,

Critical Notice/Etude critique

A Role for Reason in Science*

JONATHAN Y. TSOU

University of Chicago

1. Introduction

Michael Friedman's Dynamics of Reason is a welcome contribution to the ongoing articulation of philosophical perspectives for understanding the sciences in the context of post-positivist philosophy of science. Two perspectives that have gained advocacy since the demise of the "received view" are Quinean naturalism and Kuhnian relativism. In his 1999 Stanford lectures, Friedman articulates and defends a neo-Kantian perspective for philosophy of science that opposes both of these perspectives. His proffered neo-Kantian perspective is presented within the context of the problem of theory change or "scientific revolutions," and its main feature is a conception of scientific knowledge that requires "relativized constitutive a priori principles." The lectures make up the first part of the book; the second part of the book, "Fruits of Discussion," is a further elaboration and defence of the ideas advanced in the lectures. The resulting book serves as a useful sequel to Friedman's impressive historical studies in Foun'" Michael Friedman, Dynamics of Reason: The 1999 Kant Lectures at Stwiford University (Stanford, CA: CSLI Publications, 2001), xiv + 141 pp. Page references in the text refer to this book unless otherwise noted. Dialogue XLII (2003), 573-98 © 2003 Canadian Philosophical Association/Association canadienne de philosophie

574

Dialogue

dations of Space-Time Theories (1983), Kant and the Exact Sciences (1992), and Reconsidering Logical Positivism (1999). In the preface, Friedman tells us that this book represents the philosophical viewpoint that he has arrived at as a result of completing these works (pp. xi-xii). As such, it is not surprising that the prominent themes of the book are ones that have occupied Friedman's attention in the past, viz., the importance of a priori principles in the exact sciences (Part 1, §3; Part 2, §§1-2), the conventionalism of the logical positivists (Part 1, §2; Part 2, §4), and, more generally, an articulation of what remains defensible in neo-Kantian philosophy of science. The aim of this article is to explicate Friedman's proposed perspective and to evaluate his arguments against Quinean naturalism and Kuhnian relativism, respectively. I begin by examining the positivist considerations that motivate Friedman's neo-Kantian perspective. In particular, I examine Reichenbach's pre-conventionalist views on physics and Carnap's philosophy of linguistic frameworks focusing on the relativized a priori articulated in these works. I subsequently examine Friedman's assimilation of the positivists' conception of the relativized a priori with Kuhn's theory of scientific revolutions, an assimilation that Friedman employs to articulate his own conception of "dynamical" or "relativized" a priori principles in scientific knowledge. Finally, I explicate and evaluate Friedman's arguments against Quinean naturalism and Kuhnian relativism. 2. A Positivist Background The philosophical perspective defended by Friedman draws upon positivist philosophy of science as much as it does from neo-Kantian philosophy of science. Unlike many contemporary philosophers of science, Friedman is largely sympathetic with mature ideas of positivist thought (see Friedman 1997, 1999). In particular, he draws upon ideas forwarded in Hans Reichenbach's neo-Kantian philosophy of physics and Rudolf Carnap's syntactic-conventionalist view of scientific theories or "philosophy of linguistic frameworks." The significance of these ideas, for Friedman, lies in the positivists' articulation of a relativized a priori, a central feature of his own neo-Kantian perspective.! 2.1. Reichenbach's Revision of Kant: The Relativized A Priori Within the positivist tradition, the idea of a relativized a priori finds its origins in Reichenbach's first major work, The Theory of Relativity and A Priori Knowledge (originally published in 1920). In this work, Reichenbach considers the implications of Einstein's relativity theory for Kant's philosophy of science with the aim of modifying the latter to be consistent with the former. Central to Reichenbach's modification is a rejection of Kant's synthetic a priori in favour of a relativized a priori. Reichenbach (1965) argues that Kant's a priori judgements carried two related senses:

