“ Rehg brilliantly summarizes the major debates in philosophy of
at Saint Louis University. He is the translator of Jürgen
science over the nature of discovery, explanation, and persuasion
Habermas’s Between Facts and Norms: Contributions
that have occurred over the last half-century while also providing
to a Discourse Theory of Law and Demo�cracy (1996)
the most informed critique of Habermasian discourse theory yet to
and the coeditor of Deliberative Democracy: Essays
appear. His judicious use of real-life examples drawn from Fermilab
on Reason and Politics and Pluralism (1997) and The
and elsewhere and his extension of argument theory to include the
Pragmatic Turn: The Trans�formation of Critical Theory
social visions that frame current hot-button academic disputes
(2001), all published by the MIT Press.
regarding global warming and intelligent design make his book extremely timely and indispensable reading for laypersons trying to assess scientific journalism and reporting. Rehg’s riveting account of cogency will revolutionize the way critical theorists, logicians, and rhetoricians think about arguÂ�mentation in general and, more parÂ� ticÂ�ularly, about how we should assess the social institutionalization of collaborative research and debate in science proper for years to come.” DAVID INGRAM, Loyola University Chicago “ In bringing the disparate poles of Habermasian argumentation theory and the thickly descriptive approach of ethnomethod ology together in the form of a new ‘critical contextualism,’ Rehg has provided a realistic basis for overcoming the ‘science wars’
Studies in Contemporary German Social Thought
divide between the philosophical and the social scientific—the prescriptive and the descriptive—approaches to understanding science.€Just when ‘science studies’ has begun to explore a ‘normative turn,’ Rehg’s clearly written and rigorously argued book provides a new interdisciplinary framework and point of departure for charting the way ahead that neither practitioners of science studies, philosophers of science, nor science policymakers can afford to ignore.” LENNY MOSS, Department of Sociology and Philosophy, University of Exeter
COGENT SCIENCE IN CONTEXT
WILLIAM REHG is Associate Professor of Philosophy
The MIT Press | Massachusetts Institute of Technology | Cambridge, Massachusetts 02142 | http://mitpress.mit.edu | 978-0-262-18271-3
philosophy of science
COGENT SCIENCE IN CONTEXT
COGENT SCIENCE IN CONTEXT
The Science Wars, Argumentation Theory, and Habermas
WILLIAM REHG
Recent years have seen a series of intense, increasingly acrimonious debates over the status and legitimacy of the natural sciences. These “science wars” take place in the public arena—with current battles over evolution and global warming—and in academia, where assumptions about scientific objectivity have
The Science Wars, Argumentation Theory, and Habermas
been called into question. Given these hostilities, what makes a scientific claim merit our consideration? In Cogent Science in Context, William Rehg examines what makes scientific arguments cogent—that is, strong and convincing—and how we should assess that cogency. Drawing on the tools of argumentation theory, Rehg proposes a multidimensional, contextsensitive framework both for understanding the cogency of scientific arguments and for conducting
REHG
WILLIAM REHG
cooperative interdiscipli�nary assessments of the cogency of actual scientific arguments.
Rehg first shows how argumentation theory, with
methods for evaluating arguments that draw on disciplines ranging from logic to rhetoric, can provide an interdisciplinary lens through which to view the issues in the academic science wars. He then closely examines Jürgen Habermas’s argumentation theory and its implications for understanding cogency, applying it to a case from high-energy physics. A series of problems, however, beset Habermas’s approach. In response, Rehg outlines his own “critical contextualist” approach, which uses argumentation-theory categories in a new and more context-sensitive way inspired by ethnoÂ� graphy of science. Critical contextualism not only responds to the academic debates but also has relevance for the study of debates in the public arena, as Rehg demonstrates with a case study of National AcadÂ� emy of Sciences panels appointed to study the possible links between diet and health.
Cogent Science in Context
Studies in Contemporary German Social Thought (partial listing) Thomas McCarthy, general editor Theodor W. Adorno, Hegel: Three Studies Theodor W. Adorno, Prisms James Bohman, Democracy across Borders: From Dêmos to Dêmoi James Bohman, Public Deliberation: Pluralism, Complexity, and Democracy James Bohman and Matthias Lutz-Bachmann, editors, Perpetual Peace: Essays on Kant’s Cosmopolitan Ideal Hauke Brunkhorst, Solidarity: From Civic Friendship to a Global Legal Community Jean Cohen and Andrew Arato, Civil Society and Political Theory Maeve Cooke, Re-Presenting the Good Society Pablo De Greiff and Ciaran Cronin, editors, Global Justice and Transnational Politics: Essays on the Moral and Political Challenges of Globalization Jürgen Habermas, Between Facts and Norms: Contributions to a Discourse Theory of Law and Democracy Jürgen Habermas, The Inclusion of the Other: Studies in Political Theory Jürgen Habermas, The Liberating Power of Symbols: Philosophical Essays Jürgen Habermas, Moral Consciousness and Communicative Action Jürgen Habermas, The New Conservatism: Cultural Criticism and the Historians’ Debate Jürgen Habermas, The Philosophical Discourse of Modernity: Twelve Lectures Jürgen Habermas, The Postnational Constellation: Political Essays Jürgen Habermas, On the Pragmatics of Communication Jürgen Habermas, On the Pragmatics of Social Interaction: Preliminary Studies in the Theory of Communicative Action Jürgen Habermas, Religion and Rationality: Essays on Reason, God and Modernity Jürgen Habermas, The Structural Transformation of the Public Sphere: An Inquiry into a Category of Bourgeois Society Jürgen Habermas, Truth and Justification Joseph Heath, Communicative Action and Rational Choice Axel Honneth, The Critique of Power: Reflective Stages in a Critical Social Theory Axel Honneth, The Struggle for Recognition: The Moral Grammar of Social Conflicts Tom Huhn and Lambert Zuidervaart, editors, The Semblance of Subjectivity: Essays in Adorno’s Aesthetic Theory Elliot L. Jurist, Beyond Hegel and Nietzsche: Philosophy, Culture, and Agency Cristina Lafont, The Linguistic Turn in Hermeneutic Philosophy Jeff Malpas, Ulrich Arnswald, and Jens Kertscher, editors, Gadamer’s Century: Essays in Honor of Hans-Georg Gadamer Christoph Menke, The Sovereignty of Art: Aesthetic Negativity in Adorno and Derrida Brian O’Connor, Adorno’s Negative Dialectic: Philosophy and the Possibility of Critical Rationality Claus Offe, Modernity and the State: East, West Claus Offe, Varieties of Transition: The East European and East German Experience Kevin Olson, Reflexive Democracy: Political Equality and the Welfare State Kirk Pillow, Sublime Understanding: Aesthetic Reflection in Kant and Hegel William Rehg, Cogent Science in Context: The Science Wars, Argumentation Theory, and Habermas William Rehg and James Bohman, editors, Pluralism and the Pragmatic Turn: The Transformation of Critical Theory Albrecht Wellmer, Endgames: The Irreconcilable Nature of Modernity Rolf Wiggershaus, The Frankfurt School: Its History, Theories, and Political Significance For a full list of books in the series, please see http://mitpress.mit.edu
Cogent Science in Context The Science Wars, Argumentation Theory, and Habermas
William Rehg
The MIT Press Cambridge, Massachusetts London, England
© 2009 Massachusetts Institute of Technology All rights reserved. No part of this book may be reproduced in any form by any electronic or mechanical means (including photocopying, recording, or information storage and retrieval) without permission in writing from the publisher. For information about special quantity discounts, please email
[email protected]. This book was set in Baskerville by SNP Best-set Typesetter Ltd., Hong Kong, and was printed and bound in the United States of America. Library of Congress Cataloging-in-Publication Data Rehg, William. Cogent science in context : the science wars, argumentation theory, and Habermas / William Rehg. p. cm.—(Studies in contemporary German social thought) Includes bibliographical references (p. ) and index. ISBN 978-0-262-18271-3 (hardcover : alk. paper) 1. Habermas, Jürgen. 2. Science—Philosophy. 3. Debates and debating. 4. Persuasion (Rhetoric). I. Title. B3258.H324R444 2009 121—dc22 2008029433 10 9 8 7 6 5 4 3 2 1
Contents
Preface and Acknowledgments Introduction: Science Wars, New and Old
vii 1
I The Argumentative Turn in Science Studies
15
1 Science as Argumentative Practice
17
2 Kuhn’s Gap: From Logic to Sociology
33
3 Closing the Gap: Three Rhetorical Perspectives on Science
57
Postscript I: The Return of the Logical: Achinstein’s Realist Theory of Evidence
81
II Integrating Perspectives: Habermas’s Discourse Theory
99
4 Habermas’s Critical Theory and Science: Truth and Accountability
101
5 Habermas’s Theory of Argumentation as an Integrated Model of Cogency
129
6 Argumentation at Fermilab: Putting the Habermasian Model to Work
163
Postscript II: Who’s Afraid of SSK? The Problem and Possibilities of Interdisciplinary Cooperation
195
vi Contents
III Toward a Critical Contextualist Framework for Interdisciplinary Assessment
211
7 Adjusting the Pragmatic Turn: Lessons from Ethnomethodology
213
8 Three Dimensions of Argument Cogency—A Contextualist Case Study
241
9 Critical Science Studies and the Good Society
269
Notes
297
References
313
Index
337
Preface and Acknowledgments
In this work I reflect on the cogency of scientific arguments. Although I approach that issue primarily as a philosopher, the issue itself is a matter of concern for many groups—not only the scientists who make arguments when they adduce evidence in support of hypotheses, but for anyone concerned about the basis of scientific claims or interested in understanding that basis: laypersons, policymakers, science journalists, scholars of science and technology studies (STS), critical social theorists, rhetorical scholars, and argumentation theorists. Although many members of these groups do not consider themselves philosophers, they nonetheless have an interest in the philosophical issues raised by scientific inquiry. I therefore address this book not only to professional philosophers but also to members of these other groups as well—in the hope that they do not find the philosophical terrain overly taxing. The growing importance of science for policy-formation and lawmaking, as well as the increasing need for interdisciplinary work, poses tough philosophical problems that are not well served by superficial treatment. The approach I defend is both critical and contextualist in character. In calling the approach “critical,” I do not mean that it takes a dismissive, skeptical attitude toward the sciences and scientific argumentation; rather, “critical” here has the sense of a discriminating evaluation. In labeling the approach “contextualist,” I emphasize the sensitivity of such evaluation to the particular contexts—the particular experimental locales and subdisciplinary communities of discourse—in which arguments are made and assessed. At the same time, I formulate the critical contextualist approach as a comprehensive framework within which analysts from different disciplines can cooperate in the
viii Preface and Acknowledgments
critical evaluation of scientific arguments. The endless parade of case studies within STS has led some of its members to ask: to what end? The present study, I hope, provides an answer: a way to see how case studies can fit together within a larger vision oriented toward the assessment of cogent science that serves both the production of knowledge and the good of society. This work is the product of engagement with many groups and individuals. I am thankful to a number of people who provided helpful feedback on earlier drafts: Garth Hallett, S.J., and three anonymous reviewers for going through the entire book, Kent Staley and Richard Blackwell for feedback on chapters dealing with the philosophy of science and physics, and Thomas McCarthy for his ever sage advice on the introduction and overall framing of the project. The History and Philosophy of Science reading group at Saint Louis University, organized with the help of Kent Staley, Aaron Cobb, and Scott Crothers, commented on two chapters. A number of people provided feedback on earlier versions of chapters or parts of chapters that appeared in talks and articles: Alison Wylie, Paul Roth, James Bohman, Peter Achinstein, Steve Fuller, Scott Berman, David Bogen, Douglas Marcouiller, S.J., Michael Barber, S.J., Walter Jost, Walter Ong, S.J., Matthias Lutz-Bachmann, Andreas Niederberger, Jean Goodwin, and J. Anthony Blair. I also thank participants in the following occasions for feedback on papers that eventually worked their way into book chapters: the 1999 and 2004 meetings of the Philosophy of Social Science Roundtable (St. Louis, 1999); the Science Studies Seminar at the University of Oslo (Oslo, Norway, 1999) and the STS Colloquium at MIT (2002); philosophy department colloquiua at Saint Louis University (2000, 2004), the Goethe University of Frankfurt, Germany (2005), and Loyola University-Chicago (2006); the Critical Theory Roundtable (St. Louis, 2004); and the Society for the Social Study of Science (4S) (Pasadena, 2005). Finally, I am especially grateful to comembers of three interdisciplinary panels of which I was a member, and to the audiences for those panels. The first, at the 2002 meeting of 4S in Milwaukee, included William Keith, James Collier, and Steve Fuller. The second panel, held at the National Communication Association meeting in Chicago (2004), was organized by Jean Goodwin and included J. Anthony Blair and Robert Asen, with Lynn Clarke as chair. This panel proved especially important for chapter 8 of the book, and I thank Jean for inviting me into such a fruitful exchange that continues to this day. The third panel, organized by Kent Staley and including Henry Frisch and Deborah Tollefsen, dealt with scientific collaborations; it was held
ix Preface and Acknowledgments
at the 2004 meeting of the Philosophy of Science Association in Austin, Texas. Research for the book itself began during my stay as a Visiting Scholar at the Science, Technology, and Society Program at MIT in 2001–2002, funded by a Saint Louis University SLU2000 Research Leave Grant. At MIT I found a very hospitable environment: as chair, Merritt Roe Smith graciously welcomed me into the STS Program, and Kenneth Oye of the MIT Political Science Department invited me into his colloquium on science and technology policy. For particular aspects of research, I owe thanks to Alison Wylie, who generously provided me with a copy of an article by Jean Gero, and to the Council for Agricultural Science and Technology (CAST) for a copy of their 1982 report on the National Academy of Science’s Diet, Nutrition, and Cancer. For research assistance during the years I worked on the book, I thank Taki Suto, David Packman, Jonathan Nelson, Paul Leisen, and Yong Li. There is, however, an older debt I would also like to acknowledge. I doubt that this book would have come about but for teachers who stimulated my interest in science and for the opportunity to work as a research chemist in the mid-1970s. For their stimulating teaching and mentoring, I am particularly grateful to Joseph Sens, Rubin Battino, and Michael Smith. For the opportunity to engage in professional research, I am deeply thankful to Robert E. Sievers, with whom I worked over an exciting two-year career transition that took him from Aerospace Research Labs and Monsanto Research Corporation in Dayton, Ohio, to the University of Colorado in Boulder. My gratitude extends not only to Bob for his professional leadership but also to his entire family for their gracious hospitality and friendship in Boulder. This book incorporates some previously published material, most of it heavily revised. Material in chapter 5 first appeared as a chapter in Rhetoric and Hermeneutics in Our Time (edited by W. Jost and M. J. Hyde; Yale University Press, 1997). The postscript for Part II first appeared in Philosophy of the Social Sciences (March 2000). Earlier versions of chapter 7 appeared as a chapter in Pluralism and the Pragmatic Turn (edited by W. Rehg and J. Bohman; MIT Press, 2001) and then, after substantial revision, in Economic Policy under Uncertainty (edited by P. Mooslechner, H. Schuberth, and M. Schürz; Edward Elgar Publishing, 2004); chapter 7 also includes material from an article published in the Canadian journal Symposium (2005). Some material in chapter 8 first appeared in the journal Informal Logic (2005). I gratefully acknowledge the work of the editors and referees of these venues.
x Preface and Acknowledgments
Finally, there are a number of people whose support goes beyond the tangibles. Eleonore Stump has provided insistent and ongoing encouragement to bring the project to a conclusion. My Jesuit superiors Douglas Marcouiller, Ralph Huse, Timothy McMahon, and Frank Reale have all been supportive over the years—as have so many Jesuit brothers in communities that welcomed me as I worked on various phases of the project: at the Weston School of Theology in Cambridge, Massachusetts; at the Sankt Georgen community in Frankfurt am Main; at Loyola University-Chicago; and here at Saint Louis University. Finally, my department chair, Ted Vitali, C.P., has not only provided release time but has been a constant source of support and good humor. Without the moral support of these people, I doubt this project would have reached completion.
Introduction: Science Wars, New and Old
Today one can hardly avoid noticing an odd polarization in attitudes toward the natural sciences. What makes the polarization odd is its place. It does not appear so much as a split between two areas of the globe—between an allegedly rational, secular West and religious, tradition-bound non-Western cultures. Rather, the polarization is found precisely in those Western cultures that have been historically most committed to the advancement of science and technology. This is especially true in the United States. Among the international leaders in science and technology, the United States has nonetheless been the host of intense, increasingly acrimonious debates over the status and legitimacy of the natural sciences. Among the most publicly visible debates are those over the relationship between science and religion. Here evolutionary biology has been a particularly effective lightning rod. Atheistic scientists employ Darwinian modes of naturalistic explanation in arguments against religious belief, while Christian fundamentalists advance a literal reading of Genesis as the basis for a “creation science” that deserves equal time in science education. “Intelligent design” theorists take a somewhat different tack, arguing that improbable biological complexity requires supernatural explanations of evolution. Like the naturalists they oppose, both creationists and intelligent design theorists lay claim to an idea of “science.” In doing so, they advance a view about what counts as a scientific argument—what sort of evidence or explanatory inference provides good reason for accepting a claim about the natural world and its origins. However, conflicts over the nature of scientific argument and explanation do not arise only when religion gets involved. The potential for polarization
2 Introduction
exists whenever the sciences claim to uncover avoidable risks that call for political or personal response. The political and corporate resistance to the claims of the International Panel on Climate Change provides an example. Similar to the evolution debates, here too an oppositional atmosphere developed in which the trend of opinion in the science community by 2007—consensus on the reality of warming and growing acceptance of its likely source in human fossil-fuel emissions—ran into a small group of skeptics, including some scientists, whose influence on the debate exceeded their numbers (see Lahsen 2005; Weart 2003). But unlike the evolution debates, the players were not attempting to redefine science itself so much as arguing about the strength of evidence and whether it warrants potentially costly economic measures. For the public at large, however, such public dissension raises questions about how to assess controversial scientific claims. The debates over evolution and global warming involve more than the mutual misunderstanding that often separates scientists and nonscientists (Snow 1961). Although misunderstandings abound, the recent debates reflect a distinct issue, an issue that must arise for any culture committed to taking the sciences seriously. Scientific research is bound to generate claims that impinge not only on cherished religious beliefs but also on lifestyle, health, diet, environmental quality, and the like—and thus on the pursuit of personal, economic, and political interests. In such contexts, the rationality of scientific arguments deserves special scrutiny. As the debates show, scientific evidence can be contested: the evidence normally involves uncertainties; often enough it rests on assumptions that later prove incorrect. And are not scientific claims merely “theories” or “hypotheses” rather than facts? Do not scientists too have their biases? So say the creationists about evolution, and so claim the skeptics about global warming. In this book I tackle the deeper question that underlies controversies like those described above: what makes a scientific claim merit our consideration? In academic contexts, versions of this question have been at the forefront of interdisciplinary controversy for the last half-century. The rationality debates—or, as they now tend to be labeled, the “science wars”—have been a contentious affair involving the disciplines that contribute to “science studies,” which include inter alia the history and philosophy of science, and the sociology of scientific knowledge. The public controversies over evolution and global warming (among others) display the new, public face of the academic science wars.1
3 Introduction
The science wars should hardly surprise us, given what is at stake. On the one hand, the sciences are widely perceived as tremendously successful: in contributing to technological development and improving health care and the quality of life, in understanding the mysteries of nature. On the other hand, in contemporary pluralistic societies, “science” is one of the few intellectual authorities that nearly everyone still acknowledges. In general, the prevailing popular assumption has been that properly conducted science establishes the facts in an “objective” manner, which remains neutral with respect to value-laden decisions for which those facts are relevant. To be sure, as the latest science wars demonstrate, lay publics can withhold recognition of specific disciplines (evolutionary biology) or contest specific claims (regarding climate change) when these conflict with cherished beliefs or particular interests. These widespread assumptions about scientific objectivity and the neat divide between fact and value, science and policy, are precisely what came under fire in the academic science wars that had been brewing since the 1960s. Upon closer inspection, the vaunted “scientific method” proved to be a dubious oversimplification that belied the messy, often controversial pastiche of local techniques, shortcuts and biased judgments, ad hoc decisions, and social negotiations that characterize the actual conduct of inquiry. Like other areas of human endeavor, the sciences exist and develop as social practices—exercises in embodied social rationality, which the various empirical branches of science studies have documented through a series of sociological and historical case studies. This trend has challenged defenders of science to develop more realistic conceptions of scientific rationality. This challenge confronts anyone who still regards the sciences as capable of playing a legitimate role in contemporary life. If the claims of science do not issue straight from the mouth of nature, amplified without distortion by scientific method, then on what basis do they deserve acceptance? Influenced by the study of controversy in the sciences, many science studies scholars answer that question by referring, implicitly if not explicitly, to the argued character of scientific claims. Scientists strive to support their claims with good reasons of one sort or another: observational evidence, experimental tests, mathematical and theoretical considerations, and so on. The two debates I described above involve conflicting views about how we ought to understand the process and standards of argument-making in the sciences. In light of such conflicts, we can put the central question of this book more precisely: what is it that makes
4 Introduction
scientific arguments cogent, and how ought we reasonably to assess that cogency? This question calls for a prescriptive or normative reply. The empirically minded disciplines in science studies generally strive to avoid prescription. But descriptive-explanatory analyses leave open the unavoidable problem of assessing the reasonableness of the scientific claims that increasingly confront us with potentially relevant information for public and personal decision making. From administrative policymakers designing environmental regulations to courts grappling with DNA tests, from legislatures debating hazardous material controls to international bodies negotiating fossil-fuel emission standards, from publics deliberating limits on stem cell research to individuals deciding on cancer detection and treatment, increasingly we find ourselves confronted by questions in which reasonable choices require us to assess the relevant scientific findings. Whether one accepts such findings as input for choice, puts them aside as overly uncertain, or contests them as irrelevant, one takes a normative stance on what makes scientific arguments strong or cogent. Within science and technology studies (STS), one finds different approaches to scientific argumentation (see Keith and Rehg 2008). Evidence theories tend to focus on the logical relationships linking statements of evidence with hypotheses, though some take a more dialectical approach that stresses the severity of tests a hypothesis has passed (cf. Popper 1980; Mayo 1996; Achinstein 2001). Social epistemologists typically examine the ways in which scientific argumentation is socially organized (Longino 1990; Solomon 2001), in some cases drawing on the rhetorical tradition as well (Fuller 1993). Ideas of rhetoric have also entered into sociological and historical studies of scientific controversy and theoretical development (Shapin and Schaffer 1985; Beller 1999). This pluralism in the approaches to scientific argumentation leads me to propose a more comprehensive framework that can integrate these various initiatives at a conceptual level and foster greater cross-fertilization and interdisciplinary exchange within science studies. For a comprehensive normative framework I look to the study of argumentation (“argumentation theory”) as an interdisciplinary endeavor that provides a set of categories—drawn from logic, linguistics, dialectic, rhetoric, and so on—for the description and evaluation of arguments. My thesis is that categories and questions from the study of argumentation can provide a broad framework or heuristic that will help us critically assess the information and
5 Introduction
claims issuing from the empirical sciences. Although philosophers have done much work on the nature of inductive evidence and justification, systematic efforts to bring the full scope of argumentation studies explicitly to bear on the study of science remain rare. The best-known confrontations between argumentation studies and science have gone by the label “rhetoric of science.” In this vein, Steve Fuller’s Philosophy, Rhetoric, and the End of Knowledge (1993; cf. Fuller and Collier 2004) makes one of the more ambitious attempts to date at working out a general rhetorical-dialectical framework for cross-disciplinary exchange in STS. But as I explain in chapter 1, rhetoric and dialectic do not exhaust the resources of argumentation theory. Moreover, though Fuller and I converge on particular points (see esp. the end of sec. 1 in chapter 9 below), his central focus differs from mine. He does not provide a detailed treatment of cogency as a normative concept, but rather places the brunt of his normative analysis on how to hold scientific policy discourse accountable to democratic institutions. By contrast, I arrive at democratic accountability only by way of a broader normative analysis of cogency. In turning to argumentation studies for a normative framework, I assume that even in noncontroversial settings, scientific hypotheses, results, and forecasts must take shape as arguments and find their way in social processes of argumentation. Indeed, scientific findings and hypotheses become matters of public knowledge only through a process of “critical discussion” or argumentation—the construction, presentation, scrutiny, and reception of scientific arguments, that is, claims supported by evidential reasons. Scientists put forth such arguments publicly via a range of venues—lectures, preprints and published articles, internal reports, monographs. In doing so, they expect other scientists—at the very least, journal referees—to take a critical attitude, asking whether the evidence and reasoning supports the author’s claims. In speaking of the “critique” or “critical assessment” of scientific arguments, then, I do not mean a skeptical dismissal or rejection of the argument, but rather a process that lies at the heart of good science and science-intensive policy. Both in the sciences and in the various contexts in which the sciences meet society, one must examine scientific claims for their strength—is the finding conclusive, probable, or a mere possibility based on a limited amount of research? In policymaking, one must also ask whether the science is relevant. The popular press often omits the nuances and qualifications connected with research findings, or it overdramatizes findings that scientists themselves consider highly tentative. Critical assessment, then, requires an accurate evaluation of the
6 Introduction
strength and relevance of the evidence for the claim—in other words, an evaluation of the cogency of the scientific arguments. Cogency as a Boundary Concept The notion of argument strength or cogency provides the conceptual focal point in this book. I examine the various approaches to the study of science— philosophical, sociological, rhetorical—as models of cogent argument in the sciences. In other words, I treat cogency as a “boundary concept” that illuminates points of convergence and divergence within science studies, and hence possibilities and obstacles to greater interdisciplinary cooperation. According to Julie Klein (1996, 50–51), such concepts are “powerful sources of cross-fertilization” between two (or more) disciplines. As she goes on to explain, boundary concepts “are negotiable entities that simultaneously delimit and link particular territories. In cognitive terms, they facilitate hybrid intellectual work. In social terms, they facilitate intergroup alliance.” For example, concepts of “text” and “discourse” have fostered interdisciplinary exchange in the rhetoric of science, in which theorists have combined resources from rhetoric, literary studies, linguistics, and composition in studies of scientific texts and controversies (ibid., 66–70).2 To serve this cross-fertilizing role, a boundary concept must have some purchase in the different disciplines. Toward this end, it helps if such concepts are on the one hand “loose” or “weakly structured in common use,” yet on the other hand “strongly structured” at the interdisciplinary site (Star and Griesemer 1989, 393). The boundary concept, as I understand it here, must at one level allow a sufficiently flexible or broad interpretation that covers different methods and uses across disciplines, yet at the level of actual exchange and cooperation it must facilitate translation across disciplinary boundaries and generate specific research questions, points of debate, and possibilities for hybrid analysis that combine resources from different disciplines.3 The concept of cogency has just the properties we need for a boundary concept of the above sort. This supposition finds support in both the standard usage of the term and its use among argumentation theorists. As to standard usage, the word “cogency,” when applied to arguments, refers to the “convincing quality,” the “logical or persuasive force” of arguments; thus “cogent” arguments have “power to compel assent or belief,” they are “argumentatively forcible, convincing.”4 These definitions of “cogent” link a normative idea, the strength or logical character of good reasons, with a psychological effect on
7 Introduction
audiences, namely, the perception of a persuasive force that is not easily resisted. Thus the idea of cogency sits at the boundary between psychological effect and rational content. Moreover, the broad association of “cogency” with persuasiveness suggests that cogent arguments include not only logically valid deductions but also inductive arguments with sufficient probability (or plausibility) to persuade. This observation, along with the etymological association with force (deriving from the Latin root), suggests that cogency might sometimes come in degrees, that perhaps in some cases an argument we consider cogent, stronger than a set of counterarguments, is later met by a still stronger, hence more cogent rebuttal. Among argumentation theorists, “cogency” does not have a generally established technical sense—unlike “validity” or “soundness,” which normally refer to specific properties of deductive arguments. Rather, we find a range of uses similar to what we saw in the standard definitions. In their survey of argumentation studies, van Eemeren, Grootendorst, Snoeck Henkemans, et al. (1996) identify theories (a) that take “cogency” as a broad term that covers the ideas of “soundness,” “validity,” and “strength” (ibid., 155); (b) that understand “cogency” as a comparative measure of dialectical strength (ibid., 90); and (c) that use “cogency” in relation to the rhetorical effect on an audience (ibid., 32). Informal logicians sometimes use “cogency” as a term for the strength of an argument’s content, albeit understood to include at least one audiencerelative property, rational acceptability (see, e.g., Govier 2005; Vorobej 2006). On the whole, then, argumentation theorists tend to use the term as a broad synonym for argument strength and/or persuasiveness. We thus have at least three reasons for regarding the idea of cogency as a promising boundary concept for an interdisciplinary argumentation theory in the service of a critical science studies. First, as the standard definition and common use imply, “cogency” is more or less synonymous with a range of expressions that refer to the strength or convincing quality of arguments; it thus has the breadth and flexibility to cover a large territory of approaches to argument evaluation. Second, the term sits precisely at the key point of contention in the science wars, namely at the point where normative and empirical categories come together—where reasons display their psychological and sociological effects. Third, in contrast to terms such as “valid,” the word itself is not overburdened with technical expectations connected with specific modes of argument analysis (e.g., formal logic). Thus the term is relatively open to theoretical articulation at the sites of interdisciplinary exchange.
8 Introduction
Theoretical articulation, as I understand it here, aims to specify the properties or features of arguments that make them cogent. To do so, one must look to argumentative practices. When participants in such practices refer to some arguments as cogent or convincing and others as weak and unconvincing, they operate with at least an inchoate notion of cogency-conferring properties; in some areas, such as mathematics and the sciences, what makes an argument cogent might be precisely defined, provided we can identify a range of more or less equivalent evaluative terms (“highly probable,” “convincing,” “conclusive,” etc.) with “cogent,” or with different degrees of cogency. In attempting to articulate these inchoate notions, or to translate discipline-specific standards of argument into a general account of cogency, one engages in a theoretical enterprise. Because many of the initiatives in science studies involve approaches to argumentation in the sciences, we can construe them as contributing, at least potentially, to a theory of cogency. Some contributions lie primarily at a descriptive or explanatory level: most sociologists and historians, and many rhetoricians, strive to articulate the norms and values that scientists themselves employ in their argumentative practices, or to explain why scientists found some arguments more convincing than others. Philosophers, on the other hand, have traditionally sought to provide normative or prescriptive accounts of scientific argumentation. The final burden of my analysis is to elaborate a multidimensional conception of cogency that pulls these different approaches together, integrating logical, rhetorical, and sociological tools for purposes of cooperative critical assessment of scientific arguments. In pursuing this task, I hope to avoid an overly holistic kind of integration. As Klein (1996, 11–15) uses the term, “holistic” approaches tend to rest on metaphysical assumptions or overly strong claims of unification that ultimately undermine interdisciplinarity—the attempted theoretical unification cannot bear the strains of disciplinary difference. As an example, Klein (ibid., 13) cites the unity of science approach of the early twentieth century; some versions of Marxist social theory also seem to fit her description. For purposes of this study, I take the holist mistake to reside in an overreaching attempt to unify perspectives on cogency in a single conception of the norms of scientific argumentation in general. So understood, a tendency to holism appears in certain features of Habermas’s discourse theory, which I take up in Part II. The problems associated with holism ultimately lead me, in Part III, to propose a contextualist approach to conceptual and interdisciplinary
9 Introduction
integration. In doing so, I assume that what we need is not so much a single one-size-fits-all as a capacious set of concepts and questions that allow theorists to collaborate in the study of science, even when their deeper epistemological presuppositions differ. This assumption is reinforced by the observation that the so-called scientific method does not reduce to a single set of rules governing all of science. What we call “science” today is not a unitary enterprise, even if there are some broad family resemblances and perhaps some minimal shared commitments among the different sciences. In any case, I focus in this book mainly on the natural sciences, as they are generally identified—physics, chemistry, geology, biology, and the like—and the health sciences. Argumentation studies promise us the complexity and openness required for such an interdisciplinary and multiperspectival endeavor. As the comprehensive study of modes of analyzing and evaluating arguments and argumentation processes (both in formally structured institutional settings and in everyday life), argumentation studies are both internally complex and open to cooperation with a wide range of other disciplines. The internal complexity arises from the fact that argumentation theory, as a field of study, attends to all the dimensions of arguments and argumentative discourses—not only logical but also dialectical, rhetorical, social-institutional, and so on. We can see this internal complexity in the field itself, in which scholars from a range of disciplines, including inter alia logic and philosophy, speech communication and rhetoric, linguistics, composition, and computer science, engage in interdisciplinary exchange.5 Outline of the Argument The project I have been introducing above integrates interdisciplinarity into a theoretical framework for analyzing and evaluating scientific argumentation as an embodied social practice. Using the notion of cogency as the overarching idea, I look to argumentation studies for a framework that brings together a set of more specific boundary concepts, a constellation of common concerns and questions shared by those disciplines that study scientific practices. Such concepts make it possible to bring different disciplinary tools to bear on the study of scientific argumentation. In the first chapter, I clarify this conceptual framework in relation to a number of related theoretical developments: the “practice turn” in science studies, that is, the focus on science as a material practice; the rhetorical turn in the study of scientific texts; and cross-
10 Introduction
fertilization between rhetoric and argumentation studies. I then close the chapter by addressing ambiguities in the way some key elements of the framework have been aligned by argumentation theorists. The basic evaluative perspectives that argumentation theorists have inherited from the ancient Greeks—logical, dialectical, and rhetorical—do not neatly align with another, more recent triad: argument as product, procedure, and process. By modifying the “perspectivist” framework appropriately, we obtain a useful set of tools for studying different conceptions of cogency in science studies and fostering greater interdisciplinary cross-fertilization. To demonstrate the interdisciplinary usefulness of this framework, in chapters 2 and 3 I show how it emerges from developments within science studies itself. Taking Thomas Kuhn as the pivotal figure, I show how the perspectivist framework can illuminate issues in science studies since his 1962 Structure of Scientific Revolutions (see Kuhn 1996). In opposing his social-institutional analysis to logical empiricism, Kuhn opened up a gap within argumentation theory, that is, a situation in which two perspectives on the cogency of scientific argumentation opposed one another across an area of unexplored territory—the area in which arguments are persuasive but not logically compelling. After elaborating on these developments in chapter 2, I go on in chapter 3 to discuss three rhetorics of science that responded to “Kuhn’s Gap” with analyses of persuasion that might reconcile the logical and the sociological. These fall along a spectrum of possibilities that stretch from Marcello Pera’s abstract normative dialectics at one end, through Lawrence Prelli’s more concrete, communitarian rhetoric, to Bruno Latour’s antinormative, sociologically informed approach at the other end. The postscript to Part I updates the story by examining a more recent version of the logical perspective: Peter Achinstein’s realist theory of evidence. Achinstein’s approach is relevant because it suggests a principled distinction between a definition of cogency as an objective property of arguments and the social process necessary for accurately assessing such cogency. His analytical approach supplies an instructive contrast to Kuhn and rhetoricians of science, who include social process not merely as a means to assessing cogency but as partly constitutive of cogency itself. Taken together, these views mark out different positions and perspectives on scientific argumentation—they advance diverse views about the cogency of arguments, about the most significant aspects of such argumentation, and about the rationality of science. I survey these different theories to show how certain broad categories of argumentation theory have already emerged
11 Introduction
autochthonously within science studies. To this extent, the study of argumentation provides a suitable multidisciplinary framework for understanding the issues and concepts at stake in the assessment of scientific arguments. More importantly, the survey reveals specific tasks and challenges confronting the attempt at an interdisciplinary normative framework. Foremost is the question of how one bridges Kuhn’s Gap and integrates the different perspectives on cogency in a way that maintains the possibility of interdisciplinary cooperation. In Part II I begin to tackle those further tasks, with the aim of developing an argumentation-theoretic framework that can guide an assessment that is at once critical and interdisciplinary. For this purpose, Jürgen Habermas’s discourse theory provides one of the most developed attempts at a normative social-institutional theory structured around a perspectivist model of argument evaluation. Habermas’s theory deserves consideration in the present context for at least three reasons. As a second-generation member of the Frankfurt School of critical social theory, he inherits from his forebears an interest in the critical, yet nonskeptical, study of society and its various institutions, including science. As I explained earlier, “critique” here does not mean rejection or dismissive skepticism, but rather a discriminating evaluation that partly relies on scientific methods and findings. This leads to a second reason: like the earlier Frankfurt School, Habermas has an interest in an interdisciplinary approach that draws on a range of disciplines, sciences, and theoretical perspectives for input. Finally, in contrast to his forebears, he explicitly depends on a comprehensive argumentation-theoretic framework. In Part II, then, I move from an overview of science studies to the groundlevel of a particular theory: with the move to Habermas we can see how the broad argumentation-studies framework interacts with substantive theorizing to generate critical assessments open to interdisciplinary input. Because Habermas has developed his discourse theory most fully in the areas of morality, politics, and law, further elaboration is necessary to bring it to bear on the sciences. To prepare the theoretical background, in chapter 4 I recount his communicative, formal-pragmatic approach to the problem of social order, his pragmatic-realist theory of truth, and his larger emancipatory vision. In chapter 5 I elaborate Habermas’s theory of argumentation in detail. In doing so, I flesh out his rather thin conception of rhetoric with more substantive Aristotelian categories. I also argue that Habermas’s pragmatism, in contrast to Achinstein’s realism, commits him to a multiperspectival—and
12 Introduction
irreducibly social—conception of cogency. As serving the communicative aims of argument-making, the dialectical and rhetorical properties of argumentation as a social process are not simply a means for accurately evaluating cogency, which would then reside solely in process-independent merits of the argument product. Rather, properties of the social process of evaluation are among the merits in virtue of which arguments should count as cogent. But it remains unclear whether Habermas’s model fully reconciles the gap Kuhn opened up between logic and sociology. Habermas’s dialectical and rhetorical perspectives are social only in an idealized sense; the status of the social-institutional perspective, understood in the sociological sense, remains ambiguous in his model—it remains unclear whether that perspective makes a distinct normative contribution to the idea of cogency. In chapter 6, I analyze a case of argument construction in high-energy physics, the writing up of the 1994 paper announcing evidence for the top quark at Fermilab. That case serves in part to illustrate the different aims of critical assessment. It also points to a difficulty in Habermas’s theory that is related to the ambiguity above. Although in many respects favorable to his model, the case involves an element of compromise that he cannot readily accommodate. In the Postscript to Part II I show how the perspectivist framework might allow a rationalist like Habermas to engage in interdisciplinary cooperation with the strongly relativistic and skeptical sociology of scientific knowledge (SSK) developed by the so-called Edinburgh School. Here we see how the use of argumentation theory can pave the way for interdisciplinary cooperation in the service of critical assessment. However, the move one must make to enable such cooperation points to a deeper problem in Habermas’s approach—a holistic tendency that arises from his particular mode of integrating perspectives and that underlies the difficulties his model has with social context. In Part III, I defend a contextualist alternative to Habermas’s approach. To do so, I confront a further challenge of interdisciplinary engagement, namely the challenge posed by nonskeptical but radically contextualist ethnographic approaches, in particular the ethnomethodological study of scientific work (ESW). Taking their cue from Harold Garfinkel and colleagues, ethnomethodologists such as Michael Lynch, Eric Livingston, and others have argued for a radical particularism that appears strongly at odds with the universalism of Habermasian discourse theory. Engaging this challenge at its more radical level leads me to a critical contextualist approach to cogency.
13 Introduction
In chapter 7, I set out the contextualist challenge and develop its implications for the ideas of truth, objectivity, and reasonable dialogical process. The analysis suggests that Habermas’s idealizing pragmatic presuppositions (inclusiveness, equal voice, noncoercive process, etc.) should be understood as rhetorical potentials implicit in argumentative practices. As such, they are effective only in conjunction with detailed substantive argumentation—a point I illustrate by recalling feminist critiques of science. Interpreting process ideals this way renders the normative framework for critical assessment more contextsensitive. But does it also require a potentially relativistic “indifference” incompatible with critique? That is, must one always simply accept local standards of rationality as the final word, immune to critical evaluation by higher courts of appeal? In reply, I describe the stance critical theorists must adopt to incorporate the insights of ESW without abandoning their commitment to critical assessment and falling into relativism. In chapter 8, I apply the critical-contextualist approach to a case study of three expert committee reports sponsored by the National Academy of Sciences (NAS). The case analysis turns on a distinction between two levels of argument-making process that correspond to different initiatives in science studies and in argumentation theory: the difference between local transactions and public arenas. Using a variety of source materials and argumentation-theoretic tools, the study shows how the participants were concerned with cogency at three levels: the reports’ content, the quality of the committee transactions, and the public merits of the reports. Further examination of the debates surrounding the reports and NAS procedures reveals important tensions in the rhetorical use of process ideals—tensions that refer in turn to competing visions of the contract between science and society. In chapter 9, finally, I recapitulate the book as a whole and clarify the prescriptive character of critical contextualism in relation to substantive critical assessment. As a general framework for argument evaluation, critical contextualism amounts to a metacritical procedure that follows arguments through the network of relevant contexts. By way of further clarification, I close with some tentative explorations of three kinds of controversy, each of which poses distinct challenges for contextualist analysis: interdisciplinary scientific controversies, science-intensive policy debates, and the exchanges between atheistic champions of evolutionary biology and Christian believers. The latter two cases are especially interesting for the questions they raise about the relation
14 Introduction
of scientific arguments to discursive contexts outside the sciences. Although the contextualist approach points toward a deliberative democratic approach to such controversies, it leaves open important questions regarding the political organization of deliberation and the specific institutional designs that would be appropriate for critical discussions involving scientific questions. How one answers such questions partly depends on one’s vision of the good society. The contextualist framework does not fully specify that vision.
I The Argumentative Turn in Science Studies
1 Science as Argumentative Practice
The normative framework I propose in this book links the assessment of expert claims with the notion of cogent scientific argumentation. The idea that scientific practices depend centrally on social processes of argumentation and not simply on experimentation, is, I believe, rather widely accepted. That argumentation can provide a sufficiently comprehensive framework in which to understand the sciences is a more ambitious assumption. In this chapter I provide some initial clarification of what I mean by science as a set of “argumentation practices.” I also introduce the conceptual framework I employ in Parts I and II for analyzing scientific argumentation. After some preliminary orientation (sec. 1), I describe the “rhetorical turn,” which led to the rhetoric of science, and the emergence of argumentation studies (secs. 2, 3). These developments in the study of rhetoric and argumentation led a number of theorists to propose a multiperspectival framework for the evaluation of argumentative practices; this framework, appropriately clarified, can serve as a heuristic for understanding developments in science studies (secs. 4, 5). 1
Scientific Inquiry and Argumentative Practices
Approaching scientific inquiry as an argumentative practice immediately suggests a possible objection, which I want to dispel at the outset. The objection goes as follows. Granted, there are occasional scientific controversies, and granted, science articles employ specific types of rhetoric and can be interpreted as arguments of one sort or another. But as a general framework, an argument-centered approach seems overly textual and abstract—just one
18 Chapter 1
more variant on the old “science-as-knowledge” approach, which misses the materiality of “science-as-practice” (Pickering 1995; also 1992b). To press the point, one might ask whether arguments and argumentation capture the process of inquiry and “logics of discovery” that lie at the very heart of scientific knowledge-production. Must not an approach based on the assessment of scientific arguments ultimately reproduce—to be sure, in more complex and sophisticated terms—the old disjunction between context of discovery and logic of justification, and in such a way as to privilege that latter? The short answer to the last question is “no.” For the long answer, one must actually attempt to work out an argumentation-theoretic framework. But the short answer has a number of considerations in its favor already. As we shall see in the next section, the developments that shaped argumentation studies in recent decades are precisely of the sort that undermines a discovery– justification distinction, at least in the logical-empiricist sense. Rejecting formal logic as an adequate theory of argument, many argumentation theorists today strive to situate arguments in their practical contexts. Although they generally understand these contexts as discursive or intellectual, in the natural sciences we must also consider them material contexts: evidential arguments are typically about what one can do with materials in a laboratory, or about what one can observe in the physical world. This does not mean that no distinction remains between scientific argumentation and experimentation. What it does mean is that argumentative practices in the sciences are partly material practices. Experimental practices of inquiry thus intertwine with argumentation, even at the very concrete level of contingent material “resistances” (Pickering 1995, 51; Galison 1997). The success of scientific arguments is measured by their relation to experimental practices and not simply by standards for the logical composition of articles. We can see this at a number of levels of scientific practice, beginning with the local research site or laboratory. As the novice scientist soon learns, one of the first challenges is to arrive at stable, reliable experimental methods and results in one’s own laboratory: mastering the material situation, therefore, is partly constitutive of argument construction, for without reproducible results one has no evidence to report as reasons in support of one’s hypothesis. At the very least, one must get one’s instrumentation and observational methods to function properly. To a large extent, the daily work of bench-top science is oriented toward solving specific experimental problems, solutions that presuppose one has gained sufficient mastery of the relevant laboratory techniques to obtain
19 Science as Argumentative Practice
results that are both reproducible and trustworthy. Only if one answers such questions of detail—for example, in the area of pollutant testing, questions regarding such mundane details as the best sample size and reagent concentrations, possible interferences, optimal instrument settings, and so on—can one acquire the empirical evidence that can adequately support a conclusion. This daily struggle with the physical world in the laboratory or in the field is thus oriented toward the development or construction of an argument—indeed is part and parcel of the constructive process, where “construction” simply refers to putting together the evidence required to support a publishable result. Experimentation ultimately aims beyond the lab, however: experimental practices are heavily oriented toward the production of public knowledge, and to reach that goal findings have to be presented in a convincing manner as publicly acceptable arguments (Ziman 1968). Here public acceptability is not measured by publication alone, but more pertinently by the usability of one’s findings and arguments for the research of other scientists (cf. Hull 1988). The ongoing concern with classified military and corporate research testifies to the value of this traditional orientation. The orientation of everyday laboratory practices toward the production of arguments is also evident in the development of a research proposal. Although one might consider the proposal itself as a kind of argument, here I am interested in the tacitly projected argument at which the proposal aims. To formulate a research proposal, the researcher generally must (a) identify a problem or question that is (or can be) of interest to other scientists (and perhaps certain groups of nonscientists) and (b) have an idea about how to go about answering the question or solving the problem (cf. Montgomery 2003, chap. 11). In identifying a question or problem, one commits oneself to arriving at some kind of conclusion; in proposing an approach or method, one commits oneself to some kind of argument that will support the conclusion that addresses the question. The research proposal, in other words, is the first step in a process of constructing an argument that the researchers hope will have a place in a broader dialectic of inquiry within the subdiscipline or area of research. Thus the dialectic of inquiry that constitutes science as a substantive intellectual process—of research in response to a question, which in turn opens up further questions leading to further research—sets the argumentative context in which the proposal is supposed to make sense. Inquiry is dialectical insofar as it involves an interlocking series of substantive moves—communication of results or
20 Chapter 1
arguments in one venue or another—in which later moves respond to or build on earlier ones. The series can be controversial, involving objections, replies, and rebuttals, but it can also have a more irenic character, involving a series of studies that gradually close in on establishing some result or hypothesis. If we can understand the broader dialectic of scientific inquiry or progress as argumentative, and if we can understand inquiry within the laboratory or research team as part of argument construction (even if it is not only that), then there is no reason to split inquiry, the process of discovery, from scientific arguments and argumentation. Although an argumentation-theoretic framework provides a context in which to assess the cogency of scientific claims—and hence is in that sense a context of justification—the notion of argumentation I employ here takes in, as part of its substance, the discovery process itself. 2
The Rhetorical Turn
As I mentioned earlier, the area of research that goes by the name “rhetoric of science” has gone the farthest in applying argumentation theory to the study of science. But the rhetoric of science emerged as part of a broader “rhetorical turn” and the so-called new rhetoric associated with it. These developments were motivated by dissatisfaction with the positivist and logic-centered approaches in the study of inquiry and argument. Among philosophers, this movement was led by attempts to go beyond formal-logical analysis to the analysis of informal and noncompelling arguments (Perelman and OlbrechtsTyteca 1969 [Paris ed. 1958]; Toulmin 1958; Naess 1966 [Oslo ed. 1947]). I will say more about these initiatives below, in connection with argumentation studies. Among rhetorical scholars and speech communication theorists, dissatisfaction centered on the rather wooden neo-Aristotelian mode of rhetorical criticism dominant in the mid-twentieth century (see Black 1978; Wenzel 1987, 103–104). In these fields, Kenneth Burke was a prime mover. On the one hand, theorists who engaged in these developments desired a richer set of analytic tools and foci; in particular, they wanted to get beyond the neo-Aristotelian focus on the speaker’s use of logos-centered rhetoric in narrow occasional contexts. For their part, philosophers wanted to bring in normative and hermeneutic perspectives that were more context-sensitive than logical analysis. On the other hand, the rhetorical turn also involved an extension of the focus of rhetorical analysis beyond its traditional subject matter of
21 Science as Argumentative Practice
civic discourse. Emboldened by Burke’s idea that rhetoric operated wherever symbols were involved in communication to induce cooperation, rhetorical scholars turned their sights on a range of texts and domains, including science, which had been hitherto considered nonrhetorical (Simons 1989, 1990; Gross 1996; Pera and Shea 1991; Harris 1997). The rhetorical turn has led to interdisciplinary cooperation involving the disciplines of literary studies, composition, linguistics, sociology, history, philosophy, and communication studies.1 As Klein (1996, 66–70) notices, the rhetorical turn has spawned a number of important “boundary concepts” (“discourse,” “rhetoric,” “text”) that facilitate cross-fertilization. Even analytic philosophers of science have begun to take the rhetoric of science seriously (e.g., Kitcher 1995). Among students of argument, the rhetorical turn has also been a site of contestation. In the perception of some scholars (e.g., Schiappa 1995; Kauffeld 2002), rhetoricians have tended to emphasize persuasive effect to the point of eschewing normative evaluation. This objection points to an ambiguity in the very idea of “rhetorical criticism” (cf. Lucas 1981): in what sense is it critical? According to Black (1978, 47, 77ff.), the old-style neo-Aristotelians tended to take the speaker’s ends for granted, thereby focusing rhetorical criticism on the effectiveness of means. Indeed, much rhetorical criticism primarily explains why a given argument succeeded or failed with a given audience. But as some theorists have pointed out (including Black), rhetorical analysis does not necessitate such a restriction (Wenzel 1990; Leff 2002). One might, for example, take certain rhetorical means, such as appeal to emotion, and develop normative standards for distinguishing reasonable and unreasonable, and not merely effective or ineffective, rhetorical appeals (cf. Walton 1989, chap. 4; Tindale 1999). Thus a rhetoric of inquiry need not be at odds with ideals of objectivity (Keith and Cherwitz 1989). Other critics have objected to the emphasis on the intentions and agency of the speaker to the disregard of other types of forces, such as deeper linguistic structures, that shape the speaker’s discourse in unintended ways (e.g., Black 1978, 35; Gaonkar 1997a,b). However, the vast range of discourse-analytic methods now available for rhetorical criticism (see Sills and Jensen 1992) suggests that there are ways beyond this overemphasis on agency. Moreover, a study of the Sophists casts doubt on an overly strict association of rhetoric with the goal of persuasion (Tindale 2004, chap. 2). On a broad view, therefore, rhetoric studies “all the ways by which meaning is created symbolically among people” (Wenzel 1987, 106).
22 Chapter 1
Still other scholars have expressed dissatisfaction with the overemphasis in the rhetoric of science on the analysis of particular texts and their construction—an imbalance that probably stems from the dominance of literary disciplines in the rhetorical turn. This dissatisfaction has generated interesting attempts to move beyond the textual product. For example, Myers (1990) and Blakeslee (2001) both examine argument construction as a social process in which authors shape their arguments in light of interaction with different audiences: journal editors and referees, members of other disciplines, and so on. I mention one final criticism, which might also apply to argumentation studies and thus can serve as a transition to that topic. Gaonkar (1997a,b) targeted the globalizing move that extended rhetorical analysis to any discourse, including science. Specifically, he wondered whether the rhetoric of science, in its current form, genuinely illuminates scientific discourse rather than merely adding a veneer of useless new jargon to scientific arguments that are clear enough as they stand. For my purposes this challenge raises the question of how rhetorical analysis—or more generally, an argumentation studies framework—can help in the critical assessment of scientific arguments and claims. The answer must await later chapters. 3
Rhetoric and Argumentation Studies
The rhetorical turn and the emergence of argumentation studies share common origins, but that does not mean we should equate the two movements. As a field of study, argumentation theory involves a number of different ways of investigating argumentation, not all of which are rhetorical. Logicians, for example, are concerned with truth-preserving structures, not persuasion or discourse in context. Acknowledging this does not foreclose the possibility that for some purposes, the rhetorical perspective—given its holistic character and openness in principle to a wide range of analytic tools—might do best at bringing together these various perspectives into a unitary analysis of actual argumentation. To be sure, to avoid the problematic kind of holism described in the introduction, one should not, in employing the tools of another discipline, forthwith reduce that discipline to a subcategory of rhetoric. In light of such considerations, I use the term “rhetoric” to designate a specific perspective on science, albeit a perspective whose interpretation, scope, and relation to other perspectives vary according to different theories of science and thus remain an open question.
23 Science as Argumentative Practice
I use the term “argumentation studies,” on the other hand, as an umbrella to cover the multidisciplinary complexity I briefly described in the introduction. This multifaceted study of argument goes back to the classical Greek traditions (logic, dialectic, rhetoric), but along with the rhetorical turn it experienced a rebirth in the mid-twentieth century (see Cox and Willard 1982; Wenzel 1979). Again, commentators typically cite Toulmin’s Uses of Argument (1958) and Perelman and Olbrechts-Tyteca’s New Rhetoric (1969; Paris ed., 1958) as leading the way. Toulmin maintained that most argumentation as it actually occurs in various fields such as law, science, and so forth is not analytical but “substantial,” dependent on inductive moves governed by standards specific to the field of inquiry or institution. Perelman and Olbrechts-Tyteca likewise drew attention to context, in this case the rhetorical context as defined by the particular audience and its presuppositions. Thus, argumentation aims not at deductive demonstration but at gaining (or increasing) the adherence of an audience.2 Because both initiatives linked the study of arguments with contextual awareness—in one case institutional, in the other rhetorical—they could not take the logician’s purely formal approach to argument evaluation. To develop an account of good arguments, that is, theorists had to pay attention to the empirical details of the different contexts in which arguments occurred. We can describe the shift initiated by Toulmin, Perelman and OlbrechtsTyteca, as well as Arne Naess and others, as a move from formal to “informal” logics.3 The move to informal logics was driven not only by theoretical concerns but also by broad dissatisfaction with the standard logic texts for teaching critical thinking: standard logical methods, many contended, were inadequate for understanding and teaching real argumentation (Govier 1987, chap. 1).4 To get a sense of what “informal” means in this context, it helps to review the main uses of “formal.” The term “formal” admits of a range of meanings (van Eemeren et al. 1996, 236–245, 263–271; Johnson 2000, 119–120). For present purposes, I consider as paradigmatically formal those deductive logics that rely solely on an axiomatic syntax that abstracts from all content and context (e.g., Kyburg 1968). Barth and Krabbe (1982) designated deductive rule systems as “formal2,” in contrast to Plato’s Forms (“formal1”) and rules of dialogical procedure (“formal3”). As we shall see with logical empiricists such as Carl Hempel, one can also develop formal inductive logics that are purely syntactical (chap. 2). Such logics are both prescriptive and purely formal: they provide a set of rules
24 Chapter 1
defining valid sentential entailments, and these rules operate independently of anything beyond the syntax itself and the formal semantics to which it may be linked. Informal logics, on the other hand, introduce a certain amount of substantive content into the norms of argument assessment. Consequently, they allow argumentation theorists to handle a broader range of everyday argument forms in a less artificial manner: the vast range of argument forms that includes not only simple induction but also analogical arguments, inference to best explanation, casuistic reasoning, narrative, and so on. The inferential properties of such arguments depend on substance: one must understand the interrelated meanings of terms as well as background information that resists complete formalization. To evaluate the substance of informal arguments, however, one must attend to the interpretive subtleties of arguments in their social contexts. Thus the shift to informal logic also involves a turn to the social practices of argumentation. Arguments—as premise–conclusion packages or claims supported by reasons or evidence—should thus be distinguished from “arguing” as a kind of social interaction (O’Keefe 1982). Many argumentation theorists now view arguments as the products that issue from the process of argumentation as a social practice (e.g., Johnson 2000). Still others distinguish three normative perspectives on argumentation: the logical product (open to both formal and informal approaches), the dialectical procedure or method, and the rhetorical process (Brockriede 1982; Wenzel 1990; Tindale 1999, 2004). Joseph Wenzel’s elaboration of this framework has been particularly influential: “rhetoric helps us to understand and evaluate arguing as a natural process of persuasive communication; dialectic helps us to understand and evaluate argumentation as a cooperative method for making critical decisions; and logic helps us to understand and evaluate arguments as products people create when they argue” (Wenzel 1990, 9, my emphasis; see also Wenzel 1979, 1987). This multidimensional framework has been widely accepted among argumentation theorists (though they tend to use the term “procedure” rather than “method”). The upshot is a loose alignment of the following triads, where each row represents one dimension of, or perspective on, argument that interpenetrates the other two: rhetorical perspective
arguing
social process
dialectical perspective
argumentation
cooperative procedure
logical perspective
argument
product
25 Science as Argumentative Practice
The term “argument” has both a narrow and comprehensive usage here. As one dimension, “argument” refers to the package of reasons supporting a conclusion; as a multidimensional social practice, “argument” takes in all three dimensions. After drawing up the alignment, Wenzel (1990) characterizes each perspective in terms of its typical purposes, scope and focus, situation, resources, standards, and roles. Wenzel (1979, 83, 85; 1990, 12) grants that these categories only “roughly” align, that the framework serves merely as hermeneutic starting point; moreover, these three perspectives are not exhaustive. The value of such “perspectivism,” as I shall designate it, lies in its hermeneutic and evaluative breadth, and thus in its serviceability as a heuristic open to a range of approaches and foci that make up argumentation studies as a field. Although the three perspectives do not exhaust the approaches, they do seem to capture the central normative perspectives on argumentation. Perspectivism thus provides a kind of heuristic for reading developments in science studies over the last half-century, as I show in the next two chapters. Moreover, as a set of normative perspectives on argument evaluation, this framework might be taken as a multidimensional account of cogency: the different ways one can understand or assess the cogency of arguments. Before pursuing these ideas, however, I want to tinker with the framework by introducing some further distinctions. In fact, there is considerable slippage between the columns above. After noticing some points of nonalignment, I suggest a more serviceable multidimensional heuristic. 4
Critical Analysis of the Perspectivist Alignment
To employ perspectivism as a heuristic, it helps to notice, and then repair, certain points at which the alignment of columns does not do justice to the actual practices of argument evaluation. I am not concerned so much with the middle column: “argument,” “argumentation,” and “arguing” strike me as far too fluid and interchangeable for marking technical distinctions. The two other triads (logic, dialectic, rhetoric; product, etc.) are more readily linked with operative distinctions in argumentation studies. However, their alignment is anything but tight. I first notice some points of slippage and suggest further distinctions in the framework, and then conclude, in the next section, with some broader clarifications of how I understand the framework as a heuristic. Start with the most obvious loose spot, the alignment between logic and product. Even if we accept an informal pluralist logic that recognizes more
26 Chapter 1
forms than deduction and induction, the logical perspective cannot bear the sole responsibility for assessing argument products, as Wenzel (1990, 19) recognizes. The logical assessment of an argument requires, at the least, that we first interpret a text as a set of reasons or grounds intended to support a claim or conclusion. That is, one must first analyze the argument content as having a determinate structure—as one or another kind of deduction, or as an inductive inference, or an analogy, narrative, and so on. If one is dealing with real arguments, this task requires a certain amount of rhetorical analysis: one must, for example, understand the particular audience context, shared tropes, figures, genre conventions, and so on (cf. Wenzel 1987, 108). Only then can one apply logical standards for appraising the degree of support provided for the claim. But one may not stop with that. Ralph Johnson (2000) has made a convincing case that appraisal of the product must also examine the argument in relation to existing counterarguments, further questions and consequences, and so on—what he refers to as the product’s “dialectical tier.” And why stop with a dialectical analysis of the product? To understand the persuasive force of arguments as products, one must return to the rhetorical perspective. It seems, then, that we can, and for some purposes must, apply each perspective to an assessment of the product. Second, we should acknowledge some plasticity in the triad of productprocedure-process. The “product” that is subjected to logical analysis can vary, depending on one’s theoretical interests. The word itself suggests a focus on the outcome of a completed process of argumentation. But argumentation theorists are also interested in the different arguments that arise on the way to that outcome. As reason-conclusion packages that can be assessed using the same methods used for assessing the final product, these intermediate arguments fall under a broader sense of “product.” Moreover, what counts as the final product can differ with analytic perspective. Rhetorical scholars often train their methods on the actual substance of written articles, whereas logical empiricists construed the relevant product as a set of observation statements abstracted from context and placed in a (probabilistic) inductive relationship to a hypothesis. Similarly, if “procedure” refers to all the ways of critically testing and discussing hypotheses, then it should encompass laboratory procedures, meetings of a research team, referee procedures, conferences, published debates, and so on. Science as a social “process” likewise has both narrower and broader boundaries: process within the research team, within a subdiscipline or problem area, within a nation, and so on. Nor are process and
27 Science as Argumentative Practice
procedure so clearly demarcated: whereas one can at least identify the product as an entity (e.g., an article) distinct from the process, formal procedures and informal processes of arguing intertwine in ways that resist separation. Third, notice that the dialectical perspective seems to include two kinds of methodical rules or procedures. As Wenzel defines it, “the dialectical perspective embraces all methodological, procedural approaches to organizing argumentative discussions. The focus . . . is on rules, standards, attitudes and behaviors that promote critical decision-making” (Wenzel 1990, 16). Wenzel seems primarily to have in mind institutionally defined procedures: “dialectical situations are consciously planned or designed. . . . [they] are often institutionalized by the creation of special forums, e.g., courtrooms, legislatures, and the regular meetings of learned societies” (ibid., 18; see also 22). This fits with the traditional association between dialectic and rules of debate (Rescher 1977). However, it leaves unclarified the relation between such institutional procedures and the natural dialectic of objection-and-reply in ordinary critical discussion. Theorists who propose dialectical models of dialogical argumentation have studied the latter context extensively (e.g., van Eemeren and Grootendorst 1992; Walton 1989, 1998). In fact, the largely tacit dialectical rules of ordinary discourse may play the more fundamental role, inasmuch as they provide the standpoint from which to criticize institutional procedures. Such criticism can lead to procedural reforms or even violations. In law, for example, a jury might consider a line of testimony relevant and thus disobey a judge’s instruction to disregard it on procedural grounds. The tacit rules that operate in ordinary critical discussion may have more in common with idealized conditions of rational discourse than institutionalized procedures. In fact, early on Wenzel (1979) construed the “ideal speech situation” elaborated by Habermas (1971/2001) as a contribution to the dialectical perspective.5 We examine this idea more extensively in later chapters. To anticipate that discussion, the ideal speech situation is a regulative ideal rather than a fully realizable “situation”; as such, it involves a set of counterfactual “pragmatic presuppositions” of rational consensus. That is, if participants in argumentation are to consider a consensual outcome rational, then they must presuppose that the outcome has been (or could be) accepted by participants who took part as equals in an open discourse free of coercion or hidden compulsions (Habermas 1990, 89; 1993, 41–57). According to Habermas, such presuppositions constitute our tacit self-understanding as participants in any argumentative practice, and thus apply both to institu-
28 Chapter 1
tionalized discourse and to our informal attempts at critical discussion in everyday life. Institutionalized rules for the conduct of discussion and debate, however, differ from such idealizations. In contrast to pragmatic presuppositions, institutional procedures are contextually determinate—and often include mechanisms for reaching closure in the face of ongoing disagreement.6 It is not that the name itself matters so much—whether we call an aspect of argumentation dialectical, rhetorical, or institutional—but that the analyses move at different levels of contextualization. They also involve different standards of criticism: a legal proceeding, for example, can be procedurally correct from the standpoint of legal standards, but involve subtle forms of coercion that violate an important dialectical idealization. Fourth, notice that the rules governing formal procedures tend to be of two sorts. Some apply to the flow of speech acts and judgment of content, that is, what kinds of statements and questions each side may introduce, when certain argumentative moves may or must be made, which argument counts as stronger, or conclusive, and so on. Rules of formal debate tend to consist of such rules. But there are also rules that assign different roles to the various participants. In courtrooms, for example, we differentiate the roles of plaintiff and defendant, counsel or attorney, judge and jury. Presumably these role divisions are designed to foster an impartial critical testing of the claim under dispute. But again, we see two different kinds of standard here. The dialectic of content, or what we might call the “intellectual dialectic,” is subject in the first instance to rules determined by inferential connections between statements, as well as by pragmatic and illocutionary obligations and commitments that different types of speech acts bring with them (e.g., in a debate, an objection calls for response; an assertion obligates one to provide justification if asked to do so). The dialectical rules that regulate the positioning of the participants, by contrast, pertain to persons: their attitudes, disposition, different powers and duties, and so on. These complications—above all the slippage between the two triads—lead me to suggest that we simply break the one-to-one alignment between productprocedure-process and logic-dialectic-rhetoric. The first triad, let us say, designates different dimensions or elements that are woven together in argumentative practices. Each of the three dimensions is publicly available for investigation, that is, each dimension presents science studies with sites or data open to description and thus available for study, regardless of one’s theoretical
29 Science as Argumentative Practice
orientation. Theoretical concerns might, of course, affect exactly how one describes or what one picks out for a given dimension (recall the different approaches to the product). We can understand then the logic-dialecticrhetoric triad as different argumentation-theoretic perspectives that employ more or less distinct sets of analytic and evaluative tools for assessing the dimensions of argumentation. But we need not limit perspectives to the traditional three. Indeed, we might be able to handle some of the complications in the categories of dialectic and procedure by recognizing something like a social-institutional perspective. Such a perspective certainly plays a dominant role in science studies. To be sure, this modification does not handle all the conceptual difficulties. As Blair (2005, 142) points out, one can easily find the same normative standard appearing in more than one perspective. There is also a potential problem of cross-perspectival conflicts that requires us to determine which perspective provides the overriding standards. The first difficulty I address in chapter 5 by showing how perspectives can be internally linked to each other; the second difficulty must await Part III. 5
A Heuristic Framework for Science Studies
I offer the above critical analysis as a way to bring the perspectivist framework closer to the kinds of analyses one finds in the study of scientific argumentation. By noticing such complications and making some additional distinctions and modifications, we render the framework more serviceable as a heuristic. Such a multidimensional/perspectival framework allows us to see how scholars bring different analytic perspectives to bear on science. These perspectival differences lead scholars to focus on different dimensions of science and to describe those dimensions differently. The different approaches to the product in logical empiricism and recent rhetoric of science provide one of the clearest examples of this. We should also expect differences in how theorists describe the procedures and processes of science. We need not settle the precise reference of these terms in advance, but can simply allow science studies scholars to speak for themselves about these dimensions and how they are to be described. In using argumentation studies this way, we allow the different approaches in science studies to further shape argumentation-theoretic categories, which are somewhat vague and ambiguous in any case. (In fact, this sort of feedback from science studies already informs the critical analysis in section
30 Chapter 1
4 above.) In illuminating science studies, the argumentation-theoretic heuristic develops in ways that take us beyond Wenzel’s model. In other words, perspectivism as a heuristic framework does not function as an architectonic, a predefined grid into which we squeeze the various initiatives in science studies. Rather, I employ these categories primarily as a set of opening questions and boundary concepts that can illuminate developments in science studies and thereby foster contacts across disciplinary boundaries. As a heuristic, this framework has one direct specific questions such as the following to the theories, case studies, and proposals in science studies:
• What does the approach say about the products of science, and what does it say about the argumentative process—the various rule-governed procedures (experimental, dialogical, and institutional) and broader social processes— from which those products emerge? • Which analytic-evaluative perspective(s) does the approach seem to take or possibly contribute to—logical, dialectical, rhetorical, social-institutional, or something else? • To what extent does the theorist address normative issues, that is, provide us with standards for critical assessment, and how do these standards relate to the particularities of actual practices: to social-institutional contexts, culture, political interests, and the like?
• What conception of cogency does the approach employ or presuppose? Each of these questions is open to further differentiation in light of findings. As a set of questions, argumentation-theoretic categories have a role to play in interdisciplinary work: they supply relevant clues for spotting ways in which science studies contribute to the study of scientific inquiry as a set of argumentative practices. I do not mean that argumentation theory is what science studies are really about, unbeknownst to many of its practitioners. Rather, I want to notice potentially relevant work in science studies for understanding science as an argumentative process, thereby opening up sites of potentially fruitful interdisciplinary interplay between argumentation studies and science studies. This heuristic is not entirely innocuous, as readers have no doubt already noticed. The introduction described some basic assumptions or ideas that inform the argumentation studies heuristic: the idea of a discriminating (nonskeptical) critical evaluation, the idea of scientific inquiry as involving argu-
31 Science as Argumentative Practice
mentative practices, and the idea of an interdisciplinary, multiperspectival study of science. Each involves some thin theoretical commitments that place specific tasks on the theorist and rule out certain alternatives. First, as I understand it, the project of critically assessing scientific argumentation requires us to take the participant’s point of view on argumentative practices seriously. The three traditional perspectives, at least, are those that experienced participants take toward their own argumentative practices: successful arguers are (more or less) aware that cogent arguments must be supported by good (logical) reasons and defended against (dialectical) challenges in a manner (rhetorically) persuasive to their interlocutors. As a critical theoretical stance, then, the perspectivist framework prohibits a global skepticism toward scientific argumentation in general. This does not necessarily require one to adopt a metaphysical realism, but it does require one to recognize certain argumentative obligations incumbent on participants (e.g., that serious objections to a claim call for a reply, that claims should be supported by good reasons). Thus the theorist must take a hermeneutic approach oriented toward disclosing the norms operative within scientific inquiry. Second, the call for an interdisciplinary approach further constrains and clarifies the role of argumentation studies. If we are interested in interdisciplinary cooperation, then hegemonic claims for a particular discipline or perspective are counterproductive. In starting with argumentation studies as a framework, then, I do not claim that argumentation theory represents the master discipline, or that it exhausts everything there is to know about scientific inquiry. Rather, the idea is that argumentation-theoretic categories are particularly well suited for certain purposes, namely for pulling together a range of approaches and disciplines for the assessment of scientific argumentation. But each of these disciplines also enjoys its own stature, objects of study, and methods, which are not reducible to their function for argumentation studies. The commitment to a comprehensive approach also means that we cannot accept reductionist claims on the part of a single perspective, focus, or discipline. Thus, certain skeptical approaches that dismiss or reduce the logical perspective to rhetoric, or to sociological explanation, are at odds with the argumentation studies framework I propose. This does not mean one cannot incorporate the empirical and descriptive findings of reductionist and skeptical approaches, but only that the specifically skeptical or reductionist thesis presents a challenge—one would presumably have to detach such theses from the
32 Chapter 1
other aspects of the approach. How this is possible I explore in the postscript to Part II. The use of such a multiperspectival framework raises questions about the coherence and integration of perspectives in a given assessment. How, if at all, are the various perspectives integrated into a critical assessment in any given case? And how does one manage to integrate, for purposes of assessment, the relevant science studies models and theories that may differ in important philosophical commitments? At the point of making an assessment in an actual case, one usually must accept some thicker theoretical commitments. Its thin theoretical commitments notwithstanding, argumentation studies as an interdisciplinary heuristic does not settle many of the substantive details of critical assessment.
2 Kuhn’s Gap: From Logic to Sociology
In this and the next chapter, I use clues from the broad categories in chapter 1 to identify the various ideas of argumentation that have arisen autochthonously within science studies after Kuhn. These clues provide a framework organized around the different perspectives—logical, dialectical, rhetorical, and social-institutional—on argumentation as a social practice whose “processes” and “procedures” generate particular kinds of “products.” With this framework as a kind of heuristic, one can discern in responses to Kuhn an emergent order in the various perspectives that scholars of science have taken on processes of scientific argumentation. More specifically, Kuhn’s analysis of science opened up a gap between logical and social-institutional perspectives, a gap that rhetorics of science attempted to bridge. The gap has been described in various ways: the rational versus the social, internal versus external history. Here I characterize the gap in terms of argumentation theory. I begin with the logical empiricist side. 1 1.1
Logical Empiricism: A Formal-Logical Perspective Logical Empiricism and the Intrinsic Merits of Argument
Logical empiricism tends to be associated with a sharp split between the context of discovery and the context of justification. Even if this represents an oversimplification of a much richer story (Nickles 1980; Richardson and Uebel 2007), there is no denying that logical empiricists like Hempel and Carnap spent considerable effort on reconstructing the rationality of confirmation and explanation as abstract formal relationships between hypothesis and evidence,
34 Chapter 2
explanans and explanandum. Whatever fell outside the purview of deductive and inductive logic they relegated to pragmatic decision and convention, hence to psychology and sociology. The problem of induction posed a major philosophical challenge to the logical empiricist project: how one could justify universal statements about nature on the basis of limited evidence (Salmon 1967). The theories of confirmation that logical empiricists developed in response to this problem can, I suggest, be read as prescriptive theories of evidential argument. Insofar as the probabilistic models of the “degree of confirmation” provide a measure of evidential support for a hypothesis, one can understand them as models for assessing the cogency of the argument products of scientific research. These theories are interesting for my purposes because they define one end of the spectrum of normative approaches to inductive arguments, namely, an approach that ties the cogency of arguments to their intrinsic formal merits— the quality of their internal structure, abstracted from the substantive content, context, and methods of inquiry. This is not to say that the logical empiricists actually used the term “intrinsic merits.” Nor do I claim that they understood their formal-logical reconstructions of confirmation as a complete theory of evidential argument. However, as logics of evidential justification, their formal models of confirmation can be read as potential contributions to a normative theory of argument cogency. 1.2
Hempel’s Qualitative Definition of Confirmation
We find one of the clearest examples of this approach in the “purely syntactical” model of confirmation that Carl Hempel advanced in his most formalistic phase. Hempel conceives confirmation solely in terms of the intrinsic formal merits of argument, that is, solely in terms of structural relationships between the evidence and hypothesis-conclusion (Hempel 1943, 1945; Hempel and Oppenheim 1945/2000; cf. Rescher 1997). I do not claim that Hempel’s syntactical model represents his more complete and considered views, which increasingly emphasized the pragmatic aspects of inquiry (see Wolters 2003). Again, my aim here is simply to take this model as setting one end of the spectrum of views about argument cogency. Hempel presents both a qualitative and a quantitative reconstruction of confirmation. He understands each as a formal definition, in the sense of a logical explication of the concept of confirmation. The qualitative definition
35 Kuhn’s Gap
tells us, in formal-logical terms, what it means for evidence to confirm or disconfirm a hypothesis. As he understands his definition, it provides a criterion (the “satisfaction criterion”) for evidential support and, more broadly, for relevance, given that both confirming and disconfirming data vis-à-vis a hypothesis H count as relevant for the testing of H (Hempel 1945). The quantitative definition tells us to what degree “a given hypothesis is confirmed by certain data” (Hempel and Oppenheim 1945/2000, 135). The point of departure for Hempel is the hypothetico-deductive (HD) model of hypothesis testing. Although the HD model is open to various interpretations (Achinstein 1985), the basic idea suffices here: to test a hypothesis H one first deduces observable consequences O from H and other auxiliary assumptions E (bridge principles or correspondence rules that link H with O, assumptions about instrumentation, experimental conditions, and so on). One then designs an experiment to determine whether O is actually observed under the assumed conditions. If the experimental results do not match (i.e., are logically incompatible with) the deduced consequence, then by modus tollens at least one statement in (H & E) is false. If O is observed, then H is confirmed to some degree. But what exactly do “confirmation” and “falsification” mean here, logically speaking? This is Hempel’s question. To answer this question, Hempel presupposes a formal-logical reconstruction of inductive arguments. For present purposes, the following reconstruction suffices: (P1) If we accept hypothesis H and conduct experiment E, then we should observe outcome O. (P2) (C)
We conduct experiment E, and observe (do not observe) outcome O. Therefore, H is confirmed (falsified).
The conclusion assumes one continues to accept O and E. Following Hempel (1945), I put aside for the moment the problem of underdetermination connected with auxiliary hypotheses; Hempel simply regarded it a pragmatic decision to reject the hypothesis rather than some other assumption or the observation. Hempel (1945) finds that previous attempts to provide criteria for confirmation encounter difficulties (e.g., the “prediction criterion” does not cover the
36 Chapter 2
confirmation of existential hypotheses1), and he offers his satisfaction criterion as a definition able to handle a range of hypothesis forms (generalizations, existential hypotheses, etc.). This criterion is not itself an argument but a formal definition of what counts as confirming evidence for H or disconfirming evidence against H. Here I am interested in how it illustrates the idea of intrinsic formal merits. The criterion depends on the idea of “developing” a hypothesis for a given dataset (see Hempel 1945, 107–112). Consider the simple example hypothesis H: all ravens are black. The “development” of this hypothesis, or (dHN), is a statement of the hypothesis for a finite set of objects. We thus obtain a deductive relationship between H and its development: (D1) From the hypothesis (H) that all ravens are black, it follows that (dHN): for this sample of n objects, if object 1 is a raven, then object 1 is black, and if object 2 is a raven, then object 2 is black, . . . if object n is a raven, then object n is black.2 One can now define the “direct” confirmation of H in terms of a second deductive relationship, namely between an observation statement O and (dHN). This states the “satisfaction” of H for the finite dataset: (D2) From the observation (O) that in this sample of n objects, object 1 is a raven and is black, . . . object n is a raven and is black, it follows that (dHN). In Hempel’s model, a logical entailment of H (stated in D1) specifies what will count as relevant evidence for testing H, and (D2) states the confirmation relationship. The HD argument can now be reformulated as follows: (P1) If (H) all ravens are black, then (dHN): in sampleN (of n objects) all ravens are black. (P2) (O): here is this sampleN of n objects, and each raven in sampleN is black. (P3) If O, then (dHN). (C) Therefore, H is directly confirmed for this sampleN. In this reconstruction, (P1) and (P3) are logically true, and (P2) is relevant. To be sure, (C) does not follow deductively from the three premises as such, but
37 Kuhn’s Gap
only if one also accepts Hempel’s definition of confirmation: If O implies (dHN), then O directly confirms H. Besides avoiding certain technical difficulties in other models, Hempel’s satisfaction model is purely formal in the strictest sense—as syntactical, it abstracts entirely from the content or object domain to which the hypothesis, its development, and its dataset pertain. This is close to Carnap’s understanding of “formal”: “A theory, a rule, a definition, or the like is to be called formal when no reference is made in it to the meaning of the symbols (for example, the words) or to the sense of the expressions (e.g. the sentences), but simply and solely to the kinds and order of the symbols from which the expressions are constructed” (Carnap 1937, 1). One of the effects of this abstraction is the well-known “paradox of the ravens”: if like Hempel we insist that whatever confirms a hypothesis confirms its logical equivalent, then H above is confirmed not only by the observations of black ravens, but by nonblack nonravens and by black nonravens as well (Hempel 1946). Because confirmation is defined formally, the associated ideas of relevance and support expand to include data that our intuitions tell us are irrelevant for testing H. Hempel seems willing to accept that our contrary intuitions on this point are simply misguided, though he also suggests that one can eliminate the counterintuitive result by bringing in background information (Hempel 1945, 18–21). Some theorists have argued that the latter move can be reconstructed in terms of a quantitative approach to confirmation that would remain formalistic (Maher 1999, 57–58; cf. Kyburg 1970, 166–168). Other limitations with this model are well known. The conclusion follows only if one accepts the auxiliary assumptions. In science many of these assumptions are theoretical and instrumental, but even in our simple example, we make assumptions about the conditions of observation (e.g., proper lighting), the identity of the ravens (we assume we observe ravens and not something else), and their state (e.g., we assume none are diseased or unusually dirty). As the logical empiricists point out, observation statements are themselves hypotheses—these do not provide “an absolutely fixed and unshakeable basis” for testing scientific theories (Hempel 1937/2000, 52). At the least, the competence of the observer, or various auxiliary assumptions that go into the experimental observation, might be doubted or overturned by future advances. Thus Hempel (1945, 112–113) speaks of falsification and confirmation as “relative” to the observation in (P2), which is a hypothesis in its own right, confirmed relative to observation reports.3
38 Chapter 2
1.3
A Purely Syntactical Degree of Confirmation
As a qualitative definition of confirmation, Hempel’s satisfaction model tells us nothing about the strength of confirming evidence, that is, the degree of confirmation.4 Although he eventually came to consider quantitative formalizations of the degree of confirmation as dubious (Hempel 1966, 45–46), he did make the attempt (Hempel and Oppenheim 1945/2000). We can read this in argumentation-theoretic terms as a further elaboration of the idea of evidential support, specifically as a definition of the relative strength of support. Here I primarily want to explain why Hempel and Oppenheim regard their quantitative model no less than the qualitative one as purely syntactical. The Hempel-Oppenheim model, like other objective Bayesian models such as Carnap’s, relies on a determinate distribution of individual objects or states of affairs across the possible kinds (state descriptions) generated by the descriptive vocabulary of the chosen language. But unlike Carnap’s model, which makes this assignment by an a priori fiat, Hempel and Oppenheim’s frequency model determines it in relation to available evidence (Hempel and Oppenheim 1945/2000, 158). More specifically, they define the degree of confirmation dc (for a hypothesis H, given evidence E) as a probability measure “pr” determined by the optimum distribution, ∆E, relative to evidence E: dc(H, E) = pr(H, E, ∆E) ∆E is the most likely frequency distribution of objects across the possible descriptive states, given the available evidence. That is, ∆E is the distribution that would make E most probable in comparison to the probability of E on other possible distributions. For example, if ten random drawings (with replacement) from an urn have yielded 5 black and 5 red balls (the evidence), then the most likely ∆E is (0.5, 0.5). That is, in this urn—as a “universe” exhausted by two states (red, black)—the evidence so far tells us that most likely, objects are distributed equally in each state. In this simple case, the strongest predictive hypothesis about the next drawing follows as a matter of course: that the next drawing will yield a black ball has a dc = 0.5. Naturally, the degree of confirmation for hypotheses about domains described by more complex vocabularies will not be so straightforward.5 Even for simple universal hypotheses as in the raven example, one must reckon with indefinitely large object domains; in such cases the probabilities are limit values obtained for an indefinitely increasing sample size.
39 Kuhn’s Gap
We can think of dc(H, E) as a definition of the inductive strength of arguments for H on the basis of evidence E. Assessing such strength involves three main steps:6 (1)
The relevant data for testing H is E.
(2)
E is most likely on frequency distribution ∆E.
(3)
Therefore, dc(H, E) = pr(H, E, ∆E).
If we may use the qualitative definition to pick out what counts as relevant data in (1), then all three steps follow simply on the basis of logical and mathematical calculation. The details of the requisite calculations in (2) and (3) need not detain us here. In fact, the quantitative definition has a number of problems,7 and Hempel and Oppenheim (ibid., 160) are unsure whether it agrees with the qualitative concept on the meaning of confirmation and disconfirmation. The authors concede the limits of their analysis, which essentially captures the “statistical version of the principle of induction” that a rational gambler would employ, assuming relative frequencies remain “fairly stable” over time. Thus the model “refers only to one of several factors which enter into an objective appraisal of the soundness or reliability of an empirical hypothesis.” A more complete concept, on their view, would have to take into account the number and variety of evidential tests of the hypothesis (ibid., 159–161). In any case, the important point is that the quantitative model, like the qualitative definition, explicates confirmation in terms of intrinsic formal merits. As Hempel and Oppenheim (ibid., 158) put it, their model is purely syntactical in the sense that, once the evidence is given, the degree of confirmation (dc) of H by evidence E “is completely determined by the formal, or syntactical, structures of the two sentences [H and E] alone.” That is, the degree of confirmation follows analytically from a given vocabulary L for describing the universe of objects under investigation, a set of accepted observation statements in L about those objects, and the mathematical calculations required to arrive at ∆E and pr(H, E, ∆E). The advantage that Hempel and Oppenheim see in formalistic reconstructions is the objectivity it allows. With a quantitative measure, one does not have to rely simply on the scientist’s intuition for estimating how strongly the
40 Chapter 2
evidence supports a hypothesis. Relying on intuitions “would be highly unsatisfactory; for first, it would clearly jeopardize the objectivity—in the sense of intersubjectivity—of scientific procedure.” In addition, an intuitive approach would misconstrue the “widely accepted” objectivist view of confirmation, namely the view that “statements about confirmation assert nothing regarding an observer’s subjective appraisal of the soundness of a hypothesis; rather, they concern a certain objective relation between a hypothesis and the empirical evidence with which it is confronted.” This relation “is of a purely logical character in the sense that once a hypothesis and a description of certain observational findings are given, no further empirical investigation is needed to determine whether, or to what degree, the evidence confirms the hypothesis” (ibid., 136). In other words, assessment becomes entirely mechanical and impersonal. 1.4
Logical Empiricism and Pragmatic Context
In focusing so closely on Hempel’s most formalistic phase, my aim is to clarify the formalist extreme of the various perspectives on the argument product and its merits. But in doing so, I do not mean to imply that Hempel denied the importance of additional factors in real scientific argumentation, or that he considered such factors as necessarily nonrational. Indeed, the label of “pragmatic empiricist” might better describe the trajectory of his theorizing (Wolters 2003). In any case, he realizes that the “logical phase” of hypothesis testing—the various deductive and probabilistic moves that confront hypothesis with observation statements and determine degree of confirmation— presupposes a number of pragmatic moves that complicate the formal model of argumentative merits. Interestingly, in laying out a more complete picture Hempel (1945, 114–119) hints at a procedural-process account of argument construction, in which the logical analysis of the argument product is sandwiched between two pragmatic phases. In the first phase, scientists conduct experiments, carry out observations, and gather evidence; after making the logical analysis (phase two), in the third phase they make a pragmatic decision to accept or reject the hypothesis on the basis of the evidence. The rules for acceptance and rejection are not, in other words, immediately identifiable with the logic as such. Yet Hempel grants that acceptance rules might be open to rational reconstruction (ibid., 117–119). This suggests that formal logic does not, after all, exhaust the rationality of induction.
