is the same in both cases; i.e., the equations cannot be distinguished from each other in terms of algebra. “Thus, each time that he [the physicist] solves a problem about the distribution of stationary temperatures, he solves by that very fact a problem in electrostatics, and vice versa” (Aim 97). The advantage of this holomorphism between different areas is a twofold one: it provides for intellectual economy (“since it permits one to transfer immediately to one of the theories all the algebraic apparatus constructed for the other”) and it also “constitutes a method of discovery” (Aim 97). Duhem, working through the immense

82

¨ LOTHAR SCHAFER

materials of the history of cosmology from antiquity to modern times (Duhem 1906– 1913, 1913–1959, 1985), could show that the development of the theory of universal gravitation could be reconstructed by consequently applying the principle of analogy. Duhem’s methodology of science is strict anti-inductivist in its orientation. No theory can be established inductively starting from singular facts and ending with generalized laws and universal theories. His criticism of induction anticipates most of the arguments that Popper in his Logic of Scientific Discovery incorporated into his methodology of falsification. Duhem is also advocating the thesis of theory ladenness of observation. Facts cannot serve as fundamentals in science because facts can play their role only if they are interpreted in the light of some theory or other. Theories have to be granted priority over empirical data, which may confirm or disconfirm the former. Advocating physics as autonomous, Duhem’s methodology of physics is not only anti-inductivist but also antimetaphysical in its orientation. He denies the possibility to promote physical research by orienting it on “metaphysical principles” as in a general cosmology. In the Cartesian orientation of science, physics was depending on the priority of metaphysical principles from which special laws should be deduced. Yet, this version of deductivism would amount to putting physics in chains. Since metaphysical statements are irrefutable on empirical grounds, they open up a field of permanent quarrel and dispute. They therefore would import dissent into physics by destroying the very basis to progress. The metaphysical part attaches itself to the descriptive part of physics “like a parasite” (Aim 32). If we accept progress as a true aim of science, we have to cut off the metaphysical roots of physical theory as it were and establish physics as autonomous discipline.9 Autonomy, then, means independence of physical theory in two respects, independence with respect to empirical facts (no facts can dictate which theory we have to accept) and independence with respect to any metaphysical foundations (no philosophical system should be accepted as providing axioms for the deduction of physical hypotheses). What help for theory construction can the physicist expect from the side of the logician? From a logical point of view, the hypotheses that are to serve as a basis for physical theory have to satisfy three conditions: (a) a hypothesis shall not be a self-contradictory proposition; (b) the different hypotheses which are to support physics shall not contradict one another; (c) hypotheses shall be chosen in such a manner that from them taken as a whole mathematical deduction may draw consequences representing with a sufficient degree of approximation the totality of experimental laws. (Aim 220) In other words, the set of hypotheses should be locally consistent and empirically adequate. This is all the physicist has to respect, and logic leaves him unguided in a vast domain of theory choice. Duhem gives a vivid description of the epistemic situation of the physicist:

AUTONOMY VERSUS DEVELOPMENT: DUHEM ON PROGRESS IN SCIENCE

83

Well now! Confronting the physicist there extends, farther than one can see, the innumerable multitude and unordered pack of experimental laws with nothing yet available to summarize, classify, and coordinate them. He has to formulate principles whose consequences will yield a simple, clear, and orderly representation of this frightening total of observational data; but before he can judge whether the consequences of his hypotheses attain their object, before he can recognize whether they yield a methodic classification of, and a picture resembling, the experimental laws, he must constitute the entire system from his presuppositions; and when he asks logic to guide him in his difficult task, to designate which hypotheses he should choose, which he should reject, he receives merely this prescription to avoid contradiction, a prescription that is exasperating in the extreme latitude it allows to his hesitations. Can such unlimited freedom be useful to a man? (Aim 221)

Duhem asks, “Will not such freedom be the most embarrassing of all vexations?” (Aim 220). Yet this question has but a rhetorical meaning. The description above is not giving a true picture of the cognitive situation of the physicist. In reality, research does not take place in the abstract space as defined by the conditions of logic. Research activities are, rather, situated into contexts of different traditions that narrow down the range of options open to the researcher in a particular situation. In fact, traditions can exert influences of such a strong kind that the mind of the scientist is not only limited in his choices but, rather, determined by it.10 Thus history shows us that no physical theory has ever been created out of whole cloth. The formation of any physical theory has always proceeded by a series of retouchings which from almost formless first sketches have gradually led the system to more finished states; and in each of these retouchings, the free initiative of the physicist has been counseled, maintained, guided, and sometimes absolutely dictated by the most diverse circumstances, by the opinions of men as well as by what the facts teach. A physical theory is not the sudden product of a creation; it is the slow and progressive result of an evolution. (Aim 221)

Concerning the development of science, Duhem advocates continuity. Spontaneous generations, radical changes, scientific revolutions have no place in his reconstruction.11 Wherever someone diagnoses a particular episode as revolutionary, we have an inadequate assessment, which, upon closer inspection by a competent historian of science, shall be converted into a judgment of continuous development.12 Only ignorance about the early states of development can lead to the wrong assumption that a revolutionary innovation had taken place.13 The uninformed mind judges about the generation of physical theory like the child about appearance of the chick from the egg. The child sees a sudden creation where the biologist observes the last stage of a long development of the embryo inside the egg. “This continuity of tradition is not visible to the superficial observer due to the constant breaking-out of explanations which arise only to be quelled” (Aim 33). Traditions generate continuity in the development of science14 . The historian of science can unroll, as Duhem says, “the continuous tradition through which the science of each epoch is nourished by the systems of the past centuries, through which it is pregnant with the physics of the future” (Aim 270). In fact, in his concluding remarks Duhem reduces the freedom of theory choice almost to strict determinism when he says,

84

¨ LOTHAR SCHAFER Logic leaves the physicist who would like to make a choice of a hypothesis with a freedom that is almost absolute; but this absence of any guide or rule cannot embarrass him, for, in fact, the physicist does not choose the hypothesis on which he will base a theory; he does not choose it any more than a flower chooses the grain of pollen which will fertilize it; the flower contents itself with keeping its corolla wide open to the breeze or to the insect carrying the generative dust to the fruit. In like manner, the physicist is limited to opening his thought through attention and reflection to the idea which is to take seed in him without him [. . . ] It is only when the physicist begins to see clearly a new hypothesis received but not chosen by him that his free and laborious activity comes into play; [. . . ] it is not up to him to conceive a brand new idea, but it is very much up to him to develop this idea and to make it bear fruit. (Aim 256f).

However, Duhem’s intention is not to deny the creative and productive activity of the physicist. He just wants to make clear that this activity does not take place in a historical vacuum and not in individual arbitrariness, but that it is situated in concrete historical traditions that function as determining factors. This double aspect of freedom and necessity is expressed in the method of analogy. Reasoning by analogy has to start with previous knowledge. It has to rely on ideas that are familiar and have proved to be useful in a particular field of research. These ideas are, then, per analogiam carried over in a new domain. Applying familiar ideas to new domains implies usually modifications in the inherited body of knowledge; every genuine development of science does not only add new materials to former knowledge but does single out certain sections as no longer tenable. New knowledge, if new it is, will negate some part or other of the received knowledge.15 By consequently applying the method of analogy, the productive scientist is able to separate the fruitful germs of the inherited body of knowledge from the parts to be discarded as falsified thereby opening the road for future development.

B. ANALOGIES AS HEURISTIC TOOLS

To illustrate how Duhem understood the method of analogy as a powerful tool for the advancement of science, I shall sketch some of the decisive steps in the development of the theory of universal gravitation as reconstructed by the historian of science Duhem. This is at the same time his favorite example of continuous development of physical theory. No doubt, many of his assertions deserve closer analysis or have been revised by recent research in the history of science.16 But I will leave the historians’ corrections aside, as it is not my aim here to give the best account of the development of the theory of universal gravitation. Rather, I seek to expose how Duhem understood analogy as a tool for making progress in science. Quite generally, present-day historians of science are rather reluctant to pick up the thread of such a long story as he had spun and prefer projects on smaller scales. The range of his project may seem to be too ambitious and too risky. Nevertheless, it strikes me as the type of history of science that cannot be dispensed with. Indeed, interest in studies of smaller scale is inspired by the hope that they provide materials to retell the long story more accurately.

AUTONOMY VERSUS DEVELOPMENT: DUHEM ON PROGRESS IN SCIENCE

85

1. The Analogy Between Gravitational and Magnetic Attraction According to the physics of Aristotle heavy objects like stones fall down because of their internal nature. Heavy bodies are endowed with substantial forms, i.e., with a natural tendency to move by itself to the center of the universe, which is the “natural place” of heavy bodies. Light bodies like fire on the other hand move by their very nature upward, their natural place being located at the circumference of the cosmos. If objects reach their natural places, motion will stop because they have reached the perfection of their inner forms. It is only through “violent motion” (directed against the inner nature of the object) that bodies move in directions other than they are naturally inclined to move. Aristotle’s theory of motion is part of his cosmology. It depends on the representation of cosmos as a finite sphere with the earth located in the center and the fixed stars at the outer circumference. The new cosmology of the renaissance with its concept of an infinite universe (and a heliocentric planetary system) is cutting the ground from under the feet of Aristotle’s theory because in an infinite universe there is neither center nor periphery, neither up nor down. How, then, do we have to think about the tendency of heavy bodies to move downward? We do so by drawing on analogies. The first analogy makes use of the phenomenon that iron and magnetic stones tend to unite. The fall of heavy objects toward the earth is analogous to the attraction of the iron toward a magnetic stone. Like is attracted by like according to an ancient principle (Empedocles), and in consequence of this every homogeneous body has in itself a tendency to preserve its integrity. If parts of a body are removed, they tend to reunite and restore the original shape of the body. According to this analogy, gravitation is freed from the assumption of a finite cosmos and the “natural places” of the elements; however, it is still thought as a teleological process (striving for the perfection or completion of the body) and it is not universal, but depending on specific powers (natures) in the respective bodies, specific for the sun, the moon, the earth, etc. 2. The Analogy Between Attraction and the Periods of Tides The next steps to the universalization of gravity stem from another analogy: the correlation of the periods of the tides with the attractive forces of the moon and the sun. This was an idea developed and used earlier by astrologers. Galileo refused this idea just because of its odious origin and tried to show the periods of the tides as effected by the rotation of the earth. His astrological opponents argued (correctly) against it that the interval between two high tides should, then, be equal to half a sidereal day, whereas in fact it was equal to half a lunar day. Credit for the correct theory of the tides must go to the astrologers and physicians.17 They produced the new idea that the moon and the sun exert the same attractive influence on the waters of the earth, whereas earlier theories had relied on specific “sympathies” (e.g., between moon and humidity). Roberval is formulating the insights gained through the combination of the two analogies in general form. He speaks of a universal medium that is distributed all over the cosmos, between the bodies and inside of them as well, in a homogeneous way and which is responsible for the mutual attraction of all bodies in the universe. He

86

¨ LOTHAR SCHAFER

published this theory in 1644, yet not in his own name but in a book that he attributed to Aristarchus of Samos (Aim 243). However, so far nothing has been said about which law the bodies attract each other, how the attraction would vary if the distance between the bodies would change etc. A third analogy comes into play to solve this question.

3. The Analogy Between Gravitation and Propagation of Light The analogy between the influences emanating from astral bodies and the light emitted by them was well established among the astrologers and physicians when Kepler started to work with it. “Perhaps no astronomer has insisted more on this analogy than Kepler did” (Aim 245). Combined with the analogy, Kepler’s general approach to astronomy proves helpful: Kepler is the first astronomer (since Plato and Aristotle) connecting astronomical theory with dynamics. Since antiquity astronomy had been confined to geometry, better: to pure kinematics. The celestial spheres, the cycles and epicycles, were designed to describe the motions of the celestial bodies as means to calculate their positions, but they did not belong to celestial physics or cosmology. In his New Astronomy (1609), Kepler promotes just a new physics of heaven as expressed in the title: Astronomia nova aitiologetos, sive physica coelestis. His aim is to give not only a correct description of the paths of the planets but also a theory about the forces that cause the planets to move on their orbits. It is just the combination of kinematic and dynamic considerations that result in Kepler’s determination of the true orbit for Mars. According to Kepler, the sun is the source of the power which moves all the planets. In a way, he is shifting Aristotle’s “first mover” from the outer sphere of the cosmos to the center of the sun (the center of the planetary system). In the same way in which the sun is the source of light, it is also the source of moving power for the planets. Therefore, Kepler is taking the position of the sun and not the center of the planetary orbit as point of reference for his calculations. The first question he has to answer is, then, why the planets move in orbits, at the center of which we do not find the sun, as should be expected. Kepler is prompted to assume in addition to the power emanating from the sun a second one seated in the planet. Again, it is for dynamical reasons that Kepler is giving up the idea of uniform circular motion, which had been accepted as valid from Plato to Copernicus and Tycho de Brahe: when the planet is closer to the sun it should move faster, when in greater distance it should slow down. The circular form of orbit Kepler does not question at that time. The New Astronomy preserves as a true document the tentative and erroneous attempts of Kepler to bring Tycho’s data about the positions of Mars in accordance with an orbit that would meet the requirements of his dynamics.18 It is only through the analogy between the attraction and propagation of light that Kepler is led to adopt the elliptical orbit for Mars. In the same way in which the light rays from the sun illuminate the surrounding planets, the moving power of the rotating sun is carrying the planets along like the spokes of a wheel.19 When Kepler conceives the moving force (virtus motrix) in analogy with the light emanating from the sun,

AUTONOMY VERSUS DEVELOPMENT: DUHEM ON PROGRESS IN SCIENCE

87

he is well aware of the fact that he is in a difficult situation. It is well known20 that the intensity of light varies inverse to the square of the distance from the source. But Kepler’s “law of areas” (the first law of planetary motion says: the radial vector from the sun to a planet sweeps out an area proportional to the time during which the planet’s motion is observed) has taught him: “The moving power to which a planet is subjected varies inversely simply with its distance from the sun” (Aim 245). Kepler takes great efforts to overcome the discrepancy of his dynamics with the laws of light propagation in the course of which he is led to his third law (“The squares of the periods of revolution of the various planets are proportional to the cubes of the major axes of their orbits”), which provided the decisive step for the formulation of Newton’s theory of universal gravitation. It should, however, still need the efforts of Borelli, Hook, and Huygens that Newton could build the keystone of the scientific building, which again only after 20 years of waiting for experimental confirmation he dares to present to the public. Duhem sums up: The most diverse considerations and the most disparate doctrines arose in turn to make their bid for the construction of celestial mechanics; [. . . ] comparisons of weight with magnetic action, as well as the affinities between the light and the mutual actions of heavenly bodies. In the course of this long and laborious birth, we can follow the slow and gradual transformations through which the theoretical system evolved; but at no time can we see sudden and arbitrary creation of new hypotheses. (Aim 252)

What does Duhem’s lesson in the history of science teach us about science? First, we have to recognize the context: Duhem is reconstructing the historical development of the theory of universal attraction in order to substantiate his criticism of the “inductive method” that Newton allegedly had used when discovering the inverse square law. In spite of Newton’s explicit declaration that he had reached his theory without invocation of general (fictive) hypotheses just by means of “inductive generalization” of Kepler’s observational laws (i.e., propositions drawn from phenomena), Duhem is questioning Newton’s account of the discovery and replacing it by the historical reconstruction of theory formation.21 Second, hand in hand with criticizing inductivism as an inadequate base for the understanding of science, Duhem proposes new methodological insights and theses about the methods and concepts actually employed in science. The most important of it is his thesis that facts do not play the role Newton’s inductivism is assigning to them. Facts cannot dictate the scientist which hypothesis he should formulate and accept since scientific theories are empirically underdetermined as such.22 The same class of data can be represented by different laws that logically contradict each other but meet the same accuracy of measurement and are, therefore, equivalent on empirical grounds (Aim 134; 152). Third, history of science has more to offer than materials and illustrations of episodes of discovery. Studying its history will yield the true understanding of science, of science as an activity, of science as a project of continuing research. The

88

¨ LOTHAR SCHAFER

demarcation of science from ordinary, from philosophical, from religious ways of experience may yield an abstract concept of science, a definition even. But definitions are sterile and characterize a corpse instead of a living entity. If we want to study science as it really develops we have to include the analysis of the concrete situations in which the scientist is working, first of all his being embedded in certain traditions. It is history that reveals “a correct and clear view of the very complex and living organization of this science” (Aim 269). Fourth, history of science provides the best introduction to science for students, the best method to teach science. Duhem claims: “The legitimate, sure, and fruitful method of preparing a student to receive a physical hypothesis is the historical method” (Aim 268). The best way to lead a student to the level of competence for engaging in actual research consists in recapitulating the historical genesis of theories presently entertained, in giving a “summary but faithful exposition of the vicissitudes which preceded its adoption by science” (ibid). The idea behind that is that we have to think of mankind as an individual who always learns and never dies.23 Fifth and finally, history of science should not be contrasted with logical analysis in the usual sense. In other words, one should not contrast logical analysis—as explicating the rationality of science—with the study of history, which reveals the creative and irrational acts of science as developing over time. Duhem strongly advocates an alternative position: “To give the history of a physical principle is at the same time to make a logical analysis of it. The criticism of the intellectual processes that physics puts into play is related indissolubly to the exposition of the gradual evolution by which deduction perfects a theory” (Aim 269f.). History of science in Duhem’s sense clearly means “internal history,” a study of scientific episodes in the light of structural features that are typical of science. In other words, Duhem advocates an intimate connection between history and philosophy of science in much the same way in which Lakatos was to recommend it.24 Both agree in that they favor a reduced and condensed version of internalist history of science.25 Yet neither of the two sees the mutual interdependence as a strict symmetric relation between history and logic of science. While Lakatos sees a priority on the side of methodology, to demarcate what is internal to science and what external, Duhem appears to favor the other side, apparently for reasons that transcend methodology and concern the epistemic attitude of the scientist vis-a-vis his enterprise. The active scientist, be he concerned with the logical structure of his theories or with testing some hypothesis or other in order to improve upon the empirical adequacy of the system, cannot help to oscillate between dogmatism and skepticism. Only history of science can keep him from falling into the extremes. “Every time the mind of the physicist is on the point of going to some extreme, the study of history rectifies him by means of appropriate correction [. . . ] History thus maintains him in that state of perfect equilibrium in which he can soundly judge the aim and structure of physical theory” (Aim 270). The study of the history of his discipline, then, will stabilize the scientist’s attitude as a good researcher and enable him to develop an adequate understanding of science.26

AUTONOMY VERSUS DEVELOPMENT: DUHEM ON PROGRESS IN SCIENCE

89

C. THE SCIENTIST AS A JUDGE: TRAINING, EXPERIENCE, AND GOOD SENSE

As Duhem said earlier: no logic, no facts, no principles of metaphysics can direct the scientist in selecting one theory rather than another to accommodate the experimental data. To be sure, the traditions in which the scientist is brought up and trained do limit the range of the options open to him to a considerable extent. But they cannot do away with the situation of choice altogether. What can the scientist rely on in a situation, when he has to estimate whether the fit of theory and experimental results are satisfactory or not, whether he should continue to reach a better agreement by further refinements of the hypotheses presently at work or, rather, by discarding them altogether and trying to get to his aim by conceiving an alternative set of hypotheses. The scientist will rely on his personal experience. Experience in this sense is different from knowledge. It cannot be transferred through teaching. It can be gained only in personal practice. If the experienced physicist finds himself in a situation where the experimental results are in disagreement with the predictions of the theory and he has to decide which way to follow, he will, as we say, rely on his intuitions. His situation is very similar to the one of a medical doctor who meets a patient. The doctor cannot open the body and check each organ one after the other (as a watchmaker can with a watch); he has to establish his diagnosis by taking into account all the symptoms that might be relevant for determining the cause of the ailment. Only the experienced doctor (not the learned medicus) is in a good position to come up with a correct diagnosis and proposal for therapy. Only the experienced physicist will be a good judge and come up with corrections leading to progress. However, when speaking of a “good judge,” Duhem has more in mind than the experienced practitioner. The competence to judge as an expert rests on his “good sense.”27 There are two situations where Duhem draws on the faculty of good sense in physics. Both are concerned with the assessment of adequacy of certain adjustments. In the first situation, the scientist has to decide how to revise the system of hypotheses in order to optimize the fitting of prognoses and experimental findings. In the second situation, he has to evaluate whether the fit is satisfactory or dissatisfactory. To document the second case, Duhem could well have continued to refer to Kepler’s struggle for the orbit of Mars, when a difference of 8 minutes arc did destroy Kepler’s expectation to have found the true orbit.28 At the end of chapter 16 of the Astronomia nova Kepler seems to have reached an almost perfect agreement between his hypothesis (the oval orbit) and the observational data. Yet, surprisingly, chapter 19 opens with a dramatic exclaim: “Who would have thought it possible? This hypothesis [. . . ] is nonetheless false.” The hypothesis in question had to be rejected because of another attempt to demonstrate its correctness. As Kepler reports, he selected two positions for Mars, which Tycho had measured with special precision, and was shocked to learn that they did not fit. And when he tried to adjust his model to them, he did not succeed but rather found out that the observations deviated from the calculated positions up to 8 minutes arc. This would have been an acceptable deviation to astronomers

90

¨ LOTHAR SCHAFER

before Tycho, who had to work with observations of lesser accuracy. But not for Kepler: Since the divine benevolence has vouchsafed us Tycho Brahe, a most diligent observer, [. . . ] it is fitting that we with thankful mind both acknowledge and honor this benefit of God [. . . ] For, if I had thought I could ignore eight minutes of longitude, in bisecting the eccentricity I would already have made enough of a correction in the hypothesis found in chapter 16. Now, because they could not have been ignored, these eight minutes alone will have led the way to the reformation of all astronomy. (Kepler 1992, p. 286)

Kepler, trusting the accuracy of Tycho’s observations, found the deviations unacceptable and decided that his hypothesis had to be replaced by a new one, which brought him, finally, to project the elliptical orbit. As if he has this episode in mind, Duhem remarks: “The sound experimental criticism of a hypothesis is subordinated to certain moral conditions; in order to estimate correctly the agreement of a physical theory with the facts, it is not enough to be a good mathematician and skillful experimenter; one must also be an impartial and faithful judge” (Aim 218). Concerning the power of judgment (good sense) with respect to the choice of hypotheses, Duhem draws on Foucault’s well-known experiment from the history of optics that he analyzed and used earlier (Aim 186–190) to demonstrate the impossibility of an experimentum crucis in physics. Before Foucault’s experiment (1849), optical theory had presented itself as a battlefield where the adherents of the emissionist doctrine (Biot) and the proponents of the wave theory (Fresnel) were fighting each other, each side after a blow coming up with corrections to rescue their approach and rejecting the arguments of the other side. But after Foucault’s experiment, which had shown that light traveled faster in air than in water, Biot resigned and did no longer adhere to the corpuscular hypothesis. According to Duhem, this was not dictated to him by logic. “Strictly, pure logic would not have compelled him, to give it up [. . . ] but by resisting wave optics for a longer time Biot would have been lacking in good sense” (Aim 218). When the scientist decides to discard a theory and to adopt another one, he does not just follow some rule of logic or of scientific methodology but makes a decision in the usual sense of the word. He acts in a situation of uncertainty, and he may go wrong. Yet Duhem claims that at the same time, his choice is not arbitrary and irrational but guided by “good sense.” There are good reasons that support the decision, reasons which—in Pascal’s words—reason does not know; in other words, there are too many reasons that the geometric mind is unable to deal with and that “they do not impose themselves with the same implacable rigor that the prescriptions of logic do. There is something vague and uncertain about them; they do not reveal themselves at the same time with the same clarity to all minds” (Aim 217).29 Sooner or later the good sense of the scientists will end the state of indecision and clear the ground for the fruitful continuation of research. Good sense can operate under worse or better conditions. The conditions are better if scientists can adopt a self-critical attitude and allow for sound reasoning by freeing their minds from the vanities of profession.

AUTONOMY VERSUS DEVELOPMENT: DUHEM ON PROGRESS IN SCIENCE

91

Generally speaking, according to Duhem, progress in science is due to two extralogical means, the methodic use of analogies across the borders of a field of research and the faculty of good sense. The knowledge of the specialist needs support from the cultivated mind if progress is to be accomplished. In every phase of empirical science, sooner or later the situation will come up when observations and theoretical calculation will disagree. To overcome the situation of disagreement is the aim of science. Logically speaking, the design and acceptance of new hypotheses of which the scientist expects to achieve a better representation of empirical laws amount to acts of free choice.30 The experienced scientist with good sense looking back into the tradition that led science to its present state is sensitive to open his mind for metaphors and ideas that are on time and wait as it were to be developed and made to bear fruit. To support this view, Duhem can refer to simultaneous discoveries that occur independently of different research groups or individuals. The performances of the productive mind in science are in Duhem’s view quite at distance from spontaneous and free creation. The context of discovery is not without rules, although the geometric mind of normal science is in no position to grasp and apply them. The subtle mind of the scientist—guided by analogies and good sense—selects and develops ideas suggested to him through critical studies of theories of the past. To sum up, Duhem’s plea for an autonomous physics rests on the demarcation of the logical structure of physical theory, based on conventions independent of any metaphysical or inductive support. In this sense, Duhem’s conception of science “is positivist in its conclusions as well as in its origins” (Aim 275). Duhem’s antimetaphysical bias is confined to the bounds of science; it is meant to shield science from metaphysical and religious influence and apologetic demands in order to protect its autonomy. However, if we want to understand science as a goal-oriented activity; i.e., as aiming at progress toward a unified theory, two expansions become important: (1) Analysis of the historical development of theories, which means a shift of interest to and inclusion of the context of discovery and (2) adoption of a metaphysical principle, which allows us to understand science as an empirical enterprise which will in the long run approximate the order of the real world, as expressed in Duhem’s notion of a natural classification toward which physical theories will converge. The importance of the DJ distinction in Duhem’s philosophy of science is, then, limited to defending the autonomy of science and in this respect it retains its function. In contrast, the project of philosophy of science in general—the project of accounting for the aim and structure of science—requires a much broader domain of investigation, viz. inclusion of history of science, of common sense, and even of metaphysical principles. NOTES 1. In this paper, many aspects of Duhem’s philosophy of science can only be mentioned but not discussed. For a closer view I refer the reader to Sch¨afer 1974b; Diederich 1974; Jaki 1984. 2. Apart from Duhem himself, Einstein seems to be the most prominent scientist who gives credit to Duhem’s philosophy of science (esp. his holism) as helpful for his scientific work. See Hentschel 1987; Howard 1990.

92

¨ LOTHAR SCHAFER

3. To be sure, what Duhem has to offer as “method of discovery” is a far cry from the expectations connected once with the idea of an ars inveniendi. There is no algorithm that will produce new discoveries in an almost mechanical way. Yet, if we look back at the development of physical theories through Duhem’s eyes, each step of the creative scientist can be seen as guided by good reasons—even if there are involved what Whewell called “happy thoughts” (Schickore, this volume). In hindsight, discoveries can be seen as gained by rational procedures. 4. First published as a series of articles in Revue de Philosophie (1904–1905), the monograph came out in Paris 1906 (second, enlarged edition 1914). A translation into German, Ziel und Struktur der physikalischen Theorien, appeared as early as 1908 in Leipzig, prefaced by Ernst Mach (New editions Hamburg 1978, 1993). My quotations are taken from the English translation, The Aim and Structure of Physical Theories, by Philip Wiener, first published in 1954 at Princeton University Press. The page numbers in the text (indicated by “Aim (page)”) refer to the First Athenaeum Edition 1962. 5. The presence of Duhem’s ideas in the early days of the Vienna Circle (“Protocircle”) is expressed most clearly by Philip Frank (Frank 1949). Duhem’s positivism was a peculiar one. It was positivistic in so far as it was antimetaphysical and fought against the subordination of science under metaphysical principles of any kind. But Duhem was at the same time anti-inductivist and never adopted a positivism of facts, which was fundamental in Comte’s positivism and also a pet idea of neopositivism in Vienna. In fact, Duhem always thought that theories have priority over facts—a doctrine that is closer to Kantianism and was shared by Poincar´e, too. Therefore Duhem’s book could inspire in Vienna philosophers of science of different orientation: hardcore positivists/logical empiricists (Feigl, Carnap), critical rationalists (Popper), and conventionalists/holists (Neurath) (cf. Stadler 1997, for a systematic discussion of logical empiricism on the basis of French conventionalism, see Diederich 1974). 6. Agreement with experiment is the sole criterion of truth for a physical theory (Aim 21). 7. A physical theory is not an explanation. It is a system of mathematical propositions, deduced from a small number of principles, which aim to represent as simply, as completely, and as exactly as possible a set of experimental laws (Aim 19). 8. Which of the variants (or conflations of them) espoused by Hoyningen–Huene (this volume) does Duhem make use of, and which of them is he prepared to defend against criticism? Variant 2 seems to fit best, and Duhem certainly would defend the slimmed and rejuvenated version of section 6. Yet at no time would he have agreed that the DJ distinction separates what is irrational from what is rational in science or that the episodes of discovery should be of no concern for the philosopher of science. 9. In this respect Duhem is in line with tenets of Ernst Mach and the position of the “neopositivists” in the Vienna Circle (see Frank 1949, p. 15). Duhem’s antimetaphysical position seems to stem from his work as a physicist who favors general dynamics as basic theory; hence, his attack on mechanism, i.e., on the privileged status of mechanical explanations in physics (Duhem 1980). Concerning the hostile attitude toward mechanism, Duhem sees himself in line with Macquorn Rankine, Ernst Mach, and Wilhelm Ostwald—and with Henri Poincar´e, who held a more moderate position (Aim, Appendix, 317). In France, at the end of the 19th century the dogmatic positivism of Comte had been replaced by different conventionalist positions such as in the popular writings ´ of Poincar´e, Edouard le Roy, and Gaston Milhaud, which influenced Duhem considerably. As the frequent references show, Duhem is influenced profoundly and in many respects by the philosophy of Blaise Pascal: (1) Concerning the essentials of the methodology, which Blaise Pascal had developed against the foundationalist system of Descartes (Sch¨afer 1974a), (2) Concerning the distinction between the two modes of reason (esprit g´eom´etrique and esprit de finesse), (3) Concerning the skepticism with respect to metaphysical foundations of science, and (4) Concerning the status of his fideistic commitments beyond the limits of science. 10. Duhem faces a problem which is common to all methodologies that operate with free conventions (i.e., with nominal definitions instead of essential definitions) as starting points. How can we arrive at statements that have empirical content? Duhem’s answer is that in fact the scientist is not free in

AUTONOMY VERSUS DEVELOPMENT: DUHEM ON PROGRESS IN SCIENCE

11.

12.

13.

14.

93

his stipulations, but guided by traditions saturated with experiences of the past. Duhem speaks about one and the same act, which—seen from a logical point of view—is free, yet—seen in historical perspective—is almost a dictate of the epistemic situation, in which the scientist is working. Others introduced two types of decision: free stipulations and enforced decisions where the latter are meant to curb the arbitrariness of the former. So when Reichenbach introduces “entailed decision” to counterbalance the conventionalism of his “volitional bifurcations” (Richardson, this volume); so, when Fleck is pairing “active associations” and “passive associations” and defines as the goal of empirical research to maximize the latter (Fleck 1979, pp. 10, 95f.). Duhem’s continuity thesis is provocative first and foremost with respect to the origin of modern physics in the 17th century. It was common conviction that after the short period of science flourishing in Greek antiquity (6–4th century B.C.) science decayed and became stale under the dominance of the Christian faith. Only when the systems of the dark Middle Ages were overthrown by the thinkers of the Renaissance, science could be reborn—now in the form of experimental science. This clich´e of the revolutionary origin of modern science was destroyed when Duhem produced evidence from medieval manuscripts that the principles of Galileo’s theory of motion had been anticipated to a large extent by the nominalists, teaching in Paris in the 14th century. “The mathematical skill acquired from reading the ancient geometers was used by Galileo and his contemporaries to refine and develop a mechanical science whose most essential principles and propositions had been formulated during the Christian Middle Ages” (Duhem 1987, p. 339). If we want to study the origins of modern science, we have to “go back to Plato” (Duhem 1969, 5). According to Kuhn’s reconstruction (Kuhn 1962), science develops in such a way that periods of continuous growth are repeatedly interrupted with revolutionary brakes, which seems to clash with Duhem’s continuity thesis. Yet, since continuity, as Duhem uses the term, is restricted to the accumulation of “empirical laws” and does not include general theories, the two positions are not incompatible, as Ariew and Barker have shown (Ariew and Barker 1992). Without renouncing the continuity thesis, Duhem cites episodes which include fundamental change in science and sound as if he were advocating a revolutionary approach: “ [. . . ] perhaps some day by acting differently, by refusing to invoke causes of error and take recourse to corrections in order to reestablish agreement between the theoretical scheme and the fact, and by resolutely carrying out a reform among the propositions declared untouchable by common consent, he will accomplish the work of a genius who opens a new career for a theory [. . . ]. The history of physics shows us that very often [sic!] the human mind has been led to overthrow such principles completely, though they have been regarded by common consent for centuries as inviolable axioms, and to rebuild its physical theories on new hypotheses” (Aim 211f.). Principles and high-level theories are not subjected to the thesis of continuous development! Closely connected with the continuity thesis is Duhem’s conviction that the aim toward which theory development strives is what he calls a “natural classification” of experimental laws. Which is to say that theories, although they are free creations of the mind, will yield in the long run not a purely artificial system but a classification that “reflects the ontological order of nature.” This conviction, however, cannot be justified on the grounds of his methodology and is a metaphysical assumption in the strict sense—even if, as Duhem says, the working scientist feels compelled to believe that he is approximating the real order of things, whenever he succeeds in his attempts to describe different phenomena under a unified theory (Aim 24–27). Some have argued that the metaphysical assumption of natural classification (being directed against a positivist, instrumentalist, and conventionalist understanding of science) would qualify Duhem among the scientific realists (see Maiocchi 1990; Lugg 1990; Petrus 1996). However, these interpretations run counter to basic teachings of Duhem, most of all against his conception of physical theory. While scientific realists hold that empirically successful predictions allow us to believe that the theoretical entities postulated by the theory exist, nothing of this sort can be claimed on the basis of Duhem’s conception. His thesis that in the long run the classification of experimental laws will approximate a natural order entails no commitment to the existence of unobservable entities; theories cannot be interpreted realistically. In spite of the adoption of the metaphysical assumption above, Duhem remains a strict positivist, a strict

94

15.

16.

17.

18.

19.

20. 21.

¨ LOTHAR SCHAFER conventionalist, a strict instrumentalist concerning his conception of physical theory. Yet, looking at science from outside he will say that the positivist understanding of science is incomplete. “The study of the physical method cannot reveal to the physicist the reason that induces him to construct the physical theory” (Duhem 1987, p. 337). Understanding of science can and should be completed by invoking the extrascientific (metaphysical) principle: “The ideal form of physical theory is a natural classification of experimental laws” (Aim 297). Poincar´e, Le Roy, and other conventionalists are not criticized for entertaining an “exasperated conventionalism” (Maiocchi 1990, p. 390), but for being insensitive and obstinate to the need for completion of scientific methodology by metaphysical assumptions. Duhem, then, is not following a “middle way” between conventionalism and (scientific) realism, as McMullin wants it (McMullin 1990). Rather, he is at the same time methodological conventionalist and metaphysical realist. The metaphysical principle adopted by Duhem proves upon closer inspection that Duhem commits himself not just to one kind of metaphysics or another, but to the ontology of Aristotle. According to Aristotle, natural objects are ordered in a hierarchical order of species and genera. To determine what a thing is means to specify its genus proximum and differentia specifica. This is appealed to by Duhem’s “natural classification.” He justifies his option for Aristotelian metaphysics on two different levels: (1) Aristotelian physics (dynamics) incorporates ordinary experiences and is as it were common-sense cosmology; (2) Duhem recognizes a strong similarity between features of general thermodynamics (his basic physical theory) and essential doctrines of Aristotelian physics (Aim, Appendix, pp. 305–311). However, it is difficult to see how this type of order, which seems to be molded on the biological order of organisms, shall be applied to the ordering of experimental data or to phenomena governed by universal laws. One can doubt that Duhem’s move to accept a metaphysical principle that gives sense to physics as an empirical enterprise is indeed singling out Aristotelian ontology or cosmology as the one metaphysics most adequate for it. Duhem’s continuity thesis does not imply that all experimental laws have to be conserved through theoretical changes. “When the progress of experimental physics goes counter to a theory and compels it to be modified or transformed, the purely representative part enters nearly whole in the new theory, bringing to it the inheritance of all the valuable possessions of the old theory, whereas the explanatory part falls out in order to give way to another explanation” (Aim 32, emphasis added). After Duhem’s pioneering work, Anneliese Maier, Ernest Moody, Alexandre Koyr´e, Marshall Clagett, and E. J. Dijksterhuis have devoted extensive studies to the topic of medieval precursors to modern science (see Clagett 1969). “But it is due to the physicians and astrologers of the sixteenth century that we must attribute the precise and fruitful idea of decomposing the total tide into two tides of the same nature though of unequal intensity, one produced by the moon and the other by the sun, and to explain the diverse vicissitudes of ebb and flow by the agreement and disagreement of these two tides” (Aim 241). We tend to believe that once Kepler had realized that the path of Mars could not be a cycle was very easy for him to test in his next trial the next like conic section, viz. the ellipse. However this hypothesis was far beyond standard reasoning of astronomers at the time, and many astronomers after Kepler’s publication took the elliptical orbit to be but a scandalous slip of a physicist in astronomy. Therefore Kepler did work for a long time with an oval orbit for Mars. Still more astonishing is the fact that in that period Kepler works with elliptical orbits for reasons of calculation, but he does not dare understand them as the true paths of the planets. The pulling power emanating from the sun is, then, not acting along the radius from the sun to the planet, but perpendicular to it. The force making the planet to move around the sun is still conceived according to Aristotle’s model of a horse in harness. Duhem cites as his source a work on optics that is attributed to Euclid and claims that Kepler gave a demonstration of the theorem (Aim 245). To be more precise, Duhem first shows that Newton’s account is unacceptable on purely methodological grounds. The principle of universal gravitation cannot be understood as a generalization of Kepler’s laws because it contradicts them (Aim 193). From the view of Newtonian mechanics,

AUTONOMY VERSUS DEVELOPMENT: DUHEM ON PROGRESS IN SCIENCE

22.

23.

24.

25.

26.

27.

95

Kepler’s laws can be assigned the status of good approximations and the deviations from the values calculated on the basis of the former can be explained. This is due to the fact that the mass of the sun is very large compared to the masses of the planets. Yet on theoretical grounds Kepler’s and Newton’s theories are incompatible, and hence the latter cannot be a generalization of the former. Having shown this, Duhem proceeds in his criticism of inductivism by adducing the slow historical genesis of the theory of universal gravitation in which, as he shows, inductivism plays no role. Lakatos acknowledges this as Duhem’s “historiographical falsification of inductivism” (Lakatos 1971). The histories of scientific development can be told only in hindsight. We must know the final version of the theory and then strip off the earlier versions all parts that do not contribute to the formation of the next state. The historian of science will retain “only what prepared the way for the Newtonian theory, by neglecting systematically everything not tending to that goal” (Aim 222). Duhem’s notion of history is of the type which Nietzsche had described as “critical history,” where seeing from today we divide in the past between the parts which we can affirm and accept (the parts that we see as our past) and the parts that we condemn and negate (the parts that we will not recognize as our past, see Nietzsche 1956, p. 229f.). Duhem arrived at this position (including his “holistic” conception of theories and the denial of the possibility of an experimentum crucis) already in the 1890s, while analyzing an experiment in optics that was regarded to be a crucial experiment (see Brenner 1990). Pascal used this idea to illustrate the process of permanent progress in science: “De sorte que toute la suite des hommes, pendant le cours de tous les si`ecles, doit eˆ tre consid´er´ee comme un mˆeme homme qui subsiste toujours et qui apprend continuellement” (Pascal 1963, p. 232). In adopting this view, Duhem seems also to appeal to Haeckel’s biogenetic law, which states that ontogenesis consists in a brief and rapid recapitulation of phylogeny. Duhem suggests that the intellectual development of a student should “imitate the progress through which man’s knowledge of science has been formed” (Aim 268). Duhem seems to express his personal experience as a teacher of physics, which is ranked by him on the same high level as his experience as a man of research. “It was thus through the necessities of teaching, under their urgent and constant pressure, that we were led to produce a conception of physical theory marked different from what had been current till then [. . . ] Our ideas about the nature of physical theory are, therefore, rooted in the practice of scientific research and in the exigencies of teaching” (Aim, Appendix I. p. 278). Note the difference to Fleck and Kuhn! Both Kuhn and Fleck refer to the teaching and training of novices in science as a rite of initiation, as conditioning the student in conformity with the paradigm accepted by the scientific community at that time. In contrast, Duhem emphasizes the pressure exerted through the critical attitude of the students: “Among our students, many of whom are today colleagues of ours, the critical sense was hardly asleep” (Aim 277). This pressure challenges and eventually leads him to his new conception of physical theory. Lakatos acknowledges his debts to Duhem for both his conventionalist methodology (“Methodological falsificationism is a brand of conventionalism,” Lakatos 1970, p. 104) and his integration of history and philosophy of science (Lakatos 1971. For a more detailed discussion, see Sch¨afer 1974b, pp. 193–222). According to Duhem, in writing internal history one should neglect “all merely accidental facts, e.g., the name of an author, date of discovery, an episode or anecdote, in order to dwell only on historical facts appearing essential to the physicist’s eyes, and only on circumstances in which the theory was enriched by a new principle or saw an obscurity or erroneous idea disappear” (Aim 269). This matches well with Lakatos’s advice that the historian of science present in the text all that is in agreement with the structural aspects of science and add in footnotes to what extent actual history deviates from its rational reconstruction (Lakatos 1971). Whewell seems to have entertained similar ideas (see Schickore, this volume). Following the Duhemian influences on the philosophy of Otto Neurath, Don Howard is led to conclusions very close to the ones above (see Howard, this volume). With the faculty of good sense Duhem appeals to Pascal’s distinction between esprit g´eom´etrique and esprit de finesse. They both define different ways of correct reasoning (diverses sortes de

96

¨ LOTHAR SCHAFER

sens droit). The geometric mind is needed in accordance with the demands of logic in a deductive system, like Euclidean geometry in axiomatic representation. The mind of finesse is needed in very complex situations where a multiplicity of (conflicting) rules are to be observed and the conditions of their application are not spelled out explicitly, as in questions of style. The motives that direct our choices in complex situations “constitute what is appropriately called good sense” (Aim 217). 28. For details, see Stephenson 1987. 29. Note the similarity of Duhem’s description of theory choice to Kuhn’s description of paradigm change, where “values” like precision, simplicity, consistency, explanatory power, etc., are involved, yet their application is so complex and subtle that they do not yield a unique result (Kuhn 1962, pp. 184–186). 30. Duhem never renounced or mitigated his basic conventionalism as can be seen in his “personal account” of 1913: “The principles are laid down as pure postulates, arbitrary decrees of human reason; they are considered to have successfully fulfilled their role when they yield numerous consequences that conform to experimental laws” (Duhem 1987, p. 334).

REFERENCES Ariew, Roger and Barker, Peter (1992), “Duhem and Continuity in the History of Science”, Revue International de Philosophie, 182: 323–343. Brenner, Anastasios (1990), “Holism a Century Ago: The Elaboration of Duhems’ Thesis”, Synthese 83: 325–335. Clagett, Marshall (ed.) (1969), Critical Problems in the History of Science (Madison: University of Wisconsin Press). Diederich, Werner (1974), Konventionalit¨at in der Physik (Berlin: Duncker und Humblot). ´ Duhem, Pierre (1906–1913), Etudes sur L´eonard de Vinci, 3 Vols. (Paris: Hermann). Duhem, Pierre (1913–1959) Le Syst`eme du monde. Histoire des doctrines cosmologiques de Platon a` Copernic, 10 Vols. (Paris: Hermann). Duhem, Pierre (1962), The Aim and Structure of Physical Theory (trans. P.Wiener) (Princeton: Princeton University Press). Duhem, Pierre (1969), To Save the Phenomena (Chicago: University of Chicago Press). Duhem, Pierre (1980), The Evolution of Mechanics (trans. M. Cole) (Alphen aan den Rijn: Sijthoff & Noordhoff). Duhem, Pierre (1985), Medieval Cosmology (transl. R. Ariew), Chicago. Duhem, Pierre (1987), “An Account of the Scientific Titles and Works of Pierre Duhem”, Science in Context I: 333–348. Fleck, Ludwik (1979), Genesis and Development of a Scientific Fact (trans. F. Bradley and T. Trenn) (Chicago: University of Chicago Press). Frank, Philip (1949), Modern Science and its Philosophy 2nd (Cambridge, MA: Harvard University Press). Hentschel, Klaus (1987), “Einstein, Neukantianismus und Theorienholismus”, Kant-Studien 78: 459– 470. Howard, Don (1990), “Einstein and Duhem”, Synthese 83, 363–384. Kepler, Johannes (1992), NewAstronomy (trans.W. Donahue) (Cambridge: Cambridge University Press). Kuhn, Thomas (1962), The Structure of Scientific Revolutions (2nd edn. 1970) (Chicago: University of Chicago Press). Lakatos, Imre (1970); “Falsification and the Methodology of Scientific Research Programmes”, in Lakatos, I. and Musgrave, A. (eds.), Criticism and the Growth of Knowledge (Cambridge: Cambridge University Press), pp. 91–195. Lakatos, Imre (1970), “History of Science and its Rational Reconstructions”, PSA: Proceedings of the Biennial Meeting of the Philosophy of Science Association 1970: 91–136.

AUTONOMY VERSUS DEVELOPMENT: DUHEM ON PROGRESS IN SCIENCE

97

Lugg, Andrew (1990), “Pierre Duhem’s Conception of Natural Classification”, Synthese 83:409–420. Maiocchi, Roberto (1990), “Pierre Duhem’s The Aim and Structure of Physical Theory: A Book Against Conventionalism”, Synthese 83: 385–400. McMullin, Ernan (1990), “Comment: Duhem’s Middle Way”, Synthese 83: 421–430. Nietzsche, Friedrich (1956) “Vom Nutzen und Nachteil der Historie f¨ur das Leben”, in Werke (K. Schlechta (ed.)), Vol. I (M¨unchen: Hanser). Pascal, Blaise (1963) “Pr´eface sur le trait´e du vide”, Oeuvres compl`etes (Lafuma (ed.)) (Paris: Seuil). Petrus, Klaus (1996), “Naturgem¨aße Klassifikation und Kontinuit¨at, Wissenschaft und Geschichte”, Journal for General Philosophy of Science 27: 307–323. Sch¨afer, Lothar (1974a), “Pascal und Descartes als methodologische Antipoden”, Philosophisches Jahrbuch 81: 314–340. Sch¨afer, Lothar (1974b), Erfahrung und Konvention: Zum Theoriebegriff der empirischen Wissenschaften (Stuttgart: frommann-holzboog). Stadler, Friedrich (1997), Studien zum Wiener Kreis (Frankfurt am Main: Suhrkamp). Stephenson, Bruce (1978), Kepler’s Physical Astronomy (New York: Springer).

VOLKER PECKHAUS

PSYCHOLOGISM AND THE DISTINCTION BETWEEN DISCOVERY AND JUSTIFICATION

1. INTRODUCTION

There is a certain analogy between the discovery–justification distinction (DJ distinction) in the philosophy of science and the genesis–validity distinction in epistemology and the foundational discourse in logic.1 The investigation of this analogy may reveal the tight relation between Hans Reichenbach’s famous distinction and earlier modes of argument in foundational debates in post- and neo-Kantian times. Starting point of this paper is, however, not Reichenbach, but Karl R. Popper who had claimed in his Logic of Scientific Discovery that the way theories are set up cannot be subject of logical analysis, but it does also not require logical analysis (Popper 1934, p. 6). Popper himself associates this claim with Immanuel Kant’s distinction between quid iuris and quid facti questions, a distinction Kant had taken from the juridical practise of his time (Kant 1787, B 116). Quid iuris questions are concerned with the justification of legal claims, quid facti questions with the matters of fact. The method of justifying legal claims is called “deduction”, again a term from juridical terminology. Critical philosophy is mainly concerned with pure concepts, i.e., concepts independent from all experience, the deduction of which cannot recur to any experiences and has therefore to be transcendental. Metaphysics, and with it the germ of philosophy proper, is erected on exactly such transcendental concepts. It follows that quid facti questions are irrelevant for philosophy, which is mainly concerned with quid iuris questions, i.e., with justification. The matters of fact, e.g., the (factual) genesis of ideas, are no philosophical problem. Kant’s determination of the domain of philosophical activity is clearly a precursor of Hans Reichenbach’s distinction between the “context of discovery” and the “context of justification”. For Reichenbach, as well, philosophy or, as he says, epistemology, is only concerned with justification. It aims at a “rational reconstruction” of the scientific process, coming close to the manner a mathematician presents a new proof in an article or a physicist publishes his considerations on the foundations of a new theory (Reichenbach 1938, pp. 6–7). This is in agreement with Popper’s earlier remark that the way something new is discovered is of no interest for epistemology, but only for empirical psychology (Popper 1934, p. 6). In mentioning psychology Popper obviously takes up elements of the heated debate concerning the so-called “psychologism in logic” towards the end of the 19th century which was related to the foundational debates in logic after Hegel’s death in 1831. 99 J. Schickore and F. Steinle (eds.), Revisiting Discovery and Justification, 99–116.
C 2006 Springer. Printed in the Netherlands.