Reason in Science 575

(1) necessary and valid for all times, and (2) "constituting the concept of the object" (p. 48). Reichenbach's revision of Kant's a priori consists in, first, unlinking these two senses and, second, rejecting the necessary aspect while retaining the "constituting" or "constitutive" aspect. For Reichenbach, the constitutive aspect of the a priori represents the conceptual framework that is required before meaningfully objective (i.e., intersubjective) empirical judgements are possible in physics. Reichenbach writes: "Indeed, there cannot be a single physical judgment that goes beyond a state of immediate perception unless certain principles about the description of the object in terms of a space-time manifold and its func7 tional connection with other objects are made" (1965, p. 77). What Reichenbach finds defensible in Kant's critical philosophy is the idea "that the object of knowledge is not simply given, but constructed, and that it contains conceptual elements not contained in pure perception" (ibid., p.49). Accordingly, all objects of scientific knowledge are constructed, and these constructions require a "conceptual schema" to create the object. Reichenbach writes: "the conceptual schema, the category, creates the object; the object of science is therefore ... a reference structure based on intuition and constituted by the categories" (ibid.; emphasis added). Reichenbach's resulting picture of scientific knowledge includes two distinct components. First, there are formal components contributed by reason that Reichenbach calls "axioms of coordination" or "coordinating principles." Second, there are empirical components supplied by the sensible world called "axioms of connection." Axioms of coordination are constitutive a priori prip.ciples that must be presupposed in order to frame objectively meaningful empirical laws (i.e., axioms of connection). For Reichenbach, the lesson of Einstein's theory of relativity is that theories of physics, from Newtonian mechanics to special relativity to general relativity, require axioms of coordination (viz., laws of physical geometry) as necessary presuppositions to frame axioms of connection; however, these coordinating principles change in the transition from one theory to another. Reichenbach emphasizes that in the context of any given scientific theory there is a sharp and fundamental distinction between principles of coordination and principles of connection: [T]he principles of coordination are determined by the nature of reason; experience merely selects from among all possible principles. It is only denied that the rational component in knowledge remains independent of experience. The principles of coordination represent the rational components of empirical science at a given stage. This is their fundamental significance, and this is the criterion that distinguishes them from a particular law, even the most general one. A particular law represents the application of those conceptual methods laid down in a principle of coordination. (1965, p. 87; emphasis added)

576

Dialogue

On Reichenbach's view, axioms of coordination serve as the background framework for axioms of connection; however, axioms of coordination are revisable and evolve with the development of empirical science. These coordinating principles are constitutive a priori insofar as they secure both the objective meaning and intersubjective communicability of empirical judgements; however, such principles are "relativized" insofar as they are only a priori relative to a particular theory in some historical context. In this sense, science is possible because of constitutive a priori principles. It is important to note that Reichenbach initially conceives of axioms of coordination as relativized synthetic a priori jUdgements. This is a point of ambiguity in Reichenbach's book that becomes clarified in a subsequent correspondence between Schlick and Reichenbach. 2 What is at issue in this correspondence is the precise nature of axioms of coordination. Whereas Reichenbach conceives of his relativized a priori principles as synthetic, Schlick finds Reichenbach's axioms of coordination to be indistinguishable from Poincare's conventions (i.e., analytic a priori principles). As such, Schlick urges Reichenbach to adopt the terminology of Poincare and the theory of conventions (also see Schlick 1979). Schlick's suggestions to Reichenbach were evidently influential since in all subsequent works Reichenbach drops his Kantian terminology of aprioricity in favour of Poincarean conventionalism (e.g., see Reichenbach 1958, 1969). 2.2. Carnap's Philosophy of Linguistic Frameworks and the Relativized A

Priori Friedman maintains that the most-developed articulation of the positivists' formal notion of the relativized a priori appears in Carnap's philosophy of linguistic frameworks in works such as The Logical Syntax of Language (originally published in 1934) and "Empiricism, Semantics, and Ontology" (originally published in 1950). In these works, Carnap provides a poignant statement of his conventionalism and epistemological relativism, arguing that all standards of "correctness," "validity," and "truth" are relative to the choice of a formal language or linguistic framework (i.e., such standards are relative to the logical rules of a particular linguistic framework). As expressed in his "Principle of Tolerance," Carnap (1937, §17) maintains that the choice of a formal language is a conventional choice that can only be decided on practical (as opposed to theoretical) grounds. Friedman maintains that within this conventionalist picture, Carnap provides the most mature articulation of the relativized a priori that occupied Reichenbach's pre-conventionalist ideas (pp. 31-32, 72). In "Empiricism, Semantics, and Ontology," Carnap argues that an assertion is meaningful (i.e., meaningfully answerable as true or false) if it is asserted within the given rules of some conventionally stipulated linguistic framework. Carnap calls these sorts of questions or assertions