41 Kuhn’s Gap
Other logical empiricists also recognize this pragmatic context in which the logic of justification is grounded. The vocabulary itself constitutes a major decision that lies outside formalization (Carnap 1949, 124–125; cf. Hempel 1945, 118). Moreover, as noted earlier, scientists choose whether to accept or reject disconfirming evidence; observation statements do not automatically thrust themselves on scientists but must be accepted by the science community. To label such decisions as “pragmatic” implies that scientists make them on the basis of their usefulness for a particular set of interests—hardly an intrinsic formal merit of the impersonal sort we saw above. To be sure, the major desideratum for a language of science—at least an observation language—is its conduciveness to the intersubjective empirical repeatability of observations (Carnap 1932/1995). That granted, the role that theorists like Carnap and Hempel attribute to pragmatic interests comes surprisingly close to the social constructivism that set off the science wars in the 1970s (cf. Friedman 1998, 2003). For Carnap the problem is especially acute. Because he makes the degree of confirmation depend on a frequency distribution that is determined a priori, in advance of empirical work, he must situate his logical reconstructions not only with respect to the pragmatic choice of vocabulary, but also with respect to the choice of “inductive method,” which determines the particular mathematical weightings in the distribution of possible outcomes (Carnap 1952). These choices are not trivial, for they determine the degree of evidential support that a given hypothesis has. However, if the choice is pragmatic, what makes it epistemically rational, and not simply rational in view of particular social interests? (Cf. Salmon 1967, 75–79.) To close this section, let me summarize some implications of Hempel’s logical reconstruction of confirmation for the evaluation of cogent evidential arguments:
• Hempel’s model represents a highly focused logical perspective on argument cogency. Like other logical empiricists, he takes a prescriptive approach focused on the intrinsic formal merits of argument products, abstracted from their content, context, and the pragmatics of inquiry. • To apply his model, however, one must first interpret the content of actual arguments (e.g., in articles, review essays, and the like) by reconstructing the argument as a set of sentences formulated in a language amenable to formalmathematical manipulation.
• What count as intrinsic merits are the virtues of relevance, support, strength of support, and valid structure: premises must be relevant, which is to say they must
42 Chapter 2
either support a hypothesis or support its rejection, and the moves that take one from premises to conclusion must accord with the demands of deductive and inductive logics, which also determine the strength of evidential support.
• These virtues are intrinsic formal merits in the sense that they are identifiable entirely through the syntax of the reconstructed argument. Thus anyone who understands how to translate an argument into the requisite logical calculus can determine whether that argument has the merits of relevance, support, and validity. Intrinsic merits thus presuppose a double abstraction: not only from the context of discovery, but also from the particular substance of the premises. • Arguments also have pragmatic merits insofar as the evidential premises are intersubjectively acceptable to scientists: formulated in an intersubjectively useful language and well supported by actual observation reports. Concern with the psychological persuasiveness of concrete arguments would most probably fall within this pragmatic dimension. By making the acceptability of premises a pragmatic or conventional matter, Hempel, like other logical empiricists, injects a context-dependent, sociological element into his account of scientific inquiry. He thereby introduces a division between the logical and pragmatic aspects of cogency. (Its differences with logical empiricism notwithstanding, Karl Popper’s falsificationism makes a similar move.) So long as the language in which observational premises were formulated was neutral vis-à-vis competing hypotheses, and so long as the logical framework of comparison remained purely formal, then the pragmatic side of inquiry did not undermine the possibility of an impartial comparison of the relative strength of the arguments for one hypothesis over its competitors. By vividly displaying the fragility of these assumptions, Kuhn turned this division into a contentious gap in the analysis of scientific argumentation. 2
Kuhn’s Challenge
Kuhn’s 1962 Structure of Scientific Revolutions (3rd ed., 1996) attacked the logical empiricists’ obsessive focus on the context of justification to the detriment of the discovery process, which empiricists regarded as resistant to logical treatment.8 In Kuhn’s eyes, the models of justification developed by logical empiricists and falsificationists proved historically inadequate even as a normative account of science (see Hoyningen-Huene 1993, 245–252). In what follows
43 Kuhn’s Gap
I want to recall these well-known developments from an argumentationtheoretic point of view. From that perspective, Kuhn had his sights trained on the philosophy-of-science version of the formal-logic tradition that Toulmin and Perelman and Olbrechts-Tyteca had criticized only a few years earlier— the attempt to capture the rationality of science in terms of deductive and inductive logics that by and large abstracted from the contexts and contents of the different sciences. By calling attention to the dynamics of scientific discovery, Kuhn made scientific practices rather than abstract rules of method and logical inference the focal point of analysis. He thereby brought the substantive, contextually embedded argumentative processes of science to the fore. In doing so, he indirectly helped to rehabilitate the American pragmatists’ concern with the process of inquiry.9 To recall Kuhn’s challenge, I start with his account of “normal science,” that is, science as routinely practiced according to a particular disciplinary paradigm (Kuhn 1996, chaps. 2–5). Kuhn’s analysis of normal science provides a sophisticated evolutionary account of scientific reasoning based on examples (Nickles 2003a). In contrast to logical empiricists, Kuhn held that practitioners of normal science looked not to formal models of inference for their normative standards, but rather relied on skills akin to a craft expertise, which they first acquired by learning how to recognize and solve the specific types of problems that constituted the core “exemplars” of the current paradigm. In argumentation-theoretic terms, normal science proceeds in a casuistic manner: scientists solve problems from case to case by modeling them on the paradigmatic exemplars with which they have familiarity (cf. Barnes 2003; Nickles 2003a). Familiarity with such exemplars enables scientists actually to use the various equations, instruments, and other conceptual and methodological resources that make up the broader “disciplinary matrix” of the paradigm (Kuhn 1996, postscript). The paradigm thus constitutes a kind of disciplinary worldview with practical consequences: the paradigm is at once a way of seeing problems and evidence and a way of proceeding in the solution of those problems. Given the manner in which paradigms shape the very perception of problems and evidence, Kuhn’s treatment of major scientific change—the so-called revolutionary science that leads to a new paradigm—could hardly avoid controversial implications (see Lakatos and Musgrave 1970). If argumentative practices in ordinary science depend so heavily on paradigmatic presuppositions for their intelligibility and cogency, then arguments in which those very presuppositions are at issue seem to lose their rational footing. Rationally
44 Chapter 2
cogent argumentation across paradigms becomes problematic, even dubious, raising the threat of relativism. This is especially acute for the traditional view of scientific revolutions as exemplary of scientific insight and reason. Kuhn only exacerbated matters by declaiming the “incommensurability” of competing paradigms. This thesis seemed to have a number of distinct meanings: that paradigms conceive the relevant problems and standards differently, that scientists working in different paradigms see the world (or data) differently, and so on. According to Zammito (2004, chap. 3), philosophers tended to read Kuhn in light of the philosophy of language. From that perspective, incommensurability involved semantic holism, such that the meaning of terms changed across paradigms. This apparently relativist view undermined the possibility of impartial testing of competing theories by recourse to observations in a theory-neutral vocabulary. The philosophical reception of Kuhn had a number of unfortunate consequences, one of which was the focus on a dubious philosophy of language that missed Kuhn’s more enduring contribution: his emphasis on science as a practice. The problems posed by revolutionary change are first of all practical in nature: as Rouse (2003, 113) explains, more important than differences in beliefs are “differences in how they [i.e., scientists] comport themselves, what is expected of them, and what is at issue and at stake in their dealings with things.” Consequently, scientists talking across paradigms “cannot grasp the point of what the others are doing or recognize the force of their arguments.” As I argue below, this kind of misunderstanding issues precisely from the values that scientists share across paradigms. I do not mean to deny that Kuhn endorsed semantic incommensurability. Indeed, many of his claims in Structure support the radically relativist and skeptical picture of science that such incommensurability seems to entail. But other statements show that Kuhn wanted to avoid such radical conclusions about the rationality of revolutionary change. As a result, Structure creates a curiously Janus-faced impression of scientific change, particularly when read together with his 1969 postscript (Kuhn 1996, 174–210). Borrowing Kuhn’s own metaphor, Pera (1994, 8–10) likens Kuhn’s model of science to a gestalt figure. Kuhn’s “Duck,” as it were, foregrounds the nonrational and unpredictable aspects of science, telling us for example that revolutionary science involves conversion experiences and acts of faith (Kuhn 1996, 148–151, 157ff., 204). Kuhn’s “Rabbit,” on the other hand, foregrounds the reasonable character of science—the idea that scientists are ultimately persuaded by reasons, that such reasons include cognitive values, and that science develops increasingly
45 Kuhn’s Gap
powerful problem-solving techniques and predictive accuracy (Kuhn 1996, 158ff., 184–186, 199, 205–206). As it turned out, Kuhn preferred the Rabbit (Kuhn 1970b, 259–266; Sharrock and Read 2002). Nonetheless, the tensions in Kuhn’s text reveal the ambitious nature of his project: to relocate the context of discovery inside the context of justification, and thus to hold together two images of science that resist easy assimilation in a single perception. To do so, Kuhn must provide an account of scientific change that combines cogent reasons and social psychology without denying either its due. For this purpose the logical empiricist model is inadequate: scientific arguments “cannot be made logically or even probabilistically compelling” for scientists outside the paradigm, so logic cannot account for reasonable theory change (Kuhn 1996, 94). To rescue the reasonableness of science, then, Kuhn alludes to persuasion, good reasons, and arguments piling up over time: “because scientists are reasonable men, one or another argument will ultimately persuade many of them” (ibid., 158; see also 152ff., 204). Although deep theoretical shifts in science cannot be justified in formal-logical terms, Kuhn tells us, they do involve good reasons and shared cognitive (or epistemic) values, which include (but are not exhausted by) accuracy, consistency, explanatory scope, simplicity, and fruitfulness (Kuhn 1977, 321–322). The motivating reasons can also include “more subjective and aesthetic considerations” (Kuhn 1996, 156; see also 155–159, 165–167). So if reasonableness means having plausible reasons for a belief or choice, then scientists are reasonable in the minimal sense that whatever their sociological or psychological background motivations, in interparadigmatic debate they can appeal to reasons and evidence of one sort or another that make sense in light of shared values. In a revolutionary shift, scientists have good, but not logically compelling, reasons for choosing one paradigm over another (Kuhn 1996, 185, 199–200). Kuhn’s attempt at integration thus appeals to revolutionary science as an argumentative process, in which reasons are persuasive in virtue of their psychological impact on an audience or addressee. This move promises to combine psychology and reasonableness, inasmuch as we can think of the latter as a kind of psychological receptivity, an ability to be moved by reasons. But Kuhn must say more. If “persuasive arguments” are to move scientists actually to shift paradigms, then scientists must be more than minimally reasonable in this psychological sense: they must also make comparative judgments by sorting out the better arguments. As Kuhn himself recognizes, “To persuade someone . . . is to convince him that one’s own view is superior and ought therefore
46 Chapter 2
supplant his own” (1996, 203). This suggests that scientists must exercise a cognitive capacity that responds to normative or evaluative standards of cogent argument. It is precisely here that Kuhn’s notion of paradigm generates a distinct kind of underdetermination that arises from the difference between epistemic standards and variable psychological response across a group.10 As HoyningenHuene (1993, 249–250) puts it, communal values are individually shaped. Consequently, differences in interpretation and weighting can lead different scientists to assess the same evidence differently. In fact, we misconceive the problem here if we simply reduce it to individualism. Persuasion for Kuhn does not primarily refer to an individual process; rather, he is interested in the overall movement of the science community (i.e., the relevant subdiscipline in which a paradigm shift occurs).11 This suggests that the cognitive side of reasonableness rests precisely on the communal character of normative standards of evaluation, as distinct from their individual psychological effect. However, individual variation reveals something about the values themselves: as plural, vague, and open to interpretation, shared epistemic values are not sufficiently determinate to single out the stronger argument in revolutionary debates. Now the problem of underdetermination arises precisely at the communal level: how do such values add up to better reasons that are “ultimately decisive for the group” (Kuhn 1996, 199)? The key to Kuhn’s solution lies in his remarks on the temporal process of persuasion. He suggests that one side wins out because over the course of time it can meet challenges and accumulate arguments to the point where dissent becomes unreasonable—“only blind stubbornness can at the end account for continued resistance” (Kuhn 1996, 204). Although an individual’s fully internalizing a new paradigm presupposes a conversion based on “neural reprogramming,” being initially persuaded by the arguments for a view need not depend on such adherence and the reprogramming that comes with it (ibid., 201–204). The possibility that outsiders can be initially drawn into a paradigm precisely by reasons and evidence gives sense to Kuhn’s claim that cogent arguments are “ultimately decisive” for scientists. The persuadable outsiders, Kuhn goes on to note, tend to be the younger members of the profession who have not yet invested the bulk of their careers in defending the old paradigm (ibid., 90). Hoyningen-Huene (1993, 239–245) sheds further light on this process by distinguishing three types of reasons that come into play at different points in the process, for different players. The first adherents to a new paradigm are
47 Kuhn’s Gap
motivated, negatively, by the anomalies that resist solution within the older paradigm; positively, they have a kind of nonrational faith in the future promise of the new approach. As these adherents work out the new paradigm, however, its superiority in solving long-standing anomalies and other problems gradually becomes evident to more scientists—our persuadable outsiders—who might respond to aesthetic considerations as well. In the final phase of the transition, the epistemic virtues of the new paradigm have become sufficiently clear, on the basis of the evidence, so as to swamp individual differences in the evaluation of evidence and weighting of epistemic values; at that point further resistance simply betrays “blind stubbornness.” Although paradigm shifts can incur cognitive “losses” (e.g., reduced explanatory scope, abandonment of certain problems), Kuhn (1996, 169–170) takes such shifts as progressive on the whole, mainly in terms of problem-solving capacity and quantitative accuracy. This analysis suggests that in hindsight and in the long run, scientists can see that the new paradigm has the more cogent arguments in its favor. However, by the time this collective judgment of relative cogency comes to predominate within the science community, everyone (but the “unreasonable” and usually aged holdouts) is working within the new paradigm, which settles questions of the interpretation of epistemic values (cf. Kuhn 1996, chap. 11). This retrospective judgment, then, is not made from an impartial standpoint: the deck is now stacked in favor of the successor. Thus the decisive moment—when scientists on the whole actually compared arguments from within each of the competing paradigms and chose the new—has already passed. The interesting question for an account of the argumentative process, then, is how persuasion occurs within the transitional phase itself—the microprocesses that generate agreement on the new paradigm. Kuhn did not elaborate a full answer to this question in Structure, whose schematic “phase model” focuses primarily on the macrostructure of scientific development.12 But what he did say implies that an adequate answer must do justice to at least two points. First, the above description of the temporal process of change suggests that the shift in the community’s allegiance is a socially distributed effort, led by risk-takers willing to gamble on the new paradigm’s future promise and later joined by others (perhaps predominantly, at first, the younger members) until the bulk of the active community accepts the new paradigm. As a socially distributed process, the shift in allegiance does not result from a uniform application of a single set of rules of logical method. Although the individual choices for the new paradigm might all accord with shared epistemic values,
48 Chapter 2
the specific considerations that actually sufficed in each individual case to motivate allegiance are various—not only different weightings of shared values but also different career-oriented calculations and the like. Second, Kuhn’s turn to practice implies that the trend within a community toward a new paradigm involves an emergent collective judgment that the new paradigm presents the more interesting and fruitful framework for doing science. Scientists are not simply assessing the logical force of a set of evidential and aesthetic statements that argue for one paradigm or its competitor. Although the arguments involved in a paradigm shift involve linguistic (and other representational) structures, such arguments draw their cogency for practitioners largely from the scientists’ experience of the research activities, the laboratory and theoretical practices, that adherence to the new paradigm enables. Thus, the evidence that accrues in favor of the new paradigm is not simply a growing set of evidential statements but a growing body of scientific practice, that is, a widening circle of scientists who discover the possibilities the paradigm opens up for their own research. The foregoing considerations explain Kuhn’s replacement for logical standards as the normative basis of good science, namely the science community itself as a group of practitioners. That is, when we ask Kuhn for a normative account of cogent science—what makes a given theoretical innovation count as a cognitive gain, in the midst of ambiguous and competing epistemic values—he appeals to the social status quo in the community: “scientists should behave essentially as they do if their concern is to improve scientific knowledge” (Kuhn 1970b, 237). This imperative use of “should” imbues the institutional mechanisms of science with a normative status: Some of the principles employed in my explanation are irreducibly sociological. . . . In particular, confronted with the problem of theory-choice, the structure of my response runs roughly as follows: take a group of the ablest available people with the most appropriate motivation; train them in some science and in the specialties relevant to the choice at hand; imbue them with the value system, the ideology, current in their discipline . . . and, finally, let them make the choice. . . . Whatever scientific progress may be, we must account for it by examining the nature of the scientific group, discovering what it values, what it tolerates, and what it disdains. (Kuhn 1970b, 237–238; cf. 1970a, 21)
There is a relatively innocuous reading of this passage, to be sure. If “progress” in science means that a later theory rests on more cogent arguments than its predecessors, then one might read Kuhn as making a statement about the most favorable institutional context of discovery: he appeals to the social
49 Kuhn’s Gap
autonomy of science, the idea that science generates more cogent theories, and thus progresses best, when left to itself. That idea has a long pedigree (Proctor 1991) and is quite compatible with the logical empiricist distinction between discovery and justification. But Kuhn wants to bring the process of discovery into the context of justification and thus rejects purely formal, process-independent measures of cogency. And given his understanding of paradigms as shaping the individual interpretation of epistemic values, he also lacks substantive measures, an informal logic, that could pick out the more cogent theory from a perspective apart from the actual process of scientific development—the collective shift that actually takes place in the allegiances of practitioners.13 Kuhn’s model thus requires us to read the above passage as a more radical statement, as a call for substituting a social-institutional perspective on the process of argumentation for the logical perspective, whether formal or informal, on its products. The community as a socially organized whole provides the argumentative norm for good science. More precisely, the socially organized body of practitioners constitutes the authority whose collective judgment certifies which paradigm better realizes epistemic values over all. This collective judgment thus operates as the norm. As a result, Kuhn seems to leave us (nonscientists) with a deferential trust in the sociology of the science community—though commentators debate this point.14 In any case, if the authority that normatively governs theory choice in science is sociologically constituted, then—deference notwithstanding—Kuhn’s model invites social scientists to set up shop inside the science community. 3
The Sociological Response: Two Approaches
Having rejected the formal-logical perspective on scientific argument as inadequate, Kuhn opened the door to social-institutional perspectives on scientific argumentation as a social process. As a perspective on argumentation and not simply on the social organization of science, Kuhn’s social perspective encouraged sociologists to go beyond Merton (1973) and develop substantive sociologies of scientific knowledge (SSK).15 Kuhn’s approach suggested that the content of science was deeply intertwined with the problem of social order within the science community: in collectively solving cognitive problems scientists also solved problems of action coordination and ongoing social organization (Barnes 1982, 2003). As already mentioned, paradigm shifts involve a certain organization of efforts and persons, and once complete they
50 Chapter 2
determine who belongs to the community of active researchers, how novices are trained, and so on. Moreover, Kuhn’s efforts to read history on its own terms encouraged historians and SSK theorists to reject the traditional historiography of science that reads the past as mere preparation for the present—as a history of progress in which superstition gives way to an accumulating stock of truth. To carry out this project, SSK theorists generally adopted a “symmetry principle,” the requirement that one apply the same descriptive and explanatory methods equally to the content of putatively good and bad science. For many SSK practitioners, symmetry went hand-in-glove with a relativist, social constructivist approach to scientific knowledge (Barnes 1983; Barnes and Bloor 1982; Collins 1981; Pinch 1993). What relativism entailed varied with the approach. According to one of the sharper formulations (Collins 1982, 302), relativism meant that the content of good science was finally determined, not by rational method probing natural reality, but by contingent social circumstances: in different circumstances, “correct scientific method” could yield different answers to the same question. Even in its milder forms, the meaning and implications of relativism proved sufficiently contentious to set off the rationality debates in science studies (Hollis and Lukes 1982; Brown 1984; Radder 1992). The various SSK initiatives split up into diverse camps, however. I close this chapter by briefly describing two different approaches that sociologists took in the wake of Kuhn, the rule-skeptical approach of the Strong Program and the particularist approach of certain ethnographers, above all ethnomethodologists. As post-Kuhnian sociologies of science, these two research programs hardly exhaust the range of initiatives (see Hands 2001; Pickering 1992a,b; Knorr-Cetina and Mulkay 1983). I focus on them for two reasons. First, their opposition further displays the influence that unresolved tensions in Kuhn have had on science studies. Second, these two approaches pose particularly challenging and instructive obstacles to interdisciplinary cooperation for the social-critical approaches I focus on in later chapters. As an introductory background to the more detailed analyses to come, the following is confined to some broad strokes. Both approaches assume we understand science adequately only when we see how its content is connected with social order—with the conditions of ongoing social organization and intelligible interaction. Moreover, each understands its approach as Wittgensteinian in spirit (Lynch 1992; Bloor 1992). The
51 Kuhn’s Gap
rift between them stems in part from ambiguities in the Wittgensteinian tradition to which Kuhn alluded. Specifically, rule skeptics and particularists draw different lessons from Wittgenstein regarding the proper sociological understanding of the normativity of scientific practices, that is, the status of rules in scientific practice. This difference in interpretation is aligned with opposing research goals.16 Members of the Edinburgh School call for a “Strong Program in the Sociology of Scientific Knowledge” committed to the causal, typically macrosociological, explanation of scientific change in terms of various social conditions.17 They align this agenda with a “finitist” theory of meaning (Barnes 1982, 27– 40; 1983; Bloor 1983; 1997, chap. 2). According to finitism, the proper extension of a concept (or rule) to a new case is not uniquely determined by an “intrinsic” meaning of the concept or by past practice, the record of previous applications. The meaning of a concept is thus finite in the sense of being coextensive at any given point in time with the actual cases, or solved problems, on which the practitioners reached agreement regarding the application of that concept. Applied to scientific argumentation, the result is a kind of indeterminacy at the growing edges of a research domain: how one ought best to extend science is underdetermined by inherent theoretical content, past usage, and evidence (natural phenomena). To explain theory change, then, one must look to sociological modes of causality: interest constellations, distributions of power, social networks, and the like. Unlike Kuhn, who looked primarily to social dynamics within the sciences, sociologists like Barnes and Bloor quickly saw that even good science—what would count as progressive in Kuhn’s sense—is not isolated from broader social and cultural influences. The possible social influences that determine theory choice in any particular controversy stretch from narrow professional (and technical) goal-orientations at one end of the spectrum to political and broad class interests at the other (MacKenzie and Barnes 1979). In support of this thesis, Strong Programmers have generated a vast literature of sociological case studies, primarily of controversies, in which scientists clashed over the direction of inquiry and theorizing (Shapin 1982). For example, MacKenzie and others have sought to show how the development of statistics in Britain was influenced by the interest of leading proponents in social eugenics (MacKenzie 1978; MacKenzie and Barnes 1979). Shapin and Schaffer (1985) argue that the debate between Hobbes and Boyle over the methodology of natural science was aligned with the protagonists’ social-political
52 Chapter 2
commitments: the intellectual contest was simultaneously a struggle over social order in science and the sociopolitical order in seventeenth-century England. The rule skepticism inherent in this program lies in the rejection of explicit norms of rational method (rules, theoretical concepts, methods, etc.) as providing the real, that is, causally effective, basis of theory change in science. For rule skeptics, the “real sources of constraint” in scientific practice come not from formulated rules but from the naturalistic complex of biological, psychological, and sociological conditions that operate causally to keep social practices on track and to direct practitioners toward consensus (Bloor 1997, 20). Thus rule skeptics downplay the normative dimension of scientific practices as explanatory of theoretical development in science. Although they acknowledge that scientists make and respond to arguments, perceptions or judgments of cogency ultimately express established conventions or “social factors”—as Bloor (1984) puts it, “epistemic factors really are social factors.” Ethnomethodological particularists reject rule skepticism as a misreading of Wittgenstein’s insight into the internal relation between formulated rules and social practices. Particularists agree with Strong Programmers that the normative standards that guide the science community cannot be fully represented by general norms of rationality: no set of formulated rules of method, however laden with qualifications and provisos, can fully anticipate the myriad contextual conditions of possible application. Unlike Strong Programmers, however, they do not draw the rule-skeptical conclusion that one cannot understand the dynamics of scientific practices through the normative aspects of such practices. On a particularist understanding of the relation between rules and practice, there is no need to go beyond the normative self-understanding of practitioners and invoke a causal explanation of the development of science. They also reject philosophical analyses that attempt to prescribe normative standards for science from on high: to assess the reasonableness of a major theoretical innovation in science, one must focus on the substance of the particular case and “engage in the scientific argumentation itself, not in some separate philosophical adjudication” (Sharrock and Read 2002, 67–68). On this view, the internal normative dynamics of scientific practices, including their practical know-how, equip properly trained participants with all the guidance they need to make reasonable judgments about the best way to extend their practice. The research task for ethnomethodologists of science, then, is simply to notice and describe the various concrete rationalities that participants themselves employ and that are more or less unique to each local practical context (e.g., a particular biology
53 Kuhn’s Gap
laboratory; see Lynch 1985). To accomplish this, they take pains to immerse themselves in the rationalities of those they study and thus try to avoid imposing methodological assumptions that diverge from those that competent practitioners employ in the local context. This description-centered research agenda has led some ethnomethodologists to a radically contextualist, antitheoretical approach to scientific practices (e.g., Lynch 1993). Although ethnomethodology takes the normative dimension of science seriously, the restriction to participant-level description leads to a principled “indifference” on the part of ethnomethodologists: they do not prescribe standards of reasonableness for science. Because ethnomethodologists are not rule skeptics but take care to describe operative norms, we might call their approach “descriptive-normative” to distinguish it from prescriptivenormative approaches (e.g., logical empiricism). The former would seem to preclude a social-critical attitude toward science. Not that the Strong Program is known for its social-critical perspective on science. Although its explanatory narratives can undermine a naive respect for scientific authority, for the most part the Strong Program has confined itself to sociohistorical analysis and empirical case studies. The antiprescriptive character of SSK in both of these variants widened the gap opened by Kuhn between logical and social-institutional perspectives. For economy of expression I shall call this “Kuhn’s Gap.” I use this expression in a double sense, as referring to both a theoretical and a disciplinary situation. Theoretically, this “gap” refers to the unmediated opposition between two perspectives on scientific argumentation: on the one side, analyses of cogent argument in terms of formal or substantive properties of the product of argumentation; on the other, a focus on the social-institutional contexts and process from which cogent arguments emerge. The social-institutional analyses might include a prescriptive claim of the sort that Kuhn makes about how inquiry ought to be organized; alternatively, it might, as in the case of SSK, restrict itself to descriptive and explanatory claims about which arguments in fact carried the day and how a given outcome resulted from a given set of socialinstitutional conditions. The theoretical gap between these two perspectives arises from the tendency of their proponents to pit them against one another, as though we must choose between them without exploring the space lying between them for the possibility of mediations. Although Kuhn does not deserve the sole blame for his reception among sociologists—indeed, he vigorously rejected SSK (Zammito 2004, chaps. 5–6)—I do not think we should let him entirely
54 Chapter 2
off the hook for his role in creating this gap. He desperately wants a conception of informal argument that can save the rationality of revolutionary science without invoking logical compulsion. But instead of looking to the dialectical and rhetorical traditions, his dubious philosophy of language forces him to move to the macrodynamics of scientific institutions. He does, to be sure, spend considerable ink on issues of interpretation and understanding across paradigms, and his analysis of epistemic values provides a basis for a rhetorical perspective (as I show in the next chapter). But rather than develop that perspective, at the end of the day Kuhn rescues the rationality of science by displacing reason from logic to autonomous social-institutional organization. The disciplinary sense of Kuhn’s Gap refers to the situation within science studies as a field, where the opposition between perspectives became associated with a gap between disciplines—primarily between philosophers committed to logical analyses of evidential inference, on the one side, and sociologists interested in socially situated, rhetorically effective argumentation, on the other side. To be sure, this considerably oversimplifies the situation. On each side of the disciplinary gap, one can see forays into the middle ground. Well before Structure, Popper tied his method of conjecture and refutation to the dialectical tradition (Popper 1989, 313n4; cf. Lakatos 1976). The rational-historicist responses to Kuhn (e.g., Laudan 1977; Lakatos 1970) would later develop this dialectical perspective in more substantive terms. On the other side, SSK has often linked its social-institutional perspective with ideas of rhetorical argumentation (e.g., Shapin and Schaffer 1985). But the mutual acrimony one finds in the post-Kuhn literature on both sides testifies to a real gap that called out for more effective mediation (see Zammito 2004, chap. 5). Although the debates over scientific rationality were often contentious, they have spawned some interesting bridging attempts (e.g., Brown 1977; Fortun and Bernstein 1998; Longino 1990, 2002; Solomon 2001). If we understand the disciplinary gap in terms of argumentation theory, thus as a problem connected with a theoretical gap between perspectives on cogency, then the prospects for greater interdisciplinary cooperation in science studies depend on a sufficiently capacious model of cogent scientific argumentation, a multidimensional framework that can mediate the oppositions described above. To that end, one must not only negotiate differences in favored perspectives (logical, dialectical, etc.) but also different understandings of what it means to offer a theory of cogency. I close with some brief remarks on how I propose to deal with the latter issue.
55 Kuhn’s Gap
Start with Hempel. His analysis of confirmation, taken as a view about cogency, purports to provide a definition in the sense of a logical explication of what it means for inductive arguments to be cogent. As Hempel understands it, such a definition operates as a prescriptive criterion: a way of determining whether, and to what degree, evidence supports a hypothesis. Hempel’s definition thus tells us whether or not an argument qualifies as cogent in virtue of certain intrinsic formal features, which in turn justify our regarding the argument as cogent without referring to the process that produced it. By contrast, Kuhn rejects the idea of process-independent features of arguments in virtue of which one argument counts as more cogent than its competitors. Moreover, his social-institutional perspective on cogency does not give us a logical explication. Like Hempel, though, Kuhn does give us a set of prescriptive conditions whose satisfaction qualifies an argument as cogent. That is, an argument qualifies as cogent in virtue of its long-run social-psychological effect on a properly organized, autonomous community of experts. We might read SSK as offering a similar sort of theory, minus the prescriptive element—a theory, that is, of the social conditions under which arguments come to count as cogent for a given community of scientists who are institutionally recognized as experts. We can then distinguish SSK theorists who simply describe the relevant social—largely interpersonal—conditions from those who understand those conditions as exerting a causal effect on scientists’ judgments of cogency. Each of these views, therefore, connects cogency with some set of conditions or features of argumentation—logical, social-psychological and institutional, interpersonal or causal. As such, each goes beyond the everyday sense of the term “cogency,” which refers to the strength or powerfully convincing character of an argument. Notice, however, that the various approaches trade on different aspects of everyday usage, which seems to include both a rational and psychological component, as I noted in the introduction. Consequently, the term allows one to develop theories of cogency that highlight and expand different aspects: the rational side as involving an evaluative or prescriptive element, the (social-)psychological side as a descriptive-explanatory sociological category, or both. One can then organize these components differently: a prescriptive theory, for example, might lead with the rational, holding that an argument ought to be psychologically powerful because of its inherent rational properties (e.g., Hempel), or it might reverse the relationship, holding that an argument ought to be considered rationally cogent because it has a
56 Chapter 2
social-psychological force for the community (Kuhn). SSK provides similar variations in a nonprescriptive register. I take it that in going beyond the surface meaning of cogency, these different accounts all want to say something illuminating about the actual practice of scientific inquiry and discourse. Here is where we can find common ground, then: by reading the different initiatives in science studies as striving to say something about which conditions and features of scientific argumentation lead, or ought to lead, scientists to regard arguments as cogent. Logical empiricism failed in precisely that regard: formal-logical structure provides an insufficient basis for understanding even the rational side of cogent scientific argumentation. In any case, the appeal to practice allows us to regard theories of cogency as attempts to explicate the “social practice of cogency,” so to speak—the social-practical structures that underwrite the ascription of cogency in scientific argument-making. As providing different ways of understanding the practical ascription of evaluative terms like “strong” and “cogent” to arguments, such theories attempt to articulate a conception of cogency, where “conception” refers to an elaboration or interpretation of the practices in which a given surface “concept”—such as “cogency” in its everyday sense— plays a key role (cf. Dworkin 1986, 68–76). Given this common ground, the question is whether we can construct a broader framework that can bring these different conceptions of cogency together in fruitful interdisciplinary exchange. Kuhn’s allusions to reasonable but noncompelling persuasion points to the rhetorical tradition as a resource for such a cooperative project. In the next chapter, then, I examine three rhetorics of science, each of which represents an ambitious attempt to overcome Kuhn’s Gap.
3 Closing the Gap: Three Rhetorical Perspectives on Science
Kuhn’s Gap initially appears in his Structure of Scientific Revolutions as an apparent tension in his attitude toward scientific change. On the one hand, he rejects the formal-logical reconstructions of logical empiricism, stressing conceptual incommensurability and institutional structures in science; on the other hand, he appears reluctant to abandon the logical perspective altogether, insisting that scientists are reasonably persuaded by good reasons, and thus by the content of arguments. He resolves this tension by looking to the social-institutional dynamics of science as collectively rational and thus progressive: collectively and in the long run, scientists respond to good reasons. But in this solution, autonomous social-institutional dynamics effectively absorb the logical perspective without first exploring the area between them. Kuhn’s remarks on persuasion via good but noncompelling arguments, however, pay lip service to that unexplored area, traditionally understood as the domain of rhetorical argumentation, which typically involves elements of dialectic. So it comes as no surprise that dialectical and rhetorical accounts of science eventually stepped in to bridge the gap between logical and socialinstitutional perspectives. In this chapter I examine three such models as attempts to bridge Kuhn’s Gap. Each presents itself as a rhetoric of science in which dialectical elements play a greater or lesser role. I make no claim here to cover the rhetoric of science as an interdisciplinary field (see Simons 1989, 1990; Keith and Rehg 2008), much less the rhetorical perspective in general, which takes in a vast range of approaches and analytic techniques (Sills and Jensen 1992; Bizzell and Herzberg 2001). Much of the best work in the rhetoric of science consists in detailed case studies (see Harris
58 Chapter 3
1997), but for present purposes, I want to focus on three particularly ambitious theoretical initiatives.1 Each of these explicitly billed itself in terms of rhetoric and offered a different style of rhetorical analysis that addressed the problem in Kuhn by filling out the microdynamics of persuasion and theory change. Marcello Pera and Lawrence Prelli explicitly took up issues raised by Kuhn. Though making less of Kuhn’s work, Bruno Latour wanted to overcome the kind of gap opened by Kuhn. In choosing these three rhetorical theories, I do not claim that they offer the most compelling mode of rhetorical analysis; I choose them because their emergence buttresses the reading of Kuhn above: because they address Kuhn’s Gap, one can locate their conceptions of rhetoric within the broader argumentation-studies framework introduced in chapter 1. Moreover, the three together cover a spectrum regarding the role of normative standards in the rhetoric of science. They also represent a range of disciplines: Pera wrote as a philosopher, Prelli as a speech-communication theorist; Latour, though himself a philosopher, initially worked in close collaboration with sociologist Steve Woolgar. 1
Pera’s Rhetoric as a Dialectical Perspective
Pera is explicitly concerned to fill the gap that Kuhn’s account opens up between logic and sociology—indeed, the duck–rabbit interpretation of Kuhn is Pera’s (1994, 8–10).2 He agrees with Kuhn that scientists cannot get by with deductive and inductive arguments alone, but must at times resort to rhetorical arguments, that is, persuasive arguments that are “neither formally stringent nor empirically compelling” (Pera 1991, 35). Such arguments become necessary, for example, when one aims to open up a new way of doing science or a new line of inquiry (e.g., Galileo, Darwin), or to open entrenched assumptions to critical testing, or when interpretive issues arise, or even when one wants to make a hypothesis initially plausible. Pera’s (1987) study of Galvani’s theory of animal electricity provides an example. In his initial attempts to make the hypothesis plausible, Galvani used a range of informal arguments, such as the structural analogy between the Leiden jar and the behavior of frog legs in the presence of a metallic arc. Galileo resorted to ad hominem arguments, analogies, pragmatic considerations, and so on (Pera 1994, chap. 3; Pera 1991, 36–42). Unlike Kuhn, however, Pera commands a deep knowledge of the field of argumentation theory. Following Perelman and Olbrechts-Tyteca, Pera defines
59 Closing the Gap
rhetoric in relation to persuading particular audiences: rhetoric encompasses those arguments that “are addressed to specific interlocutors with specific systems of expressed or implicit beliefs, and they aim at changing such systems” (Pera 1994, 106). The task of determining the reasonableness of rhetorical argumentation, what makes such arguments “good,” Pera assigns to dialectics, which he calls “the logic of rhetorical arguments” (107). As the “logic of science,” scientific dialectics delivers the canons that validate such rhetorical arguments (133; cf. 103). Consequently, normative dialectics plays the central role in Pera’s response to Kuhn: his approach exemplifies more a dialectical than a rhetorical perspective. In recognizing the rhetorical and dialectical aspects of science, Pera conceives science as a game with three players: the “inquiring mind,” or scientist proposing a hypothesis; nature; and the group of scientists that test the hypothesis in debate (133). Although one can think of the inquiring scientist as engaged in a kind of dialogue with nature, asking questions of it, nature’s response is settled only through debate within the scientific community— nature “only speaks within the debate and through the debate” (11). The dialogue that has epistemic authority, in other words, is the “living conversation” among scientists (Pera 2000, 58). This last point defines the challenge for Pera. In ascribing ultimate authority to the science community, he sides with Kuhn in rejecting the idea of grounding science in an “impartial arbiter” provided by universal rules of method. Thus Pera opposes “methodologists” like Popper, Lakatos, and Laudan. Although these theorists can be read as taking a dialectical approach,3 in Pera’s eyes they are still chained to the idea of impersonally rational demonstration. By contrast, his own dialectical view links victory to the argumentative capacity of competitors. For these reasons, while the methodological view is committed to defining the concept of impartial arbiter, an arbiter independent of the specific trial he is called in to judge, the dialectical view is committed to defining the concept of honest victory without an impartial arbiter, a victory to be adjudicated by no other resources than the ones offered by the discussion. (183)
It is not some rule of rational theory choice but rather “the debate that establishes that A’s arguments are stronger than B’s” (185). The challenge is to make this move without falling into the relativism espoused by “countermethodologists” like Rorty and Feyerabend: if there are no universal rules of method independent of the contingencies of historically situated controversy, then on
60 Chapter 3
what basis may we consider outcomes epistemically rational and progressive? How does one avoid the relativist implication that a different set of historical circumstances might have produced an equally cogent, but different, answer to the same scientific question? In reply, Pera must show how a dialectical model can avoid relativism without relying on transhistorical, universal rules of method. The key to his solution lies in the idea of a scientific tradition of argumentation—stretching from the ancient Greeks onward—defined by distinctive types of “dialectical factors” that vary historically in their content but nonetheless allow scientists to resolve controversies on an objective, epistemically rational basis (140–141). If this gambit succeeds, then Pera’s dialectical perspective on cogency incorporates both logic and sociology: on the one hand, “the weight of facts, stressed by Popper and Lakatos, does not vanish but is added to that of other elements”; on the other hand, “psychological and sociological factors, considered vital by Feyerabend and sociologists of science, operate through the dialectical filter that regulates scientific debate” (135). Dialectical factors thus provide Pera with the means to go beyond Kuhn by elucidating the microdynamics of scientific debate in a way that avoids relativistic implications. These factors, which constitute the available epistemic resources of scientific debates, include both “substantive” and “procedural factors” (Pera 1994, chap. 4). Both of these categories pertain most directly to science as a conceptual process—to the interplay of experiments and ideas, with personalities and social organization largely disregarded. Substantive factors include not only facts, theories, and assumptions, but also the “epistemic values” we find in Kuhn, “commonplaces of preference” that rank such values, and various presumptions (e.g., that nature has a mathematical structure, so one should prefer mathematical explanations). Many of these substantive factors are built into the practice of science: “taken together, they define scientific tradition” as “a tight web of practices, norms, ways of thought, forms of argumentation, systems of belief, and so on” (112). The procedural factors are the rules for conducting and adjudicating debates. Rules of conduct Pera draws from Rescher’s dialectics (Rescher 1977), whereas rules for adjudication are “refutation schemes” that trade on logical, probabilistic, or pragmatic considerations (Pera 2000, 56–57; 1994, 120–126). Pera’s procedural rules directly govern not the social position and characteristics of participants, but rather the substance of what participants may reasonably state or ask. This substance itself is spelled out by the substantive factors described
61 Closing the Gap
above—the various elements of the conceptual content that informs science as a process. Rules of conduct define the admissible moves and countermoves in dialogical exchanges. For example, if scientist A states that a given experiment E supports hypothesis H, then a critic may respond by citing further evidence that rules H out (122). Rules of adjudication determine which side has the upper hand at a given point in the debate. For example, the debate is judged in A’s favor if side B ends up rejecting a presumption that B had previously accepted, or if B cannot resolve a problem that B recognizes as a problem. In line with the second rule, Galvani temporarily won his debate with Volta over animal electricity in 1794 because an experiment proposed by Volta supported Galvani’s view. Confronted by a challenge he himself acknowledged as legitimate and unanswerable, Volta withdrew his theory (Pera 1994, 125; see also Pera 1992). In addition to substantive and procedural factors, Pera also explicates some key considerations for appraising scientific arguments, that is, for making comparative assessments of the cogency of argument products. These explications define such notions as relevance, validity, and argumentative strength, and thus they have a normative character. Particularly important is Pera’s explication of what it means for an argument to be stronger than its competitor in the same “dialectical situation,” that is, in a given state of discussion with a particular set of available substantive factors: argument A is stronger than argument B “if the dialectical strategy in favor of the conclusion of A is based on more substantive factors or on more important substantive factors than those contemplated by the dialectical strategy favoring the conclusion of B” (120). It is important to note here that the judgment of the community, rather than an independent standard, defines which factors are “more important” (118–120). “Dialectical strategies” include the various ways one can put together substantive factors in an argumentative contest of competing theories so as to win the debate according to the rules of adjudication. As Pera summarizes it, a theory T has a “winning dialectical strategy” in its favor when the proponents of T, starting with the substantive factors and the common ground shared with their opponents, and following the procedural rules of scientific discourse, force the opponents to assent to the theory, withdraw from the debate, or remain silent (121).4 Thus if one side runs short of substantive resources for defending its view (and so simply repeats their arguments), or violates the procedural rules (e.g., ignores the arguments in favor of T), then the other side has won (127).
62 Chapter 3
Given this explication of argument strength, Pera can insist that the rational quality of arguments decides matters in science.5 Theory change is rational, that is, insofar as the winning theory in a debate meets the following acceptability condition: “A theory T is rationally acceptable only if it is supported by valid arguments, or if the arguments supporting T are stronger than those supporting T′” (144). A valid argument is a substantively decisive (winning) argument for a universal audience, whereas a stronger argument is stronger relative to competing arguments in a specific dialectical situation and its “configuration” of substantive dialectical resources (119). The ideas of comparative strength and rational acceptability allow Pera to distinguish his model from relativistic and strategic approaches that reduce scientific progress to de facto theoretical change. According to Pera, relativists like Rorty and Feyerabend render normative appraisal superfluous, once one has described the actual historical transformation in an area of science (138). But their relativism rests on a hasty inference from the context-dependence of normative standards to the claim that scientists lack the shared epistemic resources for the rational adjudication of controversies. Pera insists, on the contrary, that relativism does not follow from such context-dependence if one understands scientific debates, including interparadigmatic controversies, as “infratraditional” at a higher level of abstraction. His view involves two key assumptions. First, even in interparadigmatic debates in which the two parties differ in the configuration of dialectical factors they accept, they nonetheless “are members of the same tradition and therefore share many of the same factors.” Second, Pera assumes that participants “have the will and desire to converse” so as to settle the debate: “one interlocutor has to defeat the other.” Consequently, they will “begin to build an overlapping area between their respective configurations” that will allow them to reach closure. That is, given these shared substantive resources and the comparative criterion above, “there is no reason to think that a debate of this kind cannot be held and brought to a conclusion” (140)—indeed, a conclusion that is rationally acceptable in the epistemic sense of “rational.” Notice that Pera’s solution implies that interparadigmatic argumentation depends heavily on the inventive (rhetorical) capacities of participants to find common ground between their opposed paradigms, which involve somewhat different substantive factors (140). If Pera succeeds in bridging Kuhn’s Gap, then he does so by contextualizing the dialectical perspective on cogency in the actual tradition of scientific discourse. Thus the science community itself, as carrying on that tradition, con-
63 Closing the Gap
stitutes the normative arbiter of cogent argumentation. Like Kuhn, Pera holds there is no specific “configuration of substantive factors” that could provide an independent standard of rationality common to all scientific contexts (112; also 129–132, 139). Unlike the countermethodologists, however, he does not take such variation as grounds for skepticism or for reducing science to mere convention (132–133, 136–142). One reason he can avoid the relativist trap, he believes, is that his model does not define the more progressive theory simply in terms of just any de facto consensus. Rather, a winning theory enjoys a “presumption of progress” in its favor only if the consensus follows “after a rational discussion based on the substantive factors of scientific dialectics” (187, 188, emphasis added). Indeed, Pera grants that a given controversy might lead to consensus on the weaker theory; but he insists that there is no independent standard—neither an a priori rule nor a metaphysical truth6—but only further actual debate, by which one can identify such an outcome as problematic (185, 187). In other words, the community functions as normative arbiter only insofar as its consensus accords with the community’s sense of the procedural and substantive demands of rational debate, and only insofar as the outcome reflects the rules for adjudicating such debates. In saying this, Pera does not want to prescribe these normative features as independent standards of rational discussion: they function as aspects of cogency only insofar as the science community recognizes them. He clearly holds this view of substantive factors, and he also seems to regard procedural factors as likewise contingent.7 But in explicating these standards as he does, he nonetheless takes the prescriptive stance of a virtual participant in scientific discourse. So he must at the least be offering a normative interpretation of scientific practice from an internal perspective. Pera can grant this without retracting his key point: his model links cogency with features of both the content and procedures of actual scientific argumentation in such a way that neither can function apart from the other. On Pera’s view, as on Kuhn’s, one cannot understand cogency simply in terms of features of arguments that satisfy standards independent of the discourse itself. However, whereas Kuhn looked to the institutional features of the scientific process for his account of cogency, Pera focuses more squarely on science as a tradition of actual debate: the process of debate within the science community is not simply a process of assessing cogency but rather is partly constitutive of what counts as the more cogent argument. This reading is supported by the explications we saw above, according to which the community’s
64 Chapter 3
collective judgment of the relative importance of different substantive factors constitutes the measure of the relative strength of competing arguments. Moreover, what counts as the winning argument is partly defined by what actually happens in the discourse: the opponents withdraw, concede, or run out of new things to say. Elsewhere Pera holds that rhetoric plays a “constitutive role” in the argumentative “ascription of truth” to a theory (150)—hence his emphatic claim that “outside debate there is no rule” that would define what counts as the better argument in support of a theory (185). At the same time, if the judgment of the science community (i.e., the collective trend of opinion in the community) can sometimes be seen in hindsight to get it wrong about which arguments are more cogent, then the collective judgment at any point in time cannot exhaust the idea of cogency. I thus suggest that one might summarize Pera’s conception of cogency as follows: an argument A is more cogent than B just in case (a) the community judges A to be stronger than B after rational discourse, as defined by the accepted dialectical factors, and (b) that judgment is not reversed at a later stage of rational discourse (although A might be superseded by argument C). It follows from this conception that at any stage of discussion and inquiry, successful arguments can enjoy at most a presumption of being more (or less) cogent than their competitors. In sum, Pera’s conception of cogency attempts to bridge Kuhn’s Gap by conceiving the reasonableness of scientific change in terms of the historically contingent, but dialectically rational substance of the broader tradition of scientific argumentation and the communities that carry on that tradition. With dialectics as a rational bridge, Pera’s model links the logical perspective to a broader sociocultural perspective on science as a discursive practice responsible to dialectical obligations. Notice, however, that his analysis focuses mainly on abstract procedures for regulating the flow of discussion about certain types of content, as defined by the substantive factors germane to scientific practices. The specifically social aspects of such discussion remain largely unanalyzed: we find no analysis of how scientific communities should organize their members for conducting dialectically adequate discussions. Pera himself concludes his 1994 study with the claim that his model is more an “intersubjectified” than a “socialized” account of scientific progress (189). To bridge Kuhn’s Gap at the disciplinary level, however, we clearly need something like the latter, that is, an account of the social-institutional context and the norms that might govern that context and support its rational features. In Part II, we see that
65 Closing the Gap
Habermas provides such an account of social organization—albeit an account that relies on more ambitious, transhistorical idealizations. Scholars who seek a more concrete rhetorical analysis of the particular audience contexts might also object to Pera’s dialectical model (Harris 1995). A thicker rhetoric need not lose its normative ballast. The normativity to which I refer goes beyond the weak evaluative sense that rhetorical analyses tend to have insofar as they analyze the function and effectiveness of various rhetorical devices on the target audiences. This minimal evaluation does not give us the account of reasonable scientific argumentation that Kuhn alluded to but did not spell out—for that one needs a normative model. In this context, Prelli deserves attention precisely because he wants to develop a prescriptive rhetoric that would fill out the gap left open by Kuhn. That is, unlike Pera, who locates the norms of reasonableness and cogency at the dialectical level, Prelli elaborates properly rhetorical standards of reasonable scientific argumentation. 2
Prelli’s Communitarian Rhetoric of Science
Prelli (1989a, 1–8)8 conceives Kuhn’s Gap as an opposition between the formally logical “face” of science and its “other-than-formally-logical” counterpart. Though described in various ways, the latter generally encourages doubts about the rationality of science. In reply, Prelli proposes a rhetorical framework for grasping the reasonable features of the non-formal-logical face. Like Pera’s dialectics, Prelli’s rhetoric focuses not only on textual products, but also on the procedural dimension of argumentation. However, he goes beyond Pera in delineating the specifically rhetorical substance of cogent scientific arguments. Taking his cue from Kenneth Burke, Prelli highlights the “suasory use” of linguistic symbols, and hence the role of terminology and word choice, in scientific argumentation as a situated response to rhetorical “exigences,” that is, problems or gaps in knowledge (13–23). From this vantage point he can bring both Aristotle’s Topics and classical stasis theory to bear on Kuhn’s paradigmmodel. Both topical analysis and stasis theory provide ways of systematizing rhetorical “invention”—the construction of persuasive arguments (Prelli 1989a, chaps. 3–5). Aristotelian topoi provide the speaker with a taxonomy of possible “sayables” or commonplaces—potentially persuasive themes, topics, and lines of argument. Stasis theory, on the other hand, attempts to delineate the fundamental stases—the possible sticking points or issues in dispute upon which
66 Chapter 3
the speaker can take a stand. To construct persuasive arguments, the speaker surveys these symbolic resources in order to choose (a) a rhetorical aim (or exigence) appropriate for the situation and audience, (b) the specific issue on which to take a stand, and (c) the “sayable” lines of argument that are likely to persuade the audience. Applied to science, topical and stasis theory show how Kuhn’s paradigmatic “disciplinary matrix”—scientists’ shared stock of generalizations, models, values, and exemplars—provides a basis for rhetorical argumentation (87–90). In Prelli’s hands, rhetorical theory yields a rather ambitious taxonomy of scientific stasis points and topics that pertain to scientific persuasion generally, and which one can further specify for different disciplinary contexts. For example, topoi that suggest lines of argument include problem-solution topics (e.g., different kinds of competence, precision, explanatory power), evaluative topics (e.g., consistency, simplicity, elegance, fruitfulness), and exemplary topics (exemplars, analogies, metaphors) (185–215). As to stases, Prelli produces a chart of sixteen issues (formulated as questions) by crossing four types of “superior stases” with four types of “subordinate stases” (146). The former identify the basic rhetorical aims or ends germane to science, that is, the four types of problems that constitute rhetorical exigences for science, namely ambiguities about whether something exists (evidential stasis), about the meaning of theoretical constructs and phenomena (interpretive stasis), about the significance of claims for inquiry (evaluative stasis), and about the “scientific actions” required for fruitful inquiry in some domain (methodological stasis) (123; emphasis removed). The subordinate stases then specify the basic different ways of dealing with each of those four exigences—as a conjectural, definitional, qualitative, or translative issue. Thus, approaching the problem of evidence conjecturally yields the issue, “Is there evidence for x?” whereas approaching it qualitatively gives us the question, “Which empirical judgments are warranted by available evidence?” A methodological exigence, approached definitionally, yields the question, “What does it mean to apply procedure x correctly?” And so on. Prelli’s analysis converts the various conflicting and ambiguous epistemic values noticed by Kuhn—just the values that led into difficulties of incommensurability—into a set of guidelines for framing persuasive arguments. Thus the rhetorical task confronting the scientist in discussion is to make an apt choice of which issue (or chain of issues in more complex arguments) to address, a choice that will be appropriate in light of both subject matter and
67 Closing the Gap
audience context. Appropriate choices are not simply effective but also normatively reasonable. The source of normativity in Prelli’s model lies in the ends constitutive of scientific practice. The basic idea is that scientists consider an argument reasonable only if they can see how it responds to one or more of the four rhetorical exigences that define the legitimate rhetorical purposes of scientific argumentation. Although these ends define different rhetorical situations, they all are subject to an overarching “rhetorical, pragmatic criterion governing the logic of reasonable scientific discourse” in general, in effect a rhetorical definition of cogency, which I’ll refer to as RC: to be judged reasonable and persuasive in any specific situation, scientific discourse must be perceived as identifying, modifying, or solving problems that bear on a specific scientific community’s maintenance and expansion of their comprehension of natural order. (122–123)
By elaborating a kind of substantive rhetorical “logic” inherent in science as a normative practice, Prelli gives us what Kuhn promised but did not quite deliver: an account of the microdynamics of persuasion that is more substantive and contextual than formal logic and more normative than psychology and sociology. Persuasive scientific arguments, that is, depend on substantive elements—discipline-specific rhetorical ends, issues, and lines of argument— that have a normative character for scientists. Topics and stases allow one critically to evaluate scientific rhetoric not simply for its psychological effectiveness, but for its ability to meet the normative demands posed by the relevant ends, topics, and issues. At the same time, this normative evaluation includes the social-psychological dimension insofar as normatively adequate inventive choices are also situationally effective choices as well: the speaker/author ought to select which of the various ends, issues, and arguments will actually work, given the current state of discussion and controversy, consensus and dissensus, in the group of scientists to whom the argument is addressed. Thus, as Prelli shows in his case illustrations throughout the book, the critical analysis of actual argumentation is highly context bound: one must examine the participants’ concrete rhetorical choices and their employment in the actual rhetorical situation. By showing how effectiveness partly depends on normative appropriateness and vice versa, Prelli overcomes the traditional dichotomy between rational argument and effective communication. As a result, Prelli tackles the problem we saw in Kuhn: to explain how good reasons can become better reasons and arguments that lead participants to
68 Chapter 3
change their views in cases of controversy. An argument might be good in the sense of fitting with accepted stases and topoi, and thus being dialectically admissible, but it becomes better—actually persuades, is more cogent than competitors—only if it responds insightfully to the actual constellation of positions held by members of the audience. To give an example at the contested boundaries of science and nonscience, Prelli turns to McLean v. Arkansas, a federal court decision that invalidated a law requiring equal time for “creation science” in classrooms (234, 219–236). Prelli argues that the plaintiffs won the day against the creationists partly by addressing the “demarcation exigence” generated by ambiguities in the status of creationism as a science. This rhetorical strategy opened up lines of argument based on the Mertonian model of science as committed to the institutionalized virtues of communality, skepticism, universality, and disinterestedness. Each virtue in effect functioned as a topos for constructing arguments that appealed to a conventional view of science and was thus more likely to be favored by the judge. Consequently, the burden of proof fell on the creationists, but their position simply did “not have the content” for a convincing reply: they could not show signs of communality (e.g., a record of refereed articles), their biblical literalism appeared at odds with skepticism, and their resistance to well-accepted scientific claims showed a lack of universality (i.e., adherence to impersonal criteria of acceptability). Thus the plaintiffs not only gauged the weaknesses in the content of creationism, they also aptly assessed their judicial audience. As the analysis of McLean v. Arkansas shows, Prelli’s approach does not rely solely on logic and dialectics for its normative ballast, but rather has its own properly rhetorical norms. He thus provides a more robust rhetorical perspective than does Pera. The case study especially highlights the use of ethos (appeal to character) by the plaintiffs, inasmuch as one of their central arguments showed that creationists did not display the character of scientists.9 Philosophers of science such as Kitcher (1995) implicitly recognize the importance of ethos- and pathos-based arguments (see also Krips, McGuire, and Melia 1995). Appeals to ethos play a significant role in the experimental report, understood here as an argument for accepting some finding. Although the reader’s acceptance of a new finding requires trust in the author’s competence, this trust can acquire a rational basis through certain rhetorical devices, some as simple as a note on the author’s affiliation with a respected research institution. But ethos-appeals in the Aristotelian sense are better exemplified by various moves internal to the report, which display the author’s “experimental virtuosity,” for
69 Closing the Gap
example by anticipating worries about possible artifacts, disturbances, and so on (Kitcher 1995, 55). Kitcher also implicitly refers to a scientific use of pathos, in a broad sense that would include attempts to stir up readers’ interest in some finding or new theoretical construct (ibid., 56ff.). Both Pera and Prelli are concerned with scientific argumentation as it involves persuasion, but notice that Prelli conceives of the “rhetorical procedure” precisely in relation to communication within the science community, more specifically in relation to fostering greater mutual understanding among scientists. Because stases “constitute a rhetorical procedure for working out understandings of where discussants substantively agree and disagree,” stasis analysis “can further communication among adherents of quite different paradigms and orientations, allowing them to ‘make sense’ of each other’s claims” (183). Pera, by contrast, devotes the bulk of his analysis to the considerations that determine the stronger argument in a field of competing arguments; his analysis of audience remains thin, more or less reduced to opposing “sides” that accept somewhat different substantive factors as grounds for the competing theses they defend. His dialectical perspective has the audience implicitly in view, but focuses above all on the movement of ideas, that is, the conceptual process. Although Prelli also focuses heavily on conceptual process, his rhetorical perspective leads him to attend to the social process as well. The audience is not simply reduced to opposing sides but to actual personalities, people who not only hold a view but also have biases, personal idiosyncrasies, different levels of credibility in the science community, and so on.10 These differences are not necessarily oppositions, of course. Rather, Prelli’s analysis might supplement Pera’s. Specifically, one can read Prelli’s rhetorical criterion (RC) as elucidating the rhetorical substance of the rational discourse referred to in Pera’s dialectical model of cogency. Nonetheless, this move still leaves much unsaid about the social process of science—the various institutional mechanisms such as peer review, funding structures, gatekeeping, and so on. By linking a normative conception of cogency with audience psychology, Prelli’s rhetoric adds a social layer not found in Pera’s. But like Pera, he does not fully bridge the disciplinary gap: he does not link persuasion with specifically sociological aspects of science. Before examining Latour’s attempt to do just that, I mention some further potential problems in Prelli’s model. First, one might regard his systematic four-by-four grid of stases along with detailed lists of topoi as overly ambitious and unnecessarily tidy. Although he admits that his list of topoi does not exhaust possible lines of argument, he seems
70 Chapter 3
to regard a systematic table of stases necessary for evaluating the adequacy of rhetorical choices: without a sense of the alternative possibilities, this evaluation lacks normative orientation (159, 217). In any case, Prelli’s approach has been criticized as rather mechanical and less than informative (Gaonkar 1997a, 65–73). Even if the criticism is fair, in Prelli’s defense it seems to me the three main rhetorical tasks—selecting an exigence and specific issue and developing situationally reasonable lines of argument—can help with the analysis of cases. In relation to the project of critical assessment, a further criticism might be leveled against both Prelli and Pera. Recall again Prelli’s rhetorical criterion for reasonable scientific discourse (RC): “to be judged reasonable and persuasive in any specific situation, scientific discourse must be perceived as identifying, modifying, or solving problems that bear on a specific scientific community’s maintenance and expansion of their comprehension of natural order” (122– 123). This standard articulates the normative character built into the contemporary practice of scientific argumentation as a communal enterprise. To this extent, his rhetoric of science has a communitarian slant: it approaches rhetorical analysis of scientific argumentation in relation to science as a discourse community that shares a set of substantive standards and presumptions (Taylor 1996, 105). The same can be said for Pera. Neither theorist adopts a standpoint external to the science community or appeals to transcendental norms. Besides the question of whether there is an overarching science community that shares the normative commitments Prelli ascribes to it (or a single scientific tradition sharing broad types of dialectial factors, as Pera holds), there is the question of whether and how a critical stance toward those norms is possible. This second question becomes especially pertinent in boundary disputes such as McLean v. Arkansas, when opposing communities of discourse are at odds.11 As I mentioned, the anticreationists won the case, in part, because they appealed to the conventional Mertonian ideal of science. On what basis, in Prelli’s model, can such an ideal and its institutional norms be opened to critical assessment? The danger lurking in such a communitarian approach is that it “underestimates the potential ideological functions of science in contemporary culture” (Taylor 1996, 106). For those interested in the project of nonskeptical but critical assessment of scientific claims, this last problem reopens the door to a disabling relativism. Here relativism appears as a kind of conventionalism, the view that each community of discourse constitutes a context of argumentation whose normative
71 Closing the Gap
conventions and evaluations are self-sufficient, immune to outside review, as it were. How then does one reasonably adjudicate arguments across contexts? As I explain in chapter 4, Habermas criticizes Toulmin’s argumentation theory for a similar problem. Not unlike Pera and Prelli, Toulmin (as Habermas reads him) draws his norms of cogency from the conventions specific to each disciplinary field of argument. According to Habermas, such approaches are vulnerable to the charge of relativism insofar as they eschew universal norms of cogent argumentation. 3 Latour’s ANT Rhetoric: How to Win Friends and Influence Society/Nature Of the three rhetorical approaches considered in this chapter, Latour’s goes the farthest toward a sociologically informed rhetoric of science. By embedding rhetoric in actor network theory (ANT), Latour and his colleagues bring to their analyses the resources and concerns of both interest-based and ethnographic sociology: he incorporates both micro- and macrosociological perspectives (see Latour 1999a). Though his methods of textual analysis appear to stem mainly from semiotics, he casts much of his analysis in the language of rhetoric. I thus suggest that from the standpoint of argumentation theory, we can read Latour as a distinct conception of the rhetorical perspective that absorbs the sociological substance lacking in Pera’s and Prelli’s analyses. Latour’s antinormative approach to cogency also contrasts with their concern with good reasons. To be sure, the labels of “rhetoric” and “sociology” may be applied to Latour’s ANT only in a qualified sense, as I explain below. In what follows I focus mainly on three studies that together form a coherent response to Kuhn’s Gap. In Laboratory Life (1986; first ed. 1979), Latour and coauthor Steve Woolgar describe this gap as the dichotomy between the intellectual (or technical) and social dimensions of science (ibid., 22–27). Their solution regards the technical as always also social. They label their approach “social constructivist,” albeit in a “special sense”: the “social construction of scientific knowledge” refers to the process by which scientists “make sense of their observations,” that is, how scientists “use socially available procedures for constructing an ordered account out of the apparent chaos of available perceptions” (ibid., 32, 33). Though not labeled as rhetorical, their analysis is centrally concerned with the persuasive process, and in light of Latour’s later, explicitly rhetorical analysis in Science in Action (1987), one can see the relevance
72 Chapter 3
of this persuasion-talk for a rhetoric of science. The third study, The Pasteurization of France (Latour 1988; French ed., 1984), further illustrates Latour’s approach with a case study. In that work one also finds an emphatic statement of his antinormative approach to cogency. I start with the microsociological side of Latour’s rhetoric of science. In Laboratory Life, Latour and Woolgar analyze the production of scientific papers in terms of “literary inscription,” that is, reading and writing processes of various sorts, which include various sorts of traces and charts produced by laboratory instruments and their recording devices. Because “the function of literary inscription is the successful persuasion of readers,” such inscription has a rhetorical aim. The authors go on to note, however, that “the readers are only fully convinced when all sources of persuasion seem to have disappeared.” Thus the goal of scientific persuasion is to transform statements with a qualifying “modality” (e.g., “There is evidence for x,” “So-and-so has found that y”) into unqualified factual statements (e.g., “x is y”) (ibid., 75–76). Latour’s sociology of persuasion flows from this transformatory rhetorical goal. In Laboratory Life we get a detailed ethnography of the messy process of constructing an argument—the various attempts to order the chaos of inscriptions, interpersonal negotiations with colleagues, and so on—as it unfolds in a particular laboratory. In Science in Action, Latour begins from the opposite side of this process, when readers confront a finished product, the polished scientific article. He unravels the text of the article strand by strand so as to expose the rhetorical devices specific to science, which he now names as rhetorical (Latour 1987, chap. 1). The analysis uncovers a number of such devices: citation of authorities, inclusion of data and figures, strategic use of literature references (i.e., qualifying or “modalizing” cited works in a manner favorable to the author’s aims). The rhetorical aim is to overwhelm and isolate the skeptical reader by demonstrating the sheer number of “allies” and resources the author can enroll in support of the article. Latour thus distinguishes scientific rhetoric (from traditional or “old” rhetoric) by the greater number of “external allies” it enrolls. The old rhetoric,12 he claims, “used to be despised because it mobilized external allies in favor of an argument, such as passion, style, emotions, interests, lawyers’ tricks and so on.” Its dispassionate appearance to the contrary, science takes this aspect of rhetoric and runs with it: “the difference between the old rhetoric and the new [i.e., scientific] is not that the first makes use of external allies which the second refrains from using; the difference is that the first uses only a few of them and the second very many” (ibid., 61).
73 Closing the Gap
By using the label of “external allies” to describe the whole range of argumentative strategies—both traditionally problematic ones such as “lawyers’ tricks” as well as traditionally upright moves, including the report of experimental findings—Latour rejects a normative rhetoric. He thus describes scientific rhetoric as a kind of strategic warfare or competitive sport. Not surprisingly, Machiavelli’s “political ethnography” provides the exemplar that motivates ANT as a method for science studies, which should simply follow scientists around through their various interactions and describe what they do, without concern for morality or idealizations (Callon, Law, and Rip 1986, 5; Latour 1987, 124ff.; 1988, 3–7). Thus Latour compares the strategic use of references to “Byzantine political schemes.” The strategic rules for citation, he says, “are simple enough: weaken your enemies, paralyze those you cannot weaken, . . . help your allies if they are attacked, ensure safe communications with those who supply you with indisputable instruments”—and so on (Latour 1987, 37). Although one could recast these various moves in normative dialectical terms, Latour’s antinormative description of such moves places them on a par with political maneuvers that do not belong to the dialectical perspective. Latour’s antinormativism seems to stem from complex motivations.13 Here we should notice how it serves an explanatory strategy that ANT shares with SSK. Like other sociologists of scientific knowledge and social constructivists, Latour’s procedure reflects the symmetry postulate that came to define postMertonian sociology of science, that is, the prescription to treat all claims, true and false, rational and irrational, as equally open to sociological explanation (Bloor 1991, 5–7; Pinch 1993). Latour, however, wants to radicalize symmetry so that neither the intellectual (as transparent response to natural phenomena) nor the social—neither Nature nor Society—is taken as explanans; rather, these “two poles are a single consequence of a common practice that is now the single focus of our analysis” (Latour 1992, 281; cf. 1993, chap. 4). This intention is expressed in his third and fourth rules of method (Latour 1987, 99, 144): Rule 3 Since the settlement of a controversy is the cause of Nature’s representation, not the consequence, we can never use this outcome—Nature—to explain how and why a controversy has been settled. Rule 4 Since the settlement of a controversy is the cause of Society’s stability, we cannot use Society to explain how and why a controversy has been settled.
74 Chapter 3
Latour (ibid., 144) then continues: “We should consider symmetrically the efforts to enroll and control human and non-human resources.” Neither the voice of nature as revealed by rational methods nor fortuitous alliances with social interest groups plays a privileged role in Latour’s explanation. These rules of method are understandable insofar as Latour aims above all to explain the technical and cultural success of science rather than assess it critically. Notice that, unlike some strands of SSK, he rejects the reduction of arguments to social causes, as Rule 4 above makes clear. His model does not deny that scientists succeed experimentally by bringing nonhuman resources to bear in support of their theories.14 But such laboratory success counts in the success of science only insofar as it becomes rhetorically effective in a broader social context. Latour associates ANT with a controversial ontology that regards both natural entities and human beings equally as agents in the process of science. Both are “actors” (or “actants”) in a network of actors that together explain socially successful science. Thus ANT counts as “sociology” only in the sense of a science that tracks such human–nonhuman associations (1988, 205).15 Although this move might appear ontologically bizarre, it appears less so if we understand it in methodological terms, as an implication of Latour’s use of semiotics. As employed in the Pasteur study, Latour’s semiotic method focuses simply on the texts, what writers said, as a way of discerning how the process of alliance-making eventually led to the name of “Pasteur” becoming identified with bacteriology and public hygiene in France.16 Latour (1988, 35–37) finds textual evidence supporting the idea that writers themselves began to treat microbes as agents in the social interaction of human beings—hence as actors that human beings had to take into account if they wanted to succeed in their social aims. As one writer put it at the time, “Society can exist, live, and survive only thanks to the constant intervention of microbes.”17 With the discovery of microbes, society was seen to contain an additional actor—or rather, millions of additional agents whose actions human beings had to control. By taking this metaphor literally, Latour allows the writers of the time, rather than sociological theory, to define what counts as “social.” This allows him to give a specifically rhetorical explanation of the argumentative cogency of a scientific idea: how acceptance depends on effective redefinitions of social agency. These clarifications bring us to the macrosociological side of Latour’s analysis. So far we have seen that on the microside, his rhetoric has something to say both about the argument product, that is, the article and its various devices,
75 Closing the Gap
and about the argumentation process, understood as a constructive process that depends on a variety of agents, both human and nonhuman. The macrosociological dimension of Latour’s rhetoric—the analysis of broader networks of local connections—emerges when we follow the dissenting reader of an article a step farther, to the point where further effective dissent requires one to build a “counterlaboratory,” that is, mount a competing research program (Latour 1987, 79). Otherwise the skeptic cannot counter the experimental trials of strength the authors use to support the claims in their article. By situating his analysis of the article in controversies that involve competing research laboratories, Latour shows how effective argumentation depends on overwhelming opponents at the level of material process (ibid., chap. 2). His textual analysis shows that the strength of scientific rhetoric depends on the laboratory resources (including the nonhuman entities under investigation) that scientists can muster in support of their claims. However, to transform a controversial claim into a fact requires social resources beyond the laboratory. Specifically, the researchers must get other people, groups, and institutions interested in the claim (ibid., chap. 3). This process of enlisting others depends on the ability of the scientists to “translate” their “fact-building” aims into goals and interests that attract these broader social allies. In addition, the scientists must also keep their allies “in line,” lest they distort the original claim beyond all recognition. Now the Machiavellian comparison becomes quite concrete: fact building, like Machiavelli’s empire building, depends on “a series of choices about alliances”—the difference is that scientists and engineers are “subtle enough to include in their repertoire of ploys human and non-human resources, thus increasing their margin for negotiation” (ibid., 124, 125). Thus the rhetoric that appears at the microlevel in the article as a result of local experimentation and negotiation ultimately succeeds—leads from a qualified claim to a fact—only if it links up with and/or creates macrolevel networks of associated interest groups. At that point, “the fact which has been built becomes indispensable; it is an obligatory passage point for everyone if they want to pursue their interests” (ibid., 132). Latour’s study of Pasteur provides a good illustration of the kind of rhetorical translation strategies required for effective network building. For Pasteur to succeed, he had to get his various allies to work together. Part of this involved developing dependable field applications of his laboratory methods. As Latour (1983; cf. 1988, chap. 2) puts it, Pasteur had to convert France itself into a laboratory before he could make credible claims about his anthrax vaccine.
76 Chapter 3
And for that he had to make inoculation a routine matter, one that nonscientists would not be likely to botch. Pasteur thus had to get country veterinarians, farmers, sheep, and microbes to work in synch. As Latour glosses the idea, Pasteur had “to find ways to make both farmers and the bacillus predictable” (Latour 1987, 124). Pasteur’s success also depended on uptake by the public health authorities. For this task he found a solid ally in the hygiene movement, which had already set the agenda in Europe: a concern with the spread of disease in large cities and the implications for public health, sanitation, worker disability, and so on (see Latour 1988, chap. 1). To make a long story short, Pasteur’s microbial theory gave the hygienists the rhetorical ally they needed to counter public resistance: a clear causal hypothesis about the source of disease, backed by a nationally known scientist. The rhetoric of the hygiene movement involved a series of socially powerful translations that linked bacteriology with the public interest and national pride, in the process making claims for Pasteur’s bacteriology that went far beyond what his experiments actually warranted. The public impression of Pasteur’s experiments thus combined with the social influence of the hygiene movement to produce results beyond the reach of either one alone.18 Latour’s analysis clearly has implications for a conception of cogency, which in turn reveals his response to Kuhn’s Gap. Although he tends to use terms such as “force,” “potency,” and “strength” rather than “cogency,” one can read his critique of the logical and dialectical perspectives (1988, pt. 2, chap. 2) as a rejection of intrinsic and process-independent notions of cogency. Thus, according to Latour, “nothing is by itself either logical or illogical. . . . No set of sentences is by itself either consistent or inconsistent; all we need to know is who tests it with which allies and for how long. . . . ‘Reason’ is applied to the work of allocating agreement and disagreement between words. It is a matter of taste and feeling, know-how and connoisseurship” (ibid., 179; my emphasis). As to dialectic: “Since nothing is inherent in anything else, the dialectic is a fairy tale. Contradictions are negotiated like the rest. They are built, not given” (ibid., 180; my emphasis). What matters is not an intrinsic force that somehow attaches to words and arguments in themselves but rather the audience-specific “sociologic” which can bring about others’ agreement. Thus “nothing is by itself logical or illogical, but not everything is equally convincing. There is only one rule: ‘Anything goes’; say anything as long as those being talked to are convinced” (ibid., 182). To be sure, once allies are marshaled in support of certain sentences, not anything goes, but only what counts for others as con-
77 Closing the Gap
vincing, that is, persuasive—perhaps to the point that they consider a person logically compelled to say just one thing (ibid., 182–183). But this outcome is not something dictated by intrinsic merits of argument, something inherently rational about the argument itself or the process of argumentation.19 In fact, Latour rejects any prescriptive view of cogency, that is, any conception that would depend on a distinction between right and might. As he explains his explanatory goals, he “decided to see how knowledge and power would look if no distinction were made between force and reason” (Latour 1988, 153).20 Thus, we “cannot distinguish between those moments when we have might and those when we are right”; nor are there, “on the one hand, those who bow to the force of an argument, and, on the other, those who understand only violence” (ibid., 183, 233). Consequently, his conception of cogency is entirely factual: “The strongest reason always yields to reasons of the strongest.” This is the supplement of goodness that I would like to take away. The reasoning of the strongest is simply the strongest. (Latour 1988, 186)
The more cogent argument is simply the one that de facto succeeds in “trials of strength”—or as he also calls them, “trials of weakness.” Such trials cover a spectrum running from the more peaceful sort at one end to “total warfare without ritual, purpose, or preparation” at the other. “Somewhere in between,” he surmises, “lies the great game of rhetoric” (ibid., 183). Aside from the “three textbook cases” of rational persuasion (ibid., 184), trials of strength involve a mix of argumentative resources, with the victory going to the side that can enlist a larger, more resistant alliance of “weaknesses”—a network of forces and translations involving both humans and nonhumans. Latour thus explains the apparent cogency of arguments in terms of the network of actants (human and nonhuman) with which arguments are allied and through which they can successfully travel. A reader can identify many of these allies in the content of the scientific article, especially if we understand such content broadly, to include not only the technical claims regarding the various nonhuman actants, which the team has mastered in laboratory trials of strength, but also the supporting literature cited in endnotes, the team members’ institutional affiliations, and their funding sources. But many of the supporting allies are not mentioned at all, for example the various social groups, institutions, and political interests that use the scientific findings for their own purposes. So we might say that the cogency of the scientific
78 Chapter 3
argument—the article as the product of inquiry—is constituted by its ability to enlist in its support, and travel (translate) through, heterogeneous material, social, and political networks. The greater its powers of travel, the more cogent the argument. One can now see how Latour attempts to overcome Kuhn’s Gap: he levels out the distinction between normatively sound argumentation, on the one side, and psychologically and institutionally effective rhetoric, on the other. However, whereas Prelli tied these two sides of argumentation together in a way that made effectiveness part of a prescriptive criterion of cogency, Latour makes just the opposite move, rejecting the normative perspective in principle. Although he goes considerably beyond Pera and Prelli in analyzing the socialinstitutional aspects of scientific inquiry, his solution ties together the technical and the social only by eschewing the attempt to mediate the normative demands of the former with the descriptive-explanatory realities of the latter. If bridging Kuhn’s Gap at this level means linking prescriptive and descriptive perspectives without denying either its integrity, then Latour’s approach does not adequately bridge the gap. Although his “principle of irreducibility” holds that “[n]othing is, by itself, reducible or irreducible to anything else” (Latour 1988, 158), his strategic model of cogency reduces the normative perspective to the factual, as he admits: “it becomes necessary to admit this first reduction: that there is nothing more than trials of weakness” (ibid., 191). To be sure, Latour (1988, pt. 2) is motivated by a kind of moral outrage at the arrogance and militaristic legacy of Western science. To that extent, he aims at a sweeping mode of critique that profoundly questions the core categories of modernity on which scientific institutions rely for their legitimation. His antiprescriptivism thus goes hand in hand with a critique of the “modern settlement” responsible for such dichotomies (Latour 1999b, chap. 1). This does not mean he is antiscience; rather, by dismantling the Enlightenment dichotomies that legitimate Western science he hopes to open up a new way of conceiving the relation between science and politics (Latour 1999b, 2004). Latour’s stance on relativism must be understood in this context: as the reverse side of Enlightenment absolutism regarding objective truth, the “problem” of relativism vanishes once we shed our modern spectacles (Latour 1993, chap. 4; cf. 1987, 198–213). Consequently, though Latour uses SSK methods for studying “science in the making,” he rejects the principled relativism advocated by Bloor (Latour 1987, 98–100, 196). In fact, there is a more prosaic manner by which Latour’s strategic approach avoids the relativism
79 Closing the Gap
associated with Kuhn and SSK. That is, by linking cogency with de facto trials of strength and actually existing networks of alliances, Latour avoids a relativism based on incommensurable standards of reasonable argumentation: within the strategic terms of his analysis, there is no reason to doubt that any dispute in the sciences can eventually be resolved by the more “cogent” argument. However, this response to the relativistic implications of Kuhn is unlikely to satisfy argumentation theorists interested in norms for critically assessing ordinary scientific claims. 4
Concluding Remarks
In Part I, I have attempted to show how Kuhn’s Structure of Scientific Revolutions set a task for the rhetoric of science, as situated between logical and sociological perspectives on science. That is, Kuhn presented a sociological model of scientific inquiry and progress, but he wanted to hold on to the idea of reasonableness as a response to the better argument. Explanatory and particularist models of scientific inquiry, such as the Strong Program and social constructivism, picked up one side of Kuhn’s ambitions, while historically minded philosophers of science like Lakatos and Laudan attempted to develop more realistic models of scientific rationality. But sociological and philosophical attempts to resolve tensions in Kuhn remained at odds with each other. In this chapter I have examined a spectrum of views that attempt to overcome one or another version of Kuhn’s Gap by taking a rhetorical approach, broadly construed. Each rejects conceptions of cogency in terms of the intrinsic or process-independent merits of arguments. Pera and Prelli do so by construing the normative dimensions of argument more broadly, in terms of dialectical and rhetorical merits, which they analyze in ways that point toward the social character of scientific argumentation. More specifically, Pera strives to show how one can accept the Kuhnian idea that the collective judgment of the science community constitutes the arbiter of cogency—or even defines what counts as the more cogent argument—so long as that judgment issues from dialectically rational debate within an ongoing scientific tradition. Thus Pera’s rhetoric assimilates the logical and the social within a dialectical perspective on science as a conceptual process. Prelli gives us a normative rhetoric of science in the thick sense. That is, he attends not simply to the “dialectical situation” as posing a set of abstract dialectical tasks, but to specific demands of rhetorical invention in relation to features of the concrete audience. However, beyond
80 Chapter 3
their analyses of science as a discourse community—defined by a tradition of empirical discourse about nature—the social-institutional requirements of cogent scientific argumentation remain largely unaddressed in these two approaches. Moreover, because both of these approaches rely on communitarian standards of cogency, they address the relativistic implications of Kuhn only within the history of Western science. As I shall explain in Part II, Habermas considers such communitarian models of argumentative reason overly conventionalist: unless one can ground communal standards in universal ideas of reason, such approaches remain vulnerable to relativistic implications that undermine nonarbitrary critical assessment. Latour is interesting precisely because his rhetorical perspective is equally a social-institutional perspective: he attends closely to the broader social process of science as it occurs in a context populated by a range of social actors, institutions, and groups. Although Latour’s Machiavellian model leaves the normative side of argumentation behind (except as a further social fact), it does not leave nature behind: Latour includes nonhuman allies as part of the “social” landscape of scientific argumentation. By giving natural phenomena a constitutive role in persuasive argumentation, Latour acknowledges, in a backhanded sort of way, the rationalist, empirical side of Kuhn’s Gap, which highlights empirical adequacy as the primary consideration in an account of argumentative cogency. His analysis thus ties the empirical adequacy of arguments with their ability to spread materially, socioinstitutionally, and technologically. However, it remains unclear how one would integrate Latour’s strategic analyses with a prescriptive argumentation theory. None of these models succeeds in bridging Kuhn’s Gap completely. To the contrary: taken as a trio, they reproduce that gap in miniature. On one side, we have two normative models of cogent argumentation, one of them centered on the dialectical perspective, the other squarely on a rhetorical perspective in the traditional sense. On the other side, we have Latour’s antinormative theory, which measures cogency by the arguer’s factual success in forming natural-social networks: winning friends of various sorts and influencing society/nature. Pulling these perspectives together into a coherent normative conception of cogency sets the task for the second and third parts of this book.
Postscript I The Return of the Logical: Achinstein’s Realist Theory of Evidence
In chapter 2 I presented Hempel’s syntactical model of confirmation as an attempt to define the cogency of evidential arguments in terms of their intrinsic formal merits. On that interpretation, Hempel’s articles on confirmation mark out one end of a spectrum of positions on argument evaluation—a position that focuses on the abstract logical properties of the product of inquiry as a set of evidence- and hypothesis-statements. This model goes the farthest in the attempt to detach the cogency of scientific arguments from context and process. In rejecting Hempel’s logical-empiricist approach to cogency, Kuhn, Pera, and Prelli agree that the process of scientific argumentation partly constitutes the cogency of arguments in a normative sense. Process, in other words, is not simply a more or less favorable context for constructing and assessing arguments whose cogency consists of, or is even defined by, intrinsic properties of the product, whether formal or informal. Rather, the cogency of a scientific argument cannot be fully understood apart from the process that produced it. Thus the standards by which arguments count as cogent refer not only to features of the product but also to features of the process (and procedures) of argumentation. In this Postscript I examine a position that can be read as rejecting process-oriented conceptions, hence as an attempt to vindicate the logical perspective: Peter Achinstein’s objective theory of evidence, developed primarily in his Book of Evidence (2001). Achinstein’s theory lies in the tradition of attempts to understand inductive confirmation. But like a number of contemporary philosophers of science, Achinstein approaches this topic by analyzing the concept of evidence (see Achinstein 2005; Taper and Lele 2004). What interests me here is the way his
82 Postscript I
theory links evidence with a realist notion of truth. Unlike Hempel, whose analysis of confirmation admitted truth only in the weak sense of empirical adequacy, Achinstein employs a mind-independent notion of truth. If one incorporates his concept of evidence into a logical perspective on cogent products, then the metaphysical truth of evidential premises counts as one of the merits of scientific arguments. However, his metaphysical realism also requires an internal connection between cogent products and the ongoing process of inquiry as a necessary condition of construction and assessment. Though not constitutive of cogency itself, process remains necessary to generate cogent arguments. After laying out his position (sec. 1) and developing its implications for the idea of cogency (sec. 2), I close with some remarks on the aims of argument-making practices (sec. 3). 1
Achinstein’s Objective Theory of Evidence
Achinstein differs from Hempel and Carnap in that he takes the scientific article, rather than the abstract hypothesis-observation package, as the product on which he focuses his analyses. His attempt to reconstruct the inferential structure of J. J. Thomson’s 1897 article on cathode rays is a good example (see Achinstein 1991, chaps. 10–11). Achinstein begins by situating Thomson’s article in its dialectical context: the debate between particle and wave theorists over the nature of cathode rays, the phosphorescence produced when the residual gas in an evacuated chamber is subjected to a strong electrical field. In 1883 Heinrich Hertz, having conducted an experiment in which he failed to detect a charge in cathode rays, argued against the thesis that such rays were negatively charged particles. Doubting the adequacy of Hertz’s experimental design—in particular the assumption that the chamber had been sufficiently evacuated to detect the charge—Thomson conducted further experiments, using superior vacuum technology, that led him to conclude in favor of the charged particle view. The wave theorists did not deny the coincidence of electrical effects with cathode rays; the question was whether the rays themselves carried the charge. In the first of his experiments, then, Thomson aimed to show that one could not separate the electrical charge from the cathode rays. He thus connected a cathode tube (the source of the cathode rays) to an evacuated glass bulb that contained two coaxial cylinders with a small opening. Figure PI.1 depicts a simplified schematic of his apparatus (adapted from Thomson 1897, fig. 1).