100

VOLKER PECKHAUS

This paper aims at illustrating the analogy just mentioned. It is intended to show that the epistemological rigor which guided the distinction with its preference for the justificational side, may lead to a methodological impoverishment, by driving “how to” questions out of the sphere of interest of epistemology and philosophy of science. Even if one agrees that philosophy should not be concerned with individual, specific research processes, or with telling stories about certain inventions, discoveries and “heurica” situations, the empirical insight cannot be ignored showing that a scientist’s research activities are not restricted to applying the inductive method.2 They are also influenced by other methods and circumstances, like analogy, combinatorics, deductive quasi-empirical methods, or simply by accidents. If philosophy is regarded as a reflective attitude and activity it should also reflect on these questions. Such reflection may not necessarily aim at establishing a logic of scientific discovery, but can more modestly attempt to establish a general philosophy of research combining considerations on discovery and justification. It is granted that both aspects should not be muddled up. They are two distinct sides of a coin, but of the same coin. This paper will deal with this mutual relationship between discovery and justification focusing on some aspects of the 19th-century foundational debate in logic. The foundational debate as such belongs undoubtedly to the context of justification.3 But within this context, the opposition of genesis and validity occurred again. Section 2 provides a short description of the philosophical discussions in logic in the 19th century which constitute the background of the debates on psychologism. The third section presents variations of psychologism and briefly discusses Gottlob Frege’s and Edmund Husserl’s antipsychologisms. Frege, e.g., spoke of the “corrupting intrusion of psychology into logic” (Frege 1893, XIV–XV; Beaney 1997, p. 202), and Husserl devoted large parts of the first volume of his Logische Untersuchungen (Husserl 1900) to the criticism of psychologism. The fourth section discusses consequences of antipsychologism. In fighting against what Matthias Rath called “substitutive psychologism” (Rath 1994b, p. 314), Frege and Husserl also hit directions standing for a weak version of “attributive psychologism” (ibid., p. 311) in which psychology was used as a tool for gaining philosophical insights, but not as a criterion for the foundation of these insights. The anthropological logic of Jakob Friedrich Fries (Fries 1837, 1st ed. 1811), later taken up by the G¨ottingen philosopher Leonard Nelson, can be named as an early attempt in this direction. The conclusion is that the situation in the foundational debates in logic was similar to that in the philosophy of science about the possibility of a logic of discovery. In taking up strict dichotomic distinctions the advocates of a restrictive scope in philosophy of science and logic dismissed the potential of the opponent’s conceptions. 2. DEBATE ON THE “LOGICAL QUESTION”

2.1 The State of Logic in the Early 19th Century The logic discussion in post-Hegelian Germany stood under the label “the logical question” (cf. Peckhaus 1997, chapter 4). In this discussion the position of logic

DISTINCTION BETWEEN DISCOVERY AND JUSTIFICATION

101

within the system of philosophy was considered, in order to overcome the Hegelian paradigm of logic. Rationalistic positions were re-approached cautiously and critically. An essential mark of Georg Friedrich Wilhelm Hegel’s philosophical system can be seen in his identification of logic and metaphysics. It is his radical answer to the demand of reforming logic which had already been on the philosophical agenda for a long time. According to Wilhelm Windelband’s description, Hegel and his successors, on the one hand, acknowledged “the creative principle” of Kant’s transcendental logic, which had been posed against formal logic, but more resolutely than Kant they aimed at a “complete revision of the old general logic” on an epistemological basis (Windelband 1904, pp. 164–5). Hegel’s radicalism opened a wide field for polemic argument, especially due to his merging of logic and metaphysics, his indifference towards formal logic, and his depreciation of mathematics and (“positive”) sciences. In the introduction of the first volume of his Wissenschaft der Logik, he rejects the opinion that logic was the science of reasoning in general, reasoning which abstracts from all contents (Hegel 1812, II–III). By contrast, logic is, as he defines in his Encyclop¨adie der philosophischen Wissenschaften, the “science of the pure idea, i.e., the idea in the abstract element of reasoning” (Hegel 1830, p. 27). Logic therefore coincides with metaphysics, “the science of objects formed into thoughts, intended to express the essentials of objects” (ibid., p. 34). Hegel’s polemics is directed against traditional formal logic. He refers to Kant’s famous claim that logic had gone the safe course of science from oldest times on, which was obvious by the fact that it was not allowed to take any step backward since Aristotle. Kant stated it as curious, however, that logic was also not able to take any step forward, with other words, it was obviously closed and completed (Kant 1787, B VIII). Hegel now reports that Kant had praised logic as being completed so early. But if logic had not been changed since Aristotle, one should rather conclude, that it needed the more a complete recast (Hegel 1812, Introduction, XV–XVII). Hegel states that this need for a change has been felt for a long time. In form and content logic has fallen into disdain as can be seen from standard textbooks. It is still taught because one is used to its alleged importance, but not convinced that its usual contents and the occupation with empty forms had any value and benefit (ibid., XVI–XVII). Hegel speaks of the “dead content of logic” (ibid., XVII) which is without spirit (geistlos) to that degree, because its determinations are firmly fixed, so that they can only be brought into exterior relations to each other. In traditional logic, the operations in judgments and inferences are above all based on quantitative determinations, so that everything lasts on an exterior difference, on pure comparison. Logic becomes an analytic method, a conceptless calculation (ibid., XVIII). This establishes, of course, an analogy to mathematics. The kind of reasoning represented by formal logic has not without justification been identified with calculating, and calculating with reasoning. Pure mathematics has, of course, its method. This method is suitable for its abstract objects. It is, however, subordinate in respect to its scientific value (ibid., XVIII–XIX). Hegel refers to his Ph¨anomenologie des Geistes, where he had demanded that philosophy should develop a new concept of scientific method (Hegel 1807, L–LV) which cannot be borrowed, however, from a subordinate science like mathematics (Hegel 1832, XII–XIII).

102

VOLKER PECKHAUS

2.2 The Logical Question Logic became a central topic in the critical discussion of Hegel’s philosophical system after his unexpected death in 1831. The debate about the logical question was released by Friedrich Adolf Trendelenburg, the great critique of Hegelian philosophy and the dialectical method, who was professor of practical philosophy and pedagogy at the Friedrich Wilhelms University in Berlin. He brought an innovative impetus into the development of logic in the first half of the 19th century, although he criticized both, speculative logic and formal logic in Aristotelian-scholastic tradition. Nevertheless, in education he supported Aristotelian philosophy. With his Elementa logices Aristotelicae (Trendelenburg 1836), which were widely distributed as school textbook, he set the discussion of formal logic, usually called “Aristotelian”, onto a more solid fundament of sources. Trendelenburg first used the term “the logical question” in an article entitled “Zur Geschichte von Hegel’s Logik und dialektischer Methode”, first published in the Neue Jenaische Allgemeine Literatur-Zeitung (Trendelenburg 1842). According to its subtitle, the paper dealt with “Die logische Frage in Hegel’s Systeme” appealing for its “scientific getting through with”. But what is the logical question for Trendelenburg? It is formulated at the end of the article: “Is Hegel’s dialectical method of pure thought a scientific method?” (Trendelenburg 1842, p. 414). In answering this question to the negative he denies dialectics’ claim of absoluteness in logic, and with this in philosophy as such. At the same time he demands to unhinge Hegel’s system. This had to have consequences for formal logic, as well. He refers to apologists of dialectics when writing, “to the same degree as formal logic failed to fulfill its task, i.e., to understand the process of getting knowledge, one saw therein an indirect proof for the truth of speculative dialectics” (Trendelenburg 1842, p. 406). If dialectics itself was proved to be insufficient, the deficiency of the old formal logic was still given, a fact Trendelenburg never became tired to stress, but the alternative disappeared, reason enough to reconsider task and status of formal logic in the framework of epistemology. Subsequently, the term “the logical question” became standard in the discussion on a reform of logic, but it was used in a less specific form. It generally stood for the controversial debates in Germany on the shape and composition of logic. In this sense Georg Leonhard Rabus wrote in his Die neuesten Bestrebungen auf dem Gebiete der Logik bei den Deutschen und Die logische Frage: “‘The logical question’ arose from doubts about the justification of formal logic” (Rabus 1880, p. 1). The late 19th-century debate on logic was thus not really a restorative defence against criticism in the post-Kantian period; it rather continued critical approaches in that time. It aimed at a reform of traditional logic, called “Aristotelian”, which was considered insufficient. Formal logic was regarded as resistant to any reform, but, on the other hand, not as really significant for logic as a whole. The reform therefore found its expression in a shift of emphasis from the further development of logic as a methodological tool towards foundational and applicational questions. This shift led to detachment effects. In its traditional form, logic contained not only formal logic with its theories of concepts, judgments, and inferences, but also epistemological theories

DISTINCTION BETWEEN DISCOVERY AND JUSTIFICATION

103

of concept formation and methodological considerations in sections about applied logic (Wissenschaftslehre). It was regarded as propaedeutic discipline, providing the requirements for dealing with the core discipline of philosophy, metaphysics. In effect of the reform debates, parts of this collective discipline were separated from its subject matter, even from the domain of competence of philosophy as such. This is even true for its core part, formal logic, which tended towards mathematics. The philosophical foundational debate dealt above all with the question whether logic could or had to be founded psychologically or epistemologically in advance. This discussion was closely tied to the question whether psychology was the leading discipline for philosophy or, on the contrary, psychology presupposed logic.

3. PSYCHOLOGISM–ANTIPSYCHOLOGISM

3.1 Variations of Psychologism The foundational question was asked by reconsidering the evidence of logical principles. A central point was the relation between what happens in an act of reasoning and what should happen in correct reasoning. Investigations dealing with the first aspect were ascribed to psychology understood as a philosophical sub-discipline. The normative aspect was regarded as belonging to the logical domain. The demarcation between both perspectives on reasoning became the essential topic of the reform discussion. Windelband spoke of the risks of an amalgamation of logical and psychological investigations in the framework of the “inevitable relation” between logic and psychology (Windelband 1904, pp. 169–170): While the logician has to start from psychological analyses of what is really happening in a judgment, the resulting aspects are also easily and unconsciously foisted on him as criteria for the logical treatment of the subject, and once the crucial point of difference is missed, the whole logic is in danger of becoming only a branch of psychology, as was demanded and realized in former times, e.g., by Beneke. The firm demarcation against psychology is a vital issue for logic as a philosophical discipline.

Windelband refers here to Friedrich Eduard Beneke, who had classified logic as a branch of psychology in his Lehrbuch der Logik als Kunstlehre des Denkens (Textbook of Logic as the Art of Reasoning, Beneke 1832). Johann Eduard Erdmann introduced the term “psychologism” in 1870 for neutrally describing Beneke’s systematic position.4 Windelband refers also to the so-called “Psychologismusstreit” (psychologism dispute) in end 19th-century German philosophy about the relation of psychology to philosophy, especially to logic, which led finally to the constitution of psychology as an independent science in the beginning of the 20th century, when psychology left the union of philosophical disciplines.5 In the course of the debates the term “psychologism” became a polemical character, thereby leveling the differences between the various conceptions of the relation between logic and psychology. Matthias Rath distinguishes three kinds of psychologism:

104

VOLKER PECKHAUS

Attributive psychologism aims at attributing psychology to logic and to use it for solving special non-psychological problems. Jakob Friedrich Fries (1773–1843) belongs to the pre-history of this variation. In his System der Logik (Fries 1837) he demanded to supplement “demonstrative logic”, which he regarded, like Kant, as basically completed with Aristotle, with an “anthropological logic” which uses inner experience as a methodical device for answering the basic question “How can concept and reasoning be subsumed under the activities of the human mind?” (ibid., p. 3). It is only a small step to go even further and claim that logic is not able to fulfill its tasks and therefore has to be psychologized. This was the aim of the main representatives of attributive psychologism in the end 19th-century psychologism dispute, Alois Riehl and Benno Erdmann. Both wanted to save logic as a science by psychologizing it, and connecting it to scientific thought.6 In his inaugural lecture Ueber wissenschaftliche und nichtwissenschaftliche Philosophie (Riehl 1883) Riehl determines philosophy as “Erkenntniswissenschaft” (science of knowledge) (ibid., p 38). Whereas psychology is constituted as methodical guiding discipline (ibid., p. 42), logic is regarded as the descriptive part of philosophy of science (Wissenschaftslehre). Logic “describes the composition of the scientific whole from its elements and their combination.”7 Benno Erdmann determines logic as “the science of the formal conditions of scientific thought, i.e., the science of formal conditions for valid judgments on subjects of sense perception and consciousness” (Erdmann 1892, section 4.17). He therefore separates normative logic from a psychology, understood as a science dealing with matters of fact.8 In this respect, Erdmann follows Johann Friedrich Herbart, who had defined logic as the science of understanding (Wissenschaft des Verstandes) which does not investigate “according to which mental laws it may happen that we follow and firmly determine in reasoning the nature of what is thought, and, with this, rise above the game of sudden ideas and moods.” The last is reserved for psychology like everything else concerning mental incidents. “In logic, it is necessary to ignore everything psychological, because it only has to determine those forms of possible connections of what is thought, which are allowed by that what is thought according to its nature” (Herbart 1813, section 34). Contrary to Herbart, Erdmann stresses “that the psychological knowledge of the matters of fact in the process of judging is a condition for the logical decision about their validity which always deserves consideration.” In this sense “the logical ‘should be’ is dependent on being” (“ist das logische Sollen vom Sein abh¨angig”, Erdmann 1892, section 5.21). Much more radical than attributive psychologism is substitutive psychologism in which, as Rath writes, “psychology is no more seen as an inner philosophical quarry for solving of other, non-psychological problems,” but as “the basic science of philosophy on the whole, and with it, of all sciences” (Rath 1994b, p. 314). This position is already present in Beneke’s philosophical system where logic is defined as “the science of reasoning” having the task “to completely and clearly explain the forms and the genesis of our reasoning processes” (Beneke 1832, section 1). Logic is for Beneke, like all of philosophy, an empirical science, founded psychologically, and based on inner experience (ibid., section 16):

DISTINCTION BETWEEN DISCOVERY AND JUSTIFICATION

105

In this way of dealing with it, logic is therefore from the beginning applied psychology: as all other philosophical sciences have to become if they should be developed in full clearness and determinateness.

Besides Beneke’s logic, empiricist concepts like the inductive logic of John Stuart Mill can be interpreted in this sense (Mill 1843, cf. Rath 1994a, pp. 128–142). Theodor Lipps stands for the most radical variation of this kind of psychologism. In a discussion of Wilhelm Wundt’s logic, Lipps criticized as early as 1880 the normative function of logic, as it was stressed, e.g., by Benno Erdmann. Logic certainly provides the laws of “correct reasoning”, but this reasoning is at the same time a reasoning which we have to follow according to our nature. The rules of reasoning are therefore “identical with the natural laws of reasoning itself. Then, logic is [. . . ], according to this view, physics of reasoning or it is nothing at all” (Lipps 1880, p. 531). Psychology as the theory of the natural laws of reasoning is consequently identified with logic as the theory of the rules of reasoning, but Lipps goes even further: “one can ask whether philosophy can be anything else than psychology in the widest sense of the word” (Lipps 1880, p. 538). In a later book Lipps deals with logic as the theory of judgments. A judgment can, according to Lipps, only be seized in the act of judging. The only way of dealing with this act is psychological, if logic does not intend to blather blindly (Lipps 1912, p. 11). The whole theory of judgments is bound to experience, there is no pure necessity of reasoning: “With other words: logic is nothing or it is psychology” (ibid., p. 11). Rath mentions as a third variation the reflective constructive psychologism, which is determined by the attempt to place psychology as a science of its own in a reformed system of sciences. Typical representatives are philosopher psychologists like Wilhelm Wundt, Carl Stumpf, or neo-Kantian philosophers (cf. Rath 1994a, pp. 180–247.) 3.2 Antipsychologism The hardest opposition against psychologism came from mathematics. The dynamic development of 19th-century mathematics had unveiled the need for reconsidering the foundations of mathematics. Euclidean geometry had lost its paradigmatic status in mathematics, common algebra bound to the arithmetic of natural numbers proved to be insufficient for dealing with the new non-commutative “numbers” like hypercomplex numbers or vectors. The development of “pan-theories” was on the agenda, i.e., theories so general that they comprise old and new conceptions. Contrary to philosophers, the mathematicians were interested in elaborating formal logic in order to utilize it for foundational needs in mathematics.9 Within this new interest in the philosophy of mathematics inaugurated by mathematicians themselves, the logicistic mathematician Gottlob Frege, and the phenomenologist Edmund Husserl, also a qualified mathematician, represented the most prominent antipsychologistic positions, sometimes identified with antipsychologism as such. Their opposition against a psychological foundation of logic became more important than Wilhelm Windelband’s “normative antipsychologismus (Kusch 1995, p. 53) referred to above. It has

106

VOLKER PECKHAUS

long been regarded as having ended any psychological influence in logic once and for all. In his seminal Begriffsschrift (Frege 1879, English extract Frege 1997a) Frege joined the standard distinction between logic as normative science and psychology as empirical science. Frege begins his Preface with considerations on the justification of scientific truths. He divides all truths that require justification into two kinds, “those whose proof can be given purely logically and those whose proof must be grounded on empirical facts” (Frege 1879, III; transl. Frege 1997a, p. 48). He then states (ibid., III–IV; transl. p. 48): But there is no inconsistency in a proposition belonging to the first kind and yet being such that it can never be apprehended by a human mind without the operation of the senses. Thus it is not psychological origination but the most perfect method of proof that lies at the basis of the division.

It was now Frege’s project in the Begriffsschrift and in his subsequent logical and philosophical work to determine the kind arithmetical judgments belong to. His considerations in the Begriffsschrift on the relation between logic and psychology were programmatic, not critical. They were not yet antipsychologistical, but served for adjusting his logicist program of reducing mathematics to logic to the standard division between logic and psychology. In the Preface of the first volume of his Grundgesetze der Arithmetik of 1893, however, Frege spoke of the “corrupting intrusion of psychology into logic” which “seems to be infected through and through by psychology.”10 For him, the “psychological logician” is exemplified by Benno Erdmann, who had published his “Logische Elementarlehre” in 1892 (Erdmann 1892). Frege admits that usually the role of logical laws as guiding principles for thought is granted, but then emphasizes that the ambiguity of the word “law” is fatal in this context. “In one sense it states what is, in the other it prescribes what should be.” The normative reading of logical laws as determination of how one should think is unproblematic for Frege. He criticizes the assumption that the laws of thought govern reasoning in the same way as the laws of nature govern the exterior world (Frege 1893, XV; Beaney 1997, p. 202): They can then be nothing other than psychological laws, since thinking is a mental process. And if logic were concerned with these psychological laws, then it would be a part of psychology.

The relation between logic and psychology, with critical stress, was also topical in Frege’s later work. In his paper “Der Gedanke” of 191811 Frege describes it as the task of all sciences to discover the truth, and “it falls to logic to discern the laws of truth.” Again he discusses the notion of “law”. Laws of nature “are general features of what happens in nature, and occurrences in nature are always in accordance with them.” It is in this sense that he speaks of laws of truth. Laws of truth do not refer to what happens, but to what is. “From the laws of truth there follow prescriptions about asserting, thinking, judging, inferring.” Therefore, it might be possible to speak of laws of thought. Again he warns not to confuse different things. If people regard laws

DISTINCTION BETWEEN DISCOVERY AND JUSTIFICATION

107

of thought as analogous to the laws of nature, laws of thought would refer to “general features of thinking as a mental occurrence”. Such laws are, however, psychological laws. People might believe that logic deals with mental processes of thinking which follow these psychological laws. But these people are mistaken. They confuse taking to be true with being true (Frege 1918, pp. 58–59; Frege 1997b, p. 326): Whether what you take for true is false or true, your so taking it comes about in accordance with psychological laws. A derivation from these laws, an explanation of a mental process that ends in taking something to be true, can never take the place of proving what is taken to be true. [. . . ] In order to avoid any misunderstanding and prevent the blurring of the boundary between psychology and logic, I assign to logic the task of discovering the laws of truth, not the laws of taking things to be true or of thinking. The Bedeutung of the word “true” is spelled out in the laws of truth.

Edmund Husserl, who was said “to have delivered the deathblow to psychologism” on the threshold of the 20th century (Pfeil 1934, p. 179), was deeply impressed by Frege’s criticism of psychologism which was also directed against himself. His Philosophie der Arithmetik (Husserl 1891) was target of Frege’s criticism, because there he had based arithmetic on psychological foundations.12 Husserl devoted the whole first volume of his Logische Untersuchungen entitled Prolegomena zur reinen Logik (Husserl 1900) to the criticism of foundational conceptions for logic predominant in his time. The first two chapters deal with the old quarrel about the question whether logic is science or art. Husserl acknowledged normative elements of logic which becomes an art (Kunstlehre) as soon as certain purposes are included (cf. Husserl 1900, section 15, p. 47). But each normative discipline presupposes one or more theoretical disciplines as foundation in the sense (ibid., section 16, pp. 47–48), that they have to have a theoretical content that is separable from all normativeness which, as such, has to have its natural position in some theoretical sciences, already determined or still to be constituted.

Husserl saw the theoretical foundation of normative logic in formal “pure logic” which is used in its mathematical variation by the mathematician for the construction of theories and for the rigid and methodical solution of formal problems, but which also belongs to the philosophers’ domain in its function of forming theories (cf. ibid., section 71, pp. 252–254). Husserl’s description and critique of the psychologistic foundation of logic is broadly dealt with in chapters 3–10 of the Logische Untersuchungen. Starting point of his criticism is the claim that psychology provides the theoretical foundation of logic. Husserl concedes that psychology is concerned with the foundation of logic, but he doubts that it provides the essential fundament of logic. He prepares his justification by hinting at “pure logic” which has, already in the systems of Kant and Herbart, developed the essential theoretical foundation of logic independently from psychology (ibid., section 20, p. 71). Founding logic on the science of psychology, which deals with matters of fact, has empiricist consequences which can, according to Husserl, easily be refuted. First of all, there are still no exact laws in psychology. It is based on valuable, but only vague generalizations from experience (ibid., section 21, p. 72). Logical laws founded on psychological laws would therefore also be vague. Such a foundation of logical laws

108

VOLKER PECKHAUS

“would change its true sense through and through”, because the “principles” of logic, the laws of syllogistics and the other modes of inference, “are obviously genuine laws, and not ‘only empirical’, i.e., approximate rules” (ibid., section 21, p. 73). Secondly, even if psychology had exact laws, it will not do any good, because the laws of nature supposed to be exact cannot be recognized a priori, but were themselves based on inductive generalizations. Induction, however, does not establish the validity of a law, but only its more or less high probability. If laws of logic were derived from laws of such kind they would have the same status of pure probability (ibid., section 21, p. 62). Husserl stressed (ibid.): Compared with this, nothing seems to be more evident as that “purely logical” laws are altogether a priori valid. They find their foundation and justification not by induction, but by apodictic evidence.

If finally logical laws had their source of knowledge in psychological matters of fact, they would have themselves psychological content, “namely in a double sense: they had to be laws for something psychological and at the same time they had to presuppose or include the existence of something psychological” (ibid., section 23, p. 69). This is, however, demonstrably false, because no logical law implies a matter of fact, “not even the existence of ideas or judgments or other phenomena of knowledge” (ibid.). The psychologistic attempts to reform logic had, nevertheless, against Windelband’s evaluation, their effects even on formal logic. One of these influential effects of the demand that “all reasoning is judging”13 was the tendency to break the dominance of the theory of inference in favor of the theory of judgment, a tendency which can be observed in several directions of 19th-century logic, e.g., in the logic books of Hermann Ulrici, Friedrich Ueberweg, Benno Erdmann, Christoph Sigwart and others. Psychological theories were furthermore taken as direct reason for re-formulations of formal logic, as can be seen, e.g., in Franz Brentano’s influential conception. In the first volume of his Psychologie (Brentano 1874), Brentano sorted judging into psychic principles classes, together with imagining and affects like love and hate. One consequence for logic is that the interpretation of the forms of judgments in respect to their quality is dominant. With this, Brentano claimed, Aristotelian logic has been overcome.14 4. PSYCHOLOGY AS TOOL: ANTHROPOLOGICAL CRITIQUE OF REASON

After having analyzed a corpus of about 200 philosophical texts published in Germany and Austria between 1866 and 1931, Martin Kusch compiled a list of 139 authors who had been labeled “psychologists”, together with the overall number of attributions (Kusch 1995, p. 97). The “winner” is the antipsychologist Edmund Husserl with 21 attributions, immediately followed by Theodor Lipps with 20 attributions. In agreement with this, the German historian of philosophy Lutz Geldsetzer goes so far to maintain “that the philosophical current of phenomenology starting with Husserl was and kept being, contrary to the self-image of most phenomenologists, the second big

DISTINCTION BETWEEN DISCOVERY AND JUSTIFICATION

109

shape of German psychologism [after Fries].” He argues that the results of phenomenologists were received by psychologists as important contributions to their science. Furthermore, phenomenology stood in close contact to psychological research by, e.g., Franz Brentano and Alexius von Meinong (Geldsetzer 1990, p. 437). It is a matter of course that this kind of psychologism is not under debate in our context. We speak about positions concerning the role of psychology within a justificatory context, especially in the context of justifying logic. Throughout this debate, psychology is regarded as an empirical discipline, whereas logic, and with it, philosophy built on logic, is regarded as a priori science. The question is whether an a priori science can be justified empirically or whether, on the contrary, every empirical science presupposes an a priori foundation. A hardliner psychologist of the substitutive kind, but also a strict attributive psychologist would give the former answer, an antipsychologist the latter. The claim argued for in this paper is that a weaker form of attributive psychologism could accept an a priori foundation while, nevertheless, conceding psychology an important role in the foundational process. Whereas strong attributive psychologism would agree that psychology is the foundational discipline of logic, weak attributive psychologism regards psychology as a device for finding the foundations, i.e., a method of justification but not the justification itself. Rath obviously depreciated this method when saying that attributive psychologism saw psychology as an “inner philosophical quarry for solving other, non-psychological problems” (Rath 1994b, p. 314). It can be seen, however, in a more positive way. If a foundational problem can be solved with the help of empirical methods, why forbid to use them? Jakob Friedrich Fries’ epistemology gives an example for a more relaxed attitude towards the interplay between empirical and a priori methods. Popper concedes Fries a better sense for the foundational problem than most other philosophers, but, nevertheless, he used the critical label “Friesean Trilemma” (dogmatism—infinite regress—psychologist basis) for Fries’ way of dealing with it (Popper 1934, p. 60). It is an early precursor of Hans Albert’s “M¨unchhausen Trilemma” (cf. Albert 1968, p. 15). If the propositions of science are not to be introduced dogmatically, they have to be justified. If a logical justification is required, propositions can only be justified by propositions that themselves have to be justified by other propositions. So the demand of a logical foundation of these propositions leads to an infinite regress. If dogmatism and regress are to be avoided, psychologism is the result: According to Popper, Fries and with him almost all epistemologists who intended to give justice to empiricism teach that “mediated knowledge” (mittelbare Erkenntnis), i.e., the symbolical and linguistically represented sentences of science, is justified by immediate knowledge (unmittelbare Erkenntnis) represented by intuition (Anschauung) or sense perception (Sinneswahrnehmung).15 For this interpretation of immediate knowledge Popper referred to Julius Kraft who had been a pupil of the G¨ottingen neo-Friesean philosopher Leonard Nelson. He was a friend of Popper’s with whom he had long discussions on the theory of knowledge (cf. Popper 1974, p. 58), one could assume, a reliable source for Fries scholarship. Kraft writes in the context of a refutation of phenomenological intuitionism that all

110

VOLKER PECKHAUS

intuition is immediate knowledge. It does not require further reductions for proving its truth. For consciousness, the existence of the intuition and the truth of its contents is immediately present. The subject of intuition forces itself on consciousness, whereas knowledge originating in reasoning has to search for its subject by inferences (Kraft 1932, p. 120). Popper’s strong tie between intuition and sense perception which allows him to present Fries as a precursor of the Vienna Circle doctrine of protocol sentences, and the claim that all immediate knowledge only consists of intuition viz. sense perception, have no analogue in Kraft’s presentation. It appears to be necessary, thus, to go back to the writings of Fries in order to determine whether psychologism is indeed inevitable for the foundation of knowledge. In his Neue oder anthropologische Kritik der Vernunft (“New or anthropological critique of reason”), in a first edition published in 1807, the second edition appeared in 1828 (Fries 1828), Jakob Friedrich Fries presented a revision and elaboration of Kant’s Critique of Pure Reason, focusing especially on the problem of founding knowledge on its ultimate basis. The relevant section is section 70 about “Proof, Demonstration, and Deduction”. In the system of sciences, Fries writes, we do nothing else than proving propositions from principles (Grunds¨atze). These proofs show the truth of the conclusion as being inherent in the truth of the premises. In this system there is no more truth than can already be found in the premises (Fries 1828, pp. 336– 337). Nevertheless the status of principles differs in several variations of sciences. In the historical sciences each new proposition can be regarded as a new principle. Knowledge is therefore enlarged by every step. In mathematical sciences enhancement of knowledge is not due to logical proofs, but “to the possibility of composition according to the rule given in the principle” (ibid., p. 337), i.e., Kant’s construction of concepts in pure intuition. Finally, in philosophical sciences everything can be derived logically, so no truth goes beyond the truth expressed in the principles (ibid.). Fries suggests to reduce the general appreciation for proofs. He even argues against the rationalistic demand to proof everything (ibid., pp. 337–338). For him this demand originates in a misconception of the logical law of reason. Correctly understood it says that every judgment is mediated knowledge, i.e., truth or falsehood of these judgments have their cause in other judgments. Finding the cause for a given judgment must not be confused with proof. The foundational process for founding judgments on judgments has its end with first principles. According to Fries, a first principle repeats as judgment an immediate knowledge, its truth lies in its coherence with this immediate knowledge (ibid., p. 340). Concerning the truth of immediate knowledge two cases have to be distinguished dependent on the relation between this knowledge and consciousness. In empirical sciences and mathematics immediate knowledge is intuitive, i.e., its truth can be demonstrated by simply showing the respective sense perception or mathematical construction. If the relation between immediate knowledge and consciousness needs reflection, the situation is different. Reflection is needed if demonstration is not possible. This is the case in philosophy where the principles are not intuitive (ibid., pp. 340–341). Judgments like “every substance is persistent”, “each change has a cause”, “each being at the same time is determined by the interaction of substances”

DISTINCTION BETWEEN DISCOVERY AND JUSTIFICATION

111

allow us to understand the laws of nature without any reference to intuition. But laws expressed in these judgments have to be founded in immediate knowledge. Thus, the judgment can only be justified by showing the immediate knowledge it has as its basis. The method of showing this is called “deduction” (ibid., p. 342). This deduction presupposes a theory of reason which says which original knowledge we necessarily have and which principles necessarily originate from this in reason (ibid.): The principles of philosophy are without any justification in our conviction. No principle, however, may be assumed without reason, therefore we have to secure them by deduction by showing how the propositions expressed in them originate from the essence of reason.

This is the business of anthropology and, with this, of inner experience. Inner experience does not prove the truth of the principles but shows (aufweisen) these principles in reason (ibid., pp. 342–343). “I do not prove that every substance is persistent, but only show that the principle of the persistence of substance lies in each finite reason” (ibid.). 5. CONCLUSION: COMBINING THE CONTEXTS

Given Paul Hoyningen-Huene’s list of varieties of the DJ distinction (this volume) the Friesean conception can be analyzed as follows: For Fries philosophy is an activity, not a doctrine. He thereby followed the Kantian paradigm. The business of founding or justifying knowledge is thus a process consisting in the execution of two different activities, first the determination of the foundational basis, then the justification as such on this basis. This is in accordance to Hoyningen-Huene’s “standard interpretation” of the DJ distinction (version 1, discovery and justification as temporarily distinct processes). The process of determining the foundational basis consists in “showing” (aufzeigen) the immediate knowledge constituting this basis. We arrive at this immediate knowledge not by accident, but empirically by making use of our faculty of inner experience. Thus, within the process of discovery, empirical methods are used. Discovery leads to the basis from which logical justification can take start. This conception is not exactly matched by the variations presented in HoyningenHuene’s list. Versions 3 (the analysis of discovery is empirical, the analysis of justification is logical) and 4 (discovery is subject of empirical sciences, justification is subject of the philosophy of science) come close, but in deviation from these, discovery in Fries does not mean the analysis of a method, but the application of a psychological, i.e., empirical method. This application constitutes the precondition for the foundation, but it does not provide any foundation itself. Thus, it does not make foundation superfluous. The situation becomes even more complicated because we are faced with what can be called “a nested DJ distinction”. Within a context of justification, discovery and justification are both present. The situation can also well be described in respect to method, in particular as the result of an interplay between regressive analytical and progressive synthetical methods (cf. Peckhaus 2002). This description is based on a traditional distinction between two kinds of methods which can already be found, e.g., in 17th -century logic

112

VOLKER PECKHAUS

of Cartesian tradition. In La logique ou l’art de penser, first published anonymously in 1662, and today known under the name “The Logic of Port Royal”, the Jansenistic theologians and philosophers Antoine Arnauld and Pierre Nicole define in a section “On Method” (Arnauld and Nicole 1996, p. 233): The art of arranging a series of thoughts properly, either for discovering the truth when we do not know it, or for proving to others what we already know, can generally be called method. Hence there are two kinds of method, one for discovering the truth, which is known as analysis, or the method of resolution, and which can also be called the method of discovery. The other is for making the truth understood by others once it is found. This is known as synthesis, or the method of composition, and can also be called the method of instruction.

Science can thus be regarded as problem solving activity consisting of two basic parts which can be reconstructed according to an analysis–synthesis scheme. The analytical branch stands for the procedure which starts with the formulation of the problem and ends with the determination of the conditions for its solution. The synthetic branch represents the way from the conditions to the actual solution of the problem. It seems to be indisputable that science aims at presenting knowledge on the base of solved problems in textbook style. So the synthetic branch represents the main purpose of science. It is, however, deeply connected with the complementary analytical branch. Synthesis cannot be isolated, but presupposes analysis. This analysis–synthesis representation can easily be transferred to the DJ distinction. The context of discovery can be interpreted as standing for the analytical part of the scientific process, the context of justification for the synthetic side. Taking this analogy seriously, as Popper and Reichenbach might have done as well, entails to accept that justification is the main purpose of science which can, however, not be isolated, but is, on the contrary, dependent on discovery. Both aspects are intertwined, both together explain how science works, both are open for philosophical reflection. Therefore it has to be stated that Reichenbach’s distinction provides a fruitful conception for understanding science. Reichenbach was, however, misguided in claiming that philosophy should only be concerned with the justificatory aspect. NOTES 1. In this volume, Gregor Schiemann and Don Howard are pointing towards this analogy. I would like to thank Marcello Ghin, Paderborn, for helpful critical comments on a previous version of this paper. 2. For Reichenbach’s preference for induction even as a method for justification cf. Schiemann’s paper in this volume. 3. This was pointed out to me by Paul Hoyningen-Huene in private communication. 4. Usually the first edition of Erdmann’s Grundriss der Geschichte der Philosophie (Erdmann 1866) is given as a source, e.g., in Rath 1994a, p. 32, referring to Stumpf 1892. Eisler 1910 gives Erdmann the priority without naming the source. In the mentioned edition of Erdmann’s Grundriss, the term cannot be found in the passages on Beneke (Erdmann 1866, pp. 644–647). Erdmann obviously used it from the second edition of 1870 on (p. 646). It can also be found in the widely spread 4th edition which was revised by Benno Erdmann (p. 681). 5. Cf. the essential study of Matthias Rath (Rath 1994a). The story of the disciplinary separation of philosophy and psychology was investigated by Nicole D. Schmidt in her Hamburg dissertation

DISTINCTION BETWEEN DISCOVERY AND JUSTIFICATION

6. 7. 8. 9.

10. 11. 12.

13.

14. 15.

113

(Schmidt 1995). A comprehensive discussion of the debates on psychologism, especially in the first two decades of the 20th century, can be found in Martin Kusch’s book Psychologism (Kusch 1995). Cf. Erdmann 1892, also the analysis in Rath 1994a, pp. 114–120. Ibid., cf. Rath’s analysis in Rath 1994a, pp. 66–76. Cf. Erdmann 1892, section 5: “Logik und Psychologie”. Cf. Peckhaus 1999, and with emphasis on the motivations for Ernst Schr¨oder’s algebra of logic, Peckhaus 2004. For a comprehensive history of logic as foundational discipline for mathematics cf. GrattanGuinness 2000, and more programmatically, GrattanGuinness 2004. (Frege 1893, XIV–XV; Beaney 1997, pp. 201–202). For an analysis of Frege’s argument cf. Kusch 1995, pp. 30–41. The following discussion refers to Frege 1918, pp. 58–59; Frege 1997b, pp. 325–326. Cf. Frege’s review of Husserl’s Philosophie der Arithmetik (Frege 1894). There Frege writes, that Husserl’s attempt to found the notion of number belongs to a kind “in which this cleaning [of the objects from their specifics] is done in the psychological copper [Waschkessel]. [. . . ] The mixture between psychology and logic which is now so popular, gives good suds for this purpose” (Frege 1894, p. 316). Martin Kusch gives a comprehensive analysis of Husserl’s argumentation (cf. Kusch 1995). For the relation between Frege and Husserl cf. Dagfinn Føllesdal’s classic from 1958 (Føllesdal 1958, English Føllesdal 1994). The philosophical systems of Frege and Husserl are compared in Haaparanta 1994 and Hill 2000. Here are two examples: Eduard Beneke: “All reasoning is articulated in judgements” (Beneke 1832, section 19); Benno Erdmann: “In the following investigation reasoning in its wider meaning is understood as nothing else than judging” (Erdmann 1892, section 1.1). Cf. Brentano’s Die Lehre vom richtigen Schluß (Brentano 1956, pp. 29–30). (Popper 1934, p. 61). The German text is unclear in the central aspect of the distinction between intuition and sense perception. In writing “Die Anschauung, die Sinneswahrnehmung, [. . . ] ist ‘unmittelbare Erkenntnis”’ it remains open whether or not Popper identified intuition and sense perception.

REFERENCES Albert, Hans (1968), Traktat u¨ ber kritische Vernunft (T¨ubingen: J.C.B. Mohr, Paul Siebeck), 1991. Arnauld, Antoine and Nicole, Pierre (1996), Logic or the Art of Thinking. Containing, Besides Common Rules, Several New Observations Appropriate for Forming Judgement, transl. and ed. by Jill Vance Buroker (Cambridge: Cambridge University Press), first published as anon., La Logique ou l’Art de Penser contenant, outre les r`egles communes, plusieurs observations nouvelles, propres a` former le jugement (Paris: Charles Savreux 1662); last edition supervised by the authors 6th ed. (Amsterdam: Abraham Wolfgang 1685). Beaney, Michael (ed.) (1997), The Frege Reader. Selected Philosophical Writings (Oxford: Blackwell). Beneke, Friedrich Eduard (1832), Lehrbuch der Logik als Kunstlehre des Denkens (Berlin, Posen and Bromberg: Ernst Siegfried Mittler). Brentano, Franz (1874), Psychologie vom empirischen Standpunkte (Leipzig: Duncker & Humblot). Brentano, Franz (1956), Die Lehre vom richtigen Urteil. Nach den Vorlesungen u¨ ber Logik mit Ben¨utzung anderer Manuskripte aus dem Gebiete der Erkenntnistheorie aus dem Nachlaß, ed. by Franziska Mayer-Hillebrand (Bern: Francke Verlag). Eisler, Rudolf (1910), “Psychologismus”, in Eisler, W¨orterbuch der philosophischen Begriffe. Historisch-quellenm¨aßig bearbeitet, 3rd ed., Vol. 2 (Berlin: Mittler und Sohn), pp. 1088–1092. Erdmann, Benno (1892), Logik, Vol. 1: Logische Elementarlehre (Halle: Niemeyer). Erdmann, Johann Eduard (1866), Grundriss der Geschichte der Philosophie, Vol. 2: Philosophie der Neuzeit (Berlin: Hertz); 1870; 4th ed. by Benno Erdmann (Berlin: Hertz 1896). Føllesdal, Dagfinn (1958): Husserl und Frege. Ein Beitrag zur Beleuchtung der Entstehung der ph¨anomenologischen Philosophie (Oslo: Aschehoug); Avhandlinger utgitt av Det Norske VidenskapsAkademi i Oslo, Vol. II.2; English translation Føllesdal 1994.

114

VOLKER PECKHAUS

Føllesdal, Dagfinn (1994), “Husserl and Frege: A Contribution to Elucidating the Origins of Phenomenological Philosophy”, in Haaparanta 1994, pp. 3–47. Frege, Gottlob (1879), Begriffsschrift, eine der arithmetischen nachgebildete Formelsprache des reinen Denkens (Halle: Louis Nebert); reprinted in Frege 1977; English extract Frege 1997a. Frege, Gottlob (1893), Grundgesetze der Arithmetik. Begriffsschriftlich abgeleitet, Vol. 1 (Jena: Hermann Pohle, reprinted Hildesheim: Olms 1962). Frege, Gottlob (1894), Review of Husserl 1891, Zeitschrift f¨ur Philosophie und philosophische Kritik 103, 313–332. Frege, Gottlob (1918), “Der Gedanke”, Beitr¨age zur Philosophie des deutschen Idealismus 1 (1918/19), pp. 58–77; English translation Frege 1997b. Frege, Gottlob (1977), Begriffsschrift und andere Aufs¨atze. Dritte Auflage. Mit E. Husserls und H. Scholz’ Anmerkungen, ed. by Ignacio Angelelli (Darmstadt: Wissenschaftliche Buchgesellschaft). Frege, Gottlob (1997a), “Begriffsschrift: a Formula Language of Pure Thought Modelled on that of Arithmetic”, in Beaney 1997, pp. 47–78. Frege, Gottlob (1997b), “Thought”, in Beaney 1997, pp. 325–345. Fries, Jakob Friedrich (1828), Neue oder anthropologische Kritik der Vernunft, 2nd ed. (Heidelberg: Christian Friedrich Winter, 1st ed. Heidelberg: Mohr und Zimmer 1807); reprinted Fries, S¨amtliche Schriften. Nach den Ausgaben letzter Hand zusammengestellt, eingeleitet und mit einem Fries-Lexikon versehen, by Gert K¨onig and Lutz Geldsetzer, Sect. I, Vol. 4 (Aalen: Scientia 1967). Fries, Jakob Friedrich (1837), System der Logik. Ein Handbuch f¨ur Lehrer und zum Selbstgebrauch, 3rd rev. ed. (Heidelberg: Christian Friedrich Winter); 1st ed. Heidelberg: Mohr und Zimmer 1811; reprinted Fries, S¨amtliche Schriften. Nach den Ausgaben letzter Hand zusammengestellt, eingeleitet und mit einem Fries-Lexikon versehen, by Gert K¨onig and Lutz Geldsetzer, Sect. I, Vol. 7 (Aalen: Scientia 1971). Geldsetzer, Lutz (1990), “Metaphysische Tendenzen der philosophischen Entwicklung in der Bundesrepublik Deutschland”, in Die sog. Geisteswissenschaften: Innenansichten, ed. by Wolfgang Prinz and Peter Weingart (Frankfurt a.M.: Suhrkamp), Suhrkamp-Taschenbuch Wissenschaft, Vol. 854), pp. 419–447. Grattan-Guinness, Ivor (2000), The Search for Mathematical Roots 1870–1940. Logics, Set Theories and the Foundations of Mathematics from Cantor Through Russell and G¨odel (Princeton/Oxford: Princeton University Press). Grattan-Guinness, Ivor (2004), “The Mathematical Turns in Logic”, in Dov M. Gabbay/John Woods (eds.), Handbook of the History of Logic, Vol. 3: The Rise of Modern Logic: From Leibniz to Frege (Amsterdam et al.: Elsevier), pp. 545–556. Haaparanta, Leila (ed.), 1994), Mind, Meaning and Mathematics. Essays on the Philosophical Views of Husserl and Frege (Dordrecht, Boston and London: Kluwer); Synthese Library; Vol. 237. Hegel, Georg Wilhelm Friedrich (1807), System der Wissenschaft, Pt. 1: Ph¨anomenologie des Geistes (Bamberg and W¨urzburg: Joseph Anton Goebhardt); critical edition Hegel, Ph¨anomenologie des Geistes, ed. by Wolfgang Bonsiepen and Reinhard Heede (Hamburg: Felix Meiner 1980); Hegel, Gesammelte Werke, Vol. 9. Hegel, Georg Wilhelm Friedrich (1812/13), Wissenschaft der Logik, Vol. 1: Die objective Logik (N¨urnberg: Johann Leonhard Schrag); critical edition Hegel, Wissenschaft der Logik. Erster Band. Die objektive Logik (1812/1813), ed. by Friedrich Hogemann and Walter Jaeschke (Hamburg: Felix Meiner 1978); Hegel, Gesammelte Werke, Vol. 11. Hegel, Georg Wilhelm Friedrich (1830), Encyclop¨adie der philosophischen Wissenschaften im Grundrisse. Zum Gebrauch seiner Vorlesungen. Dritte Ausgabe (Heidelberg: Oßwald’scher Verlag); critical edition: Hegel, Enzyklop¨adie der philosophischen Wissenschaften im Grundrisse (1830), edited by Wolfgang Bonsiepen and Hans-Christian Lucas (Hamburg: Felix Meiner 1992); Hegel, Gesammelte Werke, Vol. 20. Hegel, Georg Wilhelm Friedrich (1832), Wissenschaft der Logik, Pt. 1: System der objectiven Logik, Vol. 1: Die Lehre vom Seyn (Stuttgart and T¨ubingen: Cotta); critical edition Hegel, Wissenschaft der Logik. Erster Teil. Die objektive Logik. Erster Band. Die Lehre vom Sein (1832), ed. by

DISTINCTION BETWEEN DISCOVERY AND JUSTIFICATION

115

Friedrich Hogemann and Walter Jaeschke (Hamburg: Meiner 1985); Hegel, Gesammelte Werke, Vol. 21. Herbart, Johann Friedrich (1813), Lehrbuch zur Einleitung in die Philosophie, K¨onigsberg: A.W. Unzer; critical edition Herbart, S¨amtliche Werke in chronologischer Reihenfolge, ed. by Karl Kehrbach and Otto Fl¨ugel, Vol. 4 (Langensalza: H. Beyer & S¨ohne 1891; reprinted Aalen: Scientia 1989, 2nd print), pp. 1–294; textcritical edition Lehrbuch zur Einleitung in die Philosophie. Textkritisch revidierte Ausgabe mit einer Einleitung, ed. by Wolfhart Henckmann (Hamburg: Felix Meiner 1993); Philosophische Bibliothek, Vol. 453. Hill, Claire Ortiz, and Rosado Haddock, Guillermo E. (2000), Husserl or Frege? Meaning, Objectivity, and Mathematics (Chicago and La Salle, IL: Open Court). Husserl, Edmund (1891), Philosophie der Arithmetik. Logische und psychologische Untersuchungen, Vol. 1 (Halle a.S.: C.E.M. Pfeffer); critical edition Husserliana. Edmund Husserl, Gesammelte Werke, Vol. 12, ed. Lothar Eley (Den Haag: Martinus Nijhoff 1970). Husserl, Edmund (1900), Logische Untersuchungen, Vol. 1: Prolegomena zur reinen Logik (Halle a.S.: Max Niemeyer); critical edition Husserliana. Edmund Husserl, Gesammelte Werke, Vol. 18, ed. by Elmar Holenstein (Den Haag: Martinus Nijhoff 1975). Kant, Immanuel (1787), Critik der reinen Vernunft, 2nd ed. (Riga: Johann Friedrich Hartknoch). Kraft, Julius (1932), Von Husserl zu Heidegger. Kritik der ph¨anomenologischen Philosophie (Leipzig: Hans Buske Verlag). Kusch, Martin (1995), Psychologism. A Case Study in the Sociology of Philosophical Knowledge (London and New York: Routledge); Philosophical Issues in Science. Lipps, Theodor (1880), “ Die Aufgabe der Erkenntnisstheorie und die Wundt’sche Logik”, Philosophische Monatshefte 16, 529–539. Lipps, Theodor (1912), Zur “Psychologie” und “Philosophie” (Leipzig: Engelmann); Psychologische Untersuchungen; Vol. 2.1. Mill, John Stuart (1843), A System of Logic, Ratiocinative and Inductive. Being a Connected View of the Principles of Evidence and the Methods of Scientific Investigation, 2 Vols. (London: J.W. Parker). Peckhaus, Volker (1997), Logik, Mathesis universalis und allgemeine Wissenschaft. Leibniz und die Wiederentdeckung der formalen Logik im 19. Jahrhundert (Berlin: Akademie Verlag); Logica nova. Peckhaus, Volker (1999), “19th Century Logic Between Philosophy and Mathematics”, Bulletin of Symbolic Logic 5, 433–450. Peckhaus, Volker (2002), “Regressive Analysis”, in Uwe Meixner and Albert Neven (eds.), Logical Analysis and History of Philosophy, Vol. 5 (Paderborn: Mentis), pp. 97–110. Peckhaus, Volker (2004), “Schr¨oder’s Logic”, in Dov M. Gabbay and John Woods (eds.), Handbook of the History of Logic, Vol. 3: The Rise of Modern Logic: From Leibniz to Frege (Amsterdam et al.: Elsevier), pp. 557–609. Pfeil, Hans (1934), Der Psychologismus im englischen Empirismus (Paderborn: Sch¨oningh); Forschungen zur neueren Philosophie und ihrer Geschichte, Vol. 5. Popper, Karl R. (1934), Logik der Forschung. Zur Erkenntnistheorie der modernen Naturwissenschaft (Wien: Springer) [on the title page: “1935”]. Popper, Karl R. (1974), “Intellectual Autobiography”, in Paul A. Schilpp (ed.), The Philosophy of Karl Popper, 2 Vols. (La Salle, IL: The Open Court), Vol. 1, pp. 3–181. Rabus, Georg Leonhard (1880), Die neuesten Bestrebungen auf dem Gebiete der Logik bei den Deutschen und Die logische Frage (Erlangen: Deichert). Rath, Matthias (1994a), Der Psychologismusstreit in der deutschen Philosophie (Freiburg and M¨unchen: Karl Alber Verlag). Rath, Matthias (1994b), “Von der Logik zur Psycho-Logik. Der Psychologismus seit Jakob Friedrich Fries”, Philosophisches Jahrbuch 101, 307–320. Reichenbach, Hans (1938), Experience and Prediction (Chicago: The University of Chicago Press). Riehl, Alois (1883), Ueber wissenschaftliche und nichtwissenschaftliche Philosophie. Eine akademische Antrittsrede (Freiburg i.Br. and T¨ubingen: Mohr).

116

VOLKER PECKHAUS

Schmidt, Nicole D. (1995), Philosophie und Psychologie. Trennungsgeschichte, Dogmen und Perspektiven (Reinbek bei Hamburg: Rowohlt Taschenbuch Verlag); rowohlts enzyklop¨adie, Vol. 556. Stumpf, Carl (1892), “Psychologie und Erkenntnistheorie”, Abhandlungen der philosophischphilologischen Classe der k¨oniglich Bayerischen Akademie der Wissenschaften 19, 467–516. Trendelenburg, Friedrich Adolf (1836), Elementa logices Aristotelicae. In usum scholarum ex Aristotele excerpsit, convertit, illustravit (Berlin: Bethge 1862, 1892). Trendelenburg, Friedrich Adolf (1842), “Zur Geschichte von Hegel’s Logik und dialektischer Methode. Die logische Frage in Hegel’s Systeme. Eine Auffoderung [sic!] zu ihrer wissenschaftlichen Erledigung”, Neue Jenaische Allgemeine Literatur-Zeitung 1, No. 97, 23 April 1842, pp. 405–408; No. 98, 25. April 1842, pp. 409–412; No. 99, 26 April 1842, pp. 413–414; separatly published as Trendelenburg 1843. Trendelenburg, Friedrich Adolf (1843), Die logische Frage in Hegel’s System. Zwei Streitschriften (Leipzig: Brockhaus). Windelband, Wilhelm (1904), “Logik”, in Wilhelm Windelband (ed.), Die Philosophie im Beginn des zwanzigsten Jahrhunderts. Festschrift f¨ur Kuno Fischer, Vol. 1 (Heidelberg: Carl Winter), pp. 163– 186.