Reason in Science

577

"internal questions." Internal questions are meaningfully answerable insofar as they are capable of being true or false given the rules of some linguistic framework. The truth or falsity of internal questions, however, is not "absolute" but relative to some linguistic framework. Questions or assertions made outside a linguistic framework (e.g., questions about the existence of entities assumed in a linguistic framework) are not meaningful, i.e., not meaningfully capable of being true or false. Carnap calls these sorts of questions "external questions.,,3 Carnap diagnoses the status of external questions or assertions as pseudo-statements because such assertions really pertain to the choice of adopting some linguistic framework. As such, external assertions are practical suggestions to adopt some linguistic framework or another. Carnap states that confusion among philosophers results from tendencies to understand external questions or assertions metaphysically: [T]he introduction of a new [linguistic framework] does not need any theoretical justification because it does not imply any assertion of reality ... it must not be interpreted as referring to an assumption, belief, or assertion of the "reality of entities." ... An alleged statement of the reality of the system of entities is a pseudo-statement without any cognitive content. ... [W]e have to face at ~his point, an important question; but it is a practical, not a theoretical question; it is the question of whether or not to accept the new linguistic forms. The acceptance cannot be judged as being true or false because it is not an assertion. It can only be judged as being more or less expedient, fruitful, conducive to the aim for which the langllage is intended. (1983, p. 250)

Here, Carnap emphasizes that the choice of adopting a linguistic framework is a conventional one. The decision involved in adopting a linguistic framework (or the answer to external questions) must appeal to practical rather than theoretical grounds. 4 Friedman argues that this picture defended by Carnap provides a further articulation of the relativized a priori (conceived as a re1ativized analytic a priori) initially forwarded by Reichenbach. Linguistic frameworks are constitutive frameworks because in these formal languages, analytic theorems provide the rules that allow for the subsequent formulation of objectively meaningful empirical laws. This idea is implicit in Carnap's distinction, in The Logical Syntax of Language, between logical rules ("Lrules") and physical rules ("P-rules"). L-rules are the formal or analytic statements of a linguistic framework, including rules of logic, mathematics, and geometry. P-rules, on the other hand, are empirical or synthetic statements of a linguistic framework whose objective meaningfulness is provided by the L-rules of that linguistic framework. In this sense, Carnap's L-rules are constitutive a priori relative to a particular linguistic framework analogous to the way Reichenbach's axioms of coordination

578

Dialogue

are constitutive a priori relative to a particular scientific theory. The distinction between L-rules and P-rules is also related to Carnap's aforementioned distinction between internal and external questions. Internal questions, which are asked within the rules of an already adopted linguistic framework, are answered on the basis of the L-rules of that linguistic framework. In contrast, external questions, which are asked outside all linguistic frameworks and concern the question of which linguistic framework or set of L-rules to choose in the first place, can only be answered conventionally, i.e., on the basis of practical considerations. In forwarding his own neo-Kantian perspective, Friedman draws strongly from Carnap's philosophy of linguistic frameworks and especially the positivists' formal conception of a relativized constitutive a priori. Inasmuch as Friedman endorses these ideas of mature positivist thought, Friedman affirms his own status as the closest approximation to a neo-positivist in contemporary philosophy of science. 3. Relativized A Priori Principles and Scientific Revolutions