83 The Return of the Logical
Evacuated bulb
Cathode tube Path of cathode ray
_
+
Coaxial cylinders with opening
Electrometer Figure PI.1
One could determine the path of the rays emanating from the tube by the phosphorescence that the rays produced on the glass wall of the larger bulb. When the rays were deflected by a magnetic field so as to enter the coaxial cylinders as shown in the figure PI.1, the electrometer connected to the inner cylinder registered a strong electrical charge; when the rays were not so deflected, no significant charge was detected at the electrometer. Thomson concluded that “this negative electrification is indissoluably connected with the cathode rays” (ibid., 295). Achinstein (1991, chap. 11) shows in detail that the HD model cannot account for the structure of Thomson’s 1897 article, inasmuch as Thomson’s experimental arrangement and his observed results were not in fact derived— and in significant measure could not be derived—from the charged particle theory along with nonarbitrary auxiliary assumptions. In Hempel’s terminology, one cannot deduce the “development” of H from H, where H is the charged particle theory. From the standpoint of argumentation theory, we may read Achinstein as rejecting the strong formalist notion of observational relevance: one cannot generate the relevant observations by deduction from
84 Postscript I
theory. Rather, Thomson had more or less to invent the experimental arrangement that would sufficiently test Hertz’s negative thesis, and he did not predict his experimental results on theoretical grounds but rather awaited the actual experimental outcome. Focusing on Thomson’s first experiment reported in the 1897 paper, Achinstein (1991, 316–317) reconstructs Thomson’s argument as a four-step inference: (1) An “experimental result claim” describes what Thomson observed in his particular experimental apparatus with its particular measuring device (i.e., he observed the magnetic deflection of the rays in a glass bulb of a specified design, he measured their electrical charge at the far end of a coaxial cylinder using an electrometer, etc.). (2) An explanatory step states the “most likely explanation” for what Thomson observed in step (1): the fact that cathode rays carry a negative electrical charge. This explanatory move involves background assumptions (e.g., regarding the nature of matter and its effects) that should have “independent warrant from similarities with other observed cases” (ibid., 317; cf. 293). (3) An intermediate conclusion follows from (1) and (2): “So probably cathode rays carry a negative charge when they are deflected by a magnet in such a way as to enter the inner cylinder of the apparatus” described in (1). (4) A concluding “inductive generalization to theoretical consequence” holds that however one deflects cathode rays magnetically, the negative charge probably follows the path of the rays. The last step is a more general consequence of the theoretical claim in step (2). Notice that in contrast to the HD model, Thomson’s argument does not deduce an observation statement from a theoretical hypothesis (the first premise in the HD model; see chap. 2, sec. 1.2, above). Rather, he begins with an observation and then moves through a theoretical explanation to a “theoretical consequence” that has implications for further experimentation: thus the “inference is not deductive but involves both explanatory reasoning and inductive generalization” (Achinstein 1991, 316). Achinstein links this structural analysis of Thomson’s argument with a normative concept of evidential justification: “if such an inference is justified (or to the extent that it is), the experimental results constitute evidence for, or confirm, the theoretical claim” (ibid.).
85 The Return of the Logical
In his later analysis of evidence, Achinstein further clarifies the normative concept of evidence by asking whether Hertz’s 1883 experiment gave him evidence for the negative hypothesis ∼H: cathode rays are not charged particles. Achinstein (2001, 18ff.) distinguishes three answers: (1) On an epistemic-situation (ES) model of evidence, Hertz did have evidence. That is, anyone in Hertz’s epistemic situation (as defined by what the available methods allowed one to know) would be justified in believing ∼H. (2) On a subjective model of evidence, Hertz’s results counted as evidence for ∼H from 1883 to 1897, but after 1897 they did not. Achinstein (ibid., 22–24) elaborates this concept primarily in descriptive terms: in effect, experimental outcome E counts as evidence for someone at a give point in time if E is true at that point and that person takes the truth of E as an evidential reason for believing the hypothesis (here, ∼H). (3) On an objective or realist model of evidence, Hertz’s results never constituted evidence for ∼H. Achinstein builds his theory of evidence around the third answer, in relation to which the first two models are derivative. Within the third model he distinguishes “veridical evidence,” which obtains only if the hypothesis is in fact true, from “potential evidence,” which can be had even if the hypothesis is not true. So veridical evidence is potential evidence that supports a true hypothesis. The category of potential evidence constitutes the realist core of Achinstein’s normative model of evidence. The key thesis holds that if E is potential evidence for a hypothesis H, then E is so simply in virtue of physical and mathematical facts, independently of whether anyone knows it or not. Achinstein connects this idea with a normative realist conception of good reasons. Potential evidence provides an objectively good reason to believe H, such that the “degree of reasonableness is an objective, non-physical, normative fact determined by . . . physical and mathematical facts. It does not depend on what anyone knows or believes. It is abstract in the sense that it is divorced from particular individuals and types of epistemic situations” (2001, 97; 96–98). To say that the “E provides a good reason to believe H” means that E is a reason to believe H rather than its negation ∼H (ibid., 24). Consequently, Achinstein’s model is probabilistic: E counts as evidence only if it raises the probability of H above 0.5. More precisely, E is relevant as potential evidence in an argument for H only if the probability of an “explanatory connection”
86 Postscript I
between E and H is greater than one-half. An explanatory connection obtains if H explains E, or E explains H, or E and H are correlated via some third explanatory fact (ibid., 148–151).1 Achinstein then analyzes explanation in terms of a content-link or substantive propositional connection between the explanans and the question raised by the explanandum (ibid., chap. 7). Putting the probability condition and realism together gives us what Achinstein regards as his “final definition” of potential evidence: E is potential evidence that H, given background assumptions B, if, and only if (i) the probability p (that there is an explanatory connection between H and E, given E & B) is greater than one-half, (ii) E and B are true, and (iii) E does not entail H (ibid., 170). Achinstein (ibid.) considers these conditions to be both necessary and sufficient. In calling these conditions a “definition,” I do not take Achinstein to be making a claim about identity of meaning but about the real nature of evidence as a normative feature of scientific inquiry and argument-making.2 In any case, each condition must hold objectively, whether anyone knows it or not. Consequently, the definition implies an empirical rather than a priori or purely syntactical model of evidence—whether E is evidence for H partly depends on empirical facts, including background facts that might not be known, or even knowable, for scientists at a given level of knowledge. Achinstein designs the above definition to avoid problems in theories such as Carnap’s. Carnap’s model allows one (in principle) to calculate the a priori degree of confirmation of H relative to E, c(H, E), on the basis of all of one’s available information, choice of descriptive vocabulary, and method of assigning weights to possible state descriptions given by the vocabulary. Achinstein sees a number of problems with this model. For one, he does not think it explains why scientists seek new information. On Carnap’s model, once I have correctly calculated that my available evidence confirms my hypothesis, nothing in the model of evidence tells me to keep inquiring. By contrast, an empirical definition of evidence does provide a reason, internal to the definition itself, for further research, if we assume that scientists seek veridical evidence (which presupposes potential evidence): they want to arrive at true hypotheses on the basis of objectively good reasons. Nor does Carnap’s model account for the controversies over the evidential force of experiments—for
87 The Return of the Logical
example, he cannot explain the Hertz-Thomson episode, in which Thomson maintained, on empirical rather than a priori grounds, that Hertz’s experiment was not evidence against the charged particle theory (Achinstein 1991, 323– 326). Achinstein’s model accounts for experimental controversy insofar as the understanding of an experiment—which includes the assumption that B is true—is crucial for whether its results count as evidence. Thus, Thomson disputed Hertz’s assumption that no unknown background facts undermined the evidential status of his results. To test his suspicion, Thomson conducted experiments with superior vacuum technology and thereby showed that Hertz’s experimental results did not in fact supply evidence for ∼H. (Whether Thomson’s results then raised the probability of H above one-half is a further issue, about which Achinstein is somewhat elusive, but the details of this calculation do not affect the main point at issue here.3) Achinstein’s model also provides a more adequate conception of relevance than we find in Carnap or Hempel. As Achinstein (2001, chap. 3) is at pains to point out, logical empiricist models define relevance either too narrowly (in terms of deductive connection) or too liberally (in terms of E raising the probability of H in one way or another). The explanatory connection condition in Achinstein’s definition of potential evidence in effect broadens deductive models of relevance and links it with raising the probability of H above onehalf. At the same time, the objective character of the explanatory connection—it obtains whether anyone knows it or not—ensures the impersonal character of relevance: it is not audience-relative. 2
Cogency as Impersonal Merits
I doubt that Achinstein himself intends to offer his theory of evidence as a theory of argument cogency. Nonetheless, there are some intuitive grounds for making this move. He understands his four-step reconstruction of Thomson’s 1897 paper as a reconstruction of Thomson’s reasoning process or inference, and explicitly considers the paper to make an argument (Achinstein 1991, 287, 316). Moreover, he interprets the definition of potential evidence as articulating such ideas as having a “good reason to believe” a hypothesis H, the degree of reasonableness in believing H, or accepting H at some level of force. Intuitively, the idea of the reasonableness of believing a conclusion on the basis of evidential reasons seems to align with the cogency of an argument—after all, we consider it reasonable to accept cogent arguments.
88 Postscript I
To construe Achinstein’s definition as an objective model of cogency, I read the conditions of evidence as merits of an empirical argument in which an arguer adduces evidence E as grounds for inferring hypothesis H, that is, as a reason that justifies a belief that H. In construing things this way, I assume on the one hand a broad notion of argument, understood as a reason or set of reasons offered in support of a conclusion. On the other hand, I assume one may align the notion of argument with notions of inference, justification, reason giving, and warrant. To be sure, both sides of this alignment raise a number of further questions, which I cannot pursue here.4 For present purposes, it suffices to consider an argument as one type of justification or warrant in which the arguer states (or presupposes) an inference. If we take Achinstein’s model of potential evidence as a standard of cogency, then two important consequences follow. First, the truth conditions on potential evidence imply a definition of cogency in terms of impersonal merits: objective metaphysical facts that obtain whether anyone knows it or not. If an evidential argument is cogent only if its conclusion rests on potential evidence, then its merits depend not only on the impersonal truth of the stated evidence, but also on the truth of unstated assumptions that those who make the argument might not be in a position even to formulate.5 Second, like Hempel, Achinstein defines cogency in relation to the product. But unlike Hempel’s definition, Achinstein’s includes a condition that is not an intrinsic feature of the product per se. Whether a scientist has potential evidence is an empirical matter, dependent on whether or not background assumptions are in fact true. But background assumptions typically are not fully stated as premises in an argument; indeed, they probably cannot be fully stated (see Polanyi 1964). If unknown background assumptions are not plausibly considered an intrinsic part of an argument, but if their truth nonetheless partially constitutes or defines the argument’s cogency, then cogency cannot be defined in terms of intrinsic merits. Note that this condition also rules out the possibility of assessing the merits of evidential arguments on the basis of their intrinsic properties, inasmuch as accurate assessment depends on further information external to the argument itself. This last point raises a question about the status of process in Achinstein’s model, which I take up below. The concept of potential evidence, however, presents a serious obstacle for a theory of argument, namely, it separates the ideas of evidence and good reasons from argumentation as a practice of giving reasons. The problem is not that Achinstein makes the truth of reasons an impersonal merit of argument,
89 The Return of the Logical
but that according to his model, a premise is not a good reason at all unless it is true. If one understands “a good reason to believe H” as a kind of justification for believing H, then the concept of potential evidence collapses the distinction between truth and justification—specifically, this model would reduce justification to truth. Although some epistemologists have defined truth in terms of justification (e.g., Wright 1992), Achinstein would take us in the opposite direction.6 In fact, Achinstein (2001, 24–25) is careful to align justification, not with potential or veridical evidence, but with ES-evidence. So for the moment let us put potential evidence aside and consider ES-evidence as a candidate for a theory of argument cogency. ES-evidence is relativized to the “epistemic situation,” defined abstractly in terms of available information, apart from whether any person is actually in such a situation. An epistemic situation, that is, delineates “an abstract type of situation in which, among other things, one knows or believes that certain propositions are true, one is not in a position to know or believe that others are, and one knows (or does not know) how to reason from the former to the hypothesis of interest” (Achinstein 2001, 20). Different epistemic situations are distinguished by the different epistemic resources available to those in the respective situations, that is, the facts and methods of inference the situation makes available; I take these resources to include the available experimental techniques and methods that put one in such a position. We can spell out Achinstein’s formal definition of ES-evidence as follows: Empirical results E are ES-evidence that H if and only if (i) E is true, and (ii) one is justified, given the epistemic resources at one’s disposal, in believing that H is true, that there is an explanatory connection between H and E greater than 0.5, and that E does not entail H. Achinstein grants that an agent who is justified in this ES-sense can also be considered reasonable (ibid., 96–97). The notion of evidence retains an impersonal dimension insofar as E must in fact be true; but unlike potential evidence, ES-evidence makes the acceptance of an argument product reasonable even though there are unknown defeating conditions, as in the case of Hertz’s argument. Background assumptions that we cannot test need not be true. If we think of cogent arguments as justifying their conclusions relative to available knowledge, and thus as providing good reasons for those in a given
90 Postscript I
epistemic situation to believe the conclusion, then we can construe ES-evidence as an impersonal model of cogency. The “normative fact” of cogency does not supervene solely on the way the world is, but depends on what is and can be known at a given point in time, given available methods of research and inference. Thus we might say that ES-evidence supervenes on the facts scientists in ES could actually state in a verbal or textual product (and may reasonably assume), given their methodological resources. However, once an epistemic situation is identified, the cogency of arguments generated within that situation is defined impersonally: it depends on the truth of E and the abstract justificatory relationships defined by the available methods of inference. The only role the audience or particular scientific community plays in this mode of cogency is to provide a kind of boundary condition that defines which impersonal facts and inferential methods constitute the epistemic situation. Although Achinstein defines ES-evidence abstractly, apart from whether anyone is actually in ES, I suggest that we may regard his model here as a normative standard for assessing whether an actual group of scientists is justified in taking E as evidence for H. ES-evidence thereby functions as an intersubjective standard for criticizing subjective evidence. Once we identify the group at issue, in principle we can delineate their epistemic situation by identifying their stock of knowledge, background assumptions, and available inferential and experimental techniques. We can then regard an “evidential argument” as any attempt to offer E (normally, along with other theoretical considerations) as grounds for inferring H. Achinstein’s definition of ES-evidence then tells us when such inferences are justified, and hence when evidential arguments are cogent, for those who make them in ES, namely, only if E genuinely counts as ES-evidence for H. The definition of evidence depends on known standards and methods of inference (in condition ii above) and on the impersonal truth of E (condition i). How does this definition apply to Thomson’s argument, as outlined above? Achinstein (2001, chap. 13) argues that Thomson had ES-evidence, as well as potential and veridical evidence, for the existence of the electron. Here I am interested in the argument reconstructed in section 1 (which does not present the entire argument Thomson made in support of electrons). On Achinstein’s model of ES-evidence, we would assess the merits of that argument as follows. First, the experimental result claim in step (1) has to be true, in a mindindependent sense: it must be true that cathode rays were indeed deflected by a magnet and that when so deflected as to enter the coaxial chamber, an
91 The Return of the Logical
electrical charge was detected, and otherwise not. Notice that this claim is not trivial: it is not a “protocol sentence” in the logical empiricist sense or merely a first-personal report of what one experiences. Thomson must understand his experimental apparatus correctly: the magnet must actually be functioning, the necessary spatial alignments must not be mistaken, the electrometer must not be malfunctioning, and so on. Second, the experimental claim in step (1) does not directly entail the hypothesis stated in (3), that cathode rays carry a negative charge. For the claim in (1) is only that when cathode rays were deflected into the coaxial chamber, an electrical charge was detected, and otherwise not. But (1) does not claim that the rays carry the charge. Third, Thomson, given what he knows, must be justified in believing that the explanatory connection stated in step (2)—that the results in (1) are best explained by the fact that cathode rays carry a negative charge—has a probability greater than one-half. As Achinstein (1991, 293, 317) notes, this explanatory claim involves background assumptions about the behavior of matter in general (e.g., that the electrical charge detected in the coaxial cylinder whenever the cathode rays enter it is not simply created there). Such assumptions are justified only if they have some independent warrant on the basis of the observation of similar cases. Given such a connection and its probability, the experimental claim in (1) is relevant as an empirical test of the hypothesis. Fourth, the inference from (1) and (2) to (3), and the inductive generalization from (3) to (4), must be logically justified, given the methods of inference available to Thomson. The estimation of probability in (2) must also be justified in light of available methods. Fifth, any known background assumptions must be justified (as noted above for the explanatory claim) but they need not be true. This permissive condition raises questions about how one distinguishes the experimental result claim from background assumptions about the apparatus. Presumably the result claim refers to actual observations, which are made under a given set of conditions that one monitors by using available methods. But I doubt one can fully immunize the truth of such observation claims from false background assumptions, particularly assumptions that underlie one’s understanding of properly functioning apparatus and desired experimental conditions. Here too one should expect to find deep assumptions about matter, and if some of these are false then the very description of the experimental results could also be false. Such assumptions seem to differ from Hertz’s assumption that he had
92 Postscript I
sufficiently evacuated the bulb so that residual gas did not interfere with his results. The falsity of his assumption meant that what he claimed to observe was not potential evidence against the particle theory; but the truth of his observed results remained intact, on Achinstein’s account, for Hertz had ESevidence in 1883. 3
Impersonal Merits and the Aims of Argument-Making
As I noted in chapter 1, argumentation theorists typically approach arguments as the products of argumentative practices, understood as a type of social practice. On this view, the argument product, a package of reasons supporting a conclusion, issues from a social process of arguing that is more or less regulated by formal or informal procedures. In this last section, I situate Achinstein’s model in the context of argument-making practices and their aims. Achinstein’s analysis of evidence, when applied to the idea of cogent argumentation, is most readily understood as a logical perspective on the products of argumentation. So construed, his model identifies the merits of arguments as including the relevance of evidential premises, sufficiency of support, and the truth and justifiability (or rational acceptability) of premises. More specifically, his model of ES-evidence implies that (a) Evidential premises E are relevant only if they do not entail H but, when taken together with background conditions B, they sufficiently support the hypothesis H; (b) E sufficiently supports H only if the available information and methods justify the greater-than-half probability of an explanatory connection between E and H, given E and B; (c) E must be true in an impersonal sense that depends on objective facts; (d) B must be justified according to the available information and methods. If these conditions obtain, then anyone in ES is justified in taking E as evidence for H, or in regarding an argument for H on the basis of E as cogent. Whether one’s ES-evidence is also potential and veridical evidence, however, depends on whether the probability claim in (b) and the background assumptions in (d) are objectively true. Because scientists can test such truth only through further research, potential evidence and veridical evidence are empirical matters. Thus a person with ES-evidence might not have potential evidence for H.
93 The Return of the Logical
However, Achinstein’s model implies that a scientist with ES-evidence but not potential evidence would be justified in believing H while lacking good reasons for that belief. This result sounds rather odd in an account of cogent argumentation, where the ideas of good reason and reasonable belief depend, at least in part, on what arguers can actually know and state.7 As it turns out, an argumentation-theoretic perspective suggests a more plausible approach, which better captures the features of science that motivate Achinstein’s analysis. As I noted above in section 1, Achinstein proposes his objective (i.e., empirical) conception of evidence to account for two important features of scientific practices, namely the potential for controversy and the commitment to ongoing research. Achinstein explains these features with the thesis that scientists want veridical evidence: true hypotheses based on true evidence and true background assumptions. Here I take this not so much as a sociological thesis as an interpretive claim about a normative aim of scientific practices. Nonetheless, it has a sociological payoff: Achinstein can explain why the science community, taken as a whole, continues to probe its experimental and theoretical assumptions by raising critical objections and conducting further tests, as Thomson did with Hertz’s experimental claims. On this view of scientific practice, ES-evidence falls short. If scientists strove only for ES-evidence, then nothing intrinsic to their practice of evidential argument would motivate them to go beyond experiments and hypotheses that were justified on the basis of the available methods and information. As a normative account of scientific practice, ES-evidence cannot account for progress. On the one hand, then, Achinstein’s concept of ES-evidence offers a plausible candidate for a logical perspective on argument cogency, inasmuch as the definition of ES-evidence gives us usable standards for determining whether someone is justified in inferring H on the basis of E. If the sciences are one sort of argument-making practice, then we need to account for the justificatory aims of evidential arguments. Whatever else one wants to achieve by putting forth an argument, the idea of justification or warrant seems to be constitutive: a set of statements or symbolic representations can count as an argument at all only if one can regard it as a set of putative reasons that are supposed to support a conclusion.8 On the other hand, if we accept Achinstein’s interpretation of scientific practices as ongoing inquiry oriented toward truth, we need the stronger concept of veridical evidence. For Achinstein, scientific practices have veristic
94 Postscript I
aims that transcend what we might be able to justify on the basis of our current information and methods. Scientists achieve those aims, and thus acquire veridical evidence, only if their background assumptions are true. To get at this truth, scientists strive to bring background assumptions to the fore for critical testing, and such testing typically requires an ongoing intersubjective process of critical discussion (and further inquiry) in which arguments confront challenges of various sorts. We can interpret the difference between ES-evidence and veridical evidence in terms of two aims of argument-making practices in science: the construction of evidential arguments that justify our taking a hypothesis as probably true, on the one hand, and the justification-transcendent aim of veridical evidence, on the other. Applied to argument evaluation, ES-evidence is important in science because it serves the aim of justification. The only accessible warrant that evidential arguments can provide for thinking we have veridical evidence lies in the justificatory merits of those arguments. Insofar as Achinstein links ES-evidence with justification, he makes the latter relative to an epistemic situation: the available methods of testing and inference. To that extent, ESevidence provides an accessible warrant for believing one has veridical evidence. But one rough spot remains in this interpretation, which prevents us from entirely aligning ES-evidence with justificatory aims: Achinstein includes the truth of experimental results E as a condition on ES-evidence. As a metaphysical condition, truth lies beyond (i.e., is not ensured by) epistemic justification per se. The truth of E, after all, is neither self-evident nor trivial, but depends on further assumptions regarding experimental conditions and apparatus, as I noted above. As the logical empiricists were well aware, one can at most take E as true insofar as it is justified in the light of corroborating observations and available knowledge. If we understand ES-evidence as serving strictly justificatory aims of argument-making, then we must replace the truth condition on E with a justifiability condition. In linking ES-evidence strictly with justification, however, we need not reject the idea of truth as unimportant for scientific argument-making. Here Achinstein’s concept of veridical evidence—which pulls together the various truth conditions one finds in the other concepts of evidence—serves to capture what I call the “representational aims” of scientific argumentation. It is debatable whether one must understand successful representation in metaphysical terms, as Achinstein does, but I leave that issue aside for now.9 For the moment notice that the distinction between justificatory and representational aims
95 The Return of the Logical
captures the features of scientific practices that Achinstein wants to account for with his concept of evidence. In their practices of inquiry and argumentmaking, scientists seek cogent evidential arguments that justify hypotheses. But they want such arguments precisely because these provide the sole indicators of truth. The justificatory merits of arguments, that is, are all we have for believing that those arguments succeed in their representational aims—that the arguments supply us with veridical evidence for claims about nature. As a fallible means to truth, arguments as justifications make sense in view of a representational enterprise whose success is measured by the world. Because this representational aim transcends—its success is not ensured by—arguments per se, the latter, no matter how cogent as justifications, remain fallible in principle and thus open to criticism in the light of further research. To summarize, then, Achinstein’s theory of evidence suggests the following distinction for a conception of cogency: (1) a (synthetic) definition of veridical evidence that captures the truth at which scientific inquiry aims: true hypotheses supported by true evidence-statements and assumptions; (2) a (synthetic) definition of ES-evidence that, if shorn of its truth condition (that E must be true), aligns cogency with properties of the evidential arguments that scientists are in a position to make and assess; insofar as those arguments succeed at providing cogent justifications, they fallibly indicate success at the representational aims of argument, stated in (1). Although neither side of this distinction refers explicitly to the process of inquiry, together they imply that the process is necessary for realizing the aims of argument-making in science. The reason is that success in the justificatory aims of argument (2) leaves open the possibility that unknown background assumptions might be false. Thus a gap opens up between the two aims in (1) and (2), giving scientists a reason, internal to the definitions as statements of the aims of argument-making, to conduct ongoing inquiry in which they elicit and probe their background assumptions. (More on this dynamic aspect below.) We can now see how Achinstein’s concept of potential evidence might serve in a theory of argument cogency. If we regard argument-making in science as oriented toward the production of cogent arguments that indicate the probable truth of the conclusion, and if we regard an argument as a more cogent indicator when it does not rest on false background assumptions, then we can say that scientists aim to produce arguments that are cogent in Achinstein’s sense of potential evidence. The concept of potential evidence thus bridges the gap between (1) and (2) above. In seeking potential evidence, scientists seek better
96 Postscript I
arguments, arguments whose cogency is objective in a mind-independent sense that does not hold for our modified concept of ES-evidence. But they also recognize that in having such arguments, their conclusions might still be false: potential evidence does not guarantee veridical evidence. Thus interpreted, Achinstein’s potential evidence provides an impersonal definition of cogency in which the mind-independent truth of the premises and background assumptions partly constitutes the cogency of arguments. The inclusion of truth in conceptions of cogency is not at all untypical among argumentation theorists. If there is a problem with potential evidence, then as I pointed out in section 2, it lies in Achinstein’s equating the truth of the premises with their status as good reasons (which he then distinguishes from justifying reasons relative to an epistemic situation). Although the word “reason” is used in various ways, it strikes me as terminologically preferable, in the context of scientific inquiry, not to conflate metaphysical and epistemological categories, as Achinstein does in talking about mind-independent reasons. Achinstein’s terminology implies that one can have good reasons for, but not be justified in accepting, a hypothesis; conversely, one might be justified without good reasons. In fact, the terminological problem disappears once we notice a further point about the concept of potential evidence. To bridge the representational and justificatory aims of scientific argumentation as a dynamic process, the concept of potential evidence must retain a connection with ES-evidence as providing justification. Scientists do not simply conjure up arguments that they hope satisfy the truth conditions of potential evidence. They strive for cogent justifications. In the actual practice of inquiry, the mind-independent “good reasons” of potential evidence must also function as an ES-justification. We can dispense with Achinstein’s locution, and still account for the dynamics of inquiry, if we say that scientists strive for potential evidence in the sense of good (justifying) ES-reasons that happen to be true. The upshot is that the most plausible definition of cogency we can find in Achinstein’s theory of evidence is one that combines the conditions of potential and ES-evidence. An evidential argument is cogent, that is, if and only if: (i) the probability p (that there is an explanatory connection between H and E, given E and B) is greater than one-half; (ii) E and B are true; (iii) E does not entail H; and
97 The Return of the Logical
(iv) one is justified, given the epistemic resources at one’s disposal, in believing (i–iii). According to this definition, scientists ought to strive to produce cogent arguments that meet this combined set of conditions, which in turn provide cogent grounds for the probable truth of the given hypotheses. This definition of cogency is impersonal: the cogent arguments that scientists strive to construct and evaluate are constituted in part by the mind-independent truth of their evidence and background assumptions. Consequently, making and assessing cogent arguments requires scientists to go beyond examining the argument in light of the current epistemic situation: they must create new situations by developing new techniques and instruments and engaging in further empirical research. Thus the dynamic aspect of Achinstein’s model emerges not in the definition of cogency itself but in the construction and assessment of arguments. The definition of cogency explicitly refers only to the product and its unstated assumptions. But to generate and accurately assess cogent arguments depends on the dialectical quality of critical testing and discourse in the science community. To the extent that scientists’ confidence in their arguments stems not simply from direct inspection of the product but also from the quality of their methods of inquiry—in the broad sense of ongoing critical testing in an open community of investigators—they assess impersonal merits indirectly, as merits relative to specific challenges made by a particular community. Confidence in arguments thus presupposes that scientists consider their procedures of inquiry and discourse to meet the kinds of standards that Pera spells out, which govern the course and obligations of discussion. Once we bring in procedural considerations, however, the question arises of whether one can define argumentative merits in wholly impersonal terms— that is, simply in terms of truth and the available logical standards of justification. If argument-making in the sciences depends for its representational success on an ongoing dialectic of critical inquiry, then successful argumentation depends on a social, communicative process. In making cogent arguments, scientists do not aim merely to represent the world with logically well-constructed justifications. As products of a social practice of critical discussion, arguments are generally made to lead addressees to accept a particular conclusion as probably true or at least reasonable. This is especially true in science, which traditionally aims to produce public knowledge. Although the logical
98 Postscript I
construction of arguments—success in the justificatory aims of argument-making— certainly helps in making the argument publicly acceptable, the logical perspective cannot entirely capture this aspect of argumentative practices. If we understand the justificatory aims of scientific arguments simply in terms of impersonal logical standards in the service of representing a mindindependent reality, then we miss the importance of the communicative aims of argument-making practices. As I have construed Achinstein’s approach, it implies that success in the communicative aims of argument is simply a necessary means in the construction and assessment of cogent arguments, whereas cogency itself is defined by the impersonal merits of truth and logical justifiability, that is, in terms of the justificatory and representational properties of the argument product. But it is not clear that we must take this approach to cogency, even if we accept Achinstein’s concept of potential evidence as a standard of cogency. If cogent scientific arguments are arguments that succeed in the core normative aims of scientific argument-making, and if those aims include public acceptability, then it seems that cogency involves not only the satisfaction of justificatory and representational aims, but also the satisfaction of communicative aims. Analytic philosophers of science have long accepted the idea that the communicative and social aspects of science are crucial to the rationality of inquiry and thus to epistemic success. In this sense, they do not reject epistemic approaches to the dialectical and rhetorical aspects of scientific discourse (e.g., McMullin 1991; Kitcher 1993, 1995, 2000). Less clear, however, is whether they want to include these conditions on rational process among the defining features of good arguments. To rephrase this question nondefinitionally: do we want to say that arguments ought to be regarded as cogent simply in virtue of identifiable merits of the product (perhaps broadened to include background assumptions)? Or should we rather say that (many if not all) arguments should count as cogent only if they have emerged from a sufficiently reasonable process of argumentation? The latter conception suggests that the same argument content could become more cogent as it held up under increasingly severe argumentative criticism, such that we do not simply come to understand the argument’s cogency more accurately; we actually make the argument more cogent. Once we bring abstract concepts of justification and good reasons into an argumentation-theoretic framework, we cannot avoid this question. In Part II, I examine Habermas’s argumentation theory with this question in mind.
II Integrating Perspectives: Habermas’s Discourse Theory
4 Habermas’s Critical Theory and Science: Truth and Accountability
In the previous chapters I have sketched a range of perspectives that emerged in the science studies literature before and after Kuhn. The various perspectives align with different approaches to the study of argumentation— logical, dialectical, rhetorical, sociological—each of which implies a different approach to the cogency of scientific arguments. If one wants to bring the full panoply of argumentation-theoretic resources to bear on critical assessment, however, then some integration, and most likely some selection, is required. In this and the following chapters I examine in detail an attempt to integrate these perspectives on argumentative practices: Habermas’s discourse theory. Habermas’s work is relevant here for several reasons. First of all, he explicitly draws on argumentation theory, attempting to integrate different aspects and standards of argumentation. In doing so he illustrates one way to bridge the gap in Kuhn, that is, to do justice to both philosophical and sociological approaches to argumentation. Second, his approach aims at critical evaluation inasmuch as the argumentation theory provides much of the normative ballast for a critical theory of society. Finally, he maintains some degree of interdisciplinary openness, a willingness to find valid ideas in approaches with which he does not entirely agree. Habermas’s critical theory also brings out important issues of theoretical strategy. The project I described in chapter 1—to develop a normative framework for interdisciplinary assessment of scientific arguments—cannot avoid certain questions. Besides the primary issue of theoretical integration, one must address the following:
102 Chapter 4
(1) questions of detail, arising from disputes over the details of how one understands specific points, say of rhetorical analysis, dialectical obligations, and so on; (2) questions of interdisciplinary cooperation, arising from deeper philosophical disputes between different approaches to argumentation; (3) questions of vision, arising from differences in how one conceives the relation between science and society, and from disputes over the desirability and feasibility of different versions of the “contract” between science and society. These questions call for choices that affect theory formation in different ways, with the most profound effects arising from items (2) and (3). Habermas’s discourse theory provides a set of answers to such questions that is both broadly amenable to the argumentation studies framework and leads to determinate critical assessments. In Part II, I elaborate Habermas’s discourse theory as a theory of scientific inquiry and argumentation. Although Habermas addressed issues of science and technology in the 1960s (Habermas 1970), at that time he had not yet elaborated a discourse theory. That theory is now in place, but its application is mainly limited to morality, politics, and law (Habermas 1996, 1998a). Thus a certain amount of work is required to see what a discourse theory of science might entail. In this chapter I provide some background on Habermas’s larger critical-theory project and explain why he regards his theory of communicative action as important for the project. This project involves a particular vision of science in society, and thus provides an answer to question (3) above. The argumentation theory itself I sketch in chapter 5. Together with the ideas in chapter 4, Habermas’s argumentation theory provides an integrated model of cogency that answers the kind of questions of detail arising in point (1) above. After illustrating (and partially testing) Habermas’s model with a case study (chap. 6), I address the question of interdisciplinary cooperation in the postscript to Part II by tackling the hard case of cooperation between a Habermasian and a relativistic SSK approach. As we move through Part II, I point out a number of problems in Habermas’s model, which ultimately stem from the way he attempts to integrate perspectives on cogency. Although his approach can account for much of the case study and allows for some measure of cooperation with SSK relativists, his way of integrating perspectives harbors a deep obstacle to interdisciplinarity—the problem that leads me to propose a contextualist alternative in Part III.
103 Habermas’s Critical Theory and Science
1
The Pragmatic Turn in Critical Social Theory
When Max Horkheimer formulated the tasks of the Institute for Social Research in 1931, he conceived of critical theory as an interdisciplinary endeavor in which social philosophy and the empirical sciences mutually informed one another. In line with their Marxist heritage, first-generation Frankfurt School theorists such as Horkheimer assumed that political economy would play an important role in this project (Horkheimer 1931/1993; Bonß 1993). Influenced by Max Weber, these theorists situated questions of political economy in the broader context of problems connected with societal rationalization (the growth of science and technology, bureaucracies, etc.). As Europe descended into the depths of totalitarianism, however, economic analysis gave way to the pressing need to rethink the concept of rationalization on which critical theory relied. Second-generation theorists such as Habermas also saw a need for a more adequate theory of democratic politics. To carry out this project, Habermas, following the lead of Karl-Otto Apel, drew more on Kant, American pragmatism, and the philosophy of language than on Marx (Habermas 1979; Apel 1980; see also McCarthy 1978; White 1989; Mendieta 2002). We can situate Habermas’s approach within the broader “linguistic turn,” which we may understand here negatively as a turn away from both the premoderns’ metaphysical concerns with the structures of being and the moderns’ epistemological focus on the subject’s access to an objective world. Positively, the turn focuses on the conditions of language use as the unavoidable context and medium for philosophical inquiry. Habermas gave the linguistic turn a specifically pragmatic emphasis by arguing that an adequate theory of linguistic meaning must go beyond the analysis of syntax and semantics, inasmuch as these do not fully account for the ability of competent speakers to use grammatically well-formed sentences in ways that are appropriate for social interaction. He proposed that this pragmatic know-how could be captured by a set of basic linguistic structures that underlay all language use—thus a “formal” or “universal pragmatics” (Habermas 1979, chap. 1). Here “formal” refers not to the syntax of deductive entailments but to the general presuppositions of the use of language.1 In his effort to rework the rational basis for social critique, Habermas argued at length that the conception of rationality dominant in the social sciences (and in earlier critical theory) captured only one aspect of the richer communicative
104 Chapter 4
rationality on which social integration depends (Habermas 1984, 1987). According to the dominant model, reason operates above all in fact-stating forms of empirical discourse, as exemplified in the natural sciences. In the practical domain, the positivist emphasis on factual truth (or empirical adequacy) translates into an exclusive focus on means-end reasoning: with ultimate ends assumed as given, the task of reason consists in the comparison and choice of efficient means for achieving those ends. Action based on this conception of practical reason is instrumentally rational (zweckrational) or, in social contexts, strategic: one approaches the choice situation, including other actors, as a set of conditions that must be controlled (as means) for the sake of the rational pursuit of one’s own ends. Habermas also found hermeneutical and phenomenological alternatives to positivism lacking (1988; 1971/2001). To get beyond truncated conceptions of rationality, he elaborated a broader conception of reason as it operates in everyday social interaction. The key to his analysis lies in the concept of “communicative action,” which he contrasts with strategic action. If actor A interacts with actor B strategically, then A tries to influence B’s behavior in a manner favorable to A’s goals by appealing to B’s desires and fears (preference orderings)—which, as “reasons” for B’s cooperating with A, are only arbitrarily related to A’s goals. In communicative action, by contrast, A presents B with “validity claims” that B is free to accept or reject on their merits. Thus if B cooperates with A, then B does so because of claims that A and B both accept as deserving recognition on the basis of mutually convincing good reasons. Such reasons are inherently related to A’s goals and freely accepted by the cooperating parties. Notice here that the phrase “validity claim,” as a translation of the German term Geltungsanspruch, does not have the narrow logical sense (truthpreserving argument forms), but rather connotes a richer social idea—that a claim (statement) merits the addressee’s acceptance because it is justified or true in some sense, which can vary according to the sphere of validity and dialogical context. Although Habermas argues that communicative action enjoys a kind of priority in social interaction, the two types of action typically intertwine. Moreover, in certain social domains, such as economic markets, more efficient social coordination is possible if we give greater play to strategic motivations (within limits). What concerns me at this point, however, is the contrast between strategic rationality and the richer conception of communicative rationality, which responds not simply to facts but also to moral and evaluative claims. In sharp
105 Habermas’s Critical Theory and Science
contrast with the positivist tradition, Habermas does not limit intersubjectively valid, or justifiable, claims to the category of empirical truth, but instead recognizes a spectrum of validity claims that also includes, at the least, claims to moral rightness, ethical goodness or authenticity, personal sincerity, and aesthetic value (Habermas, 1984, 8–23; 1993, chap. 1). Although Habermas does not consider such claims to represent a mind-independent world in the manner of empirical truth claims, they can be both publicly criticized as unjustifiable and defended by publicly convincing arguments. To this extent, validity involves a notion of correctness analogous to the idea of truth. At the core of Habermas’s theory of communicative action (TCA), then, lies the thesis that ongoing social interaction requires actors continually to make and accept validity claims on the supposition that good reasons could, if necessary, be supplied to justify the claims. Consequently, communication contains an implicit reference to argumentation or discourse, the form of communication in which actors put aside the pressures of action in order to critically discuss a validity claim that has become problematic.2 The rational basis of communication and social order thus lies in the rationality of discourse as critical discussion: if for example we accept a moral judgment as right, or an empirical claim as true, then we tacitly assume the judgment or claim would hold up in a rational discourse. To spell out the conditions on such discourse, Habermas draws on the work of Robert Alexy (1990; Habermas 1990, 86–89). But he organizes Alexy’s standards according to the familiar tridimensional perspectivist model that analytically distinguishes three aspects of argumentation—product, procedure, and process, which Habermas loosely aligns with the logical, dialectical, and rhetorical perspectives of the Aristotelian canon (cf. Wenzel 1990). To anticipate the next chapter, Habermas’s argumentation theory implies that the cogency of arguments depends not only on their logical structure and content, but also on the dialectical and rhetorical virtues of the social-institutional procedures and processes that generate them. Habermas’s pragmatic turn has significant implications for his understanding of the critical-theory project. Like critical theorists before him, he has a vision oriented by the idea of social emancipation. Unlike previous critical theorists, he looks to deliberative democracy rather than a radically altered mode of production for the social locus of emancipation. By emphasizing communicative rationality as the basis for social integration, Habermas takes an approach to critical social theory that naturally supports a democratic vision in which reasonable public deliberation plays a central role in legitimate
106 Chapter 4
politics and law. He thus criticizes theoretical approaches that reduce democratic politics to the competition of interests for power. Although he acknowledges the importance of the strategic use of power and bargaining in the democratic process, political questions ultimately involve matters of justice and the common good—matters that involve validity claims over which citizens can engage in reasonable discourse. The burden of Habermas’s “discourse theory of law and democracy” (1996) is to elaborate a normative framework for critically assessing political institutions and the discourses they support. As these introductory remarks suggest, Habermas’s critical theory rests on a number of methodological commitments. Two are especially important for the critical evaluation of scientific argumentation: the formal character of a critical argumentation theory, and the standpoint of the critical theorist. 2
Two Methodological Commitments in TCA
Habermas ultimately wants to elaborate a comprehensive normative theory of social integration for purposes of understanding and criticizing the full range of contemporary social practices and institutions (Habermas 1984, preface). Such a project, he believes, needs a more or less complete set of formal analytic categories. That is, the critic must be able to resolve the tangle of conventional and institutional practices and procedures into their core formal components. TCA supplies the framework for this formal analysis, the basis for a set of normative standards by which one can critically scrutinize the social rationality of these practices. Without such formal standards, Habermas fears, the critical project founders, possibly even falls into an ungrounded relativism. Specifically, he wants to avoid an approach that offers no higher perspective, or context-transcending basis, for critically evaluating the normative conventions of existing discursive practices and institutions as they are historically given. We can see Habermas’s concern for relativism in his critique of Toulmin. Habermas (1984, 31–38) is particularly impressed with Toulmin’s well-known model of informal logic, which is both multidimensional and aims to avoid relativism. As Habermas understands the model, it lays out a general structure for any nondemonstrative argument product. In fact, Toulmin (1958, chap. 3; cf. Toulmin, Rieke, and Janik 1984) first approached his analysis as a reconstruction of the process of argument construction.3 If we make a claim (the conclusion C) that someone challenges, then our first response typically invokes the immediate reason supporting the conclusion. If I say, “Smith deserves a
107 Habermas’s Critical Theory and Science
speeding ticket,” for example, I might support my claim by adding, “Smith’s automobile was clocked at 90 mph on the freeway,” thus citing concrete situational “data” (D). In considering this data a reason for my conclusion, I assume a “general warrant” (W) of some sort, most likely some traffic law (e.g., “If a motorist is clocked at speeds above 60 mph on the freeway, then that motorist is subject to a $100 penalty”). Although in my example the warrant and data generate the conclusion deductively, this impression is misleading. Conclusions are usually asserted with more or less force, which Toulmin explicates as the modal “qualifier” on the conclusion, expressed by words such as “normally,” “presumably,” “certainly,” and the like. We can usually say something about the conditions under which the warrant would not hold, that is, possible rebuttals (R) to the conclusion (e.g., “Smith was responding to an emergency,” or “Smith is a policeman, and was pursuing a criminal,” etc.). The qualifier means that apparent deductions (D, W ⬖ C) often conceal defeasible arguments. Moreover, the warrant is supported in turn by “backing” (B), which often involves a nondemonstrative ampliative move. In the current example, the backing would include the reasons for considering the traffic law against speeding as generally binding law. The model may be diagrammed as in figure 4.1. The force of the overall argument ultimately depends on the backing-warrant connection, and how this connection functions depends on the particular field of argumentation (law, morality, science, management, art criticism). That is, the backing-warrant link has force in virtue of the vocabulary and concepts employed in a given domain of argument, the institutional standards for adequate backing, and the aims of the enterprise. It is just at this point, the institutional basis, that Habermas detects a weak spot in Toulmin’s model.4 Specifically, the model lacks a conceptual grounding for the differentiation of fields of argumentation. This deficit leaves the analysis at the mercy of historically contingent institutional conventions: “[Toulmin] doesn’t draw the lines between accidental institutional differentiations data
(qualifier) conclusion warrant backing
Figure 4.1
unless R
108 Chapter 4
of argumentation, on the one hand, and the forms of argumentation determined by internal structure, on the other” (Habermas 1984, 35). As a result, he lumps together very different kinds of argumentation; he also fails to separate “conventional claims, which are context-dependent, from universal validity claims” (ibid., 36). Legal argumentation, for example, is regulated by a range of conventional procedures and involves a significant component of bargaining. But according to the ideal image of law, these institutionally specific aspects of law should serve the deeper demands of truth and justice, which thus provide a basis for critically evaluating legal conventions. According to Habermas, such criticism presupposes that one can distinguish truth and justice as basic (nonconventional) types of validity claims that impose certain general demands on reasonable legal discourse on the one hand and, on the other, set limits on the forms and amount of bargaining (a mode of strategic action). An adequately grounded critical theory of law, in other words, presupposes a deeper, formal basis in communicative rationality that cannot be assumed to align fully with existing institutions. TCA, with its formal system of validity claims and sharp distinction between communicative and strategic action, allows one to isolate the validity basis of legitimate institutional procedures and outcomes. The failure to make this distinction between institutional procedures and communicative rationality in general, Habermas maintains, weakens the critical potential of Toulmin’s model, which seems to accept domain conventions without question. Such ungrounded acceptance of existing institutional conventions exposes the model to relativist interpretations of argumentation. Toulmin, in other words, gets it backward: instead of starting with a formal analysis of valid argumentation as a guide for understanding institutional domains, he uses institutionally established “fields of argument” (law, science, management, etc.) as a guide for distinguishing types of argumentation. To be sure, Habermas (ibid., 38) acknowledges that a formal approach places a heavy burden of proof on argumentation theory: one must “be in a position to specify a system of validity claims” that transcends existing institutions and thus provides a nonrelativistic basis for critical assessment. In attempting to meet this requirement, Habermas presents a range of arguments, but before addressing these I must clarify a second methodological commitment. The second commitment concerns the standpoint the argumentation theorist must adopt in developing a critical framework. To get at the formalpragmatic basis of rational argumentation, Habermas believes the theorist cannot simply describe processes of argumentation as they actually unfold
109 Habermas’s Critical Theory and Science
but must adopt the perspective of engaged participants who are concerned about the rational cogency of their arguments. Like participants, the theorist must take a prescriptive or evaluative approach to argumentative practices. However, unlike participants in actual discourse, the critical theorist must go beyond the substance of arguments and articulate the formal features of cogency: what in general makes some arguments better than others. One thus attempts to reconstruct the underlying, general “pragmatic presuppositions” that participants make about cogency when they engage in substantive argumentation. In other words, the theorist aims to articulate the intuitive know-how of competent language users. To do so, one must draw on one’s everyday knowhow and experience in discourse. This orientation to everyday “communicative competence” has a peculiar effect on the status of the theorist as an “expert.” Formal pragmatics has a philosophically technical character, and the critical theorist has a philosophical expertise based on knowledge of a specific technical literature and vocabulary. But these technical tools serve for analyzing communicative practices with which socially adept laypersons are intuitively familiar. Thus the theorist’s expertise differs from the technical expertise of scientists in virtue of its internal relation to the know-how of laypersons, since that is what reconstructive theories strive to make explicit. Theoretical attempts to articulate that pretheoretical know-how are thus beholden to lay knowledge: lay participants should be able to recognize their own discursive ideals in the theorist’s reconstructions. Habermas defends this approach by contrast with the alternative approach proposed by Wolfgang Klein. As Habermas (1984, 26–31) reads him, Klein hopes to avoid relativism, and thus wants to get at universal structures of argumentative cogency that are not at the mercy of local variations. Klein, however, also wants to avoid prescription, and thus restricts himself to empirical descriptions of how participants actually argue. He assumes that such description can uncover “relatively fixed regularities . . . according to which people argue: precisely the logic of argumentation” (quoted in Habermas 1984, 30). As Habermas points out, however, this descriptive-nomological move reduces valid argumentation to rhetorically effective argumentation. Klein thereby “assimilates rules to causal regularities” (ibid.). Rather than get tangled up in the relationship between reasons and causes, one might develop Habermas’s criticism more straightforwardly by noting that Klein—again, as Habermas reads him—will have difficulty with the fact that
110 Chapter 4
participants themselves often distinguish between effective rhetoric and valid argument. To understand argumentative practices, one must elaborate the basis for this distinction as participants employ it—an analysis that Klein’s model apparently excludes from the start inasmuch as it identifies validity by describing observably effective rhetoric. Even if he can reply to this objection, a deeper problem afflicts his approach, namely the assumption that the normative aspects of social practices can be cashed out as observable regularities. This assumption has been criticized by a number of theorists. Rouse (2001), for example, insists on the difference between the dubious conception of social practice conceived as a set of regularities and the more defensible normative conception of practice (see also Brandom 1994, chap. 1). Turner (1994; cf. Lynch 1997b) has extensively criticized the causal conception of practice, particularly as a basis for sociological explanation. A normative conception, by contrast, does not aim at causal explanation nor does it assume that the rules that structure practices appear as empirical regularities (though in some cases they might). Drawing on Brandom, Rouse (2001, 190) notes that identifying the rules of a practice might in some cases always depend on reference to other rules and normative practices, without ever ending “in some kind of objectively recognizable regularity.” My aim at this point is not to resolve the above issues, but to understand Habermas’s approach. In closing this section, I notice how that approach seems to put the theorist in a rather puzzling position. On the one hand, as I mentioned above, the theorist is beholden to lay knowledge. On the other, Habermas’s critique of Klein implies that one may not simply take actual discursive practice as the decisive evidence against which normative reconstructions are tested in the search for general structures. Although formal-pragmatic reconstructions must appeal to some kind of empirical evidence in actual practice, the appeal cannot have the character of an induction that simply identifies general regularities. In other words, the theorist uses her own competence as an arguer to critically sort through the data of actual argumentative practice—therein lies the prescriptive side of formal-pragmatic reconstruction. This implies that the theorist is not beholden to lay judgments about cogency in any simple way. But the danger thereby arises that the theorist simply selects features of argumentative practices that support the formal-pragmatic model, while dismissing as aberrations and errors those moves that do not fit the model. I take up this difficulty in chapter 7.
111 Habermas’s Critical Theory and Science
In any case, Habermas is clearly confident that a participant-perspective reconstruction of argumentative practices will uncover a formal system of universal validity claims. I sketch that system in the next section. 3
The Validity Basis of Speech
As a pragmatic theory of meaning and communication in general, TCA grounds the system of validity claims as a universal basis for critical assessment. This undertaking is an ambitious one that has been subjected to a range of criticisms and undergone a number of subtle changes.5 My prime concern here is the truth claim, which according to Habermas’s model has primacy in the sciences. But before delving into his theory of truth I want to sketch the broader formal-pragmatic framework in which truth is situated. To get at this framework, I start with TCA as a theory of the meaning of speech acts. Habermas (1998b, 232) formulates the core principle of his theory as follows: “We understand a speech act when we know the kinds of reasons that a speaker could provide in order to convince a hearer that he is entitled in the given circumstances to claim validity for his utterance—in short, when we know what makes it acceptable.” By linking meaning with the acceptability of speech acts, Habermas moves the analysis beyond a narrow focus on the truthconditional semantics of representation to the social intelligibility of interaction. The complexity of social interaction then allows him to find three basic validity claims potentially at stake in any speech act used for cooperative purposes (i.e., in “strong” communicative action; Habermas 1998b, chap. 7). Habermas’s argument, which has a transcendental character, relies on three “world relations” that are potentially involved in strongly communicative acts in which a speaker intends to say something to someone about something (see Habermas 1984, 275ff.). Consider, for example, a fact-stating speech act. With it, the speaker expresses an inner world (an intention to communicate a belief), establishes a communicative relation with a hearer (and thus relates to a social world), and attempts to represent the external world. This triadic structure suggests that many speech acts involve a set of tacit validity claims: the claim that the speech act is sincere (nondeceptive), is socially appropriate or right, and is factually true. Conversely, speech acts can be criticized for failing on one or more of these scores. Thus successful speech acts, insofar as they involve these three world relations, must satisfy the demands connected with these three basic validity claims (sincerity, rightness, and truth) in order to be acceptable.
112 Chapter 4
For example, suppose an executive asks her secretary to make coffee. This request is rationally acceptable only if (a) it is factually true that the ingredients and instruments for making coffee are at hand; (b) this task falls within the secretary’s job description and the executive’s rights, which in turn satisfy universal standards of moral rightness; and (c) the executive is not making the request for some dubious ulterior motive (e.g., to humiliate the secretary). The truth claim (a) is empirically testable, the rightness claim (b) can be supported or rejected on the basis of institutionally defined role descriptions that are in turn assessable from the standpoint of justice, and the sincerity claim (c) can be questioned and defended by referring to other aspects of the executive’s behavior vis-à-vis the secretary. This transcendental argument probably does not succeed as a theory of meaning, at least in the standard analytic sense (Heath 2001, chap. 3). But it does give us a suggestive critical heuristic for exploring the pragmatics of the rational acceptability of speech acts—clearly one of Habermas’s aims, in any case. In the context of scientific communication it implies that scientific claims, as speech acts, involve more than empirical truth. Although Habermas takes the truth claim as the central, or most salient, type of claim in scientific argument, such argument at least potentially involves, in addition to truth, the claims of sincerity and rightness. Broadly construed, the former refers to the inner dispositions of the scientists themselves—not simply their honesty but also their self-understanding, their identity as scientists, their attitude of objectivity. In controversies, for example, opponents sometimes attack one another along such lines: that the opposing scientist is biased, and thus untrustworthy. The rightness claim implicit in speech acts implies that the relationship between speaker and addressee is “right” from a normative perspective. Criticisms of scientific arguments that appear to head in racist directions can perhaps be understood as attacks along the normative dimension. More broadly, the rightness dimension can be understood to include the value-laden presuppositions that underlie many scientific arguments (e.g., that the truth one asserts is “significant,” that confidence limits are appropriate from the standpoint of risk, and so on; cf. Kitcher 2001; Douglas 2000). Habermas’s argument for three basic validity claims is interesting, then, for the way it identifies an internal complexity within scientific discourse, thereby going beyond a narrow focus on representational adequacy. Habermas’s critique of social institutions, however, depends on other types of validity claims beyond these basic three. His discourse theory of democracy,
113 Habermas’s Critical Theory and Science
for example, distinguishes moral claims from ethical-political claims.6 Whereas the former appear as “ought”-statements with a universal character based on requirements of human dignity, the latter involve evaluations of what is good for a given polity, given its history and traditions. We can initially distinguish these different types through a “semantic analysis of sentence forms” (Habermas 1984, 39). But the semantic analysis finally depends on linking each validity claim with a different kind of argumentative discourse: “The meaning of the correspondingly differentiated validity claims can be explicated through specifying in each case the logical (in the sense of the logic of argumentation) conditions under which these can be established” (ibid., 39–40). That is, we can distinguish types of validity claims if we can show how each type depends on different conditions of discursive justification. To maintain his distinction between moral and ethical-political claims, for example, Habermas must show how discourses about what is morally right or wrong—about the basic principles and norms of justice—differ from discourses about a polity or group’s common good. Whereas moral discourses ultimately aim at a universal consensus of all reasonable persons who agree for the same (rather abstract) reasons, ethical-political discourses seek the agreement of citizens of a given political body or members of some group, who agree on the basis of reasons that include appeals to the group’s history and traditions. 4
Habermas’s Formal Pragmatics of Truth
According to the foregoing sketch, Habermas’s research program in formal pragmatics proceeds on the working hypotheses that (1) one can reconstruct the rational basis of communicative practices by identifying a system of validity claims such as truth, rightness, and so on, which participants presuppose in their communication, and (2) these presuppositions play a fundamental role in social integration. In regard to (1), Habermas takes the truth claim as the most important and salient for the analysis of scientific communication. That is, scientific inquiry is oriented primarily toward empirical knowledge, which in the natural sciences finds expression in warranted truth claims about nature. Although the identification of scientific inquiry with the search for truth has not gone unchallenged in the philosophy of science, Habermas is certainly on familiar ground here, and I will not question this point in what follows. For expository purposes, then, I assume that to understand scientific communication and argument, we must start with the idea that these involve speech acts
114 Chapter 4
whose acceptability depends primarily, though not solely, on the recognition of truth claims. The second hypothesis above implies that we can adequately understand Habermas’s theory of truth only if we see its sociological implications—the role that truth claims play in maintaining intelligible social interaction, or what sociologists refer to as “social order” or “orderly” social interaction. I thus start by focusing on what, for purposes of presentation, I call Habermas’s “sociological” analysis of truth. I then turn to his “philosophical” position, that is, the implications of his theory for traditional philosophical concerns. From Habermas’s reconstructive sociological perspective, the mutual imputation of accountability, or rational agency, lies at the heart of social order. Specifically, we can encounter one another as subjects, members involved in a shared practice rather than manipulable objects, only if we mutually presuppose that we could “under appropriate questioning” account for our actions (Habermas 1971/2001, 101). In other words, Habermas understands mutual accountability as a defeasible pragmatic presupposition—an imputation that participants mutually undertake but that cannot be definitively demonstrated by empirical observation. Accountability thus has the sense of a practically effective but possibly counterfactual “as if”—an idealization or “idea of reason” that has consequences for social interaction. Although we cannot guarantee our accountability, our speech and behavior in practice can display presumptive accountability in various ways, above all by displaying our capacity to give reasons for our actions and choices. When actions fail to display the marks of rational agency, others are likely to withdraw their imputation and consider the offender as irrational or unreasonable.7 Although Habermas initially associated accountability with the normative validity claim, more recently he has linked it with the “agent’s general ability to orient her action by validity claims” (2003, 95; cf. 1971/2001, 100–102). Consequently, participants display their mutual accountability and maintain an intelligible interaction through their capacity to use and respond appropriately to validity claims, including truth claims. This links the presupposition of accountability with two further ideas of reason that are especially pertinent for understanding the formal pragmatics of scientific communication: the supposition of a common objective world and the idea of objective truth. I draw here on Thomas McCarthy’s lucid treatment of these ideas, which elaborates nicely on Habermas’s approach (Habermas 2003, chap. 2).
115 Habermas’s Critical Theory and Science
McCarthy (1991, 31–32) explicitly links the presupposition of an objective world with accountability: “We are held accountable, and in turn hold our interaction partners accountable, for the transcendent objectivity of the world as invariant to discrepant reports.” Like Habermas (1984, 13–14), McCarthy draws here on Melvin Pollner’s analysis of mundane reasoning, specifically his analysis of the way this presupposition undergirds “error accounts” (see Pollner 1990). In order to maintain the world’s objectivity in the face of conflicting reports, we must account for differences by revoking one or another ceteris paribus assumption about the “community of observers” and the conditions of reliable observation. The objectivity of the world, in the sense of its intersubjective accessibility, is thus an unfalsifiable presupposition by virtue of which actors anticipate that, “all other things being equal,” competent observers should be able to reach unanimity in their factual reports (Pollner 1990, 143, 150–151). Without this presupposition, neither the problem of discrepancy nor the means used to resolve it are intelligible (ibid., 142). The idea of a common objective world depends reciprocally on the idea of truth.8 The latter is implicit in our intuitive awareness that what we consider factually true at one point, we might later recognize as false in the light of new information (McCarthy 1991, 33ff.). This awareness of fallibility points to the context-transcending force of the truth we claim in making factual assertions: truth cannot simply reduce to what is true-for-us-now, but should hold for everyone. Thus the idea of truth “keeps us from being locked into what we happen to agree on at any particular time and place”; it “opens us up to the alternative possibilities lodged in otherness and difference” (McCarthy 1991, 34). As an accountability structure, the idea of truth simply appears to draw the implications of the idea of objectivity for assertions about the world: if the unitary objective world is intersubjectively accessible, then true reports and assertions should be intersubjectively acceptable. Because of its context-transcending force, a truth claim involves a tacit commitment on the part of the speaker to be accountable beyond the particular situation in which the claim was made. In making a truth claim in a particular forum, we “implicitly assume responsibility . . . for demonstrating its rational acceptability in other relevant forums as well” (McCarthy 1994, 75). This last point—demonstrating the intersubjective acceptability of truth claims—brings us to the threshold of argumentation theory. Postponing the
116 Chapter 4
details of Habermas’s argumentation theory for the next chapter, I focus here on the idea of audience: those to whom one is accountable for truth claims, the interlocutors with whom one should be able to reach agreement. This question leads us into the relationship between truth, justifiability, and consensus. Because empirical truth claims regard a common objective world that should be accessible to all competent observers, Habermas has consistently linked such truth with the possibility of an ideal universal consensus. This initially led him to explicate the validity of truth claims in terms of a consensus theory of truth à la Peirce (see Habermas 1971/2001, chap. 5; 1973/1986; McCarthy 1978, 291–310). Although his statement of the theory was ambiguous, on one reading it suggested the counterintuitive view that a true empirical statement is true because all competent participants could accept it in an ideally rational discourse, which he seemed to regard as both necessary and sufficient for truth: “the truth condition of propositions is the potential assent of all others” (1971/2001, 89). He now grants that this view (which he describes as an “epistemic theory”) misses the core intuition that truth depends on the world, and thus is justification-transcendent. Ideal consensus is more plausibly construed as a theory of conclusive justification or ideal warrant9 than of truth (Habermas 2003, 36–42, 91, 251). McCarthy accepts the more recent view, explicating the pragmatics of scientific justification in terms of an ideal “universal audience” to which scientists tacitly hold themselves accountable: Scientists have reflexively to anticipate the scrutiny of their work by others who share the cognitive, normative, and evaluative presuppositions marking them as members of the same scientific subcommunity. If they can convince that particular audience, they have reason to expect they could convince other relevant audiences as well. That is to say, in virtue of its presumed competence, that audience can plausibly stand in for the “universal audience” of all rational beings able to judge the matter. (McCarthy 1994, 76)
Here McCarthy links the idealizing supposition of universal consensus not with truth per se but with cogent justification. Such a link, as I understand it, makes ideal consensus neither the condition of truth nor a necessary final consequence of truth, that is, the hypothetical endpoint of inquiry. On one reading of the latter ideal, when we take a hypothesis as true we presuppose that it would hold up in the long run, were scientists able to pursue inquiry as far as they fruitfully could (cf. Misak 1991).
117 Habermas’s Critical Theory and Science
But do not the ideas of objectivity and truth imply just that: that because we share a single common world, we ought eventually to be able to reconcile conflicting truth claims and reach unanimity? In fact, if one thinks only of everyday affairs (e.g., conflicting reports of traffic accidents), it is easy to overlook a potentially fragile condition built into the supposition of unanimity, namely, that the community of observers has the epistemic resources to reconcile different reports. But why presume that for all empirical truth claims, we have, or eventually could have, the resources to reach consensus? The truth of a hypothesis does not guarantee that the requisite evidential resources will in principle be available, even in the long run (e.g., if testing a hypothesis requires more energy than scientists could ever achieve, or depends on irretrievable information). So our pragmatic connection between intersubjective accessibility and acceptability is at least as much a supposition about our cognitive and linguistic resources as it is about the links joining objectivity, truth, and final consensus. If we take this point seriously, then it makes sense to connect the supposition of consensus not with truth per se but with the practice of making justifiable truth claims. Because truth claims refer to a common objective world, such claims, if true, are true for everyone. Objectivity and truth remain joined as before. But if our shared access to the objective world always depends on the available epistemic resources, then we cannot move directly from the idea of objectivity to the idea of final consensus on the truth, but only to a kind of provisional consensus on what we may justifiably take as true, given our available resources— which constitute a kind of “knowledge index” on truth claims. Here the ideal of universal consensus means that when we accept a truth claim as justified, we assume that if our collective assessment of the currently available knowledge, arguments, and methods were as reasonable as it could get and included everyone competent to speak to the matter, the claim would still stand. The idealization thus consists in the thought of justifying a truth claim on the basis of a maximally rigorous scrutiny of the available information and arguments, such that no relevant considerations were excluded or underestimated, and no mistakes in reasoning occurred. We can formulate this idea as a pragmatic presupposition of attempting to justify truth claims: (JTC) If we reasonably consider our arguments to justify our taking “p” to be true, then we must presume that our justification would prove convincing in a rational discourse that was maximally inclusive and rigorous, given current methods and knowledge.
118 Chapter 4
This thesis neither defines truth as consensus nor projects an ideal end of inquiry in which knowledge is complete, but rather posits an internal pragmatic relation—that is, a relation internal to the practice of justifying truth claims—between justified truth-claiming and the supposition of ideal provisional consensus. According to (JTC), in taking a hypothesis as true on the basis of a justifying set of arguments, we assume that competent practitioners would accept our arguments and claim, given the available level of knowledge, methods of inquiry, and shared discursive resources. This thesis simply extends the sociology of error accounts: if we share an objective world that is intersubjectively accessible through the commonly available means of inquiry and linguistic resources, and if we believe that those means and resources justify us in claiming that p, then other competent researchers, employing the same means, should also agree that p. By linking truth claims with a supposition of provisional consensus as a necessary condition of taking such claims as justified, (JTC) also commits us to the fallibilist attitude that McCarthy associates with the idea of truth. That is, the thesis implies (by modus tollens) that when the arguments we use to justify p fail to hold up in sufficiently rigorous and reasonable discourse, we may no longer consider ourselves justified—at least not in the sense of public argumentative justification—in taking p as true. Notice also an intermediate possibility that (JTC) suggests: that our arguments might be strong enough to hold up in discourse, yet fail to defeat a competing position. The arguments thus provide reasonable but not conclusive grounds for accepting p. In that case, p stands as a hypothesis meriting further investigation. Finally, as a practical accountability structure, (JTC) also commits us to a dynamic view of truth claims and inquiry, insofar as the “available resources” proviso is a moving target. As scientists develop new methods of inquiry and acquire new information, the knowledge index in (JTC) shifts, and what we took as true and justified at one point in time is once again open to question. At that point, the arguments that we once took to justify our belief that p must be reassessed. This need for reassessment distinguishes (JTC) from models that project a final state of inquiry: because (JTC) has a knowledgeindex, when we take p as true on the basis of a set of arguments A1, we only presuppose its justifiability before an ideal audience using the available information and methods M1. We do not commit ourselves to the idea that A1 must justify p relative to new information and methods M2. At the same time, because truth itself is an absolute (justification-transcendent) idea, we cannot
119 Habermas’s Critical Theory and Science
be indifferent to the fate of our truth claims when new methods emerge: the latter call for a reassessment of the claim. (JTC) does not sanction a relativistic view that would define truth in relation to available methods and information. So far I have analyzed the idea of truth as it functions sociologically, as an accountability structure. Thus (JTC) summarizes a normative social analysis of argumentative practices: it states what it means in practice for us justifiably to take p as true, not what it means for p to be true.10 I turn now to Habermas’s philosophical account. Given the sociological analysis, we should not be surprised that he endorses a pragmatic realism at the philosophical level. On this view, truth is something that a speaker claims for a statement p by uttering an assertoric speech act whose propositional content refers to the objective world, for example, “[I assert to you that] smoking increases the risk of lung disease,” where the bracketed illocutionary component is normally left unspoken. Thus Habermas identifies the statement or proposition as the primary truth bearer, contextualized by virtue of the particular speech-act utterance in which it is embedded (Habermas 1971/2001, 86; 2003, 38). This formulation points to some tangled issues of truth and language that need concern us only insofar as they illuminate the implications of Habermas’s approach for argumentative cogency. To grasp these implications we must focus not so much on the semantics of truth as on the consequences he draws from the pragmatics of truth claiming. Habermas describes his current view as “pragmatic epistemological realism” (Habermas 2003, 7; see also 1998b, chap. 8). What makes his theory realist is his insistence that the truth maker is the objective world, which exists independently of us and is the same for all. If a statement for which we claim truth is indeed true, it is so because it accurately refers to independently existing objects—albeit objects about which we can state facts only under descriptions that depend on our linguistic resources (Habermas 2003, 16, 30–36). The inescapability of language dictates the pragmatic epistemological character of his realism. Specifically, Habermas eschews the attempt to explicate the relationship between statement and world metaphysically, as a relation of correspondence. Following a common line of criticism, he regards correspondence theories as uninformative because they cannot explicate the correspondence relationship except by reiterating the propositional content (typically with the addition of emphatic qualifiers) in the explanation of that with which true statements correspond.11 A pragmatic epistemological approach, by contrast,
120 Chapter 4
gives us an illuminating explication of truth—albeit truth as it functions in action and learning. Like correspondence theorists, the pragmatist accepts the realist intuition that in making truth claims, speakers want to refer to independently existing objects or to describe states of affairs in a common objective world. The two approaches differ in the mode of explicating this intuition. On Habermas’s interpretation, correspondence realists attempt to explicate the realist intuition by projecting a standpoint beyond language and experience (2003, 216ff.). By contrast, Habermas’s pragmatic approach relies on terms that are in some sense empirically and descriptively accessible, hence open to sociological reconstruction. Thus, the realist element in the model—the idea of an independently existing, shared physical world—is explicated in terms of its pragmatic function as a necessary presupposition of social and technological practices. For example, like Pollner above, one describes the practices evident in error accounts as a kind of normative demand on interaction, a way we hold one another mutually accountable for what we take as true about the objective world. This realist presupposition operates both at the level of practical engagement with reality (or in science, the level of experimental practices) and at the level of discourse (in science, the level at which experimental results are worked up into arguments and presented to others).12 At the level of everyday practice, actors take various propositional contents as unproblematically true, acting confidently on the basis of their well-corroborated beliefs about objects in the world. In experimental practice, scientists take for granted a vast range of empirical assumptions about the object domain, their reagents, materials, instruments, and so on. To be sure, some of these presuppositions are carefully cross-checked in each experiment, but many are too obvious and well corroborated to bear checking. If something goes wrong in a routine chemical analysis, for example, I might wonder if a given reagent has lost its strength or become contaminated; I am unlikely to doubt that the basic chemistry has suddenly and arbitrarily changed for my particular experiment. At the level of practice, that is, one assumes a world of more or less stable objects that exist independently of language and are the same for all observers. The unproblematic presumption of truth in everyday behavior and experimental practice is intertwined with truth claims that have lost their unproblematic status, either because of some practical difficulty in action or experiment, or because an interlocutor has pointed out a problem (e.g., a discrepancy with
121 Habermas’s Critical Theory and Science
results from another lab). Such difficulties open up issues for critical discussion and argumentation. At the local level, for example, an upset of experimental expectations will lead the team of researchers into a critical discussion in which they try to identify the source of surprise. Conferences at which different laboratories present and criticize research represent broader discourses, the outcomes of which—the ideas gained, suggestions considered promising, and so on—then enter back into further experimentation. Here too we see the assumption of a common objective world, this time operating as the assumption that different laboratories should reach similar results insofar as they are making arguments about the same kinds of objects; within the team, the idea of an objective world appears in the assumption that true beliefs about the world should support predictable experimental outcomes, and that failures in such predictions call for further inquiry and explanation: one must revisit the statements one took as true about one’s experimental design. The foregoing analysis implies a model of inquiry in which the cogency of arguments rests partly on an internal relation between discourse (argumentation) on the one side and experience and action on the other, which in the sciences include observation and laboratory interventions. As Habermas summarizes the general model: In the spatial dimension, knowledge is the result of working through experiences of frustration by coping intelligently with a risk-filled environment. In the social dimension, it is the result of justifying one’s ways of solving problems against the objections of other participants in argumentation. And in the temporal dimension, it is the result of learning processes fed by the revision of one’s own mistakes. . . . the passive moment of experiencing practical failure or success is intertwined with the constructive moment of projecting, interpreting, and justifying. (Habermas 2003, 26; translation slightly altered)
Applying this to the sciences, we might say that the action of experimentation “in the spatial dimension” is intertwined with argumentation in the “social dimension”; together these moments generate publicly cogent empirical arguments that move inquiry along as a learning process in the “temporal dimension.” So when Habermas claims that argumentation “remains the only available medium of ascertaining truth since truth claims that have been problematized cannot be tested in any other way” (ibid., 38), we should keep in mind that scientific argumentation is deeply intertwined with practical experimental engagement in laboratory and fieldwork. But if scientific discourse is internally
122 Chapter 4
related to experimentation as a kind of experience and action, then we can also say that experiments test problematic truth claims and thereby ascertain truth—albeit only experiments that have been worked up into cogent empirical arguments.13 Thus, Habermas’s prioritization of discursive justification over experience in determining truth must be understood, in the sciences, as the claim that experiments do not interpret themselves, but must be cast as arguments. We can then say that such arguments, as internally related to laboratory “experience,” are the only means we have for determining what is probably true. 5
Vision and Critique in TCA
As I mentioned in section 1 above, Habermas’s TCA supports a specific vision of social emancipation, which sharply contrasts with the technocratic vision of twentieth-century industrial societies ruled by experts and bureaucratic elites. Like his Frankfurt School forebears, Habermas criticized technocracy, even before he developed TCA. Specifically, he attacked the impulse to reduce science-intensive policy issues to technical questions that experts alone had the competence to resolve: the technocratic vision “eliminates practical questions and therewith precludes discussion about the adoption of standards; the latter could emerge only from a democratic decisionmaking process” (Habermas 1970, 103). If empirical claims alone allow for reasonable discourse, then the choice of research agenda and new technologies should be settled not by public debate, but only by functional necessities or the strategic competition for political control. Habermas’s turn to formal pragmatics deepens and enriches his earlier critique of technocracy. TCA, that is, grounds the reasonable character of practical discourse in a universalistic analysis of language use as oriented not simply toward factual truth but also to other validity claims about the normative rightness of policy and law. Because science-related lawmaking and administration affect broader publics, not only their empirical adequacy but also their fairness and suitability for the common good—issues that involve normative and evaluative validity claims—must be justified before the appropriate audiences. Although Habermas’s account of truth implies that scientific experts play an indispensable role in such justifications, the multidimensional validity basis of science-related questions calls for broader public deliberation.
123 Habermas’s Critical Theory and Science
Thus Habermas’s discourse theory allows him to criticize technocracy by embedding technoscientific questions and discourse in a broader socialpolitical framework, specifically in a deliberative-democratic vision of an emancipated “rational society.” As we saw with his analysis of the audience and consensual ideals connected to truth, so also Habermas (1990, chap. 3; 1996, chap. 4) links normative and evaluative claims to ideals of practical discourse. I cannot elaborate these analyses here, except to point out how the type of validity claim at issue in a given discourse—the argumentative goal—specifies the audience as well as the depth and extent of consensus ideally required for a complete justification. Insofar as political issues involve moral questions of justice, discursive justification anticipates a universal consensus of all reasonable persons on both the justice claim and its supporting reasons. By contrast, evaluative questions about whether something is for the common good of a polity—what Habermas calls “ethical-political” questions—require the agreement of all citizens on the basis of reasons that appeal to shared traditions and values. In pluralistic societies ethical-political discourse would presumably have to settle in many cases for an overlapping consensus in which supporting reasons differed across different subgroups.14 Because such moral and ethical-political questions intertwine with empirical issues in political debate, such debate is typologically complex and often requires inclusive modes of public deliberation if outcomes are to count as publicly justified, and thus legitimate (Habermas 1996, chap. 4; Rehg and Bohman 2002). Habermas’s discourse theory thus situates scientific argumentation within a capacious understanding of the rational basis of social integration in general and reasonable political discourse in particular. Such a broad framework, informed by social theory, bodes well for the possibility of bridging Kuhn’s Gap, an issue I take up in subsequent chapters. To close this chapter I point out how the very complexity of the account creates certain challenges that increase the burden of argument for the critical theorist. First, to employ discourse theory for the critical assessment of real discourses, one must begin by disentangling the types of validity claims at issue and then assess how each of these is criticized and defended. But given that participants themselves often disagree or are unclear about their argumentative goals, the first step already runs into difficulties. One must, apparently, take a normative stance at the outset, namely a position regarding the goals (validity claims) that ought to be primarily at stake in a given debate. This in
124 Chapter 4
turn involves a normative conception of the institutional or social settings in which the issue arises. For example, the debates in the United States over the teaching of intelligent design theory in public schools partly turn on contested normative conceptions of scientific discourse and educational institutions. When the Kansas Board of Education dropped the condition of “natural explanation” in its redefinition of “science,” it simultaneously took a position on the kind of moves allowable in scientific arguments (e.g., in what counts as an inference to best explanation) and on the goals of biology education in public secondary education.15 Adopting a view of the aims of different institutions and their discourses means taking a position on modernity itself—the way in which modern societies have developed complexity through institutional differentiation. The normative character of this differentiation is displayed, for example, in constitutional principles like the separation of church and state (which separates religious validity claims from political discourse) and in the criticisms scientists leveled against Christian literalists who claimed to be doing “creation science” (a criticism that excludes religious claims from scientific discourse). In taking the search for empirical truth about a “disenchanted” nature as the salient validity claim at stake in the natural sciences, Habermas adopts a normative position on scientific discourse itself (i.e., he distinguishes it from legal-political discourse and religious discourse) and on the institutions that specialize in such discourse. TCA, in other words, supports a specific vision of societal rationalization and institutional organization, which contrasts not only with the technocratic vision but also with those that reject the modern “art of separation.” Habermas’s ambitious vision places a heavy burden of argument on critical theorists. Still further complications await us, however. Because actual debates so often take place in institutional settings, discourse is subject to constraints of time, place, and material resources. On the one hand, Habermas defines the rationality of real discourses of truth and justice in relation to the idealized justifiability of claims before universal audiences. On the other hand, the institutional constraints on real discourses mean that participants can at best “approximate” the idealized justification they presumably strive for and presuppose as an ideal limit of argumentation. This tension between the ideal and the real creates the problem of contextualization that motivates my proposal in Part III. For now, notice how institutionalization requires the critical theorist to adopt a normative stance on where to draw the line—that is, on the
125 Habermas’s Critical Theory and Science
normatively appropriate ways of limiting participation, apportioning resources, cutting off further discussion, reaching decisions in the face of dissensus, and so on. The problem of appropriate institutionalization is one aspect of how TCAbased critique understands the relationship between communicative and strategic action. Institutional procedures generally depend on the use of administrative power, that is, the use of institutionalized authority (or rights) to make decisions that affect the flow of discourse (e.g., in science, editors decide whether to accept or reject a manuscript, officials decide to fund grant proposals or reject them, and so on). Habermas’s association of such power with strategic action is probably a bit misleading in this context (see McCarthy 1991, chap. 6), but administrative decisions are also not communicative or discursive in the full sense, inasmuch as the decision maker’s institutional right to decide does not depend on the supposition that the decision maker could eventually convince all parties of the substantive normative rightness of the decision. Rather, the assignment of such decision-making authority is rational from the standpoint of efficiency: there is a need to get things done, even when discourses remain inconclusive. A further aspect of the relationship between communicative and strategic action arises from the mixed motives typically at work even in participants of good will. In scientific discourse, participants are usually pursuing not only argumentative goals such as truth but also personal aims, some of which are narrowly self-interested, others perhaps more altruistic. Even in searching for truth, scientists must commit to a particular line of inquiry that tends to bias their assessment of objections and alternatives. In political institutions, selfserving aims are especially intense, to the point where manipulation and deceit occur with a depressing ubiquity, so that bias and self-interest practically become the norm. Any critical analysis that aims to avoid utopianism must take this reality into account. But doing so requires the critical theorist to make further assumptions about human behavior: what sorts of self-interested behavior are likely, how much informal self-policing can one expect, what sorts of oversight mechanisms and penalties, checks and balances, are necessary? In sum, although TCA supports a particular vision of social emancipation or the “good society” (Cooke 2006), the above observations suggest that the employment of TCA for purposes of critical assessment also depends on that vision. Such a “vision” involves more than the vague affirmation of a
126 Chapter 4
constitutional-democratic ideal of social emancipation, in which justice, freedom, and equality are realized for all. As we have seen, Habermas’s vision is distinguished by the following:
• traditions of thought that serve as his primary inspirations (esp. Kant, pragmatism) and lead him to a universalistic conception of communicative reason; • a reading of modern social complexity and institutional organization as rationally progressive (hence a view about occidental history); • assumptions regarding what actors individually and collectively are capable of, good and bad (i.e., behaviors that are typically self-interested or strategic but that can also cooperate in reason-based modes of social solidarity);
• an analysis of the institutional means suitable for pursuing the vision, given the above assumptions about behavior (i.e., deliberative democracy rather than radical economic transformation à la Marx or democratization of the economy). How one actually deploys TCA, with its system of validity claims and differentiated argumentation theory, depends partly on one’s positions on the above points. It follows that the cogency of critical assessments (which after all are also arguments) partly rests on the cogency and coherence of this larger vision. The burden of proof on Habermas’s approach to critique thus appears enormous, insofar as particular critical assessments rest on an ambitious vision of modernity and sociopolitical organization in general. At the same time, to pursue this vision Habermas considers it crucial that critical theorists have at their disposal a universalistic, context-transcendent conception of reason, a conception that likewise incurs a significant burden of proof in philosophical and sociological debate (cf. Cooke 2006, chap. 3). The validity claims of truth and morality are especially important in this regard, given their orientation to the idea of universal consensus based on argumentative practices that in turn involve universal structures (more on which in the next chapter). As we saw in his critique of Toulmin, Habermas fears that without such a transcendental critical framework, critical assessment loses its leverage point for evaluating local and institutionally established practices. In other words, without TCA and its attendant discourse theory, critique bogs down in the relativity of par-
127 Habermas’s Critical Theory and Science
ticular contexts, each with its own standard of justification, understanding of truth, and so on. The danger of a relativistic contextualism, in other words, constitutes the chief theoretical threat to Habermas’s project, the virus TCA is designed to ward off. Indeed, for Habermas (1998b, 355) the threat of contextualism is “built into” the linguistic-pragmatic turn just as epistemological skepticism was built into the early modern turn to the subject. In Part III, I address that concern more fully.