PART III Revisions and Applications

PAUL HOYNINGEN-HUENE

CONTEXT OF DISCOVERY VERSUS CONTEXT OF JUSTIFICATION AND THOMAS KUHN

1. INTRODUCTION

Let me begin with a convention. I will refer to the distinction between the context of discovery and the context of justification as “the DJ distinction” (where I may note, for potentially misled younger readers, that this “DJ” has nothing to do with the music business). This paper is based on an older paper of mine (Hoyningen-Huene 1987). In the present paper, I will first recapitulate some of the topics of the older paper, and will contribute further considerations. Subsequently, I will discuss Thomas Kuhn’s ideas about justification in science. Thus will be clarified, in which sense precisely Kuhn opposed the DJ distinction. This is noteworthy, because in the 1960s and 1970s, many philosophers concluded from Kuhn’s opposition to the context distinction that he just did not understand what it was all about (and they inferred from this that he was just too uneducated as a philosopher to be taken seriously). My general line will be this: The DJ distinction, as it was used in the 1960s and 1970s, is not just one distinction, but a set of intermingled distinctions. Due to the conflation of various distinctions, the assertion of the DJ distinction contains hidden identity statements among these distinctions. This identification results in massive philosophical assumptions that are highly problematic. As a consequence, much of the discussion of the DJ distinction in the 1960s and 1970s is fairly muddled, because it is not clear what exactly is stated by its defenders and what exactly is attacked by its critics. Eventually, all parties, growing frustrated, turned away from the discussion. Earlier historical details of the DJ distinction will be provided in other chapters of this book. In section 2, I will discuss the varieties of the DJ distinction. Then, I will demonstrate how these distinctions incorporate several hidden assumptions (section 3). In section 4, I will present Kuhn’s somewhat opaque criticism of the DJ distinction as it was formulated in his 1962 The Structure of Scientific Revolutions. Section 5 is devoted to Kuhn’s positive views about justification in science. In the final section, I shall present a rejuvenated DJ distinction that might be acceptable to all parties.

2. THE VARIETIES OF THE DJ DISTINCTIONS

In this section, I shall distinguish five versions of the DJ distinction that can all be found in the literature. In order to make the DJ distinction as plausible as possible, it is

119 J. Schickore and F. Steinle (eds.), Revisiting Discovery and Justification, 119–131.
C 2006 Springer. Printed in the Netherlands.

120

PAUL HOYNINGEN-HUENE

useful to have one of the standard examples that guide the proponents, in mind. Very often, Kekul´e’s discovery of the ring structure of benzene serves as such an example. According to the standard story, Kekul´e who had pondered on the benzene structure for some time, dozed off in front of a fire place. While resting, the image of six carbon atoms forming a cycle appeared in his mind. Hence, the idea of a ring structure for a number of organic molecules was conceived. Subsequently, the community of organic chemists critically discussed whether the idea was right or wrong. In consideration of this example, the following apparently clear and plausible version of the DJ distinction emerges. Version 1 Discovery and justification are temporally distinct processes: At the beginning, something is discovered. Subsequently, it is justified (Mowry 1985, 79 calls this version of the DJ distinction the “standard formulation”). It may seem that this version of the DJ distinction does not have any empirical content, because it appears to be a conceptual consequence of the meaning of “justification” (see, for instance, Popper 1959 [1934], 31, as a clear example of this tendency). Whatever the concrete process of justification may consist of, it presupposes that there is something that has to be justified. Therefore, before the process of justification can begin, the thing to be justified has to be somehow present. Now, it is plausible that in science, anything that is in need of a justification has to be discovered; it is not simply given. At least, this approach is plausible if “discovery” is understood in a wide sense that includes “invention”. Claims in science that are in need of justification typically comprise new hypotheses, new theories, new models with certain properties, new classifications, new forms of representation, or new phenomena. It is obvious that in this particular version, the DJ distinction relies on a supposed difference between discovery processes and justification processes. If a discovery process could not be differentiated from a justification process, the distinction would collapse. There are two main objections to this distinction. The first states that phases of discovery and of justification may alternate, that is that the history of science is not a straightforward sequence of discovery and justification, but more complex (see, e.g., Feyerabend 1970, p. 70; Mowry 1985, p. 79). This presumed historical fact has been granted by some of the proponents of the DJ distinction (see, e.g., Salmon 1970, p. 71). In that, the proposed distinction is not challenged, but only a refinement is wanted. It is probable that establishing a complex item, such as a theory, happens as will be outlined as follows. First, a part of the theory is discovered and justified subsequently. Then, another part is discovered and justified subsequently, and so on, until the theory has been discovered and justified in full in this stepwise manner. The second objection is much more serious. It is doubtful that it is really possible to identify discovery and justification processes in the history of an item that is an unquestionable candidate for having been discovered and having been justified. As an example, let us take an empirical law (for more sophisticated and realistic examples, see specifically Arabatzis (this volume) and Steinle (this volume)). Clearly, empirical

CONTEXT OF DISCOVERY VERSUS CONTEXT OF JUSTIFICATION

121

laws have to be discovered (they often bear the names of their presumed discoverers), and clearly, they are in need of justification. Now, let us assume that the most recent history of an empirical law consists in the establishment of a higher degree of its quantitative accuracy due to new methods of measurement. How should these more accurate measurements be classified? Are they part of the discovery process of the quantitative refinement of the law? Or are they part of the justification process for the quantitative refinement of the law? It seems impossible to attribute these measurements uniquely to one or the other category. Thus, the identification of discovery processes as opposed to justification processes in the history of science is—at least in some cases—not possible. A typical defense of the DJ distinction against this objection allows for overlapping contexts, or even that “the process of discovery and the process of justification may be nearly identical” (Salmon 1970, p. 72). Although this may be a defense of some other version of the DJ distinction, it does not defend the version discussed here. In fact, it admits that the DJ distinction is insufficient as a distinction between processes of discovery and justification. Whatever the distinction, it is not a distinction between processes. Much of the critical discussion on the DJ distinction has focused on this version (e.g., Feyerabend), and other chapters of this book deal with it as well. As it is clear that this variant is not tenable, the DJ distinction can only be upheld in other versions than the present one. Version 2 The distinction concerns the process of discovery versus the methods (in a wide sense) of justification (or testing). Here, we have a contrast between the factual historical process and methods, considerations, procedures, etc. that are relevant to justify or to test knowledge claims (see, e.g., Feigl 1970, p. 4; Popper 1959 [1934], p. 31; Salmon 1970, pp. 68, 72; Scheffler 1967, pp. 69–73; Siegel 1980a, pp. 299–304; Siegel 1980b, pp. 369–372). Again, this distinction appears to be fairly clear. The part about the methods of justification or testing, however, is ambiguous. On the one hand, it may refer to methods of justification that were used at the time. This is probably the preferred reading by historians. These methods need to be discovered empirically, by historical work. On the other hand, the distinction may refer to methods of justification that “really” establish knowledge claims, independently of the beliefs of the historical actors. Most probably, philosophers committed to a normative philosophy of science will prefer this reading. Methods that “really” establish knowledge claims need to be justified philosophically, whatever that may mean. It is obvious that they cannot be established by any kind of historical work alone. There are problems with both readings. On the first reading, a similar concern to the main problem with version 1 of the DJ distinction presents itself. How can we distinguish historically used methods of justification from a supposed process of discovery, when such a process separate from justification often does not exist? The distinction presumes the possibility of sorting out scientific activities as belonging to either discovery or justification, and this is often impossible. This reading of the distinction does not work, therefore.

122

PAUL HOYNINGEN-HUENE

On the second reading, justification or testing is understood in a normative (or perhaps more precisely: in an evaluative) sense. Something can count as a justification or testing procedure only when its goal is attained, i.e., only if it really establishes justification or a test. On this reading, we must be in command of procedures that tell us how justification and testing must be done. At that, the DJ distinction turns into a special case of the distinction between the descriptive and the normative: historical processes (of discovery) are described, whereas claims of justification or testing are normatively evaluated. Certainly, the latter distinction is a clear one. However, important questions remain. How does one attain the norms for proper justification or testing? On what basis are the norms themselves justified? Are the norms invoked really timeless? Or are they subject to historical change? The following version of the DJ distinction is a methodological specification of the version just discussed. Version 3 The analysis of discovery is empirical, whereas the analysis of justification or testing is logical. In this version, the DJ distinction states a methodological difference on the meta-level, relative to an object-level of historical processes or justification procedures. All authors previously cited supporting version 2, also support version 3. The essence of version 3 is that descriptions have to be found empirically, whereas normative evaluations, i.e., whether or not some epistemic claims are justified, have to be carried out logically. As logic is a time-independent discipline, the justificatory procedures in this version are probably specified as not being subject to historical change; they are presumed to be timeless (unless the logical means employed for justification change over time). The next version of the DJ distinction does not introduce substantial novelty, but maps the present distinction in academic fields. Version 4 Within this version, the difference of history, psychology and sociology of science from philosophy of science is methodological: the former are empirical, the latter is logical. Empirical disciplines deal with the process of discovery, philosophy of science deals with the logical analysis of justification (testing) and is normative. Again, the quoted defenders of versions 2 and 3 also defend version 4. Although the methodological characterization of history, psychology and sociology of science as empirical is more or less unproblematic, it remains unclear, whether discovery processes (where they exist) can only be investigated empirically or not. Furthermore, the persuasiveness of the methodological characterization of philosophy of science as logical has waned substantially. Clearly, this characterization belongs to logical positivism and logical empiricism. By now, many philosophers consider these programs to be too restricted.

CONTEXT OF DISCOVERY VERSUS CONTEXT OF JUSTIFICATION

123

Version 5 Various authors introduce the DJ distinction as a distinction between different questions. However, they do not pay explicit attention to this fact most of the time. In order to promote the DJ distinction, they ask questions such as “What has happened historically during this discovery?” versus “Can a statement be justified? Is it testable?”, insinuating that the reader realizes the difference between these questions (for a clear example of this way of introduction see, e.g., Popper 1959 [1934], p. 31). In this version, the DJ distinction notes a difference between questions asked from the point of view of the meta-level. I shall discuss later whether or not the embodiment of the distinction into questions is significant. Looking back at the different versions we note that version 1 of the DJ distinction operates on the object-level: it distinguishes different kinds of historical processes. Version 2, first reading, also operates on the object-level: it distinguishes discovery processes and historically used methods of justification. Version 2, second reading, mixes object-level and meta-level: it contrasts discovery processes with normative reconstructions of justification. Versions 3–5 operate fully on the meta-level: they distinguish different kinds of analyses, meta-disciplines, or questions. 3. SOME HIDDEN ASSUMPTIONS

In the literature, versions 1–4 are typically conflated, as if we were dealing with one homogeneous DJ distinction only. What is implied in the conflation of the versions 1– 4 of the distinction? My assertion is that the conflation implicitly encloses substantial theories about discovery and about justification that the proponents of the conflated DJ distinction took more or less for granted. Let us look at the discovery and the justification sides in turn. Discovery side: The characterization of the process of discovery as subject to empirical investigation only (by psychology, history, etc.), and thereby excluding philosophy from its analysis, implies that the process of discovery has no features that can be subjected to any sort of non-empirical analysis. In other words, there cannot be a “logic of discovery” or a “rational heuristics”. This assumption was attacked especially by the protagonists of a so-called “logic of discovery” (Hanson 1971; Nickles 1980, pp. 22–25). The main argument is that there may indeed be structures of discovery that can be subjected to logical analysis; any assumption to the contrary is unfounded. Justification side: The conflation of versions 1–4 implies that there are justificatory processes in science, and that the only admissible methods of justification (testing) are logical. Philosophy of science, as a discipline, investigates this sort of justification. This position clearly reflects the program of some representatives of logical positivism (or logical empiricism) that conceived philosophy as logical analysis of language. The justification (testing) of some propositions becomes the analysis of the logical relations between this proposition and other propositions, i.e., mainly basic (or protocol) sentences.

124

PAUL HOYNINGEN-HUENE

As much as this assumption has been taken for granted in (some parts of analytic) philosophy, it is by no means philosophically innocent. One of the important implications is the following. A disagreement about the justification of an item can arise (1) out of a disagreement about basic sentences, or (2) a disagreement about conventions, or (3) by error. A disagreement over basic sentences can either be settled or emerges from diverging conventions, depending on the supposed nature of basic sentences. Thus, all disagreements arise either out of error or by adoption of differing conventions. Disagreements that are rooted in different conventions are not epistemically substantial. All epistemically substantial disagreements, therefore, are caused by error: At least one of the disagreeing parties commits a mistake, because it is impossible for both parties to be right in the same instance. In other words, a rational disagreement about justification is conceptually impossible. Put in a different idiom, the justificatory part of science is a one-person-game. This implies that in this part of science, there is no fundamental epistemic role for scientific communities, as opposed to individually working scientists. 4. KUHN’S CRITICISM OF THE DJ DISTINCTION

Towards the end of the introduction to his The Structure of Scientific Revolutions, Thomas Kuhn reflects upon the content of his book that he had just summarized in the preceding paragraphs. “History [. . . ] is a purely descriptive discipline. The theses [of Structure] are, however, often interpretive and sometimes normative. Again, many of my generalizations are about the sociology [. . . ] of scientists; yet at least a few of my conclusions belong traditionally to logic or epistemology. [. . . ] I may even seem to have violated the very influential contemporary distinction between the “context of discovery” and “the context of justification”. Can anything more than profound confusion be indicated by this admixture of diverse fields and concerns?” (Kuhn 1970, pp. 8–9)

In this paragraph, Kuhn articulates what some readers may have felt, with increasing uneasiness, when following Kuhn’s summary of Structure. Like many other paragraphs, he ends it by asking a rhetorical question that may indeed be some readers’ real question. And of course, in the next paragraph Kuhn provides a negative answer to this question: “For many years, I took [this distinction and others] to be about the nature of knowledge [. . . ]. Yet my attempts to apply them [. . . ] to the actual situations in which knowledge is gained, accepted, and assimilated have made them seem extraordinarily problematic. Rather than being elementary logical or methodological distinctions, which would thus be prior to the analysis of scientific knowledge, they now seem integral parts of a traditional set of substantial answers to the very questions upon which they have been deployed. That circularity does not at all invalidate them. But it does make them parts of a theory and, by doing so, subjects them to the same scrutiny regularly applied to theories in other fields.” (Kuhn 1970, p. 9)

This passage does not appear terribly clear to me.1 But before going into details of analysis, I want to insert a somewhat personal note. When I started writing the first draft of my book on Kuhn’s philosophy of science (Hoyningen-Huene 1993), my plan was to make this passage a central piece of my reconstruction of his theory. The reason

CONTEXT OF DISCOVERY VERSUS CONTEXT OF JUSTIFICATION

125

was straightforward. Here, at the end of his introduction, at a very prominent place, Kuhn describes a central difference of his work to the preceding epistemological tradition. This tradition lasted at least from Kant right to his days; both the logical empiricists and Popper made fundamental use of the DJ distinction. It defined the working area of philosophy of science. Thus, Kuhn seemed to be helpful enough to tell his readers what his most profound deviation from the philosophical tradition was. Therefore, this appeared to be an ideal entry point to an understanding of the philosophical underpinnings of his much discussed but—at least it seemed to me— poorly understood theory. When I told Kuhn about this plan in 1984, I was immensely surprised when he recommended that I should not focus on this passage, because it only constituted a “throw away remark”. He told me that its insertion was suggested to him by his then Berkeley colleague Stanley Cavell, in order to deal with anticipated criticism by philosophers of science. So, in a sense he did not take these remarks very seriously himself, as not being very illuminative for what he was really after. As Kuhn’s memories about details of the composition of Structure were, as he himself confessed repeatedly, not very reliable, it is worthwhile to look for confirming or disconfirming evidence for his story. Kuhn had finished a first draft of Structure that may be called “Proto-Structure” in the fall or early winter of 1960 (Kuhn 1960).2 It was mimeographed (the technical predecessor of Xeroxing) and distributed to some people, including Stanley Cavell, James Conant and Paul Feyerabend. The two paragraphs from which I quoted above are entirely missing from Proto-Structure. In their stead, there is the following note: [The final version will require one or two additional paragraphs at this point. They will describe footnote and bibliography policy, indicate the relation of this form of the monograph to its fuller version, justify the restriction to physical sciences, and attempt at least the most essential acknowledgments.] (Kuhn 1960, p. 10)

Obviously, Kuhn changed his mind about what should be inserted at this place, because what he mentions in this paragraph as required he finally dealt with in the Preface to Structure (there is no preface to Proto-Structure). In the preface, there is also an acknowledgment to Stanley Cavell that is noteworthy in our context. After gratefully noting the parallels of their views and their special mode of communication, he closes by stating that the latter “attests an understanding that has enabled him [Cavell] to point me the way through or around several major barriers encountered while preparing my first manuscript” (Kuhn 1970, xi). It is quite plausible that one of the barriers was the DJ distinction that appeared to forbid what Kuhn did in Structure, and that Cavell pointed him the way how to dismiss it. Now let us see what Kuhn’s criticism of the DJ distinction in Structure consists in. What does he mean by saying that the DJ distinction (and other similar distinctions) “[r]ather than being elementary logical or methodological distinctions [. . . ] seem integral parts of a traditional set of substantial answers to the very questions upon which they have been deployed” (Kuhn 1970, p. 9)? This is a rather convoluted

126

PAUL HOYNINGEN-HUENE

sentence that provokes the following queries. In which way can the distinctions mentioned be “deployed” upon questions? How can something that is part of an answer to a question be deployed to the very same question? What questions exactly is Kuhn talking about? These unanswered queries make the quoted sentence rather confusing. As there is no direct hint in the text on how to answer these queries, one has to make assumptions. My guess is that among the questions Kuhn mentions is the question “How does innovation occur in science?” When the standard DJ distinction (that conflates versions 1–4) is deployed to this question, the result looks as follows. First, the DJ distinction postulates that one should distinguish between a discovery part and a justification part of innovation. Then, the discovery part should be delegated to the empirical disciplines, whereas the justification part belongs to the business of philosophy of science that investigates it by logical analysis (remember the Kekul´e case where all this seems perfectly plausible). And indeed, Kuhn’s statement that this is not an application of “elementary logical or methodological distinctions”, but rather an “integral part of a traditional set of substantial answers” is right because this procedure is a part and parcel of logical positivism/empiricism. Furthermore, he is right in claiming that this substantial answer is “part of a theory”, and that this theory should be scrutinized. The theory he mentions is what I called, in the last section, the hidden assumptions built into the traditional (conflated) DJ distinction. This theory assumes that innovation is a two-step process containing discovery and justification phases, that discoveries have no structures that can be subjected to logical analysis, and that justifications can be fully understood by formal logical analysis. As is often the case with the Kuhn of 1962, philosophically he is on the right track (or rather: on an interesting track!). However, to put it mildly, he is not very explicit (because he himself rather feels his way instead of having fully analyzed it). His pertinent sentences flow nicely and seem uncomplicated, but in truth they are opaque and, ironically, by their very stylistic qualities they are easy to misunderstand. 5. KUHN’S VIEW OF JUSTIFICATION IN SCIENCE

Kuhn’s own view of justification is different from the logical positivist/empiricist picture. For them, justification ultimately uses as exclusive means formal logics and basic/protocol sentences. Before discussing Kuhn’s view, we should note that Kuhn’s explicit discussion of the DJ distinction and his deviating view takes place exclusively in the context of theory choice. Of course, there are other occasions in science were talk of justification is appropriate. Quite certainly, Kuhn would have similar views about some of them as the logical positivist/empiricists, say with respect to the justification of a certain mathematical approximation procedures. But in the case of the choice of general hypotheses, especially of theories, both the DJ distinction is most plausible and the contrast between Kuhn and the alternative philosophies is sharpest. For Kuhn, justification of theory choice in science uses means that, according to the standard DJ distinction, qualify as belonging to the context of discovery. Thus, for him the distinction must be invalid. I will not, however, pause on discussing and refuting the standard clich´es about Kuhn, namely his alleged irrationalism, subjectivism,

CONTEXT OF DISCOVERY VERSUS CONTEXT OF JUSTIFICATION

127

relativism, and so forth. Instead, I will directly present his position regarding theory choice justification. What are those means, usually counted as belonging to the context of discovery that, according to Kuhn, are relevant in the context of justification? In the end, Kuhn claims, it is a set of communal cognitive values that determines the outcome of the theory choice situation (Hoyningen-Huene 1993, pp. 147–154, 239–245). Two things are particularly remarkable about this set of values. First, this set of values changes over time and it is specific for the community in question. Thus, one may characterize a specific scientific community by the epistemic values it is committed to. Indeed, it is one of the principal sociological means to characterize any communities or groups by the specific values (or norms) that hold for them. In other words, these characteristics of scientific communities are traditionally counted as sociological and thus, cannot belong, according to the standard DJ distinction, to the context of justification. Second, although the community as a whole may be characterized by these values, each individual member of this community will specifically shape these values, both with respect to their precise content and their mutual weight. But the important fact about these individual variations of the communal epistemic values is that these differences become unimportant when the theory choice situation comes to a close. This is, in fact, analytically true. The theory choice situation only comes to a close when a consensus of the community about the best theory is reached. But each individual member of the community evaluates the candidates according to his or her individually shaped value system. So, a consensus can only be reached if in spite of these individual value differences, agreement emerges about which theory is the best and should therefore be accepted (whatever “acceptance” really means for the individual researcher in terms of commitment). In spite of their judging from slightly different viewpoints, i.e., from slightly different value systems, almost all community members come to the same conclusion regarding the winning theory. This underscores that Kuhn’s theory of justification is by no means psychological, but sociological. But there is an obvious objection to this account. The objection states that Kuhn’s account fails to really address the context of justification and instead, addresses a “context of decision”, describing a factual decision process about theory acceptance by a scientific community (Siegel 1980b, pp. 370–371; Siegel 1980a, pp. 310–312). This objection, however, underestimates the thrust of Kuhn’s account. Kuhn does not only intend to neutrally describe the actual decision procedures in science, but he also argues that this is a justified decision procedure, that this is the way that science should actually be done, or in other words, that the procedure is a rational one because there are good reasons for it. I have italicized some words that are often used by the proponents of the DJ distinction when arguing that something does indeed belong to the context of justification. Why does Kuhn think that this sort of decision procedure is actually a rational one? What sort of rationality is pertinent here? We are dealing here with simple meansends rationality. It is claimed that the decision procedure is rational because it is directed at the cognitive goals of science, that is roughly the invention and improvement of explanatory and often predictive theories. First, the epistemic values do indeed

128

PAUL HOYNINGEN-HUENE

represent the goals of science. According to these values, theories should be accurate, internally and externally consistent, as simple as possible, fruitful, and they should have a broad scope of application. These virtues of theories may be condensed into the master value of high problem solving capacity. It is also possible to see these values as an operationalization of science’s quest for truth, where the historical and communal specificity of these values reflects the particular epistemic situation of that community (Hoyningen-Huene 1992, pp. 496–499). Second, it must be shown that also the individual value differences that lead to disagreement in the phase of extraordinary science but disappear from the result of a communal theory choice, are rational means towards the cognitive goals of science. The main idea here is that these differences make a rational disagreement during the phase of extraordinary science possible. This disagreement is vital for the distribution of risk in a situation of epistemic uncertainty as no one knows, which candidate for paradigmatic theory will be successful. Thus, it is reasonable that different scientists try out a variety of possibilities in order for the community to have a wide spectrum of competing alternatives to explore (Hoyningen-Huene 1992, pp. 493–496). This closes the argument for the rationality of the theory choice decision procedure. As it was demonstrated, Kuhn does indeed engage in questions of justification of theory choice. He does not simply describe the history of science, either particularities or generalizations about its course. But then the question arises in which sense exactly Kuhn opposes to the DJ distinction because he does engage in the discussion of the rationality of justification procedures as opposed to purely historical work. In fact, this question does not only belong to Kuhn philology but raises a broader and more important issue. What remains of the DJ distinction if one removes the conflations discussed in section 3 to which also Kuhn opposed? Is there some core of the distinction that has in fact survived the attacks of historically minded philosophers and that should survive them because its philosophical substance is important? 6. THE DJ DISTINCTION, REJUVENATED

Actually, I do believe that there is a core of the DJ distinction that has, to the best of my knowledge, never been attacked in the discussion about it. This core is distributed among the versions 2 and 5 that I discussed in section 2. What I have in mind is an abstract distinction between the factual on the one hand, and the normative or evaluative on the other hand. This is a distinction of two perspectives that can both be taken regarding scientific knowledge, especially epistemic claims (but also about claims of differing characteristics such as legal, moral or aesthetic claims). From the descriptive perspective, I am interested in facts that have happened, and their description. Among these facts may be, among other things, epistemic claims that were put forward in the history of science, that I may wish to describe. From the normative or evaluative perspective, I am interested in an evaluation of particular claims. In our case, epistemic claims, for instance for truth, or reproducibility, or intersubjective acceptability, or plausibility, and the like are pertinent. Epistemic norms (in contrast to, say, moral or aesthetic norms) govern this evaluation. By using epistemic norms we can evaluate particular epistemic claims according to their being justified or not.

CONTEXT OF DISCOVERY VERSUS CONTEXT OF JUSTIFICATION

129

A perspective is a particular way of looking at an object. A perspective is actively chosen by an epistemic subject; it is not simply given by the object in question. A given perspective singles out certain aspects of the object in question as important, at the expense of others. The choice of a perspective is well expressed by a question because questions underscore the activity on the part of the questioner. Thus, it is not accidental that the introduction of the two perspectives underlying the DJ distinction is often accomplished by posing the respective questions. As I have noted in section 2, version 5, many authors do indeed introduce the DJ distinction by posing questions that make aware of the factual and the normative or evaluative perspective and their difference. In his arguments against the DJ distinction, Kuhn has never challenged the difference between the factual and the normative or evaluative perspectives; neither has Feyerabend nor any other critic. It also seems that the distinction between describing a claim and evaluating it is indeed very solid. But I should add immediately three remarks in order to prevent misunderstanding. First, the statement of the mere contrast between the descriptive and the normative or evaluative perspectives does not commit to any assumptions about the nature of facts or of descriptions, nor about the nature of norms or justification. It is only the difference between these perspectives that is abstractly stated, and not yet any content about what this difference more concretely consists in. Hence, by distinguishing the two perspectives we are not committed, for instance, to the position that norms and facts are absolutely separated and that they have nothing in common whatsoever. More to the point, it is emphatically not excluded that epistemic norms have factual presuppositions or implications, nor that facts have normative presuppositions or implications. Second, the question approach to the DJ distinction makes particularly clear that the overlap problem of the process distinction (version 1) is in fact a pseudo-problem. The reason is that different questions may receive the same answer without any danger of the two questions being confused with one another. In spite of getting the same answer, the questions “What is 5 plus 4?” and “What is 3 times 3?” remain distinct from one another. The same applies to a question that seeks a description and one that seeks a justification for some claim. Answers to these questions may be similar, even identical in some cases without blurring the difference between the questions at all. Third, the distinction between descriptive and normative or evaluative perspectives is not exhaustive. Other perspectives are also possible. In principle, there is space for those critics of the DJ distinction who claimed that it should be expanded to be threefold or even fourfold (see, e.g., Blackwell 1980; Curd 1980; Kordig 1978, pp. 114–116; Laudan 1977, pp. 108–114; McLaughlin 1982; Nickles 1980, pp. 18–22; Schaffner 1980, pp. 178– 200). If the DJ distinction is rejuvenated in the proposed way, one gets a distinction that is very lean. It is not loaded with potentially controversial philosophical theories about discovery or about justification, as the traditional conflated DJ distinction is. It is neither biased towards the program of logical empiricism, nor does it presuppose or imply controversial theories about the relationship between facts and norms. The lean distinction, thus, appears to be neutral and acceptable to everyone, at least to those who raised their voices in the extended discussion in the 1960s and 1970s. On

130

PAUL HOYNINGEN-HUENE

the one hand, it should be acceptable to the critics of the traditional DJ distinction as they either attacked the process version of the DJ distinction or the conflation of different other versions of it. But the critics never doubted the difference between the descriptive and the normative or evaluative perspectives. On the other hand, it should be acceptable to the defenders of the DJ distinction because the rejuvenated version preserves the core of what they fought for: that there is a distinct normative perspective that aims at the evaluation of scientific claims. Perhaps an acceptance of the rejuvenated DJ distinction will put an end to statements of despair like the one by Herbert Feigl: “I confess I am dismayed by the amount of—it seems almost deliberate—misunderstanding and opposition to which this distinction has been subjected in recent years” (Feigl 1970, p. 4). ACKNOWLEDGMENTS

I wish to thank all the other contributors to this volume, because they all helped improve the quality of this chapter—as far as this was possible. With respect to the English writing, I would like to thank Dr. Maya Shaha cordially for substantial improvements. NOTES 1. This passage is not only not terribly clear; it was also shocking to some logical empiricists. For instance, Wesley Salmon described his reaction to this passage as follows: “On my first reading of Thomas S. Kuhn’s The Structure of Scientific Revolutions (1962) I was so deeply shocked at his repudiation of the distinction between the context of discovery and the context of justification that I put the book down without finishing it” (Salmon 1991, p. 325). (I rediscovered this reference in chapter 11 a book manuscript by Hal Brown entitled Conceptual Systems). 2. This is what Kuhn wrote to me in a letter of 26 May 1994: a “draft that I had typed up, I should guess, in the fall or early winter of 1960”. But now it seems improbable to me that Kuhn had completed Proto-Structure before Spring 1961. According to recent archival studies in the Harvard Archives (Driver-Linn 2003, pp. 272 fn. 2), on April 22, 1961, Kuhn had sent “a draft of the Structure manuscript”, i.e., Proto-Structure, to James Conant, then President of Harvard University “with a letter inviting criticism and making an appeal for Conant’s endorsement to a publisher”. Why should Kuhn have waited for several months before sending Proto-Structure to Conant to invite his criticism, possibly to be incorporated into the final version of the book? In addition, the two letters that Feyerabend sent to Kuhn in response to receiving Proto-Structure were most probably written in Spring (or Summer) 1961 (Hoyningen-Huene 1995, pp. 353–354). Given Feyerabend’s style of work and temperament, I consider it unlikely that he delayed his reaction to Proto-Structure for several months.

REFERENCES Blackwell, R. J. (1980), “In Defense of the Context of Discovery”, Revue Internationale de Philosophie 34: 90–108. Curd, Martin V. (1980), “The Logic of Discovery: An Analysis of Three Approaches”, in T. Nickles (ed.), Scientific Discovery, Logic, and Rationality. Boston Studies in the Philosophy of Science Vol. 56 (Dordrecht: Reidel), pp. 201–219. Driver-Linn, Erin (2003), “Where Is Psychology Going? Structural Fault Lines Revealed by Psychologists’ Use of Kuhn”, American Psychologist 58 (4): 269–278.

CONTEXT OF DISCOVERY VERSUS CONTEXT OF JUSTIFICATION

131

Feigl, Herbert (1970), “The “Orthodox” View of Theories: Remarks in Defense as well as Critique”, in M. Radner and S. Winokur (eds.), Analyses of Theories and Methods of Physics and Psychology. Minnesota Studies in the Philosophy of Science Vol. 4 (Minneapolis: University of Minnesota Press), pp. 3–16. Feyerabend, Paul K. (1970), “Against Method: Outline of an Anarchistic Theory of Knowledge”, in M. Radner and S. Winokur (eds.), Analyses of Theories and Methods of Physics and Psychology. Minnesota Studies in the Philosophy of Science Vol. 4 (Minneapolis: University of Minnesota Press), pp. 17–130. Hanson, Norwood Russel (1971), “The Idea of a Logic of Discovery”, in N. R. Hanson (ed.), What I Do Not Believe and Other Essays. (Dordrecht: Reidel), pp. 288–300. Hoyningen-Huene, Paul (1987), “Context of Discovery and Context of Justification”, Studies in History and Philosophy of Science 18: 501–515. Hoyningen-Huene, Paul (1992), “The interrelations between the philosophy, history and sociology of science in Thomas Kuhn’s theory of scientific development”, British Journal for the Philosophy of Science 43: 487–501. Hoyningen-Huene, Paul (1993), Reconstructing Scientific Revolutions. Thomas S. Kuhn’s Philosophy of Science (Chicago: University of Chicago Press). Hoyningen-Huene, Paul (1995), “Two Letters of Paul Feyerabend to Thomas S. Kuhn on a Draft of The Structure of Scientific Revolutions”, Studies in History and Philosophy of Science 26 (3): 353–387. Kordig, C. R. (1978), “Discovery and Justification”, Philosophy of Science 45: 110–117. Kuhn, Thomas S. (1960), The Structure of Scientific Revolutions: Unpublished manuscript, 178 pp., referred to here as Proto-Structure). Kuhn, Thomas S. (1970), The Structure of Scientific Revolutions (Chicago: University of Chicago Press). Laudan, Larry (1977), Progress and its Problems. Towards a Theory of Scientific Growth (Berkeley: University of California Press). McLaughlin, Robert (1982), “Invention and Appraisal”, in R. McLaughlin (ed.), What? Where? When? Why? Essays on Induction, Space and Time, Explanation (Dordrecht: Reidel), pp. 69–100. Mowry, Bryan (1985), “From Galen’s Theory to William Harvey’s Theory: A Case Study in the Rationality of Scientific Theory Change”, Studies in History and Philosophy of Science 16 (1): 49–82. Nickles, Thomas (1980), “Introductory Essay: Scientific Discovery and the Future of Philosophy of Science”, in T. Nickles (ed.), Scientific Discovery, Logic, and Rationality. Boston Studies in the Philosophy of Science Vol. 56 (Dordrecht: Reidel), pp. 1–59. Popper, Karl R. (1959 [1934]), The Logic of Scientific Discovery (London: Hutchinson). Salmon, Wesley C. (1970), “Bayes’s Theorem and the History of Science”, in R. H. Stuewer (ed.), Historical and Philosophical Perspectives of Science. Minnesota Studies in the Philosophy of Science Vol. 5 (Minneapolis: University of Minnesota Press), pp. 68–86. Salmon, Wesley C. (1991), “The Appraisal of Theories: Kuhn meets Bayes”, in A. Fine, M. Forbes & L. Wessels (eds.), PSA 1990: Proceedings of the 1990 Biennial Meeting of the Philosophy of Science Association, Vol. 2 (East Lansing: Philosophy of Science Association), pp. 325–332. Schaffner, Kenneth F. (1980), “Discovery in the Biomedical Sciences: Logic or Irrational Intuition?” in T. Nickles (ed.), Scientific Discovery: Case Studies. Boston Studies in the Philosophy of Science Vol. 60 (Dordrecht: Reidel), pp. 171–205. Scheffler, Israel (1967), Science and Subjectivity (Indianapolis: Hackett). Siegel, Harvey (1980a), “Justification, Discovery and the Naturalizing of Epistemology”, Philosophy of Science 47: 297–321. Siegel, Harvey (1980b), “Objectivity, Rationality, Incommensurability, and More”, British Journal for the Philosophy of Science 31: 359–384.

THOMAS STURM AND GERD GIGERENZER

HOW CAN WE USE THE DISTINCTION BETWEEN DISCOVERY AND JUSTIFICATION? ON THE WEAKNESSES OF THE STRONG PROGRAMME IN THE SOCIOLOGY OF SCIENCE

Der geschichtliche Ausgangspunkt erscheint vom logischen Standpunkte aus meistens als etwas Zuf a¨ lliges. Gottlob Frege1

Ever since Kuhn’s The Structure of Scientific Revolutions (Kuhn 1962, 1970), many philosophers, historians, and sociologists of science have attacked the distinction between discovery and justification (the DJ distinction). It has been argued that the distinction cannot be drawn precisely; that it cannot be drawn prior to the actual analysis of scientific knowledge; that it is useless for the analysis of scientific knowledge; and that perhaps there is no such distinction at all. Other critics, instead of trying to blur or to reject the distinction, claim that we need an even more fine-grained distinction. A variety of concepts such as generation, invention, prior assessment, evaluation, test, proof, and so on, is needed, depending on the different kinds of questions we can raise concerning scientific research and its results (e.g., Nickels 1980b, pp. 18–22; Hoyningen-Huene 1987, pp. 507–509). There is more insight in the latter kind of criticism than in the former ones. For instance, by means of more fine-grained distinctions one can avoid certain problems with the term ‘discovery’. Such talk tends to imply success: In many ordinary usages, ‘to discover’ implies ‘to know’ and may therefore mean that newly discovered theories are already justified (Curd 1980, p. 201; Nickles 1980, p. 9f.; Papineau 2003, p. 141). In cases where this is appropriate, one may look for an analysis of rational ingredients in the development of scientific claims. In other cases, we need a more neutral notion such as that of the generation of a knowledge claim. Furthermore, it is problematic to say that a knowledge claim, let alone a whole theory, can be discovered at all. We may sometimes say that a certain object, property, process, cause or effect (the planet Neptune, the identity of Shakespeare, the effect of electromagnetism on light, the causes of World War I, the blind spot in the visual field, and so on) has or has not been discovered. But there are problems in saying that scientific claims are ever discovered. To begin with, it is usually odd to say that the statement “Edward de Vere was Shakespeare” was discovered, while it is appropriate to say that “The truth of ‘Edward de Vere was Shakespeare’ was discovered” or “It was discovered that Edward de Vere was Shakespeare”. Moreover, if to discover implies to know, then we must say that most knowledge claims are rarely found to be true at an instant; they

133 J. Schickore and F. Steinle (eds.), Revisiting Discovery and Justification, 133–158.
C 2006 Springer. Printed in the Netherlands.

134

THOMAS STURM AND GERD GIGERENZER

achieve communal acceptance frequently only over a long period of time. Clearly, there are cases in between the extremes. The true identity of William Shakespeare might be discovered through some hitherto unknown document the reliability of which is not doubted either by the adherents of William Shakespeare of Stratfordupon-Avon, or by the defenders of Edward de Vere, 17th Earl of Oxford. Nevertheless, with regard to many claims, we wish to be able to say that they have been developed without implying that they are accepted as true, let alone that they are true. With respect to whole theories, the latter implication has been even criticized even by moderate scientific realists, who otherwise think that some theoretical claims can be true (Kr¨uger 1983). Hence, it is often more appropriate to speak of the generation of claims or theories. Such preliminary considerations should not be understood to imply a definition of the DJ distinction that is appropriate for all the different questions we can raise concerning scientific research. Kant remarked that it is misguided to start philosophical investigations with definitions (Kant 1781/1787, A730-31/B758-59). A core element of the DJ distinction, however, can be stated and will be used in what follows. Basically, the DJ distinction may be understood as a distinction on an object-level, between different events in the research process, or as a distinction on a meta-level, pertaining to our statements about, or even the disciplines concerned with, discoveries and justifications. Some authors have combined these different versions. Here we will use a particularly lean version (see Hoyningen-Huene, this volume). The point of this version, which operates on the meta-level, is that we should distinguish between different types of questions: For any given claim p, we can always ask, “How did someone come to accept that p?” This question, which may be understood as a question about the generation or actual acceptance of a claim, differs in principle from the question “Is p justified?” A core point of this distinction is to warn us against the genetic fallacy: From the fact that a certain claim is accepted as true, it does not follow that it is true. Hence, the question of its justification arises almost naturally. What is also important about the lean DJ distinction is that it is not affected by objections that have been raised against stronger versions of the DJ distinction. For instance, it does not imply a process distinction, let alone a specific temporal order of relevant processes, such as that discovery has to precede justification. Therefore, Knorr-Cetina’s and Kantorovich’s objection that the two contexts are not strictly separable does not apply to the lean DJ distinction (Knorr-Cetina 1984, pp. 28–31; cf. Kantorovich 1993, p. 101f.). Kantorovich merely points to historical examples where discovery and justification overlap to support his claim, obviously having in mind a version of the distinction on an object-level. Furthermore, the lean DJ distinction does not imply a distinction between different disciplines, such as philosophy of science versus history and sociology of science (Nickles 1980a, pp. 8–18; Siegel 1980; Hoyningen-Huene, this volume). Our questions may be too complex to allow such neat disciplinary divisions. But, it is one thing to say that a core meaning of the DJ distinction can be defended against misunderstandings. It is another thing to show that the distinction in its core meaning is useful. What could be the relevance of the mentioned difference of questions concerning knowledge claims? Basically, there are two options here,

HOW CAN WE USE THE DJ DISTINCTION

135

an evaluative and an explanatory use. First, the lean DJ distinction may be used to emphasize the possibility of a genuinely justificatory or evaluative point of view. The lean DJ distinction supports the idea that the reasons given by scientists can be reflected critically. One does not thereby commit oneself to any specific scientific methodology; the point is only that methodological debates can make sense. This is, of course, the most widespread use of the distinction. However, we shall not consider it here but rather turn to the second, explanatory use. The lean DJ distinction may also be used in order to emphasize that justificatory reasons—as opposed to mere causal conditions—played a role in the emergence of a scientific claim and its factual acceptance in the scientific community. But if we restrict ourselves to an account of the generation of knowledge claims, why should we bother to find out whether and how these claims are justified according to the members of the scientific community? A view that is of particular interest in this regard is the so-called “Strong Programme” in the sociology of science (SP), because it implies a thoroughgoing skepticism concerning the DJ distinction. As is well known, champions of the SP are not particularly charitable towards traditional philosophical epistemology or towards normative orientations within philosophy of science. Moreover, they advocate that an explanation of science should not take into account justificatory reasons of scientists. Accordingly, they have questioned that the DJ distinction has any use at all (e.g., Barnes, Bloor and Henry 1996; Knorr-Cetina 1981; Knorr-Cetina and Mulkay 1982; Pickering 1992; Schaffer 1994). Philosophical critics frequently discuss claims of the SP concerning scientific rationality and realism (e.g., Arabatzis 1994; Fine 1996; Hacking 1999; Haddock 2004; Kitcher 1993, pp. 160–169, 184f.; Nelson 1994). However, the related and even more basic issue of how the SP treats the DJ distinction has not been addressed yet. The critics of the SP who aim to defend the project of a critical or normative evaluation of scientific claims and their credentials frequently emphasize that this project must not be conflated with the different task of explaining how scientists acquire their beliefs. In other words, they presuppose the lean DJ distinction. Already because of this debate, it is helpful to clarify how the SP criticizes the DJ distinction and whether this criticism also affects the lean DJ distinction. Moreover, it seems useful to reflect whether or not the (lean) DJ distinction is useful in explanatory contexts and not only in evaluative ones. The main constructive goal of the SP is to provide a framework for the empirical explanation of scientific research, according to which normative considerations are not supposed to play a role in the explanans statements. Against this, we shall argue that appropriate explanations of science should, at least oftentimes, take into account that scientists acquire their beliefs based upon reasons that are often reflected and made explicit. The justificatory reasons of scientists can, and indeed should, be taken seriously even if one’s concern in science studies is merely explanatory and not evaluative. Important details of the history of science, such as choices between basic theoretical concepts and claims, are not adequately explained if one does not acknowledge these reasons. Of course, the justificatory reasons scientists give for their beliefs do by no means always explain what is going on in science. It is an ongoing task for historians, philosophers, and

136

THOMAS STURM AND GERD GIGERENZER

sociologists to explore whether, and to what extent, scientific beliefs are adopted for justificatory reasons or due to other, non-rational factors. The point is that accepting this task already commits one to the difference of questions that constitute the lean DJ distinction. Viewing the justificatory reasons of scientists as causes of their beliefs does, of course, not commit one to accept these reasons or beliefs oneself. The critical or normative evaluation of those reasons and beliefs remains an additional, different task. We shall argue, first, that while Barnes and Bloor claim to be following Kuhn in rejecting the DJ distinction, they actually advance a new objection. They claim that the distinction is useless (section I). Next, we discuss some basic and general problems of their arguments and show that they do not undermine the lean DJ distinction (section II). Finally, a case study will be used in order to show positively that, and how the lean DJ distinction is useful for appropriate explanations of theory generation. Our case study comes from the cognitive sciences, but we are optimistic that similar examples can be found elsewhere (section III). I. THE STRONG PROGRAMME’S OBJECTION AGAINST THE DJ DISTINCTION

1. Social Constructivism and the SP A prefatory remark on the notion of social constructivism is in place, since the SP and social constructivism are often identified with one another. Social constructivism is also frequently characterized in such a way that it seems to undermine the DJ distinction. Social constructivists are often characterized as saying—and surely they come at least close to claiming—that no part of reality (objects, properties, effects, causes . . . ) can really be discovered, and that no knowledge claim can really be justified; rather, reality does not exist prior to its social construction or invention, and so our knowledge is not really about an independent reality either. However, there are many versions of social constructivism (Hacking 1999). It has become too fashionable in recent decades to say that something or other (nature, childhood, the subject, population statistics, quarks, madness . . . ) is socially constructed. Social constructivists have rarely reflected on what it is that could really be socially constructed, and whether constructions are really all of the same kind.2 Furthermore, not all who call themselves social constructivists have developed claims of philosophical importance. However, the defenders of the SP such as Barnes and Bloor have clearly done so. One concerns their attitude towards the DJ distinction. 2. Differences Between the SP and Kuhn What are the objections against the DJ distinction of adherents of the SP? In an influential paper, Barnes writes: “Throughout the paper the incidence and justification of new cultural elements has been thought of in naturalistic terms: the idea that there exist general ‘logics’ of discovery or of justification has been rejected. At present, social scientists and philosophers of science tend to agree that, although the origin of new ideas may be influenced by any number of contingent factors, there does exist an absolute

HOW CAN WE USE THE DJ DISTINCTION

137

logic of justification in science. Hence the sociologist may usefully study the context of discovery in science, but the context of justification therein can only profitably be investigated philosophically [. . . ]. In contrast, it is argued here that the scope of sociological explanation extends throughout the entire context of justification in natural science.” (Barnes 1972, p. 391)

As the title of his essay—“Sociological Explanation and Natural Science: A Kuhnian Reappraisal”—already reveals, Barnes claims to be transferring Kuhn’s views to the sociology of science. However, there are important differences between their views. To begin with, in Structure Kuhn advances two different criticisms of the DJ distinction. On the one hand, he thinks that the distinction is drawn too sharply; on the other hand, he claims that it is a mistake to presuppose the distinction in an investigation of scientific knowledge. As to the first point, Kuhn thinks that the DJ distinction expresses a misguided conception of the history of science and related disciplines such as the sociology of science. He writes: “History, we too often say, is a purely descriptive discipline. The theses suggested above are, however, often interpretive and sometimes normative. Again, many of my generalizations are about the sociology or social psychology of scientists; yet at least a few of my conclusions belong traditionally to logic or epistemology. In the preceding paragraph I may even seem to have violated the very influential contemporary distinction between ‘the context of discovery’ and the ‘context of justification’. Can anything more than profound confusion be indicated by this admixture of diverse fields and concerns?” (Kuhn 1970a, pp. 8–9)

Kuhn’s considerations here consist of two points: (1) He interprets the DJ distinction as a distinction between normative and interpretive disciplines. (2) He claims that some of his conclusions drawn from history, social psychology and sociology of science, belong to what has been traditionally conceived of as claims of normative disciplines, such as logic or epistemology.3 At this point, Kuhn does not argue for either (1) or (2). However, on the basis of these claims, he suggests that the DJ distinction is drawn too sharply.4 Kuhn indicates a second objection. Having suggested that the distinction is drawn too sharply, he states that it might still have “something important to tell us”, if “appropriately recast” (Kuhn 1970a, p. 9). Yet, it seems hard to apply the distinction to “the actual situations in which knowledge is gained, accepted” and so on (ibid.). He points out that rather “than being elementary logical or methodological distinctions, which would thus be prior to the analysis of scientific knowledge, they now seem integral parts of a traditional set of substantive answers to the very questions upon which they have been deployed. That circularity does not at all invalidate them. But it does make them part of the theory and, by doing so, subject to the same scrutiny regularly applied to theories in other fields.” (Kuhn 1970a, p. 9)

It is suggested here that it is a mistake to draw the DJ distinction “prior to the analysis of scientific knowledge”. But what exactly are the questions that the DJ distinction is supposed to answer? How can a revised version of the distinction come out of an empirical analysis of science? Kuhn’s considerations are not examples of crystalline clarity, and the lean DJ distinction is hardly affected by them (see Hoyningen-Huene, this volume). In any case, none of his objections is identical with the claim that the DJ

138

THOMAS STURM AND GERD GIGERENZER

distinction is useless. Indeed, Kuhn says the opposite, if in an indeterminate manner: The distinction, “appropriately recast”, might “still have something important to tell us” (Kuhn 1970a, p. 9). An additional difference between Kuhn and the SP concerns not the attitude towards the DJ distinction as such, but the possibility and the role of normative theories of epistemic justification. While there are, in Kuhn’s view, no “neutral algorithms for theory choice”, he thinks that there could be historically based norms and values of science. What normative theorists nowadays can look for are maxims of justification: Rules that guide scientific research into certain directions and that place constraints upon what steps count as rational given certain goals (Kuhn 1970b, p. 237; Kuhn 1977; Siegel 1980, p. 301f.; Hoyningen-Huene, this volume). Kuhn could accept such maxims, both for questions concerning the generation and the evaluation of scientific claims. Hence, at least a lean version of the DJ distinction could still make sense: It is one thing to ask why a person came to accept that p, and another thing to ask whether p is justified—or whether p was justified for that person, given her maxims. For the friend of the SP this is not possible. Barnes points to arguments concerning underdetermination or theory-ladenness here, arguments that are important for Kuhn’s rejection of empiricist and falsificationist methodologies as well. However, Barnes draws a stronger conclusion from these arguments: Scientists do in fact regularly “disagree on what is ‘objectively’ rational, and use conflicting criteria, based on empiricism, crude falsificationism, etc. This, in itself, is one of the most powerful arguments in favor of the position advocated here.” (Barnes 1972, p. 378f.)5

That is, since rational criteria or maxims do not explain why scientists come to believe what they believe—as when, for instance, they come to agree on a certain theory rather than another—there must be another kind of explanation for why they do end their quarrels; and the appropriate explanation should be sociological in a strong sense. It should not merely show that scientific activities might be influenced by the social contexts of science—as when grants are given for one research program but not for another. Rather, it should be shown that such contexts are also what determines the results of these activities, namely the acceptance of some scientific knowledge claims, and, ultimately, the very scientific knowledge claims themselves. This is what Barnes means when he claims that the sociological explanation “extends throughout the entire context of justification”. II. GENERAL PROBLEMS OF THE STRONG PROGRAMME

1. Misguided Worries About Theory-ladenness and Underdetermination Obviously, the viability of the SP’s position depends much upon the strength of the arguments for the underdetermination thesis, the thesis of theory-ladenness, and of other claims with similar skeptical impact. Up to today, the arguments we know from Quine, Kuhn, or Feyerabend are taken for granted by many authors. However, Larry Laudan, Philip Kitcher and others have advanced serious counterarguments to them.6

HOW CAN WE USE THE DJ DISTINCTION

139

For instance, it is reasonable to view the underdetermination thesis as a purely or primarily logical thesis: Different theories can be empirically equivalent, and so no theory can be fully determined by its empirical evidence. It hardly follows from this logical point that scientists in fact use only empirical criteria, nor that all methodologies require that they should use only such criteria in their decisions. Of course, sometimes scientists may in fact use only empirical criteria, and this may be what explains their theory decisions. Then, regardless of the logical underdetermination thesis and of other objections against empiricist methodologies, the explanation of their decisions can and should be given with reference to these criteria.7 With regard to the theory-ladenness of data, again, the familiar skeptical weight of this thesis can be weakened if one distinguishes between different forms of theory-ladenness. There are good reasons to think that at least some empirical claims can be established relatively independently of theories—in particular, many empirical claims need not be infected by those theoretical claims for which scientists are trying to find independent empirical support or refutation (Hacking 1983, chapters 9–11; Heidelberger 2003). Also, the task of explaining a scientist’s belief need not be affected by the claim of theory-ladenness. There may be many cases where a scientist comes to accept a certain belief or empirical claim thinking that his acceptance is free of any theory whatsoever (perhaps because he strongly favors an empiricist methodology). Then his methodological norm could surely be not only his justificatory reason, but also what explains his coming to accept that empirical claim. We shall not develop further such considerations here. Even if champions of the SP are unimpressed by them, there are other problems with their rejection of the DJ distinction—problems that concern the laws of the SP’s own empire.