The central feature of Friedman's neo-Kantian perspective is its emphasis on the importance of relativized constitutive a priori principles in scientific knowledge. Friedman articulates his own version of such principles in an interesting assimilation of Carnap's philosophy of linguistic frameworks with Kuhn's "paradigms." In the preface of his book, Friedman describes his project as an attempt to combine basic aspects of Carnap's philosophy of formal languages or linguistic frameworks with fundamental features of Thomas Kuhn's much less formal theory of scientific revolutions ... [in order to] articulate a conception of dynamical or relativized a priori principles within a historical account of the conceptual evolution of the sciences rather than a purely syntactic or semantic account of formallanguage(s) of the sciences. (p. xii)

Friedman aims to motivate his neo-Kantian perspective with a historical argument for a view of scientific knowledge that includes relativized a priori principles. The principle support for this perspective, he argues, stems from the historical fact that the philosophical problem of coordination, i.e., the problem of coordinating abstract mathematical structures to concrete physical phenomena, has been absolutely fundamental in formulating theories of physics (pp. 33-42, 71-82). 3.1. Kuhn's Paradigms and the Relativized A Priori In a brief yet provocative discussion, Friedman argues that Thomas

Kuhn's theory of scientific revolutions provides an informal complement to the positivists' formal conception of the relativized a priori (pp. 41-46; cf. Reisch 1991). In particular, Friedman suggests an affinity between Car-

Reason in Science

579

nap's notion of constitutive linguistic frameworks and Kuhn's notion of paradigms (in the sense of "disciplinary matrices" or "lexicons"). Consider the similarities between Kuhn's distinction between "normal science" and "revolutionary science" and Carnap's distinction between rulegoverned operations within an already adopted linguistic framework and the change of a linguistic framework (pp. 41-42). Kuhn's view of normal science, in which a paradigm prescribes generally agreed upon, often tacit, rules that are constitutive of what counts as a genuinely. "correct" or "valid" solution to a scientific puzzle is similar to Carnap's view that the logical rules of an adopted linguistic framework are constitutive of notions of "correctness" or "validity" relative to that framework. Similarly, Kuhn's view that paradigm changes, in revolutionary science, do not proceed on the basis of agreed-upon rules or algorithms is amenable to Carnap's suggestion that the choice of adopting a linguistic framework is not governed by logical rules, but must be decided on practical or otherwise conventional grounds. In support of this "positivist reading" of Kuhn, Friedman appeals to Kuhn's (1993) own acknowledgment of the affinities between his own views on lexicons (Kuhn's later term for "paradigms" or "disciplinary matrices") and the more formal conception of the relativized a priori articulated by Reichenbach and Carnap: Though it is a more articulated source of constitutive categories, my structured lexicon ... resembles Kant's a priori when the latter is taken in its second, relativized sense. Both are constitutive of possible experience of the world, but neither dictates what that experience must be. Rather, they are constitutive of the infinite range of possible experiences that might conceivably occur in the actual world to which they give access. Which of these conceivable experiences occurs in the actual world is something that must be learned, both from everyday experience and from the more systematic and refined experience that characterizes scientific practice.... The fact that experience within another form of life [i.e., another lexiconJ-another time, place, or culture-might have constituted knowledge differently is irrelevant to its status as knowledge. (Kuhn 1993, pp. 331-32, cited by Friedman on p. 43; emphasis in the original)

Besides Kuhn's explicit endorsement of the constitutive function of paradigms,5 the last sentence in this passage suggests that Kuhn is also comfortable with a form of epistemological relativism implied by Carnap's philosophy of linguistic frameworks, i.e., the view that all knowledge is relative to a choice of a constitutive framework. Friedman takes these similarities between Kuhn's view of paradigms and Carnap's view of linguistic frameworks to provide strong support for the idea that the problem of coordinating mathematical principles with empirical principles is a problem basic to science and formulating scientific knowledge. Friedman claims that Kuhn's recognition of the significance of