5 Habermas’s Theory of Argumentation as an Integrated Model of Cogency
I now take up Habermas’s argumentation theory in detail. The issues connected with Kuhn’s Gap provide the specific challenges that govern my engagement with Habermas. As I described that gap in Part I, it raises a number of questions: how to mediate between logic (good reasons) and social psychology (effective persuasion); what status the science community and its institutions should play in a conception of cogency; and how one connects descriptive and prescriptive analyses and fosters cooperation between the normative and empirical branches of science studies. As one might anticipate from chapter 4, Habermas’s commitment to a formal-pragmatic theory of meaning locates his argumentation theory in specific ways vis-à-vis the positions surveyed in Part I. On the one hand, he agrees with many philosophers of science when he grounds scientific inquiry in a realist idea of truth with universalistic normative implications for the conduct of inquiry; on the other hand, his interest in the social function of truth, along with his multidimensional understanding of validity, situates inquiry in sociopolitical contexts of various sorts. Like ethnomethodologists of science, he regards the participant perspective as the methodologically necessary standpoint for understanding the rationality of scientific practices. Although he rejects relativistic approaches, he agrees with SSK theorists who seek to situate scientific practices in their broader social and institutional contexts. In this chapter I present Habermas’s argumentation theory as a comprehensive framework designed to address the problems generated by Kuhn’s Gap in a manner consistent with the above commitments. In developing Habermas’s framework, however, I also have a basic philosophical question in view, one that bears on the conceptual side of Kuhn’s
130 Chapter 5
Gap: whether we ought to regard the cogency of scientific arguments as referring not only to features of argument products but also to the dialectical procedures and social process of argumentation. Definitions such as Hempel’s and Achinstein’s suggest a negative reply: the conditions that qualify an argument as cogent lie solely in the intrinsic and impersonal features of the product, the argument as a package of reasons that support the conclusion, given justifiable background assumptions. The three rhetorics of science in chapter 3 answer in the affirmative, as each regards the social process as constitutive of cogency. But none of these rhetorical approaches fully succeeds in bridging Kuhn’s Gap, and each takes an approach that Habermas must regard as problematic. Pera understands cogency in terms of a dialectically responsible process of communal discourse; Prelli fills out this model with further rhetorical substance that links good reasons with audience psychology. The sociality of science, however, remains underdeveloped, focused entirely on the communal culture of science. Thus neither model situates argumentation in its institutional and broader social contexts. Moreover, the communitarian approach, at least for Pera, grounds the social conception of cogency in a rejection of extradisciplinary standards of cogent argument. As I explained in chapter 4, Habermas rejects such models as overly conventionalist and vulnerable to a relativist interpretation. Latour’s conception of cogency, by contrast, incorporates the social-institutional aspects of science, understood as human–nonhuman networks both near and far. However, his antinormative strategic approach lacks the prescriptive resources that Habermas’s critical project requires. Although Habermas can agree that scientific argumentation involves strategic dimensions, these must play a subordinate role when it comes to the normative assessment of cogency. I read Habermas as also answering the fundamental question in the affirmative: he wants to develop a multiperspectival model in which arguments count as cogent in virtue of both process and product merits. As he points out, “every theory of argumentation faces the task of specifying general properties of cogent argument.” In carrying out this task, “at no single one of these analytic levels [i.e., logical, dialectical, and rhetorical] can the very idea intrinsic to argumentative speech be adequately developed” (Habermas 1984, 26). Although his talk of the “very idea” might suggest that he wants something like a definition in Achinstein’s sense of necessary and sufficient conditions, the methodology I described in chapter 4 calls for a more pragmatic interpreta-
131 Habermas’s Theory of Argumentation
tion. That is, his argumentation theory attempts to articulate the normative presuppositions of argumentative practices whose participants strive to reach consensus on justified truth claims. Habermas’s “idea” of cogency thus refers to a shared but open-textured idealization that plays a constitutive role in such practices: as shared, it makes cooperative processes of argumentation possible; as open-textured, it requires participants to engage in ongoing, and contestable, efforts at critical interpretation and clarification of the point, values, and standards that ought to inform argumentation as an essentially evaluative practice. Habermas’s argumentation theory is one such interpretation: an attempt to articulate what the rational pursuit of cogent arguments typically requires, namely that good arguments must satisfy a complex set of evaluative standards—logical, dialectical, and rhetorical. If the above reading is on target, then Habermas’s argumentation theory brings in dialectic and rhetoric not simply as aspects of the collective assessment of cogency—which is hardly controversial—but as constitutive dimensions of cogency itself. To bridge Kuhn’s Gap, however, Habermas’s model of cogency must also incorporate the social-institutional perspective as coconstitutive. If we can defend such a view as a plausible account of cogency, then Habermas’s argumentation theory presents a promising response to Kuhn’s Gap. In developing Habermas’s argumentation theory as a comprehensive framework, then, I have this specific aim in view: to see how his theory plausibly integrates, within a normative theory of cogency, (a) a logical perspective focused on argument content, (b) dialectical and rhetorical perspectives that analyze the substantive normative commitments, ethos, and psychology of science as a discourse community, and (c) social-institutional perspectives that acknowledge the strategic aspects of argumentation without negating the possibility of normatively good reasons. I begin by noticing how his model involves a set of uncontroversial internal dependency relations that integrate the logical and dialectical perspectives in the assessment of cogent arguments; here we see that what Habermas calls “rhetoric” is better understood as a further dimension of the dialectical perspective that complements Pera’s dialectics (sec. 1). These first two perspectives can be integrated with a substantive rhetorical perspective that looks to some social-psychological aspects of cogency not dealt with by Prelli (sec. 2). I then tackle the core question, arguing that Habermas’s pragmatic commitments undermine the analytic distinction between cogency
132 Chapter 5
as such and its collective assessment (sec. 3). This leaves us with the question of how well Habermas’s model incorporates a social-institutional perspective; here an analysis of credit mechanisms reveals deeper tensions between the real and the ideal in his model (sec. 4). 1 Habermas’s Argumentation Theory: Perspectives as Internally Related As we saw above, Habermas is concerned with the question of cogency, the strength of arguments: “How can problematic validity claims be supported by good reasons? How can reasons be criticized in turn? What makes some arguments, and thus some reasons, which are related to validity claims in a certain way, stronger or weaker than other arguments?” (Habermas 1984, 24) His methodology requires an answer that articulates the normative presuppositions of participants. Thus he construes the different perspectives on argumentation as “levels of presuppositions of argumentation,” which he aligns with the traditions of logic, dialectic, and rhetoric (Habermas 1990, 87; cf. Wenzel 1990). Habermas himself does not have much to say about the logical and dialectical perspectives, but his approach appears to be consistent with that of mainstream informal logic, and I will supplement his remarks accordingly. His understanding of the rhetorical perspective, however, bears closer scrutiny. 1.1
Logic and Dialectic
At the logical level, participants are concerned with the construction of “cogent arguments that are convincing in virtue of their intrinsic properties and with which validity claims can be redeemed or rejected” (Habermas 1984, 25). Like Wenzel, Habermas associates logic with the evaluation of arguments as products of argumentative practices. As examples of the normative standards operative at this level, Habermas mentions the law of noncontradiction and consistency in the application of predicates (1990, 87). Like Toulmin and other informal logicians, however, he regards argumentation in the sciences as “substantial” (ampliative) in a way that resists complete formalization. So we should take the phrase “intrinsic properties” as having a broader sense than the formal-logical merits on which Hempel focused. We might think of the logical perspective as focused on something like the “content merits” of the argument, which presumably would be defined by such standards as clarity, relevance,
133 Habermas’s Theory of Argumentation
independent plausibility of reasons, and adequacy of support. Here “independent plausibility of reasons” (i.e., acceptability, empirical adequacy) functions as the pragmatic substitute for truth. Once we understand the logical perspective to include the evaluation of informal arguments, that perspective depends on the kinds of critical questioning associated with dialectic. Normally we cannot assess the internal strength of an informal argument unless we subject it to a critical testing that takes us beyond the syntactic and semantic links between premises and conclusion. The reason for this dialectical requirement lies in the defeasible character of ampliative arguments: new information can reverse a previously probable conclusion. Consequently, assessment requires us to raise such questions as: does the argument overlook relevant information? Are its premises sufficiently precise? Have likely sources of error been excluded? Even deductive arguments are open to question insofar as one can challenge the adequacy of the vocabulary employed in the supporting reasons. Arguments are dialectically robust or “responsive,” we might say, only if they can incorporate (or could incorporate via suitable development) responses to such challenges (cf. Goldman 1994, 1999, chap. 5). We can take this point a step further by noticing that scientific arguments are, to a significant degree, constructed dialectically. For example, we can view experimental and argumentative procedures as cooperative ways of constructing products precisely by testing their evidential premises. Many laboratory procedures are designed to rule out artifacts and sources of error—a point that Deborah Mayo (1996) takes as the cornerstone for a theory of experimental knowledge. More generally, insofar as argument construction involves processes of assessment, most of what I say here about the internal dependency relations linking evaluative perspectives applies to the construction of arguments as well. Thus the logical assessment (and construction) of arguments is internally linked with the dialectical level, where arguers engage in a “ritualized competition for the better arguments” (Habermas 1984, 26). At this level, participants cooperate in a competition to produce the more cogent argument. Habermas associates this level with rule-guided “procedures” of argumentation. Such procedures define what some theorists call the “dialectical obligations” of discussants—that one should only address the issue at hand, should respond to relevant challenges, meet the specified burden of proof, and so on (cf. van Eemeren, Grootendorst, and Snoeck Henkemans 2002). By way of example,
134 Chapter 5
Habermas (1984, 25; 1990, 87ff.) mentions rules of relevance and rules for the introduction of claims, as well as the broad requirement that speakers maintain only what they believe. The details here are open to dispute. The belief requirement seems too strong for inquiries in which participants understand themselves as entertaining possible hypotheses, which they do not actually believe and which they do not present as beliefs (Meiland 1989). At the dialectical level, relevance presumably pertains to the relevance of challenges and rebuttals in view of dialogical aims, whereas at the logical level it concerns the substantive connection between reasons and conclusion.1 Although many of Pera’s dialectical factors would be compatible with Habermas’s understanding of dialectic, some caution is required. As I noted in chapter 4, Habermas distinguishes institutionally specific procedural rules from the dialectical presuppositions connected with the three central types of validity claims. Institutional procedures and rules, on Habermas’s model, belong to the social-institutional perspective, and are open to critical scrutiny from the dialectical perspective—the latter, in other words, provides the universal standards for identifying merits and demerits of institutionalized procedural designs (e.g., article review procedures, rules of discussion at conferences, etc.). Many of Pera’s rules (e.g., his rules of adjudication) are sufficiently broad that they would fall under the umbrella of Habermas’s dialectical level. Dialectic points in turn toward an intersubjective understanding of argument assessment. Evaluation of dialectical robustness has a social character insofar as the movement of challenge and response is carried forward by participants in the roles of “proponent” and “opponent.” This social, or intersubjective, aspect of dialectical evaluation is not simply an optional addition. As various argumentation theorists have maintained, dialectical argumentation is fundamentally social, even when carried on as a dialogue with oneself (Crosswhite 1996, chap. 2; Apel 1980). To be sure, for very simple modes of reasoning and argument, assessment is largely monological, grounded in each individual’s competence to recapitulate the reasoning that led the arguer to a particular conclusion. In such cases, the requisite skills are common yet individually held competences that allow individuals to construct and validate the specific arguments leading to the conclusion. The complex problems and forms of argumentation in science, however, essentially require cooperative or dialogical modes of assessment. The collection of relevant data and certification of the vast supporting evidence and auxiliary assumptions require a group effort and presuppose, moreover, the competence of past scientists. Thus we can say
135 Habermas’s Theory of Argumentation
that intersubjective acceptability in science is grounded in a cooperative group process. As an individual, I cannot in principle guarantee or validate a solution by myself alone. It follows that critical dialectical testing of a claim essentially refers to an audience of competent participants engaged in argumentation. The foregoing implies a second internal relation, which arises between the dialectical testing of argument products and social ideals that govern the procedures and processes by which such testing occurs. Whereas the first internal relation, between logical assessment and dialectical testing, linked the product with procedures regulating the rational flow of speech acts, the second links those procedures with process ideals that structure the power relationships among participants. Habermas associates such ideals with the rhetorical perspective. 1.2
The Social Conditions of Dialectical Testing
At what he calls the “rhetorical level,” Habermas construes argumentation as a process of communication in which arguers seek to gain the assent of an audience according to the standard of a universal audience. At this level, “[p]articipants in argumentation have to presuppose in general that the structure of their communication, by virtue of features that can be described in purely formal terms, excludes all force—whether it arises from within the process of reaching understanding itself or influences it from the outside— except the force of the better argument (and thus that it excludes, on their part, all motives except that of a cooperative search for the truth)” (Habermas 1984, 25). These presuppositions are “purely formal” in the pragmatic sense introduced in chapter 4, which refers to invariant structures of reasonable argumentation in general. The most important include not only the exclusion of coercion or force, but also the openness of the argumentative process (that it includes everyone able to make a relevant contribution), equality of participation (each participant has equal voice, an equal opportunity to make and criticize arguments), and nondeceptiveness (in scientific argumentation, a genuine commitment to “a sincere and unconstrained weighing of the arguments”) (Habermas 2003, 106–107). Given the centrality of these process standards in Habermas’s argumentation theory—particularly their role in his antirelativist stance—some further comments are in order. Habermas initially linked these presuppositions with the “anticipation” of an “ideal speech situation,” but he now considers that term misleading
136 Chapter 5
inasmuch as it suggests an actual state of affairs or institutional design we could empirically realize, or at least approximate (Habermas 1996, 322–323; 1971/2001, 97–100). Although he sometimes speaks of sufficiently approximating these conditions,2 one cannot positively measure or observe, but can at most presume, such sufficiency. Rather, like the ideas of truth and universal consensus with which they go hand in hand, process ideals function as “idealizations” that are at once counterfactual and practically efficacious: they cannot be empirically demonstrated as actually realized in argumentation but they nonetheless structure the mutual accountability of partners in the discursive search for truth. Process idealizations show their pragmatic effect in the doubts that evident violations awaken about the quality of a discursive outcome: “Perceived inconsistencies that provoke doubts about the genuineness of an argumentative exchange do not arise until obviously relevant participants are excluded, relevant contributions suppressed, and yes/no stances are manipulated or conditioned by other kinds of influences” (Habermas 2003, 108). Process idealizations thus function negatively, as critical standards that drive the “intrinsic dynamic of argumentation” toward ever-wider circles of (scientific) accountability. Insofar as the idea of truth implies that factual assertions involve an “absolute claim to validity”—a claim to represent a shared objective world accessible to all reasonable observers—truth claims should in principle “be justifiable in ever wider forums, before an ever more competent and larger audience, against ever new objections” (ibid., 108–109). Process ideals thus spell out the discursive conditions presupposed by (JTC), that is, the conditions that must hold if one is to link justified truth claims with a supposition of universal (i.e., ever widening) consensus. In this chapter I am more concerned with the relation between such process ideals and cogency. In a nutshell, the ideals articulate those social conditions that would foster a maximally open, reasonable, and thorough critical testing of arguments. Habermas’s ideal process standards thus have a dialectical function, and so we might wonder if their alignment with rhetoric is apt. Moreover, given the highly contextual character of most rhetorical criticism, Habermas’s understanding of rhetoric in terms of “purely formal” pragmatic presuppositions seems incomplete at best, particularly when we compare it with the analyses surveyed in chapter 3. His idealized formal approach to rhetoric does have some plausibility, to be sure. If we find ourselves jointly persuaded by some result and regard that outcome as epistemically rational, then we assume our assent has not merely been a result of a shared prejudice that would have
137 Habermas’s Theory of Argumentation
been exposed if the discussion were more open to alternative viewpoints; we assume further that the result is not merely an artifact of institutional pressures. Nonetheless, Habermas’s association of rhetoric with Perelman and Olbrechts-Tyteca’s ideal of the universal audience could, on one interpretation of that ideal, be taken as aligning communication with logic. In fact, Perelman and Olbrechts-Tyteca use the phrase “ideal audience” in a number of ways (Crosswhite 1996, 146–148). On a more plausible rhetorical interpretation, the universal audience is a “distillation of the concrete audience, comprised of the common features imagined by the arguer” (Tindale 1999, 90). Thus “the universal audience is not a model of ideal competence introduced into the argumentative situation from the outside. It is developed out of the particular audience and so is essentially connected to it” (ibid., 117). This reading introduces a degree of specificity, albeit without sacrificing critical standards to which the actual audience is beholden. Even if we read Habermas’s rhetorical perspective and its connection with the universal audience in these more concrete terms, his understanding of rhetoric remains very much a logos-centered model, or more precisely, a dialectically oriented rhetoric. To that extent, Habermas is closer to Pera than to Prelli and Latour: his “rhetorical level” primarily sets norms for the social process of argumentation with a view toward fostering a dialectically adequate testing of competing claims. In effect we have a set of dialectical process norms, cast as an ideal procedure. Moreover, as counterfactual, the norms are not only abstract but primarily negative: they ask us to test for possible blind spots—excluded persons who may have contributed an important argument, prejudices that undermine reasonable assessment, and so on. Thus Habermas’s distinction between dialectic and rhetoric boils down to a difference between two dimensions of dialectical testing. What Habermas labels as the dialectical level (and associates with “procedure” as opposed to “process”) appears to be largely concerned with rules governing the flow of discourse as a sequence of speech acts (assertion and question, challenge and rebuttal, etc.)—what Pera refers to as dialectical “procedural factors.” Habermas’s “rhetorical” level, which he associates with the “process” of arguing, regulates the relative positions, or relationships, of the participants, namely by calling for symmetry in the discursive roles open to one and all. These process standards complement Pera’s dialectics by setting forth the ideal social conditions that foster procedurally responsible, thus dialectically robust,
138 Chapter 5
argumentation. Thus the two levels that Habermas distinguishes coincide insofar as both set down rules for organizing the process of critically testing arguments. They differ mainly in the object of the rules—statements versus participants—but in fact a critical discussion requires both types of rules.3 In what follows, then, I designate Habermas’s dialectical perspective as concerned with the ideal procedures and standards that regulate both the flow and social organization of discussion. If we reposition Habermas’s “rhetorical level” as a further dimension of the dialectical perspective, then the space opens up in his theory for a more substantive rhetorical perspective. In the next section I suggest how one might develop that perspective. Rather than follow Prelli into stasis theory, however, I look to Aristotle’s analysis of ethos and pathos. That analysis, I think, provides a better opening into the social-psychological substance of argumentation. 2
Contextualizing Habermas’s Rhetorical Perspective
Habermas’s own model provides a door—one through which he himself does not pass—into a more substantive and contextualist rhetorical perspective. According to his model of cogent argumentation, arguments do not normally compel assent in the manner of logical deduction but only make assent to a claim rationally possible or reasonable. This suggests that in some cases, particularly more controversial cases, both assent and dissent may be reasonable options. A situation in which multiple reasonable options are open presupposes that at least two conditions are met: (a) participants are not in exactly identical cognitive situations, given their (more or less) different backgrounds, expertise, tacit assumptions, and training; and (b) inquiry and critical discussion have not eliminated either of two (or more) positions as untenable. Flexibility in options increases in debates in which the question of research priorities and pursuitworthiness of hypotheses are at stake; here a single individual can acknowledge more than one reasonable course of investigation. This point undergirds Solomon’s (2001) pluralist recommendations on the distribution of research agenda. In such situations, how is it—beyond the two conditions above—that some participants are rationally motivated to accept, or further pursue, a hypothesis H and others to reject H, given that everyone has followed the discussion and heard the same arguments? Although both options appear tenable, research agenda and policy choices might require some commitment to one option over
139 Habermas’s Theory of Argumentation
the other. Moreover, participants in argumentation often feel impelled to adopt one, but not the other, of two options, though both are reasonable, given the evidence. As we have seen, Kuhn provides one approach to this situation, as generated by the vague character of the epistemic and aesthetic values that shape scientific reasoning. Solomon (2001) provides an analysis in terms of the diverse “decision vectors,” some empirical but many not, that lead scientists to pursue one line of research rather than another, more or less equally reasonable, line. In distinction to these two approaches, Habermas’s overall commitments fit better, I think, with the idea of judgment: in cognitively open-ended situations such as described above, participants often must make a judgment in a nontrivial sense—for example, when one must choose between competing research proposals, make policy on the basis of current evidence, or decide which of two treatment plans for an illness is backed by the better science. Indeed, the need for such judgments can arise even when there are no deep theoretical disagreements about standards of relevance, for participants of different temperaments often form different assessments of how solid the support is for a given hypothesis. The recognition that all relevant challenges are answered, or can be answered, is not automatic: a competent judge must be neither rash, dismissing further questions and objections, nor indecisive, seeing grounds for doubt where none exist (Lonergan 1970, 284–287). The problem of judgment becomes even more acute when the standards of relevance are open to question or when there are competing interpretations of the problem. The concept of judgment, which plays a central role in rhetorical theory, allows us to supplement and contextualize Habermas’s formal model of argumentation. The key idea is this: in making a judgment between competing arguments, one takes personal responsibility for an assessment of competing plausibilities. So understood, the concept of judgment allows us to see how certain rhetorical devices are internally related to dialectic, such that rhetoric constitutes a further component of rational argumentation, that is, of the construction and assessment of cogent arguments. The internal relation emerges when we construe argumentation as a process of cooperative judgment formation. That is, participants help one another judge the arguments responsibly (thus reasonably) in the hope that the group will converge on a trustworthy or credible collective outcome. The level of consensus—whether deep, shallow, or fragmented—then provides a kind of barometer of cogency, a measure of the level of conclusiveness or inconclusiveness of the available arguments. This
140 Chapter 5
collective assessment is a cooperative achievement that cannot be reduced to individual judgments. From this angle, rhetorical standards govern the use of language in relation to the social-psychological context in which this cooperative judgment formation occurs. Such standards thus focus on the social-psychological effects of language that foster a responsible judgment on the part of the audience. I use the term “social-psychological” loosely here, as a blanket term for those speaker and audience properties that affect the course of argumentation in a particular interpersonal context. We have already seen one attempt to theorize rhetoric in relation to audience psychology and judgment, namely Prelli’s rhetoric of science. Drawing on stasis theory, Prelli elaborates what one might call the “technical substance” of the science community’s ethos: the framework of possible goals, issues, and argument types to which scientists, as members of a discourse community, are committed and upon which they rely in their attempts to communicate with specific audiences. The approach I develop below significantly differs from Prelli’s. Although we both draw on Aristotle, I am concerned more with social psychology as it directly affects judgment. Ethos indeed plays a role here, but first of all as a way of instilling confidence in the speaker’s competence rather than as a commitment to shared framing devices. Rhetorical standards, on the proposed view, tell us how to use language, symbols, gesture, images, and so on, in order to place interlocutors in a position to make responsible judgments. As so described, rhetorical standards are normative—they allow us to evaluate and criticize arguments and their presentation for their effects on our disposition for judgment. Such standards differ from (a) the logical standards that govern the semantic-syntactic interconnections between reasons and conclusion, and from (b) the dialectical standards that govern the critical testing of arguments in relation to challenges to argument content. By satisfying rhetorical standards, participants would presumably increase their ability to approximate dialectical process idealizations. So perhaps one can see such rhetorical standards as a further dimension of process idealizations. Whereas the latter govern the discursive sequencing and social structure of dialogue, rhetorical standards aim at the (ideally responsible) social-psychological use of symbolic expressions and gestures in context. Rhetorical standards also have a positive character: they do not simply rule out manipulative devices and social arrangements that undermine reasonable participation, but provide positive guidance for fostering judgment. Two of
141 Habermas’s Theory of Argumentation
the traditional Aristotelian means of “proof” illustrate the basic idea, though we need not limit the rhetorical perspective to these devices. If argumentation is to issue in judgment, then proofs of the speaker’s character (ethos) and appeals to audience emotions (pathos) may sometimes be positively required to bring about responsible judgment. Properly employed, these devices do not undermine or circumvent logos, or substantive argument on the merits, but rather complete logos by bringing about a cooperative reasonable judgment (cf. McCabe 1994). This holds for scientific argumentation as well: for example, Kitcher (1995) sees the display of competence in laboratory reports as a proof of character that encourages readers to take the findings as trustworthy. Prelli (1989b) shows how scientists explicitly draw on ideals of scientific ethos when they attempt to establish their own (or undermine others’) credibility in controversial argumentation. Given that scientists must rely on each other’s expertise in complex problems, these rhetorical ways of fostering reasonable bases of trust potentially support the assessment of arguments. Indeed, if such moves are part of the argument content that supports a more reasonable judgment in favor of a particular conclusion, then we may even say they contribute to the overall cogency of scientific argumentation as a cooperative process. Rhetorically responsible content would then enter into the definition of cogency. In any case, whether essential to the assessment of cogency or to its definition, devices such as pathos and ethos point to an epistemic conception of rhetoric—insofar as rhetoric makes (the assessment of) arguments more cogent, it does not simply (and optionally) facilitate but partly constitutes public knowledge-production. To get at the basic idea, consider an example at the boundary of science and society, a medical case in which doctors, nurses, and others are discussing whether to continue to treat a terminally ill comatose patient aggressively, on the slim chance of recovery and extension of life, or whether to scale back treatment. To simplify the case a bit, imagine the patient has no living kin.4 One side might argue that there is sufficient hope for recovery (based on the patient’s age and statistical evidence on recoveries) to justify an aggressive strategy, given the importance of prolonging life as a core commitment of medical practice. The other side could counter with the argument that it is unlikely the patient would wish such treatment and that the resources may be better used elsewhere. Here the parties may disagree over the estimation of a number of uncertain facts (chances for recovery, what the patient would desire). The discussion is further complicated by the interpretation of core medical
142 Chapter 5
values—the counterpart of epistemic values in the natural sciences. Both sides may agree about what constitutes the relevant considerations, but they weigh the plausibilities differently. Precisely because the problem calls for such a plausibility judgment, it is rational for participants to make two kinds of persuasive rhetorical appeals, in addition to the logical arguments themselves. First, in making a case for one side or the other, participants should show that as speakers, as proponents of a position, they are properly disposed to judge the arguments and counterarguments. It becomes more rational for a hearer to take one’s arguments seriously insofar as one’s presentation of the arguments gives the hearer evidence of one’s capacity to judge plausibilities responsibly. This is more or less what Aristotle called a “proof from character.” Note that character is not simply a matter of reputation but is given in the process of argumentation itself (Rhetoric 1.2.4.1356a). There are numerous ways this is accomplished. A doctor in our example might, for example, employ a disclaimer, that is, preface her argument for aggressive treatment by emphasizing that she is not the sort of doctor who cannot admit defeat when further treatment would merely prolong dying. By disclaiming opposition to a relevant consideration that motivates the other side, she signals to the others that her judgment is fair-minded and thus that she is unlikely to underestimate a particular relevant consideration. The speaker thereby provides the audience with further grounds for taking her arguments seriously, as the result of a responsible judgment on her part. More generally, a speaker’s care in the presentation of arguments and counterarguments is not simply a dialectical exercise but also a way of providing interlocutors with rhetorical grounds for accepting the speaker as a trustworthy judge of the matter. Indeed, this device can have a similar effect on the speaker herself, increasing her own confidence in the fairness of her judgment. Thus, appeal to character makes the arguments for a position more persuasive, and their acceptance more reasonable, insofar as it gives participants rational grounds for taking the speaker’s judgment of plausibilities as likely to be correct. As already noted, according to Aristotle such grounds or “proofs” are primarily displayed in the presentation of the substantive arguments, or logos. This suggests that we should see them more as a dimension of all substantive argumentation than as a separate type of argument. But the relation to logos is probably a matter of degree, insofar as some character appeals are more closely tied to the logical substance than are others (e.g., listing of institutional affiliation, educational degrees, etc., function as external character proofs; cf.
143 Habermas’s Theory of Argumentation
Lynch 2004). Consequently, a critical rhetorical analysis must gauge the particular type of character proof for its relation to logos. For example, the ability to recapitulate an opponent’s argument before refuting it is internally related to judgment in a way that other proofs of expertise, such as the listing of one’s past publications, are not. Although the latter can instill an audience with greater trust in the speaker—and thus affect how the audience weighs the speaker’s substantive arguments—the first method should be more resistant to abuse, and thus more intrinsically creditworthy, than the latter device. At the same time, given the importance of expertise and specialization in contemporary science, one may not simply dismiss extrinsic appeals as irrelevant to rational motivation. What type of character appeal is considered appropriate, and how significant such appeals are for cogency, will depend in part on the context and argument domain. The use of emotions as arguments has gained increasing respectability in recent decades (e.g., Gilbert 1994). The second rhetorical move, however, appeal to emotions, does not simply involve the use of emotions as distinct reasons. Emotional appeals in the broader sense include all the ways in which the speaker employs emotions, adjusts the emotional pitch of the argument, or selects reasons with emotional overtones, to dispose participants (including herself) to make a responsible judgment. If argumentation aims at judgments of plausibilities, and if this means steering between rash and indecisive judgments, then arguments are more cogent to the extent that they include emotional devices that help participants achieve the proper frame of mind, hitting the mean between rashness and caution. In the medical case, such appeals might emerge in the presentation of social utility arguments for scaling back treatment—that is, one might attempt to stir up compassion for suffering patients who could benefit more than the comatose patient from the available but scarce resources required for treatment. In a broad sense, scientists use the device of pathos insofar as rational argumentation always involves “hot” cognition—not simply a detached logical calculation but a human interest, say, in more elegant theories, surprising counterintuitive discoveries, and the like. We might also see attempts to capture the reader’s interest in the introduction of an article as a kind of pathos. As Kitcher (1995, 57) puts it, an “important function of the experimental report is to advertise findings to those whose research is affected by them.” One way to accomplish this is to connect the findings with certain practical dangers and opportunities. Thus, analytical chemists might motivate interest in their
144 Chapter 5
argument for a new method for detecting a certain pollutant by describing the adverse health effects associated with the pollutant. The rhetoric of science I have elaborated above, as an expansion of Habermas’s argumentation theory, is epistemic in a fairly strong sense. It does not reduce cogency simply to what works in context, but it does bring the social psychology of collective judgment into the rational content of arguments. At the very least, this model implies that an individual cannot accurately grasp the cogency of an argument without engaging in a sufficiently high-quality discourse with others, in which participants present their arguments in rhetorically responsible ways that enhance the judgment of plausibilities. This dependence of the individual on others goes beyond mere information pooling. The individual scientist depends on others, not simply to assemble all the relevant considerations, but to make a responsible judgment of their import for argumentative cogency. Earlier I limited this point to complex dialectical situations, in which scientists must trust each other’s specialized contributions in order to share expertise in collaborative problem-solving. In complex settings, the epistemic importance of rhetorical devices such as ethos and pathos emerges most clearly. But this deep social dependence might be extended to simpler cases as well. Notice that even in the simple medical case posed above, one can see a kind of irreducible intersubjectivity at work, such that an accurate assessment of cogency depends on a collective process in which rhetorical virtues foster good judgment. Suppose that everyone is able to state his or her arguments, and that everyone else can understand the relevant arguments in the sense of grasping the public meaning of statements. In other words, the pooling of information is complete—everything is on the table. Nonetheless, if arguments rest on tacit dimensions of practical engagement, then it is unlikely that each participant can grasp every argument equally well in its full depth. For example, a nurse who has known the patient over an extended period might be able to present a likely interpretation of the patient’s wishes and prospects for recovery—the nurse’s argument is on the table—but other participants can gauge its weight only insofar as the nurse has made a rhetorically effective presentation. The reason is that behind the nurse’s argument lies a body of face-to-face experience with the patient, which is not had by the other participants and is impossible fully to articulate—except perhaps indirectly, say through pathos, ethos, and other suitable devices. Even if the other participants judge that other considerations override the nurse’s argument, that argument, if it has been effectively presented, has led them to consider certain
145 Habermas’s Theory of Argumentation
aspects of the case more deeply than they did initially. Thus their final judgment would be based on a better appreciation of the counterarguments. Consequently, the dialogical outcome would exceed what any single participant could achieve by relying on a private monological assessment of all the relevant considerations. In summary, the substantive rhetoric developed above is internally related to Habermas’s dialectical perspective, and thus provides a coherent extension of his argumentation theory. Besides rendering that theory more respectable as a rhetoric of science, the proposed extension addresses the tension that generated Kuhn’s Gap, that is, the tension between good reasons and social psychology. That is, argumentative practices produce cogent argument products partly because of rhetorical moves that positively affect the psychological conditions of responsible judgment. To be sure, this move still leaves further problems unaddressed, above all the question concerning the socialinstitutional perspective and its relation to cogency. First, however, we must still ask whether this claim about the construction and assessment of arguments implies a social conception of cogency. 3
Cogency in Habermas’s Theory of Argumentation
Before proceeding further let us take stock. I have argued that Habermas’s argumentation theory implies a conception of cogency in which the three traditional perspectives are internally related in the pragmatics of argument construction and assessment: to evaluate the logical cogency of argument products, one must depend in part on a discursive process that displays dialectical and rhetorical qualities. Here the dialectical perspective concerns the basic (ideal) rules for organizing and conducting a critical discussion, understood as a social process in which interlocutors cooperate in testing arguments. The rhetorical perspective refers to the social-psychological aspects of arguments and their presentation that foster (or undermine) a collective capacity for responsible judgment formation. Thus dialectic and rhetoric have to do with both product and process, where the latter includes a procedural dimension: dialectically severe testing procedures and rhetorically responsible modes of address should produce arguments that are both dialectically robust and rhetorically reasonable. For economy of expression, let us say that the dialectical and rhetorical perspectives are concerned with the “dialogical” merits of arguments and argumentation.5
146 Chapter 5
Consequently, what integrates the three levels in the first instance is not a single definition so much as a set of internal dependency relations that link the levels together pragmatically in argumentative practices. Scientists cannot as a rule construct logically cogent arguments or evaluate their cogency without engaging in dialogically adequate processes of critical discussion in which those involved have made responsible judgments about which considerations deserve attention, which arguments are superior, and the like. The internal relations here go from logic to dialectic and rhetoric: in the construction and evaluation of logically strong arguments, scientists depend on or presuppose that they engage in dialogically adequate processes of inquiry and argumentation. A reverse relation also exists, insofar as logical standards function dialogically. Thus an argument is dialectically robust only if it meets logical standards: if, for example, I point out a logical inconsistency in my interlocutor’s response to an objection, then I employ logic as a standard for good dialectical testing. At the rhetorical level, too, we should expect that responsible judgments about cogency respond to logical and dialectical demands; good rhetoric should improve the hearer’s psychological capacity both for logically acute judgment and for listening to an interlocutor’s dialectical challenges. But these internal relations among conditions of assessment do not yet require a multidimensional conception of cogency. Following Achinstein, one might insist on a distinction between the process of constructing and evaluating arguments, on the one hand, and the impersonal merits that properly qualify arguments as cogent, on the other hand—the normative facts that supervene on logico-mathematical and physical facts whether anyone knows about them or not. There are three considerations that argue against this objection, one general, one specific to Habermas’s pragmatism, and one internal to Achinstein’s own theory. The general consideration situates cogency in the aims of argumentmaking as a social practice. According to the objection, we should conceive cogency as a measure of the impersonal representational and justificatory aims of scientific argument-making: arguments count as cogent insofar as they succeed at those aims. As I pointed out in the postscript to Part I, however, social practices of argument-making involve not only representational and justificatory aims but also communicative aims: scientists construct arguments not only in view of empirical adequacy or truth but also with communicative intent (cf. Habermas 1971/2001; 2003, chap. 2). Consequently, to be successful, an argument must succeed in all three of these aims. If an argument fails to com-
147 Habermas’s Theory of Argumentation
municate, then its content may well be representationally adequate in some abstract sense, but it will not belong to publicly acceptable knowledge—it will not count as “science.” If cogency is a measure of the success of an argument, then an argument is cogent insofar as it succeeds at all three aims, including the communicative aims. Moreover, if we identify success with the “merits” of an argument, then we should define cogency in terms of the full range of merits that identify such success: not only the impersonal merits of the argument content but also the dialogical merits necessary for its successful communication. Thus dialogical merits are not simply ancillary means for assessing the cogency of arguments, but are rather co-constitutive of cogency itself. The second consideration is specific to Habermas’s pragmatism. To avoid any misunderstanding, notice that both Habermas and Achinstein share a concern for truth as the primary goal of scientific argumentation. Moreover, both conceive of empirical truth as justification-transcendent—empirical statements have their truth-value in relation to the physical world that transcends arguments and language. Precisely this last point leads Habermas to regard cogent arguments as the only avenue scientists have for ascertaining truth, that is, as fallible indicators of where the truth probably lies. Because we lack unmediated, prelinguistic access to empirical truth, we must rely on more or less cogent argumentative indicators—for ascertaining not only the truth of conclusions but also that of empirical premises, which must be assessed via further corroborating arguments. Thus argument content only indicates that we have probably achieved our representational aims—that the conclusion is probably true on the basis of premises we consider independently acceptable and thus probably true. However, if cogent arguments are to indicate where the truth probably lies, then the merits that qualify arguments as cogent ought to be defined in terms of pragmatically manifest features of argumentation (see Johnson 2000). Because the truth of the evidential premises and background assumptions is justification-transcendent, it is not itself manifest but is only taken as probable on the basis of the manifest qualities of arguments. Thus impersonal truth is not, as such, a merit that makes an argument pragmatically cogent. Conversely, if cogency must be directly open to assessment—displayed in argument-making practices—then any manifest feature of those practices, including dialogical merits, that improves the pursuit of truth would seem to belong in a conception of cogency. The pragmatist, it seems, does not have any deep reason for distinguishing cogency itself, as a property of the argument product, from its
148 Chapter 5
assessment in dialogically virtuous social processes (see Rehg and Staley 2008). To be sure, the second consideration forces the pragmatist to clarify the idea of representational success as a merit that partly constitutes the cogency of arguments—that is, as a feature of the evidential support as opposed to the conclusion. For such success to function as a pragmatically identifiable feature of argument premises, it must refer to empirical adequacy as grounds for taking premises as true. Habermas’s pragmatism thus requires us to distinguish between “immanent” representational success as an indicator of truth, on the one hand, and success in the strong metaphysical or justification-transcendent sense, on the other. His realism then allows us to accept Achinstein’s concepts of potential and veridical evidence as formulations of such “transcendent” success, understood as the normative goal or point of scientific inquiry. Like Achinstein, the pragmatic realist assumes that scientific inquiry aims to reach true conclusions on the basis of evidence and background assumptions that are true. Just at this point, however, a gap opens in Achinstein’s theory of evidence, which suggests a third response to the objection above. In the postscript to Part I, I argued that Achinstein’s theory of evidence, as an account of scientific argument-making practices, must combine ES-evidence and potential evidence: scientists aim to produce not simply a set of metaphysically true statements, but a set of true statements that are justified for their epistemic situation. The merits that define success in these aims lie entirely in normatively supervenient features of the product as both justified and true. However, while scientists strive for potential (and veridical) evidence, they have epistemic access only to ES-evidence. Precisely this gap, Achinstein thinks, accounts for ongoing inquiry. However, he needs to say more about the relation between evidence and process. As his reconstruction of the Hertz-Thomson debate makes clear, features of the process of inquiry play a crucial role in whether or not scientists move beyond ES-evidence that is not potential evidence to ES-evidence that is potential evidence—as Thomson seems to have done. If scientists are to regard their ES-evidence as strongly indicative of potential evidence, then presumably they must have confidence in their process of inquiry. That in turn suggests that the concept of evidence ought to refer not simply to merits of the product, but also to evidence-enhancing merits of inquiry and argument as a dialectical social process.
149 Habermas’s Theory of Argumentation
In closing this section, I consider one further objection against a multidimensional conception of cogency. The problem arises from episodes in which a single scientist (or team) argues against the entire community and is eventually proved correct. That scenario suggests that cogency indeed resides simply in the merits of the content of scientific arguments. The problem here is not that the “lone genius” can reach a true conclusion on the basis of arguments the community finds unconvincing. Habermas’s distinction between truth and justification allows just that possibility. Nor is the problem one in which we have grounds for doubting the reasonableness of the community, say because of coercive or exclusionary forces. In those cases, the collective judgment of the community does not constitute a reliable authority on cogency. Rather, the difficulty occurs when an individual’s grasp of the cogency of the arguments has actually outperformed a community of peers who seem to satisfy process standards of reasonableness. If the individual genius can accurately assess cogency apart from, and better than, the reasonable community, then it seems that cogency is not irreducibly social. Conversely, if we limit cogency to the impersonal justificatory merits of the argument content, then we can say that the genius simply grasped those merits when the community did not. In fact, historical and sociological research has cast doubt on such lonegenius hagiographies, showing how great discoveries, and even the attribution of discovery, depend on social context (see, e.g., Holton 1973; Brannigan 1981). In some cases, such as Peyton Rous’s viral theory of cancer, the innovative discovery involved not so much a superior insight as a gamble in the face of major objections that could not be convincingly answered until methods had improved.6 Other cases can be explained by factors that involve not so much outright distortions of reasonableness (such as masculinist bias) as obstacles to fluid communication or dissemination of results. The “lone genius” and the broader community may simply have drifted apart in their research orientations, so that the latter no longer adequately understands the technical language of the genius. This partly explains the case of Barbara McClintock, whose breakthroughs were not fully recognized for two decades by mainstream geneticists.7 Or the high cost of scientific argumentation may generate a kind of inertia. As explained in chapter 4, argument-making in science is intertwined with experimentation. If the loner is convinced of an unorthodox hypothesis on the basis of novel experimentation, then her arguments cannot
150 Chapter 5
become publicly acceptable in the full sense until other laboratories adopt the new methods or attempt to build their own research on the new hypothesis and methods. The loner’s arguments may have appeared in journals, but the full consideration and critical testing of those arguments does not occur until the new uses of experiment spread beyond the laboratory of origin. In terms of Longino’s process criteria (see note 3 above), there has been a failure of uptake. This last observation suggests a principled response to the lone-genius objection. The genius may well have had a unique insight into a new way of thinking about nature and a novel way of doing science. However, the evidence for this insight and the claim it supports—that the new hypothesis and methods are fruitful for further inquiry—is limited largely to one’s own laboratory experience and knowledge of the discipline-specific theoretical and experimental challenges one claims to be able to resolve. Fruitfulness for inquiry, however, is not settled at the level of a single laboratory or team, but at the disciplinary level. This is evident, I think, if we recall the subtle forms of bias and sitespecific sources of error that can influence judgments at the level of the single laboratory. Only the social process of inquiry-based argumentation can warrant a corrigible confidence that such errors have been eliminated. Thus, although the lone genius may eventually prove right in her judgment, that judgment has not been fully demonstrated until it bears fruit beyond the single laboratory and the discipline actually acknowledges it. One should now see that the lone-genius objection appeals to a scenario that depends on hindsight and thus tacitly assumes the social conception of cogency it targets. That is, the scenario depicts the lone genius, in advance of the broader social process of argumentation, as having an insight primarily into the cogency of a given argument content (a hypothesis resting on experimental evidence) and how that content fits into the discipline as a dialectical field of unsolved problems. In arguing for her view, she claims that her hypothesis and methods are fruitful for further scientific practice in her discipline. But precisely that claim remains untested—and the loner’s argument remains evidentially undersupported—until other scientists can successfully incorporate her work into their own practices of inquiry. Precisely this communicative success, the collective confirmation of the loner’s insight via the social process of argumentation, is essential to the objection. What the objection misses is that the loner’s own argument makes a fruitfulness claim that can be sufficiently supported only when it actually bears up in the broader discipline: only then
151 Habermas’s Theory of Argumentation
can her arguments count as the publicly acceptable knowledge on which the objection relies. In sum, Habermas’s discourse theory implies a multidimensional conception of cogency in which social process, understood as a dialogical process, plays a crucial role. To arrive at such a conception, one must shift the question from a metaphysical to a pragmatic register, understanding cogency in terms of the pragmatics of assessment rather than abstract statements of impersonal truth conditions and logical connections. Thus the question becomes: what arguments ought we to regard as cogent indicators of the truth of their conclusions? In reply, Habermas’s theory answers: those arguments that persuade us in the right sort of way, namely in virtue of logical merits with dialogical credentials. 4
Social-Institutional Reality and Dialogical Idealizations
This brings us, finally, to the social-institutional level. If Habermas’s argumentation theory is to provide a comprehensive conception that bridges Kuhn’s Gap and brings the warring perspectives in science studies to the peace table, then his conception of cogency must include a perspective that focuses on the concrete social contexts and institutions in which scientists conduct their inquiry and discourse. Habermas does not explicitly list a social-institutional perspective as part of argumentation theory. Nonetheless, his discourse theory points toward such a perspective as an appropriate addition to the three traditional ones. This fits with his Aristotelian framework of perspectives, inasmuch as Aristotle’s Rhetoric distinguishes three different types of civic discourse, each connected with a different type of public occasion or institution. Moreover, Habermas’s theory of modernity implies that modern science, like modern law and politics, represents a cultural phenomenon with rationalizing effects on modern institutions and politics (Habermas 1984, 1987; cf. 1970). But exactly how does such a perspective function in Habermas’s discourse theory? There are at least two desiderata that help answer this question. On the one hand, his argumentation theory imposes certain conditions on how we must understand the socialinstitutional perspective. On the other hand, if we want his theory to bridge Kuhn’s Gap, then the social-institutional perspective must do justice to the social contexts and phenomena studied by SSK. In this section I first develop these general conditions (4.1) before turning to the institution of credit as an instructive example (4.2). As I hope to show, Habermas goes considerably
152 Chapter 5
further in addressing Kuhn’s Gap than the approaches in chapter 3. Nonetheless, instructive difficulties remain, which motivate developments in Part III. 4.1 The Social-Institutional Perspective in Habermas’s Argumentation Theory Given the foregoing analysis, we should expect a social-institutional perspective consistent with Habermas’s argumentation theory to involve a “level of presuppositions” that is distinct from, yet internally related to, the dialogical perspective (i.e., the dialectical and rhetorical levels of presuppositions). This is the first desideratum mentioned above. In the process of elaborating its implications we also get a sense of how well Habermas’s approach can address the second desideratum. As a level of presuppositions, the social-institutional perspective seems to include the various social, institutional, and cultural contexts interlocutors take for granted in the actual process of argumentation. These contexts come with particular constraints, some normative, some factual: that one must first make an argument before a particular team of coworkers, that one’s arguments depend on a given level of funding (which affects the kinds of experiments one can conduct), that one must formulate one’s arguments to meet specific journal requirements, and so on. In the process of arguing, scientists rely on and reproduce various social relationships and institutions; in other words, they presuppose a particular “lifeworld”—in sociological terms, they maintain social order (Habermas 1987, chap. 6). The social-institutional context does have a normative character, but it also exists as a social reality—much of which participants do not necessarily regard as especially normative or desirable but rather criticize. This last point allows us to see how the social-institutional perspective differs from the dialectical and rhetorical levels. These too are social levels of presuppositions, concerned specifically with the robust cooperative testing of arguments and the social psychology of collective judgment. But they are constituted by idealizations. This is especially clear with the ideal process standards (equal voice, inclusiveness, etc.), but it also holds for the procedural rules governing the flow of discourse in a robust critical discussion. Longino (2002, 134), for example, considers her process norms—which in effect cover the gamut of dialectical factors and not simply the social organization of discourse—as idealizations. Other dialogical approaches similar to Habermas’s make a similar
153 Habermas’s Theory of Argumentation
claim (see, e.g., van Eemeren et al. 1993). We should also regard the rhetorical standard of responsible judgment as an idealization. Although one can identify the actual use of ethos and pathos in argumentation, one can at most presume in any given case that their employment fostered a more responsible collective judgment—just as, similarly, actors must presume their mutual accountability, as explained in chapter 4. So we might distinguish the social-institutional perspective from these dialogical perspectives by linking the former with empirically verifiable contexts of discourse. The social-institutional perspective thus calls for the empirical study of the micro- and macrosocial contexts of argumentative practices in the sciences. One would examine social influences on scientific argumentation— every mode of social conditioning that SSK cherishes: socioeconomic forces, disciplinary organization, institutional mechanisms (e.g., credit), personal interests, and so on. How, then, is the social-institutional level internally related to the dialogical perspective? More precisely, how does the dialogical perspective call for, or depend on, a distinct social-institutional perspective on the assessment of cogency? A clear connection emerges if one considers Habermas’s process ideals. According to those ideals, we may have confidence that we have subjected the arguments to a severe testing only if we are also confident that all relevant contributors have been included in the discussion and have been sufficiently free and responsible in their participation. But such conditions can rarely be perfectly satisfied, given the limited, time-constrained, messy nature of real discourses; nor could any empirical indicators guarantee that we have adequately satisfied these conditions. Moreover, the idealizations themselves do not necessarily cohere, particularly when one tries to “approximate” them under constrained circumstances (see Blaug 1999; Webler, Tuler, and Krueger 2001). Efforts to maximize informational input, include more participants, ensure equal voice, and build consensus can call for incompatible institutional measures. I present a concrete example of this in chapter 8: in a debate over U.S. National Academy of Sciences expert-committee procedures, participants interpreted and prioritized the ideals of openness and noncoercion differently. And in some contexts, the ideals themselves may be in question.8 Consequently, one must specify dialogical ideals in ways that are socially and institutionally feasible. Real discourses in science are not pure dialogues, but rather sustainable social practices that depend on specific social-
154 Chapter 5
institutional conditions. Apart from the latter, the dialogical perspective remains incomplete and ambiguous. Social-institutional realization can function normatively in the assessment of arguments, however, only if one presumes that it “sufficiently approximates” dialogical conditions. Thus the dialogical and social-institutional levels presuppose each other. One can then say that the social-institutional perspective, as a critical perspective, tests the presumption of sufficient approximation by scrutinizing both the design and execution of scientific inquiry/discourse for their dialogical adequacy. One asks whether the actual process of scientific inquiry and discourse, as these are institutionally structured and carried out, warrant a confidence in the reasonableness of outcomes, specifically, a confidence that the trend of opinion in the relevant subdiscipline reflects the view that would be reached, were the discourse as dialogically reasonable as it could be for a given level of resources. If the process is sufficiently dialogical, then greater consensus indicates a more cogent argument.9 I suggest that the following statement captures the various relationships that constitute Habermas’s conception of cogency: (CH) We may regard an argument as cogent insofar as it emerges as the consensual product of a dialogically adequate argumentative practice, that is: insofar as (a) the logical merits of the argument content have been generally acknowledged (b) in an actual discourse and inquiry, sustained by a socialinstitutional context whose design and execution warrant the presumption that (c) participants have sufficiently realized a dialogical process in which they are able to make collectively responsible judgments about the construction and critical testing of competing arguments (where sufficiently realizing such a process means the outcome would not significantly change were the process as reasonable as it possibly could be). The conception links the cogency of arguments with all four perspectives: logical (a), dialectical and rhetorical (dialogical) (c), and social-institutional (b). It thus ties the normative constituents of cogency to the social contexts in which arguments actually rise and fall. On this interpretation, then, Habermas integrates social context, in the sense that SSK understands such context, into a normative conception of cogency. That was the second of the two desiderata above. Does he therefore succeed in bridging Kuhn’s Gap? In fact, his model has a two-tiered structure that creates some problems for bridging the gap as successfully as one might hope. To see the difficulty, it
155 Habermas’s Theory of Argumentation
helps to reformulate (CH) in a way that pulls apart its idealizing component, which functions as a presupposition, and its real warrant: Idealizing presupposition Scientists may reasonably judge their arguments to be logically cogent only if they presuppose those arguments would be generally accepted in dialogically adequate processes of argumentation concerned with the logical merits of the arguments at issue. Real warrant This idealizing presupposition must be warranted by observable features of the actual social organization, conduct of scientific inquiry and discourse, and level of consensus. These two components suggest a distinction analogous to the one that some epistemologists have drawn between justification and warrant. Plantinga (1993), for example, associates the idea of justification with internalist epistemology, which emphasizes the obligation of holding only those beliefs commensurate with the evidence and reasons of which one is aware and thus can adduce as justification. Warrant has a broader externalist sense that includes sources of reliability of which the knower may not be entirely cognizant but can at most presume. Broadly speaking, dialogical idealizations suggest a kind of collective internalism: a maximally thorough and transparent discussion in which all the available evidence is evaluated on the way to justifying some hypothesis. In the actual process of scientific discourse, however, we do not have full empirical access to whether we have satisfied dialogical conditions; at most we can defeasibly presume that the surviving points of view rest on dialogically cogent arguments. This presumption is thus warranted by observable conditions: it holds up so long as critical scrutiny of the organization and conduct of discourse does not uncover any obvious dialogical flaws. In any case, as a two-tiered model, Habermas’s theory involves a two-step process of indication: social context is an indication of an indication of where the truth probably lies. Diagrammatically, the model sets up a chain of connections as shown in figure 5.1. (For the sake of economy, the schematic in that figure folds the logical perspective into the dialogical: more precisely stated, the “ideal dialogical conditions” in the middle box are conditions for constructing and testing logically cogent arguments we may take as indicating the probable truth of T.) On this schematic, dialogical idealizations are in an odd place, pragmatically speaking. As a pragmatic indicator of probable truth, cogency should be
156 Chapter 5
Observable social-institutional conditions of actual discourse about theory T
warrant
a presumption we have satisfied ideal dialogical conditions that
indicate
probably, T is true.
Figure 5.1
something to which arguers have direct access, even if they have that access in the full sense only collectively. But the two-tiered model means that we do not have direct access to cogency as such, whose conceptual core lies in an ideal process of justification. Rather, we must rely on social-institutional indicators—the observable organization and conduct of discourse—as a defeasible warrant for presuming we have such a justification. As a second-order indication of probable truth, social-institutional indicators lie at one remove from justification itself, which in effect has acquired a metaphysical status. Besides the pragmatic oddity, the two-tiered model presents a potential obstacle to bridging Kuhn’s Gap at the disciplinary level. The problem arises from Habermas’s understanding of the social warrant. As described above, this warrant functions as a negation of a negation: the warrant holds so long as scrutiny of the process does not uncover nondialogue—that is, so long as we cannot show that “obviously relevant participants are excluded, obviously relevant contributions suppressed,” or that obvious manipulation occurs (Habermas 2003, 108). This understanding implies a role for the sociology of science that is dangerously close to the “sociology of error” that SSK rejected so vehemently in response to philosophers of science who relegated sociologists to the explanation of bad science. Habermas’s two-tiered model suggests such a role: sociologists should help normative theorists (i.e., philosophers) figure out if the social context has features that distort communication and reasonable argument, thus undermining warranted confidence in the cogency of arguments. Given such problems, should we not expel dialogical merits from the pragmatic conception of cogency? Here one might be tempted to follow theorists like Rorty (1989, 2001), who jettisons dialogical ideals as pragmatically otiose. In that case, one would rely simply on the logical and social-institutional perspectives for a pragmatic model of cogency. But this move too is problematic: without such ideals as equal voice, inclusiveness, and the like, how could we
157 Habermas’s Theory of Argumentation
identify features of the actual discourse that increase our confidence in its outcome? In fact, dialogical ideals play an important role in the actual organization of scientific discourse. As the case study in the next chapter will show, scientists themselves refer to such ideals when they design procedures for organizing discussion and paper writing in large collaborations. Dialogical ideals also inform procedural rules in expert committees that advise policymakers (see chapter 8). To address such difficulties we must revisit the core commitments of Habermas’s formal pragmatics. As we saw in chapter 4, he criticizes Toulmin precisely for failing to distinguish a set of formal standards—above all, dialectical process standards—that transcend specific institutional conventions. In articulating such ideals as counterfactual idealizations, however, Habermas places a crucial component of his conception of cogency beyond the empirically identifiable properties of actual discourse. The question then is whether we can construe such ideals in a way that does not relegate their normative force to a counterfactual presumption. I take up that issue in Part III. The idealized counterfactual character of Habermas’s model also creates a problem in applying it to actual discourses. If dialogical ideals are not a system one can straightforwardly employ for the design of institutions—and I agree with Blaug (1999) that they are not—then the question arises of how one knows whether a given design provides the warrant described above. In many cases, a given ideal is not only limited (e.g., not everyone is actually included), it is even contradicted (see note 8). How does one analyze such cases from the social-institutional perspective? Presumably the ideal itself does not suffice. To explore this problem more fully, I attempt to apply Habermas’s model to the institution of credit in science. 4.2
Sustainable Social Order in the Sciences: The Case of Credit
As I have interpreted it, Habermas’s discourse theory gives us, on the one hand, a conception of cogency whose logical and dialogical dimensions provide critical standards for scrutinizing actual scientific institutions and discourses in their real settings. On the other hand, it also refers to a social-institutional level as the place where argument-making practices actually unfold. As a process that involves ongoing social relationships, time constraints, material resources, and financing, real scientific discourse is subject to the existing demands of social order—the need to maintain scientific practice as a sustainable social
158 Chapter 5
organization. This is indeed a “level of presupposition” inasmuch as participants presume their institutions do not systematically undermine the search for truth. But it also presents demands of its own on argumentation, the requirement that arguments function somehow in maintaining a stable ongoing cooperation. To get a better understanding of such demands and the argumentationtheoretic problems they involve, consider the role of credit in the sciences. As is well known, the attribution of credit plays an enormous role in the growth of scientific knowledge. According to Hull (1988), science is epistemically successful in large measure because of such institutional mechanisms as credit, which channel self-interested behavior (of people with an intrinsic curiosity about nature) into productive modes of cooperation. As creatures seeking not simply truth but also their own reputation and career advancement—which in science usually means the ability to do more science—scientists have an individual, competitive interest in receiving recognition for their work. Supplementing Hull’s analysis, Kitcher (1993, chap. 8) has used game theory to show how credit can have epistemically beneficial effects under the right conditions. In terms of Habermas’s categories, these analyses link credit with strategic action, which Habermas distinguishes sharply from communicative action and discourse. If the effect of credit displays itself in strategic modes of action, then it seems that Habermas should place credit mechanisms at the level of institutional conventions that enlist strategic action for purposes of sustaining discursive practices.10 Specifically, the promise of recognition and advancement motivates scientists to undertake risks and contribute to epistemic aims. Such motivations are necessary because scientists, like most human beings, are not motivated solely by the search for truth, even when they are searching for truth. Thus credit mechanisms are part of the social organization that sustains the production of public knowledge. What makes credit an interesting challenge for Habermas’s conception of cogency arises from a particular conflict it creates: the attribution of credit requires a certain amount of secrecy in science communication, thus apparently undermining the openness required by the ideal of inclusiveness. If scientists were motivated solely by the desire for truth as a public good, then the inclusiveness standard would apply straightforwardly to communication between teams working on the same problem: they should share information without restriction in the hope that many hands make light work—allow us to
159 Habermas’s Theory of Argumentation
arrive at probable truth more quickly—without strategic concern for who gets there first and receives credit. In fact, just the opposite is often the case: it is expected that a research team may withhold its results, and refuse to share data, until it feels ready to publish; attempts by non-team members to obtain data, or team members to share data with outsiders, can provoke ethical criticism.11 This kind of conflict between an institutional procedure and process ideals is typical when we confront ideals with reality. The critical analyst must then find a justification for the departure from the ideal or else reject the procedure as a distortion. Justifications attempt to reconcile ideal and reality: on the one side, idealizations such as inclusiveness, on the other, the realities of human psychology and constraints on various resources. One then explains why the departure is necessary or helpful from the standpoint of institutional feasibility and social-psychological realities—in the case of credit, the reality of strategic, self-interested behaviors that one cannot eliminate simply by reminding participants of their idealized “pragmatic presuppositions.” One might justify secrecy norms in two steps: (1) Credit mechanisms are necessary for science to sustain itself socially, given that scientists (like all human beings) are significantly motivated by self-interest (and the need for recognition). (2) For credit to function, one must temporarily suspend the openness requirement, to prevent competitors from coopting a discovery and the credit it brings. Credit mechanisms, and norms allowing for temporary exclusion and secrecy, therefore specify the openness/inclusiveness ideal by situating its operation in a temporal framework. Such mechanisms are legitimate insofar as they make it feasible for self-interested actors to cooperate in a collective epistemic endeavor. The analysis of credit also shows that Habermas’s model, its highly idealized and formal construction notwithstanding, can draw on a range of considerations that bring the idea of cogency down to institutional earth. One specifies dialogical ideals, that is, in light of the epistemic goals of scientific argumentation and feasibility considerations. In addition, one can bring in historical considerations: which modes of organization failed and which succeeded in the history of institutional experimentation. To summarize, one might say that from a social-institutional perspective, well-designed discursive procedures aim to optimize the substantive quality of the outcome as measured by the goal(s) of the procedure and what is socially feasible in the light of a particular historical experience. If things go well, argumentative practices
160 Chapter 5
within such institutions warrant the presumption that the arguments are dialogically adequate. The key question for assessing institutionalized discourse is whether a given design and the discourses it supports satisfy this adequacy condition. To answer this question, one must engage the services of a contextualizing reason that delves into the substance of the various contextual considerations. But notice that contextualization takes a top-down approach in Habermas’s model. The critical assessment of institutional design and execution proceeds via the critique and defense of competing limitations on impossible idealizations. These represent an ideal of cogent justification, which must then reconcile itself with an imperfect contextual reality. Dialogical ideals, that is, articulate what it means for a discourse to generate epistemically cogent arguments: they provide the court of reason before which institutionalized procedures must prove their innocence, or at least their necessity. As ethnomethodologists have pointed out (e.g., Bogen 1999), Habermas’s strategy runs the risk of overlooking a more context-sensitive mode of reasonableness. I have argued above that his approach operates in a negative manner, showing that a given design does not exhibit any evident dialogical flaws. To anticipate Part III, this strategy neglects the possibility that contextualization also involves positive forms of context-sensitive reason. Moreover, it assumes that process ideals have a kind of a priori relevance, a presumptive jurisdictional force for every argumentative context, such that the burden of proof lies on showing the acceptability of contextual adjustments. Overlooked is the alternative possibility that the ideals themselves might first have to prove their relevance for a given context—that the burden of proof in critical scrutiny might lie on the other side of the ideal–reality divide. Notice, finally, that the model of cogency acknowledges the social-institutional aspects primarily at the level of macrosocial conventions; the role of microsocial contexts of argument remains unclarified. This issue, too, bears on how we understand process ideals. Before I address that topic in Part III, I want to confront Habermas’s model with an actual case of argumentation (chapter 6). This case helps clarify both the resources and limitations of his model for illuminating argumentative practice in the sciences. I then close Part II with a postscript that shows how a Habermasian model of cogency, its highly idealized character notwithstanding, still allows for interdisciplinary cooperation with relativist versions of SSK. That granted, the difficulties in Habermas’s
161 Habermas’s Theory of Argumentation
approach—the problems with context I have described in this chapter, limitations in treating the Fermilab case in chapter 6, and the hurdles for interdisciplinary exchange—point to a deeper problem in how his conception of cogency integrates argumentation-theoretic perspectives. This deeper problem, with which I conclude the postscript, calls for a different approach to an integrated conception of cogency—an approach that takes us beyond the Habermasian framework.
6 Argumentation at Fermilab: Putting the Habermasian Model to Work
I now turn to an actual case of scientific argumentation, an episode leading up to the discovery of the top quark at Fermilab. Although Habermas’s model is often criticized as overly idealized, this case displays features that fit his analysis. In writing their papers, the scientists at Fermilab sought consensus on both the conclusions and supporting evidence; their methods of discussion presuppose dialectical ideals similar to Habermas’s. This initial congeniality thus makes points of divergence all the more interesting. I use the case primarily as a way of further clarifying Habermas’s model and examining its potential for illuminating actual processes of argumentation. After an initial clarification of these tasks (sec. 1), I provide some of the technical background (sec. 2) and then describe argumentative features of the central episode, the writing of the paper claiming evidence for the existence of the top quark (sec. 3). I close by discussing implications of the case for the model of scientific argumentation sketched in chapter 5 (sec. 4). 1
The Tasks of a Case Study
In the context of argumentation theory, case studies potentially serve at least three aims. First, case illustrations further clarify, and render more precise, argumentation-theoretic categories. Cases give us a concrete sense of how these categories and norms can actually function in the sciences. This is not to say that isolated cases support a general claim about the overall prevalence of the argumentative practices exhibited in the particular cases. One obtains at most an initial confidence that the argumentative intuitions one has articulated theoretically find some resonance in actual practice.
164 Chapter 6
Clarifying the model in relation to a case thereby serves a second aim as well, that of providing a kind of preliminary test of the model. The testability of Habermas’s argumentation theory arises from his claim to articulate participant-perspective intuitions. If the various perspectives on argumentative cogency are not simply theoretical constructs but “levels of presuppositions” that participants themselves make, then we can test the model by examining how scientists themselves engage in the critical assessment of arguments. Notice that the mode of critical assessment involved in these first two aims is rather modest, requiring one merely to describe the normative attitudes of those involved rather than prescribe what those attitudes ought to be. Our assessment is critical insofar as we do not unmask the critical perspective of the participants or reduce it to a naturalistic explanation, but take it as the participants themselves do. This analysis of participant attitudes can then generate modifications in the model, in accordance with the second aim. With these first two aims of clarification and testing, the Habermasian model functions partly as a heuristic, which helps us notice and describe the explicit and tacit methods, behavior, and choices of scientists in concrete discursive situations. However, unlike the open-ended categories I employed as a heuristic in Part I, a particular model like Habermas’s is more heavily laden with theoretical commitments. And like any theoretical framework, what we notice and how we describe it depends on the categories we employ. That granted, theoretical categories do not saturate observation so thoroughly that we find only what we want. In fact, the Fermilab case contains features that are not anticipated by the theoretical framework—features that call for modification of that framework. If the first two goals can be achieved, then Habermas’s model can avoid the criticisms that Gaonkar (1997a) leveled against the rhetoric of science, namely that it simply imposes on scientific discourse a veneer of unilluminating, highly plastic—thus untestable—jargon. Insofar as Habermas’s model draws out the normative concerns and argumentative choices of scientists themselves and does not simply redescribe discourse in an alien jargon, the model is not a mere overlay but illuminates the normative issues scientists themselves recognize; thus it is open to revision in light of actual practice. As I hope to show, Habermas’s model is testable, and testable precisely against the actual normative practices of scientists. The third aim of a case study is more ambitious: the use of argumentationtheoretic categories in the directly prescriptive mode of critical assessment. This
165 Argumentation at Fermilab
more ambitious mode of critical assessment involves judgments of cogency that go beyond the modest description of the scientists’ perspectives. Consequently, this third aim appears most feasible, it seems to me, only when (a) the science is genuinely up for debate, or when it bears on controversial policy questions, and (b) we pursue the aim not solely from an outsider’s position but through an interdisciplinary dialogue involving the scientists themselves. Except for a modest suggestion in section 4, I shall not venture into this critical territory. The idea that ties the three aims together is this: having identified normative criteria that scientists themselves seem to take seriously, we have a multidimensional set of questions we can pose for scientists (and policymakers) when they make public claims about their findings. The argumentation-theoretic framework then provides a set of dialogical sites, points of orientation, which both scientists and their critically engaged addressees presumably share. 2
The Evidence for the Top Quark: General Background
For the case illustration I draw on Kent Staley’s (2004a) detailed study of the process of argument construction by the Collider Detector at Fermilab (CDF) collaboration in 1993–1994. Specifically, I examine the collaborative process of writing up the article for the first evidence of the existence of the top quark (Abe et al. 1994a,b). The science is complex, but we can get at the relevant argumentative detail without delving too deeply into the technicalities. (Some details and background I supply in the notes.) As an episode in experimental physics, the CDF collaboration represents a development in high-energy physics (HEP). As a research tradition, HEP goes back to late nineteenth-century experiments on cathode rays (fast electrons), as well as experiments in which C. T. R. Wilson studied cosmic rays by analyzing their visible traces, captured by high-speed photography, in “cloud chambers” (see Galison 1997; Kragh 1999). Experimental advances in the detection of increasingly elusive subatomic particles eventually came to depend on massive instrumentation and large interdisciplinary collaborations involving, among others, physicists, nuclear chemists, mathematicians, computer scientists, engineers, technicians, and plate scanners. HEP experiments typically rely on various methods of detecting the products of high-energy collisions between subatomic particles moving at velocities approaching the speed of light in linear and ring accelerators. The construction of detectors and interpretation of data depend on theoretical models and experiments that
166 Chapter 6
allow scientists to anticipate the kinds and probabilities of decay products and their physical properties (e.g., mass, cross section, charge, momentum, kinetic energy, decay angles). The CDF experiment employed the Tevatron colliding beam ring accelerator in which protons and antiprotons collide in a massive three-story Collider Detector. CDF scientists hoped to detect the top quark (t) by various methods, one of which searched for products of its decay to the bottom quark, that is, its b decay. The bottom quark is not detected directly, but inferred by its decay into a characteristic jet of particles and the production of leptons (in this case, electrons, electron neutrinos, muons, and muon neutrinos). This search was guided by predictions of the “standard model” in particle physics as well as by experiments that delimited the mass of the top.1 The argumentation in the CDF collaboration stemmed in large measure from the statistical character of the analyses involved in the argument for the top quark. We can get an initial feel for the general kind of problem by going back to the early days of what Galison (1997) calls the “logic tradition” in HEP, which he distinguishes from the “image tradition.” For present purposes we can distinguish these by their different modes of detection. The “image tradition” relied on media (e.g., water vapor in cloud chambers, liquid hydrogen in bubble chambers) in which single collision-and-decay events could be captured photographically in complex detail—one got a pictorial map of various particle tracks involved in the collision. By contrast, instruments in the logic tradition respond to a single property of an event (e.g., ionization produced in a gas by a particle). One understands a decay process not by looking at its visual map but by amassing data from a large number of events and analyzing it statistically. To get a sense of the logic approach, consider an early counting device, the Geiger-Müller counter, which could produce an electrical signal in the presence of charged particles (see Galison 1997, 438–454). Thus, to detect a specific decay event, the trick was to arrange a number of Geiger counters into a logic circuit, a spatial and electronic architecture designed to produce a signal in the presence of the sought-for decay. There are various arrangements, but a simple model involves a “coincidence circuit,” say of three counters arranged so that all three fire in the presence of the sought-for event. The arrangement might, for example, take advantage of the theoretical expectation that in a magnetic field the decay produces two particles moving off at a known angle from the point of decay or “vertex,” as shown in figure 6.1.
167 Argumentation at Fermilab
point of decay
counters
Figure 6.1
The partial nature of the data, however, creates a problem. In the case shown in figure 6.1, a signal indicates that all three counters have fired, which in turn implies that a charged particle passed through each counter. This is very little information about the kind of event that actually produced the signal: we cannot rule out the possibility that other types of decay, or simply three simultaneous unrelated particles, set off the counters—not to mention other sources of interference. Consequently, identification of the sought-for event requires one to distinguish “genuine” signals (or “signal events”) from background events, random noise, and so on.2 If theoretical considerations along with an understanding of detector performance allow one to estimate expected average background, then one can solve the problem statistically by recording a sufficiently large number of events; if the circuit produces a signal rate above the expected background rate, then one can calculate the probability of observing the higher rate in the absence of the signal event. This probability tells us the odds of getting a statistically random, thus misleading, excess of signal over background. Given this probability, we can assess the null hypothesis (H0) for our experiment, namely that the sought-for event was not observed. The lower this probability, the more statistically significant the excess signal, and thus the greater one’s confidence that H0 is false: one has indeed observed the signal event. Between the Geiger counter and CDF, of course, lay several quantum leaps in instrumental development. The types of counting devices proliferated, each having distinct strengths and detection capabilities (Galison 1997, 454–479, 484ff., 515–529). More importantly, methods of closely stacking counters in spatial arrays allowed scientists to track particles through space with a precision that rivaled the photographs obtained from bubble chambers.3 The detection system in Fermilab’s colliding detector combines a number of types of
168 Chapter 6
subdetectors; data collection and analysis then depends on powerful computers that execute sophisticated algorithms.4 At the same time, however, the soughtfor decay products have become increasingly rare and elusive. Consequently, the level of background interference has intensified: in effect, all the collision processes studied in the past become so much camouflage for the desired exotic species. For example, to discover the Z0 boson in the 1980s, scientists at CERN had to distinguish one event out of every ten million interactions (KnorrCetina 1999, 50–51). These complications notwithstanding, the core argument one has to make in such contexts remains the relatively straightforward statistical argument alluded to above: one must falsify the null hypothesis, which in the CDF case we may formulate as follows: (H0) “This data sample has been drawn from a population of protonantiproton collision events that is free of top-quark production” (Staley 2002, 285). Notice that H0 requires one to select a sample from the total population of collision events in the detector. To do so, one must determine the threshold values, or “cuts,” for taking an event as a “candidate” to be included in one’s sample. This decision is necessary because the sought-for event or particle can be detected in a number of ways, in virtue of different features or properties the event shares with other background events. Such features indicate a signal event but do not guarantee it. In the Geiger counter example, this decision is presupposed in the counter-circuit design itself, which takes the decay vertex as the feature indicative of the sought-for decay. In other cases, a threshold energy level of particles might serve as the cut. To assess the null hypothesis, then, we need to know how many apparent or candidate signal events we should expect if H0 is true (in which case, the candidates would all be misleading mimics). More specifically, we want to know how probable it is we would see at least one such event, or two, or three, and so on: a “null probability distribution for X,” where X is the number of candidate events. To calculate this distribution, physicists draw on theoretical knowledge, their understanding of the detector, and computer simulations. If the null probability distribution says that the probability of detecting four candidates (four “tags” in the HEP lingo) is very low, then we have a statistical argument for rejecting H0 and accepting the alternative hypothesis J: “This data sample has been drawn from a population of proton-antiproton collision
169 Argumentation at Fermilab
events that contains some top quark-producing events” (Staley 2002, 285; cf. Mayo 1996, 92–99). 3
The Writing Process as Argumentative Practice
I turn now to the episode at issue in this chapter, the writing of the paper announcing evidence for the top quark (hereafter the “Evidence paper”). As Staley (2004a, 168) points out, the writing process itself is a kind of procedure for organizing critical discussion: by “providing a textual focus for argument and consensus building, the research paper also provides a means of organizing such deliberation” (i.e., critical discussion). The writing process provides discussion with a substantive focal point, an argument content subject to specific writing demands determined by relevant logical standards, subdisciplinary conventions, topical requirements, and the like. I work through this process by dividing it up into a series of phases convenient for isolating the argumentation-theoretic aspects implicit in Staley’s account. (1) From a dialectical perspective, the research context for the top search was a very favorable one for CDF. The top quark was the only unobserved particle in the standard model, and thus empirical evidence for its existence posed the outstanding experimental challenge for the standard model. This dialectical situation made things rhetorically easy for CDF, at least in relation to the HEP audience: physicists needed no convincing that any findings on the top were of great importance. Moreover, CDF had shown in earlier experiments that their main competition, CERN, lacked the instrumental capability to detect the top, given the high collision energies required to produce the relevant decays (Staley 2004a, 104). CDF’s technology gave it a winning hand in the “competition for the better argument”—they only had to play the cards competently. In Latour’s terms, the collaboration’s “alliance” with the Tevatron gave them a massive rhetorical advantage. (2) The episode under consideration begins with the data collection known as “run Ia,” which took place in 1992–1993. Three subgroups, working independently of one another, analyzed the data using three different counting methods: the SVX and SLT groups focused on decays resulting in a lepton plus “jets” (a dense spray of particles), while the “dilepton” group searched for b decays that produced pairs of leptons.5 A fourth group took a different approach, the reconstructive analysis of the kinematics of decay events (see [4]
170 Chapter 6
below). When the three counting groups first shared results at a collaboration meeting in July 1993, an unexpected convergence emerged, suggestive of top quark production. As one team member put it, “it looked like there very well might be something there” (quoted in Staley 2004a, 135). However, the collaboration hesitated to claim a discovery. In fact, at this point the group faced a problem at the logical, dialectical, rhetorical, and social levels. These problems were somewhat acute in view of an upcoming (August 1993) conference in Ithaca, at which the audience would surely expect some report. From a logical perspective focused on the product, one asks whether the premises support the conclusion and how strongly they support it—which in turn bears on the force or status one ascribes to the conclusion. In fact, the group was uncertain about whether the convergence of findings supported a discovery claim, a weaker evidence claim (i.e., we have evidence for the top quark but it is not yet a settled issue), or no interesting claim at all. Besides doubts over the force of the conclusion, we shall see considerable debate over specific aspects of the findings and whether this or that analysis should be included as evidence. From a dialectical perspective, one confronts challenges to one’s argument. The CDF scientists were explicitly worried about whether members of their team would be able to answer the questions they would surely encounter in the wake of a strong claim at the conference. Part of the problem lay in lack of preparation, the fact that the team had not yet fully assimilated the findings (ibid.). The Ithaca conference itself made the above problems immediately rhetorical and social. The group faced a rhetorical problem insofar as they had to consider the reaction of the audience to their presentation. Would a presentation of the data be received as a discovery claim even if the group did not explicitly make such a claim? If so, then a social problem presented itself: the embarrassment that other scientists would “discover” the top quark in CDF’s own data (ibid.). Such an outcome would place the competence of the group in doubt, thereby undermining its intellectual authority among physicists. It would also weaken any credit the group would receive for the discovery, should it turn out that the data indeed supported a discovery claim. The four perspectives allow us to see that the CDF presentation at the Ithaca conference solved a four-dimensional problem. At the conference, collaboration member Paul Tipton presented only part of the data. This immediately solved, or better, dodged, the logical and dialectical problems: the group drew no solid
171 Argumentation at Fermilab
conclusions at all, and it thereby avoided having to defend a claim against objections. At the same time, this move avoided the rhetorical and social difficulties noted above: with only a partial dataset available, the audience could not find a discovery latent in the group’s presentation. But in presenting only partial data that yielded no conclusion, Tipton faced a further rhetorical challenge: why should the audience find the talk interesting? His solution involved a further rhetorical move, one we might see as combining pathos and ethos: he promised his audience specific improvements that would soon yield more complete data, concluding his talk with the provocative coda, “stay tuned.” He thus piqued his audience’s interest, but precisely by displaying his competence, that is, by giving substance to his promise by linking it with recognized technical possibilities. Meanwhile, the CDF collaboration turned to writing up the results. Paul Tipton and Tony Liss, heading a group of around ten writers, initially proposed four short articles for Physical Review Letters. Rejecting this idea, CDF opted for a longer article (the “Evidence paper”) to be submitted to Physics Review D (Staley 2004a, 139–140). After describing the procedures that structured the discussion of the drafts and how these procedures were actually carried out (3), I examine two specific debates over data analyses (4, 5). I then notice some interesting features of the consensus that finally emerged (6). (3) Given the complexity of the analyses and distribution of relevant types of expertise in the CDF collaboration, open critical discussion provided an important means of ferreting out errors (see Staley 2004a, 137). At least officially, the CDF collaboration recognized this, and thus structured writing and discussion procedures accordingly. Membership in the writing groups was open to anyone interested. The conveners (Tipton and Liss) scheduled meetings and kept the minutes. The writing procedure itself involved three broad phases or steps: 1. The conveners ask the spokespersons for the collaboration to appoint a committee of “godparents” for their paper. Modeled on journal refereeing procedures, the godparents are supposed to provide an independent, and thus objective, initial critique of the draft along with suggestions for improvement. Only after they approve a draft can the group present it to the collaboration. 2. The group then presents its paper to the whole collaboration (including scientists not involved in the top quark experiment) for open critical discussion.
172 Chapter 6
3. The collaboration then either accepts or rejects the draft as ready for submission to a journal or presentation at a conference. Acceptance requires (nearly) full consensus. In fact, the actual writing process involved procedural ambiguities, partly the result of the authors’ haste to publish, partly the result of a lack of consensus on procedure. The writers moved quickly with the four short papers, to the detriment of open discussion, as Liss concedes (ibid., 139). The result was the so-called October Massacre, the rejection of the initial proposal in favor of a single long article. Moreover, in contrast to the usual procedure, the godparents got involved in the drafting of the Evidence paper itself, instead of waiting for a finished draft. In the minds of some collaboration members, this compromised the objectivity of the godparents; however, other members thought it helped produce a better draft (ibid., 137–139). To open the discussion further, paper drafts were posted on a computer site and any member of the collaboration could send questions (via computer) to the writing group. Eventually, open meetings were held two or three times a week, which at least gave anyone the opportunity to provide input. To improve civility in discussion, Tipton had the godparent committee vet responses by writers to questions and objections (ibid., 140, 156–157). These moves represent attempts to bring a messy social, argumentative process under procedural control. The goal is clearly dialectical: to test arguments for possible errors, so that the final product can withstand criticism. Thus the scientists took a dialectical perspective on their writing process: they designed procedures according to standards of dialectical robustness. As I explain below in (4), this perspective plays a particularly important role in statistical testing procedures. But it also seems to govern the discursive procedures employed at CDF. Notice that such procedures governed both social roles and the course of discussion, a distinction that reflects the two sides of the dialectical perspective sketched in chapter 5. CDF procedures assigned different roles (godparent, convener, writer, audience), each of which involved somewhat different duties and rights (e.g., to post drafts, to respond to audience criticisms, to impose civility on discourse, to participate in open discussions). But some of these role obligations also governed the course of discussion, for example, the requirement that writers must respond to criticisms. (4) Mayo (1996, 72–73) draws a distinction between two approaches to statistical inference: “evidential-relationship” (E-R) and “testing” approaches.
173 Argumentation at Fermilab
The first is exemplified by logics of confirmation, such as Carnap’s objective Bayesianism. As I argued in chapter 2, such logics take a narrowly logical perspective on argument products: one wants to capture in quantitative terms the degree of support that a set of observation statements provides for a given hypothesis. Testing approaches, by contrast, “focus on finding general methods or procedures of testing with certain good properties” (ibid., 72). If we take discussions of drafts of the Evidence paper at CDF as illustrative, then no adequate account of statistical inference can ignore the testing approach. Concerns with the adequacy of testing methods lay at the heart of these debates. The SLT group analysis (based on “soft lepton tagging”) provides an instructive illustration (Staley 2004a, 143–148). According to the standard model, one way of spotting the top involved a decay process that generates the bottom quark (b), which then decays into leptons (muons and electrons). In this particular type of decay, the leptons are “soft” in the sense that their transverse momentum (pT, the momentum component at right angles to the collider beam axis) is low. Lepton momentum is low because the same decay process also generates the heavy W-boson, which absorbs most of the energy involved in the process; as a result, the leptons that issue from the b are low energy (and thus have low momentum). However, the range of values that pT can take in such decays depends on the mass of the top quark, which CDF had estimated but did not know precisely. Moreover, soft leptons are generated by other collision events besides top decays. Consequently, CDF had to make a decision about the appropriate cuts for pT. The decision that the SLT group took—to count as candidates those leptons with energies above 2 GeV/c— proved contentious.6 The central objection grew from the suspicion of bias in the SLT’s choice of cuts—that they had “tuned on the signal.” This type of bias represents one of the potential problems in massaging data statistically, namely that one will choose the type of statistical analysis that yields the answer one wants to find. One can avoid this suspicion by following a “no-peeking” rule, which requires one to choose the cuts before looking at the actual data. In fact, such a rule is particularly difficult to enforce for HEP detection systems. The reason is that one cannot fully determine if the system is functioning as one wants without actually seeing how it performs at collision energies likely to produce quark data.7 However, the immediate cause for suspicion lay in the fact that in 1992 the SLT group considered raising the pT cut to 4 GeV/c, but then decided to stay with the lower threshold
174 Chapter 6
of 2 GeV/c for the Evidence paper. At that threshold, there was greater background, but also a greater number of tags (seven, as opposed to four at the higher cut). Had the group chosen to stay with the lower threshold because it yielded more tags? Given how bias in the choice of cuts can operate unconsciously, a simple denial of bias does not suffice here. Thus the group responded with three main counterarguments. First, the lower threshold was consistent with the cut from an earlier experimental run. Second, the lower cut allowed them to detect products in the case that the top mass was around 100 GeV/c, just above what CDF had earlier established as a lower limit (ibid., 104–105). Finally, a comparison of two methods for optimizing signal-over-background levels did not clearly indicate either choice of cut as superior. The matter was a “judgment call,” as one physicist put it (ibid., 146). The SLT position prevailed in the end, and CDF kept the 2 GeV/c cut. Notice that the debate turns on a number of uncertainties. Besides the uncertainty inherent in most claims about motivation,8 the mass of the top quark and the optimization methods created two further uncertainties. The SLT debate thus provides a situation tailor-made for the model of rhetoric introduced in chapter 5—that is, an argumentative situation in which judgment is required. Staley’s analysis does not provide the rhetorical details that led to closure on this particular issue. However, we can see the typical result of a rhetorical situation: different CDF scientists judged differently. Some were persuaded by the SLT response, others had reservations but did not consider the question so crucial, and still others continued to have stronger doubts. In fact, these doubts seemed to be confirmed by a later experiment (run Ib), which suggested that the original SLT results were probably exaggerated, possibly as a result of bias in the choice of cuts. But even this conclusion is uncertain: the bias may have originated in the algorithm, or the group may simply have gotten an unusually high number of genuine signal events or background (Staley 2004a, chap. 5). We will see more rhetorical detail in the debate over the kinematic analysis below (5). What is particularly interesting about the SLT episode is this: it shows how statistical arguments can be inherently procedural. In one sense, this claim is rather straightforward, and would seem to hold for any experimental argument: one can assess the supporting evidence only if one trusts the experimental procedures that generated the evidence. I have a more provocative thesis in mind, however, at least for the natural sciences.