2. For Whom Is the DJ Distinction Useless? A basic problem is that it is unclear whether Barnes merely wants to say that the DJ distinction is useless within sociology, or whether he also wants to say that the distinction is useless for any investigation of science whatsoever, thus making it virtually nonexistent. On the one hand, Barnes claims that sociological explanation “extends throughout the entire context of justification”: What we call justifications and accepted knowledge claims are merely results of consensus formation. Once a sociological explanation of scientific claims has been given there is no distinct justificatory or normative point of view. On the other hand, Barnes also says more moderate things such as that the “attempts to make objective distinctions between true and false, or rational and irrational, beliefs and practices are of no interest to the sociologist” (Barnes 1972, p. 374, emphasis added; cf. 378). For many case studies about science, such moderation seems in place. One may simply try to take a strong sociological approach as far as possible in order to see what it can teach us about scientific research, without claiming that it cannot be superseded by different approaches. Moreover, even if authors say that they subscribe to the SP, they in fact often offer judgments concerning the DJ distinction differing from Barnes’ in that they do not necessarily commit themselves to the claim of a uselessness of the distinction.8

140

THOMAS STURM AND GERD GIGERENZER

However, the SP is committed to an unrestricted rejection of the DJ distinction. This can be shown by considering Bloor’s characterization of the four basic principles of sociological explanation of scientific knowledge according to the SP: 1. “It would be causal, that is, concerned with the conditions which bring about beliefs or states of knowledge. Naturally there will be other types of causes apart from social ones which will cooperate in bringing about belief. 2. It would be impartial with respect to truth and falsity, rationality or irrationality, success or failure. Both sides of these dichotomies will require explanation. 3. It would be symmetrical in its style of explanation. The same types of cause would explain, say, true and false beliefs. 4. It would be reflexive. In principle its patterns of explanation would have to be applicable to sociology itself. Like the requirement of symmetry this is a response to the need to seek for general explanations. It is an obvious requirement of principle because otherwise sociology would be a standing refutation of its own theories.” (Bloor 1976, p. 7)

These four principles are called, respectively, the requirements of causality, impartiality, symmetry, and reflexivity. It appears that at least the first three requirements have already been advanced by Barnes. First, he attacks views according to which “the culture of natural science [. . . ] has been exempted, a priori, from causal sociological analysis” (Barnes 1972, p. 373). While Barnes agrees with “idealistic” or “phenomenalistic” views according to which the rationality of actors has to be “judged by actors’ own rules and standards”, he considers it to be a mistake of authors like Peter Winch to see causal and rational explanations as “incompatible” (ibid., p. 375). Secondly, Barnes anticipates the impartiality requirement in the passage already cited above concerning the irrelevance of “objective distinctions between true and false, or rational and irrational, beliefs and practices” (ibid., p. 374). Finally, Barnes links his preference for causal explanations with a claim that closely resembles Bloor’s symmetry principle: The mistake of viewing causal explanations and explanations that cite actors’ reasons as incompatible was, Barnes suggests, probably due to the fact that causal explanations in sociology have been too “parochial” in explaining only “the irrationality of others, whilst putting action in accord with their own cognitive ideals beyond causality by labeling it rational.” Sociologists should “avoid evaluative rationality judgments, not by eschewing causal explanation entirely, but by recognizing that causal analysis is applicable, in principle, to all beliefs and practices” (ibid. p. 376). Although none of the four principles is unproblematic (e.g., Laudan 1984b; Friedman 1998), the claims that are relevant here are (2) and (3). It appears plausible that the sociologist of science, like any other empirical scientist, should abstain from value judgments concerning his subject matter and that he should treat both true and false, or rational and irrational beliefs alike as explanandum. Also, it appears plausible that both true and false, or rational and irrational beliefs can be causally explained; at least, we should always be open to this possibility. However, the SP connects this with a stronger claim: Notions such as truth or rationality are seen as having no explanatory force per se. Not only should we view true beliefs as being subject to causal explanation as well; we should also refrain from thinking that we can sometimes explain the belief of a scientist in terms of truth or rationality. Bloor claims

HOW CAN WE USE THE DJ DISTINCTION

141

that although there is “nothing wrong with using such terms as ‘true’ and ‘false’ [. . . ] the accounts given of this use [. . . ] are suspect” (Bloor 1991, p. 177). The SP is to be viewed as a part of a more general naturalistic approach in science studies, since Bloor does not wish to exclude psychological or biological factors from the explanation of scientific knowledge (Bloor 1984, p. 75). He emphasizes that the debate over “the status of the symmetry requirement lies in the clash between a naturalistic and non-naturalistic perspective. The symmetry requirement is meant to stop the intrusion of a non-naturalistic notion of reason into the causal story.” (Bloor 1991, p. 177) In other words, we should opt for a thoroughly “naturalistic” notion of rationality, according to which references to truth and standards of reasoning can never be basic in the explanation of a scientist’s knowledge claims, but must always be supported by a more fundamental, “naturalistic” account. Accordingly, explanations following the four principles are thought to be the only appropriate kind of explanation of scientific knowledge claims (cf. Bloor 1991, p. 179).9 The requirements for the new program for the sociology of science are strong indeed. Since sociology is supposed to explain scientific knowledge in a strong sense, the DJ distinction is not merely useless within sociology, but useless in general. If arguments such as those from underdetermination or theory-ladenness are viewed as leading to the skeptical conclusions about justification that Barnes and Bloor infer from them, and if principle (2) and (3) are accepted, then there are no satisfactory answers to the question of whether a scientific claim is justified—be it from the point of view of the scientist or from an outsider’s point of view who might try to evaluate the scientist’s claims. While it is not made explicit by Bloor or Barnes what version of the DJ distinction they reject, their various claims constitute an attack even on the lean version of the distinction in the explanatory guise.

3. Impartiality and Symmetry: Does, or Can, the SP Deny that Reasons Can Be Causes? How convincing is the claim that the notions of truth and rationality cannot play an explanatory role? If this claim is correct, then we have to understand scientific research without taking into account things as the following: Scientists have beliefs and goals upon which they ground their decisions and actions; they take their beliefs, at least some of them, to be things that can be true or false; they think that the truth or falsity of their beliefs does not simply follow from the ways in which they have acquired them; therefore, the question of the justification arises in a natural way as part of the scientific endeavor; and that is why they are, often enough, involved in trying to answer that question by referring to standards of rationality and make their epistemic decisions in this way. In other words, the SP is at odds with the—highly plausible—views that an agent’s reasons can be the causes of his actions, and that sometimes the most favorable explanations should be in terms of an agent’s reasons (Davidson 1963; Arabatzis 1994; Okasha 2000). Admittedly, Barnes occasionally does sound as if he accepted that we can explain the actions of scientists in terms of truth and rationality. For example, he says that

142

THOMAS STURM AND GERD GIGERENZER

sociology should take into account the agent’s point of view: “[. . . ] actors’ perceptions of efficacy or potential efficacy are important explanatory variables; objective assessments are not” (Barnes 1972, p. 379; cf. ibid., p. 375). Bloor’s talk of a “naturalistic” notion of rationality seems to point in the same direction (cf. Bloor 1983).10 But given the symmetry requirement, this cannot be correct (Hesse 1985, p. 32): The fact that scientists’ views of what they think can influence their thoughts or decisions (i.e., their own standards of evidence) sometimes explains why they pick one theory and reject another. The champion of the SP cannot account for this possibility because he does not acknowledge standards of rationality and because the same causes—social needs and interests, or other “natural” or “non-normative” factors—are supposed to explain both what we commonly single out as true and as false beliefs.11 However, a strict symmetry makes no sense from the agent’s point of view. If someone takes a certain claim p to be a reason for another claim q, then, given the acceptance of p, this agent cannot consciously and seriously believe that not-q. If an agent accepts p because (in a causal sense of ‘because’) of q, then he also commits himself to the rejection of whatever is excluded by p.12 Hence, Barnes and Bloor cannot claim that they are referring to the very psychological processes that, under another description, may be called ‘reasons’. Mentioning that scientists claim to employ certain reasons or reasoning standards, and mentioning which ones these are, is not the same as taking these standards seriously—seriously in the sense that they are acknowledged as causes of the scientist’s acceptance or rejection of claims.13 Of course, scientists may overlook rational connections between their beliefs; they may be wrong about them, and so on. Also, there are many cases where it is unclear whether we are justified in ascribing certain rational trains of thought to agents. We may tend to rationalize their behavior too much when we claim, for instance, that their thought is “implicitly” rational (McMullin 1984). The fact that one can describe a certain behavior by means of a rule of rationality does not mean that the behavior actually follows that rule. It is notorious how many scientists have not followed their own cherished methodological norms. However, it still does not follow that a scientist’s reason can never be the best explanation of his belief.

4. An Ambiguity in the Symmetry Requirement Another basic deficiency in principle (3) is the following. The claim that one should explain false and true beliefs by exactly the same kinds of causes overlooks a crucial ambiguity in such notions as ‘belief ’, ‘knowledge’ and the like. On the one hand, when we speak of a belief of a person, we might refer to an attitude (or a mental act or state) of a person towards a propositional content—Jones’ believing that it never rains in Southern California, say. On the other hand, we can refer to what is believed, or to the propositional content itself—to that it never rains in Southern California.14 There are basic reasons, coming from outside history, philosophy or sociology of science, for why we have to distinguish between propositional contents and the attitudes taken toward them. The distinction is necessary for certain possibilities in the description and explanation of thought and action. Sometimes we wish to say

HOW CAN WE USE THE DJ DISTINCTION

143

that different persons believe the same things or the same propositions. Sometimes, however, we wish to say that at a certain point in time someone merely thinks of a certain state of affairs whereas at a later point in time he comes to believe, or even know, that that state of affairs obtains. So, sometimes we keep propositional contents stable and vary attitudes over persons or points in time; and sometimes we keep attitudes stable over different persons (or over different points in time of an individual’s life) and vary the contents of their thoughts, beliefs, or desires. This is necessary in order to understand how individuals influence one another by asking and answering questions, criticizing one another, revising their claims, and so on. Ultimately, even a sociological case study of the emergence of a consensus within a scientific community cannot avoid such assumptions. To assume that there is conflict over a given claim makes sense only if one assumes that there are different scientists having different attitudes towards the same propositional content. The idea of achieving a consensus, in turn, makes sense only if the different scientists come to hold the same attitude towards a certain proposition. Now, which feature of knowledge claims is supposed to be explained symmetrically by sociological causes? Is it the epistemic attitude towards a proposition or the propositional content itself? The answer has to do with the origin of talk of a “strong” program in the sociology of science. The idea is that the SP constitutes a dramatic progress over older approaches in sociology, as developed especially by Robert K. Merton (e.g., Merton 1973, chapter 12). The “older” sociology limited its pretensions: Sociological investigations should be concerned with the social relations of a scientific community, the economic or political backgrounds for decisions concerning the directions of scientific research, and so on. With regard to scientific claims, the idea was that only errors could be explained sociologically, while true or rational claims had their explanation in scientific standards of justification. Why did older approaches in sociology limit themselves in this way? Bloor claims that the “cause of the hesitation to bring science within the scope of a thorough-going sociological scrutiny is a lack of nerve and will” (Bloor 1976, 4; cf. Barnes 1972, 373). In contrast, the SP claims to explain all scientific claims sociologically, whether they are true or false, rational or irrational. Knowledge itself is said to be an explanandum as well. However, this cannot be taken literally. For this would mean, say, attempting to explain by sociological factors the proposition that the earth moves around the sun (in contrast to why people think that the earth moves around the sun) but one would also have to explain the proposition that the sun moves around the earth. After all, according to the symmetry principle, both true and false claims would have to be explained. The problem may be resolved by noting that, as a matter of fact, nowadays hardly anyone believes that the sun moves around the earth. The SP would then only claim to explain what is factually believed within a given community. But there remain other problems. If the goal of the SP is not “merely” to explain why someone claims to know the truth of a certain proposition, but to explain “knowledge itself ”, then the SP explains why the content of such a knowledge claim is true (if it is true). But to explain that a statement or theory is true is already to justify it.15 If a sociological explanation did this, it would do what the SP (i) forbids (because of the principles of impartiality

144

THOMAS STURM AND GERD GIGERENZER

and symmetry) and (ii) says cannot be done (because of the problems of normative methodologies). The idea of getting at the content of science itself sociologically may be understood as a reaction to Mertonian “error” sociology, but it cannot be defended. If there was a “lack of nerve and will” on the side of older sociologists, it has been replaced by a conceptual confusion on the side of more recent sociologists. It follows that the explanandum of sociology has to be the attitude taken towards a propositional content—the acceptance or rejection of that content. Here it is important that Bloor states that “knowledge for the sociologist is whatever people take to be knowledge. It consists of those beliefs which people hold to and live by” (Bloor 1976, p. 5). This makes the SP look more moderate, but there is a price to be paid here. Bloor’s statement shows that he does not define knowledge in terms of justified true belief or the like, but simply in terms of beliefs that are accepted within a given community. If the explanandum is knowledge in this weak sense, then the claim that “the scope of sociological explanation extends throughout the entire context of justification” loses much of its original strength. Once a strong sociological explanation has been given for why people in fact “hold to and live by” certain beliefs, it is still sensible to ask: Very fine, but are these beliefs justified? The context of justification is by no means wholly permeated by sociological explanation; the lean DJ distinction is still alive and well. This clearly shows that the distinction supports the project of a critical or normative evaluation of scientific claims. But the point also applies to the explanatory use of the distinction: Even after a strong sociological explanation has been given for the beliefs of a scientist, it remains sensible to ask: Very fine, but how are these beliefs connected to the scientists’ justificatory reasons? Can these connections perhaps explain better why the scientist acquired the relevant beliefs?16 This result does not imply that the sociological explanation of the acceptance and rejection of scientific propositions would be a wholly unintelligible project. Many sociological case studies are concerned with cases of scientific closure or consensus formation in debates about theory choice where the evidence does not, or does not seem to, strongly dictate a particular solution. Coming to accept that p is a process that may be due to the scientist’s social or political interests. It need not involve reasons or rational conditions (understood here as those kinds of reasons which conform to some epistemological or methodological standard). It can be a non-rational, merely causal condition.17 This depends on the case in question. There also remains a difference between the older and the more recent sociological approaches to science: Sociological explanations may be directed equally at true and false, rational and less rational knowledge claims. Bloor originally claimed that various case studies—he mentioned Paul Forman’s study about physics in Weimar Germany, and others about biology and geology in the 19th century—“suggest that the sociology of science should adhere” to the four requirements (Bloor 1976, p. 7). The cautious formulation of “suggesting” was, and still is, quite in place. No particular case study can prove that the four principles are always to be preferred. Neither can it completely rule out explanations of scientific knowledge that refer to reasons, the adherence to evidential standards or the goal of reaching true beliefs as potential causal factors. Perhaps adherents of the SP should have emphasized more such cautionary remarks.

HOW CAN WE USE THE DJ DISTINCTION

145

On the other hand, how far can explanations be avoided that concentrate on the reasons of scientists? One may tell a particular story such that one explains the process of acceptance in sociological terms (where this excludes asymmetrical talk of truth and reasons); or one may look for the justificatory reasons of agents. What difference does that make? Our claim is that sometimes it matters much. It matters because leaving out the agent’s point or view, or the reasons of scientists themselves, may mean leaving out important details. In order to show this, we shall present a case study that defenders of the SP might view as filling their bill satisfactorily, or even as a striking instance of the SP. Should we find here that an appropriate explanation of the generation and acceptance of scientific claims makes essential use of rational considerations in theory generation and acceptance, this would undermine the symmetry and impartiality claims. Moreover, the rejection of the DJ distinction as useless within sociology would be weakened further as well. After all, how good can explanations be if they leave out what is essential? It may be tempting to look for cases of more unproblematic justificatory reasons in order to refute the SP’s views about explanation in science studies. Should one question that 2 + 2 = 4 is a truth mathematicians have sometimes, or even often, used in their inferences as a reason? Can one not point out that the principle of noncontradiction is often used as a good reason to reject contradictory claims? Also, are there not many relatively stable and innocent empirical claims in science that can play a role as a rational cause or constraint upon subsequent research developments? We do not doubt that such examples exist, on the contrary. However, since the SP frequently deals with consensus formation concerning the emergence and acceptance of theories or of whole theoretical approaches, we think that it is more instructive to look at more basic processes of theory formation in science. How have certain theoretical concepts and claims been developed in the first place? Why have certain theoretical approaches been accepted at very early stages of research where much empirical evidence was not yet available? Here, non-rational factors (such as certain sociological ones) might easily be expected to play a much stronger explanatory role, and justificatory reasons might be wholly absent. Once again, in order to defeat the SP, it is advisable to use a case study that is as attractive to this party as it can be. III. USING THE DISTINCTION: A CASE STUDY FROM COGNITIVE SCIENCE

1. Metaphors in the Generation and Acceptance of New Theories New theories in science are often inspired by new metaphors for the objects or processes under investigation. As Quine has noted, metaphors are “vital . . . at the growing edges of science” (Quine 1978, p. 159). One such group of metaphors comes from the realm of the methodological tools scientists use. In the history of psychology and related disciplines, this happened, especially but not exclusively, during the so-called “cognitive revolution” of the 1960s with two tools: inferential statistics and the digital computer.18 Basically, these cases possess the following structure. (1) First, there was the generation of a new theory: The tool a scientist uses suggested new metaphors of the mind

146

THOMAS STURM AND GERD GIGERENZER

to him, leading to new theoretical concepts and principles. (2) Next, there took place the acceptance of the new theory within the scientific community: The new theories were more likely to be accepted by the scientific community if the members of the community were also familiar with the new tools, and much more likely to be rejected or not even be understood if there was no such familiarity. We shall now illustrate the two examples (for further details, see Gigerenzer 1992; Gigerenzer and Goldstein 1996; Gigerenzer and Murray 1987; Gigerenzer and Sturm, forthcoming).

2. The Mind as an Intuitive Statistician Techniques of statistical inference had been known much before the 20th century (Gigerenzer et al. 1989). Nonetheless, there was little interest in these techniques in experimental psychology before 1940. By the early 1950s, however, half of the psychology departments in leading American universities had made inferential statistics a graduate program requirement. By 1955, more than 80% of the experimental articles in leading journals used inferential statistics to justify conclusions from the data (Sterling 1959). Thus, 1955 can be used as a rough date for the institutionalization of the instrument of inferential statistics—a mixture of ideas from opposing camps, those of R. A. Fisher, on the one hand, and Jerzy Neyman and Egon S. Pearson on the other—in curricula, textbooks, and editorials. The interesting fact is that, before they became familiar with inferential statistics, researchers were almost unable to develop and accept a new theoretical conception of mental functioning, namely that of the mind as an intuitive statistician. The analogy between the mind and the statistician was first proposed in the early 1940s, by Egon Brunswik, and he developed it into testable models (e.g., Brunswik 1943). In the late 1930s Brunswik changed his techniques for measuring perceptual constancies, from calculating (nonstatistical) “Brunswik ratios” to calculating Pearson statistical correlations, such as “functional” and “ecological validities.” But his analogy and the corresponding theoretical claims were poorly understood and generally rejected (Leary 1987). After the institutionalization of inferential statistics around 1955, things changed quickly and dramatically. For instance, around 1955, the traditional psychophysics of absolute and differential thresholds was revolutionized by the new analogy between the mind and the statistician. W. P. Tanner and others proposed a “theory of signal detectability” (TSD), which assumes that the Neyman–Pearson technique of hypothesis testing describes the processes involved in detection and discrimination. In Neyman–Pearson statistics, two sampling distributions (hypotheses H0 and H1 ) and a decision criterion (which is a likelihood ratio) are defined, and then the data observed are transformed into a likelihood ratio and compared with the decision criterion. Depending on which side of the criterion the data fall, the decision “reject H0 and accept H1 ” or “accept H0 and reject H1 ” is made. In straight analogy, TSD assumes that the mind calculates two sampling distributions for “noise” and “signal plus noise” (in the detection situation) and sets a decision criterion after weighing the cost of the two possible decision errors (Type I and Type II errors in Neyman–Pearson theory, now called “false alarms”

HOW CAN WE USE THE DJ DISTINCTION

147

and “misses”). The sensory input is transduced into a form that allows the brain to calculate its likelihood ratio, and depending on whether this ratio is smaller or larger than the criterion, the subject says, “no, there is no signal” or “yes, there is a signal.” Tanner (1965) explicitly referred to his new model of the mind as a “Neyman–Pearson” detector. A second example concerns theories of causal reasoning. Albert Michotte (1946/1963), Jean Piaget (1930), and others had investigated how temporal spatial relationships between two or more visual objects, such as moving dots, produced phenomenal causality. For instance, the subjects were induced to “perceive” that one dot pushes or chases another. After the institutionalization of inferential statistics, Harold H. Kelley (1967) proposed in his “attribution theory” that the long-sought laws of causal reasoning are in fact the tools of the behavioral scientist, namely Fisher’s ANOVA. Just as the experimenter infers a causal relationship between two variables from calculating an analysis of variance and performing an F-test, the man-in-thestreet infers the cause of an effect by unconsciously doing the same calculations. By the time Kelley developed the new metaphor for causal inference, about 70% of all experimental articles already used ANOVA (Edgington 1974). The theory was quickly accepted in social psychology: Kelley and Michaela (1980) reported more than 900 references in one decade.

3. The Mind as a Computer The relation between conceptions of the mind and the computer has a long history, starting from simple calculating machines developed in early modern times by Pascal and others, over Charles Babbage’s famous (though unsuccessful) attempts to build an “Analytical Engine”, which was supposed to perform algebraic transformations, up until the modern digital computer as conceived by Alan Turing. Within philosophy, the metaphor of the mind as a computer program (and the brain as its hardware) was made prominent through Hilary Putnam’s work on the status of psychological predicates, and he used Turing’s work to illustrate his functionalism about mind and brain. He argued for a distinction between mind and brain in terms of the difference between software and hardware, thus showing that mental states can be realized in quite different physical systems (e.g., Putnam 1960). This argument was able to influence the philosophical debate, which was restricted to an ontological understanding of the mind-body relation or to more principled questions concerning the autonomy of psychology in relation to, for instance, neurophysiology. But this did not help the computer metaphor to become popular within psychology. The theoretical idea of the mind as computer only really took off when the computer became a standard laboratory tool in the 1970s. Particularly influential was Herbert Simon’s and Allen Newell’s brand of information processing psychology. In the summer of 1958, psychology was given a double dose of the new information processing psychology. One was the article “Elements of a Theory of Human Problem Solving” in the Psychological Review (Newell, Shaw and Simon 1958). The other was the Research Training Institute on the Simulation of Cognitive Processes at

148

THOMAS STURM AND GERD GIGERENZER

the RAND institute. In the Psychological Review paper, the authors claim that “the program of LT [Logic Theorist] was not fashioned directly as a theory of human behavior; it was constructed in order to get a program that would prove theorems in logic” (Newell, Shaw and Simon 1958, p. 154); later on, however, they also state that Logic Theorist “provides an explanation for the processes used by humans to solve problems in symbolic logic” (163). The evidence provided for projecting the machine into the mind is mainly rhetorical. For instance, the authors spend pages arguing for the resemblance between the methods of Logic Theorist and concepts such as “set”, “insight”, and “hierarchy” described in the earlier psychological literature on human problem solving. The first general statement of Newell and Simon’s new conception of the mind appeared in their 1972 book Human Problem Solving. The authors argue for the idea that higher-level cognition proceeds much like the behavior of a production system, a formalism from computer science which had never been used in psychological modeling before: “Throughout the book we have made use of a wide range of organizational techniques known to the programming world: explicit flow control, subroutines, recursion, iteration statements, local naming, production systems, interpreters, and so on [. . . ]. We confess to a strong premonition that the actual organization of human programs closely resembles the production system organization” (Newell and Simon 1972, p. 803).

We will not illustrate how ideas of information processing changed theories of mind. It is only too natural for present-day psychologists to speak of cognition in terms of encoding, storage, retrieval, executive processes, algorithms, and computational costs. Psychologists have even begun to explain scientific discovery with programs such as BACON that attempt to induce scientific laws from data (Langley et al., 1987). The introduction of the computer as a laboratory tool in psychology began in the mid-1960s, when only a small number of psychological labs were built around computers, including: Carnegie Mellon, Harvard, Michigan, Indiana, MIT, and Stanford (Aaronson, Grupsmith, and Aaronson 1976, p. 130). As indicated by the funding history of NIMH grants for cognitive research, the amount of computer-using research tripled over the next decade. In 1967, only 15% of the grants being funded had budget items related to computers. By 1975, this figure had increased to 46%. The late 1960s saw a turn towards mainframe computers, which lasted until the late 1970s when the microcomputer started its invasion of the laboratory. In the 1978 Behavioral Research Methods & Instrumentation conference, microcomputers were the issue of the day (Castellan 1981, p. 93). By 1984, the journal Behavioral Research Methods & Instrumentation appended the word “Computers” to its title to reflect the broad interest in the new tool. During the last two decades, computers have become an indispensable research tool of psychologists. How closely was the spreading of the computer metaphor of the mind, and the communal acceptance of corresponding psychological theories, correlated to the computer becoming a research tool? This can be exemplified by reference to two institutions:

HOW CAN WE USE THE DJ DISTINCTION

149

the Center for Cognitive Studies at Harvard, and Carnegie-Mellon University. The former never came to fully embrace the new information processing psychology. The latter did, but only after a considerable delay. George Miller, the co-founder of the Center at Harvard was certainly a proponent of the new information processing psychology. Given Miller’s enthusiasm, one might expect the Center, partially under Miller’s leadership, to blossom into information processing research. It never did. In the Annual Reports of the Center from 1963 to 1969, there are only a few symposia or papers dealing with computer simulation. Although the center had a PDP-4C computer, and the reports anticipated the possibility of using it for cognitive simulation, as far as 1969 it never happened. Difficulties involved with using the tool were considerable. They even turned out a 1966 technical report called “Programmanship, or how to be one-up on a computer without actually ripping out its wires.” What might have kept the Harvard computer from becoming a metaphor of the mind was that the tool turned out to be a steady source of frustration for researchers. At Carnegie Mellon, Newell, Simon, a new information processing-enthusiastic department head, and a large NIMH grant were pushing the new psychology. Even this failed to proselytize the majority of researchers within their own department. In the late 1950s, at Carnegie Mellon, the first doctoral theses involving computer simulation of cognitive processes were being written. But this was not representative of the national state of affairs. Only in the early 1970s, information processing psychology caught on at Carnegie-Mellon University. Every CMU-authored article in the 1973 edition of the Carnegie Symposium on Cognition mentions some sort of computer simulation, the new form of psychological experiment. For the rest of the psychological community, who were not as familiar with the tool, the date of broad acceptance was years later. In 1979, Simon estimated that from about 1973 to 1979, the number of active research scientists working in the information processing vein had “probably doubled or tripled” (Simon 1979). 4. Why Were the New Cognitive Theories Generated and Accepted? The stories just outlined seem to fit quite well with the SP. It appears that the acceptance of the new cognitive theories is brought about by social factors such as the institutionalization of statistics in psychology during the 1950s, or the spreading of the computer as a laboratory tool in the 1970s. Also, these instruments were often viewed as giving psychology finally a standing as a mature or exact science, so it was perhaps no surprise that these tools informed new ideas of the mind. Telling the story in this way means to explain the adoption of the new theories not by the reasons agents advanced for those theories, but by causes having to do with the institutionalization of psychological methods, and with the public standing of psychology. Such ideas are familiar to the SP: “Much that goes on in science can be plausibly seen as a result of the desire to maintain or increase the importance, status and scope of the methods and techniques which are the special property of the group.” (Bloor 1984, p. 80)

150

THOMAS STURM AND GERD GIGERENZER

On the other hand, the story outlined has worried a realistically minded philosopher such as David Papineau: “How can cognitive scientists possibly be tracking the truth, if they can be persuaded to believe given theories by institutional developments which have no apparent connection with the subject matter of those theories? . . . It would indeed be damning if the institutional developments in question were sufficient to determine theory acceptance. But their being necessary leaves it open that other factors might also have been necessary, and in particular proper empirical support might have been necessary too.” (Papineau 2003, 146f.)

There are several problems with this objection. For instance, Papineau suggests that the acceptance of the new theoretical vocabularies and claims could be determined by proper “empirical support”. Even someone who supports realism about theoretical claims should be aware that the actual realistic interpretation of a theory is a more arduous task (Kr¨uger 1983; Gigerenzer and Sturm, forthcoming). Here, however, we wish to consider Papineau’s claim that cognitive scientists “can be persuaded to believe given theories by institutional developments which have no apparent connection with the subject matter of those theories”. While champions of the SP will probably welcome this interpretation of our case studies, it is misguided. To begin, note that although certain research tools have inspired new psychological theories, not all tools have done so, or can do so—and not even when these tools have promoted the scientific standing of psychology. In early psychological reaction time measurement, expensive precision chronoscopes were used. The so-called Hipp chronoscope gave Wilhelm Wundt and his students at Leipzig the firm conviction that they had, for the first time, turned psychology into a truly experimental science. Yet, neither such precision clocks nor, more recently, visual imaging techniques such as fMRI have inspired new theories of the mind. Some tools can play a metaphorical role, while others cannot, or not as good as others. Why? The answer lies in that the use of a tool as a metaphor of the mind is not something that occurs mindlessly. Inventing a metaphor and developing theoretical notions is certainly not something that can be justified in more familiar ways of scientific justification such as gathering empirical evidence or trying to test or support hypotheses. However, there are reasons for taking certain routes in theory generation. What is typical in the present cases is that the tools psychologists have been inspired by in their theorizing possess features that cohere with certain a priori assumptions about the mind.19 For instance, an essential precondition for the view of mind as a computer is the realization that computers are symbol transforming devices, as opposed to, say, the numerical calculators envisaged by Babbage and others. As long as computers are viewed as being restricted to the latter, and as long as mental activities are seen as more complex than numerical calculation, it is hardly surprising that the computer is not proposed as a metaphor for the mind. Newell and Simon were among the first to realize this. They not merely adopted the computer metaphor as a matter of fact, but adopted it in the light of a certain critical conceptual reflections. The symbol-manipulating nature of the computer was important to Simon because it corresponded to some of his earlier views on the nature of intelligence:

HOW CAN WE USE THE DJ DISTINCTION

151

“The metaphor I’d been using, of a mind as something that took some premises and ground them up and processed them into conclusions, began to transform itself into a notion that a mind was something which took some program inputs and data and had some processes which operated on the data and produced output” (cited in McCorduck 1979, p. 127).

There are other examples of how a priori assumptions about the mind could ground the development of new theories inspired by research tools. The metaphor of the mind as an intuitive statistician has had a long history as well. For instance, classical probability theory was not a purely formal theory but was more closely connected to the subject matter of human beliefs and decisions. John Locke thought that there are different degrees of the probability of many of our beliefs, since human beings are not able to reach perfect certainty on many topics. David Hartley developed his associationism in probabilistic terms as well. More importantly for the spreading of the metaphor of the mind as an intuitive statistician in the 20th century were further ideas at the core of the new theories. For instance, in the case of the psychology of perception, the new metaphor helped to overcome the idea of fixed thresholds of sensory inputs. A new conceptual distinction emerged: There was not any longer only either the perception of a stimulus or its lack of perception. Just as the statistical theory distinguishes two types of error, false alarms and misses, these two types of errors were also hypothesized in the mind’s functioning. The new approach made new questions thinkable, such as “How can the mind’s decision criterion be manipulated?”, leading to new kinds of experiments. A final example concerns the work of the behaviorist Edward C. Tolman, who used mazes to study animal learning (Smith 1990). For instance, rats were rewarded equally to go in both arms of a T-shaped maze, and their responses were found to be stable. Either a rat would always choose one arm, or it would consistently alternate between the two arms. Working with more complex mazes, he found that rats developed more global knowledge of the maze. When one feature of a maze was altered, the rats would rather quickly select a different route to their goal. Tolman developed the concept of a “cognitive map” to characterize this capacity, inspired by his apparatus. But it was not the apparatus as such which made the new notion plausible for him. Tolman had been taught by the neorealistic school of Ralph Barton Perry and Edwin B. Holt at Harvard. He viewed the mind as essentially embedded in its environment, as opposed to its being a private entity. A mental state such as a purpose is “out there in the behavior”, thus possessing spatial features itself. One can understand why Tolman would, at some point, find it reasonable to think of the selection of routes in terms of spatial maps, as opposed to merely begin to apply the metaphor as a matter of fact. Thus, Papineau’s worry that the explanation of theory generation presented above has no connection “with the subject matter of those theories” is unfounded. Of course, not all scientists need always be aware of the relevant a priori assumptions, such as the ones that supported Tolman’s, Tanner’s, or Simon’s new theories. Where scientists did not reflect on such basic conceptions, it is plausible that their acceptance of, and working within, behavioristic or cognitive traditions has to be explained sociologically. However, a purely sociological explanation of the generation and acceptance

152

THOMAS STURM AND GERD GIGERENZER

of theories is not merely superficial. It is, rather, essentially incomplete. It would be unable to explain many fine-grained developments within the new theoretical approaches, while looking at the scientist’s reasons often provides explanations here. Furthermore, it would overlook that there has to be an explanation for why certain basic metaphors have been chosen and others left out. If there is a story to be told at all about the generation of theories in cognitive science, this story contains prior assumptions about what counts as a possibly correct, or possibly rational, theory in the field at all. IV. CONCLUSION: THE EXPLANATORY USE OF THE LEAN DJ DISTINCTION

Our goal in this paper was not so much to argue that that the lean DJ distinction can be saved from criticisms that have been raised against various stronger versions of the DJ distinctions. We accept that it can. Rather, our goal was to look for positive uses of the lean DJ distinction, and we identified two basic possibilities: first, one may accept the lean DJ distinction in order to make clear that there is room for the view that we can pursue the task of an evaluation or critical assessment of scientific knowledge. Even after we have answered in a given case the question of how someone came to accept that p, we may always still ask whether p is justified. Secondly, the lean DJ distinction also possesses a value for the task of explaining science (scientific knowledge, research, theory formation . . . ). This we tried to support by arguing that sometimes theory developments in science cannot be sufficiently explained by purely non-rational causal conditions, nor by leaving out or not taking seriously the justificatory reasons of scientists. Now, there is no a priori reason either for the assumption that all scientific research developments are caused entirely by non-rational conditions nor, alternatively, for the assumption that in all scientific research developments justificatory reasons, or rational conditions, must play a causal role. It is most likely that we can find both cases, and mixed cases as well. If we wish to investigate scientific research developments adequately, we should bear in mind the DJ distinction, and we should view it as a continuous task to identify both rational and non-rational causes in these developments. Clearly, such a use of the DJ distinction does not—at least not immediately and without further argument—commit one to the view that history or sociology of science cannot avoid the epistemological or methodological concerns with which philosophers are frequently concerned. There may be arguments for this view. For instance, it may be that the task of showing that a scientist’s reason causally explain one or more of the scientist’s beliefs commits one to certain normative assumptions about what can possibly count as a reason. However, we do not wish to pursue this difficult topic here, important as it may be for the proper understanding of how history, sociology and philosophy of science are related to one another. We wish to end with a different point, namely that sometimes normative questions pursued within epistemology or methodology of science should not avoid considerations concerning the generation of scientific claims. Reichenbach and other advocates of the DJ distinction often assume that the task of evaluating a scientific claim can be pursued

HOW CAN WE USE THE DJ DISTINCTION

153

quite independently of knowledge about the origins of that claim (e.g., Siegel 1980, p. 303). Here we disagree. Sometimes good criticism of a theoretical assumption will profit from such knowledge, if not be impossible without it. The reason for this is not that viewing the spreading of cognitive theories on the background of the institutionalization of statistics and the computer makes one suspicious about the validity of these theories (although this history may be an eye-opener). Rather, the point is that metaphors always emphasize some aspects and leave out others. Freud famously compared the relation between the two systems of perception-consciousness and memory to the Wunderblock or “mystic writing pad”. On such a pad, consisting of a wax layer, wax sheet and a transparent celluloid paper, one can erase text by simply pulling the paper free of the wax layer. When one pulls the paper, however, at a deeper level a trace of what had been written is stored. Freud also pointed out that, unlike our capacity of memory, the pad cannot “reproduce” the erased text from inside (Draaisma 2000, chapter 1). Especially in cases of the more successful metaphors in science, such losses can easily be forgotten. The more we become aware that theoretical assumptions are often rooted in metaphors, the better we can avoid the pitfalls they contain. However, once again, this learning from the history of science is possible only if a distinction is borne in mind between the questions of how a theory is generated and whether it is justified or not. ACKNOWLEDGMENTS

We thank the other contributors to this volume and Uljana Feest for comments and criticisms that helped improve the present essay. NOTES 1. “From a logical point of view, the historical starting point appears to be accidental most of the time.” Wolfgang Carl, who pointed out this quotation to us, added to it: “but not always”. 2. During the notorious “science wars” (see Sokal and Bricmont 1998), such lack of reflection has sometimes lead defenders of the authority of science to criticize certain historians and sociologists of science in unfair ways. Not only are there various forms of social constructivism. Some of the leading authors in the field have also developed their views or changed their research interests in important ways. Thus, in the second edition of their work, Latour and Woolgar have removed the word ‘social’ from the original subtitle “The Social Construction of Scientific Facts”. Latour now claims that explaining scientific knowledge in sociological terms is problematic because we do not understand society better than nature (Latour and Woolgar 1986, p. 281). 3. Kuhn probably has in mind claims such as that empirical statements cannot alone determine scientific theories (Kuhn 1970a, p. 4); that scientific fact and theory are not categorically separable (ibid., p. 7); that we should “replace the confirmation or falsification procedures made familiar by our usual image of science” (ibid., p. 8) and perhaps to other assumptions as well. 4. In a later paper, Kuhn advances considerations that seem to elaborate upon his first criticism against the DJ distinction (Kuhn 1977; for a discussion, see Siegel 1980, pp. 309–313; Hoyningen-Huene, this volume). 5. Barnes even thinks that Popper’s views might well be in accordance with the SP, because of conventionalist elements in Popper’s views concerning the “empirical basis” of science (Barnes 1972, p. 378, fn. 9). Popper would perhaps disagree.

154

THOMAS STURM AND GERD GIGERENZER

6. E.g., Kr¨uger 1974; Heidelberger 2003; Laudan 1984a; Kitcher 1993, chapter 7; Okasha 2000 (on underdetermination and/or theory-ladenness). Hesse 1985 additionally discusses the topic of incommensurability as it informs the SP. 7. Knorr-Cetina and Mulkay write: “. . . if the thesis that scientific theories are logically underdetermined by the evidence is correct, it removes one important constraint on theory acceptance which opens the way for social science investigation. Consider the opposite case, namely that only one out of many theoretical interpretations would ever be fully consistent with the evidence. If the proponents of this hypothesis could demonstrate that a set of data uniquely and conclusively supported their interpretations . . . , they ought to be able to gain acceptance of their interpretation against the will of their opponents.” (Knorr-Cetina and Mulkay 1982, p. 3) This argument hardly shows that the explanation of (the acceptance of) knowledge claims has to be in terms of social factors; it only indicates a possibility. Furthermore, why should opponents of the underdetermination thesis have to show that a theory should be able to “to gain acceptance [. . . ] against the will” of its opponents? A lack of epistemic alternatives should not—at least not without further argument—be conflated with a lack of alternatives in matters of decision or acceptance. 8. For instance, in an important and influential study on the origins of psychology as a laboratory and applied science, Kurt Danziger writes: “In more recent times the well-known contrast between ‘context of discovery’ and ‘context of justification’ gave expression to a pervasive tendency to relegate the necessary subjective component in scientific activity to a mysterious underworld that was not susceptible of logical analysis. So there grew up a strange duality in the historiography of fields like psychology, where one kind of historical review would restrict itself to the logical succession of hypotheses and evidence while the second kind would describe the personal lives of those individuals who were the authors of the hypotheses and the producers of the evidence. [. . . ] What is missing from this account is any appreciation of the fundamentally social nature of scientific activity. What unites individual contributors is not simply their common possession of the same logical faculties and their common confrontation of the same external nature. Their social bonds are a lot more complex like that. They are related by ties of loyalty, power, and conflict. [. . . ] Once we recognize the essentially social nature of scientific activity, we are compelled to see both the ‘context of discovery’ and the ‘context of justification’ in a different light. The context of discovery is in fact a context of construction, of theories, of instruments, and also of evidence. [. . . ] As long as we limit our conception of psychological research to its purely rational aspects, we will be inclined to think of the history of that practice solely in terms of technical progress. The norms of good scientific practice will be seen as belonging to an unchangeable transhistorical realm where eternal rational principles rule.” (Danziger 1990, pp. 2–6) Does Danziger mean that the DJ distinction is useless, because we should replace a “conception of psychological research to its purely rational aspects [. . . ] where eternal rational principles rule” by an analysis of the “of the fundamentally social nature of scientific activity”? Or does he claim that the context of discovery is not a non-rational affair, thus siding with those authors who prefer more complex distinctions between generation, invention, and justification (Nickles 1980b)? 9. Hence, Kitcher’s objection to the SP is misguided (1993, p. 184f.). 10. Haddock claims that “Barnes and Bloor do not say that sociological explanations must be wholly free of normative assumptions” (2004, p. 28). This interpretation clearly overlooks the impartiality principle. 11. Kitcher interprets the symmetry principle as the claim that both true and false beliefs can have causal explanations (Kitcher 1993, p. 184f.). However, the claim is rather that true and false should be given the same causal explanations. 12. If the claim that reasons can be causes is correct, then principle (4) of the SP becomes problematic as well: If the explanation of scientific beliefs cannot always be symmetrical, then the symmetry thesis cannot be seriously applied to sociological research as well. 13. Bloor, in the Afterword to the second edition of his Knowledge and Social Imagery, tries to make room for the asymmetry from the agent’s point of view: “Our everyday attitudes are practical and evaluative, and evaluations are by their nature asymmetrical” (Bloor 1991, p. 175). Yet, he does not

HOW CAN WE USE THE DJ DISTINCTION

14.

15. 16. 17.

18. 19.

155

take such evaluative attitudes seriously as having an explanatory value; he rather insists that “dictates of reason” must always be explained in terms of naturalistic (biological and cultural) factors that avoid any commitment to non-reducible logical and evidential relations (ibid., p. 178). A similar question is raised by Haddock 2004, p. 23.—One might think that a different distinction should be invoked here, namely the distinction between truth one the one side, and taking something to be true (accepting or believing it), on the other side. Hacking moves into this direction, when he speaks of the difference between our representations of objects or facts, on the one side, and mind-independent objects or facts, on the other side; or between processes of scientific research and their products (Hacking 1999, chapters 1–3). We do not object to these distinctions, but they will probably not convince the defender of the SP. Furthermore, because of the symmetry principle we need to address the claim that true and false beliefs are to be explained by the same causes. Propositional contents can be true or false, so it seems right to consider whether these can be the explanandum of the SP. Haddock 2004, 22 correctly notes that this must be so even if the SP’s claim is to discover “local” causes of credibility—the reasons cited within particular scientific communities. Thanks to Jutta Schickore for discussions on this point. We do not characterize such non-rational conditions as “irrational”. What is at stake in the present discussion is a theoretical or strictly epistemic conception of rationality: a conception that has to do with criteria for evaluating beliefs as true or false, or with standards of correct inference or justification. According to this notion, conditions such as group pressure and other “naturalistic” conditions (in Bloor’s terminology) are themselves not irrational, but simply non-rational. Of course, there are concepts of rationality according to which it can be rational (and irrational) to conform to group pressure or to disobey strictly epistemic criteria. Rationality is often defined in terms of means-ends efficiency or of purposiveness to (natural, psychological, social . . . ) functions. It might be profitable to pursue this further, say, by considering how Bloor’s talk of “natural rationality” relates to the second, functional notion of rationality. It might also be important to reflect whether we can explicate a functional conception that is not committed to any strictly epistemic criteria or reasoning. However, this leads into complex debates over the nature of rationality (see Gigerenzer 2003; Mele and Rawlings 2004). Such a function of scientific tools does not merely occur in psychology: see Lenoir 1986. The notion of apriority used here should not cause too many worries. Basically, we mean assumptions that make possible, for instance, certain research questions or experimental procedures. This does not mean that these assumptions are unrevisable or true once and for all. However, they cannot be empirically refuted within the very experiments and empirical research programs they make possible (Brandtst¨adter and Sturm 2004).

WORKS CITED Aaronson, D., Grupsmith, E. and Aaronson, M. (1976), “The Impact of Computers on Cognitive Psychology,” Behavioral Research Methods & Instrumentation 8: 129–138. Arabatzis, T. (1994), “Rational versus Sociological Reductionism: Imre Lakatos and the Edinburgh School,” in K. Gavroglu et al. (eds.), Trends in the Historiography of Science (Dordrecht: Kluwer), pp. 177–192. Barnes, B. (1972), “Sociological Explanation and Natural Science: A Kuhnian Reappraisal,” Archives Europ´eens de Sociologie 13: 373–393. Barnes, B., Bloor, D. and Henry, J. (1996), Scientific Knowledge: A Sociological Analysis (Chicago: University of Chicago Press). Bloor, D. (1976; 1991), Knowledge and Social Imagery (London: Routledge & Kegan Paul). Bloor, D. (1983), Wittgenstein: A Social Theory of Knowledge (London: MacMillan). Bloor, D. (1984), “The Strenghts of the Strong Programme,” in J. R. Brown (ed.), Scientific Rationality: The Sociological Turn (Dordrecht: Reidel), pp. 75–94.

156

THOMAS STURM AND GERD GIGERENZER

Brandtst¨adter, J. and Sturm, T. (2004), “Apriorit¨at, Erfahrung und das Projekt der Psychologie,” Zeitschrift f¨ur Sozialpsychologie 35: 15–32. Brunswik, E. (1943), “Organismic Achievement and Environmental Probability,” Psychological Review 50: 255–272. Castellan, N. J. (1981), “On-line Computers in Psychology: The last 10 years, the next 10 years—The Challenge and the Promise,” Behavioral Research Methods & Instrumentation 13: pp. 91–96. Danziger, K. (1990), Constructing the Subject: Historical Origins of Psychological Research (Cambridge: Cambridge University Press). Davidson, D. (1963), “Actions, Reasons and Causes,” in D. Davidson, Essays on Actions and Events (Oxford: Oxford University Press), pp. 3–20. Draaisma, D. (2000), Metaphors of Memory (Cambridge: Cambridge University Press). Edgington, E. E. (1974), “A New Tabulation of Statistical Procedures Used in APA Journals,” American Psychologist 29: 25–26. Fine, A. (1996), “Science Made Up: Constructivist Sociology of Scientific Knowledge,” in P. Galison and D. Stump (eds.), The Disunity of Science (Stanford: Stanford University Press), pp. 231–254. Friedman, M. (1998), “On the Sociology of Scientific Knowledge and Its Philosophical Agenda,” Studies in History and Philosophy of Science 29: 239–271. Gigerenzer, G. (1992), “Discovery in Cognitive Psychology: New Tools Inspire New Theories,” Science in Context 5: 329–350. Gigerenzer, G. (2003), Adaptive Thinking: Rationality in the Real World (New York: Oxford University Press). Gigerenzer, G., and Goldstein, D. G. (1996). “Mind as Computer: Birth of a metaphor,” Creativity Research Journal 9: 131–144. Gigerenzer, G. and Murray, D. J. (1987), Cognition as Intuitive Statistics (Hillsdale: Erlbaum). Gigerenzer, G. and Sturm, T. (forthcoming), “Tools = Theories = Data? On Some Circular Dynamics in Cognitive Science,” in M. Ash and T. Sturm (eds.), Psychology’s Territories (Mahwah, NJ: Erlbaum). Gigerenzer, G., Swijtink, Z., Porter, T., Daston, L., Beatty, J., and Kr¨uger, L. (1989), The Empire of Chance: How Probability Changed Science and Everyday Life (Cambridge: Cambridge University Press). Hacking, I. (1983), Representing and Intervening (Cambridge: Cambridge University Press). Hacking, I. (1999), The Social Construction of What? (Cambridge, MA: Harvard University Press). Haddock, A (2004), “Rethinking the ‘Strong Programme’ in the Sociology of Knowledge,” Studies in the History and Philosophy of Science 35: 19–40. Heidelberger, M. (2003), “Theory-ladenness and Scientific Instruments in Experimentation,” in H. Radder (ed.), The Philosophy of Scientific Experimentation (Pittsburgh: Pittsburgh University Press), 138–151. Hesse, M. (1985), “The Strong Thesis in the Sociology of Science,” in M. Hesse, Revolutions and Reconstructions in the Philosophy of Science (Bloomington: Indiana University Press), pp. 29–60. Hoyningen-Huene, P. (1987), “Context of Discovery and Context of Justification,” Studies in the History and Philosophy of Science 18: 501–515. Kant, I. (1781/1787; 1997), Critique of Pure Reason, ed. and transl. by P. Guyer (Cambridge: Cambridge University Press). Kantorovich, A. (1993), Scientific Discovery—Logic and Tinkering (New York: State University of New York Press). Kelley, H. H. (1967), “Attribution Theory in Social Psychology,” in D. Levine (ed.), Nebraska Symposium on Motivation, Vol. 15 (Lincoln: University of Nebraska Press). Kelley, H. H. and Michaela, I. L. (1980), “Attribution Theory and Research,” Annual Review of Psychology 31, 457–501. Kitcher, P. (1993), The Advancement of Science (Oxford: Oxford University Press). Knorr-Cetina, K. (1981), The Manufacture of Knowledge (Oxford: Pergamon Press). Knorr-Cetina, K. and Mulkay, M. (eds.) (1982), Science Observed (London: Sage).