580 Dialogue constitutive frameworks suggests that "our best current historiography of science [viz., Kuhn] requires us to draw a fundamental distinction between constitutive principles, on the one side, and properly empirical laws formulated against the background of such principles, on the other" (p. 43; emphasis added). While this statement of Friedman does not seem warranted, or at least it places a great deal of authority on Kuhn, Friedman does give credence to this distinction in his analysis of the history of physical theories. 3.2. Friedman's Argument for Neo-Kantian Philosophy of Science Friedman argues that contemporary philosophy of science has failed to recognize the existence of constitutive relativized principles in scientific theories. In the context of the historical development of physical theories, Friedman illustrates the role that constitutive a priori principles play with reference to the development of Newtonian mechanics to general relativity, which involve the coordination of non-empirical (constitutive a priori) and empirical principles (pp. 35-39, 71-82). In Newtonian mechanics, Euclidean spatial geometry functions as the mathematical part of the theory that makes possible empirical principles such as the law of gravitation. The mathematical part, according to Friedman, is constitutively a priori because this part of the theory raises the very possibility of the empirical part by coordinating mathematical structures to experience. In the transition from Newtonian mechanics to special relativity, the background space-time structure (i.e., the constitutive a priori part) is revised in light of empirical findings. Accordingly, there is a shift in both the nonempirical and empirical parts of physical theory. In special relativity, Euclidean spatial geometry, Lorentzian kinematics, and the structure of Minowski space-time function as constitutive a priori principles relative to the theory insofar as they are presupposed in empirical theories of fields and forces such as Maxwell's equations. In the transition from special to general relativity, there is another revision in the background space-time framework. In the theory of general relativity, only the infinitesimally Lorentzian manifold structure (i.e., the space-time topology that is sufficient to admit some Riemannian metric) is constitutive a priori relative to the theory. The empirical theory realized by such constitutive principles is the particular Riemannian metrical structure determined empirically from the distribution of mass energy.6 Friedman attributes the neglect of the phenomenon of constitutive a priori principles in scientific theories, in contemporary philosophy of science, to Quine's apparent victory in the Carnap-Quine debate (pp. 40-41). In particular, Friedman argues that contemporary philosophers' acceptance of Quine's attack on the analytic-synthetic distinction has led them to blindly neglect the problematic that Carnap was attempting to address. Although Friedman states that he has no desire to defend Carnap's par-

Reason in Science 581

ticular way of drawing the analytic-synthetic distinction (p. 33), he does aim to defend the problematic that occupied Carnap's mature writings. 7 Friedman writes: [qareful attention to the actual historical development of science, and, in particular, to the profound conceptual revolutions that have in fact led to our current philosophical predicament, shows that relativized a priori principles of just the kind Carnap was aiming at are central to our scientific theories. Although Camap may have failed in giving a precise logical characterization or explication of such principles, it does not follow that the phenomenon he was attempting to characterize does not exist. On the contrary, everything that we know about the history of science ... indicates that precisely this phenomenon is an absolutely fundamental feature of science as we know it. (p. 41; emphasis in original)

This passage encapsulates the main thesis of Friedman's book, viz., that the problem of coordination is fundamental to formulating scientific knowledge. Two critical points are worth raising here. First, although Friedman's claim regarding the need for constitutive a priori principles in science is made with reference to science in general, his own analysis is limited to scientific knowledge in the domain of mathematical physics. This limitation becomes painfully apparent in the final chapter of the book (Part 2, §5) when Friedman outlines possible directions of his proposed perspective to domains such as chemistry and biology, and the discussion has very little to do with either chemistry or biology (pp. 124-29). Second, although Friedman is not obliged to defend Carnap's manner of drawing the analytic-synthetic distinction, he does seem to be obliged to provide an independent argument for drawing it since his perspective presupposes the distinction (cf. n. 7).8 Otherwise, it could be objected that Friedman's arguments fail to motivate his perspective. Despite such objections (whi::h demonstrate limitations rather than a fundamental untenability), Friedman's historical analysis does provide a strong basis for his neo-Kantian perspective. Friedman's analysis suggests a crucial functional role for constitutive a priori principles in formulating physical theories and knowledge. This neo-Kantian understanding of physical theory forms the basis of Friedman's rejection of philosophical naturalism and sociological relativism associated with Kuhn. In what follows, I explicate and evaluate the merits of Friedman's arguments against Quine and Kuhn respectively. 4. Against Quinean Naturalism Much of Friedman's book is devoted to explaining what is wrong with Quinean naturalism as a philosophical perspective for the sciences (pp. 28-46, 78-82). He argues that the anti-apriorist conception of scien-