175 Argumentation at Fermilab
To get at the main point, let us first review the basic line of argument. The SLT analysis provided a subargument in the Evidence paper (Abe et al. 1994a, 2995ff.). As such, we might approach it as an argument product, a set of evidential reasons (the data) adduced in support of a conclusion (there is evidence for the top quark). One might sketch the core argument as follows, where steps 1 and 3 come from the null probability distribution: 1. If the observed collision events are free of top quark production, then we would expect to observe on average 3.1 ± 0.3 candidate events (based on soft lepton tagging at pT ≥ 2 GeV/c). 2. We observed seven candidate events under the conditions in step 1. 3. The probability of observing seven or more candidate events in a sample free of top quark production is 0.041, under the conditions in step 1. 4. Therefore, it is highly probable that the observed collision events are not free of top quark production. If one takes the evidential-relationship approach—essentially a logical perspective on the product—and focuses solely on the internal relations between premises and conclusion, then the above argument sketch appears rather strong, assuming the details can be suitably fleshed out. The SLT debate, however, shows that simply inspecting the product as a set of statements does not allow us accurately to assess the cogency or strength of this argument, that is, whether the premises support the conclusion and how strongly they support it. The CDF scientists wanted to know the procedure the SLT group followed, indeed the actual history of decision making, which led to the cut stated in step 1 above. In particular, they wanted to know the timing of the cut decision. The timing of the decision crucially determines whether the SLT analysis provides a sufficiently severe test for its results to count as evidence for step 4. Insofar as statistical testing procedures go beyond mathematical computation and include such rules as the no-peeking (or predesignation) rule, they serve dialectical aims that cannot be reduced to logical (mathematical) relationships. Predesignation is in effect a response to a possible objection, namely that of bias in selection of cuts. This rule is more than a helpful aid to the production of arguments that might be had otherwise. Compare predesignation with a rule of inclusivity. In some contexts, the latter rule aids cogent argumentproduction insofar as it makes it more likely that a writing group will hear all
176 Chapter 6
the relevant objections, but it remains possible that the group itself hits on those objections without others’ input. That is, the group could still come up with an argument content that we would consider logically cogent. The predesignation rule, by contrast, appears to be a necessary condition for the SLT analysis to count as an argument at all, or at least as an argument whose content merits one can reasonably appraise. As Staley (2002) has argued, if the scientist has peeked at the data before setting the cuts, then the experiment includes the scientist’s motives, whose effects remain probabilistically opaque. If one is unsure about when the cuts were set, one does not know what the content of the argument actually is, because one does not know whether the experiment includes the experimenter’s motives or not. Thus the statistical significance of the data cannot be assessed as they stand. Conversely, assessing such statistical arguments inherently depends on knowledge of the procedures used to produce it. But isn’t this true of all experimental arguments? Must we not know (or presuppose) that the experimental conditions and procedures are reliable before we may trust the evidence claims adduced in the argument, or even know exactly what is being claimed? Indeed. But the no-peeking rule, although it might be included under the general rubric of “experimental procedure,” differs from procedures concerned with the use of instruments and methods of data collection. The no-peeking rule directly governs the experimenter’s psychological states—what one is allowed to know and when. This makes sense inasmuch as the rule aims to exclude an objection that targets the experimenter’s psychology, namely, the charge of unconscious bias. From the standpoint of Habermas’s process ideals, bias represents a form of internal coercion. Thus the predesignation rule links the process ideal of uncoerced discourse directly with the content of evidential argument. One would expect such a link in the human and social sciences. The SLT debate shows how important it can be in the natural sciences as well, indeed to the point of making the logical cogency of a mathematical (statistical) argument inherently depend on procedural fidelity. (5) I turn now to a second debate at CDF, that concerning the kinematic analysis. In contrast to counting methods, in the various methods of kinematic analysis one selects individual events (based on cuts), attempts to reconstruct them as top decays, and then compares the probabilities or likelihoods of the two competing hypotheses, that is, the event involves top decay or, on the contrary, the event involves merely a background effect (Staley 2004a, 113–
177 Argumentation at Fermilab
119). The reconstruction relies on mathematical techniques, computer simulation, and physical theory. The particularly acrimonious controversy that arose early in the experiment provides a revealing glimpse of the rhetorical and social aspects of scientific controversy. I start with this earlier issue. The earlier controversy was provoked by a joint effort between CDF member Krys Sliwa and two outside scientists (Richard Dalitz and Gary Goldstein) whom Sliwa approached in view of their expertise in kinematic analysis (see Staley 2004a, 119–123). The outsiders were not well received at CDF from the start, when they made presentations at CDF meetings in February and April 1992. In May the kinematics group wrote up two notes for publication, the second of which used data from an earlier experimental run (1988–1989). The first was eventually published. When Sliwa asked that Dalitz and Goldstein be included as coauthors on the second paper (which was not published), the CDF executive board denied his request, citing a policy that only CDF members may have access to data and appear as authors. Dalitz and Goldstein had already been excluded from participating in internal CDF meetings. The science press caught wind of the decision and announced that Fermilab was suppressing a discovery (ibid., 120–121). A number of argumentative issues converged in this controversy. To begin with, many CDF scientists had doubts about the trustworthiness of kinematic analysis and whether it actually had provided evidence for the top. Within the kinematics group, judgments differed. Sliwa thought “the statistical significance of my very interesting results was yet to be determined,” whereas Goldstein “became overexcited thinking that the discovery had been made” (Sliwa, quoted in Staley 2004a, 120). These different reactions reveal the effect of temperament in assessing the cogency of a potential argument. Finally, perhaps most contentious were the charges that Sliwa had acted unethically in sharing data with outsiders. Sliwa denied he shared data, whereas Goldstein claimed to have had access. What concerns me here is not whether Sliwa actually shared data with outsiders, but what the charge reveals about the rhetoric and social norms of science, such as rules governing secrecy and credit. Notice first how three issues are intertwined: decisions about inclusion, the concern for credit, and the capacity for responsible judgment. Some CDF members excluded the outsiders, it seems, partly because of concerns over credit, and partly in view of the outsiders’ alleged lack of judgment: the outsiders were perceived as overly rash because they lacked the competence to assess the results. CDF leader Alvin Tollestrup was quoted as saying “[p]eople not
178 Chapter 6
sufficiently expert in interpreting the data have jumped to conclusions” (in New Scientist, June 27, 1992, cited in Staley 2004a, 121). According to the journal Science (July 24, 1992), Tollestrup justified the exclusion of Goldstein and Dalitz with the argument that they “haven’t put in the effort and commitment that others have” and thus lacked the necessary experience with the CDF system and its complexities (Staley 2004a, 121). On one possible reading of such remarks, some CDF scientists thought that the two outsiders lacked the depth of knowledge, including perhaps tacit background knowledge, required to judge the cogency of the kinematic argument. The perceived deficiency in the outsiders’ competence undermined their credibility (for the CDF audience) in assessing the findings as a “discovery.” From a rhetorical perspective, they had not established their character or ethos as quark scientists for the CDF audience, which is to say that they had not shown CDF that they were sufficiently practiced in the competences that constituted the CDF ethos. This suspicion seems to have reinforced, and been reinforced by, perceived deficiencies in emotional disposition, insofar as we can gather from remarks referring to Goldstein’s overexcitement and jumping to conclusions. The prospect of uncovering evidence for the last remaining unobserved particle in the standard model was no doubt a thrilling one—a “hot” cognition laden with emotion. But here pathos and ethos are connected: to display competence to the CDF audience, one also had to display the appropriate level of excitement. Thus Sliwa’s contrasting reaction—that the results were “very interesting”—was closer to a level considered appropriate at CDF. Notice, however, that such judgments of competence cut both ways, in a manner that depends in part on differences in expertise. Scientists familiar with kinematic analysis tended to trust its results, as the Italian members of the collaboration would later on when it came to writing up the Evidence paper. Those less familiar with the technique, with its complex mathematics and Monte Carlo computer simulations, remained cool to its evidential force. On each side, there seems to have been a suspicion that the other side “just doesn’t get it.” As late as 1996 Sliwa was still skeptical of the discovery claim based on counting methods. He remarked that although kinematic analysis requires one to understand “nothing more than some knowledge of relativistic kinematics, parton model geometry and basic particle physics, at times it seemed to be a barrier too high for many to even try” (Staley 2004a, 123). The decisions to exclude the outsiders, both from meetings and coauthorship, show the role that the ethos of a scientific subcommunity plays in bound-
179 Argumentation at Fermilab
ary-setting. To count as one of us, you must have been with us—you must have “put in the effort and commitment that others have,” thus acquiring the requisite skills and background knowledge. Failing that, you must at least persuade us you have such skills by displaying them publicly. However, I doubt that this rhetorical norm fully explains the exclusionary policy. We cannot disregard the other issue that lay behind the concern over data-sharing: credit. This was explicit in the New Scientist article (June 27, 1992), which concluded: “If Dalitz is correct, . . . the main credit for finding the particle will go to Dalitz, a scientist outside Fermilab” (quoted in Staley 2004a, 121). We already saw the concern for credit in phase (2) above, where I discussed the problem of presenting partial data at the Ithaca conference. The kinematics controversy illustrates the link between secrecy norms and credit discussed in chapter 5: the concern for credit affects the social standing of scientists, in particular who gets into a discussion and who does not, who may see data and who may not. Before moving on, a word on the eventual fortunes of the kinematic approach. As CDF turned to writing up the Evidence paper, kinematic results continued to be a bone of contention; indeed it nearly “split the organization open,” according to Tollestrup (Staley 2004a, 150). Scientists familiar with the method, such as collaborators from Italy, considered the analysis to yield evidence for the top, whereas others thought no mention of it should appear in the Evidence paper. In the end, the collaboration arrived at a compromise solution: include the results, but do not claim they provide evidence for the top. Evidence would come from the counting experiments. However, in presenting the kinematic results as consistent with the counting evidence, as a “check,” the authors provided a kind of further reason supporting the overall argument in the Evidence paper. They did not regard the results as genuine evidence because it remained possible to interpret the kinematic findings in three ways, only one of which involved top production (ibid., 150–154). However, the element of compromise raises further interesting questions for Habermas’s argumentation theory, particularly regarding the idea of consensus. I close this section with a few words on the kind of consensus that emerged over the Evidence paper. (6) As already mentioned, the writing process at CDF was oriented toward consensus.9 Publication required the general approval of the group, and any physicist unhappy with the final product was allowed to withhold his or her name from the list of authors. Besides the goal of critical testing, then,
180 Chapter 6
discussion had to be persuasive. Persuasiveness depended not only on the content of arguments but also on features of the discussion process itself, such as its openness, thoroughness, and the shared perception that people listened to one another. According to Staley (ibid., 169), “numerous CDF physicists” expressed greater confidence in the final position as a result of the discussion process. This stemmed in large measure from the manner in which the discussion publicly displayed a severe and thorough discursive testing of results. Physicist Mel Shocket, for example, considered the process itself as productive of confidence, “even if I hadn’t been paying attention to all of the details” (quoted in Staley 2004a, 169). Thus the discussion process served the rhetorical goal of persuasion and consensus building, in part because dialectical qualities of the procedures were perceived as fostering open and thorough critical discussion. In addition, particularistic rhetorical phenomena of the sort described in chapter 5, section 2, also played a significant role, particularly when scientists faced uncertainties in the arguments. In the kinematic debate, for example, persuasion depended not only on the arguments themselves but also on characteristics of individual participants—their ability to display competence, appropriate judgment, and the like. Individual differences in background training and personal style led to conflicting assessments of cogency. Consequently, to speak of the Evidence paper as a “consensus position” is somewhat misleading, if we mean “consensus” in Habermas’s strong sense. In fact, the group did not converge on its consensus position on the basis of the same reasons, as Habermas’s model requires. As we have seen, assessments of cogency differed with respect to the SLT and kinematic arguments. To a certain extent, then, the consensus position represents a compromise that the collaborators were willing to live with. According to Henry Frisch, the kinematic data was included “partly for political reasons, partly to mollify people. . . . It was nice that [the kinematics] was there, sort of, but mostly it was put there just to make people go away” (quoted in Staley 2004a, 153). Although many authors did not share Frisch’s view, it raises an interesting point for Habermas’s argumentation theory, which I take up in the next section. 4
The Evidence Paper and Habermas’s Argumentation Theory
Aspects of Habermas’s argumentation theory should be evident from the description in the previous section. In this final section I delineate some of the
181 Argumentation at Fermilab
main ways the case displays features of his theory as well as some important divergences. To frame the analysis suitably, I regard the Evidence paper as an argument product, whose logical quality depends on how well the design and actual conduct of the writing process realized dialogical (dialectical and rhetorical) standards of cogent argumentation. After mentioning aspects of the case that fit Habermas’s model (sec. 4.1), I turn to the phenomenon of compromise and the difficulty it poses for the model (sec. 4.2). A solution in line with Habermas’s model still presents an obstacle for interdisciplinary cooperation (sec. 4.3). 4.1
The Centrality of the Dialectical Perspective
Although Habermas takes into account a range of internally related perspectives on cogency, he gives priority to the dialectical perspective. One can see this priority above all in his emphasis on process-level idealizations, which set ideal conditions on the social-interactive dimension of argumentation so as to ensure maximally robust testing of hypotheses. The logical strength of the product depends largely on the quality of the critical tests it has survived. Some of these tests are experimental, but ultimately their results must hold up in critical discussions that are open to all competent participants and free of coercion. This process-level ideal thus provides the critical standard for designing specific institutional procedures that regulate the testing and discussion process. This is not to say that process standards provide a blueprint for institutional design, but that for any actual design we should have grounds for considering it adequate in light of process idealizations. One can understand much of the CDF writing process, both its design and execution, in the above terms. In attempting to meet the logical demands on the content of their argument, scientists at Fermilab showed intense concern over the quality and openness of the writing process, and they designed various procedural mechanisms to allay such concern. Among other things, these procedures opened up the discussion to any scientists at Fermilab and required the writers to answer questions that the audience posed to them. Charges of biased results had to be countered. The paper could not go forward until the group was satisfied with its adequacy. The various rules that structured the writing process were designed to produce a logically strong argument by submitting paper drafts to severe critical testing and revision until (nearly) everyone could agree on the content. In
182 Chapter 6
adopting such rules, the CDF scientists oriented themselves toward the central dialectical standard: that cogent arguments should stand up to critical challenges in open debate. As I argued above, the dialectical situation confronted the group with a number of specifically logical problems connected with the content of their argument. To develop an argument that would prove trustworthy and that sufficiently supported the conclusion, they had to make decisions about (a) what force to ascribe to the conclusion, that is, whether they should make a discovery claim or an evidence claim, or no claim at all; (b) what to include in support of the conclusion; (c) what degree of support to ascribe to the included reasons, that is, whether a particular analysis constituted “evidence” or simply a finding consistent with top quark production. However, they could not solve these logical problems apart from a dialectical assessment of the procedures employed in collecting and analyzing data. They had to ascertain whether the procedures and testing methods were invulnerable to charges of bias and other possible errors. Thus the CDF case shows particularly well how the dialectical and logical perspectives are inextricably intertwined: the cogency of an argument depends on both logical and dialectical merits. These various rules and procedures that guided the writing process constituted part of the actual context of dialogue. From a social-institutional perspective, one can see the CDF procedures as attempting to institutionalize, at least semiformally, the two sorts of ideal dialectical standards identified in chapter 5, that is, rules of exchange that regulate the course of discussion as a critical testing process (e.g., that charges of bias had to be countered) and social process ideals that call for an open, inclusive, and uncoerced discussion in which participants have equal voice. Formal procedures of debate often match ideal rules of exchange in their literal formulation, but not in the actual conduct of discussion. Institutionalized role differentiations, by contrast, generally do not match process ideals even in their literal formulation. Although CDF did have a formal rule of inclusiveness (i.e., an open invitation to all Fermilab scientists to contribute feedback on paper drafts), the collaboration also distinguished roles such as godparent, convener, and writer from the general (inclusive) audience, to which they assigned somewhat different rights and duties. One justification for such role differences lies in the recognition of the pervasiveness of bias. The process ideal calls for freedom from internal coercion, which includes the idea that participants have sufficient interior (psychological) freedom to consider opposing views and weigh the arguments solely on their merits. The unlikelihood that most participants in actual processes of argumentation fully satisfy this ideal leads in some situations to a division of labor
183 Argumentation at Fermilab
that is supposed to enhance the collective freedom and objectivity of the group— as we saw in the role of the CDF godparents, for example.10 In Part III, I take up some problems that arise with the attempt to connect Habermas’s process ideals with actual procedures. Here I focus on the critical question raised by his understanding of the social-institutional perspective: does the design and execution of the writing process warrant the presumption that the group accurately assessed the cogency of the Evidence paper? There is no doubt that the writing process involved some contentious moments. Moreover, some aspects elicited conflicting reactions: scientists differed, for example, on whether the godparents’ involvement in the drafting of the Evidence paper enhanced or undermined the objectivity of their judgment. As I noted in section 3(6), however, at the end of the day the dialogical features of the discussion enhanced the confidence that many (though not all) team members placed in the final product. Open procedures of discussion characterized by intense question-and-answer sessions, painstaking attention to detail, careful responses vetted for civility—all this publicly displayed both the dialectical robustness of the argument as well as the group’s professional capacity for rhetorically responsible judgment. Here I understand the display of professional competence and civility as examples of ethos and pathos that enhance the capacity of participants to make responsible judgments, as explained in chapter 5: the display of competence shows that one has the proper character to judge responsibly, and a civil emotional disposition helps participants to listen to each other and avoid unnecessary contentiousness. For members impressed by the discussion, the final product emerged as a dialogically and logically cogent product, precisely because the discussion process displayed virtues associated with process idealizations of openness and noncoercion, and with the rhetorical display of professional competence or ethos. In terms of Habermas’s model, these participants took the design and execution of the writing process as warranting the presumption that their discourse was dialogically adequate. To the extent that this rather upbeat interpretation of the writing process is accurate, the Evidence paper has an irreducibly social character in the sense that (a) each author freely shares in the collective acceptance of the paper’s argument in its entirety, but in such a way that (b) no single author has complete command of the various considerations and evaluations that entered into the construction of the argument. Rather, in accepting the paper, each scientist presupposes others’ greater familiarity with, and expertise in, the respective methods of detection and analysis. This presupposition is rhetorically grounded
184 Chapter 6
in the displays of competence and judgment that different subgroups exhibited in the dialectically rigorous discussion. Thus the cogency of the product depends on the distribution of expertise and roles, which the writing process coordinated in a dialogically adequate manner. Such an argument genuinely expresses an “intellectual solidarity”—an enterprise to which individuals organically contribute somewhat different, but complementary, skills and perspectives in producing a result in which all freely share.11 This first scenario is even more clearly illustrated in the later paper announcing the discovery of the top quark after Run Ib, when much of the uncertainty vanished for CDF physicists. The writing of the “Discovery paper” proceeded more amicably, with scientists converging on a more substantive form of consensus (Staley 2004a, chap. 5). That paper, then, seems to exemplify a product whose collective character rests on the distribution of expertise and roles working together in a genuine intellectual solidarity. With the Evidence paper, however, we have seen that the above scenario was not the whole story. In the kinematics controversy, negative perceptions of the outsiders’ character and level of pathos were reinforced by exclusionary rules connected with the concern with credit. As I argued in chapter 5, one might accept such rules as temporarily justified in light of credit concerns. In that case, the more interesting rhetorical problem for a Habermasian analysis of the Evidence paper arises when we examine the communicative failures within the collaboration, evidenced by the inability of members working on different methods fully to persuade each other that they were capable of responsible judgment. This failure to meet rhetorical demands led to a compromise, whose implications deserve examination. 4.2
Compromise and the Quality of Consensus
The rhetorical features of the CDF exchanges—both those in which speakers succeeded and those in which they failed—illustrate the rhetorical perspective described in chapter 5: they show how the backers of an argument had to demonstrate to the collaboration that their personal judgments of cogency issued from a competence and temperament that put them in the position to make responsible judgments. Difficulties for Habermas’s model emerge when we consider the thought that failures on this score generated the perception, at least among some participants, that the Evidence paper rested on a political compromise. How does the presence of such compromise affect the
185 Argumentation at Fermilab
Evidence paper as a collective judgment of cogency? Behind this question lie further questions of how scientific collaborations should handle internal disagreement. In fact, the substantive rhetoric described in chapter 5 does not guarantee a consensual outcome, even when participants conduct themselves responsibly from the standpoint of ethos and pathos, that is, even when the rhetorical features of their discourse in fact enhance their capacity for responsible judgment and are mutually perceived as doing so. If the evidence is uncertain and leaves room for reasonable differences in weighting or interpretation, then consensus might not be forthcoming even under rhetorically adequate circumstances. In that case, however, the nonconsensual outcome would still provide an accurate collective assessment of cogency. More precisely, the outcome provides a measure of a deficit in the cogency of the available arguments, such that no single position has the merits to establish itself as clearly superior to the rest. If we take our experience of the actual discourse as warranting the presumption that the process was dialogically adequate, then we can say that at least two of the competing views rest on arguments with more or less equal logical and dialogical merits. Although this second scenario lacks the substantive consensus described in the first scenario above (sec. 4.1), in each case the outcome provides a good measure of the cogency of the argument product. Suppose, however, that Frisch is correct in regarding the Evidence paper as a political compromise, and suppose this compromise was necessitated by the rhetorical failure to generate a mutual perception of responsible judgment. We then have a third scenario, in which scientists reached an apparent consensus, but under conditions that render the presumption of dialogical adequacy questionable. Different authors considered the evidential force of the argument to lie in different parts of the article, and some authors thought that the article could have had more force had other analyses been more fully developed. Moreover, some doubted the capacity of others to grasp the force of the evidence. Indeed, a few did not even accept the conclusion of the article they signed. Individual assessments of the merits of the paper’s argument did not converge in a neat way. Thus the consensus on the Evidence paper was shaky, held together by a strategic maneuver that goes beyond the dialogical aspects of the writing/discussion process. In signing on to the paper, CDF scientists did not respond simply to the merits of the analyses that were tested in critical discussion.
186 Chapter 6
Although such testing clearly played an important role, CDF members also sought to solve a strategic institutional problem: how to find a solution, a consensus position, that keeps the group together in spite of disagreement over the import of data. From a sociological standpoint, CDF had to solve the problem of maintaining social order in the group—in Latour’s language, they had to maintain their strategic “alliance of weaknesses.” For solving this problem, the collaboration had the help of social pressure. Although a dissenting scientist could withdraw from the list of authors, this move was socially costly. To withdraw one’s name was perceived as “a drastic step to be taken only as a last resort.” Socially, it weakened one’s status in the collaboration (Staley 2004a, 157). Pressure of this sort apparently violates Habermas’s dialectical process idealizations, specifically the requirement of freedom from coercion, thus rendering the outcome problematic. Frisch’s assessment thus suggests a third scenario. Whereas the first two scenarios warrant a presumption of dialogical adequacy, the third undermines that presumption on two grounds. First, members failed rhetorically, not because judgmental uncertainties led to disagreement (a feature of the second scenario) but because interlocutors failed to develop confidence in each other’s capacity for responsible judgment. Second, they failed dialectically insofar as the political pressure to compromise suppressed disagreement, thereby undermining dialogical freedom. In fact, a compromise consensus is problematic even when it does not issue from coercive social pressure. Consider a fourth scenario in which a team of scientists have (as in the second scenario) engaged in a dialogically adequate process but not reached complete consensus: no member is entirely satisfied with the argument, though they consider it highly unlikely that the conclusion will be disproved in the future. So they compromise in agreeing to publish the paper in view of some other, noncoercive consideration. An expert panel, for example, might regard any sign of disunity as confusing to readers or as undermining the public effect of their argument (as the case in chapter 8 will show). Because this motive is consistent with the communicative aims of argument-making, it does not seem to have the coercive aspects of the third scenario. Nonetheless, in both the third and fourth scenarios, the team’s public consensus hides important information necessary for readers to accurately assess the cogency of its argument. Understood in terms of either of these two scenarios, the Evidence paper misrepresents the dialogical merits of the argument,
187 Argumentation at Fermilab
which is presented as a consensual product.12 This short-circuits the social conception of cogency elaborated in chapter 5. That is, (CH) assumes that the level of consensus reflects the dialogical and logical merits of the argument; when this assumption holds, the “consensus opinion” among experts is the most reliable authority for assessing cogency. As the analysis of the Evidence paper shows, however, one cannot rest secure with public displays of agreement. Rather, such displays invite further critical questions. Not only must one have confidence that the process leading to the collective product was dialogically adequate; one must also have some awareness of the depth of consensus involved, and whether that consensus involved compromises. As I note below, this requirement poses a task for both scientists and science journalists. From the standpoint of Habermas’s argumentation theory, then, both the third and fourth scenarios are problematic. Indeed, when dialogically adequate process fails to produce full consensus, the second scenario, in which disagreement is public, is a better guide than the fourth for outsiders who want to assess the force of a scientific claim. Thus, for critical assessment using Habermas’s model, one must be able to distinguish these four scenarios from one another. In doing so, one would have to examine not only the dialogical adequacy of the process, but also the relation of that process to the collective outcome, whether the level of consensus tracks the merits of the arguments. Critical assessment would thus be guided by a framework defined by different scenarios, such as those summarized in table 6.1. In scrutinizing actual scientific discourses according to the above scenarios (which I do not claim are exhaustive), critical theorists move into the third mode of critique described in section 1: they venture prescriptive claims about the ways in which scientists publicly present their views. In fact, criticisms of this sort, often targeting overhasty and unqualified announcements, are not unusual (recall the ill-fated announcement of cold fusion by Fleischmann and Pons in 1989). But Habermas’s multidimensional argumentation theory grounds a farther-reaching call for reform in the current practices of publishing and reporting science. Regarding scientific collaborations and expert panels, the model would seem to call for some acknowledgment of internal disagreements. As we shall see in chapter 8, it is not unusual for panels to advertise the procedures that make their deliberations dialogically adequate. But they tend to be less forthcoming about points of disagreement. Regarding the Evidence paper, the analysis in this chapter might ground a critical question, modestly addressed to the paper authors: in signing on to the paper without
Outcome
Public and internal consensus on full content of argument.
Public and internal dissent on content.
Public consensus masks internal dissent on content.
Public consensus masks internal dissent on content.
Scenario
I
II
III
IV
Table 6.1
Yes, for insiders: process is perceived as dialectically robust and rhetorically responsible.
No, participants regard dialectical testing and/or rhetorical responsibility as deficient, but are pressured to compromise on a single content.
No: consensus hides compromises on content merits, for sake of communicating a clear conclusion.
No: consensus is based on political pressures.
Yes: none of the competing positions is decisively superior.
Yes: cogency of one position is decisively superior.
Yes, process is perceived as dialectically robust and rhetorically responsible, and compromises do not affect judgments of content merit. Yes, process is perceived as dialectically robust and rhetorically responsible, but it does not generate full agreement on content.
Collective Outcome Tracks Merits?
Process Warrants Presumption of Dialogical Adequacy?
188 Chapter 6
189 Argumentation at Fermilab
notice of the unresolved disagreements in judgment, have you misrepresented the merits of your argument?13 Regarding science journalism, the model calls for greater care in digging into the depth and dialogical quality of the positions reported in popular science venues. This is not simply a matter of emphasizing the tentativeness of research findings. Rather, science journalists have a critical responsibility to situate reports of substantive claims in illuminating descriptions of the process and social-institutional settings of scientific argumentation. A good negative example is the so-called balanced treatment of the global warming question by the mainstream media (in the 1990s), which failed to scrutinize the credentials and sponsoring of skeptics of the consensus reached by the Intergovernmental Panel on Climate Change. Habermas’s model supports a critique of journalistic practice on this topic.14 If the above critical suggestions resonate with scientists and journalists, then Habermas’s model has proven its critical worth. At the same time, we must admit that the negative status of compromise in the Habermasian model presents a problem, on at least two counts. First of all, a general preference for the second scenario over the fourth is overhasty, particularly at science–society boundaries. In legal-political contexts, the fourth scenario has the advantage of clarity, providing policymakers with scientific arguments that are more effective politically (see Jasanoff 1990, 98, 199, 209–215). If the causes of dissent among scientists are relatively minor, then in political contexts the fourth scenario might actually be preferable to the second, particularly when failure to act on the basis of available scientific evidence poses the greater potential risks. As the global warming question illustrates, special interests can find it politically useful to overemphasize disagreement in the science community for purposes of blocking policies and laws unfavorable to their agenda. This too involves a kind of misrepresentation of the dialogical merits of the arguments. More generally, when scientific arguments enter into policy contexts, the relevance of those arguments for decision making depends not only on the cogency of the arguments themselves but also on their relation to the comparative risks of action versus inaction.15 Consequently, judging the advantages and disadvantages of these two scenarios involves a number of complications. In judging between these two scenarios, recent technical findings on judgment aggregation might also offer some suggestive insights. This literature suggests that under some conditions, a compromise solution might actually be superior
190 Chapter 6
as an indicator of probable truth.16 I will not pursue these issues here, except to note that they point toward the contextual, domain-specific character of judging between scenarios two and four. To that extent, they accord with the general tenor of my argument in Part III. Second, if SSK theorists are correct in their view that compromise and political negotiation commonly occur in scientific collaborations, then Habermas’s model would find itself in a position not unlike that of logical empiricism: offering a picture of good science with significantly limited relation to actual practice. I address this question in the conclusion. 4.3
Conclusion
In this chapter I have attempted to show how Habermas’s argumentation theory, as I developed it in chapter 5, can be used to do real work in the critical assessment of an actual case, the writing of the Evidence paper at Fermilab. This case has served three critical aims. First, it clarifies the critical model, which in turn serves as a heuristic for understanding features of the case. The CDF case shows particularly well how scientists operate with ideas about robust critical discussion not unlike the dialectical standards sketched in chapter 5. Second, the CDF case has helped test Habermas’s model. If that model represents an attempt to articulate the presuppositions of participants, then one should expect it to fit the kinds of discursive and social moves the participants themselves seem to presume as normatively appropriate. Although one can see a significant level of fit between Habermas’s model and the scientists’ practice at Fermilab, the presence of compromise presents a challenge, more on which momentarily. Third, Habermas’s model generates some specific reform proposals, both in science and science journalism. I close by returning to the second aim: how does the Evidence paper call for a modification of Habermas’s model, particularly his negative view of compromise? On the one hand, the Fermilab scientists were not entirely happy with their compromise solution, an attitude that fits with the model. On the other hand, the compromise does not seem to have hurt research at Fermilab; their next run gave them enough evidence to announce the actual discovery of the top quark. So perhaps their compromise was closer to scenario four. In any case, we should expect compromise to occur fairly often in large collaborations. Although Habermas grants the ubiquity of compromise in political discourse, does he allow it to play any positive role in scientific consensus?
191 Argumentation at Fermilab
In reply, I suggest that from a social-institutional perspective, we may understand compromise as meeting demands of sustainable social order in cooperative practices of inquiry. Specifically, we might attempt to understand compromise in a manner similar to the analysis of credit presented in chapter 5: a concession to social-psychological reality for the sake of sustainable inquiry. The difference lies in scope. Credit helps make science sustainable over the long haul as a broad institutional endeavor. It does so by motivating self-interested people with a curiosity about nature to cooperate in an endeavor in which they accord one another social recognition for their individual contributions. Compromise within scientific collaborations serves a similar end, but at the local level: it helps members of a group to stay together. If one refuses to compromise and withholds one’s name from a group paper, then one distances oneself from the majority view, thereby placing oneself at odds with the group. Specifically, one signals to others that one considers one’s own judgment superior to the collective trend of judgments within the group. Similarly, if two subgroups within a collaboration refuse to compromise and accept the other’s work, then each impugns the other’s competence and judgment. Such a refusal seriously challenges the trust and mutual recognition on which cooperation in science depends. In dissenting to publish, one in effect challenges the scientific character and temperament of those who consider the argument publishable. As a historical process, science requires participants to trust and rely on one another not simply for the sake of pooling information sporadically or agreeing on when an argument is cogent, but for the sake of sustained inquiries that may require years of research cooperation. Nonetheless, pressure-based political compromises, even if socially necessary for sustainable order in science, clash with the dialogical ideal of freedom from coercion. In the case of secrecy norms connected with credit mechanisms, the process ideal is only temporarily suspended to protect the attribution of credit; this suspension turns out to be justified as epistemically fruitful in the long run. The conflict between compromise and noncoercion is less easily reconciled with Habermas’s model, particularly in short-run contexts in which nonscientists must rely on expert opinion for making practical decisions. When an expert panel presents a “consensus view” on the state of research in an area of health, or regarding climate change, important choices are at stake. Although consensus is sometimes helpful for political decision making, it might also mislead, particularly when it issues from social pressure.
192 Chapter 6
It seems to me that a Habermasian can accept the prevalence of compromise in the sciences as a condition of its ongoing social sustainability. The critical task would then require one to discriminate between the different forces that lead to the compromise, on the one hand, and the different contexts in which the compromise occurs, on the other. The details of such analyses require a separate study; here I merely note that one might draw not only on sociological studies but also on the technical work in judgment aggregation in distinguishing contexts. One would ask, to what extent do coercive social pressures drive the compromise and how damaging are they in the given context, given its epistemic and political features? If the participants see that their differences in judgment stem from temperamental differences (e.g., risk-takers versus risk-averse personalities), or from negotiable disciplinary differences in accepted confidence levels or standards of evidence, and if the particular context gives good grounds for compromise on individual judgments for the sake of communicating more clearly to nonscientists, then one may probably accept the compromise as reasonable. One might go considerable distance with SSK in this critical project, perhaps to the point of assuming that no untainted consensus actually exists. This assumption opens all consensus positions to sociological investigation. In the end, however, critics must still distinguish arguments on the basis of merits partly defined by dialogical idealizations. Even if we take the first and second scenarios as counterfactual idealizations, we must still ask whether actual processes of inquiry and discourse warrant the presumption that compromises and social pressures, though present, have not seriously undermined the dialogical merits of a given outcome à la scenario III. To that extent, a kind of asymmetry between the scenarios remains. SSK theorists might object at this point that they will not be enlisted once more in a sociology of error—in this case, research that would relegate them to uncovering compromises that explain questionable scientific claims. Habermas’s critical project, they might object, asks them to identify consensus positions whose content has been infected with politics, which are then taken as rationally defective in comparison to “dialogically adequate” positions. This contrast has all the markings of a contrast between “good” and “bad” science. But that distinction violates the symmetry postulate at the core of SSK. If we assign SSK the job of identifying the bad science, then have we not pigeonholed it as a sociology of error? That move would hardly foster interdisciplinary receptivity.
193 Argumentation at Fermilab
The asymmetry between dialogically adequate and questionable consensus cannot entirely evaporate if Habermas’s project of critical assessment is to go on. Even if no actual consensus is compromise-free, the critique of such compromises must be oriented by the ideal scenarios in a manner useful for making further judgments—about fruitful directions of future research, wise policy, health care, and the like. Does this critical commitment pose an obstacle to interdisciplinary engagement with SSK, particularly those variants committed to relativism? If it does, then we have an interesting hard case for such cooperation, which I take up in the postscript to Part II.
Postscript II Who’s Afraid of SSK? The Problem and Possibilities of Interdisciplinary Cooperation
In the last three chapters I argued that Habermas’s argumentation theory implies a social conception of the cogency of scientific arguments. As such, his model of cogency goes a long way toward bridging Kuhn’s Gap. Conceptually, the model integrates logical, dialogical, and social-institutional perspectives on scientific argumentation. Thus the merits that constitute cogency include the dialogical adequacy of the social processes from which logically strong arguments emerge as products. To evaluate scientific arguments, then, one must examine not only their content but also the social-institutional process. The latter task calls for an interdisciplinary approach that draws on sociological analysis. To that extent, Habermas’s conceptual integration offers a framework within which one might overcome the disciplinary oppositions that arose in the wake of Kuhn’s Gap. However, difficulties that emerged in chapters 5 and 6 indicate that Habermas’s highly idealized approach still presents certain obstacles to full interdisciplinary cooperation. As I read Habermas, the strength of an argument as a justification of a truth claim ultimately depends on counterfactual idealizations. Features of the actual process at most warrant a defeasible presumption that scientific practice has not obviously violated such idealizations, thus rendering the process dialogically adequate. In a certain sense, then, dialogical adequacy is defined by intersubjective conditions that lie beyond the empirically observable world. As participants in discourse, we cannot observe such conditions but can at most presume their satisfaction insofar as what we do observe does not defeat the presumption. Habermas’s dialogical model thus requires critics to distinguish different scenarios of consensus formation in science, as I argued
196 Postscript II
in chapter 6. In doing so, they draw upon the sociology of science, but in a manner reminiscent of the division of labor that SSK theorists vehemently rejected in principle. The burden of this postscript is to examine the possibility of fruitful cooperation between Habermasian critical theorists and sociologists of science, their deep philosophical differences notwithstanding. Here it helps to frame the philosophical issue as sharply as possible, as a severe test for interdisciplinary exchange and cooperation. I thus examine the prospects for a particularly improbable collaboration, namely that between Habermas’s reason-centered critical theory and the relativistic SSK associated with the “Strong Program in the Sociology of Scientific Knowledge” developed by David Bloor, Barry Barnes, and others of the so-called Edinburgh School. These two approaches often appear to oppose one another. Habermas’s project is rationalist, insisting that universal dialogical standards, appropriately understood, are indispensable to the critical project. Strong Programmers seem to reject universal standards of reason in favor of a principled relativism.1 My question is whether a collaborative “critical science studies” (CSS) is possible involving these two approaches. I argue that it is—but that the moves such collaboration requires point to a deeper problem in Habermas’s approach. In what follows I use terms like “cooperate” and “collaborate” in a broad sense that covers a range of possible modes of cross-fertilization and exchange. At one end of the spectrum are cases in which scholars on both sides work together on joint projects: papers, conference panels, and the like. At the other end are cases in which a scholar working in one area draws on research conducted in another area; here exchange is limited to the unilateral assimilation of the other’s published work. Obstacles notwithstanding, certain similarities between the Strong Program and critical social theory in the Frankfurt School tradition point to potentially fruitful collaborations across this spectrum.2 Both approaches are known to examine the conditioning of science by social interests and values. Hence we have good reason for seeking some way of ameliorating their opposed views on the rationality of scientific inquiry. To be sure, SSK theorists have advocated different types of relativism, which critics do not always clearly distinguish, thus provoking charges of misinterpretation. However, as I read theorists like Barnes and Bloor, they insist on a kind of skepticism toward rational justification, or what I call “justificational atheism.” This view puts the Strong Program directly at odds with Habermas. If I thereby misread the Strong Program, so much the better—that
197 Who’s Afraid of SSK?
only makes my test more realistic, given the prevalence of mutual incomprehension between different camps in science studies. An oppositional misreading also ensures a more severe test, with far-reaching consequences: if Habermas’s argumentation theory can allow rationalists to cooperate with SSK relativists, then there are good prospects for a still broader interdisciplinary model involving sociologists, historians, anthropologists, feminists, and others with diverse philosophical commitments. I proceed as follows. After pinpointing the philosophical challenge for CSS posed by “justificational atheism” (sec. 1), I describe the key maneuver (sec. 2). This move opens the door to an argumentation-theoretic approach that employs a dialectical standard of immanent critique that draws on SSK analyses (sec. 3). I then illustrate the use of this standard in a particular case of controversy (sec. 4) and draw a general conclusion that points toward Part III (sec. 5). 1
Philosophical Oppositions: Reason Pure and Impure
The central obstacle to cooperation between Habermasian critical theorists and SSK arises from their apparently incompatible views of argumentation. As we have seen, for critical theorists like Habermas, argumentative engagement involves certain presuppositions, in particular the idea that reasons can have a justificatory force measured by counterfactual idealizations. If we consider ourselves rationally accountable subjects, then we must suppose, in the absence of evidence to the contrary, that our position-taking in discourse is motivated solely by the “force of the better argument alone” (Habermas 1993, 50, also 41–57). To be sure, Habermas’s conception of cogency does not reduce that force to the intrinsic or impersonal properties of the argument as a product. Rather, the idea of responding solely to the better argument has a dialogical character, referring to a rationally transparent, hence intrinsically reasonable, process. Although the dialogical force of reason never operates in a pure form, apart from various psychological, institutional, and sociocultural influences, we must assume that such a force can have a real effect on our actual discourse, with all its impurities. For purposes of this postscript, then, I assume that the Habermasian model commits us to the regulative idea of an intrinsically reasonable dialogue: a hypothetically reasonable dialogue untainted by any motive or influence that would detract from the rational construction and evaluation of arguments on
198 Postscript II
their merits (relative to the available epistemic resources). Precisely this rationalist commitment grounds the effort to discriminate between the different outcome scenarios described in chapter 6. More generally, Habermas’s project of critical assessment requires us to distinguish scientific beliefs and practices that are rationally well founded from those that are not. This project allows for a kind of agnosticism when it comes to the critical evaluation of any particular case of scientific argumentation: a studied suspicion that the outcome has been less than dialogically adequate. In examining a case of scientific argumentation or consensus, we should suspect that ideology, political pressures, or other social interests rather than reason have determined the outcome. But this agnosticism must remain open in principle to the possibility that in some cases the weight of the evidence does, or did, eventually decide the issue—as a number of historical case studies have attempted to demonstrate (Kim 1994; Galison 1987; Rudwick 1985). One thus takes a case-by-case approach, allowing that scientific controversies can terminate in a number of different ways (see McMullin 1987). A critical program committed to the intrinsically dialogical force of argumentation, even if only as a counterfactual idealization, involves an asymmetrical approach to the study of science, as I pointed out in chapter 6. Such an approach recalls the distinction between “internal” and “external histories” of science, or “epistemic” and “social” factors—just the distinction that SSK relativists have consistently rejected (Barnes and Bloor 1982; Bloor 1984; Barnes, Bloor, and Henry 1996, 25–28). In sharp opposition to asymmetrical approaches, SSK relativists adopt a form of skepticism vis-à-vis argumentative justification. In other words, they take the view of “justificational atheists”: unlike believers and open-minded agnostics, they emphatically deny that arguments could ever have an intrinsic force of their own, a force that could be distinguished from social context. Not that atheists deny the perceived force of reasons and dialogue for scientists. Rather, they deny that these have this force of themselves, “intrinsically.” Whatever argumentative force reasons, evidence, and dialogue have is borrowed from, and varies with, contingent social circumstances. This distinguishes SSK not only from Mertonian sociology of science, but also from Habermas’s approach. Unlike Habermas, SSK atheists do not view socialinstitutional conditions simply as helping or hindering scientists’ attempts to arrive at positions that match the hypothetical outcome of an intrinsically reasonable process of argumentation; rather, social contingencies are
199 Who’s Afraid of SSK?
ultimately decisive for the very substance of science, such that the same question can receive different, equally acceptable answers under different social circumstances. Similar to Rorty, they reject the attempt to legitimate science by grounding it in universal, intrinsically dialogical—hence nonrelativistic— standards of reason. Moreover, in saying that social conditions are “ultimately decisive,” atheists need not claim these are the sole determinant of outcomes or that science is unconstrained by nature. Rather, the phrase gains its sense from a context of explanation in which one wants to understand how, given the underdetermination of theories by evidence, scientists reach their conclusions. In posing the conflict between critical theory and SSK as I have, I put aside a number of related controversies.3 The important point is that this opposition rests on incompatible philosophical commitments. Whether one accepts or rejects the idea of intrinsic rationality, one adopts a sweeping epistemological assumption about the nature of scientific inquiry. For SSK relativists, accepting this idea threatens the project of explaining the content of science in sociological terms. It is thus not surprising that attempts at reconciliation by a division of labor (e.g., causal explanation of beliefs versus their validation) have failed to satisfy (Barnes and Bloor 1982; cf. Gutting 1984). For critical theorists, the skeptical rejection of normativity creates problems for critique. Although a skeptical attitude may be one we can live with as academics researching the history of science, it becomes problematic in the context of current policy debates in which scientific claims and multicultural perspectives intertwine and conflict. For CSS, an overly strong skepticism or relativism regarding argumentation would seem to undermine the ability to use science for critical purposes in the evaluation of competing social and political agendas. Not only Frankfurt School critical theorists (Marcuse 1990; McCarthy 1988; Bohman 1991) but also feminists have noted the problematic character of such relativism vis-à-vis emancipatory goals (Wylie 1996, 320–323). 2
Disconnecting Epistemology: A Method of Avoidance
As a cooperative enterprise between Habermasian critical theorists and SSK relativists, CSS appears doomed. How can believers in the intrinsically critical force of reason make common cause with justificational atheists? In reply, I propose a lateral move: adopting the epistemological equivalent of John Rawls’s method of avoidance (Rawls 1996). The idea is that an
200 Postscript II
interdisciplinary CSS should try to avoid the intractable philosophical questions raised by the rationality debates. That is, in CSS we take no position at all on the deeper epistemological question of whether good arguments have intrinsically (and universally valid) dialogical merits or, on the contrary, are always socioculturally perspectival. As argumentation theory, CSS has us scrutinize actual processes of argumentation. The influence of social factors becomes important only insofar as we can identify such forces and how they influence a given controversy and its termination. On the one hand, then, we must take argumentation seriously and acknowledge the force of arguments as we—and the participants—perceive it. On the other hand, in pursuing CSS we stop short of attributing an intrinsic, or universally binding, force to dialogically adequate argumentation. Thus, case studies that fail to uncover substantively influential social factors (e.g., Rudwick 1985, 438–445) can leave the believer’s faith untouched without forcing atheists to affirm an intrinsic rationality beyond the perceived normative force. Conversely, if an analysis identifies a determining social factor, this need not always undermine the normative quality of argumentation, as I argue below. Note that epistemological avoidance commits us neither to a general agnosticism about reason nor to a blanket dismissal of epistemology—just as Rawls’s method of avoidance does not require citizens to give up their religious convictions or to renounce theological discussion (in appropriate settings). Rather, we temporarily put aside epistemological questions for the purposes of interdisciplinary critique. This project assumes, on the one hand, that an SSK study of a given case can proceed without invoking a sweeping atheism and, on the other, that critical appraisal can appeal to less ambitious standards than the idealizations articulated by Habermas and others. In the next section I describe such a standard by drawing on dialectical theory. 3
Dialectic as an Integrating Framework
From the standpoint of argumentation theory, we may take SSK as primarily aiming at the empirical analysis of the social-institutional dimension of scientific argumentation. If such analyses are to play a role in a normative CSS based on Habermas’s argumentation theory, then we must show how they fit with his dialectical analysis of scientific argumentation. However, the use of dialectic must be sufficiently thin, so as to avoid the philosophical conflict described above. Both sides, I assume, can accept the idea of dialectical
201 Who’s Afraid of SSK?
argumentation as a social process in which the participants assess the relative plausibility of competing claims. Within that broad framework, two aspects of dialectic are central for CSS. The first is the presumptive character of dialectical argumentation (Walton 1992; Rescher 1976b, 1977). Presumption arises from the wide variety of background assumptions in scientific argumentation—regarding auxiliary theories, the instrumentation and experimental methods that generate evidence, the domain of objects under investigation, broad pictures of what the universe is like, and so on. Such assumptions are presumed to be true so long as specific grounds for doubt do not arise (see Shapere 1984, 222–227). Among these assumptions one is likely to find cultural values and biases, which can determine what counts as evidence for a hypothesis (Longino 1990). The second aspect is a standard of relevance immanently linked to the goals of scientific inquiry. Science studies theorists and argumentation theorists have employed the idea of relevance in various ways.4 The standard of relevance I employ here assumes that the following presumption governs scientific argumentation: in reaching a consensus in science, the participants must presume that none of the considerations or “factors” influencing or motivating their consensus would, if brought to light, undermine their confidence in the consensus, given the aims of inquiry. This presumption constitutes a less idealized, context-sensitive version of Habermas’s counterfactual process standards. We presume, not that we’ve approximated a dialogical ideal, but that our consensus is based on relevant motives and not undermined by irrelevant ones. We can thus define an influencing factor as “unscientific” if it is irrelevant to achieving the aims of scientific investigation, as the participants understand those aims. Thus, if the unscientific factor were to be expressed as an explicit reason for accepting a given scientific claim as worthy of consensus, that reason—even if taken as true—would fail to convince the participants in the controversy. Conversely, the participants in a consensus presume that their background assumptions or other influencing factors are “scientific”: for any such determinant of the consensus, they presume that, were it to be expressed explicitly, it would not undermine their confidence in the consensus. Four comments on this model are in order here. First, irrelevance should be distinguished from the undermining effect that arises from simply discovering that a particular background or auxiliary assumption is false. As Achinstein’s theory of evidence shows, this kind of criticism plays a central role in science as a practice of ongoing inquiry. Uncovering empirically false
202 Postscript II
assumptions also provides an avenue of critique in feminist philosophy of science (see, e.g., Wylie 1996, 328ff.). But irrelevance can be distinguished from straightforward falsehood by asking how our confidence in a consensus would fare if we regard all the factors-made-explicit-as-reasons as true. Second, the relevance standard does not impose a particular goal on science, but rather allows the participants themselves to define their goals. Hence one cannot automatically preclude metaphysical or theological factors as “external.” For example, insofar as scientists of Newton’s day accepted the theological aspects of his theory of motion, these aspects must be considered “scientific” relative to the consensus at that time. The same holds for sociological factors: as I shall argue below, it is possible that some social interests and cultural values—and not simply such standard epistemic values as simplicity and consistency—may pass muster as “scientific.” This last point should not surprise us, inasmuch as relevance—and thus the perceived normative force of arguments—is defined here in relation to the participants’ understanding of the goals of their science. Nor should we be surprised to find scientists divided over the goals of science: as “essentially contestable,” such goals are an appropriate topic of normative discourse (Baumslag 1998; Gallie 1955/1956). Indeed, the goals of inquiry are not pursued in isolation from other, possibly competing, social goals—rather, they are socially situated, pursuable only insofar as other goals and interests allow. The third point concerns the attitude of the CSS practitioner. In approaching a case of controversy or consensus formation in science, the critical argumentation theorist adopts the perspective of participants reflecting on their scientific beliefs or consensus. These beliefs, I assume, are the object either of a general consensus or, in the case of an ongoing controversy, the object of consensus for some portion of the scientific community.5 The critical analysis then proceeds as an attempt to render a case of scientific practice explicit: to raise to awareness the consensus position and the various factors that constitute its motivational basis. In making their practice explicit, reflective participants operate with the sociologically informed suspicion that, given the inconclusiveness of empirical evidence, “external” social factors may be significant, perhaps even decisive, determinants of the consensus. Suppose, now, that an SSK study confirms this suspicion. Are these social factors “unscientific”? To answer this, we must state these factors as explicit reasons for the consensus and then ask ourselves if our confidence in the consensus is thereby undermined.
203 Who’s Afraid of SSK?
Fourth, the last two points reconfigure the relation between critical evaluation and empirical SSK. That is, the model does not depend on an a priori division of labor in which sociologists are relegated to the discovery of error or irrationality in science. Here one should not be misled by my distinction between scientific and unscientific factors. Although the distinction is necessary for critical purposes, it does not predetermine the boundary between good and bad science in an a priori manner. Rather, the model ties these two designations to the participants themselves. To that extent, it promises to deliver a greater context-sensitivity than a model that begins the study of cases with a specific understanding of what counts as relevant in scientific argumentation. Moreover, it leaves open to sociological investigation the full range of scientific argumentation, both good and bad. Before moving on, a few words on a proposal similar to mine. Solomon (2001) has coined a useful term for the various factors that motivate theory choice in science: “decision vectors.” As she understands it, “decision vector” is an epistemically neutral term, covering the full range of factors that can influence the outcome of theory choice in science. Some contribute to successful science, whereas others potentially undermine it. So far, her decision vectors align with what I have referred to as “influencing factors” or “considerations.” However, Solomon understands scientific success primarily as empirical success, which is contingent on the world in a way that theoretical success is not (ibid., 16–17). In this her model differs from mine: she adopts a particular conception of relevance, predicated on the empirical goals of scientific research. Although we may assume empirical goals usually dominate in the natural sciences, the interdisciplinary model I propose does not exclude other sorts of goals. Moreover, the thrust of her critical analysis of consensus and dissent has a somewhat different target, namely the rational distribution of theoretical commitments and agenda in a given field of research. As my case study below will make clear, I am more concerned with making presuppositions and influencing conditions explicit for purposes of critically assessing particular scientific arguments that have gained some level of acceptance among a group of scientists. 4
A Case from Behavioral Biology
To illustrate the critical procedure, I draw upon Helen Longino’s analysis of two competing explanatory models in behavioral biology (1990, chap. 7). The
204 Postscript II
“linear-hormonal” (LH) model accounts for gender-associated behavioral patterns as the result of prenatal hormonal exposures. The model is “linear” in the sense that the dependent variable (certain types of behavior) “is the straightforward outcome of a unidirectional and irreversible sequence of (biochemical) events” (ibid., 135). The competing “group selectionist” (GS) model, as Longino develops it, would explain such behaviors as the result of experience-induced alterations in neural connectivities. The GS model links behavior not so much with deterministic embryonic history but with ongoing changes in brain organization that respond to the organism’s life experience and actions. According to Longino, the empirical evidence for either of the two models is inconclusive; indeed, this has to be the case inasmuch as both models depend on metaphysical assumptions regarding autonomous agency, causality, and the like. What is more, both models involve sociopolitical values related to gender (ibid., chap. 8). Specifically, the LH model assumes traditional gender stereotypes—more broadly, it assumes male–female dimorphism as biologically fixed—whereas the GS model places more importance on autonomous self-development. As Longino explains, the GS model “returns both autonomy and responsibility to the person” (ibid., 176). These different assumptions about personhood in turn have implications for social policymaking. For example, the LH model would tend to undermine policy initiatives to encourage motivated girls to study science—in general, the LH model would favor programs that maintain traditional gender roles (ibid., 165–167). Now suppose, for the sake of illustrating a CSS critique, that a subgroup advocates the LH model as theoretically superior to the GS model. Suppose further that an SSK study manages to establish that sociopolitical interests are decisive for consensus formation within this subgroup. One might then identify two broad factors, B1 and B2, as the (actual) motivating basis for the subgroup’s commitment to the LH-model; this yields the following argument-like structure of basis-supporting-consensus: Basis (B1) Plausible but inconclusive arguments can be offered in support of the LH model of behavior. (Conversely, plausible but inconclusive arguments support the GS model.) (B2) The LH model supports the sociopolitical interest in the maintenance of traditional gender roles.
205 Who’s Afraid of SSK?
Consensus position (LH1) The LH model is worthier than the GS model of our pursuit and provisional acceptance as theoretically more fruitful, that is, as the more accurate representation of nature. Note that the consensus position (LH1) is construed as a claim to theoretical (i.e., epistemic) superiority. At the same time, this claim also involves a recommendation of pursuit. The formulation thus accepts the idea that the epistemic assessment of a theory or research program looks to an “economy of research” (Rescher 1976a) in which “heuristic appraisal” (Nickles 1989) is intertwined with the practical demands of conducting an ongoing research program. The two factors, B1 and B2, are taken as the actually necessary determinants of the subgroup’s closure on the LH model. The first factor simply summarizes the status of the available evidence in the eyes of the participants—that is, they recognize that the evidence is not compelling for either model. The second factor, however, presents a potentially problematic social influence. Given the consensus position as formulated, B2 is arguably irrelevant as an explicit supporting reason and should undermine our confidence in the consensus. We have, then, a case of dialectical critique that targets a particular background assumption or social factor as unscientific. Moreover, rather than being eviscerated by SSK, this critique positively depends on a successful SSK explanation. The situation changes if the advocates of the model acknowledge the valueladen character of the goals of science and reformulate the subgroup consensus position to read: (LH2) Compared to the GS model, the LH model is worthier of our pursuit and provisional acceptance as sociopolitically more fruitful (and as possibly the more accurate representation of nature). LH2 posits both truth and sociopolitical merit as the complex goal of science. This formulation of the subgroup consensus position allows both B1 and B2 to pass muster as scientific. By positing a different goal, LH2 in turn changes the conditions of relevance. In fact, this shift in the understanding of science is precisely what Longino herself advocates—but from a feminist perspective critical of gender stereotypes (Longino 1997). Because LH2 supports such stereotypes, feminists should reject it in favor of the GS model as sociopolitically more fruitful—that is, more in the service of emancipatory goals and political
206 Postscript II
autonomy. Debate between the two models would thus center on the competing political agenda, and critique of the LH model would take a sociopolitical form, carried out, presumably, in the context of deliberative politics. The foregoing scenario, as I have analyzed it, assumes that both sides in the controversy are committed to the value of empirical adequacy. The space for fruitful sociopolitical argument opens up precisely because each camp acknowledges the inconclusiveness of its own evidence (see Longino 1997, 30). This mutual recognition may rest on straightforward empirical considerations, for example, the skimpiness or conflicting character of the data. But even in that case, the analysis does not commit us to a view of empirical evidence as untainted by social conditioning—if anything, Longino’s “contextual empiricism” suggests just the opposite, that science rarely if ever enjoys value-free evidence (Longino 1990, chap. 3). But neither does the CSS program force us to assume that scientific content is always political, all the way down. Again, CSS theorists take no position on this epistemological point but study each case for its specific modes of social influence. What matters for the analysis of the LH-versus-GS controversy, as depicted, is that both sides recognize the empirical underdetermination. But what happens if the conclusiveness of the evidence is disputed? Suppose that one party considers the empirical evidence in favor of its model to be conclusive for any competent or reasonable scientist, whereas the other party considers the arguments inconclusive for either position. Both sides may be committed to a canon of empirical adequacy, but their views of what counts as evidence may diverge too sharply for them to consider the other side’s position as reasonable. This scenario, I suspect, poses one of the most difficult types of controversy to resolve—a conflict that may well prove intractable. To some extent, the controversy generated by creationism and intelligent design theory displays such intractability. To make room for their biblically based view of creation or theism, creationists and intelligent design theorists have argued that the evidence for evolutionary biology is inconclusive and contestable—just the point that mainstream scientists reject (see Kitcher 1982). These arguments have generally failed to convince those most familiar with evolutionary science. This controversy, however, harbors deep complications. For one, it is not clear whether antievolutionists accept empirical adequacy as the primary goal of science. In the end, the debate turns on conflicting visions of modernity, society, and education (see the discussion at the end of chapter 4; for similar arguments in the biology of behavior, see Longino 1990, 178). Such conflicts
207 Who’s Afraid of SSK?
reflect a deeper social rift, a clash of worldviews that manifests itself in disagreement not only over the values that should govern science and social policies, but also over who counts as a competent member of the knowledge-making community. In such cases, critique is forced to move to the sociopolitical level, where it must rely on the standards of a deliberative politics. One final scenario deserves mention. So far I have assumed that the SSK analysis is successful—that participants accept its results as conclusive or as plausible enough to stimulate critical reflection on the science at issue. Moreover, I have also assumed that, as an immanent mode of critique, CSS does not impose its values on the participants. In fact, it is more likely that the SSK conclusion will be less than conclusive, which suggests that opponents can apply a dialectical critique to the CSS critique, thereby undermining the attempt to undermine an existing scientific consensus. If the CSS critique is itself based on an inconclusive SSK analysis, then the value assumptions of CSS become an issue, and the effectiveness of the critique is in doubt. SSK analyses, as I have presented them here, depend crucially on showing that the evidence for some scientific conclusion is not conclusive. Precisely this inconclusiveness opens the door to sociological factors. If this opening move is itself disputable in a given case, then a plausible argument can be made that the evidence for a consensus in science is indeed conclusive. Thus the attempt to apply the dialectical critique to itself will in many cases boil down to a dispute over the status of the evidence for the primary scientific conclusion at issue—the kind of deep controversy exemplified in the creationist debate.6 Even when the SSK analysis is disputed, however, critique will have succeeded in bringing the potentially ideological character of a particular scientific result into the open where it can become an object of public scrutiny. 5
Conclusion: The Problem and Possibilities of Cooperation
As the subtitle declares, this postscript has addressed “the problem and possibilities of interdisciplinary cooperation” informed by Habermas’s approach to cogency. The problem lay in the barrier separating Habermas’s ideally rational conception of dialogue from relativistic strands of science studies. I hope to have shown that possibilities of cooperation nonetheless exist: one can identify at least some areas in which Habermasian critical theorists and SSK atheists, rationalists and relativists, can join in interdisciplinary cooperation. More
208 Postscript II
specifically, a CSS that combines Habermas and SSK envisions at least three types of critique that both use and target science. The first I mentioned only briefly: a criticism of background assumptions as empirically false. Here one draws on the results of a particular science to criticize claims within the same (or a related) discipline. The second form of critique employs an immanent standard of relevance to expose ideological commitments that, once exposed, help undermine consensus. This form of critique too relies on science, in particular the sociological findings that establish the presence of ideological interests as steering a particular research program from behind. Finally, a third form of critique targets acknowledged social and political agenda that shape science and its social implementation. Inasmuch as empirical adequacy remains a consideration in this third form, the science must be taken seriously. But the resolution of conflicts at this level points beyond the analysis provided here to the kind of criticisms that inform deliberative politics. I claim neither that these three types exhaust the field, nor that other possible types of critique will always be able to avoid deeper epistemological issues. But these possibilities show that the problem is not insurmountable in principle. Notice, however, the crucial move such cooperation requires: one must set aside general philosophical commitments and focus on the close scrutiny of specific cases. This requirement gives rise to a suspicion: that the problems I identified in Habermas’s approach in the preceding chapters and in this postscript stem from the particular manner in which he integrates argumentationtheoretic perspectives in his multidimensional conception of cogency. Habermas attempts to integrate the traditional perspectives on argumentation—logic, dialectic, and rhetoric—in a single philosophical theory of cogency. As I argued in chapter 5, his integration rests on a series of internal relationships that tie these perspectives together. However, the normative force of those relationships ultimately depends on counterfactual dialogical idealizations. The latter begin to take on a metaphysical status in Habermas’s conception—no doubt very much against his pragmatic intentions. The metaphysical tendency in Habermas’s idealized conception of cogency limits its usefulness as a boundary concept that could foster interdisciplinary cooperation. These limitations in interdisciplinary potential are similar to those that Klein (1996, 11–15) connects with holist models of integration. As I explained in the introduction, holistic approaches make overly strong claims to unify different perspectives as held together by a single overarching, often covertly metaphysical scheme. Such models cannot endure
209 Who’s Afraid of SSK?
the strains of disciplinary and perspectival differences. To recover the interdisciplinary potential of Habermas’s approach, therefore, the critical theorist must relax the general philosophical claim about the ideal grounds of cogency and rely instead on participants’ judgments—what scientists perceive as cogent in the specific context at issue. In this move one can see the first hints of a critical contextualism that radically repositions the framework of critical assessment.
III Toward a Critical Contextualist Framework for Interdisciplinary Assessment
7 Adjusting the Pragmatic Turn: Lessons from Ethnomethodology
My argument up to now has attempted to demonstrate the relevance of argumentation theory for the interdisciplinary study of the sciences as embodied social practices. In Part I, I employed categories from argumentation theory to illuminate developments in science studies. There I described how theorists tend to approach the study of scientific argumentation using different analytic perspectives, which commit them to different conceptions of the cogency of scientific arguments. These differences pose a challenge for interdisciplinary cooperation, namely to bridge the gap that Kuhn created (perhaps inadvertently) between logical and social-institutional perspectives on cogency. In Part II, I explored some possibilities opened up by Habermas’s multidimensional model of cogent argumentation. I thus read Habermas as offering an argumentation theory that attempts to overcome Kuhn’s Gap at a conceptual level, integrating logical, dialogical, and social-institutional perspectives in a coherent conception of cogency. Such a model calls for interdisciplinary cooperation among philosophers, rhetorical scholars, and sociologists (among others) in the evaluation of scientific claims. The strengths of his model notwithstanding, Habermas’s emphasis on highly idealized dialogical standards also creates a serious problem for contextualization and apparently hinders collaboration with SSK theorists, particularly those committed to some form of relativism. As I argued in postscript II, this obstacle is not entirely insurmountable, provided that scholars of different persuasions leave their incompatible philosophical commitments at the laboratory door—a solution that in effect gives priority to context over general theory. The necessity of this move exposes a deeper problem in his conception
214 Chapter 7
of cogency: its philosophical ambitions inhibit its ability to function as an effective interdisciplinary boundary concept—a limitation that bears some resemblance to problematic forms of interdisciplinary holism. To extend the interdisciplinary potential of his model, one must set aside the commitment to a philosophical theory of cogency—which integrates perspectives on the normative basis of a priori pragmatic presuppositions—and focus on the participants’ judgments of cogency in particular contexts. It is no surprise, then, that the types of critique envisioned in postscript II remain largely immanent in character: in critically assessing scientific arguments, one identifies empirically false assumptions, or one exposes suppressed premises that appear irrelevant in light of accepted goals. Even the third mode of critique emerges from the identification of tacit political aims at stake in the original context. However, my vague allusions to deliberative politics as the arena for resolving the debate over competing agendas in behavioral biology and the creationist controversy reveal the need for more robust forms of critique that are not entirely dependent on the immediate context of the debate. I thus agree with Habermas and other critical theorists who insist on the necessity of normative resources that enable context-transcending modes of critical intervention (see Cooke 2006; Lewandowski 2001; Misak 2000; Benhabib 1992). In Part III, therefore, I aim to identify normative resources that can ground both immanent and external critical interventions without incurring the difficulties that come with Habermas’s idealized approach. This task will require an integrating framework quite different from Habermas’s—a framework that is a posteriori, contextualist, and interdisciplinary from the ground up, rather than a priori, universalist, and philosophical. My working hypothesis is that contextualist approaches already contain the key to the solution. In this chapter, then, I tackle the challenge of radical contextualism on its own terms—a contextualism often identified with the socialconstructivist and conventionalist forms of relativism I described in earlier chapters.1 To address this challenge without losing sight of a pragmatic, interdisciplinary approach to scientific argumentation, the ethnomethodology of scientific work (ESW) presents a more useful point of departure than the growing contextualist literature in analytic epistemology and philosophy of language.2 Although the ethnographic and analytic discussions share some common concerns (as I note below), the latter operates at a high level of abstraction, whereas the former directly engages the study of scientific inquiry. Thus, after recalling the problem of contextualization (sec. 1), I describe the
215 Adjusting the Pragmatic Turn
radically particularist contextualism espoused by ESW, whose policy of theoretical “indifference” apparently negates the critical approach necessary for a critical science studies (sec. 2). Rather than reject ESW as incompatible with critique, however, I argue that critical theorists can adopt a “critical contextualism” that incorporates radical contextualism without altogether discarding context-transcending critical standards. To do so, they must, on the one hand, understand truth and objectivity as having an indexical character (sec. 3); one can then see dialogical idealizations as rhetorical topoi with context-transcending critical potential (sec. 4). On the other hand, they must clarify the position of critical argumentation theorists as indirectly engaged participants vis-à-vis practicing scientists (sec. 5). These moves allow specific modes of critique and ameliorate worries connected with relativism (sec. 6). 1 Difficulties in Contextualizing Habermas’s Formal Pragmatics Not surprisingly, Habermas’s quasi-transcendental, abstract theory of argumentation has drawn critical fire from postmodern pragmatists like Rorty, whose “frankly ethnocentric” sensibility rebels against such universalistic models (see Rorty 1987, 1991, 2001). However, even sympathetic readers working within the Frankfurt School tradition have found problems with Habermas’s emphasis on idealized consensus (e.g., McCarthy 1991, chap. 7; Benhabib 1992). The root difficulty lies in the excessively formal character of Habermas’s pragmatics of language. He assumes that the messy details of actual communication—rhetorical elements, innuendo, ungrammatical shortcuts, and the like—can ultimately be reduced to standard sentential forms that incorporate the various validity claims described in chapter 4. As his critique of Toulmin shows, formal pragmatics is driven by the fear that critical analysis will go astray without the guidance it provides, with its idealized system of validity claims and the corresponding argumentation theory. One must then link these abstractions with “empirical pragmatics” through a rather tortuous program of contextualization (Habermas 1984, 327–328; 1979, chap. 1). The idealized process standards described in earlier chapters carry on this formalist propensity at the level of argumentation theory. Habermas assumes one can derive these standards through transcendental arguments that demonstrate performative self-contradictions in skeptical attempts to reject them (Habermas 1990, 86–94). As I argued in chapters 5 and 6, however, difficulties
216 Chapter 7
arise as soon as one attempts to link these idealizations to actual processes of controversy and argument. It remains unclear how convincingly Habermas’s formal approach can resolve these difficulties. In chapter 6 the difficulty arose in connection with the status of compromises in scientific collaborations. Habermas’s dialogical model seemingly requires us to regard any consensual compromise on argument content as problematic: the appearance of consensus conceals or distorts what the actual discourse genuinely warrants. But such compromise might be reasonable in certain contexts (e.g., when unambiguous arguments are more helpful for policymaking). In reply, I suggested that Habermas’s model might accept compromise as generally necessary for sustainable research collaboration in science: though compromise is regrettable from an ideal standpoint, under proper institutional conditions it might work out in the long run as epistemically beneficial. But short-run cases call for a tougher appraisal: critics must carefully investigate the contextual conditions of compromise, asking whether departures from dialogical ideals are justifiable in light of further context-specific considerations. How much guidance does Habermas’s formal pragmatics provide for this contextual analysis? The difficulty in chapter 6 is a special case of the kind of challenges we saw in chapter 5, which arise in the internal relation between dialogical idealizations and social-institutional realities. Because actual discourses of any scope or complexity cannot simultaneously approximate, let alone fully satisfy, idealized standards, contextualizing adjustments become necessary—for example, decisions about who may participate, when to consider evidence sufficient to warrant a policy decision, when to compromise—decisions not directly based on the ideal standards. Consider, for example, the ideal that everyone is free to question any assertion whatever (provided one gives grounds for the questioning). In scientific discourse, this rule is problematic: even with justifications one cannot plausibly question just any assertion in science; rather, a sense of what claims are open to contest partly defines disciplinary competence (see Polanyi 1983, chap. 3). In other cases, ideals can conflict with one another under actual conditions, thus requiring decisions about how to prioritize dialogical demands. For example, do we include more participants and relax our expectation of consensus, or vice versa? Thus the dialogical perspective on cogency remains partial and somewhat indeterminate until it hooks up with a social-institutional analysis of actual conditions of feasibility. From the social-institutional perspective, participants
217 Adjusting the Pragmatic Turn
must refer back to dialogical ideals, asking how much their experience of actual, institutionally structured discourse warrants a confidence that the outcome is dialogically adequate. But they cannot answer that question simply by looking to the ideals, which ex hypothesi underdetermine institutional design. So which features of the actual discourse warrant our confidence in making such a counterfactual projection? How much may we reasonably depart from ideal requirements without losing our warrant? In chapter 5, I pointed to three general considerations that help us answer such questions within the boundaries of Habermas’s approach: the specific epistemic goal or mix of goals of a given discourse, the costs and feasibility of approximating different process standards, and past experience of institutional experimentation. By drawing on these considerations, one attempts to construct a justification for institutional mechanisms, structures, and strategies (e.g., credit, secrecy rules, consensual compromise) that limit or apparently violate dialogical ideals. Such justifications attempt to show that the given institutional feature or strategy of execution optimizes, or at least preserves, the dialogical and substantive quality of discourse under real conditions, as compared to known alternatives. Dialogical quality, however, is defined here entirely by counterfactual idealizations. Consequently, the burden of proof always lies on the side of institutional realization: the various mechanisms and strategies must prove themselves as necessary or acceptable limitations of the ideal. Until those contextualizing proofs are forthcoming, the presumption lies with dialogical ideals as having jurisdictional force over all discourse. But suppose we question this presumption. What if precisely the dialogical idealizations must first prove their relevance in the given context? ESW leads us to confront this question. 2 The Ethnomethodological Challenge: Accountability and Indexicality The above difficulties indicate the need for a better account of the contextualizing rationalities that participants employ in actual discourses. To address this need, McCarthy (1994, chap. 3) cautiously enlisted the heterodox brand of sociology developed by Harold Garfinkel and coworkers (Garfinkel 1963, 1967; Heritage 1984). Ethnomethodology distinguishes itself from mainstream sociology by paying close attention to the messy, largely nongeneralizable details of social interaction in local contexts. In contrast to Parsonian sociology,
218 Chapter 7
which accounted for social order in terms of norms internalized and applied by compliant members of the group, ethnomethodology emphasizes the agency that members exercise in actively constructing orderly social interaction in concrete situations. Social order arises not from the mechanical application of rules to situations whose intelligibility is given ahead of time, but as an ongoing, occasioned accomplishment in which members actively constitute—and ongoingly transform—the sense of their situations. Shared normative expectations and general interpretive schemas certainly play a role in social interaction, but members employ them creatively, engaging various ad hoc methods of sense making, the indefinitely vast repertoire of “procedures” for weaving general schemas and rules together in an orderly interaction that continually adjusts to and reconstitutes changing circumstances. In chapter 2, I sketched the chief consequence of this approach for the understanding of scientific rationality: an ethnomethodology of scientific work (ESW) that rejects abstract philosophical models of scientific method in favor of the close description of how scientists maintain the methodic sense of their inquiry as an ongoing, situated achievement whose rational character cannot be captured by formal reconstructions. In this section I lay out some of the key features that characterize ESW as a mode of ethnographic sociology, beginning with its approach to accountability. One can understand the key elements of ESW by starting with that idea. (2.1) As exercises of agency, methods of sense making support the mutual imputation of accountability necessary for meaningful social interaction. In contrast to Habermas, however, ethnomethodologists are less interested in accountability as a pragmatic idealization than in the myriad ways by which actors display mutual accountability to one another by their competent use of methods of sense making. The situated use of this repertoire of interactive competences involves normatively laden expectations—which Garfinkel once glossed as “trust”—that others will appropriately contextualize, and thereby give a definite sense to, actions or utterances that by themselves are indefinite and vague. Garfinkel (1963) famously demonstrated the reality of these expectations precisely by challenging them in “breaching experiments” in which the experimenter (Garfinkel’s student) demanded full explicitness in conversation from unwitting counterparts (often other family members or friends). The moral indignation provoked by such noncooperation reveals, as the flipside of the accountability displayed in the active mastery of behavioral moves, a corresponding receptive accountability for taking those moves in the appropriate
219 Adjusting the Pragmatic Turn
way. Actors hold one another accountable not only for their words and deeds but also for the situationally appropriate interpretation of words and deeds. (2.2) The ethnomethodological analysis of accountability as an active, contextualizing capacity goes hand in hand with a strong claim regarding the indexicality of meaning. According to ethnomethodology, the indexicality of language is universal: not just the standard indexical expressions (“this,” “I,” “you,” “now,” etc.), but all language—and all meaningful behavior—acquires a definite sense only in the concrete situation. The full meaning of any expression or possible behavior as a social action—as a move in social interaction— requires the exercise of creative, contextually appropriate, and mutually recognizable agency on the part of participants. Social conventions, general norms, and rules of method provide interaction with its normative sense, but not in a rigid, invariant manner. Rather, these are “variously available” to competent actors, who employ them to structure their interaction in a variety of ways according to the circumstances (Garfinkel 1967, 33). One might, for example, refuse to return an acquaintance’s greeting to indicate indignation over a previous slight (Heritage 1984, 117; Sacks 1975, 64–72). Here the violation of the greeting convention does not simply reflect an oversight or lack of proper upbringing; rather, one gives the norm a very specific meaning that effectively communicates indignation in a local context defined by the history of two particular individuals. This indexicality is also irremediable, persisting despite our best efforts to translate indexical expressions into fully explicit, context-free “objective” statements.3 Such explications always contain further ambiguities, and in some cases the explication actually loses the situated meaning of the original utterance by eliminating elisions that hearers depend on and expect as ordinary. For example, making a child’s cry “Mommy!” more explicit as “My mommy!” changes its situated sense (Bittner 1977, 87). In other cases overexplicitness suggests a lack of facility in the speaker. Garfinkel considers the irremediable indexicality of language to be empirically demonstrated by a number of studies (Garfinkel 1967; Heritage 1984, 92–97, 144–157). This is not to deny the importance of efforts to substitute objective expressions for indexical ones in many practices, science in particular. However, Garfinkel maintains that efforts at substitution always remain “programmatic”—explication goes only as far as necessary “for all practical purposes.” Even when scientists define categories in terms of necessary and sufficient conditions, they still rely on ad hoc mechanisms in applying general categories to cases.4
220 Chapter 7
Accountable actors know how much explicitness is required for practical purposes and can appropriately contextualize one another’s sayings and doings. These skills allow them to manage the indexicality of language and behavior by treating their meaning as “reflexively tied” to the concrete situation. Here “reflexivity” refers not to self-awareness but to the specific manner in which the determinate sense of norms and general categories depends on the actual occasions of their use. Even scientific texts and representations (figures, graphs, etc.) are not exempt: “The organization, sense, value, and adequacy of any representation is ‘reflexive’ to the settings in which it is constituted and used” (Lynch and Woolgar 1988, 109; cf. Jordan and Lynch 1992; Button 1991; Widmer 1986). To claim that scientific claims or texts are irremediably indexical flies in the face of the standard conception of scientific objectivity and universality. Thus Eric Livingston’s study of mathematical practices of proving offers a good illustration of the idea, inasmuch as mathematical proofs appear most universal, objective, and anonymous—a point Livingston does not deny.5 Rather, he insists that the ability of a proof to travel beyond the context of origin does not reside in the formulated proof alone, the written or diagrammatic “proofaccount.” Such accounts are partial descriptions that are intelligible only in relation to the practices of a particular culture of proving and the actual “lived work” of proving. An effective proof-account supplies just those relevant concrete details that allow other members of the mathematical culture to contextualize the account and “do” the work of proving. In analogy to one’s grasp of a gestalt figure in a set of lines and shapes, appropriately contextualized material cues allow members to make the embodied, spatiotemporal moves required to see the “gestalt of reasoning”—and thus grasp the ideal object and its properties—that the proof-account describes. However, the appropriate contextualizing methods are so ordinary—for example, knowing the sequence of physical actions involved in the construction of a diagram—that competent members take them for granted as “essentially uninteresting” (Garfinkel 1967, 7–9). Concrete methods do not deserve explicit comment, indeed such commentary would only distract, and might even mislead, competent practitioners. These practical moves are seen but unnoticed, they are “witnessed but ignored” (Livingston 1987, 56). Ethnomethodologists strive to notice just these moves—to describe the concrete, mundane details of practice through which alone general accounts, as “glosses” that leave out the details, acquire their local, definite sense and relevance for actual
221 Adjusting the Pragmatic Turn
scientific practice (Garfinkel and Sacks 1970, 343–345, 362–366). This is not to deny the objective, general character of mathematical and scientific results, but rather to ground objectivity in the corporeal, social practices through which it is achieved. (2.3) Livingston’s treatment of mathematics exemplifies a general point regarding the reflexive relationship between practice and the formulated “accounts”—for example, mathematical proofs, scientific articles—in which practitioners talk about and evaluate their practice. Garfinkel (1967, 1) summarizes the idea as the “central recommendation” of ethnomethodology, namely “that the activities whereby members produce and manage settings of organized everyday affairs are identical with members’ procedures for making those settings ‘account-able.’ ” As he goes on to explain (ibid., 33), this recommendation has the status of a research policy for how one views social practices: “Any [social] setting organizes its activities to make its properties as an organized environment of practical activities detectable, countable, recordable, reportable, tell-a-story-aboutable, analyzable—in short, accountable.” One assumes that in everyday practices members do and say things that they can observe, and they continually contextualize and update these sayings and doings so that their interaction makes sense. Just these ongoing contextualizing activities or “procedures” make their sayings and doings “reportable” in formulations, and thus “account-able.” It follows that formulated accounts, like general norms and categories, are “reflexively tied” to the concrete occasion of their production, and have a rational character only in the concrete context: recognizable sense, or fact, or methodic character, or impersonality, or objectivity of accounts are not independent of the socially organized occasions of their use. Their rational features consist of what members do with, what they “make of” the accounts in the socially organized actual occasions of their use. Members’ accounts are reflexively and essentially tied for their rational features to the socially organized occasions of their use for they are features of the socially organized occasions of their use. (Garfinkel 1967, 3–4)
This text could serve as a summary for any number of ESW findings. However, the central recommendation goes hand in hand with a further methodological policy that poses a radical challenge for critical approaches to scientific argumentation. (2.4) To get at the creative process of ongoing sense making and how actors account for that process, ethnomethodologists attend very closely to the
222 Chapter 7
concrete, locally situated details of social interaction and language use. In fact, starting with a set of abstract formal structures as the presupposed basis of rational interaction—the approach taken by Habermas’s formal pragmatics—can lead the theorist to miss the concrete “rationalities” or “endogenous logic” of actual social practices (Bogen 1999, chap. 2). Consequently, the more radical versions of ESW reject a priori methodologies and theories as a basis for analysis in favor of “ethnomethodological indifference,” the policy of “abstaining from all judgments of [the] adequacy, value, importance, necessity, practicality, success, or consequentiality” of the local practices being studied (Garfinkel and Sacks 1970, 345). According to Michael Lynch (1997a, 371– 372), this policy “assigns epistemic privilege to no single version of social affairs, including sociology’s own professionally authorized versions.” As a result, one’s own ideas about rationality are set aside in order to attend more closely to the situated “methods” or “procedures” that members themselves use to produce social order. Indifference thus forbids criticism based on the investigator’s own moral, political, or epistemological beliefs. This poses a potentially serious obstacle to the appropriation of ESW by critical theorists, an issue I take up below. As a research policy, ethnomethodological indifference is supposed to keep the researcher maximally open to the full range of rational agency as a contextualizing capacity. One should not confuse this policy with a disinterested “pure empiricism” or positivism that pretends to rely on presuppositionless inductive generalizations. On the contrary, ESW presupposes its practitioners can articulate the a priori know-how or “rational intuitions” they share with their subjects of study as competent language users (Coulter 1983; cf. Lynch 1993, chap. 6). For example: without a tacit grasp of the adjacency-pair structure, ethnomethodologists would be at a loss to follow a conversation in which answers count as answers precisely because they follow questions as a matter of convention. At the same time, the explicit formulation of such structures results, not from armchair speculation, but from “a great deal of study of actual transcribed materials” (Coulter 1983, 366). One strives to notice, and perspicuously describe, the tacit methods of sense making that are visible in the transcribed material. Descriptive adequacy is achieved insofar as the investigator has described members’ methods in such a way that “anyone” can “replicate the observations described” (Lynch 1993, 208). The challenge for critical theorists is this: although ESW does not deny that scientists regard their accounts as general, its central recommendation prohibits
223 Adjusting the Pragmatic Turn
any generalizing theory about science, whether the theory involves a general sociological explanation of science or a traditional epistemology that provides general definitions of “representation,” “observation,” “truth,” and so on. Such theories are, after all, accounts, but accounts that either forget or pretend to overcome the methodological obstacles posed by an irremediable indexicality.6 The honest alternative to grand theory, according to Lynch, is a deflationary descriptive approach that “respecifies” traditional concepts as “epistopics.” The investigator simply notices and describes the variety of situated ways in which actors actually do such things as represent, observe, and so on—but resists “all efforts to build general models and to develop normative standards that hold across situations” (Lynch 1993, 306). Applying this point to argumentation, we should say that its rationality lies in the practical, local achievement of cogent arguments. Formulated rules of argument and idealizations such as Habermas’s pragmatic presuppositions are glosses, shorthands that acquire their intelligibility and relevance only in relation to the situated rationalities, the practical know-how of local practices. Competent arguers must discover each time the concrete methods, the situated rhetorics, by which they can argue reasonably. Consequently, one cannot simply invoke formal structures or idealizations to account for the rationality of argumentation. Herein lies a demanding challenge for critical theorists who look to argumentation theory for their normative standards. Although critical theory involves much more than argumentation theory, the nonarbitrariness of its critical analyses crucially depends on a normative theory whose rules or idealizations supply standards for assessing actual cases of controversy and consensus formation. But critical theorists can employ such standards, it seems, only if they believe their idealizations properly account for the rationality of argumentation in general—just what deflationary ESW prohibits, given its doctrine of indexicality. This poses a dilemma for critical theorists: on the one hand, the project of emancipatory critique seems to require a conceptual framework that, precisely because of its formality, provides a stable platform for critical assessment. On the other hand, if indexicality applies to such ideas, then no such platform exists that could set critical theorists above the social fray. One can, to be sure, formulate such ideas, which engaged arguers might then invoke. Arguers can employ these, however, only as glosses that they must contextualize in the much richer local rationalities of practice—whereupon the ideas of reason take on all the local interpretive contingencies the theorist hoped to avoid.
224 Chapter 7
Of course, the idea that the application of general practical principles requires context-sensitive judgment goes back at least to Aristotle. In moral theory, we find a range of answers to the problem of application (see Hooker and Little 2000; Davis 2004). Here we must ask whether ethnomethodological contextualism entails a radical particularism that denies the evaluation-guiding character of general standards of scientific argumentation. If so, then dialogical idealizations and general logical norms cannot serve the critical assessment of cogency in context. Where, then, do critical theorists find dialogical and logical standards in light of which local arrangements might be criticized? Presumably, critics are limited to immanent modes of critique as varied as the contexts under scrutiny and ultimately reinforcing the polyglot of taken-for-granted local rationalities, which remain impervious to outside criticism. Because external critique must presuppose some foreign context for its cogency, critical arguments would appear arbitrary in the local context under scrutiny. We thus find ourselves back at a kind of incommensurability associated with Kuhn, facing the spectre of relativism that Habermas wanted to exorcise with his formal pragmatics: if standards of good argument are always and only relative to local contexts, then whether a given argument counts as a cogent justification depends entirely on the context in which it is assessed.7 An overarching, critical appeal to reason becomes impossible. 3 Incorporating the Radical Challenge I: Ideas of Reason as Indexical Critical theorists can respond to the above dilemma, I argue, if they can synthesize the attentiveness of ethnomethodologists with the engaged stance of participants. In this and the next section I develop the first step in this synthesis: the revised understanding of truth, objectivity, and dialogical idealizations that results when we take ESW seriously. As it turns out, the contextualist sense of these ideas harbors a context-transcending potential. In section 5, I describe the engaged stance; sec. 6 develops the modes of critique this revision allows. From an ESW standpoint, ideas of truth and objectivity function as modes of mutual accountability. In fact, as I noted in chapter 4, critical theorists like McCarthy draw on the ethnomethodology of error accounts for a pragmatic understanding of factual objectivity as a specific mode of accountability: “We are held accountable, and in turn hold our interaction partners accountable,
225 Adjusting the Pragmatic Turn
for the transcendent objectivity of the world as invariant to discrepant reports” (McCarthy 1991, 31–32). The pragmatic notion of truth simply draws the implications of objectivity for assertions about the world: if the unitary objective world is intersubjectively accessible, then true reports and assertions should be intersubjectively acceptable. Radical ESW suggests that we take this pragmatic analysis of truth and objectivity a step further and regard these “ideas of reason” as indexically conditioned, such that their semantic content is reflexively tied to their particular disciplinary (or subdisciplinary) contexts. That is, scientists typically operate with tacit, discipline-specific assumptions about objectivity and truth in light of which standards of argumentative accountability are specified as contextually relevant for the given discipline. At the local research site, scientists frame their arguments for publication so that these will find acceptance not only by other members of their research team or problem area but also before broader disciplinary audiences for which those claims have some relevance or applicability. They thus strive to make publishable, locality-transcendent claims that hold up among peers working in similar or overlapping areas of research (Garfinkel, Lynch, and Livingston 1981; Lynch 1985; Livingston 1995). In grasping the accountable cogencies of their lived work, scientists in a particular laboratory observe and grasp the reportability of their findings for other members of their subdiscipline or discipline—the objectivity of claims vis-à-vis the local laboratory. For example, the idea of a unitary objective world appears in experimental physics and chemistry primarily as a uniformity supposition, the idea that the basic laws and makeup of the universe are the same for all times and places. The locality-transcending truth of such claims is understood accordingly: anyone anywhere who reproduces the given experimental conditions should arrive at the same observations, and if despite their best efforts they do not, then the truth of the claim is in doubt. These inflected suppositions of objectivity and truth define accountability among physicists and chemists: if others repeatedly fail to replicate one’s findings, then one’s own competence eventually comes into question. In geology, by contrast, objectivity involves not so much spatiotemporal uniformity as the supposition of a single earth history. Whether an established finding at one site (e.g., anomalous iridium deposits at the K-T boundary layer in Italy) also holds in corresponding strata elsewhere is not a supposition but an open research question (Alvarez 1997).