HOW CAN WE USE THE DJ DISTINCTION

157

Kr¨uger, L. (1974), “Wissenschaftliche Revolution und Kontinuit¨at der Erfahrung,” Neue Hefte f¨ur Philosophie 6/7: 1–26. Kr¨uger, L. (1983), “Empirismus oder Realismus—eine Alternative in der Wissenschaftstheorie?,” in G. Frey and J. Zegler (eds.), Der Mensch und die Wissenschaften vom Menschen (Innsbruck.) Vol. II, pp. 569–587. Kuhn, T. S. (1962; 1970a), The Structure of Scientific Revolutions (Chicago: University of Chicago Press). Kuhn, T. S. (1970b), “Reflection on my Critics,” in I. Lakatos and A. Musgrave (eds.), Criticism and the Growth of Knowledge (Cambridge: Cambridge University Press), pp. 231–278. Kuhn, T. S. (1977), “Objectivity, Value Judgment, and Theory Choice,” in T. S. Kuhn, The Essential Tension (Chicago: University of Chicago Press), pp. 320–339. Latour, B. and Woolgar, S. (1979; 1986), Laboratory Life (Princeton: Princeton University Press). Langley, P., Simon, H. A., Bradshaw, G. L. and Zytkow, J. M. (1987), Scientific Discovery (Cambridge, MA: MIT Press). Laudan, L. (1984a), Science and Values (Berkeley: University of California Press). Laudan, L. (1984b), “The Pseudo-Science of Science?,” in J. R. Brown (ed.), Scientific Rationality: The Sociological Turn (Dordrecht: Reidel), pp. 41–74. Leary, D. E. (1987), “From Act Psychology to Probabilistic Functionalism: The Place of Egon Brunswik in the History of Psychology,” in M. G. Ash and W. R. Woodward (eds.), Psychology in TwentiethCentury Thought and Society (Cambridge: Cambridge University Press), pp. 115–142. Lenoir, T. (1986), “Models and Instruments in the Development of Electrophysiology, 1845–1912,” Historical Studies in the Physical Sciences 17: 1–54. McCorduck, P. (1979), Machines Who Think (San Francisco: Freeman & Co.). McMullin, E. (1984), “The Rational and the Social in the History of Science,” in J. R. Brown (ed.), Scientific Rationality: The Sociological Turn (Dordrecht: Reidel), pp. 127–163. Mele, A. and Rawlings, P. (eds.) (2004), The Oxford Handbook of Rationality (Oxford: Oxford University Press). Merton, R. K. (1973), The Sociology of Science (ed.) by Norman W. Storer (Chicago: University of Chicago Press). Michotte, A. (1963), The Perception of Causality (London: Methuen). Nelson, A. (1994), “How Could Scientific Facts Be Socially Constructed?,” Studies in the History and Philosophy of Science 25: 535–457. Newell, A., Shaw, J. C. and Simon, H. A. (1958), “Elements of a Theory of Human Problem Solving,” Psychological Review 65: 151–166. Newell, A. and Simon. H. A. (1972), Human Problem Solving (Englewood Cliffs: Prentice-Hall). Nickles, T. (ed.) (1980a), Scientific Discovery, Logic, and Rationality (Dordrecht: Reidel). Nickles, T. (1980b), “Introductory Essay: Scientific Discovery and the Future of Philosophy,” in Nickles 1980a, pp. 1–59. Okasha, S. (2000), “The Underdetermination of Theory by Data and the ‘Strong Programme’ in the Sociology of Knowledge,” International Studies in the Philosophy of Science 14: 283–297. Papineau, D. (2003), “Comments on Gerd Gigerenzer,” in M. C. Galavotti (ed.), Observation and Experiment in the Natural and Social Sciences (Dordrecht: Kluwer), pp. 141–151. Piaget, J. (1930), The Child’s Conception of Causality (London: Kegan Paul). Pickering, A. (ed.) (1992), Science as Culture and as Practice (Chicago: Chicago University Press). Putnam, H. (1960), “Minds and Machines,” in S. Hook (ed.), Dimensions of Mind (New York: New York University Press), pp. 138–164. Quine, W. V. O. (1978), “A Postscript on Metaphor,” in S. Sacks (ed.), On Metaphor (Chicago: University of Chicago Press), pp. 159–160. Reichenbach, H. (1938), Experience and Prediction (Chicago: University of Chicago Press), 1938. Schaffer, S. (1994), “Making Up Discovery,” in M. Boden (ed.), Dimensions of Creativity (Cambridge, MA: MIT Press), pp. 13–51.

158

THOMAS STURM AND GERD GIGERENZER

Siegel, H. (1980), “Justification, Discovery and the Naturalizing of Epistemology,” Philosophy of Science 47: 297–321. Simon, H. A. (1979), “Information Processing Models of Cognition.” Annual Review of Psychology 30: 363–96. Smith, L. D. (1990), “Metaphors of Knowledge and Behavior in the Behaviorist Tradition,” in D. Leary (ed.), Metaphors in the History of Psychology (Cambridge: Cambridge University Press), 239–266. Sokal, A. and Bricmont, J. (1998), Fashionable Nonsense (New York: Picador). Sterling, T. D. (1959), “Publication Decisions and their Possible Effects on Inferences Drawn from Tests of Significance or vice versa,” Journal of the American Statistical Association 54: 30–34. Tanner, W. P., Jr. (1965), Statistical Decision Processes in Detection and Recognition (Technical Report) (Ann Arbor: University of Michigan, Sensory Intelligence Laboratory, Department of Psychology).

THOMAS NICKLES

HEURISTIC APPRAISAL: CONTEXT OF DISCOVERY OR JUSTIFICATION?

1. INTRODUCTION

Many have noted the irony of the English title, The Logic of Scientific Discovery, of Karl Popper’s expanded translation of his Logik der Forschung (1934). Given Popper’s use of the distinction between context of discovery and context of justification (the DJ distinction), there is no such thing as a logic (or method or even rationality) of discovery. Yet the book is nearly 500 pages long!1 But instead of once again looking at how the DJ distinction discouraged attention to what Popper termed “the initial stage, the act of conceiving or inventing a theory” (Popper 1959, p. 31), I shall examine the privileged context of justification. I maintain that context of justification can be separated into two components, which I shall term epistemic appraisal (EA) and heuristic appraisal (HA). EA attends to truthconducive features of justification and decision-making, while HA attends to a variety of heuristic and pragmatic considerations relating to economy of research. In my view we lack good accounts of both EA and HA. Mainline treatments of EA by logical empiricists, Popperians, Bayesians, and others who articulate theories of justification, as licensed by the received DJ distinction, overlook major features of research2 ; and those same philosophers have pretty much ignored HA, except for the occasional throwaway line. In this paper I shall waive the shortcomings that I find in standard EA itself in order to make HA the center of attention. If I am right, standard EA leaves us with a severely limited account of scientific decision-making, even when the key materials are already assembled and on the table for appraisal. HA is as important as EA in making and justifying scientific research decisions. Do not philosophers of science have a responsibility to investigate this sort of evaluation and justification also? HA evaluates the promise or potential fertility and feasibility of further work on a problem, research program, theory, hypothesis, model, or technique. HA estimates the likely return on investment in expensive equipment, in reorganizing a laboratory, or even in adding a new member to the research team. It may be possible to factor HA itself into predominantly heuristic and predominantly economy-of-research concerns, but I shall not attempt that here. We do know that heuristics can be strongly linked with economy. In any case HA extends to larger institutional arrangements and government and corporate science and technology policies as well, e.g., the decision whether to start a new proteomics institute or society with its own journal, the Rockefeller Foundation’s decision, in the early 20th century, to fund biological research, and the U.S. Congress’s decision to fund, then not to fund, the superconducting supercollider. 159 J. Schickore and F. Steinle (eds.), Revisiting Discovery and Justification, 159–182.
C 2006 Springer. Printed in the Netherlands.

160

THOMAS NICKLES

However, for simplicity of direct comparison with EA, I shall often couch my points in terms of the HA of hypotheses, models, theories, and research proposals and programs.3 It would appear that HA has one foot in context of discovery and one foot in context of justification yet belongs to neither as defined by the standard DJ distinction. Do we need a third “context” in our distinction? In his contribution to this volume, Paul Hoyningen-Huene teases out several DJ distinctions, or at least many different ways of interpreting and applying the distinction(s) originally drawn by Hans Reichenbach (1938) and other analysts such as Popper. Hoyningen ultimately defends a “lean” version of the DJ distinction, one that he believes no one (neither the logical empiricists and Popperians nor their critics) has disputed. This is the distinction between a descriptive perspective and a normativeevaluative perspective. In this understanding, ‘context of discovery’ refers to the descriptive perspective.4 It concerns questions seeking descriptive answers, presumably questions about how scientists actually designed and performed the experimental tests as well as how they hit upon the hypothesis being tested. Meanwhile, ‘context of justification’ refers to normative-evaluative issues, such as whether or not the work meets (or met) the scientists’ own standards (object level evaluation) and whether it meets our standards (metalevel evaluation). Hoyningen adds that the DJ distinction, so described, is not exhaustive, that there may be room for additional perspectives. He does not list them by name but he presumably means activities that various philosophers have labeled prior appraisal, preliminary evaluation, plausibility assessment, promise, pursuit, fertility assessment, and heuristic appraisal. Does the existence of an irreducible HA challenge the lean DJ distinction as well as the traditional distinction? If so, should we accommodate it by introducing a third perspective? I shall attempt to answer these questions after explaining in more detail what I mean by HA and after outlining my reasons for denying that HA reduces to EA and suggesting reasons why HA has been neglected. A systematic review of previous attempts to deal with HA issues will have to await another occasion5 , nor do I have space here to say much about what positive accounts of the various dimensions of HA might look like. 2. HEURISTIC APPRAISAL (HA) AND EPISTEMIC APPRAISAL (EA)

Imagine that you are a member of a national grants panel whose job it is to evaluate research proposals and to select a fraction of them for funding. What do you look for? Surely you look for more than reasons for thinking the proposed hypothesis or model is true. Among other things, you assess the feasibility of the project, the quality of the research design, and the principal investigator’s competence to carry through. As for the hypothesis, its being scientifically interesting is at least as important as its being true. It must address an extant or newly formulated problem or engage in potentially valuable exploration of some kind, and it must promise benefits for future research. Something significant must be at stake. A hypothesis can be important and interesting

HEURISTIC APPRAISAL: CONTEXT OF DISCOVERY OR JUSTIFICATION?

161

even if it is already known to be false. And if the proposal is for a conference or for a piece of equipment, for example, questions of evidential support and truth do not arise directly at all. How might you have reacted in 1948 to the proposal to spend a large sum of the taxpayers’ money to build an atomic accelerator of unprecedented size (the Cosmotron), capable of reaching the energy of cosmic rays, at the largely undeveloped new Brookhaven Laboratory on Long Island, given that many of the design ideas were still sketchy and untried and would have to be worked out as the construction took shape? This was a risky project facing both unsolved and unknown problems. And, assuming that it could be done, would the tremendous cost be worth it compared to other projects that might be funded, given that physicists had never before worked at this energy?6 Here you would have faced a double HA task: the HA of the eventual but still unknown applications piggybacked upon the estimated scientific and technological value and feasibility of building the accelerator itself. Now look at the other end of the grants process. Imagine that you are a research scientist with good training who has already completed some successful work. While still immersed in your current project, you and your associates are beginning to think about what your next project might be, for which you shall, of course, need grant support. How do you select the project? How do you frame the grant proposals? Here again traditional confirmation theory will be far from sufficient. Traditional EA does help to shape the field and its problems, of course. And you may have a pilot project in which the preliminary data are promising (preliminary EA). But you still need to show why it is important to work on this project, given that yours is competing with many other proposals. EA alone cannot decide questions of opportunity cost, either for you or for the grant committees. Heuristic appraisal (HA) is the term that I apply to the cluster of considerations that are either distinct from epistemic appraisal (EA) or that enable us to evaluate the same information from the standpoint of future fertility. HA evaluates the promise, the future potential (including what is at stake), the problem-solving capacity7 , or what we might call the “opportunity profile” of a claim, technique, proposal, etc. If I am right that HA is a distinct form of evaluation in context of justification and that HA is indispensable to research, then standard confirmation theory has it practically backwards. What justifies a scientist’s decision about what to work on now, or next, is not its past—its confirmation track record—so much as what it promises for future innovation. The question is not “What have you done for me?” or even “What have you done for me lately?” but “What can you do for me tomorrow?” Scientists, like farmers and business people, are future oriented. Economic forecasts and the “futures” market are more important today than what happened last year. Research decisions are investments in the future. Good research is highly adaptive to changing situations, including local contingencies. In a word it is opportunistic. As Einstein once put it, “The scientist . . . must appear to the systematic epistemologist as an unscrupulous opportunist.”8 EA alone is not highly adaptive in this way. HA is far more sensitive to opportunity—and to lack thereof.9

162

THOMAS NICKLES

The flavor of HA versus EA is also captured by Streater and Wightman (Streater and Wightman 1964, p. 1), who report about the once-suspect quantum field theorists: Cynical observers have compared them to the Shakers, a religious sect of New England who built solid barns and led celibate lives, a nonscientific equivalent of proving rigorous theorems and calculating no cross sections.

EA and HA agree that something is missing here, but their perspectives differ. EA emphasizes the lack of predictive testability, whereas HA worries that there is not much that scientists can do with the theory in question.10 An absolutely true theory is no help if scientists cannot use it for anything. At some stages of research, fertility is more important than either conceptual purity or epistemic credibility. Some of our most fertile theories, models, and programs have been known in advance to be false and even inconsistent (or otherwise incoherent) at some point—a mortal sin for logical empiricist confirmation theories, Popperian corroboration theories, and their successors. Familiar examples from Einstein’s three most famous papers of 1905 will serve as reminders.11 In his quantum hypothesis paper, suggestively titled “On a Heuristic Point of View [heuristischen Gesichtspunkt] about the Creation and Transformation of Light,” Einstein deliberately employed Wien’s blackbody radiation formula rather than Planck’s, even though Wien’s was already refuted by the experimental results that confirmed Planck’s. This ploy enabled Einstein easily to derive the existence of free quanta in the high frequency limit. His Brownian motion model of that same year contained an inconsistency, yet the model was extremely influential in helping to establish the atomic-molecular theory of matter. Third, Einstein’s special theory of relativity invoked clocks attached to rigid rods, yet there can be no rigid rods within relativity theory (Brush 1968, p. 18; Nickles 1980). In milder cases, scientists employ models that are simplified approximations or idealizations. We could cite any number of such examples.12 There are also plenty of examples in which a presumably true theory or correct result is unfruitful as a future research site because it leads nowhere, is impossibly difficult to use (as suggested by the Streater and Wightman quotation), or, while significant in itself, basically finishes off a field of research. An example of the last is David Hilbert’s solution, in 1893, of the last major problem of invariant theory, leading him to move on to greener pastures. The pastures need be only relatively greener. Physicists such as Max Delbr¨uck, who moved into biology and eventually founded the field of molecular biology, did not presume that there were no interesting physics problems left, only that they could make a greater contribution to the other field, which seemed ripe for their sort of expertise. Another example is Francis Crick. Later, both men moved on to still other fields, and encouraged their prot´eg´es to do the same, because they believed that the main creative phase of molecular biology was about to play itself out.13 Let me put the basic point of this section in a different way. Discussions of statistical testing commonly distinguish Type I error from Type II error. Rejecting a true hypothesis is a Type I error, while accepting a false hypothesis (in relation to the null hypothesis) is Type II. The first is a false negative, while the second is a false

HEURISTIC APPRAISAL: CONTEXT OF DISCOVERY OR JUSTIFICATION?

163

positive. We can draw a similar distinction in terms of HA rather than EA. Rejecting for further research a more fertile hypothesis than a competitor would be a Type I error, and accepting a relatively sterile hypothesis would be Type II. The point is that there is no correspondence of epistemic Type I and Type II errors with heuristic Type I and Type II errors. Either type of epistemic error may concern a hypothesis with high or low fertility, and vice versa.14 One other difference for now. Statements and their logical or probabilistic relations just “sit there” or exist in some abstract sense. They do not wear their importance on their sleeves, not even their epistemic importance. In that regard, all truths are equal. But in science they are, of course, not equal. To which truths (and nontruths) should scientists in a given situation attend? If the norm “Seek the truth!” provides little guidance, what should scientists do? My answer: They need to perform HA. It is HA, in my broad sense, that provides the “focus” or “attention” to some problems, claims, and practices above others, the directive toward what it is worth trying next. The “What next?” problem arises in a variant form in rich scientific contexts in which there is so much information that the problem becomes one of “knowledge pollution.” How do scientists pick out what is salient when choosing a problem, then searching the literature and calling upon their background knowledge? At this point, HA becomes more important than EA, for the sought-for, salient results of previous EA are, epistemically speaking, hidden among zillions of other results that are perhaps equally good, epistemically speaking. Aside from a bit of lip service to fertility, most philosophers have treated justification as a purely epistemic affair, in effect making this part of philosophy of science a subdomain of epistemology in a rather narrow sense that excludes concern for future innovation. Probably the most common view is that since scientists seek the truth, theory of justification should concern itself with methods or features that are truthconducive when applied to materials already produced by some kind of discovery or inventive construction process. For example, the central question about simplicity (aside from how to define it) has always been whether or not it is truth-conducive. Even when philosophers distinguish different levels of acceptance of a claim (or a practice), these levels are often epistemic levels—degrees of tentative epistemic acceptance. While ‘promise’ can mean ‘promise of being true’, or ‘promise of being more accurate than what we now have’, it often means something quite different, something not directly correlated with truth, something that is not truth-conducive in any simple way, as we have seen. Even those philosophers who have proposed a middle, historical or logical stage between original generation and final justification (‘preliminary evaluation’, ‘plausibility assessment’, ‘fertility assessment’, etc.) have tended to treat HA as just anticipatory EA. However, evaluating fertility is distinct from evaluating possible truth or possible future degree of confirmation. I certainly do not wish to deny that scientists are concerned with truth, both everyday local truths (“The voltage regulator is working properly.” “This sample has been contaminated.”) and high-level truths. Scientists work very hard to convince themselves and others that a phenomenon they are observing or producing is not an artifact, as in the fascinating cases of Dufay and Zeeman discussed in this volume by Friedrich

164

THOMAS NICKLES

Steinle and Theodore Arabatzis. However, truth and purely epistemic justification do not figure as heavily in scientific decision-making as most confirmation theories and theorists would have us believe. Scientists are not content (as many philosophers seem to have been, from the Greeks to this day) to become spectators of the universe, with their models and theories being representational objects of epistemic admiration and aesthetic enjoyment. Rather, they want and need tools that they can use to get on with their work. If philosophical accounts of science are supposed to describe and explain basic patterns of scientific investigation, HA demands more attention; for standard varieties of confirmation theory by themselves explain very little. One can, of course, treat them as purely normative ideals, but insofar as they fail to explain what scientists do, and why, the question of their relevance to actual historical and contemporary scientific practice becomes urgent. To be sure, as I have characterized it, HA is a loose, catch-all category that includes a great diversity of things, including pragmatic considerations as well as strictly heuristic ones. Eventually, these need to be sorted out, and it is doubtful whether any elegant, unitary account will be possible. However, I do not see that as an objection to my central thesis in this chapter, that we need to study HA as well as EA in order to understand how science works—and that several aspects of HA intersect genuinely philosophical concerns and responsibilities. On the contrary, the more dimensions to HA, the more grist for my mill! In section 3, I shall summarize and extend my reasons for denying that HA can be reduced to EA or that it is somehow derivative from EA. It will turn out that some HA does depend upon EA, but even there HA evaluates the situation from a different perspective, as I shall contend in section 4. 3. IS HA REDUCIBLE TO EA?

My answer is “no,” for several (highly overlapping and criss-crossing) reasons. Since there are many conceptions of EA and HA, individual reasons will carry more or less weight, depending on the context. 1. HA is needed before EA even begins, for at least four reasons. First, HA is necessary for problem choice prior to the stage of hypothesis formation. It is also likely to be useful in various ways during hypothesis formation as well as in searching for ways to test the hypothesis once available. Fourth, scientists must deem the hypothesis worth testing. Testing has a cost. Contrary to Popper and many others, not every hypothesis or model remains on the table until it is explicitly rejected. Scientific claims are in effect sterile unless and until someone finds them worth testing or at least worth citing. Like biological organisms, they contribute nothing by simply not being refuted (not dying). They must in effect mate and reproduce themselves. The Physical Review is a journal that now includes thousands of pages per year. Do all the hypotheses, models, and computational techniques proposed therein become and remain active research sites at the frontier

HEURISTIC APPRAISAL: CONTEXT OF DISCOVERY OR JUSTIFICATION?

2.

3.

4.

5.

6.

7.

165

of research until explicitly refuted? No. Most of them fall “deadborn from the press.”15 An eliminative methodological strategy of Popper’s sort would be hopelessly uneconomical.16 Later in the research cycle, HA is necessary even after EA is “finished.” The most successful theory or model may be abandoned as an active research site if the relevant experts believe that it has exhausted its resources, its problem-solving or exploratory potential. This does not, of course, mean that it will be useless to further research and no longer cited. After all, classical mechanics is still useful for many purposes although not an active research site. Good research is highly adaptive to changing circumstances. HA is far better at tracking those opportunistically than EA alone. We could plausibly amplify this point by contending that scientific research itself is a kind of evolutionary process that depends on exploiting “ecological” opportunities of the research context, but I shall leave the more contentious claims for another occasion. The scope of HA is larger than that of EA. HA is ubiquitous: it figures at all stages of research. EA evaluates the epistemic warrant of descriptive claims. HA can evaluate those same claims from heuristic and pragmatic perspectives (and can also piggy-back on epistemic evaluations themselves: see section 4), but HA evaluates many kinds of things besides descriptive claims—problems, techniques, equipment, personnel, etc. (See also point 22 on “internal” versus “external” factors.) Specifically, unlike EA, HA weighs prudential reasons as well as evidential and testimonial ones. Evidential reasons increase the (subjective) probability of a claim, as can testimonial reasons (as when reliable experts publish their results). Prudential reasons are those that, in themselves, do not alter the probability but take into account what is at stake, the investment risk, the utilities as well as the epistemic probabilities. In this context, HA asks not (or not only) “Is it true (or probable)?” but “So what? What is at stake if we succeed or fail?” Moreover, the probability of making research progress is not the same as the chance of increasing the epistemic probability of any specific claim or even the probability of moving closer to the metaphysical truth about the world. A good question for a more refined treatment of HA is whether these pragmatic considerations can be factored into a strictly heuristic component and a prudential component. EA tends to be optimizing, or at least oblivious to economy of research and financial economy. EA preaches that, insofar as truth is the goal, we should spare no expense and should be satisfied with nothing less. HA is satisfying17 and economy-oriented. HA is indispensable for the “rational” setting of goals and standards since it evaluates them in terms of realizability and fertility. For example, Thomas Kuhn criticized Paul Feyerabend’s call for proliferation of major alternative theories as hopelessly unrealistic on both cognitive and pragmatic grounds (material realizability). He also rejected Popper’s methodology as something that would kill science as we know it if implemented, partly for the lack of heuristic guidance that it furnishes as opposed to Kuhnian normal science with its exemplars, learned

166

8.

9.

10.

11.

12. 13.

THOMAS NICKLES

similarity relations, and resulting unanimity of agreement. Another example is Popper’s assertion that all methodological rules are conventions, yet that some conventions are more fruitful than others.18 He thereby seemed to concede that choice of methodology is underlain by HA. I believe that this is true and worth emphasizing as a separate point. HA underlies EA in the sense that which mode of EA scientists choose comes down, in part, to a fertility-feasibility assessment. A main reason that the risky method of hypothesis replaced old-style inductive methodologies during the 19th century was that it was judged more fruitful, even more feasible, for theoretical research (Laudan 1981; Nickles 1987a). Given our present topic, this change is especially interesting, since the method of hypothesis, formerly treated as mere heuristic scaffolding, now became the epistemic centerpiece of scientific method. And it was hypotheticalism that in turn inspired the DJ distinction, which does not fit well with an inductivist methodology. (See Gregor Schiemann’s contribution to this volume.) Indeed, we can even question the heuristic fertility of the DJ distinction itself, at both the object level of scientific work and the metalevel of historical and philosophical commentary. HA concerns a different logical modality than EA. EA evaluates the actual problem solving and predictive success to date of a model, theory, research program, etc., whereas HA evaluates its possibilities, what we may call its problem-solving potential or exploratory potential (in the case of pure experimental or theoretical exploration). A related point is that the grammar of HA differs from that of EA. EA makes assertions about logical relations among claims that actually exist, and normative judgments based on those; whereas HA requires subjunctive thinking of the form, “If we were to pursue research program X, we would probably be able to accomplish Y.” HA is more rhetorical than EA in terms of the useful employment of tropes such as analogy, simile, and metaphor. These are notoriously weak modes of reasoning when it comes to justification, yet they can provide invaluable “intuition pumps” in contexts of innovation and HA and legitimate modes of persuasion in making research choices. For example, Kuhn’s modeling of new problems (or research puzzles) upon exemplars by means of what he calls “the learned similarity relation” is better described as rhetorical rather than logical. Rhetorical persuasion is also essential to paradigm choice in times of crisis. EA is past oriented, toward the “track record” to date. HA is future oriented, toward future development. EA is the basis of reports on what has been accomplished, at what cost. HA is the estimate of what can be accomplished in the future, in how much time and at what cost. EA is theory-centered, whereas HA is practice-centered. EA alone does not yield an adequate demarcation criterion. HA can help. Science cannot be defined simply as a body of established knowledge or even that plus a standard method of justification (EA). At the frontier of research, by definition, there are no experts in the sense of people who possess detailed knowledge of that domain, but there are experts in the sense of those who know how to

HEURISTIC APPRAISAL: CONTEXT OF DISCOVERY OR JUSTIFICATION?

167

proceed and (more to the present point) who can fairly reliably compare the future prospects of competing research proposals for that domain. This is a major difference between scientific research programs at the frontier and claims put forward by fringe movements and fraudulent groups such as creationists (Rouse 2003). At a time when the difference in positive EA is small (between alternative solutions, e.g., to the problem of the origin of life), the difference in HA can be large. 14. HA itself employs heuristics, often quick-and-dirty heuristics of a kind that EA cannot employ, as point 15 makes evident. 15. EA requires less constructive work than EA, since it concerns the logical and mathematical relationships among theoretical and data claims already on the table; whereas HA can be highly constructive, involving sketches of future models, theories, experimental designs, and the like, based on those items already on the table. The difference here is no longer so clear cut for us as it was for the logical empiricists, since we now appreciate how much modeling may intervene between already available data and theory, and we recognize that later investigators may reconstruct this relation by reanalyzing the data and its fit with theory. However, there remains a significant difference. To state this point in terms of a logical difference: present-day EA is mainly consequentialist—comparing hypotheses with regard to the truth of their predictive consequences (in the context of relevant auxiliary assumptions). By contrast, HA is heavily generative, reasoning in a tentative and sketchy way toward possible future results.19 Fully developed HA includes what we might call discovery sketches of the sort that one finds in good research proposals and in published papers, as hinted in James Watson and Francis Crick’s famous understatement: It has not escaped our notice that the specific pairing we have postulated immediately suggests a possible copying mechanism for the genetic material (Taylor 1965, p. 256).

16. As epistemic evaluation, EA is chiefly concerned to answer questions about epistemic credibility, correctness, degree of confirmation or degree of corroboration of a theory, whereas HA is primarily concerned to answer the questions, “Where do we go from here? What would be a good project to do next?” “Is the project feasible for anyone right now? For us?” In other words, HA addresses more pragmatic issues than does EA. 17. HA thus addresses an underdetermination problem at least as serious for working scientists as the logical underdetermination of theory by data plus logic advertised by Pierre Duhem (1954), W. V. Quine (1951), and Kuhn.20 The standard underdetermination problem is an epistemic one of which theory to accept. But even when the epistemic situation is fully determined, as we have seen, the question of what to work on next remains to be answered. Epistemic determination leaves future scientific action underdetermined. Contrary to logical empiricism and Popperianism, this last point holds even when the epistemic determination is negative rather than positive. That is, even knowing that a theory is false, indeed, even knowing it to be inconsistent or otherwise incoherent in some respect, does not automatically doom it from further

168

18.

19.

20.

21.

THOMAS NICKLES

consideration. Scientists may yet decide, on the basis of HA, that it is highly pursuitworthy. EA death is not HA death, and vice versa. After all, no major development ever entered the scene completely free of difficulties. Eliminating them is what pursuit is for. Deciding that a defective theory or model is worthy of further pursuit amounts to launching or continuing a research program. As traditionally understood, EA is insufficient to support accounts of scientific work in terms of research programs or research traditions. The EA of the logical empiricists and Popperians was intended to evaluate individual hypotheses or theories, and the same is true of typical Bayesian accounts. Positive EA implied that a theory should be tentatively accepted, further pursued if incomplete, and used in further research, while negative EA implied the opposite. Of course, one can also evaluate the track record of a research program to date, as with Imre Lakatos’s tally of novel predictions and predictive successes of one program against another; but, as his work makes clear, nothing at this EA level forces scientists either to continue or to abandon the research program itself. HA, an appraisal of the programs’ available resources and potential, is needed to do this. Standard EA cannot account for scientists’ choosing underdeveloped research programs over “war horse” programs with a tremendous record of success. (I further discuss this innovation problem in the next section.) Indeed, I claim that EA alone cannot account for much of the activity of scientists in any field. This is not to deny the importance of EA but, rather, to emphasize the need for HA as well. HA is required to solve the epistemic closure problem (already the focus of Rudner 1954 on value judgments). This is yet another way in which HA underlies EA, in this case, specific applications of EA. Typically, individual applications of EA require HA before detachment is justified, in order to answer the research counterpart of the standard risk-analysis question, “How safe is safe enough?” There are always various alternatives and objections that could, logically speaking, be pursued. But, at a certain point, HA determines that they are not worth pursuing further, not likely to bear fruit, and it is in this manner that epistemic closure is achieved, unless and until some new consideration is forthcoming.21 And similarly for larger questions: How degenerating must a research program be before we should abandon it? At what point should we seriously consider a paradigm switch? Answers to these questions typically require HA as well as EA. They are hard enough to answer in any case. To sum up, EA and HA amount to different “control theories” for scientific activity. (A control theory is a theory of “the locus of control”—the key variables that control a process or system.) Neither EA nor HA alone is adequate by itself. We need logistics not just logic (logistic)! EA captures only some of the factors that govern judicious scientific decisions and that define the active research sites and the research frontier. Too many philosophers have written as if EA is the controlling evaluation that regulates scientific activity, given sufficient funding and the absence of political interference. But EA alone cannot determine the relative importance of different truths or the direction that science should take.

HEURISTIC APPRAISAL: CONTEXT OF DISCOVERY OR JUSTIFICATION?

169

Without HA, research has limited ability to focus, limited ability to recognize salience, especially since opportunistic research may alter its goals, standards, and research methodologies in medias res. EA alone has trouble furnishing starting rules and stopping rules. Why bother pursuing this hypothesis at all, given that we cannot possibly test them all? And how much confirmation is required before we can “detach” a scientific conclusion, given the stakes of doing so? Again, this last point deserves a separate entry. 22. EA proper is restricted to consideration of what used to be called “internal factors,” whereas HA can take into account “external factors” as well. HA takes into account material resource issues that are “externalities” as far as EA is concerned. In deciding what to work on next (which may, of course, be simply a continuation of their current work), scientists must take into account such external factors as whether their research is likely to be funded; whether the lab director or department head will look favorably upon this project; whether enough laboratory space, equipment, and expert technical assistance is available; whether the necessary samples or data can be obtained from other labs; and so on. These issues affect the feasibility of real research just as much as the internal, technical feasibility of the project. I conclude that EA is neither necessary nor sufficient for HA in general, nor is there any regular correlation between EA and HA. The same item may have a high EA but a low HA, and vice versa. Defenders of the various versions of EA will have plausible responses to some of the items above, but I know of no account of epistemic justification that comes close to handling all of them. Here is one example of such a response, and my reply. The usual objection to point 22 is that philosophy is primarily concerned with epistemic issues and does not presume to provide a complete explanation of scientific activity, that external factors are legitimately left as externalities or exogenous variables, to be addressed by other specialties. There is a grain of truth to this objection. However, it seems strange that philosophers who consider the problem of the growth of knowledge central should make innovation itself an externality! Their stance is analogous to that of standard economists who regard innovation as the driver of the economy yet treat innovation as an externality (Freeman and Soete 1997). In both cases, the experts leave the key factor of innovation as a black box that they take no responsibility for opening. The result is an account that, at a certain point, becomes magical, an unexplained explainer, a deus ex machina. What is the source of, or the justification for, this externalizing of innovation— whether in original context of discovery or in HA? My answer: It is the DJ distinction, coupled with the internal–external distinction and with what we might call “romantic hypotheticalism.” This last is the Popperian view that fertile scientific theories, musical works, and poems are the work of inarticulate genius. Several logical empiricists (not to mention later analysts) also found this view attractive. Yet it denies philosophy of science the very resources necessary to deal with one of its central problems. In fact, this observation leads us to question the fertility of the traditional DJ distinction

170

THOMAS NICKLES

itself. If, indeed, the problem of the growth of knowledge is central to understanding scientific research, then what justifies imposing the old-style DJ distinction so as to exclude not only creative activities in the context of discovery but also HA? What justifies restricting justification to the fit between a hypothesis and the phenomena, when every mature scientific discipline has other resources as well? Having declared the independence of HA from traditional theory of justification, the logical next step would be a section showing that it is also independent of context of discovery as the traditional DJ distinction characterized it. For those who minimize the importance of HA as a separate topic are likely to argue that it collapses either into context of justification on the one hand or context of discovery on the other. As a matter of fact, HA is not totally independent of context of discovery, since some forms of HA coincide with what may become the next cycle of innovation. But since HA’s raison d’ˆetre is its normative-evaluative function, it ill fits the so-called context of discovery. 4. WHY HAVE PHILOSOPHERS NEGLECTED HEURISTIC APPRAISAL?

Why have philosophers given HA a tiny fraction of the attention devoted to EA? I shall briefly mention several contributing reasons, then make two extended comments. 1. The received DJ distinction together with its analytical presuppositions (noted by Hoyningen in his chapter) links philosophy of science more closely to traditional philosophical problems of knowledge, justification, logic and even metaphysics than to actual scientific practices of evaluation and decision-making. The distinction reinforces the idea that philosophy is a non-naturalistic, a priori discipline not answerable to careful empirical studies of scientific behavior. 2. The traditional DJ distinction, especially when viewed through the lens of Popperlike romantic hypotheticalism and the internal–external distinction, rules out HA as impossible since it combines discovery and justification, two mutually exclusive things. If possible at all, it does not belong to philosophy. To be sure, norms and values are a traditional philosophical domain, but the logical empiricists were only minimally interested in value theory as a philosophical topic. For them it was non-cognitive, and, especially in their American phase, they located philosophy of science closer to epistemology and further from value theory. Logical norms were norms enough. 3. Although philosophy of science is parasitic upon the sciences as its subject matter, philosophers and scientists have different professional interests. Philosophers of science are naturally interested most in those aspects of science that provide work for philosophers themselves, where they enjoy a kind of professional privilege— as in logical analysis. HA is not one of those places. HA requires evaluations and judgments of a subtle kind performed by expert scientists themselves and not (so far) captured within a formal system. 4. One need not harbor logical empiricist sympathies to regard ‘heuristic appraisal’ as oxymoronic. For how can we appraise something in terms of results not yet

HEURISTIC APPRAISAL: CONTEXT OF DISCOVERY OR JUSTIFICATION?

5.

6.

7.

8.

9.

171

produced? In such cases aren’t we really pretending to evaluate nonexistent results? How can we anticipate discoveries before they are actually made? After all, if we can predict them in sufficient detail to evaluate them by means of HA, then have we not already made them? As Popper liked to say, “We cannot know now what we shall only know later.” Insofar as research is serendipitous, this problem is even worse. On an evolutionary model of innovation, every adaptive development depends upon the exploitation of chance contingencies in the production of variants and in the “environment” (Nickles 2003). These cannot be known in advance in science any more than in evolutionary biology. Whether or not we invoke an evolutionary model, forecasting the future in any field is notoriously risky. Done systematically, it would seem to require a “science” of futurology, when all we have in that field is pseudoscience. We have substantial empirical evidence that such forecasts are extremely unreliable. They can be humorous, as when leading figures in 1900 made such predictions as that, in the year 2000, aircraft would take the form of large balloons pulled between cities by wires.22 Careful studies have also questioned the mechanism used to award National Science Foundation grants, e.g., the peer-review system (Cole et al. 1978). The lure of technical formalism has worked against attention to HA. Many analysts believe (have believed) that EA is susceptible to crisp formal treatments, whereas HA is so variegated and context-relative that only a fuzzy, case-by-case approach is possible. And when formal logical and semantic theories of confirmation failed, subjective Bayesian confirmation theory stepped in to renew the formalist dream. Bayesian confirmation theorists assert that they can accommodate HA in the prior probabilities of their Bayesian calculi.23 I disagree, since their probabilities are epistemic, thus restricting HA at best to prior epistemic appraisal. Also, simply claiming to accommodate HA intuitions does not thereby furnish an account of them. Many hold that HA is based on intuitive hunches, the inarticulate wisdom of expertise rather than on explicit reasons—certainly more so than EA. Even if these intuitions are somewhat reliable, there is not much that philosophers (or anyone else) can say about HA. HA keys on the opportunistic dimension of scientific practice, whereas philosophers, unlike historians and science studies practitioners, have always tended to avoid the particular, contingent, concrete, and local in favor of the general, necessary, and abstract. The current interest in moral luck is an exception that proves the rule, since ‘moral luck’ is practically an oxymoron for traditional moral theories. The same holds for ‘methodological luck’ in philosophy of science. Scientists flexibly mold goals and values themselves to fit opportunity, making philosophical methodologists look like prigs in comparison. Traditional epistemology, methodology, and theory of rationality stand to working science in roughly the way the “mirror of princes” moral tradition in early political philosophy stood to actual political practice. This was the point of that great pragmatic opportunist— Machiavelli—in The Prince. Corresponding to the real politicians’ ragioni di stato, we have the enterprising scientists’ ragioni di ricerca.

172

THOMAS NICKLES

10. HA does depend upon EA in some ways, suggesting to some that all HA is derivative from EA, for example, that HA is just preliminary EA.24 11. A good deal of the importance of EA itself lies in its heuristic value. Conflation of the EA and HA perspectives has fostered the illusion that EA is where all the action is. These reasons go far toward explaining why HA has been an unpopular topic. I cannot respond to all of them here other than to say that, given its ubiquitous importance to scientific work (not to mention technology and innovation and policymaking of any kind), the difficulty of analyzing HA should not deter us from doing our best to understand it. Two extended comments will have to suffice. Comment one. To those who contend that HA, insofar as it is of philosophical interest, collapses into EA, I reply that EA has unfairly co-opted several features that properly belong to heuristic appraisal (HA). A historical example is William James’s conflation of truth and fertility, in his so-called pragmatic theory of truth (James 1907, Lectures II and VI). James emphasized verification as a process (veri-fication or truth-making) and often made heuristic points using epistemic terms. He spoke of truth as “leading,” as fruitful, as having “cash value.” Although he should be applauded for his emphasis on heuristic fertility, his critics were right to accuse him of conflating EA and HA (in both its heuristic and pragmatic dimensions). I claim that a good deal of the value of EA itself is heuristic value rather than strictly epistemic value. Often enough, EA has seized the credit to which HA is entitled. A great many epistemic results are not important or interesting in themselves.25 From the scientists’ point of view, the central significance of these results is to direct future research. What is really at stake is not the truth or falsity of that particular claim in itself but its directive value—which EA alone cannot supply.26 Settling an epistemic claim can focus research in one direction rather than another and can establish the reliability of some research tools over others, but EA itself does not provide this shift of focus. Consider several examples in which trouble has arisen as a result of an overemphasis on narrowly epistemic considerations. Most often mentioned is what Larry and Rachel Laudan (1989) term “the innovation problem.” Standard EA cannot explain why scientists would ever abandon an accomplished theory with a strong record of explanatory and predictive success in favor of an upstart model that so far has little empirical support and that may suffer from conceptual problems as well. Clearly HA can in principle solve this problem, provided that we recognize that HA can outweigh EA in decision contexts of the kind we have already canvassed.27 The Laudans also raise the problem of metamethodology: cases of methodological disagreement raise the question of what justifies choices of methodological goals and standards. Larry Laudan (Laudan 1996) has written extensively on this topic, defending what he terms “normative naturalism.” As a pragmatist, I am sympathetic to this approach; but here, too, HA is essential to an adequate solution. The problem arises only because metamethodology is too closely linked with traditional epistemology.

HEURISTIC APPRAISAL: CONTEXT OF DISCOVERY OR JUSTIFICATION?

173

To repeat, scientists opportunistically adapt their goals and standards to take advantage of the resources available to them, to whichever ones look fruitful to future research. Laudan (Laudan 1981) agrees, there observing that scientists tend to invoke methodological principles as weapons to defend their own programs against others. What is striking here is that heuristic and pragmatic considerations that wielders of the traditional DJ distinction relegated to the context of discovery now triumphally return from banishment to help to undergird epistemic theory of justification. Now consider a different sort of case: so-called crucial experiments. Are they really epistemically crucial? Often they are not, as the vagaries of the history of theories of light attest. Yet they can be heuristically crucial in showing that, at a certain point in the development of the subject (available techniques, apparatus, mathematics, etc.), it is clearly more fruitful to develop this line of approach rather than that one. Early in the 19th century, wave analogies and Fresnel’s mathematics combined to produce rapid and fruitful developments even though the classical wave theory eventually fell on hard times. (However, many of the mathematical techniques introduced remain invaluable to this day.) The crucial experiments in the history of light turned out not really to be epistemic branch points revealing the final truth about the nature of light—but they were indeed heuristic branch points.28 Insofar as scientific progress, like the business economy, is a matter of creative destruction (Schumpeter), it seems more appropriate to speak of crucial experiments in heuristic than epistemic (“final justification”) terms. Another example is the debate over whether novel prediction carries special weight. “Yes,” say John Herschel, William Whewell, the Popperians and Lakatosians,29 and some Bayesians. “No,” say John Stuart Mill, John Maynard Keynes, and the Laudans. I believe the latter are correct as regards epistemic weight. But, obviously, novel prediction carries extra weight of a different kind insofar as it opens up new research areas, as when the novel prediction is a previously unknown phenomenon or a phenomenon previously assigned to a different scientific domain and/or covered by another theory or model. (Strong, cross-theory linkages can be of great value to EA as well as to HA and are one of the things that standard confirmation theory has trouble handling.) The special weight is heuristic, not epistemic. Novel predictions get a higher HA than ordinary predictions, but not a higher EA. Finally, consider the strange case of Fritz Rohrlich’s (Rohrlich 1965) reconstruction of the classical theory of the electron. At first this appears to be a fruitless effort to resolve some conceptual problems and thus to achieve a better EA of a field already long abandoned for other reasons. (Dilemma: an electron is either a point particle or not. If yes, we have infinite charge and mass densities; and if no, then we have an extended body that should blow itself apart by mutual repulsion, which increases exponentially as the distance between charges decreases.) So why did a good physicist take the trouble to rehash old and largely forgotten issues? Because he believed that the exercise would clarify current problems. In short, Rohrlich believed that his reconstruction would be a valuable heuristic for current work. Indeed, there is a strong tradition of utilizing recently failed theories as heuristic guides to their own

174

THOMAS NICKLES

successors, e.g., the many ways in which classical mechanics has been employed in developing and using the new quantum mechanics (Bohr’s correspondence principle, Hamilton’s equations, Ehrenfest’s theorem). Comment 2. Why this persistent tendency to conflate HA with EA? Disentangling HA from EA is particularly difficult for philosophers, given their professional bias (in a non-pejorative sense of ‘bias’), which makes epistemology and a hankering after metaphysics the central concern. Although philosophy of science is parasitic upon the sciences as its subject matter, philosophers and scientists have different professional interests and orientations (Nickles 2002a). This helps explain why HA itself receives a negative HA from philosophers. Much traditional philosophy has been a super-optimizing enterprise, a “to the max” enterprise, seeking absolutely the greatest good, the very best form of government, the ultimate truth, perfect rationality (logical consistency, no Dutch books), and so on. Few philosophers worried about the feasibility, likelihood of success, and cost of these goals until Charles Peirce, in his path-breaking work on economy of research and abduction. It was Peirce who first posed the question of pursuitworthiness in a developed way. Ernst Mach also had plenty to say. However, the advent of Carnap’s brand of logical empiricism, together with the DJ distinction, left little place for such considerations, since there seemed no way to capture them within the confines of the first-order predicate calculus. Pragmatism’s concern for heuristics and economy of research fell by the wayside. Thus, the logical empiricists could still seriously propose a principle of total evidence as a constraint on inductive inference. Also impossible, as it turns out, was their insistence that logical consistency is a sine qua non of successful research. For we possess no routine method for checking the consistency of a large number of independent propositions.30 Given that logic was at the center of their philosophical method, and that, for them, inconsistency was the worst logical sin, we can understand the logical empiricists’ concern. But this worry was largely an artifact of their philosophical methodology—specifically, the paradoxes of the material conditional (that from inconsistent premises one can derive any and all propositions)—rather than a research stopper in science itself. Of course, inconsistency is a serious defect of any scientific theory, but an epistemic defect is not necessarily a heuristic defect. The classical pragmatists, Peirce, James, and Dewey, rejected what they termed “the spectator theory of inquiry” for an account of learning and knowing that directly engages and eventually changes the world by employing experimental intervention (e.g., Dewey 1929). I agree with their assessment that not only the ancient Greeks but also many modern philosophers look only at the “big theory” products of scientific research, at an aesthetic distance. Science replaces traditional metaphysics in furnishing our world picture. Hence, the dominance of theory-centered, representational philosophy of science. Only recently and reluctantly have many philosophers of science come to regard detailed empirical information about scientific practice as just as important as the traditional tools of logical reconstruction—partly as a response to competition from social studies of science (Rouse 2002). Such a stance is perfectly appropriate

HEURISTIC APPRAISAL: CONTEXT OF DISCOVERY OR JUSTIFICATION?

175

for philosophers qua philosophers, who fear that their dreams of a final theory will never materialize in their lifetimes; but it interferes with their understanding of how science itself works. Working scientists also entertain such dreams (Weinberg 1994), but more urgent to their line of work are the pragmatic issues of what to do next and how to do it. Many traditional philosophers look backward, toward retrospective justification, whereas scientists and pragmatists look forward, again giving more attention to HA than to EA.31 Philosophers are too much like Frank Lloyd Wright’s floo floo bird. I am referring to the peculiar and especial bird who always flew backward. To keep the wind out of its eyes? No. Just because it didn’t give a darn where it was going, but just had to see where it had been.32

Scientists are not so much interested in final justification of truth claims as in getting results reliable enough to continue on to the next stage of investigation. Epistemologists ask, “Is it true?” while scientists ask, “Is it new?” Epistemologists ask, “How do you know that the cat is on the mat? or that Jones owns a Ford?” Whereas scientists ask, “Is it is fruitful to search for magnetic monopoles or for string theories or for mechanisms of group selection?” Most philosophers are directly truth-seeking, whereas scientists are problem-seekers and solvers. Scientists thrive on finding new problems, while most philosophers of science have treated problems as mere obstacles rather than as research achievements. 5. CONCLUSION: HEURISTIC APPRAISAL AND THE DJ DISTINCTION

It is time to attempt answers to our opening questions. We have seen that the standard DJ distinction leaves no room for HA. Should we therefore abandon the DJ distinction altogether or should we instead insert HA as a third branch of the distinction? I conclude that the original DJ distinction is indeed untenable (as several other papers in this volume have also demonstrated), nor will it help to interpose a third context either logically or temporally between the context of discovery and the context of justification. However, before taking the radical step of rejecting all DJ distinctions, without replacement, let us consider Hoyningen’s minimal DJ distinction, the distinction of a normative-evaluative perspective from a descriptive perspective, whether object level or metalevel. Hoyningen intends his distinction to be so minimal as to avoid controversy. It would be unfair to expect it to have the philosophical “bite” of traditional invocations of the DJ distinction, since the purpose of his move to minimality is precisely to unload the logical empiricist and Popperian baggage from the DJ distinction. Those who wish to employ a stronger distinction must explicitly defend it. Is there any bite left at all? Does the minimal distinction do any useful work? One valuable purpose of the original DJ distinction was to guard against committing genetic fallacies. It is a genetic fallacy to conclude that the source or conditions of origin of something forever determine its character. That something has dubious beginnings does not necessarily make that thing itself defective. People whom

176

THOMAS NICKLES

we distrust can nevertheless produce valid arguments, and the blind, trial-and-error process of biological evolution—and scientific investigation!—can produce wonderfully complex designs. Although the logical empiricists were not much interested in deep scientific change, they did recognize that the final product of the discovery process could be very different in logic, content, and intellectual motivation from the early probing. Their DJ distinction could therefore function as a kind of rule of detachment. A suitably tested and theoretically anchored scientific product, the logical empiricists insisted, can be detached from the process that produced it. There is surely something right about this position. (Some qualifications may be necessary, but we need not consider them here.) Scientific papers early in the discovery/innovation process already begin this task of weaning results from their conditions of origin—by reporting the research design and results in a logically streamlined way, by omitting mention of all the blind alleys, accidents, and irrelevant contingencies on the way to the final results. There is nothing wrong with this: such rewriting and rethinking is an essential part of the research process itself. It is not mere rhetorical overlay on the “real” scientific work. Nor is it mere weaning; for as the original results are rederived, reinterpreted, and linked with other results, they often gain in theoretical depth as well as epistemic status. This reconstruction process may continue via citations by later authors and discussion in later work, summary papers, annual reviews, handbooks, textbooks, and teaching. Notice that this fruitful sort of reconstruction is not a logical reconstruction in a formal language, accomplished by philosophers, but a technical reconstruction and repositioning performed by scientists themselves. Hoyningen’s liberal distinction no longer makes context of justification (as defined by the DJ distinction) external to empirical science itself. The question then becomes whether Hoyningen’s minimal DJ distinction still serves to block genetic fallacies and to permit this sort of detachment. It seems to me that it can do this, since it is a descriptive/normative distinction. At the detachment stage, it is the normative evaluation that counts. My purpose in this chapter has been only to expand what counts as evaluation and normative detachment to include HA. Thus I accept Hoyningen’s distinction. Rejecting all forms of the DJ distinction, root and branch, is a mistake.33 The next question is whether the minimal DJ distinction excludes or discourages HA. In my view it does not. The minimal DJ distinction is so philosophically neutral, so weak, that it places no restrictions on what counts as evaluation. In particular, it does not restrict evaluation to EA or to evaluation in terms of any specific goals, standards, or methodologies. That being the case, the minimal DJ distinction creates no need to introduce HA as a distinct “stage” of research in either a temporal or a logical sense. However, HA does represent a “perspective” (to employ Hoyningen’s term) distinct from both EA and nonevaluative elements of context of discovery. It is also logically distinct in a broadly modal sense, being concerned with future possibilities rather than with past and present actualities. Perhaps the simplest thing to say is that Hoyningen’s evaluative perspective divides into two kinds of evaluation, EA and HA, although, as we have

HEURISTIC APPRAISAL: CONTEXT OF DISCOVERY OR JUSTIFICATION?