582

Dialogue

tific knowledge suggested by Quine's epistemological holism and "web of belief" picture provides an entirely unsatisfactory account of scientific knowledge when one actually considers the history of science (viz., mathematical physics). As previously alluded to, Friedman also argues that Quine's attack on the analytic-synthetic distinction is entirely irrelevant to the "phenomena of coordination" motivating the positivists' original conception of the relativized a priori. Against Quine's suggestion of locating philosophy or epistemology within science, Friedman's ultimate aim is to reaffirm a meta-scientific role for philosophy of science. In this section, I review and evaluate Friedman's arguments against Quine. I conclude that Friedman's analysis presents a strong challenge to Quinean naturalism while providing, conversely, a strong bid for his own perspective. 4.1. Quine's Naturalistic Holism

At the heart of Quine's (1969) naturalism or "naturalized epistemology" is a holistic view in which all knowledge is conceived of as being empirical within an interrelated "web-of-belief." This picture of knowledge is motivated by Quine's well-known rejection of the analytic-synthetic distinction in "Two Dogmas of Empiricism." Once it is "recognized" that there is no special status for certain types of knowledge traditionally thought to be analytic or a priori, then the idea of epistemology naturalized, viz., that philosophy itself should be set within empirical science, gains merit. That is, the recognition that all knowledge is synthetic or empirical and that no statement is immune from revision implies that traditional foundationalist projects that aim for certain knowledge are hopeless. 9 Accordingly, the ideal of philosophy occupying a privileged position to analyze science (and, more generally, knowledge) must be given up. Quine's own prescription is to relocate philosophy (or "epistemology") as a branch of empirical science. In an interview, Quine describes his position as follows: "Epistemology ... is science self-applied. It is the scientific study of scientific process. It explores the logical connections between the stimulation of the scientist's sensory receptors and the scientist's output of scientific theory" (Quine, cited in Pyle 1999, p. 20). Quine's argument is that if epistemology is to become a plausible project, then it must no longer be conceived as a normative project but as a descriptive one (viz., as a scientific project). In this somewhat debased but pragmatic role, philosophy will be capable of drawing upon the best available scientific resources in describing what knowledge is. Quine's holist picture of knowledge is motivated by instances of revolutionary science (a point that Friedman rightfully emphasizes motivates the positivists' own articulation of the relativized a priorz) and a Duhemian picture of theory testing. On Quine's view, the total system of science, a vast web of beliefs, faces the "tribunal of experience" as a corporate body. While Quine grants that some beliefs will be more entrenched (e.g., beliefs about mathematics and logic) while other beliefs lie at the periphery of the web

Reason in Science

583

(e.g., beliefs about biology), he maintains, on the basis of Duhemian underdetermination considerations, that when a particular scientific theory is inconsistent with empirical data (a "recalcitrant experience") this evidence cannot suggest which of the beliefs ought to be revised (see Quine 1970, pp. 5, 7).10 In describing "empiricism without the dogmas," Quine writes: The totality of our so-called knowledge or beliefs ... is a man-made fabric which impinges on experience only along the edges.... A conflict with experience at the periphery occasions readjustments in the interior of the field .... But the total field is so underdetermined by its boundary conditions, experience, that there is much latitude of choice as to what statements to reevaluate in the light of any single contrary experience.... If this view is right, it is misleading to speak of the empirical content of an individual statement. ... Furthermore, it becomes folly to seek a boundary between synthetic statements, which hold contingently on experience, and analytic statements, which hold come what may.... Conversely, by the same token, no statement is immune to revision. (1980, pp. 42-43, cited by Friedman on pp. 28-29)