226 Chapter 7
ESW thus leads to a more deeply contextualist pragmatics of truth and objectivity. Although contextualism and disciplinary differences have been linked with pluralist metaphysics of truth and nature (e.g., Michael P. Lynch 1998; Dupré 1993), the pluralism described above is a pragmatic one that does not preclude attempts at reconciliation at a higher level of integration (e.g., a single earth history that follows laws of physics and chemistry; or hierarchically integrated levels of description). Moreover, the claims that scientists advance as objectively true in one discipline are potentially relevant for work in other areas of research. This potential context-transcendence of discipline-specific claims is realized when, for example, geologists draw on physics or physicists take geological phenomena into account in the construction of sensitive experiments. We can thus distinguish locality-transcendence, which extends only to those further sites of reproduction that share the background assumptions of the particular subspecialty, from the broader context-transcendence exhibited insofar as local findings are taken up by scientists in other fields and by nonscientists in other social domains. Context-transcendence is possible because scientists work in the awareness of numerous contexts and practices simultaneously—the broader science community, the immediate institutional setting (university, corporation, government lab), funding agencies, technologists, and numerous other interested lay publics. Each of these contexts enters into the lived work of scientists and shapes their local practice and methods of accountability. Because these different contexts involve different practices of argumentation, with different notions of relevance and cogency—in a word, different practices of accountability—the context-transcendence of scientific findings first proves itself in these broader extradisciplinary spheres. Consequently, any adequate attempt to formulate idealizations or rules of argumentation must first of all attend to the obvious and the not-so-obvious differences in how members in different settings actually employ such idealizations in their production of order. In sum, the supposition that the reality we apprehend and understand here and now, in a particular local practice, is the same reality for everyone has a somewhat different practical sense that varies with context and practice, albeit in a way that does not preclude relevance beyond the original context. The same goes for the ideas of truth and rational accountability: what counts as “true enough” for the purposes at hand, and what adequacy requirements one must meet before a claim is taken seriously, will vary according to context. If
227 Adjusting the Pragmatic Turn
the above observations hold true within science, we should expect them to hold a fortiori for contexts in which scientists must interface with nonscientists. The conflict of background expectations and modes of accountability that can arise in such situations—for example, situations in which scientific experts appear in court, or participate in popular media presentations—have been well documented (Goldstein 1986; Jasanoff 1995). The context-transcendence of a truth claim assumes that such obstacles can be overcome. Thus a scientific truth claim assumes, not so much the counterfactual assent of an ideal audience, but rather the potential relevance and contextualizability of that claim in an indefinite range of scientific and extrascientific contexts. 4 Incorporating the Radical Challenge II: Dialogical Ideals as Rhetorical Potentials We turn now to process ideals. Ethnomethodology suggests that we can incorporate greater context-sensitivity into high-level process idealizations if we conceive them as possible recurrent discursive moves whose determinate sense—the meaning for the actual practice of science in a specific locale (laboratory or research team) and context (research domain)—depends on the demands specific to the locale and context in which the ideals are actually invoked. For example, the idea of inclusiveness, that anyone capable of making a relevant contribution should not be excluded from discourse, remains uselessly indeterminate until one identifies (a) a specific individual or group that (b) in a specific discipline has been arbitrarily and systematically excluded, but that (c) is capable of making a relevant contribution. The last specification (c) in turn requires both evidence of competence and a determination of relevance. Each of these specifications requires argumentation in its own right that goes beyond simply invoking the bare norm of inclusion. To establish a contribution as relevant, one must delve into the substance of the science itself. Similar demands arise with the norm of equality: besides identifying the subordinated group, and so on, one must provide a convincing measure of equal voice. None of these moves is given in the abstract idealization but requires further dialectical argumentation and close reflection on actual practices. Such efforts are necessary to overturn entrenched presumptions that the status quo sufficiently approximates standards of reasonableness—that existing exclusions and inequalities do not violate good process but are justified by the lack of ability
228 Chapter 7
in the excluded parties to make relevant contributions. To be useful for critical rather than conservative purposes, then, process standards must be explicitly invoked in a rhetorically effective manner that makes sense in the specific context (cf. Cramer 2003). As an example, consider the concern for greater participation of women in science. To a large extent, this concern issues from moral demands, or equity requirements, such as the demand for equal job opportunity (see Wylie 1997). But here I am interested in the epistemic function of ideas of inclusion and equality, that is, the role they play in cogent scientific argumentation. To the extent to which epistemic considerations motivate the drive for more women in science, we should find advocates of inclusion arguing that excluding women from science undermines epistemic goals. To make this case, one cannot simply invoke abstract ideals but must reverse traditional exclusionary views by arguing that women are equally competent and can indeed enrich scientific understanding.8 Examining the literature on women in science (see Schiebinger 1997, 1999), we find that proponents of inclusion have in fact made such arguments, albeit along different lines. At one end of the spectrum are those who champion distinctive “ways of knowing” for women. For example, the undeniable association of the entry of women into primatology with the critique of masculine bias and enriched understanding of female primates has been attributed to the distinctive ability of women scientists to empathize with female primates.9 Others have argued that positioning in a marginalized group, or explicit commitment to feminist goals, makes women (or anyone similarly positioned or committed) more sensitive to dubious background assumptions in the dominant scientific paradigm. On this type of approach, specifically feminist political-cultural perspectives, rather than gender or sex as such, constitute the epistemically relevant parameters for inclusive science (see, e.g., Harding 1986; Longino 1990). More prosaic arguments are also available. At the very least, one can argue that excluding women impoverishes science simply by reducing the pool of talent. Or one might notice gender-associated tendencies. A study by Alison Wylie (1997, 88–96) suggests that women reshaped archaeology not so much because they brought an explicitly feminist perspective with them, but simply because they had an interest in gender questions; this led to different research agenda rather than to a radically different kind of archaeology. Joan Gero has made a sociological argument against segregation and inequality in archaeol-
229 Adjusting the Pragmatic Turn
ogy. Specifically, she links the poor representation of women archaeologists in research on Paleo-Indian hunting practices with a tendency of scientists in that area to disregard relevant findings in other areas (edge-wear analysis) in which women archaeologists tended to specialize (Gero 1993; cf. Wylie 1997, 84). Gero’s analysis thus suggests that the uneven gender distribution within different areas of archaeology fostered an impoverished understanding of paleocultures—specifically, a view that associates early cultural development primarily with developments in hunting technology. My first concern here is not so much with the tenability of the above arguments; indeed, each of these feminist approaches has been subject to criticism. Rather, I emphasize the historical fact that the feminist call for inclusion was linked with substantive epistemic arguments—arguments that based the call for inclusion on detailed sociological analyses linking male dominance with bias in a field, that appealed to role models of outstanding women scientists (e.g., Jane Goodall in primatology), that invoked evidence for the topical relevance of the specific contributions of women scientists, and so on. The feminist philosophy of science suggests the general point that process idealizations have a determinate sense for actual practice only as they are rhetorically usable and effective in specific contexts for specific epistemic purposes, and in connection with substantive dialectical arguments that debunk existing models and establish alternatives. This has at least two implications. On the one hand, ideals of inclusiveness, equality, and noncoercion in scientific argumentation refer not to some ideal universal audience but always to specific features of a specific institutional arrangement in some particular domain or locale (cf. Tindale 1999, 87–93; Crosswhite 1996, 140–164). On the other hand, one cannot simply appeal to process idealizations as one would to norms with a universal legislative force that directly applies to concrete cases. Process ideals are indeed normative, but in the manner of potential rhetorical sites for critical interventions into existing practices and conventions in the science community. To render process idealizations more context-sensitive, then, I suggest we view them as enduring sites of contest and reflection in social life—potential questions or rhetorical topoi that in principle remain open to contest and thus can never be disregarded by practitioners as finally settled. Who is admitted, who is excluded, what counts as equal voice, how coercion differs from legitimate constraint are questions that are always potentially up for discussion, and whose resolution requires detailed argumentation focused on the specific
230 Chapter 7
features of the relevant area of inquiry. Indeed, questions such as these can arise for any social order whose constitutive practices and membership conditions depend on shared normative expectations. Insofar as members regard a social order as legitimate, they presuppose that the operative answers to such questions are adequate. Thus to refer to process idealizations as rhetorical topoi does not so much deny their status as pragmatic presuppositions as specify it: (a) process idealizations represent enduring (potentially normative) issues of social organization, including the social organization of scientific practices; (b) they are rhetorically available (as a possible means of persuasive argumentation) insofar as the corresponding vocabulary and cultural tradition provide thematic possibilities and historical precedents; (c) they have actual normative force for existing practices only insofar as practitioners can make convincing arguments for their contextspecific relevance by linking them with substantive considerations connected with the context at issue. In other words, as abstract ideals they do not enjoy presumptive applicability to practice; rather, their proponents must meet domain- and locale-specific burdens of proof. 5
Engaged Participation: Avoiding Indifference
Can critical theorists appropriate the above findings without forfeiting critical purchase? This question brings me to the second step in my proposal for a more context-sensitive critical science studies. To answer the question one must first determine whether critical theorists can acknowledge the indexicality of objectivity, truth, and other argumentative idealizations without adopting a full-blown ethnomethodological indifference. I suggest they can insofar as they combine the substantive findings of ESW with the engaged attitude of the participants—an attitude that ethnomethodologists themselves take seriously, albeit without taking sides. As ESW and other ethnographic studies reveal, scientists engaged in discursive interaction are anything but indifferent chroniclers of their situated methods, which they normally fail to notice (see Gilbert and Mulkay 1984). Like ethnomethodologists, critical (argumentation) theorists strive to notice such situated details; like scientists, however, they take the standpoint of participants who are interested in the correct assessment of potentially controversial scientific arguments. This does not mean that critical theorists must become scientists or public intellectuals and engage directly in this or that controversy. Although they may (and at least some should) do that,
231 Adjusting the Pragmatic Turn
there is also an indirect or “vicarious” mode of engagement available to critical theorists working in academic settings (see Rehg 2002, 32–37). This indirect engagement is evident in those controversies in which participants explicitly invoke argumentative ideals as part of their advocacy. Scientists themselves invoke such ideals, particularly in interdisciplinary controversies (e.g., Officer and Page 1996; Prelli 1989b), when they appeal to ideal models of method and cogency in order to make sense of their discursive practices, to produce orderliness in the controversy, and to persuade—in sum, as an accountability procedure. Argumentation theorists are indirectly involved in these debates insofar as directly engaged participants draw upon formulated ideals of argumentation. Scientists may acquire such ideas in a number of ways—in undergraduate philosophy or critical thinking courses, through contact with philosophers of science, from science textbooks, works by public intellectuals and science journalists, and so forth. These observations suggest that the context for argumentation theory arises from a dialectic of indirect, more or less disengaged theorizing that characterizes academic practices, on the one hand, and the directly engaged uses of formulated argumentative ideals by participants in controversy, on the other. By formulating indexically sensitive idealizations that participants find relevant to their situated accounting procedures, academic theorists avoid a disconnected top-down approach. But theorists meet the more radical contextualist challenge only when they recognize formulations as no more than potential accounting procedures that acquire a definite sense insofar as they inform members’ situated assessments of cogency in actual controversies. One thereby avoids mistaking them for foundations or rules with free-standing jurisdictional force over actual argumentative practices (cf. Lynch 1993, 187). Argumentatively effective ideals enjoy not transcendental necessity, but a situated practical necessity that is defeasibly acknowledged by actual participants who are committed to the reasonable resolution of disputes. 6
Sufficient Leverage for Critique?
To complete the argument for my proposal I must dispel the worry that arose at the end of section 2: can contextually indexed ideas of reason as described above provide sufficient critical leverage, or must they fall captive, finally, to the status quo precisely because they depend so heavily on participants’ situated—and often ad hoc—methods? I first sketch some modes of critical
232 Chapter 7
assessment that the proposed approach allows (6.1), then close by adducing some general considerations that should mitigate the related worry of relativism (6.2). (6.1) The context-sensitive approach I have elaborated above points to specific modes of criticism, some of which are immanent, others external. The latter modes nonetheless remain within a contextualist framework inasmuch as they arise from relations of potential relevance that link different contexts with one another. The first possibility I have already described above, in the feminist critique of masculinist bias in the sciences. Many of these criticisms attempt to show how a more egalitarian and inclusive representation of women in this or that scientific discipline or area of research has (or would have) a positive effect on inquiry. Insofar as this mode of criticism depends on the accepted understanding of epistemic goals and a close analysis of local and disciplinary practices, it proceeds in an immanent fashion. However, some feminists have also targeted accepted epistemic values, such as theoretical simplicity and homogeneity (e.g., Longino 1996, 1997). Insofar as the latter sort of critique has an external character vis-à-vis established practices, it faces a heavier burden of argument. In fact, I suspect Longino’s recommendations challenge the philosophy of science more than actual scientific practice. The mode of critique I described in postscript II (which draws on Longino) is perhaps a clearer case of external feminist critique. To be sure, that critique begins immanently, drawing on sociological analysis to make explicit the political values that drive competing research agenda. But the critic thereby raises a question for science policy whose answer requires a shift to a political context of debate. Two further possibilities proceed immanently by exploiting the indexicality of formal accounts and idealizations. The straightforward mode attempts dialectically to undermine or refute simplistic ideals of scientific method often invoked by experts in contexts of controversy (see Lynch 1999, 227–229). The relevant critical moves should not be difficult to imagine, at least in broad outline: essentially the critic adduces more detailed descriptions of situated methods as a counterargument against justifications based on formal ideals of method (see, e.g., Evans 1999). Exactly how such counterarguments work depends on the particular context. One can, for example, imagine sciencebased policy debates in which descriptively enriched critique undermines official policy justifications that assume positivistic models of scientific objectivity.
233 Adjusting the Pragmatic Turn
Or one might target the use of positivist ideals of method, as Anderson, Hughes, and Sharrock (1988) have done in economics. The third mode employs this dialectic ironically. Here the critic challenges science-based justifications as failing to meet ideals of scientific method. I call this move ironic because the critic’s real aim is to elicit justifications that elaborate the situated rationalities and local discretionary judgments that the justification did not at first attend to. The critique thus has the maieutic function of eliciting from the participants themselves (1) the recognition of the insufficiency of formal ideals and (2) a further articulation of the actual, concrete reasonableness of their practices. Of course, it may turn out that the articulation lacks plausibility—but then that too would constitute a critical advance in the controversy.10 The fourth possibility is one that Lynch considers compatible with indifference. By choosing to study a group whose practices are assigned a marginal status by the dominant culture, ethnomethodologists “alert us to possible alternative rationalities,” thereby casting doubt on official versions of reason and valorizing the marginalized culture (Lynch 1997a, 375). To be sure, for critical theorists, the choice of research topic is not an “indifferent” matter. It makes a critical difference, for example, whether one chooses to valorize the epistemic practices of Aryan supremacists, government-sponsored expert panels, or citizen-led initiatives. Thus, from the standpoint of a normative critical theory, the partisan commitment built into the research program calls for some justification, a sociopolitical “accounting.” If that accounting is not simply to beg the question in favor of one sort of rationality over the others, then it must proceed in a manner external to the practices it considers for possible valorization. All of the above types of criticism trade on the context-dependence of the standards and ideals that inform our notions of the “reasonable”—what counts as evidence, as a cogent argument, a sufficiently inclusive process, and so on. Of these, the second and third proceed immanently, leading practitioners to reflect critically on and revise inadequate conceptions of, or assumptions about, their own practices of inquiry and argument. As an immanent mode of critique, feminist philosophy of science goes a step further to draw practical implications from accepted assumptions, for example, implications for improving the organization and process of inquiry. The fourth mode of critique, however, as well as certain forms of feminist evaluation, is external, calling for a sociopolitical accounting whose standards
234 Chapter 7
of cogency are those of contexts other than the context under scrutiny. A similar point holds for external forms of feminist critique. These modes point to larger questions that require the critical theorist to take a position on competing conceptions of science, its research goals, and its legitimate place in society. Although these larger questions can emerge within the sciences, they are especially prevalent in debates at the science–society boundary (e.g., debates over stem cell research, intelligent design theory in science education). When scientific argumentation leads into larger issues having political, moral, or religious dimensions, the essentially contested character of the reasonable becomes painfully obvious. Neither critics nor participants can then avoid the question of social vision. At that point, a nonarbitrary critical analysis must be justifiable in relation to a specific vision of science in relation to society: which epistemic practices merit our attention as scientific and how those practices should relate to other social domains in a “good” or “emancipated” society. As science studies scholars have recognized, value-laden questions of social vision cannot be cordoned off from the content of scientific argumentation by drawing a sharp boundary between science and politics.11 Although scientific inquiry and political debate involve distinct forms of argumentation, at the interfaces of science and society they come together in various ways, some controversial among science studies scholars. To clarify the core issues that arise in this fourth mode of critique, it suffices to notice some relatively straightforward ways in which nonepistemic considerations and values seep into the content of policy-relevant scientific argumentation. Consider the social consequences of “inductive risk,” that is, of mistakenly accepting or rejecting policyrelevant hypotheses. Douglas (2000, 559–560) argues that “non-epistemic values are a required part of the internal aspects of scientific reasoning for cases where inductive risk includes risk of non-epistemic consequences. In these cases, value-free science is inadequate science; the reasoning is flawed and incomplete.” Consequently, we must modify our “understanding of what can count as a good argument.” Douglas identifies a number of points where research methodology rests on nonepistemic values: the choice of standards for judging statistical significance, the use of human subjects, the interpretation of data in light of background assumptions, and so on. The public dispute over the methodology of AIDS research provides a well-known case, in which activists effectively argued for changes in the testing requirements on AIDS drugs (Epstein 1995, 2000; Fabj and Sobnosky 1995). Here the kind of modified understanding of cogency that Douglas proposes to philosophers of science
235 Adjusting the Pragmatic Turn
emerged in the actual practice of argumentation. Postscript II provided a second example: Longino’s analysis of explanatory models in behavioral biology shows how metaphysical assumptions can steer research in directions that have consequences for social policy. Although inductive risk might not exhaust all the entry points for social values (see Kincaid, Dupré, and Wylie 2007), it opens the door to the broader point at stake in the fourth mode of critique, namely the possibility in some cases that the standards for judging cogency are internally related to evaluations of the social costs of error—evaluations that depend in turn on a vision of the good society. I doubt that the critical contextualist program entirely determines the content of that vision. But the possibility of such value-laden content arising within scientific argumentation should not surprise critical contextualists. On the one hand, the contextualist analysis of truth shows that the local, context-specific meaning of claims to truth contains a reference to other locales and contexts, in which the truth claim may be at least potentially relevant. This relevance clearly obtains in the case of AIDS research, which explicitly targets a reality experienced by those who suffer from the disease. On the other hand, the activists’ challenges called into question the broader social adequacy of the standard testing methods in clinical trials: in effect, the scientists’ dialogical ideals had to prove their relevance and adequacy in a broader context. The contextualist approach shows how such challenges are possible in principle, inasmuch as scientific argumentation is constituted at the local level through an interleaving-and-nesting of diverse contexts. As so constituted, locally produced arguments are situated in relation to a complex social reality, which harbors possibilities of criticizing and revising local assumptions in light of broader social demands. Douglas’s analysis of inductive risks fills in this broad account by identifying specific pathways of relevance, namely internal relations that link technical choices with nonepistemic social values. The possibility of critically scrutinizing the science used in legal and political decision making thus arises precisely from the way ideas of truth and cogency function in scientific argumentation. In adducing scientific arguments to justify policy, one insists on the probable truth and practical relevance of the findings for settings beyond the artificial laboratory contexts of origin. Consequently, an adequate justification requires one to make the science plausible in extrascientific contexts, where nonepistemic considerations might in some cases penetrate the content of cogent science. In such cases, disciplinary expertise
236 Chapter 7
alone cannot guarantee the expert’s authoritative status for assessing the cogency of scientific arguments.12 Consequently, controverted questions at the science–society interface call for modes of dialogue that bring together the multiple locales and contexts that constitute the interface as such. If so, then critical contextualism supports a deliberative democratic model of scienceintensive policy formation, for deliberative democratic procedures are designed for just such cross-contextual argumentation and dialogue.13 I explore this democratic connection further in the last chapter. (6.2) I close with some remarks on the relativist worry expressed by some critical social theorists. Of course, many science studies scholars find no cause for concern in relativism and some even welcome it. One difficulty in assessing these views is determining just what a given theorist actually means by “relativism.” Martin Kusch’s position, which draws both on science studies and analytic epistemology, frames the issue appropriately for our purposes. According to Kusch (2002), a consistent contextualism should embrace conceptual relativism, which denies “that there is some neutral or privileged ground from which such criticism [of accepted epistemic standards within a community] can be carried out” (ibid., 274). Positively, Kusch’s relativism extends the meaning finitism endorsed by the Strong Program to truth itself: the meaning of truth “at a given time and for a given [epistemic] community is more or less fixed by the accepted exemplars, judgements of similarity, interests and goals” (ibid., 275, emphasis in original). Thus there is no single uniquely correct definition of truth to provide a basis for judging the adequacy of its use in novel situations. Although truth has the broad sense of fitting or corresponding to the facts, “ ‘fit’ and ‘correspond’ have many and diverse exemplars and do not pick out a unique one-to-one relationship between language (mind) and world” (ibid., 219). Kusch’s relativism recalls the conventionalism that characterizes Strong Programmers’ interpretation of Kuhn and motivates Habermas’s critique of Toulmin. A conventionalist approach to the sciences seems to imply that the cogency of interparadigmatic argumentation is defined entirely in social-institutional terms, according to the demands of social order in science communities, thus disabling reasonable outside criticism. In response to that worry I have shown how ideas of objectivity and truth function within local contexts in a way that points toward the potential relevance of truth claims for broader contexts. Thus, when scientists believe their findings have implications for other contexts—other areas of science, or law, or policy—they strive to support
237 Adjusting the Pragmatic Turn
their claims by using arguments they hope are cogent abroad. And because at least some scientists have some familiarity with other contexts of relevance, they normally can make such arguments. This does not mean they always succeed in those attempts. But the fact that standards of cogency differ across different disciplinary and social contexts does not mean that, in principle, one cannot find cogent arguments that make the original claim rationally acceptable in other contexts. In fact, Kusch seems to grant this point. If relativism means that reasonable cross-contextual understanding and dialogue is impossible because the standards of each context are “self-vindicating” and thus beyond outside critique—an idea he finds at issue in MacIntyre 1988—then Kusch disavows relativism (2002, 273–274; cf. 134). Consequently, Kusch rejects strong notions of incommensurability that render reasonable cross-contextual argumentation impossible; whether two given contexts operate with commensurable or incommensurable standards is a matter that must itself be judged from a third context (ibid., 245–246; 277–279). So perhaps critical argumentation theorists might accept Kusch’s link between contextualism and relativism without fear of unemployment. Interestingly, Hales (1997) provides formal-logical grounds for this conclusion. According to his modal “logic of relativism,” the thesis that everything true (or untrue) is true (or untrue) relative to some perspective or context is not selfcontradictory—unlike the simplistic relativist thesis that “everything is relative,” full stop. His “consistent relativism” is thus opposed to the absolutist thesis that true statements are true independent of any context. But it remains compatible with a different sort of absolutism, namely the idea that at least some, and possibly all, true statements are true in every perspective or context.14 This formal analysis thus allows the kind of cross-contextual moves and idea of truth for which I have argued. At the same time, there are at least two reasons for hesitating to adopt Kusch’s relativistic contextualism. The first lies in his commitment to meaningor rule-finitism. This idea takes us back to the debate between the Strong Program and ESW. The debate is complex, involving conflicting interpretations of Wittgenstein and sociological method, and I cannot delve into details or the substantial body of literature here.15 Rather, I note what seems to me the core difficulty with finitism from a pragmatic point of view: it fails to understand what happens in learning a rule and what actually is learned, namely, a practical operation (see Sharrock and Button 1999).
238 Chapter 7
The classic starting point is Wittgenstein’s analysis of learning to count by twos, but the idea generalizes to language and normatively guided activity in general. Finitism is motivated by the problem of explaining how the student manages correctly to extend a given rule, such as nk+1 = nk + 2, beyond the finite set of examples set forth by the teacher (2, 4, 6, 8, etc.). The problem arises, however, because one assumes that the rule can guide future cases only if it somehow “contains” them in advance, whether we think of the “container” as a formulation of the rule, an intention, or a causal mechanism developed in learning the examples. None of the options holds up under scrutiny: a causal mechanism fails to explain the normative character of rule following, and it can hardly be said that the abstract meaning of a rule, or the content of the student’s intention in accepting the rule, “contains” the infinity of possible future cases. One is thus led to conclude that the meaning of the rule is finite, limited to the actual cases. Behind that assumption, it seems to me, lurks an overly contemplative notion of learning, focused on acquiring the literal meaning of a rule. One understands the counting-by-two rule by getting into one’s head the explicit meanings of the words or symbols, but no conceptual analysis of those symbols will give you numbers like 794 and 796. Consequently, after grasping the rule one has to figure out how to “apply” the rule in a separate step, for which the rule as a literal semantic entity provides no real guidance. If we return to the idea of meaning as use, however, and think of such learning as more akin to learning how to tie one’s shoelaces or how to swim (cf. Sharrock and Button 1999, 208 note 14), then we must say that in learning the rule one has an insight into the practical operation that constitutes the basis for the examples one has been shown. “Insight” here refers not to acquiring a mental content but to seeing how to perform an operation, to do something. If one does not see in the teacher’s examples how to go on with new cases, then one has not learned the rule to begin with. The second reason for hesitation lies in Kusch’s level of analysis. Although he allows for cross-contextual discussions, his analysis does not systematically attend to the ways in which scientists typically operate in multiple contexts, each having somewhat different exemplars, similarity sets, and interests. Thus his analysis of the relation between the epistemic subcommunities of scientists and other scientific and lay domains remains rather thin and unsatisfactory.16 Although his conception of an epistemic community allows for a range of scopes, his analysis of relativism tends to lie
239 Adjusting the Pragmatic Turn
at the level of societies or cultures in a way that neglects their epistemic complexity. To be sure, we should not ignore the broader cultural level as a context. In fact, the larger questions I noticed earlier in this chapter emerge most strongly at that level. The various cross-contextual exchanges that allow for outside criticism of scientific arguments occur largely within a culture in which arguers presuppose a deeper agreement on modern notions of reasonableness. One might thus raise the further objection that the entire approach remains bound to a particular context, that of Western modernity and its conception of science (see Cooke 2006). Although a full answer to this objection is beyond the scope of this study, I make a few remarks at the end of chapter 9. Before turning to that larger question, I want to illustrate, and further develop, the critical contextualist approach with the help of a detailed case study.
8 Three Dimensions of Argument Cogency—A Contextualist Case Study
What makes an argument cogent? Argumentation theorists have generally answered this question by supplying a list of general standards—standards of formal and informal logic, rules of dialectical testing and rhetorically effective persuasion. In many of these approaches, context enters in primarily through the rhetorical perspective, which focuses on the ways in which persuasiveness depends on the “rhetorical situation,” above all on properties of the particular audience. Beyond these rhetorical particularities, contextual differences tend to be marked off rather broadly, according to one or another typology of discourses: conversational, scientific, legal, and so on.1 Habermas’s typology of validity claims, each linked with a specific form of discourse, provides a good example of that approach. The analysis in the previous chapter goes beyond the usual approach in emphasizing the context-sensitivity of argumentation. If we take ESW seriously, then we should regard standards of cogency as context-dependent in at least two ways: the context, locale, and occasion of use determine not only which general standards are relevant but also the situated meaning of those standards.2 This approach does not dismiss general logical and dialogical norms, for these remain potentially relevant as possible resources on which arguers can draw when they formulate accounts that gloss their situated procedures of argumentation. But these normative resources do not operate in the manner of laws that automatically have jurisdiction over any and all arguments. Rather, their relevance must be demonstrated for the locale and context in question. Such demonstrations, however, proceed by way of further arguments that are cogent in the given context. Two things keep this model of scientific
242 Chapter 8
argumentation from collapsing inward and sealing off each context with its own standards of cogency: the context-transcending notion of truth and the fact that scientists typically operate with multiple contexts of relevance simultaneously in view. Consequently, the contextualist model allows both immanent and context-transcending modes of critique. By drawing on the situated disciplinary know-how of scientists, immanent critique targets simplistic selfunderstandings of argumentation or aspects of the social process of inquiry that are indefensible in light of accepted disciplinary goals. By confronting arguments made in local contexts with wider disciplinary and social audiences, external critique challenges participants to show how their local arguments deserve a hearing beyond the context of origin—and thus have context-transcending merits. Both modes of critical assessment nonetheless remain context-bound insofar as each must refer to some context. Starting with the argument at its local point of origin, the critic must always look to some particular context of cogency— the broader discipline, the administrative context of choosing among competing research agenda, the legal-political context of science-based policy, the democratic public sphere, and so on. This procedure suggests a broad heuristic for answering the question of cogency in context. In this chapter I introduce that heuristic and apply it to a case study. The critical contextualist heuristic that I propose here further elaborates the basic framework of distinctions that inform the kind of context-sensitive approach to cogency proposed in chapter 7. For the case study I look to a series of expert panels appointed by the National Academy of Sciences (NAS) to study the possible links between diet and health. Such panels provide an excellent site for studying the ways in which scientists assess the merits of published arguments.3 Located at the boundary of science and society, expert committees typically operate in a regulatory context that differs from that of research science in certain respects (Jasanoff 1990, 76–80). Many of these differences, however, heighten the demands on argument evaluation, and thus prove useful for understanding the cogency of scientific argumentation.4 Of particular interest here is the manner in which expert panels work within a range of contexts: not only that of the panel itself as a locale and occasion for deliberation, but also scientific contexts from which the arguments under scrutiny originate, administrative institutions to which panels are accountable, regulatory agencies that rely on expert reports in setting policy, and broader lay publics for which the report findings are relevant.
243 Three Dimensions of Argument Cogency
A critical contextualist case study is most fully realized as an interdisciplinary collaboration. Although this chapter has benefited from a certain amount of actual interdisciplinary exchange,5 the case study here mainly involves a “unilateral” interdisciplinarity in which a single author draws on input from other perspectives and disciplines. The case study thus remains more of a sketch meant to further clarify the critical contextualist approach and its relation to the traditional perspectives (logic, dialectic, and rhetoric). I will, however, conclude by drawing an implication from the contextualist approach, which supports a specific position in a debate over expert panel procedures at NAS. For the social history of the case and for some rhetorical and sociological aspects of my analysis, I rely on Stephen Hilgartner’s study of the NAS reports on diet and health.6 Beyond that, my analysis depends on further argumentation-theoretic analysis of primary texts, using dialectical and rhetorical tools I have described in earlier chapters. After introducing the case as he reports it (sec. 1), I spell out the critical contextualist framework as it emerges both from theoretical developments and from the case itself, which reveals three directions of concern on the part of participants (secs. 2, 3). I then delve further into the case to reveal tensions within the framework that participants in the case struggled to integrate (secs. 4, 5, 6). I close with some tentative conclusions for critical assessment (sec. 7). 1
Hilgartner on NAS
In his study of NAS advisory committees, Hilgartner applies ethnographic techniques, above all Goffman’s dramaturgical analysis of self-presentation, to the rhetorical analysis of three NAS-commissioned reports on diet and health. Although dramaturgical sociology normally studies face-to-face interactions, Hilgartner shows how Goffman’s concepts of front and back regions, impression management, and information control can illuminate the texts of expert panel reports (see Goffman 1959). As public self-presentations of the expert committee, the texts are exercises in impression management, written to enhance the panel’s credibility by emphasizing the reasonableness of its procedures and conclusion. Conversely, expert reports conceal the messy “backstage” negotiations and compromises that occur in closed committee sessions. Information about the discursive processes that generated the texts remains generally unavailable.
244 Chapter 8
Welding dramaturgical sociology to actor network theory and rhetorical studies of narrative, Hilgartner attempts to explain the different public receptions of three studies: whereas the 1982 Diet, Nutrition, and Cancer was relatively successful, the 1980 Toward Healthful Diets met with severe public criticism and the 1985 draft revision of the Recommended Daily Allowances (RDAs) was scuttled after backstage controversies became public. The key to understanding the different fates of these studies lies in the panel’s success at impression management, which in turn depended partly on the institutional networks of expertise they could enlist in their support. Because the 1980 panel lacked important alliances (e.g., there was no epidemiologist on the panel), critics enjoyed a rhetorical superiority that allowed them to cast doubt on the quality of the backstage discourse: the report was thus criticized as catering to the food industry. The 1982 study, by contrast, presented itself as strongly allied; this, along with the detail of the report itself (over 400 pages of analysis), supported the impression of impartiality and rigor; critics thus found it more or less impossible to impugn the panel’s self-presentation by calling its backstage into question. In the case of the 1985 draft, however, information about the backstage debate leaked into the press, and NAS’s impression management completely collapsed. Critics and NAS spokespersons subsequently struggled over the proper narrative of collapse: NAS claimed it was reasonable to cancel publication given the inconclusiveness of the evidence; critics, including panel members themselves, charged NAS with caving into political pressures that were extraneous to “good science.” Although Hilgartner’s study remains at the level of a descriptive-explanatory analysis, he is aware that the public credibility of advisory panels ultimately raises normative questions for assessing expert advice. Thus the failure of the 1985 draft casts doubt on the character of NAS, leading to questions about whether the procedures designed to ensure “the integrity of the Academy’s work [had] failed.” Did the failure of the draft “signal that public trust in the Academy was undeserved?” (114). Hilgartner’s dramaturgical model can inform a critical approach to the issue of public credibility by making us more aware of the rhetorical techniques that inform the writing of expert reports. However, the model cannot fully answer the normative questions raised by critical assessment. Indeed, Hilgartner does not answer the very questions he raises: he does not venture a judgment, for example, about which of the competing narratives of the draft’s cancellation was more accurate, or whether, more broadly, NAS deserves public trust.
245 Three Dimensions of Argument Cogency
There is good reason for his silence on such questions: measures of dramaturgical effectiveness do not supply adequate normative standards for assessing the empirical reliability of expert advice. When scientists tell me that smoking contributes to lung disease, or when my doctor advises me to accept one form of cancer treatment over an alternative, I do not simply want a good public performance from them. I want the truth, so far as the available scientific evidence indicates where the truth lies. Effective performances might give me confidence that scientists or doctors know what they’re talking about, but these can also mislead. We thus have reason to seek normative assistance at this point. As it turns out, Hilgartner’s rhetorical analysis, when supplemented by the further resources from argumentation theory, suggests a normative framework amenable to the contextualist approach. 2
Three Dimensions of Argumentation in the NAS Reports
In earlier chapters I have operated with the well-known distinction among three dimensions of argumentation (product, procedure, and process) that some theorists align with traditional analytic perspectives (logical, dialectical, and rhetorical). But as I argued in chapter 1, this alignment has problems. Still more important for my purposes, the three dimensions of argumentative practice do not capture important differences in the scope of argumentative contexts, that is, the difference between argumentation as a local, situated endeavor and argumentation as a broader social process that moves across different contexts. In chapter 7, I described how scientists themselves, working up arguments at a local level, have an eye on their transmission within the discipline and beyond. If a contextualist approach to assessment suggests any general framework, it should be one that recognizes differences in scope or levels of context. This supposition also fits with broad developments in contemporary sociology of science and argumentation theory. Sociological studies support the broad conclusion that argumentation as a microprocess differs in some important respects from the macroprocesses through which arguments travel across a large body or public; the critical analysis of these two levels will likely require different sets of tools. In argumentation studies, Goodwin (2005) has pointed to a broad scholarly division, two areas of research that align with this distinction between micro- and macroprocess. On the one hand, some
246 Chapter 8
argumentation theorists have studied argumentation as a locally situated discursive process, taking a dialogical or rhetorical approach. Scholarship in this direction has been largely framed in terms of conversational “transactions” (dialogues, critical discussions) and the “rhetorical situation,” standardly defined in terms of the particular audience, the exigence (the issue that generates the speech), and the constraints and resources available to the speaker.7 On the other hand, public sphere theorists have studied argumentation as a broadly dispersed public process. Such analyses tend to focus on institutional structures that affect the quality of public debate: the role of the media, cultural presuppositions, “special” interests, and the like as they bear on the quality of public discourse, especially democratic deliberation (e.g., Habermas 1989, 1996; Willard 1996; Goodnight 1982). These observations suggest we divide the dimension of process according to its local and public contexts. We can then distinguish three interrelated dimensions of argumentation: the argument itself (the product), the local transactions in which arguers produce and engage arguments, and the public networks and arenas through which arguments spread and reach a large number of people. Reciprocal relations of influence connect these three dimensions: the textual product, for example, typically reflects at least some features of the transaction as well as the broader publics to which it is addressed. Widespread cultural assumptions and stereotypes are, in one sense, properties of a “public,” but they have their effect at the level of local transactions. An a priori specification of all these interactions is not the point, however; rather, I propose this framework as a heuristic for opening up aspects of argumentation relevant for context-sensitive critical assessment. One can see the above dimensions at work in Hilgartner’s analysis of Diet, Nutrition, and Cancer (DNC). In studying the modes of impression management that NAS employs to establish its public credibility, Hilgartner lays bear some of the rhetorical techniques for packaging expert advice for “travel abroad,” as it were: expert reports are rhetorically designed to find acceptance beyond the local setting of the committee, that is, by policymakers and interested lay publics. Hilgartner thus frames his analysis in terms of the dynamic observed in chapter 7, namely the concern that participants in local settings—here, experts involved in committee transactions—have for broader contexts. He then goes on to describe two types of rhetoric that serve to establish trustworthiness abroad: an “empiricist rhetoric” and a rhetoric of expertise. The first (following Gilbert and Mulkay 1984) employs “a linguistic register
247 Three Dimensions of Argument Cogency
that conveys the impression of simply allowing evidence to speak for itself” (Hilgartner 2000, 51). This need not imply that the evidence speaks conclusively. DNC reaches negative or cautious conclusions for links between cancers and many of the dietary components studied. In summarizing the findings for naturally occurring carcinogens, for example, the panel concludes that according to the available studies, “there is no evidence that any of these substances individually makes a major contribution to the total risk of cancer in the United States.” The report hastens to add that this lack of evidence does not mean such compounds pose no diet hazard at all. Rather, “[f]urther investigations are necessary” (NAS 1982, 258). I call this an example of empiricist rhetoric because it contains no indication that the evidence is open to opposing interpretations: as the situation currently stands, the available carcinogenic studies do not support an evidence claim. An empiricist rhetoric, in other words, strikes an impersonal tone, suggesting that the weight of evidence, or lack thereof, is unambiguous; the facts themselves drive the committee to its conclusion. The rhetoric of expert judgment, on the other hand, employs qualifiers that indicate some uncertainty on whether the available evidence supports a given conclusion or not, so that the committee had to exercise a deliberative agency of its own in order to reach a conclusion. The rhetoric of expertise thus strikes a more personal tone. Although texts can be ambiguous about which rhetoric is operative, explicit references to the committee are a good indicator of the rhetoric of expertise. A good candidate is the statement “In the judgment of the committee, evidence from epidemiological and laboratory studies suggests that high protein intake may be associated with an increased risk of cancers at certain sites” (NAS 1982, 6). Distinguishing these two rhetorics is not entirely straightforward, an interpretive ambiguity scientists sometimes exploit.8 What matters for present purposes, however, is that one can see in such texts passages in which evidential arguments are reconstructed impersonally, and others in which the deliberative agency of the committee explicitly emerges. In the first case, the authors of the report employ a purely content-focused mode of presentation: just examine the evidence presented by this set of studies, they say, and it should be clear that such-and-such conclusion follows regarding the link between cancer and diet. In doing so, the committee presents an argument, the product of its reconstruction and summary of the available studies (each of which is itself an evidential argument). Presumably, readers are to take the merits of the content
248 Chapter 8
of the committee’s argument as sufficient grounds for accepting the conclusion. In the second case, the authors refer to themselves as the arguers (who make a judgment, who reach or fail to reach a conclusion). Thus the mode of presentation invites us to trust the conclusion because we trust the reasonableness and expertise of the arguers. Why should we trust the committee’s expertise? Here Hilgartner refers us to parts of the study that attempt to certify the committee’s expertise, overall balance of perspectives, and impartiality. We find this material in the front matter and in the appendix of DNC, as well as in brochures explaining NAS committee procedures designed to ensure objectivity (43–70). In dramaturgical terms, this rhetoric attempts to establish an authoritative public identity as the “front stage” for the committee’s reporting performance. This staging already begins on the title page, where we read that the report issues from the National Academy of Sciences, whose public identity places it “at the pinnacle of American science” (44). Whereas the empiricist rhetoric focuses the reader’s attention entirely on the content of the argument product, the expert-judgment rhetoric tacitly relies on claims about the quality of the transactional process through which the committee produced its arguments. In the 1980 study, assurances of transactional quality failed to convince many readers, and in the 1985 study the process itself broke down in full view of the public. These two documents thus failed as attempts at public argumentation. Diet, Nutrition, and Cancer, by contrast, was quite influential. According to Hilgartner (38; 165n106), the 1982 study “was widely perceived as supporting significant change in the American diet. . . . it focused scientific attention on the subject and helped consolidate interest in diet as a cancer-prevention strategy.” Moreover, as Hilgartner (chap. 3) makes clear, it succeeded publicly because of both its detailed, comprehensive content and the ability to present the committee transactions as meritoriously objective and balanced. 3 From Description to Critique: The General Normative Framework So far I have attempted to show how Hilgartner’s dramaturgical analysis of Academy advice involves the dimensions of argument content, transactional situation, and public sphere. NAS committee discussions represent transactional situations involving a small number of participants; transactions of the
249 Three Dimensions of Argument Cogency
committee (and of the NAS board) aim to generate products, arguments that summarize and assess empirical findings as warranting a set of conclusions regarding diet and health; these arguments are framed so as to engender the committee’s credibility before a wider public. I now want to bring out the normative features of the case study. Although Hilgartner himself does not venture a normative or prescriptive analysis, he is clearly aware of such concerns, and his case study describes the normative attitudes of the participants. My approach depends on the attitude described in chapter 7: to engage in critical (i.e., prescriptive) analyses of expertise, science studies scholars must weld their social-scientific study of scientific expertise with the normative attitude of participants who want to address their coparticipants in discussions of mutual concern. I take as my normative entrée the standards to which NAS procedures appeal, in order to draw some lessons for internal critique. In this section I lay out the general normative framework, which raises the problem of integration. We can then examine the NAS procedures in more detail (sec. 4) and their relation to public debate (sec. 5). The framework I introduced above distinguished content, transaction, and public. This suggests three dimensions of normative concern, which participants in the case in fact displayed:
• At the level of the arguments as products, participants are concerned that the content of expert advice is empirically adequate, a reliable basis for practical choices. This concern is reflected in the empiricist rhetoric that presents the evidence as leading straightforwardly to some conclusion. Formal and informal logic, theories of evidence and confirmation represent the main attempts to explicate the normative bases of this rhetoric. • In their argumentative transactions, participants are held accountable for the reasonable quality of their deliberations. This is evident in the procedures themselves, which display a concern for the committee’s balance, comprehensiveness, and freedom from pressure (more on which below). • Participants are concerned that their arguments win acceptance by wider publics. Toward this end, they cast their arguments as evidentially well grounded and present themselves, in their transactions, as adhering to procedures designed to ensure impartiality, well-informed expert judgment, and the like. This mode of self-presentation assumes that the public shares the ideal of reasonableness presented by NAS and thus is more likely to accept an
250 Chapter 8
argument that emerges from well-designed transactions of responsible experts who weigh the evidence on its merits. According to this three-dimensional framework, participants can evaluate the strength or cogency of an expert argument according to (a) the argument’s content, (b) the quality of the transactions that produced the argument, and (c) the ability of the argument to appeal to a wider reasonable public that finds it relevant, thought-provoking, or convincing. Each dimension deserves further comment. Because the framework is contextualist, the idea of transactions plays a central role. As I use the term, “transactions” refers to immediate situated encounters of argument and participant: argumentation as a local interaction that transpires on a particular occasion defined by some set of circumstances. Paradigmatically, transactions involve two parties (or a small group) engaged in a dialogical process of argumentation in which participants are physically present; as a symbolically mediated exchange between persons, transactions simultaneously involve both person–person and person–argument encounters.9 By modifying these elementary components—medium of exchange, participants, and arguments—we can broaden the idea to include transactions of various sorts: those in which only one party speaks (public addresses), electronically mediated exchanges, solitary reading of an article (in which the two parties remain more or less constructs for each other). In each case, the transaction involves an exchange or transfer of ideas, which if minimally successful enhances understanding between persons. The crucial question for a normative model of cogency at the transactional level is this: how well do features of the local transaction serve to place interlocutors, or the author and reader/ hearer of a text or speech, in a position to make a responsible judgment regarding the cogency of the argument? Dialogical models and some rhetorical analyses (such as that in chapter 5) attempt to spell out the relevant features: civility, open-mindedness, freedom from coercive pressures and emotional blocks, and so on. If such features can be identified in a particular transaction, then an argument product that passes muster in that transaction enjoys “transactional merits.” This holds both for productive transactions (in which arguments are created) and receptive transactions (in which arguments from one context are accepted in another). Conversely, we may not attribute such merits to arguments that issue from productive transactions distorted by bias, coercion, and the like. When a receptive transaction is distorted, the argument
251 Three Dimensions of Argument Cogency
likewise lacks transactional merits, though here the lack does not reflect badly on the argument itself. That is, if an argument is accepted or rejected in an unreasonable transactional engagement, then that transaction is something like a failed critical test or badly conducted experiment. At issue in transactions are the contents of arguments. In constructing and criticizing arguments, participants focus on “content merits,” the properties one can identify in the argument text that justify our regarding its conclusion as probably true. Taken as symbolic products of one or another sort, however, arguments have contents only insofar as they sustain interpretations. To interpret an argument, arguers must understand it as having a particular structure—at a minimum, they must differentiate between premises and conclusion—and they must supply tacit premises and background information, and so on. To assess the content merits of a given argument, therefore, arguers must appropriate and contextualize the text according to their understanding of the general and domain-specific topical and logical requirements that apply to the argument. Because transactions are situated in broader disciplinary fields and social domains, participants refer beyond their immediate local setting both in constructing and evaluating arguments. In constructing arguments they draw on inputs from this broader field: common motifs, assumptions, stereotypes, public knowledge, tacit background, and so on. At the same time, they assess those arguments with a view to their broader acceptance beyond the local transaction. This means that they desire a product whose content merits will hold up—be appropriately interpreted and accepted—across a range of transactional locales. To achieve this acceptance, arguers in a local transaction must have an eye on the operative domain standards of cogent content. Local transactions thus feed into macrosocial discourses that involve interested publics: scientists in a given area of research, citizens debating prospective legislation, laypersons concerned about a particular topic, and so on. Thus public spheres of various types and levels emerge from myriad face-to-face exchanges, local speeches and presentations, individuals watching a public speech anonymously, and the like. Arguments that win broad acceptance across a well-structured social space of multiple local (and reasonable) transactions enjoy what I will call “public merits.” One condition for public merits, then, is that an argument can travel: people in different transactional locales and domains10 can engage the argument and accept it. How far an argument travels thus depends on how much it can be
252 Chapter 8
tailored to specific audiences without distortion of its content. Technical scientific arguments, because of their highly specialized nature, usually do not travel much beyond the relevant subdiscipline. As technically formulated, such arguments have public merits within a limited domain—which is not to say that the lay public lacks grounds for accepting the conclusions (more on this below). To some extent, such arguments can be adapted for different audiences—science popularizations both simplify the original arguments and fill in background that scientists would find unnecessary to make explicit. To speak of an argument traveling this way assumes that we can identify some core that persists through the modifications, or that we can see sufficient similarity across the various formulations to regard them as making the same basic argument. How to specify the conditions of similarity for argument contents remains an open research question in argumentation studies. But as criticisms of both the Executive Summary of DNC and press releases show, scientists consider similarity important when scientific claims are repackaged for public consumption. However, the capacity to travel is not a sufficient condition for public merits. The macrosocial space through which the argument travels—the networks and aggregates of transactional contexts—must also be one that sustains collective reasonableness. To assess the public merits of an argument, one must therefore analyze the composition and structure of this social space in the light of a normative model of public reason. If the social space is well structured, then we have social-institutional grounds for considering that argument cogent. Because these grounds involve sociological and cultural information, the ascription of public merits is (a) independent of the merits we can identify in the argument itself on the basis of logical and topical standards11 and (b) differs from the transactional merits we can attribute to the argument as persuasive in this or that particular locale (although in ascribing public merits, one assumes that overall or on average, the transactions that make up a given public have been sufficiently reasonable). Consider, again, the case of technical scientific arguments. As a layperson, I might have little ability to assess the contents of those arguments, and I might know nothing of transactions within the research team, at science conferences, and so on. But if I have reason to believe that the social-institutional structures of science (education, networks of cooperation, funding methods, review procedures, etc.) support collective reasonableness (and tend to correct unreasonable individual transactions), then I have some grounds for believing scientific claims that have won wide accep-
253 Three Dimensions of Argument Cogency
tance among scientists. Such arguments may support conclusions that deserve broad acceptance on the basis of their public merits within the sciences. To summarize: if each of the three dimensions—content, transaction, and broader publics—is subject to normative standards whose satisfaction counts in favor of the argument at issue, then one can speak of arguments having three kinds of merits, depending on which dimension one refers to. Content merits can be identified in the text of the argument itself by applying various analytic tools to an interpretation of that text. An argument has transactional merits to the extent that it wins acceptance in a local dialogue or exchange conducted in a way that fosters reasonable judgment. The conditions for ascribing transactional merits vary according to the particular transactional context—the capacities of the participants to process information, their background knowledge, local conventions of argument, and so on. An argument has public merits insofar as it can travel across different transactional locales whose macrosocial arrangement and aggregate conditioning sustain collective reasonableness. Each of these types of merits, I claim, potentially matters for participants in locally situated occasions of argumentation.12 I say “potentially” in order to leave open the possibility that different types of participants rely more on one dimension than another when they assess scientific arguments. Thus we might expect experts to focus more on content merits, and the lay public more on assumptions about transactional quality or the public organization of science. But that division of labor does not represent an iron-clad rule: citizens have sometimes become quite well informed about the technical aspects of scientific claims, and scientists worry about transactional quality, as the NAS and CDF cases demonstrate. A division of labor, however, seems to presuppose some principle of integration. Indeed, by calling these elements of cogency “dimensions,” I imply the possibility of integration, as though cogency might emerge as a single coherent object from tridimensional assessment. The framework thus raises the question of integration, or at least calls for more discussion of the interrelation among the different types of merits. Yet we should hesitate to expect a general answer to this question if we take a contextualist approach—particularly if we want to avoid the danger of interdisciplinary holism and the problems that beset Habermas’s rather strong mode of integration. In fact, even as a general idea, the contextualist framework itself discourages easy solutions to the problem of integration, for example, the answer that
254 Chapter 8
cogent arguments should ideally enjoy all three kinds of merits. Perhaps internal relations similar to those described in chapter 5 between logical and dialogical merits tie together content, transactional, and public merits. But that will depend on the context and kind of argument rather than on an a priori analysis. If we must integrate the contextualist model, then best to do so along procedural lines, as I propose in the next chapter. In any case, the task of integrating (or not integrating) these different sorts of merits always falls to the participants engaged in specific occasions of argument construction and evaluation. Those occasions provide the starting point for critical analysis. To address the problem of integration, then, we should first return to the case at hand and examine the difficulties that the participants themselves encountered in that task. 4
Transactional Process Standards at NAS
I begin with the official procedures governing NAS committees. As I mentioned above (sec. 2), Hilgartner analyzes the rhetorical use of such procedural features as a means of enhancing the public authority of the committee. Closer examination of the procedures, however, shows that their rhetorical effect depends on dialectical ideals that, similar to Habermas’s, attempt to regulate the transactional process within committees in a way that contributes to more reasonable or “objective” outcomes. The dialectical ideals are in turn closely tied to the Aristotelian idea of ethos, which I presented in chapter 5. The procedures at issue are alluded to in the opening “Notice” and Preface of Diet, Nutrition, and Cancer (NAS 1982, ii, v–viii). There we find, among other things, several references to the process and procedures that produced the report:
• the report’s approval by the Governing Board whose members come from the Councils of the National Academy of Science, the National Academy of Engineering, and the Institute of Medicine; • the description of the composition of the committee, whose thirteen members “were chosen for their special competences and with regard for appropriate balance” (NAS 1982, ii); • a listing of the thirteen specialties, whose “multidisciplinary” breadth “has served to ensure comprehensive coverage of the scientific literature and to provide a broad perspective to the committee’s conclusions” (ibid., v);
255 Three Dimensions of Argument Cogency
• a reference to appendix A, where we find the “institutional affiliations and major research interests” of the members (ibid., v, 453–455);
• a statement to the effect that the committee was “aided by extensive consultation with scientific colleagues” and by special technical conferences and public meetings (ibid., v–vi); • the acknowledgment of special debts to specific consultants and staff (ibid., vii–viii);
• an acknowledgment that the issues are matters of controversy, involving interpretive judgments, but that “the committee has attempted to present the evidence as objectively as possible and to indicate the range of scientifically acceptable interpretation” (ibid., vi); • a statement that the report has been subjected to outside review according to “approved” procedures (ibid., ii). Hilgartner’s (chaps. 2–3) rhetorical analysis of these statements draws on Latour’s actor network theory: by citing the prestigious institutional affiliations of members and extensive consultation, the committee announces the strength of its alliances and silences potential skeptics. He also points out that the rhetoric of “comprehensiveness” and “balance” aims to present committee members as “representatives of domains of expertise.” The panelists “do not speak for particular organizations or constituencies; they represent the viewpoints of their specific areas of knowledge” (48). This rhetoric establishes the committee’s credibility by associating its transactions with broader scientific publics. Notice, however, that the rhetoric is effective for the lay public only if the latter sees the committee’s multidisciplinary composition and institutional alliances as the right sort of connections, namely those that are consistent with an image of scientific rationality that the public finds credible. To examine this case from a participant perspective, then, one must engage in a normative argumentation-theoretic analysis foreign to Latour’s approach. From the standpoint of argumentation theory, one can see that the standards undergirding this rationality have a dialectical character: the idea of open, critical discussion as a path to cogent arguments. In making its dialectical ideals public, the committee report in effect attempts to establish its trustworthy character or competence—thus a rhetorical appeal to ethos in the Aristotelian sense, which has a closer connection with collective reasonableness than the notion of impression management that informs Hilgartner’s
256 Chapter 8
approach. This dialectical-ethotic interpretation becomes clearer if we dig into the NAS procedures that underlie the statements above. Here I extend Hilgartner’s analysis of recent NAS brochures and guidelines for committee members. I focus in particular on the following requirements: the outside review process, the statement on conflicts of interest and bias, the confidentiality requirements, and the stress on reaching consensus. (1) From a dialectical standpoint, arguments are cogent only if they can withstand severe critical tests. The outside review process is designed to provide such tests.13 The questions that reviewers are encouraged to pose target comprehensiveness (“Are all aspects of the charge fully addressed?”), the adequacy of the substantive analyses that support the conclusions (“Are the conclusions and recommendations adequately supported by evidence, analysis, and argument?. . . Are data and analyses handled competently?”), and impartiality of judgment, particularly at those points where “value judgments” enter in (“If any recommendations are based on value judgments or the collective opinions of the authors, is this acknowledged? . . . Is the report fair? Is its tone impartial and devoid of special pleading?”). Committee members are then required to respond to such challenges of the outside reviewers. (2) Dialectical treatments of argumentative processes stress the importance of certain dispositions that foster the participants’ capacity for impartial judgment—above all their freedom from internal pressures and biases. We see this concern expressed in the form “On Potential Sources of Bias” (reproduced in Hilgartner, 64), which explicates the Academy’s understanding of “balance”: acknowledging that the issues addressed by experts rest partly on “professional value judgments” and not just on technical arguments, the form explains that because sources of individual bias can be subtle, candidates for membership must disclose not only monetary ties but also “any such factors that in your opinion might reasonably be construed as potentially compromising your independence of judgment in matters within the assigned task of the group.” With this information in hand, the Academy then attempts “to appoint members in such a way as to represent a balance of potentially biasing backgrounds or interests.” This selection rationale implies a realistic recognition of the difficulties individuals have in freeing themselves from biases that can influence judgments subconsciously. Thus the impartiality of the committee’s deliberations emerges more as a collective result of its composition than as individual achievements of each member.
257 Three Dimensions of Argument Cogency
(3) The concern for the freedom of the committee also emerges in the confidentiality requirements. Although some NAS meetings are open to the public, committee meetings in which the members are actually developing their report remain closed to the public. Moreover, the members themselves are prohibited from revealing the matter of these sessions to outsiders, nor may anyone disclose findings until the report is finished and the Report Review Committee has signed off. As Hilgartner explains (57–58), confidentiality is “a way of shutting out pressure groups and publics. . . . the Academy’s system of closure aims not only to furnish committees with a protected environment where they can operate independently but also to deprive vested interests of a space for influence peddling.” Thus we see here a concern for the committee’s freedom from external pressures. The exclusionary character of confidentiality, however, reveals a tension in NAS ideals—that is, an apparent conflict between inclusiveness and freedom–that provides a target for critique, as I explain below. (4) Some dialectical theories of argumentation, such as Habermas’s, take consensus as the goal of critical discussions. Although the role of consensus in science has been challenged both as a description of science communities (Gilbert and Mulkay 1984) and as a norm of research (Solomon 2001), scientists themselves often consider it important, as we saw in chapter 6. In the NAS report, consensus functions rhetorically to reinforce credibility. Thus the committee process is structured along the lines of a consensus conference: the members are urged to arrive at consensus, though the Academy makes provision for minority reports (Hilgartner, 23, 51–52). This sets a very high burden of proof on report conclusions, but it also gives the impression that the committee, given its broad composition and procedural design, has reached a conclusion that is widely acceptable across the relevant scientific disciplines. The consensual character of reports, in other words, works together with the diverse composition of the committee, confidentiality, and the review requirements to help establish the idea that the committee’s local transactions reflect a wider public of experts—what Polanyi (1969; cf. also 1951) called the “republic of science”—such that the conclusion has not been skewed by one constituency or by interpersonal dynamics unique to the committee’s transactions. In chapter 6 we saw how the use of such dialectical procedures can increase the participants’ confidence in the cogency of consensual outcomes. Hilgartner emphasizes the other side of this process: in advertising its methods, NAS also aims to increase the confidence of outside addressees of its reports.
258 Chapter 8
If the panel and public share the dialectical notion of reasonableness that underlies such procedures, then the use of such procedures can have a rhetorical effect both inside and outside the group that follows them. Here “rhetorical” has the substantive sense similar to that of chapter 5: on the one hand, the procedures place panel members in the position to make responsible collective judgments; on the other hand, by displaying those procedures the panel communicates to addressees its adherence to an ethos that makes them trustworthy judges of cogency. As it turns out, however, this rhetoric was not entirely successful, even with the 1982 report. 5
The Reception of DNC by Scientists
Let us take stock. In terms of the heuristic I proposed above, Diet, Nutrition, and Cancer employs a rhetoric that attempts to establish the public credibility of its arguments. Behind this rhetoric lies a multidimensional, normative notion of cogency, which emerges when one considers the report together with Academy procedures. The Academy clearly has a concern for the content merits of the report: conclusions should be sufficiently supported by the evidence; all the relevant findings should be taken into consideration; referee challenges must be satisfactorily met. At certain points in the argument, however, the committee must exercise its considered judgment as an expert deliberative body. Thus the rhetoric of expertise invokes the quality of the deliberative situation as relevant for the assessment of argument cogency: insofar as the deliberative transactions are reasonable and impartial, readers may consider the committee’s argument cogent in virtue of its transactional merits. But the claim to transactional merits rests, in this case, on the way in which the report and the Academy procedures present the transactional setting as a microcosm of the wider scientific public. The report thus presents its argument as having public merits within the relevant scientific communities, and thus as meriting a broader acceptability among still wider publics of administrators and consumers. This rhetorical strategy assumes that the audiences for the report operate with a similar notion of cogency. As I argued in section 4, the presentation of the report as having public merits within the sciences depended heavily on a set of process ideals. Hence the rhetoric assumes a broad agreement on these ideals and their interpretation. Criticisms of the report, however, show that this assumption is problematic. Although scientist-critics tended to focus on content merits of the report, in particular its policy conclusions, lay critics had
259 Three Dimensions of Argument Cogency
already raised concerns about the process itself. In this section I consider the scientific reception. For a sense of the reception of DNC among scientists I turn to the report prepared by the Council for Agricultural Science and Technology (CAST) in response to a request by some legislators (CAST 1982).14 Unlike the NAS panel, the CAST report contained forty-five separate reviews of DNC written by individual scientists (or a team of scientists in one case). Participants in the CAST report targeted various aspects of the report: the main content, summaries and news releases, and the process. The chief objection, raised by nearly three-fourths of the reviews, concerned the committee’s tentative recommendations for healthy diets (“Interim Dietary Guidelines,” NAS 1982, 14–16). The NAS committee, in fact, had been charged with three goals: not only to review the “state of knowledge” but also to propose recommendations for diet and for future research (NAS 1982, v). However, given the uncertainties in the existing evidence, making dietary recommendations appeared to many scientists as unwarranted. This objection targets the content merits of the practical argument that moved from empirical claims to practical policy guidelines. Other reviewers took aim at the content merits of the arguments supporting the evidential claims. Aside from a few reviewers who regarded the report as more or less a disaster, criticisms of the content of the report tended to be limited to specific flaws: the misinterpretation or neglect of studies, faulty generalizations, and so on. The fact that the CAST scientists did not converge on any single flaw (aside from the questionable dietary recommendations) speaks to the content merits of DNC. Although the reviewers focused heavily on the content, a number of them noticed transactional features. Many praised the comprehensive and balanced character of the study, sometimes referring explicitly to the diverse composition and competence of the committee (see, e.g., CAST 1982, 27, 29, 35, 65– 66, 75). Others, however, criticized procedures (ibid., 72) or even charged the committee with having “a blind side or bias” (George V. Mann, in CAST 1982, 49). Some scientists were motivated to examine the transactional situation as a way of explaining the discrepancy in conclusions between the 1980 and 1982 reports (e.g., CAST 1982, 29, 39). Whereas the 1980 report eschewed recommendations as unwarranted by the evidence, DNC held that some interim dietary guidelines were justified. Owen Fennema (CAST 1982, 19–20) explained this difference in terms of two judgmental inclinations, which stem in turn from the different contexts, specifically the differing institutional charges. The 1982 committee, “abetted by its charge, has ventured from the
260 Chapter 8
scientific community’s time-tested approach of deriving virtually irrefutable conclusions from hard facts, to the less reliable and more contentious approach of composing judgments and educated speculations from factual associations that may or may not be causative in nature.” Scientists reading the report relied on both content and transactional features in their assessment. They also attended to issues connected with the public merits of the report. Here one finds another large area of concern, namely problematic shifts in content between the report itself on the one hand and its Executive Summary and various news releases on the other. As attempts to translate the technical arguments for broader consumption, the summary and releases omitted crucial information and qualifications, and in the case of some releases made incorrect claims. A number of reviewers were also concerned that the discrepancy between the 1980 and 1982 reports would confuse the public (e.g., CAST 1982, 30, 33–34, 39, 63). In sum, scientists assessed DNC in terms of each kind of merit: content, transactional, and public. Notice, however, that in criticizing the committee’s transactional process, they did not question the process ideals themselves that lay behind NAS procedures; rather, transactional criticisms called for greater fidelity to those ideals: less bias, greater disciplinary diversity, and so on. Another group of outside critics, however, had already begun to challenge NAS procedures in a way that rendered the ideals themselves problematic. These criticisms in effect questioned the public merits of NAS reports by casting doubt on their transactional merits. 6
The Lay Reception: Tensions in Process Ideals
Comparing Academy procedures with argumentation studies, one may see that the claim to transactional merits depends on three types of process ideal: inclusiveness, freedom, and equality. Claims of the committee’s comprehensiveness and balance rely on an ideal of inclusiveness: the discourse includes all the relevant scientific perspectives, thus ensuring that any arguments that emerge have taken all the relevant evidence into account. As I noted above, however, “balance” was also associated with the idea that no single interest or bias controlled the outcome. Although individual members of the committee might be subject to distorting biases, the collective judgment of the committee as a whole is impartial. Impartiality in the sense of balance suggests equality among the diverse perspectives: the idea that each member has an equal voice
261 Three Dimensions of Argument Cogency
in the deliberation, that is, each has an equal opportunity to make the best case possible for arguments from different perspectives, so that the various sources of evidence receive due, or “evenhanded,” consideration.15 The confidentiality of the committee buttresses this ideal of impartiality by protecting members from outside pressures. Confidentiality requirements thus promote the freedom of the deliberators to consider the arguments purely on the scientific merits of their contents. Criticisms of Academy procedures, however, show that the ideals of inclusiveness and freedom allow conflicting interpretations of what count as dialogically adequate transactions. This is evident from Phillip Boffey’s early critique of the confidentiality requirement. According to Boffey (1975, 257), not confidentiality, but openness—in effect, a sunshine rule—should govern transactions: “The Academy should open its advisory proceedings to public inspection.” He justified this with two considerations. The first notes the difficulty of assessing transactional merits simply from content alone: confidentiality makes it “extraordinarily difficult to monitor the performance of the Academy committee. Even experts find it hard to evaluate a committee’s judgment without access to the original data on which the judgment is based.” Interestingly, Boffey argues that confidentiality can lead to the very problem it attempts to prevent, a committee operating under the influence of outside pressures: “the public has no way to judge whether special interests have influenced the committee’s deliberations.”16 Boffey (1975, 258) goes on to rebut the Academy’s rationale for confidentiality with a second consideration, namely the dialogical benefits of his sunshine proposal: The excuse generally given for this secrecy is that the committee members can speak more frankly, and with less fear of outside pressure, if the public is barred. But if the process were opened up more—if all meetings and records were made public—then the committees might benefit from unexpected insights volunteered by the interested public, and the public could assure itself that the project was being well handled.
Boffey’s argument against confidentiality thus makes two moves. On the one hand, his sunshine rule is supposed to resolve the apparent conflict in NAS procedures between inclusiveness and freedom from coercion: to assess Academy reports, lay readers must have the kind of access that allows them to assess the freedom of the transactions that generate those reports. On the other hand, Boffey suggests that “interested” lay publics provide a source of relevant contributions, and thus should be included in the process more directly.
262 Chapter 8
Boffey acknowledges that independence of judgment is important for expert panels, but he sees the main threat in government officials having access to meetings, not in post hoc publication of meeting records. The danger, that is, lies not in public access but in giving “government funding agencies much greater power than they now possess to influence a committee’s deliberations” (ibid.). Thus he disagrees with NAS over the main threat to the freedom of committee transactions. This disagreement—which stems in large part from his analysis of the institutional position of NAS and its subservience to the status quo (ibid., 246–251)—leads him to a different interpretation of inclusiveness. The relevant perspectives that must be represented for comprehensiveness and balance include adversarial points of view and lay publics, and not just scientists sympathetic to industry and agency goals (ibid., 255–257). Boffey’s (and others’) criticisms instigated some procedural reforms at NAS in the 1970s and 1980s (Hilgartner, 25). By 1997 the concern with confidentiality had led to lawsuits and court rulings that required modest changes in NAS procedures in accordance with an amended version of FACA (58–60). Behind such concerns one can see a dispute over the way expert panels are constituted in relation to democratic participation and oversight—an issue that has a long history and has recently gained increasing attention in STS (see Chilvers 2008; Lengwiler 2008; Turner 2003b; Collins and Evans 2002). The analysis of Boffey’s critique shows that such disputes can involve interpretive conflicts over dialogical ideals—here, freedom and inclusion. In fact, a contextualist approach reveals that such conflicts—which should come as no surprise at this point— actually involve two interesting tensions: (1) The tension that most readily aligns with the problem of science and democracy appears in the case study as a conflict of publics: to which public should we hold expert transactions accountable? The foregoing analysis shows that Academy procedures and rhetoric rely on an interplay of two publics. NAS attempts to establish the credibility of expert advice for lay publics by presenting committee transactions as a microcosm of the relevant scientific publics: the range of perspectives, balance of potential biases, submission to outside reviews, and the like all serve to present the committee’s judgment as representative of the collective trend of opinion in the broader scientific community, that is, the disciplines most relevant for assessing the content of evidential arguments about links between health and diet. Thus, the rhetoric of NAS reports presents the report as a product with public merits of a particular
263 Three Dimensions of Argument Cogency
sort, namely as an argument that holds up across the network of scientific publics. In other words, the public of scientific experts is the certifying authority that vouches for the content merits of the report and to which the committee is accountable for its assessment of the available evidence. At the same time, NAS reports are addressed to lay publics of administrators, citizens, politicians, and the like. Thus the rhetoric of expert credibility described above is ultimately designed to win acceptance by lay publics. This rhetoric envisions a deferential relation between experts and laypersons: you lay readers can trust our argument, as a reconstruction of the relevant literature, because we represent the collective trend of scientific opinion—our argument has public merits within the science community. In fact, the content of DNC would not challenge the lay reader much more than much popular science journalism; the committee’s methods for assessing different studies and reaching overall conclusions are laid out fairly straightforwardly, and do not include any deep analysis of the different statistical methods (NAS 1982, chap. 3). At the level of content assessment, the real question for the lay reader is whether the committee has actually covered all the relevant studies and reasonably interpreted their import. Boffey challenges NAS’s deferential vision of public merits. In calling for more inclusive committees—for example, for the inclusion of scientists with activist and adversarial positions, for open transactions, and for deeper lay involvement and input in the committee process—he also makes a broader argument for a more democratic, participatory mode of expertise and sciencerelated policy formation. His argument involves two claims, one sociological, the other normative. First, he challenges the sociological assumption behind the Academy rhetoric of scientific credibility. That rhetoric, as described above, presumes that Academy panels, as representative of the scientific public, are politically disinterested. Boffey’s study of a number of prominent NAS reports leads him to the opposite conclusion: the Academy has rarely led the way in addressing societal problems or in challenging the status quo. Rather, “it has often allowed itself to be used as a shield by those intent upon preserving business-as-usual” (Boffey 1975, 247). The “chief reason” for this, he argues, “is undoubtedly the Academy’s orientation toward serving the rich and powerful institutions that do not generally encourage activism in behalf of reform” (ibid., 249). In effect, Boffey challenges the self-presentation of the Academy as representing the disinter-
264 Chapter 8
ested republic of science; the reality, on his view, is that committees typically represent economically and politically interested parties aligned with power. Second, he makes a prescriptive argument, recommending as a solution to the sociological problem that NAS “reorient . . . its relationships with other organizations.” Specifically, the Academy should align itself with the broader lay public. NAS “should designate itself a national ombudsman,” so that “every Academy committee should approach its task as if it were representing the public rather than offering consulting services to a particular agency” (ibid., 261, 262; my emphasis). Along with other recommendations,17 this fundamental shift in the Academy’s institutional position and mission would not make NAS less objective but rather should “enhance its independence, objectivity, and social utility” (ibid., 253; cf. 261)—in effect, enhance the public merits of its reports. Boffey directly challenges the NAS rhetoric of credibility: in his view, the wider lay public (working together with scientists) constitutes the authority for certifying the public merits of expert advice. As a broad claim, Boffey’s prescriptive thesis finds support in the contextualist analysis in chapter 7; it also represents the dominant trend of opinion among STS scholars and “participatory appraisal” researchers (Chilvers 2008; Lengwiler 2008; Jasanoff 2004). Boffey’s sunshine proposal, however, is less convincing from a contextualist perspective—a point that brings us to the second tension. (2) At one level, the conflicting views about confidentiality involve a socialpsychological question, namely whether closed sessions and confidentiality promote the committee’s independence or, on the contrary, provide cover for various sorts of biases and influences to have their way. Behind this question, however, lies a deeper tension between the transactional situation in which committee arguments are produced and the wider publics to which those arguments are addressed. This tension arises from the different demands on the production of social order at these two levels. In generating consensual positions within local settings, participants must manage very specific, local problems of mutual intelligibility and cooperation, problems that stem, at least in part, from the particular personalities of those directly involved. From this local perspective, opening the record of meetings to public scrutiny radically alters the transactional situation: the participants who are physically present must now reckon not only with nonpresent sponsors, colleagues, and the like, but also with an indefinite audience of unknown membership and extent. This shift makes the task of maintaining mutual intelligibility for all participants essentially
265 Three Dimensions of Argument Cogency
unmanageable, inasmuch as the committee members have no effective means of gauging the reactions of, or adequately explaining themselves to, unknown virtual participants. On the other hand, the reactions of sponsors and the like may be all too predictable, leading members to tailor their remarks accordingly—hence the concern that public access would threaten independence and freedom of exchange. Thus, from a transactional standpoint, concealing backstage negotiations and conflicts from direct public monitoring serves to keep the social problems that come with local argument construction manageable. Conversely, the polished, smooth appearance of scientific reports employs a public idiom, a standardized form of presentation whose vocabulary and modes of argument are designed precisely for addressing wider contexts of argumentation, where the wealth of situated, ad hoc resources that help arguers manage face-to-face interaction and ensure mutual intelligibility are simply not usable. Rather, at the level of public argumentation authors must rely on standardized models of, and assumptions about, the intended audiences; such models are built into the Academy’s public idiom as a mode of self-presentation. Although it is plausible to hold that one must assess transactional merits in the evaluation of argument cogency, it is not immediately obvious that Boffey’s sunshine proposal would contribute to better transactions. Such access might not be desirable, and it also might not be necessary for transactional oversight. The discussion in chapter 6 suggests an alternative to public committee meetings, namely anonymous disclosure within reports of points of compromise in the committee (or perhaps even voting tallies, if a committee resorted to majoritarian procedures at particular points). Such disclosures do not go as far as minority reports (which NAS allows) but they apprise readers of transactional realities that potentially affect the judgment of merits. In any case, the integrity and freedom of the transactional process might well require limitations on openness. This conclusion assumes that the broader deliberative process allows for indirect methods of lay oversight and opportunities for lay intervention (e.g, regarding agenda, committee selection, review). 7
Concluding Remarks
I began this chapter with the question of cogency. In reply, I propose a normative framework that directs attention first of all to the different levels or contexts in which we can evaluate arguments for their cogency: the oral or written text
266 Chapter 8
of the argument, which provides the context for interpretations of its content as a rational justification; the local transactions in which, paradigmatically, interlocutors engage face to face in the construction, evaluation, and modification of the argument; and the broader publics for which the argument has potential relevance. The model ties the idea of cogency to the meritorious properties of each of these contexts. This three-dimensional model is similar to the perspectivism of Wenzel and Habermas in that it provides a broad heuristic for assessing the cogency of arguments. Unlike their perspectivism, however, the contextualist model takes different contexts as its leading idea rather than three components (product, procedure, process) or the traditional modes of evaluation (logic, dialectic, and rhetoric). Though the traditional modes are certainly relevant for assessment, how they are relevant depends on the argument domain and level of context. Consequently, the model does not suffer from the misalignments that afflict perspectivism. Nor does the model require every argument to have all three sorts of merit—that too is a context-sensitive matter. Though scientific arguments normally aim at content, transactional, and public merits, some arguments might be inherently “intimate,” incapable of public transmission. I also leave open the possibility that for some evaluative purposes it may suffice to examine only one type of merit, even if we could in principle assess the argument more comprehensively. For example, if the argument is obviously compelling on its face, then transactional and macrosocial factors may not add much, if anything, to its normative evaluation (though they may be relevant for explaining other facts about the public impact of the argument). Other cases might involve contexts of evaluation that lie between local transactions and publics: is a public address in a large auditorium one multiperson transaction, or is it the sum of as many transactions as there are listeners? Perhaps we should regard the address as a distinct sort of context that has certain features in common with both local transactions and complex publics. The CDF collaboration also has distinctive features: a kind of public (of different work groups) that produced a single argument-text. In any case, the proper approach to such intermediate cases depends not on idealizing reconstructions of tacit pragmatic presuppositions but on paying close attention to the actual case and the normative concerns of those involved. The above provisos should make us hesitate to propose a general standard of cogency that would integrate the three types of merits into a single conception that has prescriptive force for any argument. The contexts themselves are
267 Three Dimensions of Argument Cogency
nested, to be sure: texts are normally developed and evaluated in transactions, which in turn are situated in broader social domains. But whether and how these domains constitute publics through which the arguments should travel remains an open question. And the importance of reasonable transactions for generating cogent arguments might also vary. Within science studies, the importance of what I call transactional merits is debated. Some models of science seem to regard transactions merely as points for the origination of arguments whose cogency depends primarily on the public structure of science rather than on the reasonableness of the local transactions.18 The CDF and NAS case studies belie those models, at least as attempts to articulate normative attitudes of participants. But perhaps other cases might show that transactional merits are not always necessary for producing substantively cogent arguments. How likely that scenario is, and in which domains of argumentation, are questions I cannot pursue here. Rather than start with an integrated prescriptive definition of cogency, the critic must delve into the particular case and first become familiar with the normative concerns of the participants themselves and how they attempt to integrate those concerns. In the NAS case one can see all three contexts in play. One can also see how participants employ those concerns. Specifically, NAS scientists invoked transactional merits (within the panel) and public merits (within the science community) as part of the content of their argument before lay publics—hence as a way of constructing an argument that would travel beyond the science community. In their critical assessment of DNC, some CAST scientists scrutinized its transactional merits. Others contested the panel’s claim to public merits within the sciences, that is, the claim that the panel represented the trend of scientific opinion. Lay critics like Phillip Boffey went still further, contesting the interpretation of transactional quality at NAS. His criticism buttresses the argument in chapter 7, that the meaning of dialogical ideals must be established by substantive arguments in context. In a word, the participants themselves, both lay readers and scientists, defenders and opponents, already acted as critics who assessed arguments in light of the three sorts of merits. On balance, DNC seems to have held up reasonably well to such critique. Insofar as one can judge from the CAST report, scientific scrutiny uncovered no single devastating flaw in content or transactional merits (again, the dietary recommendations aside).19 And Boffey’s criticisms of the committee’s transactional structure are open to rebuttal in light of a contextualist approach. Thus the committee seems to have done a passably
268 Chapter 8
good job with the content and transactional merits of their argument. And the relatively benign reception of DNC overall speaks in favor of public merits, assuming a reasonably structured social network. To be sure, I have not provided a full-scale study, so the basis for this judgment remains quite limited. I offer it primarily to illustrate the kind of critical assessment the contextualist approach allows. In making such critical assessments, argumentation theorists enter the fray at the same level as critical participants: the contextualist framework does not bestow privileged status on those who adopt it. But neither does it leave them at the mercy of whatever the participants happen to believe about cogency. Rather it requires them to find persuasive modes of critique such as those presented in chapter 7, critical evaluations that draw on the argumentative resources available from the range of contexts that come together in the discussion. The three dimensions serve as a heuristic in this endeavor, helping critics notice the various normative and empirical demands on argumentation-incontext and their potential interrelationships—demands connected with the topic-specific aims of argumentation, with transactions conducted within particular institutional frameworks and involving specific social-psychological dynamics, with lines of potential relevance for broader publics, and so on. Because these normative demands always have their full sense and force relative to some context, whether that of the local occasion, a given discipline or institution, or the democratic public, their reasonableness is tied to one or more contexts. As I noted at the end of chapter 7, this unavoidable contextspecificity does not exclude the possibility that some demands are “absolutes” in the sense of holding in every context, or at least every context at issue. Other demands, though they lack such universality, might not be reasonably in dispute in a given case. Such demands, universal or uncontested in context, provide the critic with standards that can ground reasonable critical assessment in context. The real challenge for critique arises when standards are contested, either in their relevance or proper interpretation—as in the public critique of NAS procedures. The critic must then find a basis for taking a justified stand on a particular interpretation of merits, or on the cross-contextual relevance of certain transactional standards, and so on. As evident with Habermas’s critical theory (see chapter 4, sec. 5), the necessary justification of critique can involve a series of deeper issues. Of these, the most challenging, I suspect, is the vision of science in society.
9 Critical Science Studies and the Good Society
The previous chapter opens the door to a range of further questions. Most point to issues beyond the scope of this book, but I hope to address some of them at least briefly. Before doing so, however, I recapitulate the overall thrust of my argument and its implications for interdisciplinarity (sec. 1). I then take up some further issues that serve to further clarify certain aspects of the critical contextualist model (sec. 2), and I close with some tentative explorations of the more difficult questions that have been lurking in the background since chapter 7 (sec. 3). 1 Recapitulation: A Critical Contextualist Answer to Kuhn’s Gap This book has been motivated by a series of challenges connected with postKuhnian science studies. At a broad level, we confront the challenge of understanding the rationality of scientific knowledge-production as it actually operates, namely in the form of materially conditioned social practices whose reality belies simplistic ideals of rational method. In response to this broad challenge, I have approached scientific inquiry as a socially embodied constellation of argumentative practices. The leading questions then turn on a normative conception of cogency: what makes scientific arguments cogent and how ought we to assess that cogency? The increasing number of science-intensive policy and health issues that confront us today at the public and personal levels give these theoretical questions a contemporary urgency. The opposing answers that science studies scholars have given to these questions followed fault lines discernible in Kuhn, fueling the rationality debates and subsequent “science wars” that pitted disciplines against one another. In
270 Chapter 9
Part I, I analyzed those developments from the standpoint of argumentation studies, which led me to identify two specific challenges posed by “Kuhn’s Gap.” These defined the chief tasks for subsequent chapters. At a conceptual level, Kuhn’s Gap calls for a comprehensive conception of cogency that can integrate the prescriptive perspectives favored by philosophers and the descriptive perspectives of social scientists. At a disciplinary level, the integrated model of cogency should provide a framework for fruitful interdisciplinary exchange and cross-fertilization. As Klein’s remarks on interdisciplinary holism imply, the two tasks are linked: a comprehensive conception must not make overly strong claims of conceptual unification if it is to serve the goal of interdisciplinary cooperation. In Part II, I examined Habermas’s discourse theory as a possible response to the conceptual and interdisciplinary challenges posed by Kuhn’s Gap. Extended to the area of scientific argumentation, Habermas’s discourse theory provides a comprehensive integration of analytic perspectives on argument evaluation. However, he attempts to achieve such integration within a single philosophical theory of cogency that is anchored, finally, in a set of universal process idealizations together with the idea of objective truth. Although Habermas’s theory is multidimensional and captures important features of scientific practice (as the case study in chapter 6 shows), it makes an overly strong claim of conceptual unification. Consequently, the range of perspectives and commitments in science studies puts his theory under considerable strain. These strains emerged in the hard case of cooperation examined in the second postscript: Habermas’s approach allows cooperation with relativistic SSK theorists only in virtue of a method of philosophical avoidance analogous to that advocated by Rawls for pluralist democratic deliberation. Moreover, his conception relegates the actual organization and conduct of scientific argumentation to secondary status, thereby creating problems for understanding the rationality of real practices. Again, the basic difficulty lies in the fact that Habermas grounds the rationality of scientific argumentation in an idealized theory of dialogical process, which has normative priority: although scientific arguments have a logical structure that justifies the conclusion in an immediate sense, the strength of a justification is measured against a set of dialogical ideals whose satisfaction we can never definitively establish by examining the actual process. The latter can at most warrant a defeasible presumption of dialogical adequacy, and hence of the justification of argued claims.