177

seen, HA typically depends on some work that falls within context of discovery. But so, after all, does EA, which needs data statements (which are descriptive) even to get started. Whether or not we regard HA as a “mixed” perspective distinct from Hoyningen’s two, it does not threaten his distinction, which is not exhaustive. In my opinion, failure to attend to HA in both its heuristic and pragmatic dimensions has stunted the field of philosophy of science. Faithfully adhering to the DJ distinction, the internal–external distinction, a rather na¨ıve and romanticized hypotheticalist model of theoretical science, a non-naturalistic conception of philosophy, and putting truth-seeking epistemological questions first, philosophers of science have conceded large territories to computer scientists and artificial intelligencers on the one hand and to science studies practitioners on the other. The former investigate the solvability of various kinds of problems by various methods, including heuristic methods, and attend to the relative efficiency of alternative methods, a topic belonging to economy of research. The latter investigate scientific practices of all kinds, without focusing solely on the giants of physics, and they take more seriously the forward looking perspective of scientists themselves (e.g., Latour 1987). My point here is not at all that philosophy, properly done, would have rendered these new fields unnecessary. Rather, it is to regret that we have been so slow to join in these investigations, so slow that philosophers are now often ignored as “old fashioned.” We may hope that the new construal of the DJ distinction will help us to reclaim some of the ground that Peirce, Mach, Duhem, Poincar´e, and others once plowed.

ACKNOWLEDGMENTS

Thanks to Jutta Schickore, Friedrich Steinle, and the entire group for encouraging me to focus on heuristic appraisal and the future of the DJ distinction. I here develop previous work supported by the U.S. National Science Foundation. Thanks are due also to the Max Planck Institute for the History of Science in Berlin for sponsoring this project. In Reno I have had many productive conversations with Yoichi Ishida.

NOTES 1. In somewhat the same vein, Popper reportedly would enter his lecture hall on the first day and announce to the students, “I am a professor of scientific method, and I have a problem. There is no such thing as scientific method!” 2. For one thing, in standard EA most of the attention goes to hypothesis testing against the data and very little to the later refinement and theoretical imbedding of results in a rich fabric of mutually supportive claims and practices. We tend to get low-level hypothesis testing at one extreme and radical Quinean holism at the other. For another, standard DJ applications give a misleading account of experimental research as the cut-and-dried method of testing sharp, mathematically formulated hypotheses. This despite the fact that logic of justification for Reichenbach was the final philosophical product of the process of justification: not the process of scientific testing itself but the logical reconstruction of the relations between theory and evidence. Against the view of experimental work as routine, Hans-J¨org Rheinberger (Rheinberger 1997, p. 27) quotes Ludwik Fleck:

178

THOMAS NICKLES [I]f a research experiment were well defined, it would be altogether unnecessary to perform it. For the experimental arrangements to be well defined, the outcome must be known in advance; otherwise the procedure cannot be limited and purposeful.

3. 4.

5. 6.

7.

8.

9.

10. 11. 12.

13.

Adds Rheinberger, “Experimental systems, together with the scientific objects wrapped up in them, are inherently open, if bottlenecked, arrangements. Their movement is such that it cannot be predicted if they are to retain their character as research devices” (p. 74). The disadvantage of this is that it perpetrates a theory-centered approach to philosophy of science, one of the things that attention to HA itself can help to overcome. The old labels, ‘context of discovery’ and ‘context of justification’, become somewhat misleading. In addition, I employ the term ‘discovery’ in a broad way that includes innovation of all kinds without commitment to strong epistemological or ontological realism. See Nickles 1989 for a beginning. The Cosmotron was, of course, built. Physicists convinced the Atomic Energy Commission that they would find good uses for it. “Build it and we will come.” If I remember the story that Sam Goudsmit used to tell, Enrico Fermi testified that a significant discovery costs about $10,000 (in 1948 dollars). For a serious history of Brookhaven, see Crease 1999. There are many dimensions to HA. An experimentalist may engage in free exploration without any particular hypotheses or problems in mind. There are even different dimensions of problem-solving capacity. One sort of positive HA indicates how an approach may solve or circumvent a major difficulty that has hitherto blocked progress. Another sort judges that a research program is capable in principle of solving all of the problems that can arise in a given domain. Adherents of a Kuhnian paradigm make the latter judgment, that the set of exemplars (successful problem solutions already available as models) span the space of the domain in the sense that all problems can be construed as variants of one or more problems already solved. In Einstein’s reply to Lenzen and Northrop, in Schilpp 1949, p. 684, quoted by Pickering 1984a, p. 3. Of course, the logical empiricists had their own opportunistic interests, as I tried to bring out in Nickles 2002a. See also section 4 below. In my view, all innovation involves at least small elements of luck or serendipity, a view that makes HA even more obviously indispensable—and more difficult. See section 4 below and Nickles 2003; Nickles in press. For examples from high-energy physics, see Pickering 1980; Pickering 1984a; Pickering 1984b. In 1905 Einstein received his PhD, published Einstein 1905a, Einstein 1905b, Einstein 1905c and two other papers as well. A fairly good year! For example, consult Suppes 1962; Shapere 1969; Cartwright 1983; Wimsatt 1987; Herfel 1995; Meheus (ed.) 2002 and the references therein. For my fuller discussion of such issues, see Nickles 2002b. An important recent contribution is Frisch (2005). In fact, both men took up brain research. Here is Gunther Stent, from an unpublished but recorded interview by Istvan Hargittai (Budapest, January 13–15, 2003): In the fall of 1949—during my second and last year at Caltech and three years before Watson and Crick’s discovery of the DNA double helix—Max [Delbr¨uck] announced to his disciples that the search for the mechanism of biological self-reproduction was now in “good hands.” What he meant by this locution was that the quest for its solution would soon be over and that our present line of work was about to turn boring. He revealed to us that the future of vanguard biology now lay in understanding the brain, which he considered to be the last frontier. So to prepare us for that future, he assigned to each of us a set of neurobiological papers that we were to present to our colleagues.

Stent himself eventually went into the developmental neurobiology of leeches. My thanks to Professor Stent for a copy of the interview. 14. William James made something like this distinction in “The Will to Believe,” section section vii and following, but in a way that jumbled together epistemic and heuristic considerations, as critical epistemologists were quick to point out.

HEURISTIC APPRAISAL: CONTEXT OF DISCOVERY OR JUSTIFICATION?

179

15. There is much socio-historical support for this point. Popper’s view smacks of the old history of ideas that views history as the unfolding of logical relations, whereas social historians emphasize that cultural resources, including scientific hypotheses, do not just automatically endure. They must be actively reproduced. Cultural evolutionists will add that this reproductive process almost inevitably introduces variant understandings and practices. Personal style and astute career-building moves can help, too. Writes Gunther Stent (Stent 1998, p. 241f), The more banal cause [than theft] for the failure to get due credit for one’s discovery is, as in my case, the personal lack of the qualities needed to have it make an impact. Originality and inventiveness, though necessary, are not sufficient for making a mark in science: One must also have the intuition, stamina, and, above all, the self-confidence necessary to exploit one’s inventions and present them as a salable package.

16. By contrast with Popper, Kuhn (Kuhn 1962, 1970) contended that we learn from our problemsolving successes (exemplars) rather than from what Popper considered our “mistakes.” Popper’s position was a little more complicated than indicated in the text. Popper (Popper 1959, p. 53f) stated that a hypothesis that has “proved its mettle” cannot simply be dropped rather than explicitly refuted or replaced by something better. But that still leaves us with the problem that what is in fact the true hypothesis may seem too bizarre to bother testing, if it gets formulated at all. The history of science shows how often the truth lies beyond the horizon of plausibility and even the horizon of intelligibility of anyone living at a particular time. 17. Gigerenzer et al. 1999 show that there are other kinds of cognitive economy than Herbert Simon’s satisfying. 18. (Popper 1959, p. 55). For example, “My only reason for proposing my criterion of demarcation is that it is fruitful: that a great many points can be clarified and explained with its help.” The logical empiricists and Popperians often helped themselves to heuristic considerations while denying them an official place in their methodologies as items for analysis. Imre Lakatos and his successors tried to remedy this defect in Popperian methodology while remaining within a broadly Popperian epistemological framework. 19. For the generativist/consequentialist distinction, see Laudan 1981, chapter 10 and Nickles 1987a. My major complaint against standard EA is that it is too consequentialist. 20. As Kuhn 1962, sections IX and X, put it, experimental observations plus logic do not dictate major theory or paradigm choices. Laudan 1996, chapter 2 argues that the more dramatic consequences that Quine and others have drawn from the underdetermination point are based on fallacious reasoning. 21. This point is different from point 6, in which HA helps determine the very choice of EA method. 22. See Walter 1993 and Cerf and Navasky 1998. 23. For an insightful early attempt to incorporate HA in a Bayesian framework, see Salmon 1967, chapter VII. 24. There are varieties of dependence. HA obviously depends on a background of well-established claims (those enjoying positive EA) to keep it honest concerning what seems possible. It also cues on the EA of specific claims. Both a positive and a negative EA in such a case can generate new problems or difficulties for HA to evaluate. (Interesting new problems are research accomplishments, by contrast with difficulties in a current model, for instance.) HA can key both on the scientific content of the claims in question and on their epistemic status. 25. This is why EA alone provides a poor control theory: it puts the “locus of control” in the wrong place. An extreme case is Popper’s stereotype of Baconian inductivism, which assigns equal epistemic value to all empirical facts. 26. A similar point can be made wholly within EA. An empirical fact may not be important or interesting in itself, but it may help to confirm or refute a lawlike claim or theoretical model. 27. The Laudans state their solution in terms of “epistemic dominance,” with the plate tectonic revolution as the central example. Sometimes the evidence comes to favor one theory over another for both parties in a dispute, even though their evidential standards differ. This epistemic solution works only in special cases, I maintain. Elsewhere, the Laudans recognize the importance of HA.

180

THOMAS NICKLES

28. Or consider the Thomson family’s engagement with electrons. J. J. Thomson received the Nobel Prize for his work with cathode-ray tubes in the 1890s demonstrating that electrons are particles; whereas his son, G. P. Thomson, earned a Nobel Prize for his work on electron diffraction in the 1920s demonstrating that electrons are not particles. And both men were right! 29. For the latter, the point is not that novel prediction confers greater probability on a hypothesis but that novel prediction is the only evidence that counts at all in testing. 30. Cherniak 1986 shows that a truth table check (or any other known check) of the mutual consistency of only 137 arbitrarily chosen, independent statements is physically impossible. So we have the irony that the very attempt to make methodologies rational, by having them require logical consistency, is itself irrational, since failsafe consistency checks are impossible. In this respect, the American pragmatists took the efficiency movement more seriously. 31. Scientists are, of course, also concerned with EA. The greatest scientific achievement is to have a law, constant, or effect named after one. Both EA and HA considerations come into play in choosing sample size or how much power one’s experiment will have. 32. From a 1938 speech on the backwardness of architecture in the U.S. and elsewhere. Quoted by William Safire, New York Times, 5 April 2004. 33. Some of the so-called friends of discovery made this mistake, but I departed from them at this point. See, e.g., Nickles 1985.

REFERENCES Brush, Stephen (1968), “A History of Random Processes I: Brownian Movement from Brown to Perrin,” Archive for History of Exact Sciences 5: 1–36. Cartwright, Nancy (1983), How the Laws of Physics Lie (Oxford: Oxford University Press). Cerf, Christopher, and Victor Navasky (eds.) (1998), The Experts Speak (New York: Villard). Cherniak, Christopher (1986), Minimal Rationality (Cambridge, MA: MIT Press). Cole, Stephen, Leonard Cole, Leonard Rubin, and Jonathan R. Cole (1978), Peer Review in the National Science Foundation (Washington: The Academy). Crease, Robert (1999), Making Physics: A Biography of Brookhaven National Laboratory, 1946–1972 (Chicago: University of Chicago Press). Dewey, John (1929), The Quest for Certainty, reprinted in the series “John Dewey: The Later Works,” Vol. 4, J. Boydston (ed.) (Carbondale, IL: Southern Illinois University Press, 1988). Duhem, Pierre (1954), The Aim and Structure of Physical Theory, translated from the 1914 French edition (Princeton: Princeton University Press). Einstein, Albert (1905a), “On a Heuristic Point of View about the Creation and Conversion of Light,” reprinted in translation in D. ter Haar (ed.), The Old Quantum Theory (Oxford: Pergamon Press, 1967), pp. 91–107. Einstein, Albert (1905b), “On the Movement of Small Particles Suspended in a Stationary Liquid Demanded by the Molecular-Kinetic Theory of Heat,” reprinted in translation in R. F¨urth (ed.), Albert Einstein: Investigations on the Theory of the Brownian Movement (New York: Dover, 1956), pp. 1–18. Einstein, Albert (1905c), “On the Electrodynamics of Moving Bodies, ” reprinted in translation in The Principle of Relativity (New York: Dover, n.d.), pp. 37–65. Freeman, Chris, and Luc Soete (1997), The Economics of Industrial Innovation, 3rd ed. (Cambridge, MA: MIT Press). Frisch, Mathias (2005), Inconsistency, Asymmetry, and Non-Locality: A Philosophical Investigation of Classical Electrodynamics (Oxford: Oxford University Press). Gigerenzer, Gerd, et al (1999), Simple Heuristics that Make Us Smart (Oxford: Oxford University Press). Herfel, W., W. Krajewski, I. Niiniluoto, and R. W´ojcicki (eds.) (1995), Theories and Models in Scientific Processes (Amsterdam: Rodopi), pp. 137–149.

HEURISTIC APPRAISAL: CONTEXT OF DISCOVERY OR JUSTIFICATION?

181

Hoyningen-Huene, Paul (this volume), “Context of Discovery versus Context of Justification and Thomas Kuhn.” Hull, David (1988), Science as a Process (Chicago: University of Chicago Press). James, William (1907), Pragmatism (New York: Longmans Green). Kuhn, Thomas (1962), The Structure of Scientific Revolutions, 2nd ed., enlarged, 1970 (Chicago: University of Chicago Press). Latour, Bruno (1987), Science in Action: How to Follow Scientists and Engineers Through Society (Cambridge: Harvard University Press). Laudan, Larry (1981), Science and Hypothesis (Dordrecht: Reidel). Laudan, Larry (1996), Beyond Positivism and Relativism (Boulder, CO: Westview Press). Meheus, Joke (ed.) (2002), Inconsistency in Science (Dordrecht: Kluwer). Nickles, Thomas (1980), “Can Scientific Constraints Be Violated Rationally?,” in T. Nickles (ed.), Scientific Discovery, Logic, and Rationality (Dordrecht: Reidel), pp. 285–315. Nickles, Thomas (1985), “Beyond Divorce: Current Status of the Discovery Debate,” Philosophy of Science 52: 177–206. Nickles, Thomas (1987a), “From Natural Philosophy to Metaphilosophy of Science,” in R. Kargon and P. Achinstein (eds.), Kelvin’s BALTIMORE LECTURES and Modern Theoretical Physics: Historical and Philosophical Perspectives (Cambridge: MIT Press, 1987), pp. 507–541. Nickles, Thomas, (1987b), “Lakatosian Heuristics and Epistemic Support,” British Journal for the Philosophy of Science 38: 181–205. Nickles, Thomas (1989), “Heuristic Appraisal: A Proposal,” Social Epistemology 3: 175–188. Nickles, Thomas (2002a), “The Discovery-Justification (D-J) Distinction and Professional Philosophy of Science: Comments on the First Day’s Five Papers,” in J. Schickore and F. Steinle (eds.), Revisiting Discovery and Justification. (Max-Planck-Institut f¨ur Wissenschaftsgeschichte), Preprint 211, pp. 67–78. Nickles, Thomas (2002b), “From Copernicus to Ptolemy: Inconsistency and Method,” in Meheus (2002), pp. 1–33. Nickles, Thomas (2003), “Evolutionary Models of Innovation and the Meno Problem,” in L. Shavinina (ed.), International Handbook on Innovation (Amsterdam: Elsevier Scientific Publications), pp. 54–78. Nickles, Thomas (in press), “The Strange Story of Scientific Method,” in J. Meheus and T. Nickles (eds.), Models of Discovery and Creativity. Dordrecht: Springer, 2005. Pickering, Andy (1980), “The Role of Interests in High-Energy Physics: The Choice between Charm and Colour,” in K. Knorr, R. Krohn, and R. Whitley (eds.), The Social Process of Scientific Investigation (Dordrecht: Reidel), pp. 107–138. Pickering, Andy (1984a), “Against Putting the Phenomena First: The Discovery of the Weak Neutral Current,” Studies in History and Philosophy of Science 15: 85–117. Pickering, Andy (1984b), Constructing Quarks: A Sociological History of Particle Physics (Chicago: University of Chicago Press). Popper, Karl (1959), The Logic of Scientific Discovery (Expanded translation of Logik der Forschung, 1934) (New York: Basic Books). Quine, W. V. O. (1951), “Two Dogmas of Empiricism,” reprinted in From a Logical Point of View (Cambridge, MA: Harvard University Press, 1953), pp. 20–46. Reichenbach, Hans (1938), Experience and Prediction (Chicago: University of Chicago Press). Rheinberger, Hans-J¨org (1997), Toward a History of Epistemic Things: Synthesizing Proteins in the Test Tube (Stanford: Stanford University Press). Rohrlich, Fritz (1965), Classical Charged Particles: Foundations of their Theory (Reading, MA: Addison-Wesley). Rouse, Joseph (2002), How Scientific Practices Matter: Reclaiming Philosophical Naturalism (Chicago: University of Chicago Press). Rouse, Joseph (2003), “Kuhn’s Philosophy of Scientific Practice,” in T. Nickles (ed.), Thomas Kuhn (Cambridge: Cambridge University Press), pp. 101–121.

182

THOMAS NICKLES

Rudner, Richard (1954), “Remarks on Value Judgments in scientific Validation.” Scientific Monthly 79: 151–153. Salmon, Wesley (1966), The Foundations of Scientific Inference (Pittsburgh: University of Pittsburgh Press). Schilpp, P. A. (ed.) (1949), Albert Einstein: Philosopher-Scientist (Evanston, IL: Library of Living Philosophers). Shapere, Dudley (1969), “Notes Toward a Post-Positivistic Interpretation of Science,” in P. Achinstein and S. Barker (eds.), The Legacy of Logical Positivism (Baltimore: Johns Hopkins University Press), pp. 115–160. Stent, Gunther (1998), Nazis, Women and Molecular Biology: Memoirs of a Lucky Self-Hater (Kensington, CA: Briones Books). Streater, R. F., and A. S. Wightman (1964), PCT, Spin and Statistics, and All That (New York: Benjamin). Suppes, P. (1962), “Models of Data,” In E. Nagel, P. Suppes, and A. Tarski (eds.), Logic, Methodology and the Philosophy of Science (Stanford: Stanford University Press), pp. 252–261. Taylor, J. H. (ed.) (1965), Selected Papers on Molecular Genetics (New York: Academic Press). Weinberg, Steven (1994), Dreams of a Final Theory (New York: Random House). Walter, Dave (1992), Today Then: America’s Best Minds Look 100 Years into the Future on the Occasion of the 1893 Columbian Exposition (Helena, MT: Farcountry Press). Wimsatt, William (1987), “False Models as Means to Truer Theories,” M. Nitecki and A. Hoffman (eds.), Neural Models in Biology (Oxford: Oxford University Press), pp. 23–55. Wimsatt, William (forthcoming), (Piecewise) Approximations to Reality (tentative title), (Cambridge, MA: Harvard University Press).

FRIEDRICH STEINLE

CONCEPT FORMATION AND THE LIMITS OF JUSTIFICATION: “DISCOVERING” THE TWO ELECTRICITIES

In this essay, I examine the possible use of the distinction between discovery and justification for the analysis of research practice, and what we can learn, in turn, from this analysis for an assessment of the uses and limits of this distinction. First, I illustrate that nonstandard uses of experiments and processes of concept formation reveal the “process-interpretation” of the DJ distinction as inappropriate. Turning to what is often regarded as the core of the DJ distinction—the differentiation between genesis and validity—I shall focus on the role it plays within science. Moreover, by taking processes of concept formation seriously, new and hitherto unrecognized limits of justification become visible. In particular, justification turns out to be more genuinely bound to history than is usually assumed. To illustrate and flesh out my general claims, I shall first provide an analysis of a specific historical episode: the purported “discovery” of the two electricities in the 1730s. CHARLES DUFAY AND THE TWO ELECTRICITIES

Electricity comes twofold: plus or minus, positive or negative. This fact provides the basis for our handling of electricity, be it in the laboratory when using a voltmeter, or in everyday life when renewing the battery of a camera, or helping the neighbor starting his car on a frosty morning. We learn about the bipolarity in our first encounters with electricity in school, and take it as a basic feature of nature. This has not always been so. Still in the early 18th century, there was no other idea than to speak of electricity as a uniform thing. It was defined as the virtue of some materials to attract small bodies in their neighborhood after being rubbed. Even the spectacular news of Francis Hauksbee’s new glassrod (by which electric effects could be more stably produced than before), and Stephen Gray’s report that the attractive virtue could be transferred to other bodies, did in no way affect this basic understanding. But only a few years later, as many histories of electricity tell us, a Paris academician named Charles Cisternay Du Fay “discovered” that electricity in fact was twofold and thus changed the foundations of a whole research area. In what follows, I shall briefly sketch and discuss the pathway by which such an extraordinary innovation was achieved. Charles Dufay, an early 18th-century, brilliant academician, acted as director of the Paris Jardin du Roy, a botanic garden central both for European taxonomical research and for Paris education of medical doctors. Besides running the garden, Dufay did research in many different fields, from astronomical calculations to studies 183 J. Schickore and F. Steinle (eds.), Revisiting Discovery and Justification, 183–195.
C 2006 Springer. Printed in the Netherlands.

184

FRIEDRICH STEINLE

of dew, from taxonomical classification to experiments on luminescence. Some time in 1733, he switched to electricity, which was soon to become his largest single research subject.1 As a research field, electricity (and this means always what we now call static electricity) was at a difficult stage during that period. More than hundred years of research in all European countries had produced a multitude of different and puzzling effects. I want to mention just a few of the problems that made the field appear unstable and confused: some materials could be electrified by rubbing, others sometimes, again others not at all. Sometimes bodies could be made electric by contact with others, sometimes they could not. Sometimes electricity acted as attraction, sometimes as repulsion, and sometimes sudden changes of attractive and repulsive effects occurred. Dufay did most extensive experiments, going through all sorts of variations. He used a vast number of different materials, single or in combination, and also varied the shape, temperature, color, moisture, air pressure, and the experimental setting: two bodies in touch, in close neighborhood, in large distance, being connected by a third, and so on. His work led to remarkable results, such as the bold claims that all materials except metals could be electrified by rubbing, and that all bodies except for flames could receive electricity by communication. But still he was left with serious problems with respect to attraction and repulsion and their sometimes sudden change. I shall take a more detailed look at how he tackled this problem by a most systematic experimental enterprise.2 In a first step he intended to clarify whether or not repulsion existed. By varying experimental conditions, he analyzed the conditions under which repulsion occurred. When he succeeded in many cases, he asked himself whether the objection that the observed repulsion was only an effect of overlapping attractions could be sustained. The answer was clearly negative: repulsion occurred even when the arrangement of other bodies in the vicinity was changed most drastically. Thus the repulsive effect was clearly shown to exist. In a particularly delicate arrangement, Dufay was able to keep a very light leaf of gold hovering at a distance of more than one foot over an electrified glass tube for several minutes, even when he moved through the room with the tube. When the force of the tube weakened, the leaf would lower and finally fall down to the floor. Highly characteristic of the spirit of his search for laws, Dufay now turned to the question of when exactly repulsion occurred in contrast to the usual attraction—after all, attraction had become more or less the defining feature of the electric virtue! Again, he approached the question experimentally, by varying many parameters of the arrangement: the manner of electrification (rubbing, communication), the degree of electrification, the size of the electrified body, the material of the electrified body, the basis/support of the electrified body (more or less idio-electric material), and the constellation between the bodies involved (larger or smaller distance). For a long time, the results were merely puzzling, appeared unstable, and did not combine to anything resembling a regularity or correlation. Dufay could not formulate the conditions of when attraction and repulsion occurred, let alone when, as sometimes observed, a sudden switch between the two actions took place. This latter effect, however, was

CONCEPT FORMATION AND THE LIMITS OF JUSTIFICATION

185

ultimately the first to be conceptualized in a rule. For specific constellations Dufay was able to formulate regularities: r Electrified bodies always attracted small unelectrified bodies. r However, when these small bodies approached the electric one and became themselves electrified by communication, the former attraction turned into repulsion. Dufay found this regularity of attraction–contact–repulsion valid without exception, and it could account for many of the effects he had obtained. Nevertheless, it was definitely restricted to the interaction of every two bodies when one had been electrified by communication from the other. In other cases (when the two bodies had been electrified independently of each other, e.g.), the regularity did not hold; they still appeared irregular. Dufay did not stop. He found a crucial hint in an experiment with the hovering gold leaf. When he approached this leaf with a third electrified body, strange effects resulted. If this third body was made of glass, it repelled the leaf strongly, as the tube did. But when it was of copal, it strongly attracted the leaf! Although this result was most puzzling (“. . . me d´econcerta prodigieusement,” Dufay 1733, p. 464) and gave a strong hint that the material used was crucial here. In pursuing that trace very persistently, Dufay increasingly became aware of regular behavior that could not, however, be formulated in the usual language of electricity. To express this behavior, Dufay finally made a radical proposal: rather than electricity in general, one should more accurately speak of two electricities. The regularity then was that electrified bodies repelled all those that had the same electricity, but attracted all those that had the other. As the experiments showed, the electricities maintained their character even when communicated to other bodies. Thus the above regularity of attraction–contact–repulsion turned out to be just a special case of the now general regularity. According to Dufay, which of the two electricities a body obtained by rubbing depended only on its material. Thus the dichotomy of electricities induced a division of all materials into two classes. At the same time, the electricities could be characterized according to these classes—Dufay named them vitreous and resinous electricities, respectively, according to prominent representatives of the two classes of materials. We have no explicit notes about what exactly convinced him of the power of these new concepts. Presumably their immediate “success” was central here: Dufay emphasized that, with these concepts and the regularities they allowed him to formulate, he could understand not only his own, quite numerous experiments, but also those reported by others. The concepts enabled him to do exactly what previously had been impossible: to order the whole field in such a way that stable regularities could be formulated. The proposal of two electricities was radical news indeed, on many levels. Not only could Dufay ask whether the new classification of materials might point to new fundamental properties of matter; much more immediately, it became clear that the field of electricity appeared totally transformed, at least for Dufay. In handling electrical effects, in designing, conducting, and evaluating experiments, new considerations became prominent. New questions arose, questions that previously had simply been inconceivable because the very categories they used did not exist.

186

FRIEDRICH STEINLE

It is instructive to look at the specific level of knowledge aspired to (and eventually obtained) in Dufay’s research. He knew well the various theories that attempted to explain electric effects by air currents, by humorous components of matter, by specific electric effluvia, etc. However, these theories played no role in his research. His aim was definitively not to consider the hidden causes of electric effects, but to order them on the phenomenological level, to establish regularities, and eventually to find the appropriate concepts that would allow him to do so. In his view, this constituted an essential precondition for the search for the hidden causes. It would mean “attempting the impossible,” he emphasized, if one were to search for causes without previously having discovered the large number of phenomena, and having ordered them along a “few simple principles” (Dufay 1733, p. 476). From what Dufay actually did, I interpret this as the task of establishing regularities. And in that context, there was no space for “theories” about hidden causes. The search for correlations and regular dependencies dominated his enterprise. Sensitized by his former work, moreover, particularly in botany and luminescence, he was specifically (and more than other researchers of electricity) aware of how essentially such work depended on working with appropriate concepts and categories. He was ready to question and revise existing categories, to shift or even replace them with others that he found more appropriate. And it was here, in the realm of categories and classifications, that his fundamental innovations were located. Before Dufay, there was only an electric virtue, defined by the attractive action onto small bodies nearby. With Dufay’s innovation, the field looked much different. There were two species of electricity now, defined not only by their action on small bodies nearby, but essentially by a mutual interaction which could be either attraction or repulsion. Moreover, that dichotomy induced a classification of all materials according to their susceptibility to those two electricities when subjected to friction. RESEARCH PRACTICE AND THE DJ DISTINCTION

On the basis of the specific historical case, I shall now turn to a more general discussion of the virtues and vices of the DJ distinction. First, I shall examine its use for an analysis of the practice of scientific research. Dufay’s experimental research cannot be grasped in any reasonable way by the standard view on experiment: there was no theory to be tested or to be refined, and even the conceptual framework that guided the experimental activity turned out to be instable. By contrast, the experimental series described above is a striking case of a type of experimenting that I have called “exploratory” elsewhere.3 Far from being a mindless playing around with the apparatus, exploratory experimentation can be characterized by definite guidelines and epistemic goals. The most prominent characteristic of the experimental procedure is the systematic variation of experimental parameters. The first aim is to find out which of them do affect the effect in question, and which of them are essentially required. The central goal is to formulate empirical regularities or laws about those dependencies and correlations. In many cases, moreover, the existing concepts and categories come out to be inappropriate for that purpose. Thus the revision of existing

CONCEPT FORMATION AND THE LIMITS OF JUSTIFICATION

187

concepts and categories becomes a topic, and the formation of new ones that allow a stable and general formulation of the experimental results. It is here, in the realm of concept formation, that exploratory experimentation has its most unique power and importance. Acting and conceptualizing stabilize or destabilize each other on every step. Exploratory experimentation is far more common in scientific research than is usually realized. This perspective opens up, however, only when we take a serious look to research practice, as opposed to later tales about it. It could sound tempting, and this proposal has frequently been brought up to me, to connect the contrast between exploratory and theory-driven experimentation to the distinction between discovery and justification. Exploratory experimentation would belong to the context of discovery, theory-oriented experimentation to justification. If that holds, there would have been a systematic reason why exploratory experimentation was not treated in philosophy of science. But a closer look shows quickly that such an approach does not fit. Exploratory experimentation is concerned with developing regularities and appropriate concepts. If it is successful, this success consists in formulating ever more general laws. One may well ask whether such laws have then been discovered or justified: after all, in common language we often speak of laws as having been “discovered,” by Galileo, Boyle, Hooke, or Mariotte, for example. As soon as we try to clarify our concepts, however, such talk immediately becomes inappropriate: at the moment when laws are formulated in the research process, they are discovered and justified at the same time. Even if a researcher had initially just a speculation of a possible empirical law, she would conceive this law as being “discovered” only in the moment when it was fully supported, i.e., justified. The distinction does not work here, as also Theodore Arabatzis (this volume) shows very clearly in other cases. Moreover, taking into account the second crucial element of exploratory experiment—the possible formation of new concepts—the point becomes even more fundamental. It is a truism that concepts cannot be “tested,” be found true or false, but they can only prove themselves, be demonstrated to “do good work,” whatever that may exactly mean in specific cases. If one wants to transfer a DJ-like distinction to concepts at all (which one had better not, since it will easily create confusion), one had again to say that they are formed and corrobated at the very same time. Dufay’s case is a good example: what did he “discover”? He became aware that the new concept of two electricities served well to order the whole field by general laws, an insight that he could gain only with all empirical evidence present in his mind. The concept was formed only with a huge background of results in mind that served for corrobation. Formation and corrobation went along in the very same process. DJ-like distinctions do not make much sense here. Observations of such a type—and one could easily add many more cases—show clearly that the DJ distinction provides no help whatsoever for understanding research practice in experimental science. Neither is it appropriate to give a chronological account, nor does it grasp different types or aspects of research activity.4 As a means to grasp the process of science, the DJ distinction is useless and may even hinder an appropriate understanding.5

188

FRIEDRICH STEINLE

GENESIS AND VALIDITY IN SCIENCE

Despite the negative result as to a specific (though not uncommon) interpretation by philosophers,6 we cannot easily dismiss the DJ distinction as a whole. Too obviously it plays a major role within science itself, up to this day. And if philosophy of science aims at dealing with real science as contrasted to some idealized science, there is reason enough not to drop the DJ distinction too easily, but rather to turn to science and to have a look how the DJ distinction is used there specifically. Even a quick glance then makes clear that it is not the “process version” of the distinction that is present in science: scientists do not sort their activities in terms of discovering and justifying. But scientific activity does not only consist in doing research in the laboratory or the field. Another and essential activity consists in communicating the results to the community, to a wider public, and to science students. And it is here that some kind of DJ distinction is explicitly brought up . . . “I do not care,” one of my university physics teachers used to say, “about how Newton historically found his law of gravitation; I’m just interested in why it is valid.” He formulated an attitude that is quite typical for scientists (and many philosophers): he pointed to two distinct types of questions: how did a certain insight (a theory, law, fact, . . . ) come about? And why should we believe it, what are the reasons for support? It is the age-old difference between genesis and validity that is addressed here or, in other words, the “lean” version described by Hoyningen-Huene in this volume. Regardless of whether this was exactly Reichenbach’s original intention or not, I think this is the main point why the DJ distinction still plays such an important role. “Discovery” stands emblematically (though with a badly misleading term, as Arabatzis sharply points out in this volume) for the process of successively forming theories, laws, and facts, “justification” stands for the activity of arguing for the validity of these results. Indeed, this is what we find in all activities concerned with presenting scientific results. Science education is illustrative here. A student of, say, microbiology, solidstate physics, cosmology, or plate tectonics, learns a systematic body of knowledge, supported by mathematical reasoning, observations, and experimental results. More often than not the empirical evidence provided for a theory is different from what had been important in its historical development. The system is presented as standing in itself, bearing its whole evidence by present-day arguments, completely dehistoricized and decontextualized. It is not even visible why and how the generation process should or could bear on the understanding. In textbooks we have perhaps the clearest manifestation of the idea that scientific knowledge can be presented and established— or, in other words, be justified—without any regard to its generation and historical development. Here, we have a clear distinction between genesis and validity, and it is here that we learn it in early stages of our encounters with science. Textbook accounts, as particularly crystallized and solidified presentations of scientific results, contrast most sharply to the actual research process in the laboratory, in the field, or at the desk. Major parts of scientific activity, however, take place right in between the two extremes, such as all activities of communicating scientific results

CONCEPT FORMATION AND THE LIMITS OF JUSTIFICATION

189

to ever broader communities. Once it comes to conveying the results of an individual researcher or a small working group to a broader community, these results change their character. In preprints, papers, books, etc., the historical pathway, the provisional character, the limitations and uncertainties, and all local settings are taken out of the picture step by step. The presentation and the arguments successively and deliberately distance themselves from the original setting. It is in processes of decontextualizing scientific results, of presenting, discussing, solidifying, and finally teaching them, that a distinction between generation and justification is successively and actively introduced. Significantly enough, Reichenbach himself pointed to this feature when he explicitly connected justification to the activity of scientists’ presenting their results to the community.7 To use Fran¸cois Jacob’s and Ludwik Fleck’s metaphors, it is the very transition from night science to day science,8 which is crucial here, with all the intermediate stages of twilight. And it is perhaps only in textbook accounts and in accounts for a broader public that there appears something that could be called pure justification. The closer we come to research practice, the less the two are separated and separable. Thus, the separation between questions of formation and justification is by far not as clear-cut and omnipresent as it appears in the public picture. Scientists at the “research-front” are often much more aware of the historical nature of their enterprise than the common picture of science suggests. It is within the process of communicating, presenting, and teaching scientific results that a clear-cut distinction between formation and validity is actively introduced. What are the reasons for doing so? There might be the pragmatic reason of efficiency of science education: if a science student had to start with all history of the field in question, it would probably take her too long to reach a stage to be able to do original, present-day–level research.9 More principally, our intuitions about the character of knowledge come in here: characterizing knowledge as justified belief attributes justification an absolutely central role. The above-mentioned activities of scientists are driven by the idea to present the knowledge as firm as possible, i.e., to collect systematic arguments of justification, regardless of whether they played a role in knowledge generation or not. Thus knowledge, at least in this understanding, tends by its very nature to be decontextualized, i.e., stripped from the specific time, place, and process by which it has been generated. Whether we do well in conceiving knowledge in such a way is a matter of discussion10 and cannot be treated here, but it is beyond question that such an ideal of knowledge indeed plays a major guiding role within scientific research.

CONCEPT FORMATION AND THE LIMITS OF JUSTIFICATION

A disquieting question, however, still remains open. What exactly is the significance of this separation of justification from genesis, and how deep does it go? Is it just that we always can pose two different types of questions toward scientific claims (as the “lean” DJ distinction tells us), and that scientists for pragmatic reasons prefer the question of validity in some of their activities? Or do we have a more fundamental assumption in

190

FRIEDRICH STEINLE

the background here: the assumption that genesis and justification can principally be separated in such a way that validity can be established in total independence of the genesis (except that there is a [trivial] temporal sequence: in arguing for validity of a result, the result itself has to be there, i.e., first to be generated)? I take it that it is the latter, additional, and mostly implicit assumption that provides the hard rock on which much of the self-understanding of science is built. Moreover, it is this assumption that is often taken as marking the core difference between history and philosophy of science: history deals with genesis, philosophy with validity. And as science can (supposedly) argue the validity of its results without regard to its history, philosophy of science can (supposedly) be entertained without regard to history of science. By contrast, I argue that the separability of justification and genesis does not extend as far as this assumption suggests and that there are firm but mostly invisible limits to justification. There is a fundamental, if largely overlooked, epistemic feature that points to an inherent historicity of justification and its results. The core issue is concepts and language. I take “concepts” here in a quite general sense, as “elements of thought,” like words are elements of language.11 As such, concepts provide the foundation not only of forming theories, but already of spelling out individual empirical claims or experimental results. My argument then, to put it shortly, is this: justification, as arguing for the validity of results, shares an essential feature with argumentation in general: it necessarily has to make use of concepts that are regarded as stable, i.e., that are not put in question themselves. Thus justification is always relative to a specific conceptual framework, and this relativity transfers to the validity of the results that have been justified. Justification at any point in time takes advantage of all specifications and differentiations of that conceptual system, but is also limited by its limits. Moreover, as long as we are bound to our present conceptual system, these limits are “invisible” and can be seen only by “stepping outside.” The central means to do so is tracing back the concepts to their formation and their history. Hence, the attention turns to these concepts and conceptual frameworks, and to the question of how they are formed and stabilized—a question that has not found too much attention in history and philosophy of science so far. If it turned out that even processes of concept formation do allow a clear separation between generation and justification, we would not encounter difficulties with an overall separation. But I claim that this is not the case: not only have concepts been formed and stabilized in historical processes, but they continue to bear traces of these processes as well: their historicity is irreducible. The central argument is provided by analyzing the way by which concepts are stabilized and established. As I have emphasized above, concepts can, by their very nature, not be proved or disproved in the same way as theories. They have to “prove themselves” (“sich bew¨ahren” in German), or to “prove their worth,” and this procedure may fundamentally differ from explicit argument. Typically, it is achieved by pointing to their “successful” use, where “success” may refer to a broad variety of aspects: be it in improving the handling in the realm in question, or in allowing for wider generalizations than before, or in some other respects. Sometimes the procedure is not articulated at all. Forming specific concepts and establishing them against alternative concepts involve processes other than explicit argumentation,

CONCEPT FORMATION AND THE LIMITS OF JUSTIFICATION

191

processes that involve decisions and that are typically connected to specific historical constellations and practices. The process of stabilizing and establishing concepts, the decision to prefer a certain concept rather than an alternative one, is bound to specific historical constellations that comprise not only the epistemic constellation, but also practical uses and cultural norms, values, and tendencies. And the concepts thus formed and established bear traces, though invisible, of these historical particulars: once they are established, they become part of the very language of the realm in question. Alternative concepts disappear out of sight and are no longer discussed. Because of the very nature of concepts as elements of thought, the decisions made in the process of establishing become invisible, and thus become part of the concepts themselves. Thus concepts bear an irreducible historicity, they “have memories,” as Hacking put it (Hacking 2002, p. 37). To be sure, they cannot be reduced to their history, but at the same time history cannot be stripped off them. Among other things, this means that to assess the scope, the (often implicit) connotations, the limits, and the inherent tendencies of concepts, a closer look at their history is unavoidable. One of the few to acknowledge this point and even to turn it productively into scientific insights was Ernst Mach, who, in his “historical-critical” exposition of mechanics, analyzed the then well-established notion of absolute space in its historical roots and was thus able to remove this “monstrous concept” (“Begriffsunget¨um”) once for all from mechanical reasoning.12 It was not a theory that he rejected, but, more fundamentally, a concept that had just been taken for granted, and the removal of which opened new horizons indeed for the development of mechanics. Significantly enough, he did so by tracing back the concept to its history. Half a century later, and in a different field, the bacteriologist and epistemologist Ludwik Fleck made a similar point when he traced the history of such a fundamental concept as “infection disease” and showed exactly which decisions and cultural factors were involved in its acceptance and spreading (Fleck 1935 [1980]; Fleck 1979). Much stronger than Mach, moreover, Fleck highlighted how deeply social processes are involved in such developments. With some specifications, the tendency to understand the implications and scope of concepts by studying their history is even visible in William Whewell’s philosophy of science that he explicitly based on the history of the fundamental ideas of science (Whewell 1840), as Jutta Schickore points out in this volume. The historicity of concepts weighs all the more as we are usually not aware of it. We learn concepts during education, we normally just use them, and we use them as given and unproblematic. The limits of our conceptual systems are normally invisible: what is beyond the scope of our language seems unspeakable and, unthinkable. However, such a conceptual system has once been actively formed and stabilized in an earlier time, and these processes usually involve decisions. For systematic reasons, we are normally “blind” to these historical formation processes, processes that have opened specific perspectives, but have, as a rule, at the same time foreclosed the view on alternative accounts. Processes of concept formation are not only inherently historical, but they also have far-reaching effects on fundamental epistemic level. This insight sets definite limits to the separability of genesis and validity. Justification of a theory, e.g., as presented in a textbook, may well be independent of the genesis of that theory,

192

FRIEDRICH STEINLE

but certainly it is not independent of the genesis of the concepts it uses. Justification and validity can be separated from genesis only if the conceptual framework on which they rely is taken for granted and left untouched. At the moment, however, when the conceptual framework is taken into account and opened for discussion, the genesis and the historicity of concepts come in and remain there irreducibly. In this perspective, justification and genesis can no longer be neatly separated. As I have emphasized, I do not deny the possibility and even necessity of justification as systematically pulling together arguments for the validity of scientific claims. What comes out now, however, is that this procedure is not as absolute as often regarded.13 It is always relative to a conceptual framework that is taken as given. The irreducible historicity of these concepts transfers to justification and gives it an inherent historical character. Of course, it is different time frames that are made up here and there. On a short-term view, the stability of concepts is mostly taken for granted, and put into question only in exceptional situations. On a long-term perspective, however, the historicity of concepts becomes visible as a historicity of justification. Rather than denying the possibility of justification, this insight points to definite limits of justification and of the associated knowledge claims. FROM CONCEPT TO FACT: ESTABLISHING THE TWO ELECTRICITIES

In my last section, I shall illustrate these general considerations by coming back to the case of Dufay and the two electricities. It is most remarkable to see how Dufay’s new concept made its way among those who worked with electricity. The first historical observation is the nearly complete absence of an explicit discussion of the topic. This is all the more significant as the proposal did not disappear. On the contrary, it soon showed up, but in a very specific manner. Only 5 years later, the Leyden professor Pieter van Musschenbroek, the writer of a most important physics textbook of the period, presented the new concept as unproblematic. “Experience has taught us that electrical power is of a twofold manner . . . ,” he told the reader.14 It is worthwhile noting that he did not say, “Dufay has made this claim,” but “Experience has taught us.” Of course, this was not a denigration of the effort and achievement of Dufay— it was rather a significant indication of the specific level of knowledge that was addressed here. The existence of the two electricities was presented as indisputable, as a “fact” of nature, as we would say—and indeed, this quickly became the common view.15 Another five years later, in the first textbook ever devoted exclusively to electricity, the Petersburg academician Johann Gabriel Doppelmayr went even further, and took the twofold nature of electricity as a very part of its definition (Doppelmayr 1744, p. 1). And this was characteristic for a general attitude. In textbooks of the 18th century, the twofold nature of electricity was often introduced as defining feature, derived “from experience,” and only later in the text were the specific laws of attraction and repulsion presented (Frercks 2004). Georg M. Bose, who reinvented the electric machine so as to stimulate the wave of fascination with electricity, based his invention explicitly and essentially on the fact of two electricities (Bose 1744). By

CONCEPT FORMATION AND THE LIMITS OF JUSTIFICATION

193

the mid-1740s, thinking in terms of two electricities, or two states of electricity, had become basic for most of the users and researchers: not as a theoretical thing to be discussed, but as an undisputed empirical fact that was essential for guiding, spelling out, and understanding of everyday acting with electricity. It is sometimes objected that Benjamin Franklin, with his famous new theory, insisted exactly on the existence of only one, not two electric fluids. But with this proposal he addressed a different epistemic level. His goal was to explain with a microscopic theory the obvious and undisputed fact that electricity always had to be considered as twofold. In view of his theory, he renamed Dufay’s vitreous and resinous electricities as positive and negative, but this just highlights how fundamentally and undisputedly he maintained the basic dichotomy itself. Such a way of “reception,” of “filtering in,” is strikingly different from the way in which theories are properly received, both by the absence of explicit discussion and by the quick introduction of the concept not as something to be problematized, but as a fact, taught by experience. The process of stabilizing and establishing the concept of two electricities provides a striking case of how concepts may be established, or taken as valid, not by explicit arguments, but by other, less explicit pathways, such as their success in facilitating the practical handling. At the same time, the importance and consequences of such a process can hardly been overestimated. The once actively formed concept framed all further thinking and acting, research and techniques, on a fundamental level. Ever since Dufay’s achievement, it would have been very difficult, if not impossible, to conceive electricity other than twofold or bipolar. Any debate about, and justification of, the many electric, galvanic, and electromagnetic theories of the 18th and 19th centuries was conducted within this conceptual framework— a framework that itself was no longer subject of discussion. Taking such processes of concept formation16 and their epistemic significance seriously shows how justification of specific scientific claims relies on, and is relative to established conceptual frameworks, frameworks that themselves have historically been generated, and now remain undisputed. This insight brings an irreducible historical aspect into the picture and thus delineates strikingly the limits of separating validity from genesis.

ACKNOWLEDGMENTS

I thank Jutta Schickore and the other contributors to this volume for discussion, Uljana Feest for reading and critique, and the Fritz Thyssen Foundation (Cologne) and the Max Planck Institute for the History of Science (Berlin) for supporting my research.

NOTES 1. Heilbron 1979, chapter 9 gives a brief account of Dufay’s research. Dufay’s eight M´emoires at the Paris Acad´emie Royale des Sciences between 1733 and 1737 provide the main source of my account, along with some archival material; cf. also his own English summary (Dufay 1734). 2. He presented these researches in Dufay 1733. 3. For a detailed account, see Steinle 1997; Steinle 2003 or Steinle 2005, chapter 7.

194

FRIEDRICH STEINLE

4. Some philosophers, such as Paul Hoyningen-Huene, do readily accept my negative result. It is not so clear, however, whether such a result remains without consequences even for Reichenbach’s original DJ distinction, in particular for the very enterprise of rational reconstruction, when taking considerations about language and concepts into account, as I shall do below. 5. This point supports similar results in this volume, such as by Arabatzis, Nickles, and Potthast. 6. For details, see Paul Hoyningen-Huene (this volume). 7. Reichenbach 1938, pp. 6–7, see also the quotes given by Schiemann and by Richardson (this volume). 8. (Jacob 1997, chapter VII). Jacob refers explicitly to Fleck here. 9. It would be worthwhile to consider, however, whether including some history in science education would not significantly enhance its efficiency. 10. One may think of contextualism in epistemology, as presented in Williams 1996, for example—an elaboration of this approach for scientific knowledge would be a most interesting challenge. 11. This basic understanding runs through the recent philosophical discussions and is rather independent from the various views about what it means to have a concept, etc. (Smith and Medin 1981; Prinz 2002; Gurova 2003). 12. (Mach 1883; Mach 1960). I am grateful to Don Howard for reminding me of Mach in this context. 13. While this point has already been observed in general (Shapere 1980, e.g.), my analysis renders it more specific by pointing to concepts and language and the limiting factors. 14. “die Erfahrung hat uns gelehrt, dass die electrische Kraft von einer zwiefachen Art sey . . . ” (Musschenbroek 1739. cf. the German translation: Musschenbroek and Gottscheden 1747, p. 242). 15. (Winkler 1745, p. 5; Musschenbroek and Gottscheden 1747, p. 242; Gralath 1747, p. 208). The latter work is a huge and most comprehensive (natural and chronological) “History of Electricity,” which most later historians of electricity, from Priestley to our time, like to take as source. 16. Of course, the case is not singular. For similar examples, one might think of how the mechanical and astronomical debates of the 18th and 19th centuries were shaped by Newton’s concepts of mechanical force and of absolute space, and how only in the late 19th century, i.e., three centuries after their formation, these concepts became seriously questioned, and new horizons opened. One might also mention the concept of chemical reaction that emerged only in the 17th century and succeeded against other, competing concepts of chemical processes (Klein 1994). In both cases the fundamental character needs no extra emphasis.