Quine maintains that because empirical evidence "spreads" over all beliefs in the vast web that is the totality of science, all beliefs encompassed in this web, holistically conceived, equally face the "tribunal of experience." And it is precisely in this sense, for Quine, that all knowledge or beliefs have the status of being empirical or synthetic. 4.2. Friedman on the Poverty of Naturalism Friedman argues that Quine's picture of scientific knowledge is entirely inadequate in the face of the historical development of physics (pp. 35ff.). He claims, moreover, that Quine's holism only gains its force by completely ignoring the history of mathematical physics. Friedman writes: The increasingly abstract character of the mathematics applied in our most fundamental physical theories, the characteristically modern problem of coordination between abstract mathematical structures and concrete physical phenomena, and the Kantian idea of constitutively a priori principles functioning to mediate between the two are thus three distinguishable aspects of what is at bottom the same conceptual situation. And it is only by ignoring this situation entirely that we can, on the cOntrary, arrive at Quinean holism. Indeed, Quine himself arrives at epistemological holism by focussing exclusively on problems in the foundations of mathematics, with no real concern for the foundations of modern mathematical physics. (p. 81)

Here, Friedman highlights a weak point in Quinean holism insofar as it seems peculiar to conceive of knowledge in the domain of mathematical physics as being a large web of beliefs that equally face the "tribunal of

584

Dialogue

experience." In particular, Quine's picture of empirical testing, in which even the mathematical background of a physical theory equally faces the "tribunal of experience" seems to be at odds with the actual testing of theories in physics. In an earlier article against Quinean naturalism, Friedman voices his complaint as follows: The fundamental problem is that general relativity is not happily viewed as something like a large conjunction, such that one conjunct is given by Einstein's field equations, another ... by the Kleinian theory of transformation fields, and a third ... by the Riemmanian theory of manifolds-where we then view Eddinington's experimental results ... as spreading empirical confirmation over the entire conjunction. Rather, the mathematical background of Einstein's theory functions as a necessary presupposition of that theory, as a means of representation or a language, as it were, without which the theory could not even be formulated or envisioned as a possibility in the first place. (1997, p. 12)

Here, Friedman emphasizes the problem of coordination that motivates his neo-Kantian view. This aspect of scientific knowledge, he argues, is completely obscured by Quine's "beguiling form of epistemological holism" (p. 35). Defenders of holism might object to Friedman's historical argument by appealing to Quine's notion of entrenchment. It could be maintained that the presuppositions of physical theories (Le., the geometrical principles) are simply more-entrenched parts of knowledge. In the transitions from Newtonian mechanics to general relativity, these entrenched beliefs are revised in the face of recalcitrant experiences. 11 This rejoinder to Friedman, I think, completely misses his point. Friedman's challenge for the naturalist, quite simply, is that Quinean holism cannot do justice to the nature of scientific knowledge in the domain of mathematical physics; it is not a challenge for the naturalist to provide an ad hoc reconstruction of mathematical physics in Quinean terms. Friedman's point is that the notion of entrenchment is not an appropriate distinguishing feature of scientific knowledge at all. His own suggestion is that What characterizes the distinguished elements of our theories is ... their special constitutive/unction: the function of making the precise mathematical formulation and empirical application of the theories in question first possible. In this sense, the relativized and dynamical conception of the a priori developed by the logical empiricists appears to describe ... conceptual revolutions far better than does Quinean holism. (p. 40; emphasis in original)

To be clear, Friedman's argument against naturalism gains its force through an appeal to the history of physics. Although Friedman presents the development of physics through a neo-Kantian lens, the naturalist

Reason in Science

585

cannot simply re-describe the situation in Quinean terms, which would beg the very question at issue. The issue at hand is whether the problem of coordination, i.e., the problem of coordinating mathematical principles to physical phenomena, is a phenomenon that actually exists in the formulation of scientific knowledge. In the case of mathematical physics, Friedman presents a strong case for adopting his own perspective against QUine's naturalistic holism. 12 Against Quine's suggestion of locating philosophy within science, Friedman's neo-Kantian perspective proposes a distinct role for philosophy (as "meta-frameworks" or "meta-paradigms") in generating and suggesting Carnapian external questions (pp. 43-46). This neo-Carnapian picture of science can be illustrated with reference to a three-tiered framework for analyzing scientific knowledge advanced by Friedman (pp. 45, 105). On this picture, scientific knowledge operates on three levels that are constantly evolving and in "dynamic interaction" (p. 45): (1) The "base level" of empirical laws and tests. (2) The "conceptual level" of a priori principles that are constitutive of empirical laws and tests, i.e., mathematical principles and assumptions that frame and make possible objectively meaningful empirical laws and tests. (3) The "philosophical level" where philosophical concepts and prin-