271 Critical Science Studies and the Good Society
To be sure, dialogical ideals like Habermas’s are certainly relevant for participants in scientific argumentation—those ideals provide means by which participants make sense of their argumentative practices and account for the cogency of their arguments. But by defining them as counterfactual idealizations that transcend the observable realities of practice, Habermas creates difficulties in understanding their reasonable contextualization. To deal with such difficulties, in Part III I attempted to rethink his ideals along more thoroughly contextualist lines inspired by ethnomethodology. The key move involves a shift from the traditional evaluative perspectives (logical, dialectical, rhetorical) to a context-oriented framework. Rather than start with the idea that a cogent argument must satisfy a specific set of logical, dialectical, and rhetorical standards, I subordinate the traditional perspectives to the different levels of context that condition the meaning and relevance of the standards those perspectives highlight. As a result, the social context and process of argumentation are not simply an afterthought to an ideal conception of dialogue (toward which social reality must strive); rather they lead the way in the assessment of cogency. One understands the meaning and relevance of dialogical ideals precisely through the level of context for which they have relevance. One thus understands ideals substantively, in their rhetorical, context-specific senses, from case to case. Similarly, the idea of a single objective world, along with the specific pragmatic commitments entailed by making truth claims about that world, also acquires an indexical component, such that its specific meaning shifts with differences in disciplinary context. Notice, however, that the presupposition of a single world also helps bridge different contexts of discourse: if participants in two contexts take themselves to be referring to the same world, then they must regard their claims as consistent in principle—at least in some broad sense of compatibility. By shifting to a contextualist framework, the conceptual oppositions I described in chapter 2 are tightly integrated without falling into an overly strong philosophical unification. For the contextualist mode of integration does not rest on a single multidimensional definition that holds for all cases; rather, it involves a heuristic commitment to examine the context-specific ways in which the three kinds of merit potentially interact. Thus to understand the content merits of a given scientific argument, we must attend to the transactions in which arguers interpret those arguments, and that analysis in turn requires us to understand the different publics for which those contents are
272 Chapter 9
potentially relevant. Logical standards of various sorts are certainly an important measure of content merits, but to understand which standards, and to see how they function, one must situate the argument content in its relevant social contexts, taken here in a broad sense that includes both the local transactions and the broader disciplinary domains. When scientific arguments are addressed to broader lay publics, content merits again shift in their substance: now we must see how well scientists and their spokespersons (journalists, etc.) have translated technical arguments without losing the content they consider important to communicate. What counts as a logically strong argument thus depends on the range of relevant social contexts to which the argument is addressed and within which it might be relevant. Dialogical ideals are likewise linked very closely with the social contexts in which they are relevant. I have offered a number of examples in which disciplinary publics constitute the relevant contexts for understanding these ideals. In chapter 7, I noted how ideals of inclusiveness and equal voice can be brought to bear in the feminist critique of scientific disciplines. Such critiques target problems in the public merits of scientific theories: critics point out how the underrepresentation of women in a given discipline (e.g., primatology) has a detrimental effect on the cogency of its arguments. Such criticisms attempt to make dialogical ideals rhetorically effective by linking them with detailed substantive arguments that draw on domain-specific features and goals of the science in question. The CDF case might also be analyzed in terms of the interplay between dialogical ideals and public merits. The collaboration itself constituted a kind of public comprised of diverse subgroups associated with different experimental methods and procedural roles. The scientists established a set of discursive procedures for structuring dialogically meritorious transactions among members in the collaboration. However, in writing a single Evidence paper as the consensus position, the group made the strong claim that the paper’s argument had been fully accepted by each of these groups—a claim that it enjoyed public merits within the collaboration as a complex set of transactional locales and occasions. CDF members seemed to have different views about the appropriateness of that claim. If it did rest on a pressure to compromise, then there are grounds for criticizing the paper as misrepresenting the public merits of its argument by suppressing deficits in particular transactions, that is, those involving scientists who agreed to some parts of the paper only under pressure. But unlike Habermas’s model, the contextualist model may not stop with that
273 Critical Science Studies and the Good Society
immediate negative assessment based on the bare idea of public merits. Rather, one must also analyze the broader contexts and publics to which the paper was addressed, asking whether the paper violated the norms and goals of communication within the sciences. The potential “violation” here must be assessed contextually: given the need for ongoing cooperation in research collaborations on the one hand and the norms and goals of knowledge-production on the other, can public communication in the sciences tolerate the level of compromise found in the Evidence paper? Chapter 8 provided yet a further example, in which NAS advertised the dialogical quality of committee transactions as providing grounds for the public merits of its claims across the relevant scientific disciplines. The NAS procedures were designed and presented as grounding the confidence that committees represent the republic of science. As an explicit part of the NAS report, the procedural claims entered into the content of the report’s argument as a kind of “indirect evidence” for the quality of the substantive claims regarding diet and health, which were based on direct evidence garnered from a survey of empirical studies (see Staley 2004b). Thus the report represents a case in which public and transactional merits were closely intertwined in a manner that entered into the content of the argument, at least at the margins (e.g., in the preface). In other case studies, analysts have focused more closely on what I call transactional merits. Such studies are often critical, bringing dialogical ideals to an assessment of the transactions in question. One may read Mara Beller (1999) along such lines. She closely scrutinizes the transactions of the small group of theoretical physicists who engaged in the development of quantum mechanics. Specifically, she critically examines the microprocesses in which individuals constructed and reacted to arguments under the powerful influence of Niels Bohr, who had the status of an authoritative father figure. According to Beller, Bohr’s powerful aura discouraged physicists from making sober assessments of his arguments. In my terminology, a bias within the group undermined the reasonable quality of the transactions in which readers engaged with Bohr and his arguments—to the detriment of the development of quantum theory. Beller links this deficit in transactions with a critique of the content merits of Bohr’s arguments. Although his arguments were accepted, Beller disputes both their content merits and transactional merits. The literature on the search for the structure of DNA provides a second example of transactional troubles. Here too one can see dialogical ideals used
274 Chapter 9
in the critical assessment of the transactions among the scientists involved in this quest as it unfolded in two laboratories in England in the early 1950s. The transactions between male scientists and Rosalind Franklin have been assessed in various ways. Some see Franklin as the victim of male bias (see Delamont 2003); others, such as her colleague Maurice Wilkins, identified personality differences or simple failures to communicate (see McElheny 2003, 39; Judson 1996, afterword 1). Prelli (1989a, 250–256) has argued that Watson and Franklin clashed because of different research temperaments. Each assessment identifies transactional deficits, which affected not so much the fate of knowledge as the fates of individual scientists. The above examples are meant to illustrate some of the possibilities of conceptual interdependence that tie traditional logical and dialogical standards to different levels of context in particular cases. In these cases, the contextualist model requires us to connect the sorts of rational standards emphasized by philosophers with the different social contexts studied by social scientists and others. In emphasizing these connections, the contextualist approach goes beyond Habermas in reconciling the conceptual oppositions in Kuhn. By integrating conceptual oppositions in context rather than through a priori philosophical unification, the contextualist model also promises to go further than Habermas’s theory in fostering interdisciplinary cooperation. Habermas’s strong commitment to universal ideas of reason poses a potential obstacle for cooperation with SSK theorists who understand themselves as relativists. As I argued in the postscript to Part II, there are ways of dodging that obstacle, which allow critical theorists like Habermas to engage in casecentered cooperative projects with relativistic sociologists. The contextualist model, by contrast, starts with the indexical character of the ideas of reason. Although I do not believe this commits the contextualist to relativism, it also does not directly challenge the relativist at a philosophical level. One does not start with the idea that a given norm of rational argumentation has universal a priori validity for all cases and therefore applies eo ipso to the case before us. Rather, ideals of argumentation must prove their relevance in context for the participants themselves. This starting assumption is much closer to an empirical social-scientific approach; indeed, the critical theorist needs the help of empirical sociological methods and input in order to understand certain context-specific features of the case at hand. A contextualist approach to critical assessment thus requires some level of interdisciplinarity with other science studies scholars. Interdisciplinarity can
275 Critical Science Studies and the Good Society
take at least two forms. The more common we might call “unilateral,” exhibited when a theorist in one discipline appropriates findings and ideas from other fields and incorporates them into his or her own approach. More challenging is the “multilateral” interdisciplinary cooperation that involves collaborative projects of one sort or another, from discussion groups and multidisciplinary conferences, through occasional consultations and joint papers, to interdisciplinary research teams with an ongoing shared agenda that reshapes the self-understanding of the member disciplines (see Klein 1996). One can find both types of cross-fertilization within and between science studies and argumentation studies.1 Thus critical contextualism, as a context-sensitive critical science studies (CSS), is deeply committed to interdisciplinarity.2 Indeed, the present study has benefited from both unilateral and multilateral interdisciplinary engagement (see chapter 8, note 5). But contextualist CSS also goes beyond SSK in certain ways. Some of these should already be evident—for example, the rejection of a principled relativism or an ethnomethodological indifference. Here I want to note two further characteristics of contextualist CSS that distinguish it from some forms of SSK. First, unlike some approaches (e.g., Woolgar 1983, 1988) it normally prohibits a skeptical or ironic attitude toward the normative concerns of the scientists involved in the case under investigation. Rather, it has us assume that scientists are normally those with the background and skills for assessing the content merits of technical arguments addressed to expert audiences. In contexts involving lay audiences, CSS requires us to understand the situated local methods and public standards to which lay participants subscribe. In adopting this attitude as a starting point, critical theorists are similar to ethnomethodologists. This sympathetic attitude is a starting point, however, not a final destination. The critical orientation in CSS marks the second feature that differentiates it from nonprescriptive and acritical versions of SSK. CSS does not exclude the possibility of eventually criticizing technical arguments and scientific methods on their own grounds—thus it does not exclude the possibility of scientists and others becoming critics of disciplinary methods and findings, nor the possible relevance of critical lay arguments.3 Nor does it stop with the reasoning methods of lay audiences.4 Contextualist CSS commits us to making critical assessments in light of context-specific demands, but that does not put critics entirely at the mercy of whatever participants happen to think is reasonable in their context. That context, after all, typically implicates a range of wider
276 Chapter 9
contexts and social domains, from which critics can draw in ways that make sense in and transform the initial context. Thus, a contextualist CSS welcomes interdisciplinary collaboration, but it also does not attempt to hide its critical commitments that differentiate it from SSK and pose potential sites for theoretical debate. Such debate involves a different sort of interdisciplinarity from the two forms above, both of which focus on a cooperative critical assessment of cases. Theoretical debates that characterize this third mode of interdisciplinarity deal with deeper philosophical differences that separate science studies scholars in their respective understandings of cogency and approaches to the study of scientific argumentation. For such debate to be fruitful, one should not assume that the field of science studies must converge, in the end, on a single comprehensive framework that resolves all the theoretical and methodological conflicts. Rather, one must enter the fray of contention, staking out one’s basic theoretical and practical commitments. Those commitments involve more than specific methods of analyzing and appraising scientific argumentation, however. The most important questions for science studies scholars, particularly those engaged in contextualist CSS, concern (1) the critic’s sense of the most pressing and important problem situations at a given time in history and (2) the critic’s vision of science and its relation to society. These are the larger issues that I want to address briefly below. Here I note that the interest in interdisciplinarity shapes the manner in which one pursues these broader questions and develops one’s commitments. The idea is not so much to construct a theory that can win the day against other theories as to elaborate a range of cooperative projects with other positions in the field—alliances and collaborations that are available as possible contributors to critical assessment. To accomplish this one must have a sufficient sense of one’s own theoretical standpoint and its limitations, as well as a willingness to meet other positions halfway insofar as one can enlist them in the critical program. If we view the field of science studies—or more broadly, science and technology studies (STS)—as ultimately a practical endeavor, oriented toward socially and politically relevant models of the complex interactions that link science, technology, and society today, then in a certain sense STS scholars are all in the same boat, arguing about the three issues above: how to approach cogency, which problems deserve primary attention, and what direction science and technology ought to take in today’s society.
277 Critical Science Studies and the Good Society
In debating these questions, critical argumentation theorists engage in a process of theoretical exchange at once agonistic and cooperative. On the one hand, to develop plausible responses to the three issues one must critically evaluate the range of initiatives in STS for their strengths and weaknesses, and one must argue for and against different models—the agonistic mode, or what Steve Fuller (1993) has associated with the dialectical side of interdisciplinarity. By taking a critical, discriminating approach toward alternative models in STS, one further defines one’s own stance as a critical argumentation theorist. On the other hand, the aims of interdisciplinary exchange and collaboration give rise to a more constructive, dialogical mode of exchange, which Fuller (1993, 1995) links with rhetoric as the art of forging apt partnerships at the propitious moment. Here one aims not so much to destroy opponents or reduce them to silence as to enlist them as allies in the critical project, even if only temporarily and for limited purposes. For this mode of engagement, one’s theoretical position counts as stronger, as based on the better argument, insofar as it allows for a greater range of collaborative endeavors without compromising the basic integrity of the critical project. 2
The Prescriptive Character of Critical Contextualism
The prescriptive force of the contextualist model of cogency lies in the first instance at a procedural level. The model, that is, prescribes a context-sensitive methodology for critique. But it leaves open what specific standards and arguments count as relevant and cogent at the three levels of context—the standards that define content, transactional, and public merits. As a model of cogency, contextualism holds that cogency is generated within contexts rather than brought to contexts by a priori norms that have prescriptive force in themselves. At first glance, then, the prescriptive force of the model primarily lies at a higher, metacritical level, where it operates as a guide for making specific normative models of logic, dialogue, and rhetoric context-sensitive. In effect, this was my procedure in modifying Habermas’s dialogical ideals in chapter 7. If one must put the critical contextualist approach into a single integrated statement, then let it be a procedural statement such as the following, which summarizes the basic moves involved in critically evaluating scientific arguments for their cogency, on the defeasible assumption that all three sorts of merits normally come into play:
278 Chapter 9
Cog(A): To assess the cogency of an argument A, (1) start with the context of origin CO: assess the content, transactional, and public merits of A as it is interpreted in CO, according to (a) the logical, dialectical, and rhetorical standards relevant in CO, and (b) the goals of scientific argumentation in CO; (2) evaluate the broader public merits of A: ask whether there are further relevant contexts CR for assessing the cogency of A (e.g., related scientific disciplines, technological and medical contexts, interested lay publics); if there are, then assess A according to the standards and goals relevant in CR; (3) situate the critique, and settle conflicts between (1) and (2): ask whether the relation between science and society, or the goals and problems currently relevant for science in society, call for assessment of A from the standpoint of further evaluative contexts CE. If the answer is yes, then assess A in relation to standards relevant in CE. The first step is immanent to the context of origin, which includes the local laboratory transactions and the broader contexts or publics—the problem area or subdiscipline—explicitly addressed by arguments produced in those transactions. Though modestly external vis-à-vis the context of origin, step (2) involves moves that participants in CO typically expect and in some cases even welcome, for example, when scientists have an eye on possible technological applications of their work. Step (2) also becomes important in interdisciplinary research and controversies and in the communication of technical arguments that have relevance for interested lay publics: policymakers, health-care personnel, private citizens deciding about diet regimens, and so on.5 Introductions to technical scientific articles sometimes allude to these additional contexts of relevance, but normally such articles primarily target the immediate disciplinary contexts relevant for step (1). Expert reports, by contrast, represent a complex case: a context of origin (for the report itself) whose members work at the boundary of steps (1) and (2). We should not be surprised to find that evaluations in the first two steps sometimes diverge: arguments that emerge as cogent from the immanent perspective in step (1) can prove wanting in (2).6 This conflict necessitates step (3), in which the critical theorist makes the broader model of science and society explicit, on the assumption that the conflict between (1) and (2) calls for reflection on deeper questions regarding the aims and methods of science, their social positioning, and so on.
279 Critical Science Studies and the Good Society
We might regard the argument over NAS procedures in chapter 8 as an example of a conflict among contexts. The NAS committee employed an interpretation of dialogically adequate process that was widely accepted among scientists: though some CAST critics challenged its execution, they accepted the basic idea. As representing the republic of science in miniature, the committee assumed that the appropriate demos for evaluating the scientific evidence was technical, constituted primarily by the different disciplinary perspectives. The committee thus accepted the ideals of scientific dialogue operative in scientific contexts and carried them over to its own transactions, the immediate context of origin for its report. In doing so, they assumed its procedural ideals would carry weight for wider lay publics as well: the further contexts of relevance they aimed to address.7 Boffey’s critique challenges NAS assumptions with an alternative interpretation of adequate dialogue. At stake are two different interpretations of dialogical ideals, understood from different contexts: one the context of origin, namely the panel with its need for confidentiality and a balance of scientific perspectives, the other that of an interested public concerned about the misuse of confidentiality to ensure institutionally “safe” reports that do not rock the boat. Sheila Jasanoff (1990, 21) poses the relevant question for such disagreements: “who should be designated as final arbiters of quality?” She refers here to science-intensive policymaking whose content and procedures become controversial. All three sorts of merits appear to be at stake in the controversies she studies: the content of regulatory science, the procedures governing expertcommittee transactions, and the public acceptability of the regulatory outcome. Rephrased in the framework of contextualist CSS, her question asks, what context of evaluation should provide the authoritative standards and goals for assessing these merits? When actors in different contexts reach competing evaluations of these merits, what authoritative basis do we have for a critical adjudication of the debate? Behind Boffey’s criticisms, then, lurk larger issues about the appropriate context of evaluation: how science is positioned in democratic contexts. Whereas the panel operated with a technocratic model, assuming that policy recommendations are the preserve of scientists and bureaucrats, Boffey proposed a more open and inclusive democratic model. Does contextualist argumentation theory, understood as a metacritical framework, help decide the question? Or does an answer to Jasanoff’s question rather require critical theorists to move from metacritique to critique, looking somewhere beyond the contextualist model for a set of nonarbitrary resources for adjudicating conflicts of the sort she documents? In brief, “yes” and “yes.”
280 Chapter 9
Yes to the first question, inasmuch as the metacritical framework summarized in Cog(A) offers some guidance for both the substance and process of critical assessment (e.g., my reply to Boffey’s recommendations in chapter 8). Yes to the second inasmuch as metacritique does not fully determine critical assessments. An answer in full to these questions would take us into the vast literature on policy-relevant science, science policymaking, expertise, and democracy, all bearing on the “contract” between science and society. That topic puts us at the edge of my project, which I can only reconnoiter with a view to clarifying some substantive and process implications of the contextualist approach.8 In the present context, when I speak of the substance of critical adjudication, I mean the specific logical, dialectical, and rhetorical standards one employs to respond to Jasanoff’s question (thus my use of “substance” here covers both “substantive” and “procedural” standards that are often distinguished within each perspective). Negatively, contextualist CSS proscribes the attempt to seek that content simply by appealing to a transcendental context of evaluation or set of a priori universals. Indeed, the debate between Boffey and NAS turned precisely on the proper interpretation of the kind of dialogical ideals on which transcendental models typically rely. But the prescription against transcendental arguments does not preclude the possibility that resources exist within the web of contexts shared by Boffey and NAS—which include empirical studies of how expert committees operate, the likely sources of bias, and so on. Rather than seek a transcendental foundation, context-sensitive critical theorists seek the bases of authoritative critical adjudication within the web of contexts and shared beliefs that are unproblematic for the case at hand. We already have an example of this approach in feminist philosophers of science, whose critique of the social organization of research connects dialogical ideals with substantive, discipline-specific arguments. By the process of critical adjudication, I mean the kind of argumentative process participants ought to consider appropriate for reaching an adequate critical position, that is, a cogent argument. In chapter 7, I briefly noted that critical contextualism favors a deliberative democratic approach to controversies involving the relation between science and society. But that statement suggests too narrow an idea of deliberative democracy for my purposes—as though we were speaking mainly of public debates about the direction of science, regulatory science, and the like. I will say something about such debates below (sec. 3), but first I want to expand the idea of deliberative democracy into a flexible heuristic able to capture the shifting senses
281 Critical Science Studies and the Good Society
of “demos” and “deliberation” for different sorts of controversies and arguments. Although a number of STS scholars have called for a more democratic science and technology (see, e.g., Fuller 2000b; Kleinman 2000; Sclove 1995), the heuristic expansion of “demos” I propose here is simply a metacritical requirement to specify the scope of a demos by the relevant contexts for which the controversial issue or claim is potentially relevant. Thus the process of argumentation should involve deliberative exchanges in which members from the relevant contexts strive to overcome conflicts between specific, contextdependent standards. The kind of controversy provides an initial specification of the relevant contexts. To illustrate the idea, I close this section by considering interdisciplinary scientific controversies, which remain largely within the demos of the natural sciences. These exhibit a structure similar to the conflicts Jasanoff analyzes: conflicts between contexts of origin CO and obviously relevant contexts CR (i.e., steps 1 and 2 in Cog(A) above). In such debates, the former are those contexts in which a given argument originates, the latter are those further (sub)disciplines to which the claim is addressed beyond the discipline of origin. Substantively, the contextualist model requires critical theorists to develop their evaluative stance in light of all the relevant contexts of evaluation (which might include further contexts CE, which go beyond CO and CR). The dinosaur extinction controversy provides a good example of an interdisciplinary controversy, as the prominent disputants included nuclear physicists, archaeologists, and volcanologists, among others (see Alvarez 1997; Officer and Page 1996; Glen 1994). The controversy can be traced back to a 1980 article in the interdisciplinary journal Science (Alvarez et al. 1980). The article crossed disciplinary boundaries: though its lead author was a physicist (and the second author a geophysicist), the article made the paleontological claim that linked the extinction of the dinosaurs to a massive meteor impact. The ensuing debate was complex, involving disagreements over how to interpret geological evidence (e.g., whether “shocked” quartz was evidence for an impact) and how to weigh not only conflicting pieces of evidence but also different gaps in evidence. Discussion was hindered by prejudices regarding disciplines: the prestige of physics and disdain toward some disciplines seem to have played a problematic role in the debate (Officer and Page 1996, 9ff., 77–81). Moreover, scientists’ knowledge of other disciplines tended to be out of date, limited to stereotypes acquired in their education or through the popular science press (Clemens 1986, 1994).
282 Chapter 9
Without delving into detailed analysis, one can see in this controversy at least three types of obstacles in the attempt to move an argument from its context of origin to other contexts. First, there are differences in disciplinary ethos and the ascription of scientific competence, as some of the disdainful attitudes of physicists toward paleontologists revealed. Such differences have to do with differences in the context-specific rhetorical standards relating to competence. Further obstacles involve differences in dialectical standards, in particular the “substantive dialectical factors”—background assumptions and the like—that inform acceptable scientific argument. Scientists not only tend to misunderstand (or to operate with outdated conceptions of) the substance and methods of other disciplines, they can also encounter strong ontological assumptions that hinder cross-disciplinary exchange. Thus the proponents of the impact hypothesis ran up against an entrenched predilection in paleontology for uniformitarian (i.e., gradualist) models of geologic and evolutionary change (Alvarez 1997, 48–59; Raup 1994). Third and finally, the controversy shows how scientists in different disciplines weigh different sources of evidence differently, most likely preferring evidence from their own discipline—the kind of empirical bias that Solomon (2001) has noted. More generally, it is reasonable to assume that in interdisciplinary scientific controversies each of the disciplines has somewhat different, context-specific standards for the content merits and transactional merits of arguments. If that is so, then it would seem to count in favor of a hypothesis H that the evidential arguments for H hold up across these different contexts. But one must also ask whether journalistic influences helped or hindered interdisciplinary exchange and understanding in the development of positions that enjoyed (or lacked) public merits across disciplines. To pursue critical assessment in such controversies, one might move through a series of steps. First, one might ask whether arguments developed and accepted in one context—thus judged as cogent in that context according to its standards—were packaged for travel in a way that both preserved content and allowed them to receive a fair hearing in other contexts. (This assumes that evidence can be tailored to meet standards of cogency in multiple contexts without distortion of the argument content.) In the course of critical analysis one might discover missed opportunities, inept transactions, arbitrary disciplinary prejudices, and animosities that unnecessarily exacerbated conflict. Conceivably one could discover new arguments that might advance the debate.
283 Critical Science Studies and the Good Society
Suppose, however, that standards differ in deeper ways across conflicting contexts. In that case, the contextualist model has the critic take a further step and address the question of adequate argumentative process, asking whether scientists engaged in interdisciplinary dialogue on their methods and relevant standards for assessing the merits of their arguments. I suspect that within the natural sciences, such dialogue would primarily involve settling differences in assessing the content merits of arguments. The transactional merits of such exchanges are also important, but I suspect that transactional standards are less likely to be under dispute than they are in politicized regulatory contexts involving expert advice. As part of their dialogue over standards, participants might also have to mediate disciplinary differences in the indexical meaning of truth or empirical adequacy. The contextualist model assumes that the primary goal of scientific argumentation is truth, understood in the pragmatic sense of representationally (or empirically) adequate knowledge of the objective world. As I argued in chapter 7, however, the pragmatics of truth—the exact sense of objectivity and the mutual accountability one incurs in making truth claims—can vary from discipline to discipline. If such differences become a point of contention, then interdisciplinary dialogue would have to find some kind of mutually acceptable integration of perspectives: participants would have to agree on the pragmatic satisfaction conditions implicit in regarding a claim as objectively true. A full discussion of this point is beyond the scope of this chapter, but I suspect a key question concerns the kind of replication or confirmation a hypothesis implies is possible, given its status as a claim about the objective world. A third critical step arises when the critic suspects that further contexts have relevance, beyond those obviously involved in the interdisciplinary controversy. This step becomes necessary when a context or perspective appears relevant for a given scientific discussion but has been excluded from that discussion. When interdisciplinary debates remain within the sciences, this step can arise when someone observes that some area of research has implications for the issue at stake but has been neglected by participants. It can also arise when debates about standards of evaluation lead to reflection on the nature of science and acceptable methods, the difference between “good” and “pathological” science (Officer and Page 1996). As Kuhn observed, these debates tend to arise in revolutionary periods, that is, when deeper methodological assumptions are at issue. However, contextualist CSS approaches such
284 Chapter 9
controversies with resources not available to Kuhn—above all, a more sophisticated model of context-sensitive argumentation and a pragmatic understanding of truth and objectivity. 3
At the Boundaries: Between Science and the Good Society
The most difficult questions that call for a third phase of critical evaluation arise at the boundaries of science and society. Here we find Boffey’s critique, and here Jasanoff raises her question of “who arbitrates” conflicts between contexts of evaluation. Here too we see nonepistemic values emerge within the very content of science, as I already mentioned in chapter 7, section 6, in the discussion of inductive risk. I close with some provisional remarks on these boundary contexts. First I examine contexts of regulatory science (1) and then return briefly to the controversy between science and religion, with which I opened this book (2). (1) Science-intensive policymaking typically relies on expert committees whose members come from a range of disciplines. Thus the critical analysis of expert committees involves, to begin with, the steps described above for interdisciplinary discussion. Jasanoff’s (1990) study of regulatory science at the Environmental Protection Agency and the Food and Drug Administration confirms the importance of adequate argumentative process for such committees, that is, the importance of building consensus in the science community on the interpretation of evidence. Otherwise the committee lacks scientific credibility, rendering it vulnerable to charges of bias or “sloppy science.” Consensus building in turn is fostered by a nonadversarial process in which committee members develop a capacity for negotiation and compromise on what counts as substantively reliable science: a readiness to accept “incremental adjustments among divergent scientific viewpoints” (ibid., 234). For this process, it is important that the committee include representatives from all the relevant disciplines. Scientists themselves take a third critical step rather straightforwardly when they object to a committee for its failure to include a particular disciplinary expertise they judge relevant to the topic. Inadequate representation was one reason for the poor reception of the 1980 diet and health report (see chap. 8; Hilgartner 2000, 89–90). So far we have a picture of the expert committee as a case of interdisciplinary scientific dialogue writ small, the republic of science in miniature. From
285 Critical Science Studies and the Good Society
that standpoint, a further feature of many such committees makes sense, namely the kind of confidentiality rules we saw in chapter 8. Jasanoff (1990, 236) explains this feature in terms of “boundary work”: credible expert advice depends not only on consensus and negotiation but also on securing a privileged discursive space as one to which “authorized personnel” alone are admitted. By screening the public from its internal compromises and presenting itself as a neutral judge of evidence, a committee makes it more difficult for critics to “deconstruct” its findings as merely a product of partisan interests. This kind of boundary work builds on assumptions that, according to Jasanoff (1987, 199), all the stakeholders in policymaking tend to accept: that science should be politically neutral, that scientists (in consensus) are the best judges of good science, and that scientists should stay out of policymaking itself. The 1982 NAS report on diet and cancer was relatively successful in the tasks of consensus building and boundary work. From a critical contextualist standpoint, however, difficulties arise at this point. To begin with, expert committees sometimes make policy “recommendations.” The diet and cancer panel was charged with making dietary recommendations; in other cases, experts are in effect asked to judge whether the evidence warrants regulatory action.9 Granting experts some say in policy formation is probably motivated, at least in part, by the wide recognition that regulatory science cannot be entirely separated from policy considerations. That does not mean that no difference exists between scientific truth claims and practical decisions, or that all science is inherently or necessarily political. Rather, these two realms intersect in policy-relevant science in a complex manner that resists attempts to draw a precise boundary. This is especially the case in areas of risk assessment, which can blend science and policy even when scientists do not make explicit recommendations. The fuzziness of the boundary here stems from factors involving inductive risk: the inconclusiveness of the evidence on some questions, the contestable use of secondary evidence (e.g., extrapolation from animal studies to human health risks), the decision about how conservative an approach one should take, whether action should be based on a risk-averse attitude or not, and so on.10 However, because these interpretive and methodological decisions are not dictated by science alone, they already involve an element of policy. Although methodological policies are not necessarily inherently political, they do involve nonepistemic values. Consequently, they often function as lightning rods for politically and economically motivated controversies (see Jasanoff 1990, chaps.
286 Chapter 9
6, 9). It is thus more accurate to say that methodological choices are “contingently political”: they can have political implications that are contingent on the circumstances. For example, if regulators decide to accept animal studies as evidence for human cancer risks, then they take a risk-averse approach that will likely saddle businesses with economic costs that might be avoided by a different method of risk assessment. The public policy implications of scientific advice make such technical decisions relevant not simply for policymakers but for broader publics. Sometimes even soberly factual claims can have a political dimension. For example, given the associations between diet, culture, wealth, and gender, factual correlations between diet and health have serious consequences for business, medical practice, and lay publics (see Campbell 2004). Thus, contextualism as a metacritical norm requires that the process of policymaking somehow include these further contexts. At the same time, the contextualist model also supports Jasanoff’s recognition that successful policymaking is aided by nonadversarial relationships between experts and bureaucrats working behind closed doors. Given the manner in which scientific and regulatory contexts overlap, it helps if members—both science advisors and administrators—can cross boundaries in the effort to reach policy-relevant conclusions (see Jasanoff 1990, chap. 5). Given the uncertainties of data, requisite compromises, and other transactional particularities involved in such exchanges, confidentiality rules make sense. As Jasanoff (1987, 198) observes, when internal compromises come to light in politically charged environments, critics tend to overstate the socially constructed character of science advice for political ends. Thus it is not entirely surprising that Jasanoff concludes her 1990 study by noting that U.S. regulatory science involves a “creative dialectic” that oscillates between technocratic and democratic models of the relation between science and society (1990, 228). How one actually structures deliberative policy-formation to reflect these legitimate concerns depends on a range of contextual factors (ibid., chap. 11), as well as ingrained national temperaments, with their associated “civic epistemologies,” that is, their culturally specific modes of evaluating and appropriating public knowledge (Jasanoff 1991, 2005). She thus denies there is one best procedure for effective regulatory science. Jasanoff’s findings confirm the contextualist preference for case-specific critical analysis over sweeping prescriptions. However, contextualist CSS—as well as Jasanoff’s more recent work (see her 2003)—also calls for a deliberative democratic framework for policy-relevant science, thereby challenging the
287 Critical Science Studies and the Good Society
technocratic side of the “creative dialectic.” Precisely because such science is laden with policy implications and produces findings with relevance for broader lay contexts, the scientific arguments confront a normative, argumentationtheoretic demand that they be able to travel beyond technical contexts and prove their broader public merits in interested lay contexts. Contextualist critique thus rejects technocratic and elitist models that altogether exclude substantive lay participation in administrative policy formation. But that broad demand for inclusion does not settle the further question regarding the appropriate venue and structure of inclusive deliberation. Unlike interdisciplinary controversies within the sciences in which all parties are supposedly oriented toward empirical truth, policy-relevant scientific arguments move across fundamentally distinct cognitive domains: in the journey from narrowly empirical to broader practical-political contexts, one moves from realms of argument concerned primarily with what is the case to realms concerned with what we ought to do. Again, to distinguish the practical and empirical domains does not mean that their boundary is sharp. Rather, the distinction highlights a deep shift, not simply in the standards of argumentation but also in its goals and circumstances. Consequently, contextualist democratic inclusion does not necessarily mean that lay publics must participate in the direct evaluation of the empirical merits of scientific argument. And as I argued in chapter 8, contextualists need not immediately accede to demands for more open expert panels; rather the practical details of inclusion depend on a comparison of alternative divisions of institutional labor. To be sure, there may be cases when such direct inclusion is appropriate, for example when expertise depends on local nonscientific sources of knowledge, or when lay concerns affect methodology.11 However, these modes of participation are possible because lay participants are sometimes able to establish the relevance of their contributions for argumentative goals that experts either share or can accept as part of their technical practice. In drug testing, for example, ethical concerns legitimately arise, which affect the substance of medical experimentation. Such cases aside, technical arguments normally become relevant for lay publics in virtue of their practical implications. Those publics thus have a legitimate concern that the experts have adequately assessed the scientific evidence, which is to say: that the experts’ technical arguments have transactional and intrascientific public merits. This concern is legitimate for two reasons. On the one hand, accurately assessing the content merits of an argument
288 Chapter 9
depends partly on the transactional and public contexts in which the assessment occurs. If scientists claim to identify a correlation between diet and health, we certainly want them to have gotten the evidence right. On the other hand, precisely because empirical arguments can have policy implications, we want to rule out any suspicion that the experts have operated from a politically biased position that would lead them to neglect or downplay relevant data or counterarguments to their claims. Lay publics thus have a legitimate interest in how scientists and administrators arrive at their technical conclusions—an interest, that is, in the transactional and organizational conditions affecting the process of argument construction and evaluation in regulatory agencies. These conditions provide a kind of indirect evidence for the content merits of technical science advice. As I argued in chapter 8, NAS scientists and administrators recognized that argumentation-theoretic demand. Many of the controversies reported by Jasanoff (1990) also turn on such process conditions. Both the NAS rhetorical moves and much public controversy over regulatory science have a solid contextualist basis: transactional and (intrascientific) public merits mediate the relevance of scientific arguments for lay publics, thereby opening technical argumentation to a wider demos. But identifying the appropriate manner and place of such inclusion depends on close contextual analysis. Critical contextualism thus implies that adequate argumentative process must respect both the demands of technical expertise and the validity of public involvement at some level. Meeting these demands calls for a differentiated analysis of contextual relevance. Scientific arguments have interdisciplinary relevance when scientists in different areas of research can connect disciplinespecific ideas of truth within the broader common framework of an objective natural world. Those arguments have relevance for lay publics in a number of possible ways. Substantively, scientific arguments can entail specific conclusions about what natural occurrences to expect in their physical environment or bodies; in conjunction with unquestioned practical premises, they can suggest obvious courses of action. From an empirical perspective, failure of the expected events can lead laypersons to question the science itself. In some cases, such failures reveal gaps between laboratory contexts and local conditions to which the science is to apply; such gaps then invite scientists into closer collaboration with lay sources of knowledge (see note 11). From a practical perspective, when technical arguments have relevance for policy, empirical claims face broader contexts of evaluation whose standards include or depend
289 Critical Science Studies and the Good Society
on value-laden conceptions of risk, cost-benefit schedules, goal priorities, and so on. In moving into broader contexts of the above sort, the contents of scientific arguments sometimes undergo significant transformations. Some of these follow simply from the diverse background knowledges of different audiences, so that what need not be said in one context must be laboriously explained in another. Other transformations stem from the transition from technical to political contexts. For example, in a technical context scientists might make a methodological decision, say, in the choice of statistical methods of error analysis, that affects their judgment of the evidence. They might make that decision for any number of nonpolitical reasons: the chosen method might be mathematically more straightforward, or reflect contingencies of training or the conservative disciplinary ethos that prioritizes the avoidance of error. In a regulatory context, however, such a decision can end up favoring one political or economic group over another. For the argument to prove itself in the new context, it must now include a defense of the method in relation to its political implications. To summarize, then, one can say that contextualism, as a metacritical framework, replies to Jasanoff’s question of “who arbitrates” with at least one general prescription for ground-level critique: it calls for the inclusion of all the relevant contexts, scientific and lay. The authoritative arbitrator is democratic rather than technocratic. To be sure, formidable challenges confront attempts to bring experts and lay publics together in genuinely democratic exchange (Turner 2003b). Determining whether such challenges have been met requires detailed critical assessment that takes into account the full range of context-specific complications: the normative demands and goals of intrascientific argumentation, the need in expert committees for compromise (as appropriate) in the evaluation of evidence, readiness to work with agency officials, the demands of the specific organizational and political environment, norms operative in the relevant lay contexts, substantive transformations in argumentative demands, and so on. Critical contextualism thus leads us to expect the case-specific complexities that Jasanoff documents in her study and that resist general prescriptions, beyond that above. In working out these detailed assessments, the critic is partly guided by cross-contextual relevancies that specify the legitimate deliberative demos. However, that specification will no doubt leave open a range of possible alternatives for deliberative design. Although the critical contextualist model calls
290 Chapter 9
for a deliberative democratic framework, that demand arises from the way policy-relevant scientific arguments tend to spread across diverse contexts. But exactly how one includes those contexts remains open, and selecting among alternative designs might rest in some cases on one’s vision of a democratic society. I am not sure that the metacritical framework settles this deeper question of vision. (2) The controversies that have arisen around the relationship between evolutionary science and Christianity raise an especially interesting question for critical contextualism. Here I am not concerned with the fights in school boards and courts over education, whether or not creationism must be taught in secondary school biology courses. Rather, I want to say something, however brief and provisional, about a debate on one of the broader questions that lies behind the policy debates, namely whether Christianity and evolutionary theory are compatible. Among others, scientists, theologians, philosophers, and public intellectuals have carried on this debate in a variety of forums (e.g., NAS 2008; Biema 2006; Haught 2000; Miller 1999; Kitcher 1982). The different answers to this question are in effect positions on relevance: whether and how each context is relevant for the other. Darwin and his peers generally recognized the mutual relevance that linked these two contexts of argumentation (Campbell 1986, 1990). Both evolutionary theory and Christian dogma propose an answer to what appears to be the same question, namely the question of the origin of species, including the origin of the human race. But they address that question according to very different, apparently incompatible, vocabularies, methods, and standards of cogent argument. An argument that bears directly on a given question is said to be “materially” relevant for the context in which the question is at issue. But given the differences in methodologies, just that relevance is in doubt here, and the parties to this debate have taken different positions on the matter of relevance. If one recasts the typologies proposed by John Haught (2000) and Ian Barbour (2000) in terms of a contextualist notion of relevance, then three major positions emerge: Opposition. This approach understands the two contexts as materially relevant in a narrow sense that denies any contextual differences in how the idea of objectivity functions in the two contexts. The result is an either/or “turf battle” in which each side claims the other cannot correctly answer the basic question at issue (regarding the origin of species). The conflict approach
291 Critical Science Studies and the Good Society
assumes that the basic question is sufficiently precise in the two contexts as to rule out different answers as incompatible answers, and different modes of argumentation as fundamentally at odds. On the antireligion side, militant atheists such as Daniel Dennett and Richard Dawkins argue that evolutionary theory is incompatible with religious faith. On the anti-Darwin side, fundamentalist creationists insist on a literal interpretation of Genesis as an accurate description of earth history and the origin of species. Creationists in effect insist on fidelity to scriptural texts as a standard of content merits; they thus challenge standard modes of scientific argument. Intelligent design theorists, by contrast, take scientific argumentation more seriously but attempt to widen it. Rather than invoking sacred texts, they introduce supernatural explanations for evolutionary leaps that, on their view, cannot be accounted for by neoDarwinian theory. Separatism. This approach in effect denies cross-contextual relevance, holding that the two contexts do not really pose the same question, appearances to the contrary notwithstanding. Theorists taking this approach must meet a burden of proof, however: they must explain why the appearance of material relevance is misleading. There are different ways to do this. Stephen Jay Gould makes a case for science and religion as “nonoverlapping magisteria” by arguing that religious accounts of creation attempt to answer questions of ultimate meaning, whereas science attempts to answer the question of “why does it work this way” (quoted in Barbour 2000, 100). One might also attempt to resolve the apparent conflict as some Christians have (going back to St. Augustine), by holding that religious texts in apparent conflict with science should be read symbolically or metaphorically. In the Thomistic tradition, Christians have also distinguished secondary and primary causality: God as primary cause works through the secondary causes discovered by the natural sciences (ibid., 17–19, 101–103). Engagement. This third category of Haught’s covers two forms of engagement, which Barbour (2000, 23ff.) distinguishes as “dialogue” and “integration.” Forms of engagement accept the idea that science and religion are materially relevant in a broad sense, neither denying relevance nor reducing it to a narrow sense. From a contextualist perspective, this solution recognizes how indexical aspects of the idea of objectivity shift across the two contexts. In dialogical engagement, theorists in each context recognize the autonomy of each context, but also attempt to uncover similarities and differences that help each side better understand the other context and, in the process, its own
292 Chapter 9
context. Such dialogue can be fruitful for both sides. For one, it can help draw the boundary between scientific modes of argumentation and theology—the point at which theology must grant science its due, on the one side, and where scientists like Dawkins begin to speak beyond their ken and enter a different context of argumentation, on the other (see Miller 1999). In fact, this mode of engagement can readily lead to integration. For example, scientific ideas of self-organization, information, and structural and functional hierarchy have suggested ways of reconciling scientific and theological modes of explanation. As Barbour (2000, 164ff.) explains, theologians have attempted to portray God “as the designer of a self-organizing process or as a communicator of information,” or as a top-down cause. However, theologians who take this step have already begun to integrate scientific findings and modes of thought into the content of their theological arguments. The theology of Teilhard de Chardin, the process theology of Alfred North Whitehead and his disciples, and Haught’s own theological synthesis are examples of such integration (see Haught 2000). If engagement between evolutionary science and theology illustrate a distinct sort of cross-contextual relevance, then that relevance, which I have provisionally labeled broad material relevance, is distinguished by the following features: broadly materially relevant contexts (a) appear to address the same question, but (b) the elements of expertise—the technical vocabulary, methods of inquiry, and standards of cogent argument—differ significantly in each context, making cross-context communication difficult and raising suspicions that the very questions at issue fundamentally diverge. Consequently, (c) broadly materially relevant contexts raise deep questions about the precise relationship between the two contexts. These questions challenge members to clarify exactly how each context is, and is not, relevant for the other. That task in turn requires one to articulate the indexical character of the different ideas of objectivity and truth, asking whether or not substantively different truth claims can be rendered consistent. Such clarification can sometimes change standards of argument in at least one of the affected contexts, as it did for Christians who rejected overly literal readings of the Scriptures and began to rethink their ideas about God. In effect, they came to understand the objectivity of the creation account in Genesis differently from that operative in the earth sciences, paleontology, or historical reports.
293 Critical Science Studies and the Good Society
Conclusion In this final section of the book I have probed some of the further questions that arise when we bring the contextualist framework into domains of argument at the interfaces of science and society. Although these domains lie at the edge of the current study and raise complications beyond the scope of this book, they also provide the opportunity to clarify the critical contextualist model and develop questions for further inquiry. Four areas strike me as important. First, I hope to have clarified the prescriptive status and limits of critical contextualism. Although it calls for a certain procedure of critical assessment, stated in Cog(A), which has deliberative democratic implications, it does not put forth a set of universal conditions on cogency, nor does it dictate specific deliberative designs. In other words, the contextualist model does not claim that for any argument A, A is cogent if and only if the content of A holds up according to the content and transactional standards of evaluation in every potentially relevant context, such that A has public merits across all those contexts. Even when all three merits are at stake in argumentation, the exact manner in which they are at stake can vary considerably. For example, in some cases the standards of evaluation in a given context CR might not apply without further ado to arguments that originate elsewhere but are relevant for CR. On the contrary: the external arguments might provide grounds for revising the standards in CR. Something like this occurs when Christian theologians attempt to rethink their interpretation of Scriptures and understanding of God in light of scientific findings. The contentious interactions between evolutionary science and religion, as well as interdisciplinary controversies and policymaking, point to a second topic of research: the relationships of relevance that link different contexts. Although relevance constitutes an ongoing research agenda in argumentation studies (Walton 2004), theorists tend to start with the assumption that a speaker addresses a particular audience in a particular context. One thereby takes the context for granted and asks what features of an argument make it relevant (or not) for that context; the normative question of whether or not the argument ought to be assessed in that context in the first place is taken as settled. Critical contextualism reverses that assumption: for a given argument, one wants to know which contexts of evaluation are relevant. One thus asks
294 Chapter 9
about the different ways in which an argument has potential relevance for different types of context, thus rendering those contexts relevant in the evaluation of that argument. To answer this question, one can draw on the available tools for the study of relevance, but one must use those tools in new ways (Rehg 2008). Third, relevant contexts of evaluation engender transformation in the contents of arguments as they move across contexts. This raises questions about what it means for a single argument to “travel” and acquire public merits. On the one hand, there must be some point of continuity; on the other hand, transformations may be significant insofar as different contexts set different challenges: different objections must be met, different standards of evaluation applied, different rhetorical situations mastered. Finally, I have argued that context-specific complications place limits on the prescriptive reach of contextualism as a metacritical framework. Critical analysts must therefore move beyond the framework by developing substantive, context-specific evaluations that rely on available argumentative resources. This necessity raises the question with which I conclude: what in the end grounds critical assessments? At one level, the rational force of critique simply rests on the quality of the substantive arguments the critic is able to mount with the help of available resources. The critic thus draws on standards of cogency that are not at issue in a controversy. In doing so, one need not claim that these standards constitute a foundation beyond challenge or an inescapable set of presuppositions: in other cases or contexts, they may well come under attack. One presumes only that they are not at issue in the case currently under consideration. At a deeper level, however, these standards sometimes embody specific ideas of the good society and its most important challenges. For example, dialogical ideas of openness, inclusion, and the like—which we saw not only in Habermas’s discourse theory but also in scientific and regulatory arguments—might be regarded as belonging to an ideal of reasonable, autonomous cooperation that emerged only under specific cultural-historical circumstances.12 In fact, I think it more plausible to regard these idealizations as a distinctly rationalist-egalitarian interpretation of perennial issues that confront any social practice, modern or not: who is included, to what extent, under what constraints, and so on. In debating competing interpretations of such ideals and their concrete implications for institutional design, participants in effect argue over alternative visions of social cooperation in the practices
295 Critical Science Studies and the Good Society
at issue: the meaning of good science, good policymaking, and ultimately, the good society. If that analysis is on target, then critical assessment finds its ground in a vision of the good society and its relation to scientific knowledge. As a metacritical framework, contextualist CSS does not fully specify a single vision of the good society. Pushed to this deeper level, reasonable critique must argue for one vision over its competitors. The analysis of such argumentation takes us beyond the present work (but see Cooke 2006), though I suspect that effective arguments depend on innovative transformations of practices and socialinstitutional experimentation in which members acquire direct experience of alternative visions. In any case, a vision of the good society constitutes the final, encompassing context of evaluation in which all other contexts presumably emerge and find their place.
Notes
Introduction: Science Wars, New and Old 1. The rationality debates began in the late 1960s over issues in cultural anthropology and spread to the study of science, where they morphed into the “science wars” (see Wilson 1970; Hollis and Lukes 1982; Brown 1984, 2001; Zammito 2004, chap. 5). In a narrow sense, the latter phrase refers to the academic debate in the 1990s over postmodern theories of science (Pinch 2001; Ashman and Baringer 2001; Labinger and Collins 2001). I use the term in a broader sense, similar to Turner (2003a), who speaks of three “science wars”: the social critique of science in the 1930s and ‘40s, the academic disputes that began with science studies and culminated in the 1990 debates, and the more recent public controversies over the political status of expertise. 2. For a good example, see Atkinson 1999, which combines resources from the sociology of science, quantitative linguistics, and rhetoric in a study of the Philosophical Transactions of the Royal Society from 1675 to 1975. 3. Star and Griesemer (1989, 404) give the example of “species” as a boundary concept that fostered cooperation in setting up a museum of vertebrate zoology. In this case, the concept functioned as an ideal type, which “is adaptable to the local site precisely because it is fairly vague; it serves as a means of communication and cooperation symbolically—a ‘good enough’ road map for all parties” (ibid.). 4. The Compact Edition of the Oxford English Dictionary (1971). In The American Heritage Dictionary (1981), we find that cogent arguments are “not easily resisted.” 5. Interdisciplinary exchange is evident, for example, in conferences sponsored by the National Communication Association (NCA), the Ontario Society for the Study of Argumentation (OSSA), and the International Society for the Study of Argumentation (ISSA); see also the respective publications of the latter two groups (Informal Logic and Argumentation). Departments of argumentation theory can be found, for example, at the University of Amsterdam (Dept. of Speech Communication, Argumentation Theory, and Rhetoric).
1
Science as Argumentative Practice
1. If one focuses on the United States, the rhetorical turn found its main support in departments of English, literary studies, and communication. But the input came from a broader range of sources.
298 Notes
2. “The aim of argumentation is not to deduce consequences from given premises; it is rather to elicit or increase the adherence of the members of an audience to theses that are presented for their consent” (Perelman 1982, 9). 3. In fact, the informal logic movement had roots of its own, apart from the works by Toulmin and Perelman and Olbrechts-Tyteca, though it drew on those authors for confirmation (see van Eemeren et al. 1996, chap. 6; Walton 1989). 4. Van Eemeren et al. (1996, 165–166) point out that informal logic and critical thinking are not coextensive, but their cross-fertilization is evident. 5. In a later paper, Wenzel (1987, 101) expressed dissatisfaction with his earlier views, so I am not sure if he would still construe Habermas this way. In any case, Habermas (1984, 25ff.) eventually associated these conditions with rhetoric (more on that in chapter 5). 6. Habermas (1984, 33–38) is clear about this difference between institutional procedures and discursive idealizations; note also that both Alexy (1990, 166) and Habermas (1998a, 43) consider the latter to apply to argumentation or justification as such.
2
Kuhn’s Gap: From Logic to Sociology
1. The prediction criterion defines relevance in terms of deductive entailment of observation consequences from a hypothesis. But from the hypothesis that “Two-headed snakes exist,” one cannot deduce the confirming observation, “This is a two-headed snake.” Probabilistic examples show that deductive connection is also not a sufficient condition for confirmation (Achinstein 2001, 146–48; 1985). 2. In fact, Hempel does not seem to regard the deductive connection between H and its development as the crucial point, as his treatment of existential hypotheses does not rely on such a connection. Nonetheless, the connection is certainly there in the case of generalizations, and one can easily add it in the case of existential and nonquantified hypotheses. Thus, for all hypotheses that are open to direct confirmation, H entails dHN. Hempel (1945, 109) also defines the nondirect idea of confirmation of H by O: O confirms H (nondirectly) if O directly confirms J and J entails H. Thus, if H is “the next raven we see will be black,” then a sample (O) of 100 black ravens confirms H because O directly confirms J, “all ravens are black,” which entails H. 3. Hempel did not consider truth, in an “absolute” correspondence sense, as useful for argument assessment. Rather, we should speak only of empirical truth, both for hypotheses and observations, in a “doubly relative” sense, i.e., as relative to a given degree of confirmation and relative to a given point in the ongoing process of inquiry (see Hempel 1937/2000; 1990/2000). 4. According to Rescher (1958, 92–93), Hempel’s qualitative definition sets forth rules of “evidential presumption” rather than of “evidential support.” The former implies that H is “more likely than not,” whereas supporting evidence makes H “more likely than before, i.e., more likely a posteriori than it was a priori” (ibid., 84–85, emphasis removed). 5. The determination of the most likely frequency distribution normally requires the use of partialdifferential equations (perhaps numerical approximations), and typically ∆E is multivalued. In such cases, dc(H, E) also has multiple values. Strictly speaking, then, dc itself is not a probability, inasmuch as it does not satisfy all the postulates of probability theory (Hempel and Oppenheim1945/2000, 149–152, 154–155).
299 Notes
6. Hempel and Oppenheim (1945/2000, 158) state two steps, but this combines (1) and (2) in my account. 7. E.g., when applied to the testing of universal hypotheses, the quantitative model is disappointing (Hempel and Oppenheim 1945/2000, 154). If H is “all ravens are black,” then according to the quantitative model, a single observation of a nonblack raven generates dc = 0 (regardless of the number of black ravens in our sample), whereas one confirming instance is as good as a million in the absence of any falsifying observation: both sample sizes yield dc = 1.0! This result does not add anything useful to what the satisfaction criterion already told us about confirming and disconfirming evidence. According to Jeffrey (2000, 87–88), problems with “Dutch book” arguments convinced Hempel to abandon the quantitative definition; for other problems, see also Achinstein 2001 (chap. 3); Salmon 1967 (68–79, 83–96). 8. See Brown 1977; also Kuhn 1996 (chap. 6). Even without Kuhn, the internal problems afflicting formalist models like Hempel’s and Popper’s would lead to more substantive, contextual accounts in any case (cf. Suppe 1977); moreover, other philosophers of science (Polanyi 1964 [first ed., 1958]; Hanson 1958) had already made similar arguments. But Kuhn’s book drew the most attention. Note that Popper also anticipated Kuhn in some respects (see Kuhn 1970a, 1–3). 9. The pragmatist tradition played little direct role in Kuhn’s thinking, which according to Fuller (2000a, 8n16, 60–70) was shaped more by Ludwig Wittgenstein, Karl Popper, Alexander Koyré, and J. B. Conant. However, other proponents of the “logic of discovery” did draw on pragmatism (e.g., Hanson 1958), and their work in turn influenced Kuhn (see Beller 1999, 290–294). 10. By contrast, the underdetermination confronting Hempel was that formulated by Duhem, which was limited to the difficulty of isolating single hypotheses for testing in the physical sciences; Quine later extended the idea to entire language systems (see Zammito 2004, chap. 2). 11. Hoyningen-Huene (1993, 252ff.). Kuhn (1970b, 240) takes as his focus of explanation the “normal (i.e., non-pathological) scientific group.” He thus has “recourse exclusively to social psychology [or sociology] . . . a field quite different from individual psychology reiterated n times.” In fact, a number of his claims pertain to individual cognitive psychology, e.g., his remark on “neural reprogramming” (1996, 204); cf. Sharrock and Read (2002, 46–47). 12. See Hoyningen-Huene 1993, chap. 24–27; Golinski 1998, 14. It is at best debatable whether Kuhn’s two major historical studies, of the Copernican and quantum revolutions (published in 1957 and 1978, respectively), usefully illuminate the macrotheory of collective persuasion in Structure; Sharrock and Read (2002) consider them essential for understanding Kuhn, but Beller (1999, chap. 14) emphasizes the differences between Structure and the earlier Copernican study, which Kuhn says he wrote before hitting on the notion of paradigms. Kuhn also admits that his later study made no attempt to hook up with Structure (Kuhn 2000, 289–296, 313–314). 13. In some passages, Kuhn seems to consider the values of accuracy and scope as the best measures of progress; however, they allow us to see progress only over the long run (see Kuhn 1996, 205–206). 14. The deferential interpretation is argued by Fuller (2000a), rejected by Sharrock and Read (2002).
300 Notes
15. For an excellent overview, see Zammito 2004; note that I use the term “SSK” in a broad sense, as including not only interest-based sociology of scientific knowledge (e.g., Bloor 1991) but also ethnographic approaches (e.g., Knorr-Cetina 1981) and ethnomethodological studies of scientific work (dubbed “ESW” in Lynch 1992), views designated as “social constructivist.” 16. On the debate between these approaches, see Lynch 1992; Bloor 1992, 2004; Kusch 2004a,b; Sharrock 2004. 17. See Barnes 1982; Shapin 1982; Shapin and Schaffer 1985; Bloor 1991; Barnes, Bloor, and Henry 1996; for the Wittgensteinian basis, see Bloor 1983, 1997. Microsociological approaches that fit, to some extent, with an interest-based approach are found in Collins 1983, 1985 and Pickering 1984.
3
Closing the Gap: Three Rhetorical Perspectives on Science
1. Prelli (1989a, 6) sees Finocchiaro 1980 as the only previous major attempt to bring a rhetorical framework to bear on issues raised by Kuhn; but cf. Shapin and Schaffer 1985. 2. In this section, intralinear numbers refer to Pera 1994, unless otherwise indicated. 3. Lakatos (1976) and Popper (1989, 313n4) are explicit about the dialectical aspects of their approaches; Pera (1994, 4, 32, 183–185) lumps Laudan together with Popper and Lakatos (among still others) in their shared commitment to a transhistorical ideal of method whose rules would allow an impartial arbiter to rise above the actual debate and assess outcomes. 4. For example, according to the textbook account, Peyton Rous abandoned his research on cancer viruses—which would eventually win him the Nobel Prize—for a number of years, in part because his opponents had more substantive factors on their side, that is, a number of objections that Rous could not meet (Gross 1970, 913–915; Marcum 2002). As adjudicated by Pera’s model, Rous had temporarily lost the debate, and the medical science community acted rationally in rejecting the viral theory of cancer. It is not clear how well Pera’s model would deal with the possibility of social-institutional deficiencies in this case, a point I return to below (cf. Kevles 1995). 5. In one place, Pera (1994, 144) says it “never happens” that the scientific community prefers a theory whose supporting arguments are weaker than those supporting competing theories. The attached note (ibid., 213n54) is somewhat unclear but seems to imply that he means that scientists never rationally prefer a theory with weaker support. Elsewhere he grants that the weaker argument might sometimes win (187). 6. Though Pera (1994, chap. 6) believes that scientists aim for true theories, he rejects attempts to define truth in a manner external to actual debate, i.e., metaphysical (correspondence) theories of truth and ideal warranted assertibility theories. Thus, although truth is indicated by arguments, it cannot serve as an independent standard of scientific progress or cogency. 7. See Pera 1994, 121: “As far as these [procedural] factors are concerned, again, my aim is not prescriptive. I do not intend to construct a system of formal dialectics . . . but to make explicit those rules that are provided by scientific practice found in the history of science and accredited scientific tradition.” 8. Unless otherwise noted, intralinear citations in this section refer to Prelli 1989a.
301 Notes
9. By contrast, Pera (2000, 59–60) excludes appeals to ethos and pathos (emotions) from the “proper” means of persuasion. Recent work in argumentation theory shows that such a restriction is overly narrow (Gilbert 1997; Rehg 1997; Tindale 1999; cf. McCabe 1994). 10. See, e.g., Prelli’s analysis of the memory-transfer controversy (Prelli 1989a, 158–181), where he takes into account the effect of the main protagonist’s (James V. McConnell) personal foibles and how other scientists perceived his character as a scientist (ibid., 161, 295n38); Prelli also notices how the positioning of scientists in the community affected the controversial exchanges, i.e., whether the scientist was well established or not, or a graduate student, and so on. 11. Pera’s model would allow scientists themselves to criticize the substantive factors within their tradition, but still leaves open the question raised by science–society boundary disputes. 12. Note that Latour takes “rhetoric” in its usual sense: “the discipline that has, for millennia, studied how people are made to believe and behave and taught people how to persuade others” (Latour 1987, 30). 13. See Latour 1988, e.g., 216–218, 236: Latour seems to hold that the success of science historically depended on Western political power and colonial violence; today science is deeply implicated in military projects that threaten the globe. For a critique of Latour’s antinormativism, see Fuller 2000c. 14. Latour (1999b) thus insists that his approach is realist, albeit in a sense that rejects dichotomies; more on that below. 15. For a more complete elaboration of ANT, see Latour 1999b, 2005. 16. For the Pasteur study, Latour examined the articles published in three journals over an extensive period, each of which aimed at a somewhat different audience concerned with the relevant issues (Latour 1988, introduction). 17. L. Capitan, “Le rôle des microbes dans les société,” Revue Scientifique 10 (1894): 292, quoted in Latour 1988, 37 (my emphasis). 18. Here Latour (1988, 29–31) actually makes a normative claim about the cogency of Pasteur’s evidence. 19. Latour (1988, 183–184) grants that “very, very rarely, everything else being equal,” the “strength of a word” actually persuades someone. 20. As Latour (1999b, chap. 7) later explains, the distinction between might and right masks a deeper conflict between elitist expertise and democracy.
Postscript I The Return of the Logical: Achinstein’s Realist Theory of Evidence 1. The definition of explanatory connection attempts to overcome difficulties with the HD model and inference-to-best-explanation model of relevance. 2. Perhaps we could regard Achinstein’s conditions as a “synthetic definition” (following Timmons 1999, chap. 2).
302 Notes
3. Achinstein (1985, 142) admits that he has “no general theory to offer about this,” i.e., how to determine if the probability of H is greater than one-half, as required by (1) in the definition of potential evidence—only that this threshold is assumed or argued for somehow. 4. For example, some epistemologists distinguish between justification and warrant (e.g., Haack 2003, chap. 3; Plantinga 1993), but my purposes at this point do not require such a distinction; among argumentation theorists, the relation between inference and argument is also a contested issue (Johnson 2000). 5. In fact, background assumptions pose difficult problems in argument theory. On a traditional approach, charity in argument evaluation required one to fill in any missing premises that would be necessary for deductive validity. On an informal approach, however, one cannot assume that arguments must satisfy standards of deductive validity; moreover, arguments are situated in lifeworld contexts in such a way that interlocutors share background assumptions that escape full articulation (cf. Habermas 1987, chap. 6; Govier 1987). 6. Staley (2005) points out a related difficulty: Achinstein’s concept of potential evidence separates the idea of good reasons from the rationality of belief formation. 7. Achinstein (2001, 96–98) also marks the difference between justification and good reasons by distinguishing two senses of “reasonableness of belief”; see Staley (2005) for a critique of this move. 8. Argumentation theorists have, to be sure, identified multiple context-sensitive goals that motivate people to make arguments (e.g., interpersonal dominance, identity-maintenance; see Hample 2005), but my claim pertains to the rational goal inherent in adducing reasons for a conclusion, or what Asen (2005) refers to as “internal justification.” 9. Goldman (1999), for example, does not insist that a “veritistic” approach requires metaphysical realism; empirical adequacy suffices. Perhaps one could also understand representational adequacy in relation to an “ideal epistemic situation” that included all possible testing (and hence would still go beyond present justificatory resources) (cf. Staley 2005).
4 Habermas’s Critical Theory and Science: Truth and Accountability 1. See Habermas 1987, 119: “formal-pragmatic analysis aims at structures that, in contrast to the historical shapes of particular lifeworlds and life-forms, are put forward as invariant.” 2. Thus we can align Habermas’s discourse theory with the dialectical tradition in argumentation theory, which focuses on dialogue as critical discussion, though much work remains in spelling out the similarities and differences (see Feteris 2003; Walton 1998; Zuber 1998; van Eemeren and Grootendorst 1992). 3. I thank William Keith for alerting me to this point. 4. Habermas (1984, 33) grants that some aspects of Toulmin’s model suggest an alternative reading, but insists that the model ultimately is ambiguous. 5. Habermas’s use of speech act theory is one source of criticism: as Potter (1996, 11–12) notes, speech act theory is weakened by its reliance on made-up examples and its inability to deal with
303 Notes
indirection in speech (see also Thompson and Held 1982). Other critics, such as Heath (2001), draw on analytic philosophy of language. Cooke (1994) provides a sympathetic critique. 6. It is not entirely clear how Habermas’s broader list of validity claims relates to the basic three, but that question need not concern us here. 7. Ideas of reason thus have a specifically Habermasian sense here, which is inspired by Kant but not equivalent to his transcendental idealist model; see McCarthy 1991, chap. 1; Habermas 2003, chap. 2. 8. Habermas (2003, 90–91) distinguishes the “world” as the “totality of objects” from “reality” as “the totality of statable facts.” The idea of the world seems to involve that of a transcendent or mind-independent domain, which we strive to represent in fallible statements of fact. 9. Because Habermas tends to use the ideas of justification and warrant more or less interchangeably, I do not draw a distinction between these terms here; in chapter 5 I suggest a way in which his model might involve a distinct notion of warrant. 10. Habermas’s pragmatic conception of truth, as I read it here, differs from the Peircean view defended by Misak (2000; cf. Misak 1991), which I understand as an attempt to spell out what it means for a hypothesis or belief to be true. As Misak (2000, 49) puts it, a true belief is one that “would withstand doubt, were we to inquire as far as we fruitfully could on the matter.” This formulation gets at the idea of truth as transcending any particular level of inquiry, and thus it seems to project a final state of inquiry, even though Misak wants to avoid that idea. 11. Here I go somewhat beyond Habermas’s rather sparse remarks in Habermas 1971/2001, 86–87. The idea behind this criticism seems to be something like the following: If I say that “p” is true if and only if the state of affairs described by “p” actually obtains, or corresponds to the fact that p, then I simply reiterate, on the right-hand side of the equation, the original statement p with the emphasis “fact that” or “actually obtains.” Thus the correspondence relationship itself remains obscure. Breaking the sentence down into components (e.g., following Russell) still does not get one out of such linguistic reiteration. For an overview of theories of truth, see Kirkham 1992; the criticisms of metaphysical correspondence theories are summarized in Hallett 1988, chap. 2. According to Cooke 2001a,b, Habermas’s theory of truth retains a modest metaphysical view. 12. Here I extend what Habermas says about everyday practice to scientific settings; see Habermas 2003, 36–42; cf. 1971/2001, 88–89. 13. Habermas (2003, 40–42) creates a problem for himself by not noticing how much practical engagement with the world actually enters into argumentation as a source of empirical reasons. That is, he worries about explaining why discourse participants would move from discourse back to action, given the difference between how truth functions in action (as a practical certitude) and how it functions in empirical discourse (as a defeasible claim, even on the best of evidence). I think the problem arises because (a) Habermas invests action with too much certitude—in fact, we often act very cautiously or nervously, unsure if we can trust all our beliefs—and (b) he does not bring action into discourse to a sufficient extent. 14. I suspect my interpretation of ethical-political discourse might differ from Habermas’s on this last point, since he tends to distinguish argumentation (discourse) from bargaining by the idea that discourse issues in consensus on reasons and conclusions, whereas bargaining issues in compromises that parties accept for different reasons. But a number of theorists have questioned this sharp distinction (Bohman 1996; Warnke 1999; McCarthy 1998).
304 Notes
15. See Overbye 2005a; the old definition read: “Science is the human activity of seeking natural explanations for what we observe in the world around us.” Although the new definition includes the ideas of observation and hypothesis testing, it does not limit explanation to natural causes.
5 Habermas’s Theory of Argumentation as an Integrated Model of Cogency 1. Argumentation theorists have understood relevance in various ways: as a content merit (Johnson 2000), in dialectical terms (Walton 1995, chap. 6), and as part of rhetoric (Tindale 1999, chap. 4). 2. See Habermas 1993, 53: “That degree of fulfillment counts as ‘sufficient’ which qualifies our argumentative practice in a given instance as a spatiotemporally localized component of the universal discourse of an ideally extended community of interpreters that we necessarily presuppose.” Habermas then goes on to compare these idealizations with geometrical figures we can only approximate in actual diagrams (ibid., 54–55). 3. It is interesting here to compare Habermas’s dialectics with the model of objectivity proposed by Helen Longino (1990, chap. 4; 1996, 40; 2002, 128–135). As Longino (1990, 78, 197–200) acknowledges, her norms for “critical discursive interaction” are similar to Habermas’s idealizations. But notice that her four “conditions of effective or transformative criticism” pertain not only to the substance and flow of discussion (shared standards, uptake) but also to its social space (publicity, equality): (a) There must be recognized public venues (conferences, journals, etc.) for the presentation and criticism of research in its various facets (evidence, methods, assumptions, etc.). (b) There must be “uptake of criticism” in the science community, that is, scientists must pay attention to and criticize one another’s positions, and they must adjust their views in response to such criticism. (c) There must be “publicly recognized standards”—shared meanings, rules of inference, values, and aims—that members can invoke in their criticism. (d) Finally, a “tempered equality of intellectual authority” must obtain among the participants (Longino 2002, 134, 128–35). 4. The decision-making problems I describe in what follows also arise when patients are mentally alert and families are involved in the decision; for a poignant account of end-of-life decision making in the case of Susan Sontag, see Rieff 2005. In what follows, I draw on ideas in Benjamin 1990, 24–32. 5. Argumentation theorists who have elaborated dialogical models tend to emphasize dialectical rules governing the intellectual process (cf. Walton 1998; van Eemeren, Grootendorst, and Snoeck Henkemans 2002). However, the orientation of dialogue toward reasonable judgment, and the openness of dialogue models to ethos and pathos-based arguments (see Walton 1989, 1995), implies that other dialogical models, similar to Habermas’s, include a rhetorical perspective. For a rhetorically oriented dialogical approach to the history of science, see Beller 1999. 6. On Rous, see Gross 1970, chaps. 6, 23; Marcum 2002; as Solomon (2001, chap. 4) points out, scientists tend to overestimate the strength of evidence produced in their own laboratory relative to others’ findings. This kind of bias, she argues, can be fruitful for inquiry in a discipline. 7. As Keller (1983) presents her case, McClintock was using an older methodology that fewer biologists were able to understand as the field increasingly turned toward molecular methods.
305 Notes
8. For example, in debates over science funding in the early twentieth century, “best-science” elitists defended the epistemic superiority of a distribution of grants that would privilege the betterequipped research centers, a view that seems to contradict the standard of equality (Kevles 1979, 151–152). 9. I draw here on a comparison with Habermas’s democratic theory, which aims to show how the democratic process produces legitimate outcomes; that task in turn requires one to show that “insofar as the democratic process satisfies conditions of an inclusive and discursive opinion- and will-formation, it grounds a presumption that outcomes are rationally acceptable” (Habermas 2005, 107–108). 10. Habermas’s formal pragmatics also suggests a complementary, but less strategic interpretation of credit as serving scientists’ need for mutual recognition, an approach one also finds in Merton (1973, chap. 14), though Habermas (1987, 97–100ff; 1990, 199–200) differs from Merton in linking this need with communicative practices in general, and not simply with the ethos of science. 11. In a recent case, a Caltech team accused a Spanish team of astronomers of stealing a discovery by using clues incautiously posted on the Caltech website; see Overbye 2005b. Chapter 6 presents a further example of exclusionary norms.
6 Argumentation at Fermilab: Putting the Habermasian Model to Work 1. For the basic decay processes that CDF considered, see Staley 2004a, 104–107 (also his chap. 1 on the theoretical breakthrough that led to the prediction of top and bottom quarks). The socalled standard model in physics holds that ordinary matter is composed of three fundamental particles (fermions): up (u) and down quarks (d)—which combine to form neutrons and protons— and electrons. Higher-energy processes can generate four other types of quark: charm, strange, top, and bottom quarks. The electron belongs to the class of leptons, whose members are not composed of quarks; leptons also include taus, muons, and three types of neutrino that correspond to the other three leptons. Each fermion has its antimatter partner. These particles and their various derivatives are held together by four basic forces that are mediated by corresponding gauge boson particles. The table below summarizes the basic particles and forces; note that the forces align with force particles, but not with the matter particles: Fermions (matter particles)
Force
Leptons
Quarks
electrons taus muons neutrinos: electron-, tau-, muon-
up, down charm, strange top, bottom
Forces (between fermions, bosons, and derivatives)
Force particles (gauge bosons)
electromagnetic weak strong gravity
photons W, Z bosons gluons gravitons
2. For a sense of the various sources of interference in more sophisticated systems, see KnorrCetina 1999, 48–52 on CERN’s colliding detector. Besides other types of high-energy events, there are interferences inherent in the detector and electronics (noise), particles from low-energy events, and “smearing” of data distributions due to limitations in resolving power.
306 Notes
3. The silicon vertex detector installed at CDF, for example, can resolve particle tracks at the micrometer level (Staley 2004a, 90–100). 4. On the design, construction, and testing of the collider detector, see Staley 2004a, chaps. 2–3. 5. See Staley (2004a, 104–107). The SVX group was named after the silicon vertex detector that produced their data; the SLT group was named after their search target, “soft (low momentum) lepton tagging.” 6. 1 GeV/c = 109 electron volts. 7. See Staley 2004a, 86ff., 147; on testing detectors, see also Knorr-Cetina 1999, 56–60. 8. This uncertainty emerges in one SLT member’s comments: “I believe that choice was made after we’d seen the data. . . . Some people may have in their minds decided that they were going to make the cut at 2 GeV before they saw the data; I’m not sure” (quoted in Staley 2004a, 147). Staley (2002, 291) quotes Tipton: “the ability to fool yourself is pretty subtle.” 9. According to Staley (personal communication, March 25, 2004), CDF did not have a clearly defined standard of general approval, but the practice was to require full, or nearly full, consensus. 10. Role divisions in legal adjudication provide a further example; see Rehg 1994, 214–227. 11. I take the term “intellectual solidarity” from David Hollenbach, S.J. (2002), who applies it to political discourse; in the area of moral discourse, see Rehg 1994; I also draw on Durkheim’s (1964) concept of organic solidarity. 12. Speaking as a unified “we,” the authors of the Evidence paper and its summary (Abe et al. 1994a,b) describe uncertainties and conflicts in the data but make no overt mention of intrateam divergences. 13. Henry Frisch, one the CDF scientists, has raised a related question regarding authorship in large collaborations: the huge list of undifferentiated names can make it difficult for readers to identify the individuals to whom they should address specific questions about parts of the argument (Panel on Scientific Collaborations, Philosophy of Science Association Meeting, Austin, Texas, Nov. 18–20, 2004). 14. See Boykoff and Boykoff 2004; Lahsen 2005. I am indebted here to Ryan 2006. 15. For example, some argue that given the uncertainties of scientific arguments, policymakers should prefer risk-averse options, whose worst-case outcomes are the least damaging (e.g., Collingwood, Collingridge, and Reeves 1986). 16. This work centers on the so-called “discursive dilemma” (Pettit 2001; List 2006), which arises for models of argumentation that require agreement on both the reasons and the conclusion. The dilemma arises when the majority supports the conclusion but for different reasons, or when each reason has support of a different majority, such that the conclusion lacks majority support. In this context, the Evidence paper compromise represents a conclusion-based decision procedure. Whether we regard such compromise as advantageous depends on how the reasons are related to the conclusion (i.e., whether it follows only from a full set of reasons or rather follows from any one of several reasons) and on the structure of individual expertise.
307 Notes
Postscript II Who’s Afraid of SSK? The Problem and Possibilities of Interdisciplinary Cooperation 1. For succinct statements of SSK relativism, see Barnes and Bloor 1982, Barnes 1994; on the Habermasian approach, see McCarthy 1988, Bohman 1991. 2. Outside the specifically Habermasian line of critical theory, one finds modes of critical analysis that have been influenced by SSK, such as Harding 1998, Longino 2002, Rouse 1996, and Fuller 1988. For reflection on exchange from the other direction, see Lynch 1994. 3. That is: (1) I am not so concerned here with SSK theorists’ well-known agnosticism regarding truth; for argumentation theory, truth claims must be evaluated on the basis of evidence in any case, i.e., from the standpoint of justification. (2) SSK has also been criticized for its claim to avoid evaluation (McCarthy 1988). Inasmuch as SSK theorists aim in their case studies to show that rational considerations, or the weight of empirical evidence, did not decide the outcome of a controversy, they must evaluate the scientific arguments to some extent. But this minimal evaluative stance, which allows for a withholding of definitive judgment, is compatible with the more modest claim to engage in a primarily descriptive-explanatory, rather than strongly prescriptive, enterprise. (3) I do not take up the problems associated with the claim to provide causal explanations (Bloor 1991), a claim criticized by philosophers (Roth and Barrett 1990; Bohman 1991, 43–46), sociologists (e.g., Woolgar 1981; Cole 1992, chaps. 2–3), and historians (e.g., Kim 1994, chap. 1). Finally, (4) I do not explicitly address the problem of reflexivity here, although the CSS approach I propose below alleviates this problem considerably. 4. Shapere (1984, 1986) links relevance to domain assumptions, whereas Weingart (1993) defines the “context of relevance” in terms of the sociopolitical environment. For an argumentation-theoretic account that links relevance with the goals of argument, see Walton 1995, chap. 6; 2004; for a rhetorical perspective on relevance, see Tindale 1999. 5. Consensus may in some cases revolve more around practical working commitments than theoretical beliefs, as students of “science as practice” emphasize (Pickering 1992a; Galison 1997). 6. In other cases the dispute may concern whether an SSK analysis has correctly identified the significant social factors. I shall not pursue this sort of controversy here.
7
Adjusting the Pragmatic Turn: Lessons from Ethnomethodology
1. See chap. 2, sec.3; chap. 3, sec. 2; chap. 4, sec. 2 (on Habermas’s critique of Toulmin). 2. An exception is Kusch 2002, more on which later. Note that some epistemological contextualism is a specific form of semantic contextualism, focused on the context-sensitive character of epistemic vocabulary; for a range of views, see Preyer and Peter 2005. 3. According to Recanati (2004, 85), this claim is no longer controversial among philosophers of language, if by “objective statement” we mean eternal sentences: “It is more or less accepted that natural language sentences are irreducibly context-sensitive”; for a brief history of developments and the spectrum of theoretical positions, see Recanati 2005, which suggests that the ethnometh-
308 Notes
odological position is open to a range of analytic interpretations, some more literalist, others more contextualist. 4. In a study of the use of explicit coding instructions, Garfinkel found that “ad hoc considerations are consulted by coders and ad hocing practices are used in order to recognize what the instructions are definitely talking about” (1967, 22; emphasis in original; see also chaps. 6, 7). 5. The following summarizes Livingston 1987, 1999; see also Livingston 1986, 1995. 6. Cf. Sosa 2005; he argues that epistemic contextualism threatens to render analytic epistemology irrelevant or even meaningless vis-à-vis actual social practices. 7. This problem has affinities with the relativism objection that arises for attributor-based versions of epistemological contextualism, which holds that the attribution of knowledge to an individual S who claims that p depends on the attributor’s epistemic context (assuming here that p is true and S believes p on the basis of some reason). Because contexts of attribution shift independent of the justification that S has for claiming p, there is no nonperspectival fact of the matter about whether S knows that p (see Brower 1998). Kusch (2002) reaches a similar position by a somewhat different route, insisting that contextualism implies cognitive relativism; I discuss his view in the last section. 8. Note that the exclusionary views in this case were not automatic, but required explicit arguments when modern science was first emerging. Potter (2001) documents such arguments in the case of Robert Boyle; Schiebinger (1999, 69–72) traces such exclusionary views to the eighteenth century. 9. This particular type of argument is open to serious objections (see Schiebinger 1999, 5ff.; Haraway 1986); in the specific area of primatology, the argument might also reverse the causality: as Fedigan (1994) suggests, the receptivity of primatologists for feminist critique may have fostered the entrance of women into the discipline. 10. What I call the straightforward mode of immanent critique is evident in courtroom attempts to deconstruct expertise, for example, the defense’s attempt to undermine DNA-sampling expertise in the O. J. Simpson trial (see Lynch 1998). When the strategy backfires, as Lynch (2000) describes in a particular cross-examination of an expert witness, the net effect is an inadvertent use of an ironic critique; according to Lynch, courtroom experience in such trials led to a revision of the official scientific protocol for DNA-sampling methods in forensic contexts. 11. For a sociological analysis of the attempts to draw boundaries, see Jasanoff 1987; but the point is also recognized by philosophers of science. See, e.g., Kincaid, Dupré, and Wylie 2007; Kitcher 2001; Douglas 2000. For historical perspectives, see Kevles 1979 and Proctor 1991. 12. Besides the AIDS case, this point has been demonstrated for the scientific analysis of risks; see, e.g., Wynne 1989; Morgan 1993; for an extended argument, see Fischer 2000. The point is also increasingly relevant for studies of expertise (Collins and Evans 2002; Turner 2001). 13. On deliberative democracy, see not only Habermas 1996 but also Bohman 1996, 1999; Bohman and Rehg 1997; Elster 1998. For deliberative methods in environmental policy formation, see Webler, Tuler, and Krueger 2001. 14. Hales goes on to argue against Davidson (2001) that we should reject the latter’s definition of conceptual schemes in terms of their mutual incommensurability, which in turn depends on
309 Notes
the idea of translatability. His relativism thus allows for contexts being commensurable or not in different senses. 15. For discussions, see Lynch 1992; Bloor 1992, 2004; Kusch 2002, 2004a,b; Sharrock 2004. 16. For example, Kusch’s analysis of the relation between laypersons and the International Astronomical Union leads him to regard a layperson’s astronomical beliefs as “purely individual,” whereas the same beliefs of IAU members are “group-involving,” such that members undertake entitlements and commitments toward one another in virtue of their beliefs (Kusch 2002, 146). Laypersons, by contrast, enter into no such obligations when they learn something from an astronomer. This contrast strikes me as overdrawn.
8 Three Dimensions of Argument Cogency—A Contextualist Case Study 1. This model has been challenged, of course, for example by contextualists working in the area of critical thinking education; see Brown 1998; Norris 1992; McPeck 1990. 2. What Garfinkel (1967, 203) says of clinical records can thus be extended to methods of inquiry and argument: “Which [methods and norms] will be used, how they will be used, and what meanings their contents will assume, wait upon the particular occasions, purposes, interests, and questions that a particular member [familiar with the practice] may use in addressing them.” 3. For a history and overview of issues connected with science advisory committees, see Jasanoff 1990. 4. The time pressure on expert panels, for example, means that they do not have the luxury of returning to the laboratory or awaiting further data; because the policymaker requires an actionguiding answer, an evaluative judgment about the relative cogency of the available evidential arguments is more difficult to avoid (see Jasanoff 1990, 78, 80). 5. As I explain more fully in the text below, my thoughts on the general heuristic were first provoked by participation in an interdisciplinary panel (including both communication theorists and philosophers) entitled “Public Sphere Theory and the Norms of Argumentation,” held at the 2004 meeting of the National Communication Association (Chicago, Nov. 11). The panel was organized by Jean Goodwin, chaired by Lynn Clarke, and included Jean, Robert Asen, J. Anthony Blair, and me. The case study itself benefited from presentation before an interdisciplinary audience at the 2005 meeting of the Society for Social Studies of Science, held in Pasadena. 6. See Hilgartner 2000; intralinear numbers not designated otherwise in this chapter refer to this book. 7. Although Goodwin (2005, 152) borrows the term “transactions” from Black (1978, chap. 5), a number of authors have employed the term, including Garfinkel (1967, 198, 202). Those who have taken a dialogical approach to the analysis of conversational argumentation include Walton (1998) and van Eemeren et al. (1993); Bitzer (1968, 6) defined the rhetorical situation “as a complex of persons, events, objects, and relations presenting an actual or potential exigence which can be completely or partially removed if discourse, introduced into the situation, can so constrain human decision or action as to bring about the significant modification of the exigence.”
310 Notes
8. On interpretive flexibility, see Gilbert and Mulkay 1984; Hilgartner (168n37) considers the text just quoted as ambiguously situated between these two rhetorics, though the expert-judgment interpretation strikes me as more plausible; another example of such rhetoric appears at the end of the section on protein: “evidence from both epidemiological and laboratory studies suggests that protein intake may be associated with an increased risk of cancers. . . . Because of the relative paucity of data . . . and the strong correlation between intakes of fat and protein in the Western diet, the committee is unable to arrive at a firm conclusion about an independent effect of protein” (NAS 1982, 116). 9. Thus Garfinkel (1967, 198, 202) uses the term “transaction” for both person–person interactions and person–text interactions. 10. By “locale” I refer to the site of transactions; terms like “domain,” “field,” and “discipline” designate broader contexts that include many transactions and locales, whose members belong to a single domain in virtue of a common training, topical interest, etc. 11. I emphasize “ascription” here to avoid conflation with the generation of public merits, which normally depends on the fact that various audiences find the content of the argument itself inherently convincing. 12. In terms of Habermas’s distinction between participant and observer perspectives, one must say that each type of merit represents a potential normative concern from the participant perspective, even though adequate assessment—particularly of public merits and probably also transactional merits—might require information gained from an observer perspective. 13. See Review Criteria for National Academy of Sciences Reports (reproduced in Hilgartner 2000, 67) and Academy Report Preparation Procedures (ibid., 63). 14. For a more extensive overview of the criticisms, see Hilgartner (96–103). 15. “Committees are expected to be evenhanded and to examine all evidence dispassionately” (Hilgartner, 68, quoting 1996 NAS guidelines). Although this source postdates DNC, the rule fits with the concern with balance evident in the report itself (see sec. 4 above). The 1972 Federal Advisory Committee Act made balance a legal requirement; NAS committees were exempted from FACA until 1997, but the main concern was public access rather than disciplinary balance (Hilgartner, 25–26, 58–61; Jasanoff 1990, 46–48). 16. Boffey’s critique recalls the problem I described in chapter 6 with the Evidence paper: “It is difficult for an outsider to evaluate how well a committee has performed simply by reading its final report. The report seldom makes explicit the compromises that were reached, the issues that were ducked, or the biases that may have swayed committee members” (Boffey 1975, xi). 17. Boffey (1975, chap. 11) recommends a number of reform measures; in his view, the most important concern the funding and institutional charge of NAS. 18. For example, Solomon (2001) considers it a mistake to locate the rationality of the sciences in the reasonableness of individual scientists, rather than in the overall distribution of research. As I understand him, Steve Fuller also downplays the importance of reasonable transactions within the sciences; he stresses instead the broader level of institutional organization, which should ultimately answer to a democratic politics (see Fuller 1988, 2000b; Remedios 2003).
311 Notes
19. This oversimplifies things a bit. If one examines CAST along with critics outside CAST, one can see some convergence in criticisms of the committee composition (Hilgartner, 96).
9
Critical Science Studies and the Good Society
1. See Keith and Rehg 2008 for an overview of cross-fertilization between argumentation studies and science studies. Within these two areas, one finds a number of multidisciplinary conferences: e.g., the Society for Social Studies of Science (4S), which sometimes meets together with the Philosophy of Science Association and the History and Philosophy of Science Association; the International Society for the Study of Argumentation (ISSA); and the Ontario Society for the Study of Argumentation (OSSA), to name a few. 2. The idea of a critical science studies is not new, and might be applied to a range of critical theories of science (see Hess 1997, chap. 5; Longino 1990; Fuller 1988; Aronowitz 1988), at least some of which take a critical contextualist approach. Here I use the terms “CSS,” “contextualist CSS,” and “critical contextualism” for the model I propose. 3. The first sort of criticism is precisely what we should expect in scientific revolutions; but these also occur in interdisciplinary situations, e.g., the dinosaur extinction controversy (see Glen 1994). Lay critique has had an impact at certain times, e.g., in AIDS treatment and nuclear power controversies (see Epstein 1995; Gusterson 1996, chaps. 7–8). 4. For some cautions on this score, see Hample 2005, chap. 2. 5. Science journalism that reports and explains results does not necessarily imply that the lay context is relevant for assessing the public merits of technical arguments; rather the implication follows only when the arguments are relevant to political or personal decision making. Science journalism that merely informs the public of interesting results that have no immediate connection with everyday life are best understood as attempts to bring the lay public into the scientific context of origin. 6. For an example, consider Blakeslee 2001: a team of physicists developed an argument that appeared cogent in a physics context and that had relevance for biology, but it failed to convince biologists. 7. This oversimplifies somewhat by not bringing in NAS administrators, who had to approve the report. We might regard administrators as the first stand-in for the lay public, but I suspect that the better approach is to consider them a distinctly bureaucratic element of the context of origin, inasmuch as the report requires their approval. However, because NAS administrators also accepted the scientific model of dialogue, we can disregard their role for present purposes. 8. There is some affinity between the approach I sketch and that of Nowotny, Scott, and Gibbons (2001) and Gibbons (1999), but I cannot explore the similarities and differences here. 9. For example, the 1985 guidelines for the EPA’s Science Advisory Board include “Recommending, as appropriate, new or revised scientific criteria or standards for protection of human health and the environment” (quoted in Jasanoff 1990, 85). Often the policy charge arises indirectly, for example as the result of judging that a particular chemical poses a health risk, or “has the potential to be carcinogenic in humans” (EPA official Paul Lapsley, quoted in Jasanoff 1990, 146). The effect of such judgments is often to initiate regulatory action.
312 Notes
10. See Jasanoff 1987; according to Mayo 1991, one can acknowledge such factors without accepting the further relativist thesis that risk assessment is necessarily politically biased. Although judgments about how to interpret and weigh different sources of evidence involve policy choices that have implications for regulation, such judgments are still subject to objective “metascientific” criteria of adequacy. Mayo’s position, if correct, would have implications for the structuring of the risk assessment process, but it would not negate the broader point I make here. 11. Various examples of this are available in the literature; Wynne 1989, for example, studies cooperation between experts and farmers whose knowledge of local conditions had substantive relevance for the science at issue; one of the best-known is the case of AIDS activists influencing the methodology of drug testing (Epstein 1995). For an overview in the area of environmental regulation, see Fischer 2000; for attempts to develop broader implications for a theory of expertise, see Collins and Evans 2002; Turner 2001, 2003a,b. 12. Cooke (2006) takes this line, as have many others. Sen (2005), however, argues convincingly that one can find similar ideals at work in premodern, non-Western cultures.