REFERENCES Bose, Georg Matthias (1744), Tentamina electrica in academiis regiis Londinensi et Parisana primum habita, omni studio repetita (Wittenberg: Johann Ahlfeld). ¨ ¨ Doppelmayr, Johann Gabriel (1744), Neu-entdeckte Phaenomena von bewunderungswurdigen urkungen der Natur (N¨urnberg). Dufay, Charles Fran¸cois de Cisternai (1733), “Quatri`eme m´emoire sur l’´electricit´e. De l’attraction et r´epulsion des corps e´ lectriques”, Histoire de l’Acad´emie Royale des Sciences, avec les M´emoires de Math´ematique & de Physique pour la m eˆ me ann´ee: pp. 457–477. Dufay, Charles Fran¸cois de Cisternai (1734), “A Letter from Mons. Du Fay, F. R. S. and of the Royal Academy of Sciences at Paris, to His Grace Charles Duke of Richmond and Lenox, concerning Electricity. Translated from the French by T. S.MD.”, Philosophical Transactions 38(431): 258–266. Fleck, Ludwik (1935 [1980]), Entstehung und Entwicklung einer wissenschaftlichen Tatsache. ¨ Einfuhrung in die Lehre vom Denkstil und Denkkollektiv (Frankfurt: Suhrkamp). Fleck, Ludwik (1979), Genesis and Development of a Scientific Fact (Chicago: University of Chicago Press). Frercks, Jan (2004), “Disziplinenbildung und Vorlesungsalltag. Funktionen von Lehrb¨uchern der Physik um 1800 mit einem Fokus auf die Universit¨at Jena”, Berichte zurWissenschaftsgeschichte 27: 27–52. Gralath, Daniel (1747), “Geschichte der Electricit¨at”, Versuche und Abhandlungen der Naturforschenden Gesellschaft zu Danzig 1: 175–304.

CONCEPT FORMATION AND THE LIMITS OF JUSTIFICATION

195

Gurova, Lilia (2003), “Philosophy of science meets cognitive science: The categorization debate”, in D. Ginev (ed.), Bulgarian Studies in the Philosophy of Science. Boston Studies in the Philosophy of Science, Vol. 236 (Dordrecht: Kluwer), pp. 141–162. Hacking, Ian (2002), Historical Ontology (Cambridge: Harvard University Press). Heilbron, John L. (1979), Electricity in the 17th and 18th Centuries (Berkeley: University of California Press). Jacob, Fran¸cois (1997), La souris, la mouche et l’homme (Paris: Odile Jacob). Klein, Ursula (1994), “Origin of the Concept of Chemical Compound.”, Science in Context 7(2): 163– 204. Mach, Ernst (1883), Die Mechanik in ihrer Entwickelung: historisch-kritisch dargestellt. Internationale wissenschaftliche Bibliothek 59 (Leipzig: Brockhaus). Mach, Ernst (1960), The Science of Mechanics: A Critical and Historical Account of its Development. 6th edn., with revisions through the 9. german ed. Transl. Thomas J. MacCormack, introd. Karl Menger (LaSalle, IL: Open Court Publ.). Musschenbroek, Petrus van and Gottscheden, Johann Christoph (1747), Hrn. Peters von Muschenbroek ¨ Grundlehren der Naturwissenschaft, nach der 2. lat. Ausg., nebst einigen neuen Zusatzen des Ver¨ fassers, ins Deutsche ubers. Mit einer Vorrede ans Licht gestellt von Johann Christoph Gottscheden (Leipzig: Kiesewetter). Musschenbroek, Pieter van (1739), Essai de physique—Beginsels der natuurkunde (Leyden: Luchtmans). Prinz, Jesse J. (2002), Furnishing the Mind: Concepts and Their Perceptual Basis. Representation and Mind (Cambridge, MA: MIT Press). Reichenbach, Hans (1938), Experience and Prediction: An Analysis of the Foundations and the Structure of Knowledge (Chicago: University of Chicago Press). Shapere, Dudley (1980), “The character of scientific change”, in T. Nickles (ed.), Scientific Discovery, Logic and Rationality. Boston Studies in the Philosophy of Science, Vol. 56 (Dordrecht: Reidel), pp. 61–101. Smith, Edward E. and Medin, Douglas L. (1981), Categories and Concepts. Cognitive Science Series Vol. 4 (Cambridge, MA: Harvard University Press). Steinle, Friedrich (1997), “Entering New Fields: Exploratory Uses of Experimentation”, Philosophy of Science 64 (Supplement): S65–S74. Steinle, Friedrich (2003), “Experiments in History and Philosophy of Science”, Perspectives on Science 10(4): 408–432. Steinle, Friedrich (2005), Explorative Experimente. Amp`ere, Faraday und die Urspr¨unge der Elektrodynamik. Boethius 50 (Stuttgart: Franz Steiner Verlag). Whewell, William (1840), The Philosophy of the Inductive Sciences, Founded Upon Their History (London: Parker). Williams, Michael (1996), Unnatural Doubts: Epistemological Realism and the Basis of Scepticism (Princeton: Princeton University Press). Winkler, Johann Heinrich (1745), Die Eigenschaften der electrischen Materie und des electrischen ¨ Feuers, aus verschiedenen neuen Versuchen erklaret, und, nebst etlichen neuen Maschinen zum Electrisiren beschrieben (Leipzig: Breitkopf).

THOMAS POTTHAST

CONTEXTS OF JUSTIFYING AND DISCOVERING THE NATURE OF ECOSYSTEMS: FROM CONCEPTS TO OBJECTS AND VICE VERSA

1. INTRODUCTION

The distinction between a context of discovery and one of justification with regard to scientific activities (hereafter: DJ distinction) has developed a remarkable life of its own in 20th century philosophy of science and beyond, eventually becoming an influential conceptual device. Hans Reichenbach’s exposition provided the major starting point for a broad discussion, within which several versions of the distinction emerged.1 One can safely state that no such thing as one single proper DJ distinction exists. Rather, it implies a richness of different perspectives on the relation between historical and philosophical approaches to science. Drawing on a discussion about the foundation of “ecosystems” in the history and philosophy of ecology, this paper addresses two themes connected with the DJ distinction. The first is the distinction between historicized descriptive and epistemologically normative methodological perspectives. This version maps out a “disciplinary” separation: On the one hand, history of science is usually outlined as descriptive and neutral with regard to epistemological questions. On the other hand, philosophy of science is singled out as the epistemologically normative enterprise. The second theme is the sequential interpretation of the DJ distinction—the “temporal distinction”. According to this distinction, discovery and justification are consecutive and mainly separate processes in the development of science. Reichenbach’s text and the initial debates about ecosystems are located in roughly the same historical period. In section 2, my discussion will hence start from Hans Reichenbach’s original outline of the DJ distinction in his 1938 book Experience and prediction for both epistemological and historical reasons. It addresses, among others, the question of epistemological normativity and the process distinction. I shall argue that despite the explicit agenda for proper separation, a peculiar hybridization of descriptive and normative epistemology results from Reichenbach’s exposition of the distinction. The rest of the paper is devoted to the history of ecology as a case study. It exemplifies the two themes of the DJ distinction mentioned above. Section 3 discusses the DJ distinction as a model of sequential processes generating new epistemic objects in ecology. The investigation of the emergence of the “ecosystem”—and related conceptual and theoretical novelties—shows how ecological objects of that sort were in some sense first “justified” and then “discovered”, that is, empirically substantiated. Section 4 provides a comparative overview of different historical narratives 197 J. Schickore and F. Steinle (eds.), Revisiting Discovery and Justification, 197–214.
C 2006 Springer. Printed in the Netherlands.

198

THOMAS POTTHAST

of the foundations of the “ecosystem”. I show that these narratives inevitably rely on epistemological presuppositions. In section 5, notions of complete separations of “scientific content” or validity questions from “non-scientific contexts” or genesis questions are challenged in both directions: The justification of theories cannot be properly understood epistemologically without regard to the processes of their discovery. Hence the context of discovery must be understood as shaping, so to speak, also the content of systematical justifications of the theoretical text investigated from a reflexive perspective of normative epistemology. And vice versa: A fruitful historical study of cultural, social, or psychological contexts of discoveries has to take into account the systematically oriented epistemological perspective that the justificatory texts necessarily bring to bear. To be sure, the DJ distinction remains a useful regulative idea, but any sharp DJ distinction blurs, when it comes to concrete accounts of adequate reconstructions of how science is done, because modes of justification and discovery remain not completely separable also on the meta-level. At the same time the DJ distinction provides both an important analytical tool and a means for methodological clarification. Nevertheless, neither history nor philosophy of science can escape dealing with both historicized descriptive and epistemologically normative perspectives. 2. REREADING REICHENBACH—HYBRIDIZING DESCRIPTIVE AND NORMATIVE EPISTEMOLOGY

Turning to the original sources may be one of the common methodological denominators for both historians and philosophers of science. The primordial source of the DJ distinction is the introduction of Hans Reichenbach’s 1938 book, Experience and Prediction—An analysis of the foundations and the structure of knowledge. Written in English by an exiled German philosopher of the Vienna Circle teaching at a Turkish University, the book was published in the United States in the very year when Nazi Germany took political control over Prague and when the first organized pogroms against Jewish citizens occurred in Germany. This text obviously bears fascinating historical contexts. However, disregarding these contexts, in this section I will provide a close reading of the immediate textual framework of the distinction between the “context of discovery” and the “context of justification” (the DJ distinction)2 . The introductory chapter of Reichenbach’s book exposes an agenda for epistemology as a philosophical enterprise (Reichenbach 1938, p. 3–16). One might also identify the first pages of that outline as taxonomy of science studies, since philosophy of science is placed within a larger range of possible scholarly perspectives on science.3 Reichenbach attaches different agendas to different fields, all of which have science as their object of study.4 Epistemological considerations are not at all excluded from sociological investigations of science, but he creates an isolated space exclusively reserved for epistemology as a distinct approach. Epistemology is vested with three overall tasks: Description, critique, and advice. The demarcation of what is the proper domain for epistemology within each of these tasks asymmetrically separates philosophy from sociology of science. According

JUSTIFYING AND DISCOVERING THE NATURE OF ECOSYSTEMS

199

to Reichenbach, “internal” and “external” relations between human utterances and knowledge demarcate descriptive epistemology from sociology. Although sociology deals with the alleged external social settings, it also integrates the content of science and, thus, epistemological perspectives. The relation between sociology and philosophy therefore is asymmetrical. Epistemology, i.e., philosophy of science or “scientific philosophy” proper, is to deal in the descriptive part solely with the (“internal”) content of science, whereas psychology and sociology also investigate external issues. Reichenbach’s context distinction is introduced with regard to a subdivision of the descriptive task of epistemology, namely, the process of activities in science called “discovery”. Psychology may provide descriptive accounts of the actual ways of thought of more or less ingenious scientists—the “context of discovery”. But epistemological issues are completely excluded from this psychological description of discovery and remain restricted to logical questions, namely the “context of justification”. The latter stands for a descriptive but partly counterfactual rational reconstruction where epistemology provides a “fictive construction”, which justifies the previously discovered knowledge.5 After exposing the context distinction, Reichenbach moves to the tasks of epistemology beyond description. His treatment of critique in the philosophical sense and even more his notions of advice are most fascinating if one reads them as general notes of philosophical counseling. The critical task includes the analysis of science as the core of epistemology. Only here can fallacious elements inherent in the descriptive part of rational reconstruction be addressed.6 Finally, the advisory task of epistemology is raised—and immediately taken back by reducing it to the critical task of the “discovery [sic!] of entailed decisions” instead of providing advice to scientists (Reichenbach 1938, p. 14).7 Here Reichenbach reformulates the issue of epistemological advice to scientists. He concedes that epistemologists could provide help as to which path to follow in the “forest of knowledge.” But ultimately he does not want to go that far. Reichenbach suggests that the advisory task is only to exposit the “entailed decisions”—a set of epistemological “if-then scenarios” to be used as “logical signposts” (Reichenbach 1938, p. 14). The decision on taking a direction should be left to scientists in the process of doing science. Although the DJ distinction is exposed in the descriptive task of epistemology, the description of justification becomes part of the second, critical task. Simultaneously, the advisory task is reduced to the critical task of exposing epistemological discoveries—entailed decisions—that are essential for extending the critical and reducing the advisory task. What becomes apparent in that exposition is an unstable and rather unclear status of descriptive and normative issues among all three tasks of epistemology and eventually hybridization within the critical task. To sum up the sequential order in Reichenbach’s argument: (i) The DJ distinction is initially established within the descriptive task. (ii) Then descriptive epistemology, expelled in the disciplinary sense from the context of discovery and now solely committed to the context of justification, is pushed into the critical task as part of the rational reconstruction. (iii) The advisory task, after being rejected as a feasible political and epistemological normative task, is modeled on epistemological discoveries: entailed decisions. Functioning as

200

THOMAS POTTHAST

providing logical signposts, the remaining advisory task is also moved into the space of the critical task, which constitutes the overall “analysis of science”.8 From this textual interpretation of the DJ distinction as exposed by Reichenbach in his 1938 book, a historically oriented assumption can be drawn: It is precisely the lumping together of the task of the descriptive and the epistemologically normative that might have produced subsequent, different versions of the DJ distinction galore as well as its ensuing criticisms. Also it seems safe to say that the hybridization of historicized descriptive and normative epistemological issues is still disturbing to both history and philosophy of science.9 Approaches to (dis)entangle them must take into consideration that history and philosophy treat this problem in different ways. I shall get back to this point after discussing the case study. 3. EPISTEMOLOGY IN AND OF ECOLOGY: TWO JUSTIFICATORY ACCOUNTS OF DISCOVERIES

How does epistemology matter in and for the history of ecology? Most ecologists, historians, and philosophers agree that ecology used to be a very unstable professional and disciplinary field within the sciences at least until the 1950s, thus providing abundant fuel for ongoing, intense debates on fundamental epistemological issues.10 This has inspired ecologists to search for general philosophical foundations of their field, not least in order to provide a unified theoretical framework for divided and sometimes adversarial sub-disciplines. Epistemological issues were always at stake: The concept “ecology” itself was coined by Ernst Haeckel in the course of outlining a system of biology and its sub-disciplines. From the 1920s onwards, long epistemological battles unfolded about the role of experiment as the one and only way of doing science properly. These battles resonated with the development of Logical Positivism. From the 1950s, Popper’s version of hypothetico-deductivism was explicitly addressed. From the 1970s, Kuhnian notions of paradigms became fashionable in both ecology and the philosophical approaches to it. One cannot overestimate how influential Kuhn’s historical developmental scheme of a “proper” scientific discipline was for ecologists.11 Recently, Lakatos’s suggestion of fruitful research programs has also been applied to the field by ecologists and other scholars (Worthington 1975; Schwarz 2003). What is discovered in science? An standard answer from the early 20th century might have been: new celestial or micro-world bodies, laws of nature which are conceptualized in theories about natural processes of descriptive and predictive character—the more mathematical the better. How can we apply this answer to biology? There has been a long debate on the question of whether there are laws in biology (including ecology).12 Following a recent suggestion, in biology one can discover both entities and activities relevant for the respective field of inquiry (Machamer et al. 2000). Note the difference for conceptual issues. The concepts of “gene”, “population” or “ecosystem” would not be objects of discovery, but the related material entities of scientific practice. Likewise, DNA-splicing, predator impact on a population and nutrient cycling would be activities subject to discovery. In this sense, discovery appears to be ultimately linked with realist ontology: Things and processes,

JUSTIFYING AND DISCOVERING THE NATURE OF ECOSYSTEMS

201

i.e., entities and activities, are out there waiting to be discovered. It should be noted, though, that this ontic presupposition is not a necessary one here. What Reichenbach and others called “discovery” may be the same as what science studies scholars call “construction” or “making.” A discovery of this sort means the coming into being of “epistemic objects” in science on a very general level—leaving aside for a moment the ontological and metaphysical issues involved.13 What is justification? It is a process by which scientists and philosophers and historians stabilize the existence, the meaning, and the significance of an entity or an activity that they have already researched and/or want to research in the future. Thus, I opt against the understanding of a necessary sequential order of first, discovery and secondly, justification. Discovery may or may not antedate justification. The two possible sequential orderings of “justification” and “discovery” might provide insights in different formations of developments in science. As an analytical distinction within a continuum, two major ways of producing new epistemic objects and the ensuing theorizing can be identified. Both can be identified by scientists and historians as well as in epistemological treatments of ecology: We encounter justifications that precede discovery and discoveries that precede justifications (the latter case is in accord with Reichenbach’s temporal distinction).

Sequential Distinction I For ecology, the first account can be characterized as: In the beginning was the word, and the word was with the ecologist. (And word was transformed into action). There are several examples where a field of research, a concept, an ecological unit, a methodological innovation by a conceptual innovation is justified first, and only later, material research and empirical “discovery” of the hypothesized item unfold. Ernst Haeckel coined the very term “ecology” in 1866, years before ecology emerged as a substantial field of research practice. August Thienemanns’s first notion of the unity of community and habitat being an organism of higher order (“Lebensgemeinschaft & Lebensraum”, “Organismus h¨oherer Ordnung”) appeared in 1916 as a theoretical side-remark in the context of zoological taxonomy and biogeography, preceding most of his empirical research on the alleged units as units. Karl Friederichs’ subsequent elaboration of this theme, coining the term “holocoen” in 1927 delineating a special unity factor of any distinctive ecological space, alluded more to future research of a very specific epistemological kind (Potthast 2001). The most prominent neologism, ever since Haeckel’s “ecology”, was Arthur Tansley’s “ecosystem” in 1935. It emerged from a discussion on terminology as a core epistemological issue. Tansley introduced the term not in the context of empirical data, but in a philosophical debate about doing ecology and the nature of ecological objects.14 Other examples are Alfred Lotka’s ideas about a physical biology around 1925, put into practice some thirty years later by thermodynamically oriented ecosystem ecology/systems ecology (Taylor 1988; Hagen 1992, p. 125f). A famous theoretical as well as conservation-oriented epistemic object in population ecology—the dynamic flux of local populations—was first justified mathematically by Richard Levins in

202

THOMAS POTTHAST

the late 1960s and 20 years later discovered, i.e., fleshed out in empirical research as “metapopulation” (Levins 1968; Reich and Grimm 1996). Likewise, “biodiversity” in 1986 began as a disciplinary justification of tropical ecology, evolution and taxonomy with regard to conservation biology and global conservation policies. “Biodiversity” was justified before truly discovering, i.e., establishing in practice, how biodiversity as such could scientifically be grasped in detail (Potthast 1996; Takacs 1996). In all of these cases, definitions and outlines of further research as “testing” the validity of empirical and epistemological claims played a large role. One might even go as far as relating these efforts to the philosophical camps of anti-inductivism and hypothetico-deductivism. Sequential Distinction II The second type of discovery and justification is: Ecological objects are mangled out of practice, the work on and transformations of material objects already at hand. The most famous examples from the early times of ecology are Karl M¨obius’s 1877 biological communities (“Biocoenose”, drawn from studies of oyster banks) and Stephen Forbes’s 1887 “lake as a microcosm”, both emerging from studies of economically oriented zoology as the basis of fisheries. Similar investigations of reservoirs as well as volcanic lakes led Thienemann to develop his notion of nutrient cycling in 1926. The same holds for the subsequent investigations of production and productivity. Raymond Lindeman’s famous outline of the “trophic-dynamic aspect of ecology”, published in 1942, also started from a natural history investigation of a senescent pond devoid of much of the later theoretical framework. And, as a last example, a large portion of recent population ecology as well as the notion of patch dynamics as a whole departed from empirical problems of species protection and wildlife management.15 Constructing two types of processes with reverse ordering of discovery and justification is of course at odds with Reichenbach’s temporal version of the DJ distinction but it could be related to the version of empirical (discovery) versus logical (justification). Here we see the DJ distinction at work—productively generating variations that do not always relate in a coherent way. Concerning the temporal version: Given the inextricable historical synchronicity of epistemological justificatory processes already within (the context of) discovery, one might gain some useful insights juxtaposing two types of generating and justifying—as the extremes of a continuum—in order to understand the different ways in which epistemic objects and the related theories may come into being. 4. ON THE ECOSYSTEM: CONFLICTING NORMATIVE HISTORIOGRAPHIES

Ecologists have always struggled for the existence and establishment of their field as a full-fledged scientific discipline. Yet in some sense, the emergence of the ecosystem and ecosystem research followed the classical H-D model of philosophy of science: As alluded to above, concepts and theories have often been constructed in the sense of justification and subsequently subjected to empirical testing. However, in

JUSTIFYING AND DISCOVERING THE NATURE OF ECOSYSTEMS

203

the following, I show that one could go further. The labels “discovery” and “justification” in a historical account are themselves subject to epistemological, ontic, and other presuppositions or judgments. These are necessarily inherent in the narrative and cannot be identified strictly as either valid or invalid from any meta-perspective. With regard to the hybridization of descriptive history and normative epistemology in spite of the DJ distinction, or even as one result of Reichenbach’s exposition, different accounts of the inception of the ecosystem are juxtaposed. Can an ecosystem be discovered at all? Does that count as a discovery in the sense of epistemology? Following the arguments of the preceding section, it is appropriate to ask how the ecosystem and its properties have been both discovered and justified in Reichenbach’s sense of descriptive epistemology and whether these processes can be separated, given that any historical reconstruction relies on systematic presuppositions. Four narratives will be discussed with regard to the question of how the history and epistemology of the ecosystem is presented. I focus on the following issues. Given the generally accepted notion of ecosystem research emerging only after 1945 and mainly in the United States, what are the reasons for this success? Acknowledging important “forerunner” traditions of “synecology” (ecology of communities) in Germany before Word War II, how is the alleged delay of implementing the ecosystem treated? What is the epistemological and ontological status that the authors bestow upon the ecosystem in their respective narratives, and how does this status play out? In the beginning of the 1990s, the first two book-length studies were published on the history of the ecosystem in ecology. Historian Joel Hagen’s An entangled bank. The origins of ecosystem ecology draws on the famous passage on an “entangled bank” from the final paragraphs of Charles Darwin’s book The origin of species, which allude to the web of ecological relations between organisms. The narrative is thereby framed with reference, if not reverence, to Darwin and Darwinism, which does not come without some irony: Hagen concludes his book by pointing to the fact that precisely during the establishment of ecosystem ecology in the mid 20th century, a dissenting approach of “evolutionary ecology” emerged by adopting a Neo-Darwinian perspective for the field of population and community studies. From an epistemological perspective, it is rather obvious that ecosystem ecology did not adhere to the specific Darwinian biological framework. It might be added that it hardly could, because it reversed ultimate and proximate mechanisms of biological fitness and reproductive success as outlined by Neo-Darwinism and the Evolutionary Synthesis, replacing biological considerations by a purely physics-oriented thermodynamic perspective.16 Hagen sees an epistemological “failure” on both sides. As one consequence, only ecosystem ecology linked up with an environmentalist perspective and the moral dimensions of nature. The epistemological, the political and the moral dimensions of the “ecosystem” cannot be deduced from each other, but rather form an interrelated network (Hagen 1992, pp. 195–97).17 In a nutshell, Hagen’s thesis may be summarized by the motto: “math, physiological and physical analogies, and politics matter.” He sketches the successful heuristics of mathematization and systems theory, but also demonstrates how the physiological analogy of the organism eventually led to a physiology of ecosystems on the basis of thermodynamics. It is in this context that

204

THOMAS POTTHAST

notions of “self-regulation” with a metaphorical link both to organisms and machines hybridize successfully (Hagen 1992, p. 128ff.).18 At the same time, the importance of successfully implementing the ecosystem approach as part of the program of radiation studies of the U.S. military due to the Cold War is stressed. Hagen does not treat the case of Germany and thus only implicitly points to the missing acceptance of mathematics and missing financial support of ecology as “big science” in Germany. Concerning the way from Tansley’s coining of the word to a full-blown research program of ecosystem ecology, Hagen discusses the famous 1942 paper of Raymond Lindeman, now regarded as the first mature document of the ecosystem perspective in ecology proper. By providing details of the process, Hagen alludes to the ex-post construction of that perspective. After Lindeman’s untimely early death in 1942, his mentor G. Evelyn Hutchinson not only succeeded to publish the article despite other rather critical peer reviews. He edited it and also added an interpretive postscript paving the way for a distinct way of reception of the paper as an exemplary ecosystem research approach. However, it appears that Lindeman rather started his research on Cedar Bog from a descriptive natural history perspective of the biological community and its flow of matter and energy, drawing on existing studies of productivity and ecological succession, among others. Hagen does not treat the ecosystem as a peculiar ecological object different from other units like populations. Although he does not comment explicitly on its ontic status, I interpret his account as taking seriously ecosystems as real-world units.19 In this sense, the empirical substantiation of a new ecological unit went from first a theoretical justification by Tansley via a broad variety of empirical investigations of a concrete object in space and time (the now famous Cedar Bog) to an ex-post justification of the empirical discovery of ecosystems as trophic-dynamic units. The next book appeared in the year after Hagen’s publication, Frank Golley’s A history of the ecosystem concept in ecology. More than the sum of the parts. Golley, a disciple of Eugene Odum and active in the “Man and Biosphere” ecosystem research program, provides a classic participants’ account of his own field, to be sketched here by the motto: “epistemology and politics matter”. As one can readily infer from the title, he places much effort on issues of ecosystems being epistemologically and ontologically “real” by unfolding emergent properties. The claim of physical reality of the unit of organisms and their environment is not directly drawn from Tansley’s paper, in which the coning of the term appears mainly to be part of a conceptual debate criticizing an idealist ontology of climax as provided by some vegetation ecologists.20 Golley credits Lindeman being the first true ecosystem ecologist although he mentions the context of how the paper emerged, i.e., despite the post interventions brought forward by G. Evelyn Hutchinson after Lindeman’s death, alluded to above. In accordance with other authors’ accounts, Golley then explains the successful development of the agenda exposed in Lindeman’s paper in the U.S. on the basis of Cold War and “big science” business. As to ecology in Germany, Golley makes two remarkable points. First, he credits Karl Friederichs’s 1927 notion of “holocoen” as well as August Thienemann’s elaborations of “Lebensgemeinschaft & Lebensraum” starting already in 1916 as precursors very close to the ecosystem concept, despite their ontological,

JUSTIFYING AND DISCOVERING THE NATURE OF ECOSYSTEMS

205

rather than (mainly) functional, concepts of a spatially explicit “reality”. However, following Golley, these units display features of a self-regulating “organization” and thus are granted being in accord with later ecosystem ecology. Second, the lack of pragmatic and programmatic progress of ecosystem ecology in Germany after 1945 is interpreted as resulting from the negative political connotations of any “holism” in postwar Germany, because the National Socialists had employed this terminology. The histories of Hagen and Golley do not conflict, but place different emphasis on issues due to their different epistemological perspectives. Hagen places the ecosystem into a larger framework of biology whereas Golley discusses it as a proper discovery in its own right, with much emphasis on its status as a unit of its own. The next two pieces of historical accounts conflict explicitly over a slightly earlier episode, namely Arthur Tansley’s coining of the term ecosystem. They also question claims of the earlier works on epistemological grounds. Contemplating the origin and validity of concepts, ecologist and philosopher of ecology, Kurt Jax, presents a clear-cut message (Jax 1998):21 The ecosystem concept of Arthur Tansley eventually made its way mainly because of its epistemological sophistication and its anti-ontological stance. An ecosystem was not integrated as a real-world object but rather as a “mental isolate”, a methodological construct. The author juxtaposes this concept with the German concepts of “Lebensgemeinschaft” and “Holocoen”, which he chided to be loaded with a metaphysics of German Naturphilosophie and thus not applicable to produce any stable research program for scientific ecology. Jax, by the same token, questions Golley’s interpretation of the political notions of “holism” being a problem in German ecology, because none of these authors had changed their terminology after 1945 and because no debate on these matters can be found until the late 1970s.22 Rather, Jax maintains, it is the claiming of an ontological reality of ecological units on the ecosystem level that didn’t allow any feasible research program to develop. Jax finds Tansley to account only for a methodological reality of ecosystems as “systems we isolate mentally” (Tansley 1935, p. 300). Following Jax, only overcoming metaphysical or other non-scientific notions of ecosystems as “given” objects of the material world allowed the ecosystem to thrive. This contrasts not only with Golley’s account, but also with Hagen’s narrative, which both present the material reality, or at least the alleged reality as provided by the scientists, respectively, being one of the driving forces behind successful ecosystem research. Historian of science Peder Anker, in a study of Imperial Ecology in England and British colonial Africa, chided Jax for having gotten it completely wrong (Anker 2001, 296 no. 106). He points out that Tansley’s ecosystems are nothing less than only “mental isolates” because they are “coincident with physical systems” (Tansley 1935, p. 300 (footnote)).23 Anker emphasizes the context of Tansley living and writing in Oxford, where he decisively took part in a broad academic philosophical debate on the realists’ side as opposed to holistic idealists. Thus Anker describes the epistemological framework differently. For him Tansley was a realist—albeit not at all na¨ıve—with regard to the ecosystem. The opposing ontic view was a kind of Platonic idealism of holism. Anker locates Tansley in a materialist camp opposed to skeptical philologists

206

THOMAS POTTHAST

strongly contesting the increasing domination of science in the academe and beyond (Tansley 2002; Anker 2002). On this background, Tansley’s attacked his ecological foes, vegetation scientist John Phillips and his mentor, General Jan Christian Smuts, who wanted to establish a specific biological and idealist notion of holism both in ecology and in politics. For Anker the motto reads: “epistemology matters when linked with politics”. Hence, realism becomes associated with a larger British science-based movement of successful leftwing political planning and conservation. Right-wing politics of nature draw on an idealist holism, which is skeptical to the almighty power of materialist science. Although Anker makes his point about properly placing Tansley in the historical context, he implicitly takes sides with the materialists against holistic idealism. However, his critical judgment on the “reality” of ecosystems becomes visible in his critique of naturalisms of different kinds, both in the introduction and the conclusion of his book. Not epistemology or ontology, but rather the political economy of scientific knowledge claims is regarded as the major problem of the ecosystem and its history. Now, who is right? All four authors maintain that reference to the context of discovery is important if we want to get straight any normative epistemological issues (Anker 2001, p. 296 no.107; Jax 2002, p. 122).24 Concerning one major issue, I strongly opt for explanations provided by Golley and Anker. The most important success of ecosystem ecology—following the Odum Brothers—clearly was built on a realist, even ontic perspective of the ecosystem as a real-world entity. The question of the epistemological validity of such a status still prevails. If one takes the epistemologically normative position that the ecosystem is only a methodological construction and hence not as real as empirical objects like organisms and populations, then Jax’s narrative makes sense. However, reasons for this position emerged only as a result of the later history of the ecosystem, not the justification/discovery by Tansley.25 To substantiate Jax’s perspective, one might point to the telling fact that up to 1990s the major Britain-based textbook of ecology neither in its title nor in the text treated “ecosystems” as respectable ecological units but only “individuals, populations and communities” (Begon et al. 1996).26 Epistemological statements regarding the nature of ecosystems have been linked to both theory choices in ecology and environmental and political agendas. Challenging the ontological and epistemological claim that the ecosystem is solely a methodological abstraction has resulted in the claim that evolutionary biology and organisms are more adequate for ecology than thermodynamics and (bio)geochemistry as a physiology of higher order. Politically, the reification of ecosystems is often associated with technocratic or eco-totalitarian, anti-humanist agendas.27 The justification and discovery of the ecosystem took place in rather different ways in the Anglo-American and Continental European setting. Starting as an alleged “mental abstraction” in Britain, the ecosystem was transformed into a naturalistic (implicitly ontological) real-world unit in North America. The reverse happened in Germany with “Lebensgemeinschaft” and “Holocoen”: Starting with an ontological commitment to the nature of ecological units, the interpretation of the investigated results was claimed to require an also trans-scientific, somewhat idealistic approach of “really” grasping the essence of such objects. Although rather different in content

JUSTIFYING AND DISCOVERING THE NATURE OF ECOSYSTEMS

207

and theoretical issues, the pattern of epistemological reasoning appeared to be quite similar. The role of research objects (mainly lakes) and methodology did not differ before the early 1940s. Only later, a combination of theoretical decisions—adopting general systems theory and hence denying the need for a specific biological theory of ecosystems—and new techniques—e.g., radiocarbon tracers for studying the flow of matter in ecosystems—made a difference for the U.S. context. The gains and losses of this development are still contested. Ever since, the international community of ecologists has disputed the “theoretical” justification as well as the subsequent “empirical” proof of both the aforementioned options. To sum up this section: 1. History of ecology and its epistemological context: Contesting the status of the ontological and epistemological “reality” of ecosystems played a formative role for the history of justifying and discovering the ecosystem in ecology. However, the claims that on both practical and epistemological levels ontologically charged views of the reality of ecosystems scientifically failed are historically unjustified: It was precisely the reification and—maybe—counterfactual simplification of “the” ecosystem that, among others, rendered especially the ecosystem ecology research program to successfully spread from the United States. Moreover, this implies that (from today’s perspective) allegedly deficient epistemologies might nevertheless foster research programs, which eventually also turn out to produce valuable data and “better” theories. 2. Historiography of ecology and its epistemological context: Differences of historical narratives can partly be explained by meta-analyzing the historian’s epistemology. Depending on the historian’s take on the epistemological and ontological status of the ecosystem, different narratives unfold. 3. Philosophy of ecology and its historical context: A sophisticated epistemological realism debate on ecosystems still continues. It would be a task of normative epistemology to ask: Is it possible to justify the claim that individual organisms, biological species, and populations are real, in contrast to allegedly purely methodological constructions of ecosystems? In what historical contexts do different answers of these questions receive—maybe different—credibility? The description and critique of the nature of ecosystems, and political as well as epistemological advice both to and from ecosystem ecologists will continue to be contested issues for the philosophy and history of ecology. 5. DISCOVERY AND JUSTIFICATION AS SEPARATING HISTORY AND PHILOSOPHY OF SCIENCE?

At least two good reasons have been provided for re-thinking the context distinction. First, it is a significant topic “in and for” (Nickles 1980, p. 1) the history of science as well as the history of philosophy. Second, reflecting on the context distinction may foster insights in the peculiar relationship between the history and the philosophy

208

THOMAS POTTHAST

of science. But there is more. Beyond historical and methodological significance in their own right, the DJ distinction resonates with current debates about science and science studies: Does a core of scientific theories and practices exist which transcends historical, social, political, aesthetic and moral contingencies? Can there be, in other words, trans-historical epistemological justifications in science, which can be completely separated from all contingencies of discovery? Notwithstanding the opposing claim that the context distinction is outdated because there is no such thing as trans-historical justification, the distinction still makes sense beyond that extreme version. Moreover, it can help us understand the (un)stable systematic status of science and the current debates in science studies. Identifying a historical, a methodological and a systematic aspect of the context distinction is not new. Nor is it new to claim that confusion often arises because both protagonists and critics lump together different aspects of the context distinction (Nickles 1980; Hoyningen-Huene 1987; among many others). In fact, this lumping together can be traced back to Reichenbach’s ambiguity in addressing the different tasks of epistemology that I alluded to in the preceding section. The most fundamental conflict between history and philosophy of science lies within the third, the systematic aspect—the descriptive (rather: analytical) versus the normative perspective on science.28 This distinction may be regarded as a still valid juxtaposition made by Reichenbach when introducing the context distinction. History and philosophy of science—as well as the sciences as objects of their study—all grapple with the juxtaposition of the analytic versus the normative. And they do so in different ways, as I showed in the case study on ecology and the ecosystem above. The simple connection of history with discovery and of philosophy with justification does not work.29 As indicated by the publication of two volumes on Scientific Discovery in the Boston Studies in the Philosophy of Science edited by Thomas Nickles 25 years ago, the issue of discovery has returned to philosophy of science. Most philosophers acknowledge discovery as an important, if indispensable, issue to understand science.30 Assuming the significance of discovery can be taken for granted in the history of science, two questions remain: What happened to justification in the philosophy of science31 and how are epistemological issues of justification treated in the history of science? My deliberations fall into this latter area. In some sense, recent methodological developments in the history of science have brought the context distinction back in a new form: Criticism arose against accounts of histories following scientist’s ideas about the alleged process and progress of science too closely.32 As a result, large portions of scholarship in the history of science were chastised because they reified a notion of advancement of knowledge that identifies dead branches of former scientific error and failures in the grown solid tree of secure knowledge. In recent years, though, cultural and social historians of science programmatically declared that they would no longer follow narratives of progress nor presuppose judgments on the adequacy of theories and practices from the present epistemological standpoint.33 This is one possible answer to the question of what to do with justification in the history of science: simply historicize justification completely, and do not even think about who’s epistemologically—or morally—right or wrong.

JUSTIFYING AND DISCOVERING THE NATURE OF ECOSYSTEMS

209

By this token, justification as a historical phenomenon would just become another part of what Reichenbach subsumed under the context of discovery. So what is at stake in the relation between history and philosophy of science and the DJ distinction is the place and status of epistemological normativity.34 Philosophy of science is bluntly normative in an epistemological sense: It aims at reconstructing justifiable knowledge—nowadays this often includes the reconstruction of what exactly happened in the making of knowledge before and around the discovery. Some still restrict their perspective only to aspects connected to justification. The agenda for philosophy of science is set according to the best presently available epistemological criteria.35 At the same time and despite all the well-informed reflections about shifting standards for scientific evidence in a broad sense, the present criteria for the delineation of a justifiable body of scientific knowledge claims must necessarily be left unhistoricized, as well as otherwise decontextualized. There always remains a contingent historical context to be taken for granted, i.e., to be chosen not deliberately, but for good reasons to be explicitly regarded as part of the indispensable basis for the respective epistemological judgment. Many historians of science, in accordance with other branches of history and science studies, generally abhor open epistemological judgments just like nature was said to abhor a vacuum. Historiography of science nonetheless aims to get close to what “really” happened, at least, from a certain perspective. New perspectives include those approaches, which take experiments, instruments, practices and spaces of knowledge into account. And yet, epistemological neutrality is impossible. Since history of science is no longer a place from which to address issues of normative epistemological standpoints, it tends to involuntarily affirm the present state-of-the-art or na¨ıve relativism. Ignoring or rejecting epistemological evaluations creates problematic hidden premises. Here could be a meeting point for history and philosophy of science: We might think of it as a forum to address the respective blind spots of the normative epistemological and the historicist perspectives on science. Addressing the historical and normative aspects of both discovery and justification is mandatory for history and philosophy of science. Even though there are people in the field who are capable of combining both, complete separation and complete reduction to either one perspective still seems to be the case for large portions of current scholarship. One major goal of my paper has been to challenge the claim of agnosticism or equidistance to conflicting epistemological evaluations for history of science and science studies. Contrary to implicit or explicit claims, complete epistemological neutrality is not possible in writing about science. Both history and philosophy of science have to address the context of present state-of-the-art evaluations that necessarily shape both historical and philosophical approaches to science—something that we may call the context of metajustification.36 In this sense, in history of science normative assumptions are a hidden part of the descriptive; in philosophy of science necessarily incomplete descriptions shape the normative in mostly undisclosed ways. The main difference between the two fields is the way in which they approach the validity of the scientific knowledge claims. History of science is interested in both discovery and justification in order to understand the formative

210

THOMAS POTTHAST

mechanisms of science. Philosophy of science aims at an understanding of these mechanisms from the perspective of a separation between correct and incorrect ways of justifying the discovered mechanisms. In this sense, Reichenbach’s distinction prevails. And the question remains how historians should judge certain knowledge claims as justified and others as not justified—and how philosophers could integrate the contingencies and historical relativity of validity claims (Geltungsanspr¨uche) of knowledge.

6. AFTERWORD: A PROTO-ECOLOGY OF KNOWLEDGE?

It is still common to regard physics as the paradigmatic subject of philosophy of science. Concern has long been expressed that biology as a whole might not be a proper science from an epistemological point of view. Yet, ecological objects have played a major role in epistemology. Let me provide the full quote from Reichenbach: “It is a kind of logical signpost which we erect; for each path we give its direction together with all connected directions and leave the decision as to his route to the wanderer in the forest of knowledge” (Reichenbach 1938, p. 14). Also Bacon, Descartes, and some earlier philosophers of knowledge ploughed this metaphorical field. Some ten years ago, the practitioner, historian and philosopher of ecology, Frank Golley from Atlanta, Georgia, stated the following: “[Animate as well as inanimate; t.p.] nature may be organized [. . . ]. Several paths through this jungle of complexity, variability, and multiple causation were invented by ecologists” (Golley 1993, p. 22). Lots of trees standing together seem to be of significance for describing epistemological issues. Can we infer from the choice of words that nature is the wild jungle through which first ecologists invent (!) paths for crossing it? And do epistemologists erect their signposts in an already tamed jungle, a jungle transformed into a forest of knowledge? Moreover, what about the historians? Do they cut all this wood in order to infer from the annual growth rings the dendrochronology of science? To push this further: Why does Nature provide such powerful metaphors for epistemology and how does this shape its contents? Can and do the analytic tools exhibit the same inherent structure as the objects themselves? Should we construct both the objects of natural science as well as its epistemology and historiography on the basis of Selbst¨ahnlichkeit? This puzzling question shall be left to further research in the history and philosophy of science.

ACKNOWLEDGMENTS

I would like to thank the editors and the other contributors to this volume for valuable discussions. Financial support at different stages of this paper came from the Max Planck Institute for the History of Science, the Deutsche Forschungsgemeinschaft (DFG), and the Alexander von Humboldt Foundation.

JUSTIFYING AND DISCOVERING THE NATURE OF ECOSYSTEMS

211

NOTES 1. Reichenbach 1938; cf. Nickles 1980; Hoyningen-Huene 1987, and, not least, the contributions to this volume. 2. Don Howard (this volume) provides a history of philosophy account of the historical context of Reichenbach’s DJ distinction; Alan Richardson (this volume) points out the importance of the distinction addressing contexts of and for justification. Both authors point to the underlying agenda of epistemological and political responsibility of the scientific philosopher. For the purpose of this paper, I confine myself to non-contextual reading. 3. Explicitly mentioned are sociology, psychology, and epistemology. History of science is not named but the three disciplines stand here as different methodological approaches (also) dealing with past and present science. 4. According to Hoyningen-Huene (this volume), the disciplinary separation is one of the four to five possible meanings of the DJ: (1) temporal, (2) process of discovery vs. method of justification, (3) empirical vs. logical analysis, (4) disciplinary distinction resulting from the former two versions, (5) distinction of questions. At least in Reichenbach’s book, however, the two contexts do not completely separate the whole set of disciplines; see below. 5. Reichenbach 1938, p. 6 mentions critically scientific publications not purged from elements of the context of discovery by the (discovering) authors who are making their own history. 6. See Reichenbach 1938, p. 8ff.; for detailed analyses cf. Richardson, (this volume) and Schiemann, (this volume). 7. I am not sure whether the wording is just sloppy or somewhat ironic. Of course, the “discovery” meant here follows the completely logical way of epistemological analysis. 8. I agree with Don Howard’s analysis of Reichenbach de-politicizing the possible advisory role of epistemology, especially when understood as being opposed to Otto Neurath’s view (Howard, this volume). 9. Hoyingen-Huene (this volume) suggests a common denominator of a “lean version” of the DJ distinction that solely comprises the difference between descriptive and normative. However, what is at stake is the (diverging and contested view of the) exact way how this separation comes into (counter)force in philosophy and history of science. 10. And it has continued well up to the present; cf. Peters 1992; Haila 1997; Ford 2000. 11. “Paradigmatism” seems to be a highly contagious Kuhnian disease among ecologists and conservation biologists, cf. the accounts of Simberloff 1980; Trepl 1987; Pickett et al. 1992. 12. Cf. for example Mayr 1988 and Beatty 1995 for evolutionary biology as well as Peters 1992 and Weber 1999 for ecology. 13. “Epistemic objects” or “epistemic things” have a specific meaning as proposed by Rheinberger 1997. I use the term here in a broader sense acknowledging the danger of inflationary usage of both “discovery” and “epistemic objects”. I also acknowledge that different words usage is not at all a peripheral or negligible issue in science and science studies. But this is not the topic of this paper. Arabatzis (this volume) has provided a stimulating terminological scheme of “discovery”, “construction”, and “generation.” 14. For details see section 4 below. 15. Cf. Leps 1998 and section 4 below. 16. Cf. Potthast 1999 for details of the epistemological impediments to a unification of evolutionary and ecosystem theory. 17. See also the ferocious account of the epistemic-moral dimension by Worster 1994. 18. Hagen points out that ideas of ecological self-regulation can be traced back to earlier outlines in biogeochemistry from the 1910s as well as to population ecology. G. Evelyn Hutchinson seemed to be the person successfully getting these strains together, influencing Raymond Lindeman and later Eugene Odum as his students at Yale. 19. To be sure, this is no (debate on) naive realism going on; it is just to say that ecosystem ecology does refer to material objects as empirically traceable units.

212

THOMAS POTTHAST

20. “Tansley did not clarify whether he considered an ecosystem to be an object of nature or something else” (Golley1993, p. 34). Hagen rather refers to the methodological constructivist notion of Tansley. See below for further interpretations. 21. Jax also provided a book-length study on units of ecology, including the mentioned issues (Jax 2002). 22. Jax is right in his observations. However, the absence of such a discussion does not preclude that science policy decisions after 1945 did play a role regarding—to say the least—somewhat peculiar connotations of Lebensraum & Lebensgemeinschaft (Potthast 2001). 23. Jax 2002, p. 1006 also points to this contradictory claim but maintains his anti-realist epistemological and ontological interpretation of Tansley’s original intention. 24. Golley and Hagen acknowledge this point rather implicitly by framing their narrative; see above. 25. Stating their match with physical reality, Tansley mentions the “mental isolate” only in the context of pointing out that different hierarchical systems of the earth are not completely separate from each other—rather a truism today. 26. This has changed in the new millennium by moving away from units as the major approach to ecology in the succeeding textbook of the same authors (Begon et. al. 2000). 27. There is ample evidence of such tendencies on the progressive as well as the conservative sides (Anker 2001; Potthast 2001). However, the political reification of the evolutionary or ecological individual definitely can also have devastating effects on the concepts human social life (Potthast 1999; Potthast 2004). 28. According to Nickles 1980 this is the only one Reichenbach “really” intended. 29. It may have never properly worked, notwithstanding scientists, philosophers, high-school teachers, and media people who still maintain this view. The impact of Popperianism, especially among scientists, on that matter cannot be overestimated. Scientists also deal with the context of discovery, although mainly in introductions to reviews, anecdotally, or in (auto)biographic pieces. 30. Several papers of this volume approach this issue from rather different perspectives and with different conclusions. 31. It seems that the problem of logical positivism and its predecessors finding firm ground for issues of justification was one of the main driving forces pushing philosophy of science back to matters of discovery (Nickles 1980, p. 3), among many others. 32. Scientists providing full-fledged histories of their own fields of research have also been targets of this criticism. 33. In doing so, they de facto tend to affirm (the most powerful) orthodoxy due to a lack of critical epistemological perspectives. It is precisely therefore that many recent science studies—contrary to their own lip-service as well as to science warriors like Sokal and Briquemont—get along so well with their objects of study. Normative epistemology has been abandoned. 34. Moral normativity is another major issue for both history and philosophy of science. But this is beyond the scope of this paper, cf. Potthast 2000; Potthast 2001. 35. I am not addressing truth-talk or any approximation or incommensurability theses here. The mentioned issues can be and are raised within realist, pragmatist and constructivist frameworks. 36. This is a preliminary term for further critical discussion, especially with regard to already existing multiplicities of DJ distinctions.

REFERENCES Anker, Peder J. (2001), Imperial ecology. Environmental order in the British Empire, 1895–1945. Cambridge, MA and Landon: Harvard University Press. Anker, Peder J. (2002), “The context of ecosystem theory,” Ecosystems 5: 611–613. Beatty John (1995), “The evolutionary contingency hypothesis,” in G. Wolters, J. G. Lennox (eds.), Concepts, theories, and rationality in the biological sciences (Pittsburgh: University of Pittsburgh Press), pp. 45–81.

JUSTIFYING AND DISCOVERING THE NATURE OF ECOSYSTEMS

213

Begon, Michael, John L. Harper and Colin R. Townsend (1996), Ecology—Individuals, Populations and Communities. 3rd ed. (London: Blackwell). Begon, Michael, John L. Harper and Colin R. Townsend (2000), Essentials of Ecology (London: Blackwell). Ford, E. David (2000), Scientific Method for Ecological Research (Cambridge: Cambridge University Press). Golley, Frank B. (1993), A History of the Ecosystem Concept in Ecology. More Than the Sum of Parts (New Haven, CT: Yale University Press). Hagen, Joel B. (1992), An Entangled Bank. The Origins of Ecosystem Ecology (New Brunswick, NJ: Rutgers University Press). Haila, Yrj¨o (1997), “Trivialization of critique in ecology” Biology and Philosophy 12: 109–118. Hoyningen-Huene, Paul (1987), “Context of discovery and context of justification,” Studies in History and Philosophy of Science 18(4): 501–515. Jax, Kurt (1998), “Holocoen and ecosystem. On the origin and historical consequences of two concepts,” Journal of the History of Biology 31(1): 113–142. ¨ Jax, Kurt (2002), Die Einheiten der Okologie. Analyse, Methodenentwicklung und Anwendung, Theorie ¨ in der Okologie Vol. 5, (Frankfurt a.M.: Peter Lang). ¨ ¨ Leps, G¨unther (1998), “Okologie und Okosystemforschung,” in I. Jahn (ed.), Geschichte der Biologie. Theorien, Methoden, Institutionen, Kurzbiographien. 3. neubearbeitete und erweiterte Auflage (Jena: Gustav Fischer), pp. 601–619. Levins, Richard (1968). Evolution in Changing Environments. Some Theoretical Explorations (Princeton: Princeton University Press). Mayr, Ernst (1988), Toward a New Philosophy of Biology. Observations of an Evolutionist (Cambridge, MA: Belknap). Machamer, Peter, Lindley Darden and Carl F. Craver (2000), “Thinking about mechanisms,” Philosophy of Science 67: 1–15. Nickles, Thomas (1980), “Introductory essay: Scientific discovery and the future of philosophy of science,” in T. Nickles (ed.), Scientific Discovery, Logic, and Rationality. Boston Studies in the Philosophy of Science Vol. 56, (Dordrecht: Reidel), pp. 1–59. Peters, Robert H. (1991), A Critique for Ecology (Cambridge: Cambridge University Press). Pickett, Steward T. A., Thomas Parker and Peggy L. Fiedler (1992), “The new paradigm in ecology: implications for conservation biology above the species level,” in: P. L. Fiedler, S. K. Jain (eds.), Conservation Biology. The Theory and Practice of Conservation, Preservation and Management (New York: Chapman and Hall), pp. 65–88. Potthast, Thomas (1996), “Inventing Biodiversity: Genetics, Evolution, and Environmental Ethics,” Biologisches Zentralblatt 115(2): 177–185. Potthast, Thomas (1999), Die Evolution und der Naturschutz. Zum Verh¨altnis von Evolutionsbiologie, ¨ Okologie und Naturethik (Campus: Frankfurt a.M. and New York). Potthast, Thomas (2000), “Bioethics and epistemic-moral hybrids: Perspectives from the history of science,” Biomedical Ethics 5(1): 20–23. Potthast, Thomas (2001), “Gef¨ahrliche Ganzheitsbetrachtung oder geeinte Wissenschaft von Leben und ¨ Umwelt? Epistemisch-moralische Hybride in der deutschen Okologie 1925–1955,“ Verhandlungen zur Geschichte und Theorie der Biologie 7: 91–114. Potthast, Thomas (2001), “Die wahre Natur ist Ver¨anderung. Zur Ikonoklastik des o¨ kologischen Gleichgewichts,“ in L. Fischer (ed.), Projektionsfl¨ache Natur (Hamburg: Hamburg University Press), pp. 194–223. ¨ Reich, Michael and Volker Grimm (1996), “Das Metapopulationskonzept in Okologie und Naturschutz: ¨ Eine kritische Bestandsaufnahme,“ Zeitschrift f¨ur Okologie und Naturschutz 5: 123–139. Reichenbach, Hans (1938), Experience and prediction. An analysis of the foundations and the structure of knowledge (Chicago: University of Chicago Press). Rheinberger, Hans-J¨org (1997), Toward a history of epistemic things. Synthesizing proteins in the test tube (Stanford: Stanford University Press).