References
Abe, F., et al. (CDF Collaboration). 1994a. Evidence for top quark production in ¯p p collisions at 兹s = 1.8 TeV. Physical Review D 50: 2966–3026. Abe, F., et al. (CDF Collaboration). 1994b. Evidence for top quark production in ¯p p collisions at 兹s = 1.8 TeV. Physical Review Letters 73: 225–231. Achinstein, Peter. 1985. The method of hypothesis: What is it supposed to do, and can it do it? In Observation, Experiment, and Hypothesis in Modern Physical Science, ed. P. Achinstein and O. Hannaway, 127–145. Cambridge, Mass.: MIT Press. Achinstein, Peter. 1991. Particles and Waves. New York: Oxford University Press. Achinstein, Peter. 2001. The Book of Evidence. Oxford: Oxford University Press. Achinstein, Peter, ed. 2005. Scientific Evidence. Baltimore: Johns Hopkins University Press. Alexy, Robert. 1990. A theory of practical discourse. In The Communicative Ethics Controversy, ed. S. Benhabib and F. Dallmayr, 151–190. Cambridge, Mass.: MIT Press. Alvarez, Luis W., Walter Alvarez, Frank Asaro, and Helen V. Michel. 1980. Extraterrestial cause for the cretaceous-tertiary extinction. Science (New Series) 208, no. 4448 (June 6, 1980): 1095–1108. Alvarez, Walter. 1997. T. Rex and the Crater of Doom. Princeton: Princeton University Press. Anderson, R. J., J. A. Hughes, and W. W. Sharrock. 1988. The methodology of Cartesian economics: Some thoughts on the nature of economic theorizing. Journal of Interdisciplinary Economics 2: 307–320. Apel, Karl-Otto. 1980. Towards a Transformation of Philosophy. Trans. G. Adey and D. Frisby. London: Routledge. Aronowitz, Stanley. 1988. Science as Power. Minneapolis: University of Minnesota Press. Asen, Robert. 2005. Pluralism, disagreement, and the status of argument in the public sphere. Informal Logic 25: 117–137.
314 References
Ashman, Keith M., and Philip S. Baringer, eds. After the Science Wars. London: Routledge, 2001. Atkinson, Dwight. 1999. Scientific Discourse in Sociohistorical Context. Mahwah, N.J.: Erlbaum. Barbour, Ian G. 2000. When Science Meets Religion. San Francisco: Harper. Barnes, Barry. 1982. T. S. Kuhn and Social Science. New York: Columbia University Press. Barnes, Barry. 1983. On the conventional character of knowledge and cognition. In Knorr-Cetina and Mulkay 1983, chap. 2. Barnes, Barry. 1994. How not to do the sociology of knowledge. In Rethinking Objectivity, ed. A. Megill, 21–35. Durham: Duke University Press. Barnes, Barry. 2003. Thomas Kuhn and the problem of social order in science. In Nickles 2003b, 122–141. Barnes, Barry, and David Bloor. 1982. Relativism, rationalism, and the sociology of knowledge. In Rationality and Relativism, ed. M. Hollis and S. Lukes, 21–47. Cambridge, Mass.: MIT Press. Barnes, Barry, David Bloor, and John Henry. 1996. Scientific Knowledge. Chicago: University of Chicago Press. Barth, E. M., and E. C. W. Krabbe. 1982. From Axiom to Dialogue. Berlin/New York: de Gruyter. Baumslag, David. 1998. Choosing scientific goals: The need for a normative approach. Studies in History and Philosophy of Science 29A: 81–96. Beller, Mara. 1999. Quantum Dialogue. Chicago: University of Chicago Press. Benhabib, Seyla. 1992. Situating the Self. New York: Routledge. Benjamin, Martin. 1990. Splitting the Difference. Lawrence: University Press of Kansas. Biema, David van. 2006. God vs. science. Time. November 13, 2006. Bittner, Egon. 1977. Must we say what we mean? In Communication and Social Interaction, ed. P. F. Oswald, 83–97. New York: Grune and Stratton-Harcourt Brace. Bitzer, Lloyd F. 1968. The rhetorical situation. Philosophy and Rhetoric 1: 1–14. Bizzell, Patricia, and Bruce Herzberg, eds. 2001. The Rhetorical Tradition. Boston: Bedford/St. Martins. Black, Edwin. 1978. Rhetorical Criticism. Madison: University of Wisconsin Press. Blair, J. Anthony. 2005. Norms and functions in public sphere argumentation. Informal Logic 25: 139–150. Blakeslee, Ann M. 2001. Interacting with Audiences. Mahwah, N.J.: Erlbaum. Blaug, Ricardo. 1999. Democracy, Real and Ideal. Albany: SUNY Press.
315 References
Bloor, David. 1983. Wittgenstein: A Social Theory of Knowledge. New York: Columbia University Press. Bloor, David. 1984. The sociology of reasons: Or why “epistemic factors” are really “social factors.” In Brown 1984, 295–324. Bloor, David. 1991. Knowledge and Social Imagery. 2nd ed. Chicago: University of Chicago Press. Bloor, David. 1992. Left and Right Wittgensteinians. In Pickering 1992b, 266–282. Bloor, David. 1997. Wittgenstein, Rules, and Institutions. New York: Routledge. Bloor, David. 2004. Institutions and rule-scepticism. Social Studies of Science 34: 593–601. Boffey, Phillip M. 1975. The Brain Bank of America. New York: McGraw-Hill. Bogen, David. 1999. Order without Rules. Albany: SUNY Press. Bohman, James. 1991. New Philosophy of Social Science. Cambridge, Mass.: MIT Press. Bohman, James. 1996. Public Deliberation. Cambridge, Mass.: MIT Press. Bohman, James. 1999. Democracy as inquiry, inquiry as democracy: pragmatism, social science, and the cognitive division of labor. American Journal of Political Science 43: 590–607. Bohman, James, and William Rehg, eds. 1997. Deliberative Democracy: Essays on Reason and Politics. Cambridge, Mass.: MIT Press. Bonß, Wolfgang. 1993. The program of interdisciplinary research and the beginnings of critical theory. In On Max Horkheimer, ed. S. Benhabib, W. Bonß, and J. McCole, 99–125. Cambridge, Mass.: MIT Press. Boykoff, Maxwell T., and Jules M. Boykoff. 2004. Balance as bias: Global warming and the U.S. prestige press. Global Environmental Change 14: 125–136. Brandom, Robert B. 1994. Making It Explicit. Cambridge, Mass.: Harvard University Press. Brannigan, Augustine. 1981. The Social Basis of Scientific Discourse. Cambridge: Cambridge University Press. Brante, Thomas, Steve Fuller, and William Lynch, eds. 1993. Controversial Science. Albany: SUNY Press. Brockriede, Wayne. 1982. Arguing about human understanding. Communication Monographs 49: 137–147. Brower, Bruce W. 1998. Contextualism, epistemological. In Routledge Encyclopedia of Philosophy, ed. E. Craig. London: Routledge. Retrieved August 11, 2006, from http://www.routledge.com/ article/P0658. Brown, Harold I. 1977. Perception, Theory, and Commitment. Chicago: Precedent. Brown, James Robert, ed. 1984. Scientific Rationality. Dordrecht: Reidel.
316 References
Brown, James Robert. 2001. Who Rules in Science? Cambridge, Mass.: Harvard University Press. Brown, Ken. 1998. Education, Culture, and Critical Thinking. Aldershot: Ashgate. Button, Graham, ed. 1991. Ethnomethodology and the Human Sciences. Cambridge: Cambridge University Press. Callon, Michel, John Law, and Arie Rip, eds. 1986. Mapping the Dynamics of Science and Technology. London: Macmillan. Campbell, John Angus. 1986. Scientific revolution and the grammar of culture: The case of Darwin’s Origin. Quarterly Journal of Speech 72: 351–376. Campbell, John Angus. 1990. Scientific discovery and rhetorical invention: The path to Darwin’s Origin. In The Rhetorical Turn, ed. H. W. Simons, 58–90. Chicago: University of Chicago Press. Campbell, Nancy D. 2004. Credible performances: The performativity of science studies. Social Studies of Science 34: 433–442. Carnap, Rudolf. 1932/1995. The Unity of Science. Trans. M. Black. Bristol: Thoemmes. Originally published in Erkenntnis 2: 432–465. Carnap, Rudolf. 1937. The Logical Syntax of Language. Trans. A. Smeaton. London: Routledge and Kegan Paul. Carnap, Rudolf. 1949. Truth and confirmation. In Readings in Philosophical Analysis, ed. H. Feigl and W. Sellars, 119–127. New York: Appleton-Century-Crofts. Carnap, Rudolf. 1952. The Continuum of Inductive Methods. Chicago: University of Chicago Press. Chilvers, Jason. 2008. Deliberating competence: Theoretical and practitioner perspectives on effective participatory appraisal practice. Science, Technology, and Human Values. 33: 421–451. Council for Agricultural Science and Technology (CAST). 1982. Diet, Nutrition, and Cancer: A Critique. Special Publication No. 13. Ames, Iowa. Clemens, Elizabeth S. 1986. Of asteroids and dinosaurs: The role of the press in the shaping of scientific debate. Social Studies of Science 16: 421–456. Clemens, Elizabeth S. 1994. The impact hypothesis and popular science: Conditions and consequences of interdisciplinary debate. In Glen 1994, 92–120. Cole, Stephen. 1992. Making Science. Cambridge, Mass.: Harvard University Press. Collingridge, David, and Colin Reeves. 1986. Science Speaks to Power. London: Frances Pinter. Collins, Harry M. 1981. What is TRASP? Philosophy of the Social Sciences 11: 215–224. Collins, Harry M. 1982. Knowledge, norms, and rules in the sociology of science. Social Studies of Science 12: 299–309.
317 References
Collins, Harry M. 1983. An empirical relativist programme in the sociology of scientific knowledge. In Knorr-Cetina and Mulkay 1983, 85–113. Collins, Harry M. 1985. Changing Order. London: Sage. Collins, Harry M., and Robert Evans. 2002. The Third Wave of science studies: Studies of expertise and experience. Social Studies of Science 32: 235–296. Cooke, Maeve. 1994. Language and Reason. Cambridge, Mass.: MIT Press. Cooke, Maeve. 2001a. Jürgen Habermas on meaning and truth. European Journal of Philosophy 9: 1–23. Cooke, Maeve. 2001b. Socio-cultural learning as a “transcendental” fact: Habermas’s postmetaphysical perspective. International Journal of Philosophical Studies 9: 63–83. Cooke, Maeve. 2006. Re-Presenting the Good Society. Cambridge, Mass.: MIT Press. Coulter, Jeff. 1983. Contingent and a priori structures in sequential analysis. Human Studies 6: 361–376. Cox, J. R., and C. A. Willard, eds. 1982. Advances in Argumentation Theory and Research. Carbondale: Southern Illinois University Press. Cramer, P. A. 2003. The public metonym. Informal Logic 23: 183–199. Crosswhite, James. 1996. The Rhetoric of Reason. Madison: University of Wisconsin Press. Davidson, Donald. 2001. The very idea of a conceptual scheme. In Davidson, Inquiries into Truth and Interpretation, 183–198. Oxford: Clarendon. Davis, Darin Harris. 2004. Rules and vision: Particularism in contemporary ethics. Dissertation, Saint Louis University, St. Louis, Missouri. Delamont, Sara. 2003. Rosalind Franklin and Lucky Jim. Social Studies of Science 33: 315–322. Douglas, Heather. 2000. Inductive risk and values in science. Philosophy of Science 67: 559–579. Dupré, John. 1993. The Disorder of Things. Cambridge, Mass.: Harvard University Press. Durkheim, Emile. 1964. The Division of Labor in Society. Trans. G. Simpson. New York: Free Press. Dworkin, Ronald. 1986. Law’s Empire. Cambridge, Mass.: Harvard University Press. Eemeren, Frans H. van, and Rob Grootendorst. 1992. Argumentation, Communication, and Fallacies. Hillsdale, N.J.: Erlbaum. Eemeren, Frans H. van, Rob Grootendorst, Francisca Snoeck Henkemans, J. Anthony Blair, Ralph H. Johnson, Erik C. W. Krabbe, Christain Plantin, Douglas N. Walton, Charles A. Willard, John Woods, and David Zarefsky. 1996. Fundamentals of Argumentation Theory. Mahwah, N.J.: Erlbaum.
318 References
Eemeren, Frans H. van, Rob Grootendorst, and Francisca Snoeck Henkemans. 2002. Argumentation: Analysis, Evaluation, Presentation. Mahwah, N.J.: Erlbaum. Eemeren, Frans H. van, Rob Grootendorst, Sally Jackson, and Scott Jacobs. 1993. Reconstructing Argumentative Discourse. Tuscaloosa: University of Alabama Press. Elster, Jon, ed. 1998. Deliberative Democracy. Cambridge: Cambridge University Press. Epstein, Steven. 1995. The construction of lay expertise: AIDS activism and the forging of credibility in the reform of clinical trials. Science, Technology, and Human Values 20: 408–437. Epstein, Steven. 2000. Democracy, expertise, and AIDS treatment activism. In Kleinman 2000, 15–32. Evans, Robert. 1999. Macroeconomic Forecasting. London: Routledge. Fabj, Valeria, and Matthew J. Sobnosky. 1995. AIDS activism and the rejuvenation of the public sphere. Argumentation and Advocacy 31: 163–184. Fedigan, Linda Marie. 1994. Science and the successful female: Why there are so many women primatologists. American Anthropologist (New Series) 96: 529–540. Feteris, Eveline T. 2003. The rationality of legal discourse in Habermas’s discourse theory. Informal Logic 23: 139–159. Finocchiaro, Maurice A. 1980. Galileo and the Art of Reasoning. Dordrecht: Reidel. Fischer, Frank. 2000. Citizens, Experts, and the Environment. Durham: Duke University Press. Fleischmann, Martin, and Stanley Pons. 1989. Electrochemically induced nuclear fusion of deuterium. Journal of Electroanalytical Chemistry 261 (2A): 301–308. Fortun, Mike, and Herbert J. Bernstein. 1998. Muddling Through. Washington, D.C.: Counterpoint. Friedman, Michael. 1998. On the sociology of scientific knowledge and its philosophical agenda. Studies in History and Philosophy of Science 29: 235–271. Friedman, Michael. 2003. Kuhn and logical empiricism. In Nickles 2003b, 19–44. Fuller, Steve. 1988. Social Epistemology. Bloomington: Indiana University Press. Fuller, Steve. 1993. Philosophy, Rhetoric, and the End of Knowledge. Madison: University of Wisconsin Press. Fuller, Steve. 1995. The Strong Program in the Rhetoric of Science. In Krips, McGuire, and Melia 1995, 95–115. Fuller, Steve. 2000a. Thomas Kuhn. Chicago: University of Chicago Press. Fuller, Steve. 2000b. The Governance of Science. Philadelphia: Open University Press. Fuller, Steve. 2000c. Why science studies has never been critical of science: Some recent lessons on how to be a helpful nuisance and a harmless radical. Philosophy of the Social Sciences 30: 5–32.
319 References
Fuller, Steve, and James H. Collier. 2004. Philosophy, Rhetoric, and the End of Knowledge. 2nd ed. Mahwah, N.J.: Erlbaum. Galison, Peter. 1987. How Experiments End. Chicago: University of Chicago Press. Galison, Peter. 1997. Image and logic. Chicago: University of Chicago Press. Gallie, W. B. 1955–1956. Essentially contested concepts. Proceedings of the Aristotelian Society 56: 167–198. Gaonkar, Dilip Parameshwar. 1997a. The idea of rhetoric in the rhetoric of science. In Gross and Keith 1997, 25–85. Gaonkar, Dilip Parameshwar. 1997b. Close readings of the third kind: Reply to my critics. In Gross and Keith 1997, 330–356. Garfinkel, Harold. 1963. A conception of, and experiments with, “trust” as a condition of stable concerted actions. In Motivation and Social Interaction, ed. O. J. Harvey, 187–238. New York: Ronald. Garfinkel, Harold. 1967. Studies in Ethnomethodology. Englewood Cliffs, N.J.: Prentice-Hall. Garfinkel, Harold, Michael Lynch, and Eric Livingston. 1981. The work of a discovering science construed with materials from the optically discovered pulsar. Philosophy of the Social Sciences 11: 131–158. Garfinkel, Harold, and Harvey Sacks. 1970. On formal structures of practical action. In Theoretical Sociology, ed. J. C. McKinney and E. A. Tiryakian, 337–366. New York: Meredith-Appleton. Gero, Joan M. 1993. The social world of prehistoric facts: Gender and power in paleoindian research. In Women in Archaeology, ed. H. du Cros and L. Smith, 31–40. Canberra: Australian National University. Gibbons, Michael. 1999. Science’s new contract with society. Nature 402 (December 2, 1999, supplement): C81–C84. Gilbert, G. Nigel, and Michael Mulkay. 1984. Opening Pandora’s Box. Cambridge: Cambridge University Press. Gilbert, Michael A. 1994. Multi-modal argumentation. Philosophy of the Social Sciences 24: 159–177. Gilbert, Michael A. 1997. Coalescent Argumentation. Mahwah, N.J.: Erlbaum. Glen, William, ed. 1994. The Mass-Extinction Debates. Stanford: Stanford University Press. Goffman, Erving. 1959. The Presentation of Self in Everyday Life. Garden City, N.Y.: Anchor-Doubleday. Goldman, Alvin. 1994. Argumentation and social epistemology. Journal of Philosophy 91: 27–49. Goldman, Alvin. 1999. Knowledge in a Social World. Oxford: Clarendon. Goldstein, Jeffrey H., ed. 1986. Reporting Science. Hillsdale, N.J.: Erlbaum.
320 References
Golinksi, Jan. 1998. Making Natural Knowledge. Cambridge: Cambridge University Press. Goodnight, G. Thomas. 1982. The personal, technical, and public spheres of argument: A speculative inquiry into the art of public deliberation. Journal of the American Forensic Association 18: 214–227. Goodwin, Jean. 2005. The public sphere and the norms of transactional argument. Informal Logic 25: 151–165. Govier, Trudy. 1987. Problems in Argument Analysis and Evaluation. Dordrecht: Foris. Govier, Trudy. 2005. A Practical Study of Argument. 6th ed. Belmont, Calif.: Wadsworth. Gross, Alan G. 1996. The Rhetoric of Science. 2nd ed. Cambridge, Mass.: Harvard University Press. Gross, Alan G., and William Keith, eds. 1997. Rhetorical Hermeneutics. Abany: SUNY Press. Gross, Ludwik. 1970. Oncogenic Viruses. 2nd ed. Oxford: Pergamon. Gusterson, Hugh. 1996. Nuclear Rites. Berkeley: University of California Press. Gutting, Gary. 1984. The strong program: A dialogue. In Brown 1984, 95–111. Haack, Susan. 2003. Defending Science—Within Reason. Amherst, N.Y.: Prometheus. Habermas, Jürgen. 1970. Toward a Rational Society. Trans. J. J. Shapiro. Boston: Beacon. Habermas, Jürgen. 1971/2001. Reflections on the linguistic foundations of sociology: The Christian Gauss Lectures (Princeton University, 1971). In Habermas, On the Pragmatics of Social Interaction, trans. B. Fultner, 1–103. Cambridge, Mass.: MIT Press. Habermas, Jürgen. 1973/1986. Wahrheitstheorien. In Habermas, Vorstudien und Ergänzungen zur Theorie des kommunikativen Handelns, 2nd ed., 127–183. Frankfurt am Main: Suhrkamp. Habermas, Jürgen. 1979. Communication and the Evolution of Society. Trans. T. McCarthy. Boston: Beacon. Habermas, Jürgen. 1984. The Theory of Communicative Action. Vol. 1. Trans. T. McCarthy. Boston: Beacon. Habermas, Jürgen. 1987. The Theory of Communicative Action. Vol. 2. Trans. T. McCarthy. Boston: Beacon. Habermas, Jürgen. 1988. On the Logic of the Social Sciences. Trans. S. W. Nicholsen and J. A. Stark. Cambridge, Mass.: MIT Press. Habermas, Jürgen. 1989. The Structural Transformation of the Public Sphere. Trans. T. Burger and F. Lawrence. Cambridge, Mass.: MIT Press. Habermas, Jürgen. 1990. Moral Consciousness and Communicative Action. Trans. C. Lenhardt and S. W. Nicholsen. Cambridge, Mass.: MIT Press.
321 References
Habermas, Jürgen. 1993. Justification and Application. Trans. C. P. Cronin. Cambridge, Mass.: MIT Press. Habermas, Jürgen. 1996. Between Facts and Norms. Trans. W. Rehg. Cambridge, Mass.: MIT Press. Habermas, Jürgen. 1998a. The Inclusion of the Other. Ed. C. Cronin and P. DeGreiff. Cambridge, Mass.: MIT Press. Habermas, Jürgen. 1998b. On the Pragmatics of Communication. Trans. B. Fultner. Cambridge, Mass.: MIT Press. Habermas, Jürgen. 2003. Truth and Justification. Trans. B. Fultner. Cambridge, Mass.: MIT Press. Habermas, Jürgen. 2005. Zwischen Naturalismus und Religion. Frankfurt am Main: Suhrkamp. Hales, Steven D. 1997. A consistent relativism. Mind 106: 33–52. Hallett, Garth L. 1988. Language and Truth. New Haven: Yale University Press. Hample, Dale. 2005. Arguing: Exchanging Reasons Face to Face. Mahwah, N.J.: Erlbaum. Hands, D. Wade. 2001. Reflection without Rules. Cambridge: Cambridge University Press. Hanson, Norwood Russell. 1958. Patterns of Discovery. Cambridge: Cambridge University Press. Haraway, Donna. 1986. Primatology is politics by other means. In Feminist Approaches to Science, ed. R. Bleier, 77–118. New York: Pergamon. Hardcastle, Valerie Gray. 1999. Scientific papers have various structures. Philosophy of Science 66: 415–447. Harding, Sandra. 1986. The Science Question in Feminism. Ithaca: Cornell University Press. Harding, Sandra. 1998. Is Science Multicultural? Bloomington: Indiana University Press. Harris, Randy Allen. 1995. Review of Marcello Pera, Discourses of Science. Rhetoric Review 14: 207–211. Harris, Randy Allen, ed. 1997. Landmark Essays on Rhetoric of Science. Mahwah, N.J.: Hermagoras. Haught. John F. 2000. God after Darwin. Boulder: Westview. Heath, Joseph. 2001. Communicative Action and Rational Choice. Cambridge, Mass.: MIT Press. Hempel, Carl G. 1937/2000. The problem of truth. In Hempel 2000, chap. 4. First published in French in Theoria 3: 206–246. Hempel, Carl G. 1943. A purely syntactical definition of confirmation. Journal of Symbolic Logic 8: 122–143. Hempel, Carl G. 1945. Studies in the logic of confirmation. Mind 54: 1–26, 97–121.
322 References
Hempel, Carl G. 1946. A note on the paradoxes of confirmation. Mind 55: 79–82. Hempel, Carl G. 1966. Philosophy of Natural Science. Englewood Cliffs, N.J.: Prentice Hall. Hempel, Carl G. 1990/2000. The irrelevance of the concept of truth for the critical appraisal of scientific theories. In Hempel 2000, chap. 5. First published in Nuova Civilità della Macchine 8: 109–113. Hempel, Carl G. 2000. Selected Philosophical Essays. Ed. R. Jeffrey. Cambridge: Cambridge University Press. Hempel, Carl G., and Paul Oppenheim. 1945/2000. A definition of “degree of confirmation.” In Hempel 2000, 135–161. First published in Philosophy of Science 12: 98–112. Heritage, John. 1984. Garfinkel and Ethnomethodology. Cambridge: Polity. Hess, David J. 1997. Science Studies. New York: New York University. Hilgartner, Stephen. 2000. Science on Stage. Stanford: Stanford University Press. Hitchcock, David. 1992. Relevance. Argumentation 6: 251–270. Hollenbach, David, S. J. 2002. The Common Good and Christian Ethics. Cambridge: Cambridge University Press. Hollis, Martin, and Steven Lukes, eds. 1982. Rationality and Relativism. Oxford: Blackwell. Holton, Gerald. 1973. The Thematic Origins of Scientific Thought. Cambridge, Mass.: Harvard University Press. Hooker, Brad, and Margaret Little. 2000. Moral Particularism. Oxford: Clarendon. Horkheimer, Max. 1931/1993. The present situation of social philosophy and the tasks of an Institute for Social Research. In Horkheimer, Between Philosophy and the Social Sciences, 1–14. Cambridge, Mass.: MIT Press. Hoyningen-Huene, Paul. 1993. Reconstructing Scientific Revolutions. Trans. A. T. Levine. Chicago: University of Chicago Press. Hull, David. 1988. Science as a Process. Chicago: University of Chicago Press. Jacobs, Scott. 2000. Rhetoric and dialectic from the standpoint of normative pragmatics. Argumentation 14: 261–286. Jasanoff, Sheila. 1987. Contested boundaries in policy-relevant science. Social Studies of Science 17: 195–230. Jasanoff, Sheila. 1990. The Fifth Branch. Cambridge, Mass.: Harvard University Press. Jasanoff, Sheila. 1991. Acceptable evidence in a pluralistic society. In Mayo and Hollander 1991, 29–47. Jasanoff, Sheila. 1995. Science at the Bar. Cambridge, Mass.: Harvard University Press.
323 References
Jasanoff, Sheila. 2003. (No?) Accounting for expertise. Science and Public Policy 30: 157–162. Jasanoff, Sheila. 2004. Science and citizenship: A new synergy. Science and Public Policy 31: 90–94. Jasanoff, Sheila. 2005. Designs on Nature. Princeton: Princeton University Press. Jeffrey, Richard. 2000. Editor’s introduction to Carl Hempel: Selected Philosophical Essays. Cambridge: Cambridge University Press. Johnson, Ralph H. 2000. Manifest Rationality. Mahwah, N.J.: Erlbaum. Jordan, Kathleen, and Michael Lynch. 1992. The sociology of a genetic engineering technique: Ritual and rationality in the performance of the “plasmid prep.” In The Right Tools for the Job, ed. A. Clarke and J. Fujimura, 77–114. Princeton: Princeton University Press. Jordan, Kathleen, and Michael Lynch. 1998. The dissemination, standardization, and routinization of a molecular biological technique. Social Studies of Science 28: 773–800. Judson, Horace Freeland. 1996. The Eighth Day of Creation. Expanded ed. Cold Spring Harbor: Cold Spring Harbor Laboratory Press. Kauffeld, Fred J. 2002. Pivotal issues and norms in rhetorical theories of argumentation. In Dialectic and Rhetoric, ed. F. H. van Eemeren and P. Houtlosser, 97–118. Dordrecht: Kluwer. Keith, William M., and Richard A. Cherwitz. 1989. Objectivity, disagreement, and the rhetoric of inquiry. In Simons 1989, 195–210. Keith, William, and William Rehg. 2008. Argumentation in science: The cross-fertilization of argumentation theory and science studies. In The Handbook of Science and Technology Studies, 3rd ed., ed. E. J. Hackett, O. Amsterdamska, M. Lynch, and J. Wajcman, eds., 211–239. Cambridge, Mass.: MIT Press. Keller, Evelyn Fox. 1983. A Feeling for the Organism. New York: Freeman. Kevles, Daniel J. 1979. The Physicists. New York: Vintage-Random House. Kevles, Daniel J. 1995. Pursuing the unpopular: A history of courage, viruses, and cancer. In Hidden Histories of Science, ed. R. B. Silvers. New York: New York Review. Kim, Kyung-Man. 1994. Explaining Scientific Consensus. New York: Guilford. Kincaid, Harold, John Dupré, and Alison Wylie, eds. 2007. Value-Free Science? Ideals and Illusions. Oxford: Oxford University Press. Kirkham, Richard L. 1992. Theories of Truth. Cambridge, Mass.: MIT Press. Kitcher, Philip. 1982. Abusing Science. Cambridge, Mass.: MIT Press. Kitcher, Philip. 1993. The Advancement of Science. New York: Oxford University Press. Kitcher, Philip. 1995. The cognitive functions of scientific rhetoric. In Krips, McGuire, and Melia 1995, 47–66.
324 References
Kitcher, Philip. 2000. Patterns of scientific controversies. In Machamer, Pera, and Baltas 2000, 21–39. Kitcher, Philip. 2001. Science, Truth, and Democracy. Oxford: Oxford University Press. Klein, Julie Thompson. 1996. Crossing Boundaries. Charlottesburg: University Press of Virginia. Kleinman, Daniel Lee, ed. 2000. Science, Technology, and Democracy. Albany: SUNY Press. Knorr-Cetina, Karin D. 1981. The Manufacture of Knowledge. New York: Pergamon. Knorr-Cetina, Karin D. 1999. Epistemic Cultures. Cambridge, Mass.: Harvard University Press. Knorr-Cetina, Karin D, and Michael Mulkay, eds. 1983. Science Observed. London: Sage. Kragh, Helga. 1999. Quantum Generations. Princeton: Princeton University Press. Krips, Henry, J. E. McGuire, and Trevor Melia, eds. 1995. Science, Reason, and Rhetoric. Pittsburgh: University of Pittsburgh Press. Kuhn, Thomas S. 1970a. Logic of discovery or psychology of research? In Lakatos and Musgrave 1970, 1–23. Kuhn, Thomas S. 1970b. Reflections on my critics. In Lakatos and Musgrave 1970, 231–278. Kuhn, Thomas S. 1977. The Essential Tension. Chicago: University of Chicago Press. Kuhn, Thomas S. 1996. The Structure of Scientific Revolutions. 3rd ed. Chicago: University of Chicago Press. First published 1962. Kuhn, Thomas S. 2000. The Road since “Structure.” Ed. J. Conant and J. Haugeland. Chicago: University of Chicago Press. Kusch, Martin. 2002. Knowledge by Agreement. Oxford: Clarendon. Kusch, Martin. 2004a. Rule-scepticism and the sociology of scientific knowledge. Social Studies of Science 34: 571–591. Kusch, Martin. 2004b. Reply to my critics. Social Studies of Science 34: 615–620. Kyburg, Henry E., Jr. 1968. Philosophy of Science: A Formal Approach. New York: Macmillan. Kyburg, Henry E., Jr. 1970. Probability and Inductive Logic. London: Collier-Macmillan. Labinger, Jay A., and Harry Collins, eds. 2001. The One Culture? Chicago: University of Chicago Press. Lahsen, Myanna. 2005. Technocracy, democracy, and U. S. climate politics: The need for demarcations. Science, Technology, and Human Values 30: 137–169. Lakatos, Imre. 1970. Falsification and the methodology of scientific research programmes. In Lakatos and Musgrave 1970, 91–196.
325 References
Lakatos, Imre. 1976. Proofs and Refutations. Cambridge: Cambridge University Press. Lakatos, Imre, and Alan Musgrave, eds. 1970. Criticism and the Growth of Knowledge. Cambridge: Cambridge University Press. Latour, Bruno. 1983. Give me a laboratory and I will raise the world. In Knorr-Cetina and Mulkay 1983, 141–170. Latour, Bruno. 1987. Science in Action. Cambridge: Harvard University Press. Latour, Bruno. 1988. The Pasteurization of France. Trans. A. Sheridan and J. Law. Cambridge: Harvard University Press. Originally published in French, 1984. Latour, Bruno. 1992. One more turn after the social turn. In The Social Dimensions of Science, ed. E. McMullin, 272–294. Notre Dame: University of Notre Dame Press. Latour, Bruno. 1993. We Have Never Been Modern. Trans. C. Porter. Cambridge, Mass.: Harvard University Press. Latour, Bruno. 1999a. On recalling ANT. In Actor Network Theory and After, ed. J. Law and J. Hassard, 15–25. Oxford: Blackwell/Sociological Review. Latour, Bruno. 1999b. Pandora’s Hope. Cambridge, Mass.: Harvard University Press. Latour, Bruno. 2004. Politics of Nature. Trans. C. Porter. Cambridge, Mass.: Harvard University Press. Latour, Bruno. 2005. Reassembling the Social. Oxford: Oxford University Press. Latour, Bruno, and Steve Woolgar. 1986. Laboratory Life, 2nd ed. Princeton: Princeton University Press. Laudan, Larry. 1977. Progress and Its Problems. Berkeley: University of California Press. Leff, Michael. 2002. The relation between dialectic and rhetoric in a classical and a modern perspective. In Dialectic and Rhetoric, ed. F. H. van Eemeren and P. Houtlosser, 53–63. Dordrecht: Kluwer. Lengwiler, Martin. 2008. Participatory approaches in science and technology: Historical origins and current practices in critical perspective. Science, Technology, and Human Values 33: 186–200. Lewandowski, J. D. 2001. Interpreting Culture. Lincoln: University of Nebraska Press. List, Christian. 2006. The discursive dilemma and public reason. Ethics 116: 362–402. Livingston, Eric. 1986. The Ethnomethodological Foundations of Mathematics. London: Routledge. Livingston, Eric. 1987. Making Sense of Ethnomethodology. New York: Routledge. Livingston, Eric. 1995. The idiosyncratic specificity of the methods of physical experimentation. Australian and New Zealand Journal of Sociology 31, no. 3: 1–22.
326 References
Livingston, Eric. 1999. Cultures of proving. Social Studies of Science 29: 867–888. Lonergan, Bernard J. F. 1970. Insight. 3rd ed. New York: Philosophical Library. Longino, Helen E. 1990. Science as Social Knowledge. Princeton: Princeton University Press. Longino, Helen E. 1996. Cognitive and non-cognitive values in science: Rethinking the dichotomy. In Feminism, Science, and Philosophy of Science, ed. L. H. Nelson and J. Nelson, 39–58. Dordrecht: Kluwer. Longino, Helen E. 1997. Feminist epistemology as local epistemology. Proceedings of the Aristotelian Society (Suppl. Vol.) 71: 19–35. Longino, Helen E. 2002. The Fate of Knowledge. Princeton: Princeton University Press. Lucas, Stephen E. 1981. The schism in rhetorical scholarship. Quarterly Journal of Speech 67: 1–20. Reprinted in Readings in Rhetorical Criticism, 2nd ed., ed. C. R. Burgchardt, 88–107. State College, Penn.: Strata, 2000. Lynch, Michael. 1985. Art and Artifact in Laboratory Science. London: Routledge and Kegan Paul. Lynch, Michael. 1992. [Exchange with David Bloor 1992:] Extending Wittgenstein: The pivotal move from epistemology to the sociology of science. [And]: From the “will to theory” to the discursive collage: A reply to Bloor’s “Left and Right Wittgensteinians.” In Pickering 1992b, 215–65, 283–300. Lynch, Michael. 1993. Scientific Practice and Ordinary Action. Cambridge: Cambridge University Press. Lynch, Michael. 1997a. Ethnomethodology without indifference. Human Studies 20: 371–376. Lynch, Michael. 1997b. Theorizing practice. Human Studies 20: 335–344. Lynch, Michael. 1998. The discursive production of uncertainty: The OJ Simpson “Dream Team” and the sociology of knowledge machine. Social Studies of Science 28: 829–868. Lynch, Michael. 1999. Silence in context: Ethnomethodology and social theory. Human Studies 22: 211–233. Lynch, Michael. 2000. The discourse of courtroom science: An ethnomethodological view. Paper delivered at a Conference on Studies of Work, Technology, and Practice. Harvard University, Cambridge, Mass., May 5. Lynch, Michael. 2004. Circumscribing expertise: Membership categories in courtroom testimony. In States of Knowledge, ed. S. Jasanoff, 161–180. London: Routledge. Lynch, Michael, and Steve Woolgar. 1988. Introduction: sociological orientations to representational practice in science. Human Studies 11: 99–116. Lynch, Michael P. 1998. Truth in Context. Cambridge, Mass.: MIT Press. Lynch, William T. 1994. Ideology and the sociology of scientific knowledge. Social Studies of Science 24: 197–227.
327 References
Machamer, Peter, Marcello Pera, and Aristides Baltas, eds. 2000. Scientific Controversies. New York: Oxford. MacIntyre, Alasdair. 1988. After Virtue. 2nd ed. Notre Dame: University of Notre Dame Press. MacKenzie, David. 1978. Statistical theory and social interests. Social Studies of Science 8: 35–83. MacKenzie, Donald, and Barry Barnes. 1979. Scientific judgment: The biometry-Mendelism controversy. In Natural Order, ed. B. Barnes and S. Shapin, 191–210. Beverly Hills: Sage. Maher, Patrick. 1999. Inductive logic and the raven paradox. Philosophy of Science 66: 50–70. Marcum, James A. 2002. The transformation of oncology in the twentieth century: The molecularization of cancer. Proceedings of the 37th International Congress on the History of Medicine, 41–49. Galveston, Tex.: University of Texas Medical Branch. Marcuse, Herbert. 1990. The sociological method and the problem of truth. In Knowledge and Politics, ed. V. Meja and N. Stehr, chap. 9. London: Routledge. Mayo, Deborah G. 1991. Sociological versus metascientific views of risk assessment. In Mayo and Hollander 1991, 249–279. Mayo, Deborah G. 1996. Error and the Growth of Experimental Knowledge. Chicago: University of Chicago Press. Mayo, Deborah G., and Rachelle D. Hollander, eds. 1991. Acceptable Evidence. New York: Oxford University Press. McCabe, Mary Margaret. 1994. Arguments in context: Aristotle’s defense of rhetoric. In Aristotle’s Rhetoric: Philosophical Essays, ed. D. J. Furley and A. Nehamas, 129–165. Princeton: Princeton University Press. McCarthy, Thomas. 1978. The Critical Theory of Jürgen Habermas. Cambridge, Mass.: MIT Press. McCarthy, Thomas. 1988. Scientific rationality and the “Strong Program” in the sociology of knowledge. In Construction and Constraint, ed. E. McMullin, 75–95. Notre Dame: University of Notre Dame Press. McCarthy, Thomas. 1991. Ideals and Illusions. Cambridge, Mass.: MIT Press. McCarthy, Thomas. 1994. Critical Theory, chaps. 1–3, 7. Other chapters by David Hoy. Oxford: Blackwell. McCarthy, Thomas. 1998. Legitimacy and diversity: Dialectical reflections on analytic distinctions. In Habermas on Law and Democracy, ed. M. Rosenfeld and A. Arato, 115–153. Berkeley: University of California Press. McElheny, Victor K. 2003. Watson and DNA. Cambridge, Mass.: Perseus. McMullin, Ernan. 1987. Scientific controversy and its termination. In Scientific Controversies, ed. H. T. Engelhardt and A. L. Caplan, 49–91. Cambridge: Cambridge University Press. McMullin, Ernan. 1991. Rhetoric and theory choice in science. In Pera and Shea 1991, 55–76.
328 References
McPeck, John E. 1990. Teaching Critical Thinking. New York: Routledge. Meiland, Jack W. 1989. Argument as inquiry and persuasion. Argumentation 3: 185–196. Mendieta, Eduardo. 2002. The Adventures of Transcendental Philosophy. Lanham, Maryland: Rowman and Littlefield. Merton, Robert K. 1973. The Sociology of Science. Chicago: University of Chicago Press. Miller, Kenneth R. 1999. Finding Darwin’s God. New York: HarperCollins. Misak, Cheryl J. 1991. Truth and the End of Inquiry. Oxford: Clarendon. Misak, Cheryl J. 2000. Truth, Politics, and the End of Inquiry. London: Routledge. Montgomery, Scott L. 2003. The Chicago Guide to Communicating Science. Chicago: University of Chicago Press. Morgan, M. G. 1993. Risk analysis and risk management. Scientific American (July): 32–41. Myers, Greg. 1990. Writing Biology. Madison: University of Wisconsin Press. Naess, Arne. 1966. Communication and Argument. London: Allen and Unwin. Oslo ed., 1947. National Academy of Sciences (NAS). 1982. Diet, Nutrition, and Cancer. Washington, D.C.: National Academy Press. Available online at http://books.nap.edu/books/0309032806/html/index .html. National Academy of Sciences (NAS). 2008. Science, Evolution, and Creation. Washington, D.C.: National Academies Press. Available online at http://www.nap.edu/catalogue/11876. html. Nickles, Thomas. 1980. Introductory essay: Scientific discovery and the future of the philosophy of science. In Scientific Discovery, Logic, and Rationality, ed. T. Nickles, 1–59. Dordrecht: Reidel. Nickles, Thomas. 1989. Heuristic appraisal. Social Epistemology 3: 175–188. Nickles, Thomas. 2003a. Normal science: From logic to case-based and model-based reasoning. In Nickles 2003b, 142–177. Nickles, Thomas, ed. 2003b. Thomas Kuhn. Cambridge: Cambridge University Press. Norris, Stephen P., ed. 1992. The Generalizability of Critical Thinking. New York: Columbia University, Teacher’s College Press. Nowotny, Helga, Peter Scott, and Michael Gibbons. 2001. Re-Thinking Science. Cambridge: Polity. Officer, Charles, and Jake Page. 1996. The Great Dinosaur Extinction Controversy. Reading, Mass.: Helix/Addison-Wesley. O’Keefe, Daniel J. 1982. The concepts of argument and arguing. In Cox and Willard 1982, 3–23.
329 References
Overbye, Dennis. 2005a. Philosophers notwithstanding, Kansas School Board redefines science. New York Times. November 15, D3. Overbye, Dennis. 2005b. One find, two astronomers: An ethical brawl. New York Times. September 13, D1, D3. Pera, Marcello. 1987. The rationality of discovery: Galvani’s animal electricity. In Rational Changes in Science, ed. J. C. Pitt and M. Pera, 177–201. Dordrecht: Reidel. Pera, Marcello. 1991. The role and value of rhetoric in science. In Pera and Shea 1991, 29–54. Pera, Marcello. 1992. The Ambiguous Frog. Trans. J. Mandelbaum. Princeton: Princeton University Press. Pera, Marcello. 1994. The Discourses of Science. Trans. C. Botsford. Chicago: University of Chicago Press. Pera, Marcello. 2000. Rhetoric and scientific controversies. In Machamer, Pera, and Baltas 2000, 50–66. Pera, Marcello, and William R. Shea, eds. 1991. Persuading Science. Canton, Mass.: Science History Publications/USA. Perelman, Chaim, and Lucie Olbrechts-Tyteca. 1969. The New Rhetoric. Trans. J. Wilkinson and P. Weaver. Notre Dame: University of Notre Dame Press. Paris ed., 1958. Perelman, Chaim. 1982. The Realm of Rhetoric. Trans. W. Kluback. Notre Dame: University of Notre Dame Press. Pettit, Philip. 2001. Deliberative democracy and the discursive dilemma. Philosophical Issues 11: 268–299. Pickering, Andrew. 1984. Against putting the phenomena first: The discovery of the weak neutral current. Studies in History and Philosophy of Science 15: 85–117. Pickering, Andrew. 1992a. From science as knowledge to science as practice. In Pickering 1992b, 1–26. Pickering, Andrew, ed. 1992b. Science as Practice and Culture. Chicago: University of Chicago Press. Pickering, Andrew. 1995. Beyond constraint: The temporality of practice and the historicity of knowledge. In Scientific Practice, ed. J. Z. Buchwald, 42–55. Chicago: University of Chicago Press. Pinch, Trevor. 1993. Three generations of SSK. Social Studies of Science 23: 363–373. Pinch, Trevor. 2001. Does science studies undermine science? Wittgenstein, Turing, and Polanyi as precursors for science studies and the science wars. In Labinger and Collins 2001, 13–26. Plantinga, Alvin. 1993. Warrant: The Current Debate. New York: Oxford University Press. Polanyi, Michael. 1951. The Logic of Liberty. Chicago: University of Chicago Press.
330 References
Polanyi, Michael. 1964. Personal Knowledge. New York: Harper. First edition 1958. Polanyi, Michael. 1969. The republic of science: Its political and economic theory (1962). In Knowing and Being, ed. M. Grene, 49–72. Chicago: University of Chicago Press. Polanyi, Michael. 1983. The Tacit Dimension. Gloucester, Mass.: Peter Smith. Doubleday ed., 1966. Pollner, Melvin. 1990. Mundane reasoning. In Ethnomethodological Sociology, ed. J. Coulter, 138–157. Brookfield, Vermont: Elgar. Originally published in Philosophy of the Social Sciences 4 (1974): 35–54. Popper, Karl R. 1980. The Logic of Scientific Discovery. London: Routledge. First published by Hutchinson Education, 1959. Popper, Karl R. 1989. Conjectures and Refutations. London: Routledge. Potter, Elizabeth. 2001. Gender and Boyle’s Law of Gases. Bloomington: Indiana University Press. Potter, Jonathan. 1996. Representing Reality. London: Sage. Prelli, Lawrence J. 1989a. A Rhetoric of Science. Columbia: University of South Carolina Press. Prelli, Lawrence J. 1989b. The rhetorical constructions of scientific ethos. In Simons 1989, 48–68. Preyer, Gerhard, and Georg Peter, eds. 2005. Contextualism in Philosophy. Oxford: Oxford University Press. Proctor, Robert N. 1991. Value-Free Science? Purity and Power in Modern Knowledge. Cambridge, Mass.: Harvard University Press. Radder, Hans. 1992. Normative reflections on constructivist approaches to science and technology. Social Studies of Science 22: 141–173. Raup, David M. 1994. The extinction debates: A view from the trenches. In Glen 1994: 145–151. Rawls, John. 1996. Political Liberalism. 2nd ed. New York: Columbia University Press. Recanati, François. 2004. Literal Meaning. Cambridge: Cambridge University Press. Recanati, François. 2005. Literalism and contextualism: Some varieties. In Preyer and Peters 2005, 171–196. Rehg, William. 1994. Insight and Solidarity. Berkeley: University of California Press. Rehg, William. 1997. Reason and rhetoric in Habermas’s theory of argumentation. In Rhetoric and Hermeneutics in Our Time, ed. W. Jost and M. J. Hyde, 358–377. New Haven: Yale University Press. Rehg, William. 2002. The argumentation theorist in deliberative democracry. Controversia 1, no. 1: 18–42.
331 References
Rehg, William. 2008. Cogency in motion: Critical contextualism and relevance. Argumentation, online first version. Available at http://dx.doi.org/10.1007/s10503-008-9114-y. Rehg, William, and James Bohman. 2002. Discourse and democracy: The formal and informal bases of legitimacy in Habermas’s Between Facts and Norms. In Discourse and Democracy, ed. R. vom Schomberg and K. Baynes, 31–60. New Haven: Yale University Press. Rehg, William, and Kent Staley. 2008. The CDF collaboration and argumentation theory: The role of process in objective knowledge. Perspectives on Science. 16: 1–25. Remedios, Francis. 2003. Legitimizing Scientific Knowledge. Lanham, Maryland: Lexington. Rescher, Nicholas. 1958. A theory of evidence. Philosophy of Science 25: 83–94. Rescher, Nicholas. 1976a. Peirce and the economy of research. Philosophy of Science 43: 71–98. Rescher, Nicholas. 1976b. Plausible Reasoning. Assen/Amsterdam: Van Gorcum. Rescher, Nicholas. 1977. Dialectics. Albany: SUNY Press. Rescher, Nicholas. 1997. H2O: Hempel-Helmer-Oppenheim, an episode in the history of scientific philosophy in the 20th century. Philosophy of Science 64: 334–360. Richardson, Alan, and Thomas Uebel, eds. 2007. The Cambridge Companion to Logical Empiricism. New York: Cambridge University Press. Rieff, David. 2005. Illness as more than metaphor. New York Times Magazine (December 4): 52–57. Rorty, Richard. 1987. Pragmatism and philosophy. In After Philosophy, ed. K. Baynes, J. Bohman, and T. McCarthy, 26–66. Cambridge, Mass.: MIT Press. Rorty, Richard. 1989. Contingency, Irony, and Solidarity. Cambridge: Cambridge University Press. Rorty, Richard. 1991. Objectivism, Relativism, and Truth. Cambridge: Cambridge University Press. Rorty, Richard. 2001. The ambiguity of rationality. In Pluralism and the Pragmatic Turn, ed. W. Rehg and J. Bohman, 41–52. Cambridge, Mass.: MIT Press. Roth, Paul, and Robert Barrett. 1990. Deconstructing quarks. Social Studies of Science 20: 579–632. Rouse, Joseph. 1996. Engaging Science. Ithaca: Cornell University Press. Rouse, Joseph. 2001. Two concepts of practices. In The Practice Turn in Contemporary Theory, ed. T. R. Schatzki, K. Knorr Cetina, and E. von Savigny, 189–198. London: Routlege. Rouse, Joseph. 2003. Kuhn’s philosophy of scientific practice. In Nickles 2003b, 101–121. Rudwick, Martin J. S. 1985. The Great Devonian Controversy. Chicago: University of Chicago Press. Ryan, James. 2006. Balance versus “balance.” Unpublished ms.
332 References
Sacks, Harvey. 1975. Everyone has to lie. In Socio-Cultural Dimensions of Language Use, ed. B. Blount and M. Sanchez, 57–79. New York: Academic. Salmon, Wesley C. 1967. The Foundations of Scientific Inference. Pittsburgh: University of Pittsburgh Press. Salmon, Wesley C. 1983. Confirmation and relevance. In The Concept of Evidence, ed. P. Achinstein. Oxford: Oxford University Press. (Selection from Salmon, Induction, Probability and Confirmation.) Schiappa, Edward. 1995. Editor’s introduction to Warranting Assent. Albany: SUNY Press. Schiebinger, Londa. 1997. Creating sustainable science. Osiris 12: 201–216. Schiebinger, Londa. 1999. Has Feminism Changed Science? Cambridge, Mass.: Harvard University Press. Sclove, Richard E. 1995. Democracy and Technology. New York: Guilford. Sen, Amartya. 2005. The Argumentative Indian. New York: Farrar, Straus, and Giroux. Shapere, Dudley. 1984. Reason and the Search for Knowledge. Dordrecht: Reidel. Shapere, Dudley. 1986. External and internal factors in the development of science. Science and Technology Studies 4: 1–9. Shapin, Steven. 1982. History of science and its sociological reconstructions. History of Science 20: 157–211. Shapin, Steven, and Simon Schaffer. 1985. Leviathan and the Air Pump. Princeton: Princeton University Press. Sharrock, Wes. 2004. No case to answer. Social Studies of Science. 34: 603–614. Sharrock, Wes, and Graham Button. 1999. Do the right thing! Rule finitism, rule scepticism, and rule following. Human Studies 22: 193–210. Sharrock, Wes, and Rupert Read. 2002. Kuhn: Philosopher of Scientific Revolution. Cambridge: Polity. Sills, Chip, and George H. Jensen, eds. 1992. The Philosophy of Discourse. 2 vols. Portsmouth, N.H.: Boynton/Cook-Heinmann. Simons, Herbert W., ed. 1989. Rhetoric in the Human Sciences. London: Sage. Simons, Herbert W., ed. 1990. The Rhetorical Turn. Chicago: University of Chicago Press. Solomon, Miriam. 2001. Social Empiricism. Cambridge, Mass.: MIT Press. Sosa, Ernest. 2005. Contextualism, epistemic, recent work on. In Routledge Encyclopedia of Philosophy, ed. E. Craig. London: Routledge. Retrieved August 11, 2006, from http://www.routledge .com/article/P061. Snow, C. P. 1961. The Two Cultures and the Scientific Revolution. New York: Cambridge University Press.
333 References
Staley, Kent W. 2002. What experiment did we just do? Counterfactual error statistics and uncertainties about the reference class. Philosophy of Science 69: 279–299. Staley, Kent W. 2004a. The Evidence for the Top Quark. Cambridge: Cambridge University Press. Staley, Kent W. 2004b. Robust evidence and secure evidence claims. Philosophy of Science 71: 467–488. Staley, Kent W. 2005. Agency and objectivity in the search for the top quark. In Achinstein 2005, 165–183. Star, Susan Leigh, and James R. Griesemer. 1989. Institutional ecology, “translations” and boundary objects: Amateurs and professionals in Berkeley’s Museum of Vertebrate Zoology. Social Studies of Science 19: 387–420. Suppe, Frederick. 1977. Introduction, Afterword. In The Structure of Scientific Theories, 2nd ed., ed. F. Suppe, 3–232, 617–730. Urbana: University of Illinois Press. Suppe, Frederick. 1998. The structure of a scientific paper. Philosophy of Science 65: 381–405. Taper, Mark L., and Subhash R. Lele, eds. 2004. The Nature of Scientific Evidence. Chicago: University of Chicago Press. Taylor, Charles Alan. 1996. Defining Science. Madison: University of Wisconsin Press. Thompson, John B., and David Held, eds. 1982. Habermas: Critical Debates. Cambridge: MIT Press. Thomson, J. J. 1897. Cathode rays. Philosophical Magazine 44 (October): 293–316. Facsimile reprint in Stephen Wright, Classical Scientific Papers: Physics (New York: American Elsivier; UK: Mills and Boon, 1964). Timmons, Mark. 1999. Morality without Foundations. Oxford: Oxford University Press. Tindale, Christopher W. 1999. Acts of Arguing. Albany: SUNY Press. Tindale, Christopher W. 2004. Rhetorical Argumentation. Thousand Oaks, Calif.: Sage. Toulmin, Stephen E. 1958. The Uses of Argument. Cambridge: Cambridge University Press. Toulmin, Stephen, Richard Rieke, and Allan Janik. 1984. An Introduction to Reasoning, 2nd ed. New York: Macmillan. Turner, Stephen. 1994. The Social Theory of Practices. Chicago: University of Chicago Press. Turner, Stephen. 2001. What is the problem with experts? Social Studies of Science 31: 123–149. Turner, Stephen. 2003a. The third science war. Social Studies of Science 33: 581–611. Turner, Stephen. 2003b. Liberal Democracy 3.0. London: Sage. Vorobej, Mark. 2006. A Theory of Argument. Cambridge: Cambridge University Press. Walton, Douglas N. 1989. Informal Logic. Cambridge: Cambridge University Press.
334 References
Walton, Douglas N. 1992. Plausible Argument in Everyday Conversation. Albany: SUNY Press. Walton, Douglas N. 1995. A Pragmatic Theory of Fallacy. Tuscaloosa: University of Alabama Press. Walton, Douglas N. 1998. The New Dialectic. Toronto: University of Toronto Press. Walton, Douglas N. 2004. Relevance in Argumentation. Mahwah: Erlbaum. Warnke, Georgia. 1999. Legitimate Differences. Berkeley: University of California Press. Weart, Spencer R. 2003. The Discovery of Global Warming. Cambridge, Mass.: Harvard University Press. Webler, Thomas, Seth Tuler, and Rob Krueger. 2001. What is a good public participation process? Five perspectives from the public. Environmental Management 27: 435–450. Weingart, Peter. 1993. Close encounters of the third kind: Science and the context of relevance. In Brante, Fuller, and Lynch 1993, 153–174. Wenzel, Joseph W. 1979. Jürgen Habermas and the dialectical perspective on argumentation. Journal of the American Forensic Association 16: 83–94. Wenzel, Joseph W. 1987. The rhetorical perspective on argument. In Argumentation: Across the Lines of Disciplines, ed. F. H. van Eemeren et al., 101–109. Dordrecht: Foris. Wenzel, Joseph W. 1990. Three perspectives on argument: Rhetoric, dialectic, logic. In Perspectives on Argumentation: Essays in Honor of Wayne Brockriede, ed. R. Trapp and J. Schuetz, 9–26. Prospect Heights, Ill.: Waveland. White, Stephen K. 1989. The Recent Work of Jürgen Habermas. Cambridge: Cambridge University Press. Widmer, Jean. 1986. Wörtliche Bedeutung und reflexiver Sinn. Zeitschrift für Semiotik 8: 63–69. Willard, Charles Arthur. 1996. Liberalism and the Problem of Knowledge. Chicago: University of Chicago Press. Wilson, Bryan R., ed. 1970. Rationality. New York: Harper. Wolters, Gereon. 2003. Carl Gustav Hempel: Pragmatic empiricist. In Logical Empiricism, ed. P. Parrini, W. C. Salmon, and M. H. Salmon, 109–122. Pittsburgh: University of Pittsburgh Press. Woolgar, Steve. 1981. Interests and explanation in the social study of science. Social Studies of Science 11: 365–394. Woolgar, Steve. 1983. Irony in the social study of science. In Knorr-Cetina and Mulkay 1983, 239–266. Woolgar, Steve. 1988. Science: The Very Idea. New York: Tavistock. Wright, Crispin. 1992. Truth and Objectivity. Cambridge, Mass.: Harvard University Press.
335 References
Wylie, Alison. 1996. The constitution of archaeological evidence: Gender politics and science. In The Disunity of Science, ed. P. Galison and D. J. Stump, 311–343. Stanford: Stanford University Press. Wylie, Alison. 1997. The engendering of archaeology: Refiguring feminist science studies. Osiris 12: 80–99. Wynne, Brian. 1989. Sheepfarming after Chernobyl: A case study in communicating scientific information. Environment 31, no. 2 (December): 10–15, 33–39. Zammito, John H. 2004. A Nice Derangement of Epistemes. Chicago: University of Chicago Press. Ziman, John. 1968. Public Knowledge. Cambridge: Cambridge University Press. Zuber, Marta Spranzi. 1998. Dialectic, dialogue, and controversy: The case of Galileo. Science in Context 11: 181–203.
Index
Accountability, 114, 222–227, 249, 262–263, 283. See also Validity claims ethnomethodology and, 218–221 Habermas on, 114–120, 136 Achinstein, Peter, 81–98, 146–148, 201 Action types. See Theory of communicative action Actor network theory (ANT), 71–74, 244, 255 Adequacy dialogical, 154, 185–187, 195 empirical, 80, 104, 146–148, 185–188, 195, 206–208, 249 Analytic-evaluative perspectives. See Perspectivist framework; Dialectical perspective; Logical perspective; Rhetorical perspective; Social-institutional perspective ANT. See Actor network theory (ANT) Argument, 25, 246, 249, 269–272, 287–291 dialectically robust (responsive), 133–138, 145–146, 172–173, 181–183 evidential, 34, 88–90, 93–96, 176, 247 as product (see Product) Argumentation aims of scientific argument-making (representative, justificatory, communicative), 97–98, 146–147, 283 argumentative practices in science, 8, 17, 30, 43, 67, 97–98 and bias (see Bias) and context of discovery (see Context, of discovery; Experimental practice) as cooperative judgment formation, 139–141
as dialectical procedure, 24–29, 30, 133–134 in the laboratory or at the research site, 18–19, 75, 133, 141, 149 (see also Experimental practice) multidimensional nature of, 24, 28–29, 245–246 (see also Perspectivist framework; Cogency, as multidimensional) as process (see Perspectivist framework; Process) as social practice, 9, 24–25, 33, 51, 146–147, 269 substantial, 23, 34, 60–64, 132, 142–143, 267, 272–273 Argumentation studies. See Argumentation theory Argumentation theory, 4–9, 22–25, 30–32, 245–246 contextualist (see Critical contextualism) and Habermas, 129–161, 180–193, 270, 303n13, 303n14 informal logic, 20, 23–24, 106 perspectivism, 25, 30, 266 (see also Perspectivist framework) rhetorical turn, 9, 20–23 Argument-making practices. See Argumentation Aristotle, 20–21, 65–69, 140–142, 151, 224, 254–255. See also Rhetoric, Aristotelian Auxiliary assumptions, 35, 37, 134, 201–202 Barbour, Ian, 290–292 Barnes, Barry, 51, 196–199 Beller, Mara, 273
338 Index
Bias, 69, 112, 125, 149–150, 201, 250, 273–274, 280–284, 288 at CDF, 172–176, 181–182 masculine, 149, 228–232, 274 at NAS, 256, 259–264, 280 Bloor, David, 51–52, 78, 196–199 Boffey, Phillip, 261–267, 279–280, 284. See also Sunshine rule Bohr, Niels, 273 Boundary concepts, 6–9, 21, 30, 208, 214, 297n3 Burke, Kenneth, 20–21, 65 Carnap, Rudolf, 31, 37–38, 41, 82, 86–87, 173 Cases/examples AIDS research, 234–235 behavioral biology, 203–207, 214, 235 Bohr’s quantum theory, 273 CDF Evidence paper, 169–186, 190, 272–273 (see also CDF collaboration) creationist-Darwinist debate, 1–2, 68, 70, 124, 206–207, 214, 290–293 dinosaur extinction controversy, 281–282 DNA structure, 273–274 Galvani-Volta debate, 58–61 Geiger-Müller counter, 166–168 global warming (see Climate change debate) K-T boundary layer, 225 lone genius, 149–151 (see also Cases/ examples, Rous, Peyton; Cases/examples, McClintock, Barbara) McClintock, Barbara, 149–150 NAS panels on diet and health, 242–249, 254–268, 273, 279–280, 285, 288–290 Pasteur, 72–76 Rous, Peyton, 149–151 terminally ill comatose patient, 141–145 Thomson on cathode rays, 82–87, 90–93, 148 Wittgenstein’s counting by twos, 238 women in science, 228–230, 272 Case studies, tasks of, 163–165 CAST. See Council for Agricultural Science and Technology (CAST) Cathode rays, 82–85, 90–91, 148, 165 CDF collaboration, 163–193, 266–267, 272 kinematic analysis, 176–180 SLT group analysis, 173–176 CERN, 168–169 CH. See Habermas’s conception of cogency Climate change debate, 2–3, 189, 191
Cloud chamber, 165–166 Cog(A), 278–281, 293 Cogency, 2–14, 130. See also Perspectivist framework and Achinstein’s theory of evidence, 87–90, 95–97, 130 (see also Cogency, as impersonal merits; Logical perspective) as boundary concept, 6–8 (see also Boundary concepts) and costs of error, 235 and critical contextualism, 250, 266–267, 277–284 (see also Cog(A)) and ethnomethodology, 49–53, 222–224 Habermas’s theory of, 130–131, 138, 144–151, 154, 197–198 as impersonal merits, 34, 41, 55, 88, 97, 130 and Kuhnian paradigms, 42–49, 62 Latour’s rhetorical conception of, 77 (see also Rhetorical perspective) logical theory of (Hempel), 34, 40–43, 55–56, 130 as multidimensional, 8, 25, 129, 146–151, 249–254, 258, 266 Pera’s dialectical conception of, 64 (see also Dialectical perspective) Prelli’s rhetorical conception of (RC), 67 (see also Rhetorical perspective) and process ideals, 136 and Rorty, 156–157 and the sociology of scientific knowledge (SSK), 49–56 usage, 4, 7, 30, 55 Communicative action, 104–105, 111–112. See also Theory of communicative action Concrete audience, 79–80, 137. See also Ideal universal audience Confirmation, 32–41, 55, 81–82, 86, 249, 298n2 Content merits, 132, 176, 251–253, 271–275, 282–283, 287–288, 291 and the reception of DNC, 258–259, 263 Context, 13–14, 249, 253, 293–294 context-transcendence, 106, 115, 214–216, 224–227, 242 dialogical context, 104, 250 of discovery, 18, 33, 42, 45, 48–49 as domain or discipline, 227–230, 241 further evaluative contexts (CE), 278, 281 (see also Cog(A)) further relevant contexts (CR), 278, 281, 293 (see also Cog(A)) of justification, 20, 33, 41–42, 45, 49
339 Index
local, 53, 217–219, 224, 236, 242, 246, 264 as locale, 227–230, 241–242, 251–253, 272, 310n10 material, 18 of origin (CO), 278–281 (see also Cog(A)) public, 246–248, 260–265, 288 rhetorical, 23, 174, 241, 246, 294 scientists as functioning in multiple, 238, 242, 282 social context in argumentative evaluation, 24, 53, 130, 151–156, 237, 271–274 social-psychological, 140, 145, 151–152, 159, 271 Contextualism critical (see Critical contextualism) epistemological, 214, 308n7 ethnomethodological, 53, 214–231 and Habermas, 139–145, 160, 253, 266, 270–274 semantic, 302n2, 302n3 Conventionalism. See Relativism Council for Agricultural Science and Technology (CAST), 259–260, 267–269, 279 Creationism, 1, 68, 70, 124, 206–207, 214, 290–293 Credit, 157–161, 177–179, 184, 191, 217 Critical contextualism, 8, 124, 214–215, 231–239, 265–268, 284–295 and CDF, 272–273 and Habermas’s discourse theory, 270–274 and Kuhn’s Gap, 269–278 prescriptive character of, 277–284 Critical science studies (CSS), 7, 195–209, 230, 275–287, 295 Critical (social) theory, 103, 105, 101–127 and empirical sociology (ESW, SSK), 196–209, 222–223 (see also Ethnomethodology of scientific work; Sociology of scientific knowledge) and formal pragmatics, 109 CSS. See Critical science studies (CSS) Dalitz, Richard, 177–179 Dawkins, Richard, 291–292 Deliberative democracy, 105, 112, 122–123, 180, 236, 280–281, 286–290, 293 Demos, 279–281, 288–289 Dennett, Daniel, 291 Descriptive-normative approach, 53. See also Prescriptive-normative approach
Dialectic, 59, 76, 134 Dialectical level (Habermas’s), 132–138 Dialectical perspective, 24, 27, 30, 59–65. See also Perspectivist framework CDF Evidence paper and, 169–172, 181–184 (see also Cases/examples, CDF Evidence paper) and Habermas, 131–138, 145, 181–184, 304n3 and NAS, 256–258 Dialectical rules, 27–28, 59–61, 134, 182 Dialectical situation, 27, 61–62, 79–80, 144, 169, 182 Dialectic of inquiry, 19 Dialogical level. See also Dialectical perspective; Rhetorical perspective combines rhetorical and dialectical perspectives, 145 internally related to the logical perspective, 146–148 internally related to the social-institutional perspective, 151–157, 216–217 Dialogical merits, 145–148, 151–160, 184–190, 200, 250–251 Diet, Nutrition, and Cancer (DNC), 244–248, 258–268, 285 Discourse theory (Habermas), 8, 101–102, 126–127, 151–157, 270, 292. See also Validity claims discourse types, 113, 123–124 holistic character, 208–209 as implying a multidimensional conception of cogency, 151 Discursive dilemma, 306n16 DNC. See Diet, Nutrition and, Cancer (DNC) Douglas, Heather, 234–235 Dramaturgical analysis of self-presentation, 243–249 Edinburgh School, 51, 196 Eemeren, Frans H., van, 7 Emancipation, social, 105, 122–126, 199, 205–206, 223, 234. See also Good society Error accounts, 115, 118–120, 224 ES-evidence. See Evidence ESW. See Ethnomethodology of scientific work (ESW) Ethnomethodology, 50–53, 71–72, 129, 160, 213–233, 271. See also Ethnomethodology of scientific work central recommendation of, 221 Ethnomethodology of scientific work (ESW), 214–215, 217–226, 230, 237, 241
340 Index
Ethos, 68–69, 131, 140–144, 153, 282, 289 CDF Evidence paper and, 171, 177–179, 183–185 (see also Cases/examples, CDF Evidence paper) NAS and, 254–258 Evidence Achinstein’s theory of, 82–87, 89–92 and argument, 249 ES-, 89–98, 148 potential, 85–96, 148 veridical, 85–96, 148 Evolutionary theory, 1–3, 13, 43, 206, 282, 290–292 Experimental practice, 18–19, 121–122, 133 and argumentation, 18, 121–122 Expertise, 109, 138–144, 235–236, 280, 284, 287–288, 292 at CDF, 177–178, 183–184 at NAS, 244–249, 255, 258 Falsification, 35–37, 42, 115, 168 Feminism, 199, 201–202, 205, 228–234, 272, 280 Fennema, Owen, 259 Fermilab. See CDF collaboration Finitism, 51, 236–238 Formal-logical perspective. See Logical perspective Formal pragmatics, 103, 109–114, 122, 135–136, 157 and ESW, 215–216, 222–224 as theory of meaning, 111, 129 of truth (see Truth, Habermas’s formal pragmatics of) Frankfurt School, 11, 103, 196, 215 Franklin, Rosalind, 273–274 Frisch, Henry, 180, 185–186 Fuller, Steve, 5, 277 Galison, Peter, 165–167 Gaonkar, Dilip Parameshwar, 22, 164 Garfinkel, Harold, 12, 217–222 Gero, Joan, 228–229 Goffman, Erving, 243–244. See also Dramaturgical analysis of self-presentation Goldstein, Gary, 177–179 Good society, 14, 125, 235, 294–295. See also Emancipation, social Goodwin, Jean, 245–246 Gould, Stephen Jay, 291 Grootendorst, Rob, 7
Habermas, Jürgen, 11–13, 27–28. See also Argumentation theory; Critical (social) theory; Truth argumentation theory and compromise, 184–193, 216 process idealizations (see Process idealizations) sociological analysis of truth (see Truth, sociological analysis of) three-dimensional perspectivism, 266 on “world” as opposed to “reality,” 303n8 Habermas’s conception of cogency (CH), 154–155, 187, 197–198 Hales, Steven, 237 Haught, John, 290–292 (HD). See Hypothetical-deductive method (HD) Hempel, Carl, 23, 33–42, 55, 81–83, 87–88, 130–133 Hempel-Oppenheim model, 38 HEP. See High-energy physics (HEP) Hertz, Heinrich. See Cases, Thomson on cathode rays High-energy physics, 165–169, 305n1 Hilgartner, Stephen, 243–249, 254–262 Holism, 8, 12, 22, 44, 208–209, 214, 253, 270 Horkheimer, Max, 103 Hoyningen-Huene, Paul, 46 Hull, David, 158 Hypothetical-deductive method (HD), 35–36, 83–84 Ideal dialogical conditions, 155, 182, 191–192, 227–230 Idealizations, 27–28, 114, 131. See also Ideal speech situation; Ideas (of reason) as counterfactual, 114, 136–137, 157, 192–198, 201, 208, 217, 227, 271 dialogical, 155–161, 270–272 process (Habermas’s) (see Process idealizations) Ideal of an intrinsically reasonable dialogue, 197–198 Ideal of cogent justification, 160 Ideal of equal voice, 13, 135, 152–156, 182, 227–230, 260–261, 272 Ideal of impartiality, 59, 244, 248–250, 256, 258, 260–261 Ideal of inclusiveness, 152, 156–159, 175–176, 182, 227–230, 257, 260–263, 272 Ideal of uncoerced discourse, 13, 27–28, 153, 176, 182–186, 191–192, 229, 257, 260–261
341 Index
Ideal speech situation, 27–28, 135 Ideal universal audience, 62, 116, 124, 135–137, 229. See also Concrete audience Ideal universal consensus, 116–118, 123–126 Ideas (of reason), 27–28, 114, 223–227, 231, 274 objectivity, 114–117, 224–227, 290–292 truth (see Truth, Habermas’s formal pragmatics of) Incommensurability, 44, 57, 66–67, 79, 224, 237 Indexicality, 219–225, 230–232, 271, 274, 283, 291–292 Indifference, ethnomethodological, 13, 53, 215, 222, 230, 275 Induction, 34–43, 84. See also Inductive risk Inductive risk, 234–235, 284–286 Inquiry. See Dialect of inquiry Institutionalization, 68, 124–126, 134, 160, 182, 134, 154 credit as an institutional convention, 158 Intelligent design, 1, 124, 206, 234, 291 Interdisciplinarity, 4–14, 30–31 (see also Boundary concepts) in argumentation studies, 4–5, 297n5 between critical theory and SSK, 196–209 and cogency, 54 and critical contextualism, 213–214, 243, 269–278, 281–284 and Habermas, 101–103, 184–190 and perspectivism, 30–31 and the rhetorical turn, 21 between different scientific disciplines, 281–282, 284–289 and tension between critical and social approaches, 50, 192 International Panel on Climate Change, 2, 189 Intersubjectivity, 40–42, 144, 195, 225 and Achinstein, 90, 94 and Habermas, 105, 115–117, 134–135 Jasanoff, Sheila, 279–281, 284–289 Johnson, Ralph, 26 Journalism, science, 187–190, 231, 263, 272, 282 JTC, 117–119, 136 Judgment, 139–146 and Habermas, 139–146 among Kuhnian paradigms, 45–49 responsible, 140–146, 152–154, 183–186, 250
Justification. See also Ideal universal consensus and evidence, 88–90, 92–97 and truth, 95, 116–118, 147 and warrant, 88, 155–156 Justificational atheism, 196–200 Kitcher, Philip, 141, 143, 158 Klein, Julie Thompson, 6, 8, 21, 208–209, 270 Klein, Wolfgang, 109–110 Kuhn, Thomas, 33, 42–56, 139. See also Kuhn’s Gap Kuhn’s Gap, 33, 42–49, 53–54, 79–80 and critical contextualism, 269–277 and Habermas, 129–131, 145, 151–156 and Latour, 76–78 and Pera, 62–64 Kusch, Martin, 236–238 Latour, Bruno, 71–80, 130, 137, 169, 186, 255. See also Actor network theory on allies, 72–77 antinormative approach to cogency, 72, 77 Linguistic turn, 103, 127 Liss, Tony, 171–172 Livingston, Eric, 12, 220–221 Logic, 20–29, 34, 146, 166, 249 and the evaluation of products, 132 formal and informal, 23–24 of justification, 18, 41 of relativism, 237 Logical empiricism, 10, 29, 33–42, 45–46, 94, 190 Logical perspective, 24–26, 30, 33–42, 49, 55, 93. See also Perspectivist framework and Achinstein, 81–82, 92–93 CDF Evidence paper and, 170–171, 175, 182 (see also Cases/examples, CDF Evidence paper) and content merits, 33–34, 131–133 and dialectical level, 133 Kuhn and, 49 limits of, 98 Pera and, 64 Longino, Helen, 150–152, 203–206, 232–235, 304n3 Lynch, Michael, 12, 222–226, 231–233 MacIntyre, Alasdair, 237 Mayo, Deborah, 133, 172–173 McCarthy, Thomas, 114–116, 118, 217, 224–225
342 Index
Mertonian model of science, 49, 68, 70, 73, 198 Method of avoidance, 199–200, 270 Narrative, 244 NAS. See National Academy of Sciences (NAS) National Academy of Sciences (NAS), 13, 242–249, 254–268, 273, 279–280 No-peeking rule, 173–176 Normal science, 43–44 Objective world (objectivity). See Ideas (of reason), objectivity Observation statement, 26, 36–42, 84, 172–173 Olbrechts-Tyteca, Lucie, 23, 43, 58–59, 137 Oppenheim, Paul, 38–39 Paradigms, 43–49, 54, 62–69, 228, 236, 250, 266 Paradox of the ravens, 36–38, 298n2, 299n7 Parsonian sociology, 217–218 Particularism, 12, 50–54, 79, 215, 224. See also Ethnomethodology Pathos, 68–69, 141–144, 153, 171, 178, 183–185 Pera, Marcello, 44, 58–65, 69–71, 78–81, 134, 137 duck-rabbit interpretation of Kuhn, 44–45, 58 on science as a tradition of actual debate, 60–63 Perelman, Chaim, 23, 43, 58, 137 Perspectivist framework, 10–12, 24–25, 30–31, 105, 266 alignment of triads, 24–29, 105, 266 internal relations, 132–138, 144–146, 208 Plantinga, Alvin, 155 Policymaking and science, 5, 13, 165, 246, 269, 278–280, 293–295 and civic epistemologies, 286 and Habermas, 122, 138–139 political aspects of, 285–286 and risk, 189, 232–236, 284–290 Popper, Karl, 42, 54, 59–60 Practical engagement, 120, 144 Practice turn (in science studies), 9, 48 Pragmatic context, 40–41 Pragmatic merits, 42 Pragmatic turn, 9–10, 103, 105, 127 Predesignation rule. See No-peeking rule Prediction criterion, 35, 298n1
Presuppositions dialectical, 134–135 epistemological, 9 level of, 132, 152, 158, 164 normative, 131–132 pragmatic, 13, 27–28, 109, 113–117, 136, 159, 214, 223, 230, 266 social-institutional, 152 Prelli, Lawrence, 65–71, 78–81, 140–141, 274 rhetorical standard of cogency (RC), 67–70 understanding of reasonable, 67 Prescriptive-normative approach, 53. See also Descriptive-normative approach Problem of induction, 34. See also Induction Procedure, 10, 24–30, 231, 236 as associated with dialectical level by Habermas, 137 credit as an institutional, 159 in critical contextualism, 248–249 as dimension of argumentation, 133–134, 249 and ESW, 218–222 at Fermilab, 169–176, 180–187 (see also Cases/examples, CDF Evidence paper; CDF collaboration) and Habermas, 105–108, 125, 133–138, 159–160 at NAS, 242–249, 254–268, 279 procedural factors (Pera), 60–63, 137 statistical arguments as inherently procedural, 174–176 (see also CDF collaboration) Process, 9–13, 81–82, 98 Achinstein and, 87–88, 92–98 and cogency, 270–271 and critical contextualism, 270–271, 279–284 at Fermilab, 169–191 (see also Cases/ examples, CDF Evidence paper; CDF collaboration) Habermas on, 105–108, 121–122, 130–131, 135–142, 145–161, 180–191 Kuhn on, 43–49 Latour and, 71–77 at NAS, 245–246, 254–265, 279–281 Pera and, 60–63 and the perspectivist alignment, 24–30 Prelli and, 69–71 as rhetorical, 24–29, 30, 69 as social, 5, 10–12, 19–20, 49, 135–138, 150–154, 245 Toulmin on, 106–107
343 Index
Process idealizations, 135–137, 140, 153, 159–160, 181–182, 186 in tension with contextualization, 215 and ethnomethodology, 227–230 Product, 24–30, 246, 245–251 Achinstein and, 81–82, 88–92, 97–98 conditions for dialectically testing, 135–138 at Fermilab, 170–175, 179–185, 187 (see also Cases/examples, CDF Evidence paper; CDF collaboration) Habermas and, 105–106, 130, 133–135, 145, 147–148 Hempel and, 40–41 Latour and, 72 at NAS, 245–251, 262, 266 Public merits, 13, 251–254, 258–268, 277–278, 287–288, 293–294 and CDF, 272–273 Public sphere (networks and arenas), 246–251 Rational agency. See Accountability Rationality debates, 2, 200, 269, 297n1. See also Science wars Rawls, John, 199–200, 270 RC. See Cogency, Prelli’s rhetorical conception of Realism, 11, 31 in Achinstein, 82, 86, 148 in Habermas, 119, 148 Recommended daily allowances (RDAs), 244, 248 Reflexivity (in ESW), 220–221, 225 Relativism, 12–13, 80 and contextualism, 126–127, 224, 275 critical social theory and, 236–238 and ESW, 50, 213–215, 224 and Habermas, 106–109, 126–130, 160 Kusch’s, 236–238 and Latour, 78–79 across paradigms, 44 and Pera, 59–63 in Prelli’s rhetorical perspective, 70–71 in SSK, 50, 102, 106–109, 129, 160 196–199, 270, 274–275 Toulmin’s, 106–108, 126–127 Relevance, 3–6, 201–205. See also Context and Achinstein, 83–84, 87, 91–92 broad material, 291–292 and confirmation, 35–39 and contextualism, 231–237, 257–258, 271–294 cross-contextual, 268, 281–283
as a dialectical obligation, 134 and ESW, 224–230 and Habermas, 133–136, 139–145 as an intrinsic formal merit of arguments, 41–42 and NAS, 258–268 and process idealizations, 136 and SSK, 201–203 Research proposal, 19, 125, 139 Revolutionary science, 43–46, 54, 283 Rhetoric, 20–25 Aristotelian, 20, 65–68, 141, 151, 254–255 (see also Aristotle) communitarian, 10, 65, 70, 80, 130 epistemic conception of, 141 and ESW, 227–230 ethos (see Ethos) at Fermilab,169–171, 174, 177–180, 183–186 (see also Cases/examples, CDF Evidence paper) in Habermas, 130–132, 135–145, 151–154 in Klein (Wolfgang), 109–110 and Kuhn, 54–57 in Latour, 71–79, 169 logos, 137, 141–143 at NAS, 241–250, 257–258, 262–264, 271–272 pathos (see Pathos) in Pera, 58–65 and perspectivist alignment, 26–31 in Prelli, 65–71 proof from character, 142 Rhetorical criticism, 21, 70–72, 136 Rhetorical perspective, 22–26, 30, 59. See also Perspectivist framework; Pathos; Ethos CDF Evidence paper and, 170–172, 177–179, 184 (see also Cases/examples, CDF Evidence paper) and critical contextualism, 241 of Habermas, 105, 132, 137–138, 141, 145 of Latour, 71–80 of Prelli, 65–71 Rhetorical situation, 174, 241, 246, 294 Rhetoric of science, 5, 20–22, 29, 57–58, 70–72, 79, 138–145, 164 Rorty, Richard, 59–62, 156, 199, 215 Rouse, Joseph, 44 Satisfaction criterion (Hempel’s), 35–38, 299n7 Science and technology studies (STS), 4–5, 262, 264, 276–277, 281
344 Index
Science-as-knowledge, 18 Science-as-practice, 18 Science studies, 6–11, 28–32, 50, 54, 234–236, 269–270, 274–276. See also Critical science studies ANT as a method of, 73 and ESW, 213, 215, 230, 234–236 Habermas and, 101, 129, 151 Science wars, 1–3, 7, 41, 269–270, 297n1 Scientific community according to SSK and ESW, 49–53 and Achinstein, 90 and critical science studies, 202, 258–260 and Habermas, 112–114 and Kuhn, 41, 46–50 and Pera, 59, 62–65 and Prelli, 67, 70 Scientific discourse, 22, 98, 187, 216 and Habermas, 112, 121, 124–125, 155, 157, 164 and Pera, 61–63, 70 and Prelli, 67, 70 Scientific inquiry, 17–20, 30–31, 78–79, 95–96, 199–201, 214, 269 and credit, 157 and Habermas, 113, 129, 148, 154–155 Scientific method, the, 3, 9, 50, 218, 232–233, 275 Skepticism Latour and, 72, 75 Pera and, 63 and the perspectivist framework, 30–31 Prelli and, 68, 70 of rational justification, 196–200 (see also Justificational atheism) rule skepticism, 51–53 and the sociological response to Kuhn’s Gap, 50–53 SSK and, 196–199, 275 Sliwa, Krys, 177–179 Snoeck Henkemans, Francisca, 7 Social-institutional perspective, 29–30, 33, 191, 213, 216 as distinct from the dialectical and rhetorical levels, 152–153 at Fermilab, 170–172, 182–183, 191 (see also Cases/examples, CDF Evidence paper) Habermas and, 131–134, 151–157, 159 Latour and, 80 as replacing the logical perspective for Kuhn, 49, 55 NAS and, 252 and SSK, 53–57
Sociology of error, 118, 156, 192, 203. See also Sociology of scientific knowledge Sociology of scientific knowledge (SSK), 49–56, 153 and behavioral biology, 203–207 and CSS, 275–276 as including ethnographic approaches and ESW, 300n15 and Habermas, 102, 129, 153–156, 195–209, 270, 274–276, 307n3 Latour and, 73–74, 78–79 as analyzing social-institutional dimension, 200–201 as rejecting the sociology of error, 118, 156, 192 Solomon, Miriam, 138–139, 203, 282 SSK. See Sociology of scientific knowledge (SSK) Staley, Kent, 165, 169, 174, 176 Standard model (particle physics), 166, 169, 173, 178, 305n1 Stases, 65–70, 138, 140 Strategic action, 104, 108, 125–126, 130–131, 158–159. See also Theory of Communication action Strong Program in the Sociology of Scientific Knowledge, 50–53, 79, 196, 236–237 STS. See Science and technology studies Sunshine rule, 261–265 Symmetry postulate. See Symmetry principle Symmetry principle, 50, 73–74, 192–193, 198. See also Sociology of scientific knowledge (SSK) TCA. See Theory of communicative action (TCA) Teilhard de Chardin, Pierre, 292 Theoretical articulation, 7–8 Theory choice and social influences, 49, 51, 59, 203 Theory of communicative action (TCA), 104–111, 122–127 and the meaning of speech acts, 111–113 Thomism, 291 Thomson, J. J. See Cases/examples, Thomson on cathode rays Tipton, Paul, 170–172. See also Cases/ examples, CDF Evidence paper Tollestrup, Alvin, 177–179 Toulmin, Stephen, 23, 43, 71, 126, 132, 157, 215, 236 model of argument, 106–108 Toward Healthful Diets, 244–248, 259–263, 284
345 Index
Transactional merits, 251–253, 258–261, 265–268, 273, 282–283 Transactions, 246–254, 257–268, 271–274, 277–278, 282–288, 293 Transcendence context (see Context, context-transcendence) locality, 225–226 Truth, 119–120, 303n11 and Achinstein, 82, 88–98 and contextualism, 283–285, 287–288, 292 and ESW, 224–227 in Habermas’s argumentation theory, 133–136, 146–151 Habermas’s formal pragmatics of, 111–122, 147, 225 Habermas’s philosophical account of, 119–122, 303n11 Habermas’s sociological analysis of, 114–119 Hempel and, 298n3 as an idea of reason, 114 as impersonal, 87–90, 97–98, 147, 151 as justification-transcendent, 116, 118, 126, 147–148, 242 Kusch’s truth finitism, 236–238 and the idea of a common objective world, 115 Pera and, 63–64, 300n6 and universal consensus, 116–118 as a validity claim, 108, 111–113, 136 Universal pragmatics. See Formal pragmatics Validity claims, 104–116, 122–126, 132–134, 136, 215, 241. See also Discourse theory (Habermas) in relation to accountability, 114 as multidimensional, 129 morality, 126 rightness, 111–113 sincerity, 111–112 truth (see Truth) validity basis of speech acts, 111–113 Values, 201–207. See also Inductive risk epistemic, 46–49, 60, 66, 141–142, 202, 232–235, 284–285 and Kuhnian incommensurability, 42–49 as defined by Pera, 62 as employed by Prelli, 66–67 Watson, James, 274 Wenzel, Joseph, 24–27, 30, 132, 266, 298
Whitehead, Alfred North, 292 Wilkins, Maurice, 274 Wilson, C. T. R., 165 Wittgenstein, Ludwig, 50–52, 237–238 Woolgar, Steve, 58, 71–72 Wylie, Alison, 228–229 Zammito, John, 44