214

THOMAS POTTHAST

¨ Schwarz, Astrid E. (2003), Wasserw¨uste, Mikrokosmos, Okosystem. Eine Geschichte der ,Eroberung’ des Wasserraums (Freiburg i. Br.: Rombach). Simberloff, Daniel (1980), “A succession of paradigms in ecology: essentialism to materialism and probabilism,” Synthese 43: 3–39. Takacs, David (1996), The idea of biodiversity. Philosophies of paradise (Baltimore and London: Johns Hopkins University Press). Tansley, Arthur G. (2002), “The temporal genetic series as a means of approach to philosophy,” Presented before the Magdalen Philosophy Club of Oxford University on 5 May 1932, first publication prepared by Peder Anker, Ecosystems 5: 611–613. Taylor, Peter J. (1988), “Technocratic optimism, H. T. Odum, and the partial transformation of ecological metaphor after World War II,” Journal of the History of Biology 21(2): 213–244. ¨ Trepl, Ludwig (1987), Geschichte der Okologie. Vom 17. Jahrhundert bis zur Gegenwart (Frankfurt a.M.: Athen¨aum). Weber, Marcel (1999), “The aim and structure of ecological theory,” Philosophy of Science 66: 71–93. Worster, Donald (1994), Nature’s Economy: A History of Ecological Ideas, 2nd ed. (New York: Cambridge University Press). Worthington, Edgar B. (1975), The Evolution of IBP (International Biological Programme) (Cambridge: Cambridge University Press)

THEODORE ARABATZIS

ON THE INEXTRICABILITY OF THE CONTEXT OF DISCOVERY AND THE CONTEXT OF JUSTIFICATION

1. INTRODUCTION1

Before the historicist turn in philosophy of science, it was generally regarded that scientific activity takes place within two distinct contexts, the context of discovery and the context of justification. The former consists in the processes of generation of scientific hypotheses and theories; the latter in their testing and validation. According to Reichenbach, who codified the distinction, the context of discovery was the province of historians, psychologists, and sociologists and was not susceptible to logical analysis: “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” (Reichenbach 1951, p. 231). On the other hand, the context of justification was an area which could be rigorously explored and formalized and thus fell within the province of logic and philosophy.2 Popper introduced a very similar distinction in The Logic of Scientific Discovery (Popper 1968, p. 31). His notion of discovery, however, was different from Reichenbach’s (see note 12). This distinction3 historically derived from several premises. First, it was based on a conception of philosophy of science as a normative enterprise, i.e., an enterprise whose aim was to lay out rules that should govern any activity that deserves to be called science. Second, it was grounded on a conflation of scientific discovery with the generation of novel ideas. Thus, the study of discovery had to be the study of scientific creativity. Third, it rested on the widespread view that there are no rules whose application can enhance one’s creativity. The latter two assumptions precluded the possibility of a normative theory of discovery and, along with the first one, rendered impossible the philosophical exploration of discovery. Finally, the distinction required justification to be a rule-governed process so as to be the subject of a normative project. The ultimate aim of such a project would be to find logical rules in terms of which a “justification machine” could be constructed that would take over the justificatory practices of the scientists.4 All of these assumptions have for some time now been under attack and, consequently, the distinction has been undermined. To begin with, there has been a gradual shift towards a more “naturalistic” conception of philosophy of science, namely a conception that stresses the descriptive and hermeneutic aspects of the philosophical study of science as opposed to its normative ones (Kitcher 1992; Hoyningen-Huene, this volume). Furthermore, the conflation of discovery with generation has been 215 J. Schickore and F. Steinle (eds.), Revisiting Discovery and Justification, 215–230.
C 2006 Springer. Printed in the Netherlands.

216

THEODORE ARABATZIS

exposed and criticized. This point is crucial for my purposes and will be extensively discussed later. Even if this conflation and the concomitant identification of the study of discovery with the study of creativity were valid, one could still deny that creativity is an unanalyzable, totally mysterious phenomenon that precludes the possibility of a normative theory of discovery. Indeed, there has been overwhelming evidence that hypothesis generation and theory construction are reasoned processes whose explication can (and should) be carried out by philosophers of science.5 Some have even argued that it is possible to devise a normative theory of scientific discovery that would specify heuristic procedures which would improve the efficiency of scientific inquiry and, thus, facilitate the discovery process.6 Finally, the notion of justification as a rule-governed process has been challenged. Justification itself requires many discovery tasks. For example, to justify a hypothesis one needs to “discover” an appropriate test as well as the appropriate auxiliary statements to render the hypothesis testable (Nickles 1980a, p. 13; Nickles 1985, p. 193; Nickles 1990, p. 162; cf. also Putnam 1991). The distinction has also been undermined on different grounds. It has been argued that the kind of reasoning that is involved in generating a hypothesis is not fundamentally different from the kind of reasoning employed in justification (Achinstein 1980). Moreover, hypothesis generation and theory construction are extended problem-solving processes with many stages, each of which involves (partial) justification. At each particular stage one’s aim is to satisfy some of the constraints posed by the problem. The satisfaction of those constraints amounts to partial justification of the evolving solution (Langley et al. 1987; Nickles 1980a). Furthermore, justification in science often takes the form of heuristic appraisal, i.e., of evaluating the future problem-solving potential of a theory (Nickles 1985, p. 194; Nickles 1987, pp. 47–48; Nickles 1989a; Nickles, this volume). For someone who views discovery as an instance of problem solving, this form of theory appraisal amounts to judging the capacity of a theory to generate discoveries and, therefore, it is closely linked with discovery itself. Finally, Nickles has stressed the importance of “generative justification” or “discoverability”, namely a form of appraisal that justifies a claim by deriving it from already established knowledge. Justification in this case amounts to specifying a rationally reconstructed (not necessarily the actual) discovery path (Nickles 1984; Nickles 1985, pp. 194–195; Nickles 1988, p. 394; Nickles, this volume). Thus, it is reasonably established that justification and discovery are much more closely related than formerly thought. The distinction has also been attacked from a historical perspective. Thomas Kuhn, for instance, has argued that Considerations relevant to the context of discovery are ... relevant to justification as well; scientists who share the concerns and sensibilities of the individual who discovers a new theory are ipso facto likely to appear disproportionately frequently among that theory’s first supporters.7

However, many of the critics of the distinction continue to share with its proponents the same conception of scientific discovery. The context of scientific discovery, on this view, consists in the processes which lead to the formulation of new hypotheses

THE CONTEXT OF DISCOVERY AND THE CONTEXT OF JUSTIFICATION

217

or theories. In other words, both sides of the debate equate discovery either with the “generation”8 of hypotheses or with the construction of scientific theories.9 With some notable exceptions, justification is still not seen to be part of the discovery process.10 In this paper, I argue that this view of scientific discovery is misleading and without it the debate on the validity of the distinction between the two contexts could not even start. The focus of my discussion will be the discovery of unobservable entities and the discovery of new phenomena, using as examples the discovery of the electron and the discovery of the Zeeman effect. 2. THE INEXTRICABILITY OF THE TWO CONTEXTS

The term “discovery” is used to designate many different kinds of processes: the discovery of phenomena through controlled experiment (e.g., the discovery of the Zeeman effect—the magnetic splitting of spectral lines); the discovery of entities which are accessible to immediate inspection (e.g., the discovery of a previously unknown species); the discovery of objects which are not accessible to unaided observation (e.g., the discovery of the planet Neptune); the discovery of entities which are unobservable in principle (e.g., the discovery of the electron); the discovery of new properties of well established entities (e.g., the discovery of electron spin); the discovery of new principles/laws (e.g., the discovery of energy conservation); and the discovery of new theories (e.g., the discovery of the special theory of relativity).11 In all of these cases the two contexts are inextricably linked. Consider, for example, the discovery of unobservable entities. An individual or a group can acquire the status of “the discoverer” only after they have convinced the rest of the scientific community of the existence of the entity in question. A mere hypothesis to the effect that a new entity exists would not qualify as a discovery of that entity. The justification of that hypothesis would be a constitutive characteristic of that discovery.12 The context of discovery is “laden” with the context of justification because “discovery” is a term which refers to an epistemic achievement: if one succeeds in discovering something then, no doubt, this something exists.13 That this is the case is witnessed by the fact that in the historical literature the historiographical issue of scientific discovery has been discussed only in relation to entities that remain part of the accepted scientific ontology (e.g., oxygen). No historian or philosopher, to the best of my knowledge, has ever used the term “discovery” to characterize the proposal and acceptance of an entity (e.g., phlogiston) that we now believe was a fictitious one.14 The example of phlogiston is instructive. Contemporary historians and philosophers do not think that phlogiston was discovered, despite the fact that some 18th century chemists referred to phlogiston as one of the most significant discoveries in the history of chemistry. In Joseph Priestley’s words, phlogiston “was at one time thought to have been the greatest discovery that had ever been made in the science.”15 This suggests that there is a retrospective dimension to discovery. Only beliefs that have remained immune to revision can be designated with that term. “Discovery” is an evaluative category, which has realist presuppositions.16

218

THEODORE ARABATZIS

Despite the critical remarks that have been raised against the distinction between discovery and justification, one can still distinguish between the original historical mode of hypothesis generation and the “final” form of justification. These two aspects of the discovery process need not coincide. The actual path that led to the hypothesis (theory) in question might be “edited” out of the presentation of the hypothesis before the community.17 Furthermore, justification itself is a constantly evolving process: it is rarely the case that the justification of a hypothesis retains its original form. As science develops the justification of scientific beliefs undergoes continuous reconstruction.18 Thus, the distinction becomes a temporal, as opposed to a logical, one between two aspects of the discovery process.19 This brings me to the application of the term “discovery”. “Discovery” should not be confused with either “generation” or “construction”. Even though the terms “generation” and “construction” (or “extended generation”) do not preclude that the outcome of the corresponding processes is a true statement about nature, they do not imply it either. Furthermore, they carry the connotations of “creation”; with construction something comes into being as a result of human action. “Discovery”, on the other hand, implies truth. Moreover, it carries the connotations of “revelation”; some truth about nature is disclosed to a passive intellect (Stachel 1994, pp. 142, 146; Caneva 2001, p. 19). It should be noted that by using the term “construction” I do not thereby commit myself to the view that scientific facts are socially constructed (see below, p. 226). Much of the work carried out under this approach is strongly relativist and anti-realist. While this is not the place to take up the challenge posed by contemporary sociology of scientific practice, I should point out that viewing science as a constructive activity does not necessarily carry relativist or anti-realist implications. The neutrality of such a view vis-`a-vis the issue of relativism is shown by the fact that one might be able to specify canons of sound construction that would transcend the local practices of particular scientific groups. Moreover, it is conceivable that one could come along and show that sound constructions result in genuine facts about nature (realism). This possibility shows that constructivism, properly understood, is neutral with respect to the realism debate.20 In view of the distinction between discovery and construction, I would like to revise and extend the tentative classification of scientific discoveries that I offered above. Some of those discoveries (D) will be re-classified as constructions (C) or inventions (I). The utility of this revision will become apparent later. In the domain of application of the term “discovery” I will include individual observable entities (e.g., Neptune), observable natural kinds (e.g., tigers), and phenomena (e.g., the Zeeman effect). The term “construction” will apply to problems, problem-solutions, theories, theoretical entities (e.g., the representation of the electron), principles (energy conservation), and representations of unobservable properties (electron spin). Finally, I will use the term “invention” to characterize the proposal of novel theoretical and experimental techniques, and the creation of new instruments. All these different kinds of discovery and construction are interrelated. For instance, the construction of a problem (e.g., the incompatibility of two established scientific theories) might result in the construction of a new theory that will resolve the problem in question.

THE CONTEXT OF DISCOVERY AND THE CONTEXT OF JUSTIFICATION

219

It is worth exploring the similarities and differences between these kinds of discoveries and constructions. The question would be to determine whether the generative and justificatory procedures are similar in all cases and whether different kinds of discoveries are valued differently by the scientific community, i.e., whether they are assigned a different social status. It seems to me that the items in the above classification differ from each other in important respects. For instance, discovery cannot be identified with problem solving. The proposal of, say, phlogiston, solved several of eighteenth century chemical problems, but it does not count as a discovery. In what follows I will focus on the discovery of phenomena and unobservable entities, simply because these are the processes that have been significant for my own historical research. Starting with phenomena, their discovery involves the following circumstances: the observation of a novel situation and the construction of an argument to the effect that the observations obtained are not artifacts of the apparatus employed and that all perturbing factors (“noise”) have been eliminated. Furthermore, the validity of the argument in question must not be affected by subsequent theoretical and experimental developments. I will discuss further this issue later in connection with the discovery of the Zeeman effect. Proceeding to unobservable entities, their “discovery” can be seen as the first stage of the construction of their representation. During that stage scientists construct a representation of a novel entity, attempting to resolve particular (empirical or conceptual) problems. If the emerging representation provides an adequate solution to those problems, then this is taken as an indication that the corresponding entity exists. Thus, the “discovery” of an unobservable entity and the early phase of the construction of its representation are two aspects of a single process and cannot be sharply distinguished.21 The “discovery” ends when the “discoverers” persuade the rest of the community that the entity in question is real. Only in this qualified sense can one claim that an unobservable entity was discovered. Again, I defer a detailed discussion of this case for later. Further, the widespread view that a discovery is an isolated event that can be credited to a single individual is misconceived, at least vis-`a-vis those cases that mostly concern me here. Both the discovery of phenomena and the discovery of unobservable entities involve many complex tasks and, thus, cannot take place at a single moment. Furthermore, the discovery of unobservable entities is rarely the accomplishment of a single individual. These are Kuhnian insights and they are reinforced by the realization that the context of discovery comprises both the context of generation and the context of justification (Kuhn 1970, pp. 52–65; Kuhn 1977, pp. 165–177). I think, however, that Kuhn’s claim that discoveries of phenomena, which could not be predicted from accepted theory, cannot be attributed to particular individuals is not, in general, true. The criteria that enable us to claim that such a discovery has been accomplished are the criteria that are involved in judging the reliability of the experiment that exhibits the new phenomenon. Regardless of whether the phenomenon can be given a theoretical explanation, the experimental result, along with the demonstration of its validity (usually based on experimental background knowledge) constitute the

220

THEODORE ARABATZIS

occurrence of a discovery. Both of these achievements might be the product of a single scientist. One might not want to use the term “discovery” to characterize the products of scientific activity, but undoubtedly discovery occupies a central place in the scientists’ own image of their enterprise. Discoveries are seen as the units of scientific progress and are accordingly valued. Usually they are constructed in the light of knowledge that was not available to the historical actors at the time when the presumed discovery took place, and are intimately tied to the reward structure of science. They tend to be post-hoc reconstructions of specific episodes, whose aim is to propagate and reward certain practices and beliefs that are deemed significant for contemporary scientific activity (Schaffer 1986). The “discovery” of the electron provides a good example of what I have in mind. It is a discovery that supposedly took place in 1897 and was the exclusive achievement of J. J. Thomson. As I will argue below, neither of these claims can stand the test of historical scrutiny. This poses an interesting historiographical problem vis-`a-vis the aims and function of the retrospective construction of that discovery. This problem, in turn, suggests that the construction and continuous reconstruction of scientific discoveries can be fruitfully studied from a sociological perspective (Brannigan 1981; Schaffer 1986; Caneva 2001). The study of discovery transcends both the psychological exploration of scientific creativity and the philosophical analysis of scientific justification, since in many cases discoveries serve specific purposes within the scientific community. The psychological, philosophical, and sociological approaches to the study of discovery are complementary and should not be undertaken at the expense of each other (Nersessian 1993; Stachel 1994, p. 142). In what follows I will provide concrete illustrations of these historiographical and philosophical issues, by examining the discovery of the Zeeman effect and the discovery of the electron.22 3. ON THE DISCOVERY OF PHENOMENA: THE CASE OF THE ZEEMAN EFFECT

It was known since the middle of the nineteenth century that there was a close connection between magnetism and light. In the early 1890s a Dutch physicist, Pieter Zeeman (1856–1943), attempted to detect the influence of a magnetic field on the spectrum of a sodium flame. After several unsuccessful attempts he managed to demonstrate the effect in question (Zeeman 1896). He placed the flame of a Bunsen burner between the poles of an electromagnet and held a piece of asbestos impregnated with common salt in the flame. After turning on the electromagnet, the two D-lines of the sodium spectrum, which had been previously narrow and sharply defined, were clearly widened. In shutting off the current the lines returned to their former condition. Zeeman was not convinced that the observed widening was due to the action of the magnetic field directly upon the emitted light. The effect could be caused by an increase of the radiating substance’s density and temperature. Since the magnet caused an alteration of the flame’s shape, a subsequent change of the flame’s

THE CONTEXT OF DISCOVERY AND THE CONTEXT OF JUSTIFICATION Magnetic poles

221

Direction of the tube’s axis

Direction of the magnetic field

Figure 1.

temperature and density was also possible. Therefore, Zeeman tried another more complicated experiment. He put a porcelain tube horizontally between the poles of the electromagnet, with the tube’s axis perpendicular to the direction of the magnetic field (see figure 1). A piece of sodium was introduced into the tube and simultaneously the tube’s temperature was raised by the Bunsen burner. At the same time the light of an electric lamp was guided by a metallic mirror to traverse the entire tube. In the next stage of the experiment the sodium, under the action of the Bunsen flame, began to gasify. The absorption spectrum was obtained by means of a Rowland grating. Finally the two sharp D-lines of sodium were observed. By activating the electromagnet the lines became broader and darker. When it was turned off the lines recovered their initial form. Zeeman’s experimental scruples were, nonetheless, not satisfied. Remember that the experiment’s purpose was to demonstrate the direct effect of magnetism on light. Zeeman was still skeptical about whether this aim had been accomplished. The different temperature in the upper and lower parts of the tube resulted in a heterogeneity of the vapor’s density. The vapor was denser at the top of the tube and, since their width at a certain height depended on the number of incandescent particles at that height, the spectral lines were therefore thicker at the top. It was conceivable that the presence of a magnetic field could give rise to differences of pressure in the tube of the same order of magnitude and in the opposite direction to those produced by the differences of temperature. If this were the case, the action of magnetism would move the denser layers of vapor toward the bottom of the tube and would alter in this way the width of spectral lines, without interacting directly with the light that generated the spectrum. To eliminate this possibility, Zeeman performed an even more refined experiment. He used a smaller tube and heated it with a blowpipe in order to eliminate disturbing temperature differences. Moreover, he rotated the tube around its axis and thus achieved equal densities of sodium vapor at all heights. The D-lines were now uniformly wide along their whole length. The subsequent activation of the electromagnet resulted in their uniform broadening. Zeeman was by then nearly convinced that the

222

THEODORE ARABATZIS

outcome of his experiments was due to the influence of magnetism directly upon the light emitted or absorbed by sodium. The originality and ingenuity of Zeeman’s experiments consisted in the elaborate and sophisticated methods that he used in order to eliminate background “noise” and thus establish the direct relationship between the observed effect and the action of magnetism. The change of the width of the spectral lines after the activation of the electromagnet was not by itself an indisputable demonstration of a direct interaction between magnetism and light. As we have seen, certain intermediate links, interposed between the generation of a magnetic field and its effect on the spectrum of a substance, could have explained the experiment’s result and thus prevented Zeeman from postulating a direct causal connection between magnetism and light. Zeeman’s significant achievement was in the elimination of all these potentially existing links.23 Zeeman’s experimental work indicates that the role of background knowledge in experimental practice, along with the all-pervasive “noise” render experimentation a very complex process. Part of the experimenter’s task is to employ background knowledge in order to eliminate the all-pervasive “noise”, a task that requires a very subtle form of “experimental” reasoning. The display of this reasoning in the narration of experimental discoveries amounts to the construction of an argument for the validity and significance of the reported experimental results. In Zeeman’s case, his strategy in eliminating potentially distorting features of his experimental situation depended on already established experimental knowledge. The reasoning behind this strategy was displayed in the initial report of his discovery to persuade his audience that his experimental results revealed the direct influence of magnetism on light. I have already suggested that discoveries of new phenomena involve the observation of a novel situation and the construction of an argument to the effect that the observations obtained are not artifacts of the apparatus employed and that all perturbing factors (“noise”) have been eliminated. A further requirement was that the validity of the argument in question must not be affected by subsequent theoretical and experimental developments. All of those criteria are met in the episode that I have sketched. Zeeman’s novel observations were supported by considerable argumentation that aimed at establishing their validity. Furthermore, his “experimental” reasoning was not based on the high-level electromagnetic theory (Lorentz’s theory of “ions”) that was employed to explain the results he had obtained. Thus, the subsequent abandonment of that theory was not detrimental for the validity of his arguments. It was the robustness of those arguments that rendered the Zeeman effect a stable part of the experimental setting for several subsequent developments in the theory of atomic structure. I hope it has become evident that, with respect to the discovery of phenomena, justification is an essential part of the discovery process. In discoveries of this kind, the context of discovery and the context of justification are inextricably linked. In the rest of this paper, I will try to show that the same is true of the discovery of unobservable entities. I will illustrate my arguments by reference to the discovery of the electron.

THE CONTEXT OF DISCOVERY AND THE CONTEXT OF JUSTIFICATION

223

4. ON THE DISCOVERY OF UNOBSERVABLE ENTITIES: THE CASE OF THE ELECTRON24

In order to identify an event or a process as the discovery of an unobservable entity, one needs a criterion (or a set of criteria) that would enable one to say that such a discovery has taken place. Several possible stances to the problem of what constitutes a discovery of this kind can be adopted. Two of those possibilities, that I will examine here, depend on one’s position on the debate on scientific realism, a salient aspect of which concerns the grounds that we have for believing in the reality of the unobservable entities postulated by science (electrons, protons, fields, etc.). First, one might favor an anti-realist stance, i.e., maintain that one has to be at least agnostic with respect to the existence of unobservable entities. From such a point of view discoveries of unobservables never take place. To quote from an eminent contemporary representative of this approach, “scientific activity is one of construction rather than discovery: construction of models that must be adequate to the phenomena, and not discovery of truth concerning the unobservable” (van Fraassen 1980, p. 5). On this stance, “discovery” has nothing to do with truth. Rather, it is a process of constructing empirically adequate models. The unobservable entity is a convenient fiction. To put it in terms of the discovery / justification distinction, existence claims concerning the unobservable can never be sufficiently justified. On the second (realist) stance, one might propose certain epistemological criteria whose satisfaction would provide adequate grounds for believing in the existence of a particular entity. From this point of view a discovery takes place when an individual or a group has managed to meet the required criteria. As an example consider Ian Hacking’s proposal that a belief in the reality of an, in principle, unobservable entity is justified to the extent that the entity in question can be manipulated (Hacking 1983, pp. 262–266). It follows then that an unobservable entity has been discovered only if a scientist has found a way to manipulate this entity. Justification is considered an essential aspect of the discovery process and is identified with manipulability. It is evident that the adequacy of the proposed way for deciding when something qualifies as a genuine discovery depends on the adequacy of the epistemological criteria for what constitutes unobservable reality. Any difficulties that might plague the latter would cast doubt on the adequacy of the former. Although this approach to the issue of scientific discovery can be, in principle, realized, no adequate proposal of the kind outlined has been made so far. That is, no epistemological criteria have been formulated whose satisfaction would amount to an existence-proof of an unobservable entity. In particular, Hacking’s proposal that manipulability provides such a proof leaves much to be desired. Let us examine the merits and limitations of Hacking’s view vis-`a-vis the discovery of the electron. The historian Isobel Falconer has employed Hacking’s criterion of what constitutes unobservable reality in an attempt to justify the attribution of the electron’s discovery to J.J. Thomson (Falconer 1987). She challenged traditional interpretations of Thomson’s discovery that portrayed

224

THEODORE ARABATZIS this discovery . . . [as] the outcome of a concern with the nature of cathode rays which had occupied Thomson since 1881 and had shaped the course of his experiments during the period 1881–1897 (Ibid., p. 241).

Instead she argued that “[a]n examination of his work shows that he paid scant attention to cathode rays until late 1896” (Ibid.). Furthermore, [t]he cathode ray experiments in 1897 were not the origin of the corpuscle [which has been re-named electron] hypothesis; instead they acted as a focus around which Thomson synthesized ideas he had previously developed (Ibid., p. 255).

However, she did not deny a central presupposition of the traditional view, namely that the discovery of the electron was a temporally non-extended event which can be credited to a single individual. Even though the “corpuscle hypothesis” did not originate with Thomson’s experiments with cathode rays, the discovery of corpuscles (i.e., the experimental demonstration of their existence) was the outcome of these experiments. Arriving at the theoretical concept of the electron was not much of a problem in 1897. Numerous such ideas were “in the air”. What Thomson achieved was to demonstrate their validity experimentally. Regardless of his own commitments and intentions, it was Thomson who began to make the electron “real” in Hacking’s sense of the word . . . . He pinpointed an experimental phenomenon in which electrons could be identified and methods by which they could be isolated, measured and manipulated. This was immensely significant for the development of the electron theory which hitherto has been an abstract mathematical hypothesis but now became an empirical reality (Ibid., p. 276).

In terms of the methodological issues discussed above, Falconer attempts to reduce the discovery process to the precise moment when experimental verification took place, thus equating discovery with the ability to isolate, measure, and manipulate. From my perspective this amounts to equating discovery with justification. If, however, as I have argued, the context of discovery comprises both the context of generation and the context of justification, then the physicists who had formulated all those “ideas in the air” should also be considered as having taken part in the discovery of the electron. Furthermore, Thomson was not the only one who could manipulate electrons. All those who experimented with cathode rays were able to manipulate them in various ways. For example, they could deflect them by means of magnetic fields. That is, from our perspective, given that they manipulated cathode rays and that cathode rays are streams of electrons, it follows that they manipulated electrons. And this brings me back to Hacking’s criterion of what constitutes unobservable reality. To see the limitations of this criterion consider Thomson’s experiments with cathode rays. Since one could describe these experiments in terms of cathode rays as opposed to electrons, the act of manipulation could be described without even mentioning the entities that, according to present-day physics, were manipulated. Moreover, an antirealist could give an even less theory-laden description, by avoiding the term “cathode rays” and using instead the phenomenological expression “spot on a phosphorescent screen.” The only thing that we know, the anti-realist would argue, is that by activating an electromagnet Thomson could move a spot on a phosphorescent screen. Since an act of manipulation can be described without mentioning the unobservable entity that is

THE CONTEXT OF DISCOVERY AND THE CONTEXT OF JUSTIFICATION

225

(supposedly) manipulated, this act does not, by itself, imply the existence of the entity in question. Thus, given that experiments can be (re)described in phenomenological terms,25 manipulability cannot be employed, to the satisfaction of an anti-realist, for existential inferences. Whereas for Hacking manipulability justifies existence claims, for the anti-realist it is the other way around: It is the belief in the existence of, e.g., electrons, prior to the act of manipulation, that allows us to interpret that act as a manipulation of electrons (as opposed to something else).26 Since Hacking’s criterion does not provide adequate grounds for a realist position towards unobservable entities, it cannot be employed to justify discovery claims. Thus, Falconer’s claim that the discovery of the electron was Thomson’s exclusive experimental achievement is undermined.27 Besides offering a satisfactory account of the justification of existence claims concerning the unobservable, the “friends of discovery” (Nickles’ expression) should tackle two related problems, what I will call the problem of knowledge and the problem of identification. The problem of knowledge: Kuhn formulated the problem very succinctly, in relation to the discovery of oxygen: “Apparently to discover something one must also be aware of the discovery and know as well what it is that one has discovered. But, that being the case, how much must one know?” (Kuhn 1977, p. 170). Any entity that forms part of the accepted ontology of contemporary science is endowed with several properties. The electron, for instance, has a given mass, a certain charge, an intrinsic magnetic disposition (spin), a dual nature (particle versus wave), and many other features. The question then arises, how many properties must one have discovered in order to be granted the status of the discoverer of the entity in question? To give an example, Laszlo Tisza, a very well-known physicist, suggested to me that the electron was discovered in the late 1920s when C. J. Davisson & L. H. Germer and G.P. Thomson detected its wave properties. From the measurement of the electron’s wavelength it became possible to calculate its momentum. That, according to Tisza, rendered electrons directly detectable.28 Another aspect of the problem of knowledge concerns mistaken beliefs. If knowing what one has discovered is a prerequisite for being credited with the discovery, then can one be considered the discoverer of, e.g., the electron even though he entertained wrong beliefs about it? For instance, in 1897 J. J. Thomson thought of his corpuscle, an entity later identified with the electron, as a structure in the ether. Leading physicists at the time (e.g., Joseph Larmor and H. A. Lorentz) entertained similar “wrong” conceptions of the electron. To put the problem in terms of the discovery—justification distinction, how many beliefs about an entity should be justified for the entity in question to be (considered) discovered? The problem of identification: If most, or even some, of the beliefs that the putative “discoverer” had about the “discovered” entity are wrong, it is not at all evident that the entity in question is the same with its contemporary counterpart. It has to be shown, for instance, that Thomson’s “corpuscles”, which were conceived as classical particles and structures in the ether, can be identified with contemporary “electrons”, which are endowed with quantum numbers, wave-particle duality, indeterminate

226

THEODORE ARABATZIS

position-momentum, etc. The “friends of discovery” should propose some criteria that enable us to identify the original entity with its present counterpart. I have attempted elsewhere to come to terms with the problem of identification, for reasons not directly related to scientific discovery (Arabatzis 2006). Nevertheless, I will sketch below a more neutral approach to the discovery of unobservable entities, which has the advantage of avoiding that problem altogether. Because of these problems I would be extremely reluctant to base a historical narrative about the electron (or any other unobservable entity, for that matter) on the traditional, realist notion of scientific discovery. Moreover, this notion is often an obstacle to historical understanding. Consider, for example, the case of energy conservation. Several parallel developments, from the study of steam engines to theoretical mechanics to physiology, contributed to the formulation of that principle. Until recently, historians portrayed those developments as “simultaneous discoveries”. This view, however, has been plausibly challenged, because all those putative discoverers of energy conservation were concerned with different problems and came up with different theoretical hypotheses. It was only in the 1850s that these different approaches were reinterpreted as aspects of the same discovery (Smith 1999). Besides, there is a straightforward alternative that avoids these problems. One should simply try to historicize the notion of scientific discovery, by adopting the perspective of the relevant historical actors, without worrying whether that perspective can be justified philosophically.29 On this approach, one would first examine the context of generation, that is, show how a novel concept denoting an unobservable entity was introduced into the scientific literature. Then one would reconstruct the original context of justification, consisting of all the experimental and theoretical arguments that were given in favor of the existence of the entity in question. The next step would be to trace the developmental process that followed that initial stage and gradually transformed the corresponding concept. The evolution of any such concept resembles a process of gradual construction, which takes place in several stages. A realist might want to label the first stage of that process “the stage of discovery”. In that case discovery should be construed as a gradual process of consensus formation within the scientific community whose outcome is the acceptance of an existence claim (e.g., “the electron exists”) (Caneva 2001, p. 19; Stachel 1994, p. 143). This appeal to the consensus of the scientific community should not be interpreted as a social constructivist position. My approach is constructivist, in the sense that the representation of, say, the electron was gradually constructed. However, it is not a (social) constructivist one in the usual sense, which implies that consensus within the scientific community is the outcome of professional interests, the distribution of power within the scientific community, etc. 5. CONCLUDING REMARKS

I have argued that the context of discovery, if we want to retain this expression, concerns an extended process, which involves both generation and justification. In view of the inextricability of discovery and justification, what can we conclude about the

THE CONTEXT OF DISCOVERY AND THE CONTEXT OF JUSTIFICATION

227

DJ distinction? I think that my analysis undermines versions 1–4 of that distinction, as distinguished by Hoyningen-Huene. Take, for instance, version 4: the DJ distinction as an expression of the division of labour between history of science (and related “empirical” disciplines) and (“logical”) philosophy of science. I hope to have shown that an adequate understanding of the process of scientific discovery (including justification) requires an integrated historical and philosophical approach. Understanding scientific discovery is not just an empirical task. The, apparently innocent, question “when and by whom was something discovered?” is not merely a request for factual information, but requires conceptual analysis. And conceptual (not merely logical) analysis is the hallmark of philosophy. What about version 5 and the related “lean” version, advocated by HoyningenHuene? There seems to be a difference between a factual and an evaluative perspective towards scientific knowledge. It is one thing to understand how a scientific claim was generated and accepted and another to ask whether it is justified, in light of the available evidence. I wouldn’t object to this version, provided “that facts have normative presuppositions” (Hoyningen-Huene, this volume). In particular, the descriptive statement “X discovered Y” embodies an evaluative judgment, namely that the evidence presented by X demonstrated Y’s existence. It is this confluence of the descriptive and the evaluative that makes fruitful, or even indispensable, a joint HPS approach to scientific discovery.

ACKNOWLEDGMENTS

I would like to thank the other contributors to this volume, and especially the editors, Friedrich Steinle and Jutta Schickore, for their thoughtful and constructive comments.

NOTES 1. This paper draws on material I have presented elsewhere. See my Representing Electrons: A Biographical Approach to Theoretical Entities (Arabatzis 2006). 2. Reichenbach 1938 is usually cited as the primary site of the DJ distinction. However, as Nickles has pointed out, the distinction found there “is merely one between scientific activity itself and that activity as logically reconstructed” (see Nickles 1980a, p. 12). The original context in which Reichenbach put forward the DJ distinction and his evolving take on it are explored in Howard, this volume, Richardson, this volume, and Schiemann, this volume. 3. As Hoyningen-Huene points out (this volume), the DJ distinction, as presented above, has various aspects (“versions”). What I say below undermines all of those versions, with the possible exception of version five and the “lean” one advocated by Hoyningen-Huene; but more on this at the concluding section of this essay. 4. Here I am paraphrasing Reichenbach. 5. See, for instance, Nersessian 1992; Nickles 1980b; Hoyningen-Huene, this volume. 6. See, for instance, Langley et al. 1987. 7. Kuhn 1977, p. 328. For a detailed analysis of Kuhn’s criticism of the DJ distinction see HoyningenHuene 1987, pp. 508–509; Hoyningen-Huene, this volume; Sturm & Gigerenzer, this volume. 8. I borrow the term from Nickles 1980a. 9. See, for instance, Burian 1980, pp. 322–323; Curd 1980, pp. 201–202; Laudan 1980, pp. 174–175.

228

THEODORE ARABATZIS

10. The exceptions are important. See, for instance, Gutting 1980; Hoyningen-Huene 1987; Koertge 1982; Kordig 1978; McMullin 1980. 11. Reichenbach neglected the variety of scientific discoveries and used the term “discovery” only in connection with scientific theories (Reichenbach 1938, p. 7). Thus, it is not surprising that he conflated discovery with the process of developing new theories. 12. As Whewell recognized, a “happy guess” does not constitute a discovery (Schickore, this volume). Popper made a similar point, when he referred to the “tests whereby the inspiration may be discovered to be a discovery, or become known to be knowledge.” (Popper 1968, p. 31) 13. This idea has also been put forward by Nickles 1980a, p. 9. Nickles, in turn, credits Ryle (Ryle 1949, pp. 303–304). 14. A possible exception is Langley et al. 1987. 15. Quoted in Conant 1957, p. 13. 16. Cf. Potthast, this volume. It is true that the mere use of a term (or statement) does not necessarily commit its user to its presuppositions. For instance, one may say that “the sun rose today at 6:00 am”, without believing that the sun actually moved. Thus, it is possible to use the term “discovery” in a weaker, non-realist sense. In section four, below, I attempt to show how this can be done with respect to discoveries concerning unobservable entities. 17. Cf. Friedrich Steinle’s distinction between the private and the public aspects of scientific activity (Steinle, this volume). It is interesting that this was also Reichenbach’s original version of the DJ distinction. See Reichenbach 1938, p. 6. Cf. Richardson, this volume and Schiemann, this volume. 18. The function and importance of reconstruction in science have been emphasized by Nickles 1989b. 19. Note, however, that even this version of the distinction is not free of difficulties. See HoyningenHuene this volume. 20. The connections between the discovery issue and the realism debate will be explicitly drawn below. 21. This is not peculiar to discoveries of this kind. As Steinle argues (this volume), the discovery of regularities and the formation of new concepts are also inextricable. 22. Since I have discussed elsewhere those discoveries in considerable detail, my presentation will be sketchy. For a detailed analysis I refer the interested reader to Arabatzis 1992; Arabatzis 1996. 23. Zeeman’s achievement exemplifies one of Allan Franklin’s “epistemological strategies”, which “entails the elimination of all plausible sources of error and all alternative explanations”. This strategy is part of the “arguments designed to establish, or to help establish, the validity of an experimental result or observation”. See Franklin 1989, pp. 466 and 438, respectively. 24. Even though I do not believe that unobservable entities are discovered, in the traditional sense of the term “discovery”, I will continue to use this term in this section for two reasons. First, because it is used by the proponents of views that I will be arguing against. Only after having argued against those views, I might be justified in dropping it. Second, because, as I will indicate below, the term might still be used to capture a distinction between two stages. The first stage is usually characterized by ontological debates, where the existence of an entity is contended. After that stage is over, a realist might claim that the entity has been discovered (i.e., that we know that it exists). 25. See Caneva 2001, pp. 18–19 for further examples. 26. This is just one problematic aspect of Hacking’s entity realism. For further criticism of his view see Arabatzis 2001. 27. In correspondence, Falconer has suggested to me that she agrees with my “objections to using Hacking’s criteria for reality to justify ‘discovery’”. But she still “think[s] it can be used as an analytical tool, to ask what did this or that physicist contribute to our understanding of the electron; did he help to give it manipulative reality for the physicist?” I do not have a problem with using Hacking’s manipulability criterion in this way, provided that it gives a good descriptive account of how scientists construct “existence proofs” for unobservable entities. As I have argued elsewhere, however, it also faces difficulties in this respect. See Arabatzis 2006. 28. I would like to thank Prof. Tisza for discussing with me some of the ideas I develop here. 29. Cf. Arthur Fine’s Natural Ontological Attitude (Fine 1986).

THE CONTEXT OF DISCOVERY AND THE CONTEXT OF JUSTIFICATION

229

REFERENCES Achinstein, P. (1980), “Discovery and Rule-Books,” in T. Nickles (ed.), Scientific Discovery, Logic, and Rationality (Dordrecht: Reidel), pp. 117–132. Arabatzis, T. (1992), “The Discovery of the Zeeman Effect: A Case Study of the Interplay between Theory and Experiment,” Studies in History and Philosophy of Science 23, 365–388. Arabatzis, T. (1996), “Rethinking the ‘Discovery’ of the Electron,” Studies in History and Philosophy of Modern Physics 27, 405–435. Arabatzis, T. (2001), “Can a Historian of Science be a Scientific Realist?” Philosophy of Science 68 (Proceedings), S531–S541. Arabatzis, T. (2006), Representing Electrons: A Biographical Approach to Theoretical Entities (Chicago: University of Chicago Press). Brannigan, A. (1981), The Social Basis of Scientific Discoveries (New York: Cambridge University Press). Burian, R. (1980), “Why Philosophers Should not Despair of Understanding Scientific Discovery,” in T. Nickles (ed.), Scientific Discovery, Logic, and Rationality (Dordrecht: Reidel), pp. 317–336. Caneva, K. L. (2001), The Form and Function of Scientific Discoveries, Dibner Library Lecture Series (Washington, DC: Smithsonian Institution Libraries). Conant, J. B. (1957), “The Overthrow of the Phlogiston Theory,” in J. B. Conant & L. K. Nash (eds.), Case Histories in Experimental Science (Cambridge, MA: Harvard University Press). Curd, M. V. (1980), “The Logic of Discovery: an Analysis of Three Approaches,” in T. Nickles (ed.), Scientific Discovery, Logic, and Rationality (Dordrecht: Reidel), pp. 201–219. Falconer, I. (1987), “Corpuscles, Electrons and Cathode Rays: J. J. Thomson and the ‘Discovery of the Electron’,” British Journal for the History of Science 20, 241–276. Fine, A. (1986), The Shaky Game: Einstein Realism and the Quantum Theory (Chicago: The University of Chicago Press). Franklin, A. (1989), “The Epistemology of Experiment,” in D. Gooding et al. (eds.), The Uses of Experiment: Studies in the Natural Sciences (Cambridge: Cambridge University Press), pp. 437– 460. Gutting, G. (1980), “Science as Discovery,” Revue Internationale de Philosophie 131–132, 26–48. Hacking, I. (1983), Representing and Intervening (Cambridge: Cambridge University Press). Hoyningen-Huene, P. (1987), “Context of Discovery and Context of Justification,” Studies in History and Philosophy of Science 18, 501–515. Kitcher, P. (1992), “The Naturalists Return,” The Philosophical Review 101, 53–114. Koertge, N. (1982), “Explaining Scientific Discovery,” in P. D. Asquith and T. Nickles (eds.), PSA 1982. Proceedings of the 1982 Biennial Meeting of the Philosophy of Science Association (East Lansing: Philosophy of Science Association), Vol. 1, pp. 14–28. Kordig, C. R. (1978), “Discovery and Justification,” Philosophy of Science 45, 110–117. Kuhn, T. S. (1970), The Structure of Scientific Revolutions, 2nd ed. (Chicago: The University of Chicago Press). Kuhn, T. S. (1977), The Essential Tension (Chicago and London: The University of Chicago Press). Langley, P., H. A. Simon, G. L. Bradshaw, and J. M. Zytkow (1987), Scientific Discovery: Computational Explorations of the Creative Process (Cambridge, MA: MIT Press). Laudan, L. (1980), “Why Was the Logic of Discovery Abandoned?” in T. Nickles (ed.), Scientific Discovery, Logic, and Rationality (Dordrecht: Reidel), pp. 173–183. McMullin, E. (1980), Contribution to “(Panel Discussion) The Rational Explanation of Historical Discoveries,” in T. Nickles (ed.), Scientific Discovery: Case Studies (Dordrecht: Reidel), pp. 28–33. Nersessian, N. J. (1992), “How Do Scientists Think? Capturing the Dynamics of Conceptual Change in Science,” in R. N. Giere (ed.), Cognitive Models of Science (Minneapolis: University of Minnesota Press), pp. 3–44. Nersessian, N. J. (1993), “Opening the Black Box: Cognitive Science and History of Science,” Cognitive Science Laboratory Report 53 (Princeton University).

230

THEODORE ARABATZIS

Nickles, T. (1980a), “Introductory Essay: Scientific Discovery and the Future of Philosophy of Science,” in T. Nickles (ed.), Scientific Discovery, Logic, and Rationality (Dordrecht: Reidel), pp. 1–59. Nickles, T. (1980b), “Can Scientific Constraints Be Violated Rationally?” in T. Nickles (ed.), Scientific Discovery, Logic, and Rationality (Dordrecht: Reidel), pp. 285–315. Nickles, T. (1984), “Positive Science and Discoverability,” in P. D. Asquith and P. Kitcher (eds.), PSA 1984. Proceedings of the 1984 Biennial Meeting of the Philosophy of Science Association (East Lansing: Philosophy of Science Association), Vol. 1, pp. 13–27. Nickles, T. (1985), “Beyond Divorce: Current Status of the Discovery Debate,” Philosophy of Science 52, 177–206. Nickles, T. (1987), “Twixt Method and Madness,” in N. J. Nersessian (ed.), The Process of Science (Dordrecht: Martinus Nijhoff), pp. 41–67. Nickles, T. (1988), “Truth or Consequences? Generative Versus Consequential Justification in Science,” in A. Fine and J. Leplin (eds.), PSA 1988. Proceedings of the 1988 Biennial Meeting of the Philosophy of Science Association (East Lansing: Philosophy of Science Association), Vol. 2, pp. 393–405. Nickles, T. (1989a), “Heuristic Appraisal: a proposal,” Social Epistemology 3, 175–188. Nickles, T. (1989b), “Justification and Experiment,” in D. Gooding, T. Pinch, and S. Schaffer (eds.), The Uses of Experiment (Cambridge: Cambridge University Press, 1989), pp. 299–333. Nickles, T. (1990), “Discovery,” in R. C. Olby et al. (eds.), Companion to the History of Modern Science (London and New York: Routledge), pp. 148–165. Popper, K. R. (1968), The Logic of Scientific Discovery (New York: Harper & Row). Putnam, H. (1991), “The ‘Corroboration’ of Theories,” in R. Boyd et al. (eds.), The Philosophy of Science (Cambridge, MA: MIT Press), pp. 121–137. Reichenbach, H. (1938), Experience and Prediction (Chicago: The University of Chicago Press). Reichenbach, H. (1951), The Rise of Scientific Philosophy (Berkeley and Los Angeles: University of California Press). Ryle, G. (1949), The Concept of Mind (Chicago: The University of Chicago Press, 1949). Schaffer, S. (1986), “Scientific Discoveries and the End of Natural Philosophy,” Social Studies of Science 16, 387–420. Smith, C. (1999), The Science of Energy (Chicago: University of Chicago Press). Stachel, J. (1994), “Scientific Discoveries as Historical Artifacts,” in K. Gavroglu et al. (eds.), Trends in the Historiography of Science (Dordrecht: Kluwer), pp. 139–148. van Fraassen, B. C. (1980), The Scientific Image (New York: Oxford University Press). Zeeman, P. (1986), “On the Influence of Magnetism on the Nature of the Light Emitted by a Substance (Part I),” Communications from the Physical Laboratory at the University of Leiden 33, 1–8.

CONTRIBUTORS

Arabatzis, Theodore Department of Philosophy and History of Science University of Athens Panepistimioupolis (ANO ILISIA) Athens 157 71 Greece Gigerenzer, Gerd Max Planck Institut fu¨ r Bildungsforschung Lentzeallee 94 14195 Berlin Germany Howard, Don Department of Philosophy 100 Malloy Hall University of Notre Dame Notre Dame, IN 46556 USA Hoyningen-Huene, Paul Zentrum f u¨ r Wissenschaftsethik und Wissenschaftstheorie Universit¨at Hannover Im Moore 21 30167 Hannover Germany

Peckhaus, Volker Universit¨at Paderborn Fakult¨at f u¨ r Kulturwissenschaften Institut f u¨ r Humanwissenschaften: Philosophie Warburger Str. 100 33098 Paderborn Germany Potthast, Thomas Universit¨at T u¨ bingen Interfakult¨ares Zentrum f u¨ r Ethik in den Wissenschaften (IZEW) Wilhelmstraße 19 72074 T u¨ bingen Germany Richardson, Alan Department of Philosophy 1866 Main Mall-E370 University of British Columbia Vancouver, BC V6T 1Z1 Canada Sch¨afer, Lothar Philosophisches Seminar Universit¨at Hamburg Von-Melle-Park 6 D-20146 Hamburg Germany

Nickles, Thomas Philosophy Department Mail Stop 102 University of Nevada Reno Reno, NV 89557 USA 231

232

CONTRIBUTORS

Schickore, Jutta Department of History and Philosophy of Science Goodbody Hall 130 Indiana University 1011 East Third St Bloomington, IN 47405-7005 USA Schiemann, Gregor Philosophisches Seminar Bergische Universit¨at Wuppertal Gaussstr. 20 42119 Wuppertal Germany

Steinle, Friedrich Wissenschafts- und Technikgeschichte/ Historisches Seminar Bergische Universit¨at Wuppertal Gaussstr. 20 42119 Wuppertal Germany Sturm, Thomas Max-Planck-Institut f u¨ r Wissenschaftsgeschichte Wilhelmstr. 44 D-10117 Berlin Germany

Archimedes NEW STUDIES IN THE HISTORY AND PHILOSOPHY OF SCIENCE AND TECHNOLOGY 1.

J.Z. Buchwald (ed.): Scientific Credibility and Technical Standards in 19th and Early 20th Century Germany and Britain. 1996 ISBN 0-7923-4241-0

2.

K. Gavroglu (ed.): The Sciences in the European Periphery During the Enlightenment. 1999 ISBN 0-7923-5548-2; Pb 0-7923-6562-1

3.

P. Galison and A. Roland (eds.): Atmospheric Flight in the Twentieth Century, 2000 ISBN 0-7923-6037-0; Pb 0-7923-6742-1

4.

J.M. Steele: Observations and Predictions of Eclipse Times by Early Astronomers. 2000 ISBN 0-7923-6298-5

5.

D-W. Kim: Leadership and Creativity. A History of the Cavendish Laboratory, 1871-1919. 2002 ISBN 1-4020-0475-3

6.

M. Feingold: The New Science and Jesuit Science: Seventeenth Century Perspective. 2002 ISBN 1-4020-0848-1

7.

F.L. Holmes, J. Renn, H-J. Rheinberger: Reworking the Bench. 2003 ISBN 1-4020-1039-7

8.

J. Chab´as, B.R. Goldstein: The Alfonsine Tables of Toledo. 2003 ISBN 1-4020-1572-0

9.

F.J. Dijksterhuis: Lenses and Waves. Christiaan Huygens and the Mathematical Science of Optics in the Seventeenth Century. 2004 ISBN 1-4020-2697-8

10.

L. Corry: David Hilbert and the Axiomatization of Physics (1898–1918). From Grundlagen der Geometrie to Grundlagen der Physik. 2004 ISBN 1-4020-2777-X

11.

J.Z. Buchwald and A. Franklin (eds.): Wrong for the Right Reasons. 2005 ISBN 1-4020-3047-9

12.

M. Feingold and V. Navarro-Brotons (eds.): Universities and Science in the Early Modern Period. 2006 ISBN 1-4020-3974-3

13.

R.R. Hamerla: An American Scientist on the Research Frontier. Edward Morley, Community, and Radical Ideas in Nineteenth-Century Science. 2006 ISBN 1-4020-4088-1

springer.com