ciples function as suggestions for choosing one particular scientific framework over another. Reconstructing this picture of scientific knowledge in Carnapian terms, we can ask relatively straightforward internal questions about (1) whose answers will be prescribed by the rules of (2). Questions about principles in (2), however, are external questions and they are asked on level (3). Against Quine's (1969) suggestion to locate philosophy within empirical science, as a branch of psychology, Friedman's claim is that philosophy can still occupy a distinct meta-scientific role, offering suggestions drawing on philosophical resources for choosing among competing scientific or constitutive frameworks (p. 46). Moreover, he argues that this philosophical level plays an especially fruitful role during periods of scientific crisis. 13 5. Against Kuhnian Relativism In the most ambitious section of his book, Friedman addresses the problem of the "arationality of science" implied by Thomas Kuhn's picture of theory change where sociological factors playa key role in theory choice (pp. 47-68, 93-103).14 Friedman opposes the relativism of Kuhn and, in particular, Kuhn's suggestion that sociological factors, such as persuasion or

586 Dialogue

conversion, explain theory choice better than rational reasons. Like other commentators on Kuhn, Friedman's strategy against this picture is to articulate the sense and extent to which theory choice is objective and rational. In the following section, I explicate Kuhn's (1977) argument that theory choice is inevitably arational with reference to the two main theses that motivate it, viz.: (1) the "incommensurability of values" thesis and (2) the "incommensurability of meaning" thesis. Subsequently, I examine and evaluate Friedman's counter-argument for "scientific rationality." I conclude that Friedman's arguments fail to give good reasons to reject Kuhnian relativism. 5.1. Kuhn on Values and Paradigm Choice: An A rational Picture of Theory Change

An overlooked and important aspect of Kuhn's picture of science is his philosophical thesis on the incommensurability of values (Kuhn 1977; 1983; 1996, pp. 94-110, 198-207). Through an examination of the problem of values in science, Kuhn (1977) concludes that there is no "rational" basis for theory choice. Kuhn's argument begins with a consideration of the epistemic values that may be employed to judge between competing paradigms. Kuhn provides a "non-exhaustive" list of "defensible" values or standards as follows: (1) empirical accuracy; (2) consistency, i.e., internal and external consistency; (3) explanatory scope; (4) simplicity; and (5) fruitfulness, i.e., puzzle-solving efficacy. Kuhn notes that two sorts of problems arise when applying such values to judge between competing paradigms. First, there is an ambiguity regarding these values insofar as the terms can be interpreted and applied differently by scientists. Thus, the application of such values in paradigm choice will inevitably involve the application of both objective (shared) and subjective (personal) values. Second, when considered together, these values may conflict, e.g., accuracy conflicts with scope or accuracy conflicts with consistency. Again this suggests that paradigm choice will necessarily involve the application of SUbjective values insofar as certain values will be weighted more heavily than others. Kuhn appeals to these two observations to establish the claim that there exists no objective (i.e., intersubjective) basis for theory choice, i.e., there is no objective "meta-perspective" of values that can be appealed to in order to choose among competing paradigms. Presented more formally, Kuhn's argument that there is no rational basis for theory choice can be reconstructed as follows: 1. Comparisons between competing paradigms must appeal either to an objective set of values or to the shared values of a particular paradigm. 2. It is impossible to appeal to an objective set of values because such a meta-perspective does not exist (see previous paragraph).

Reason in Science 587

3. Thus, theory choice must appeal to the shared values of one particular paradigm. 4. Thus, to argue for the merits of a paradigm is to appeal to values that a paradigm is premised upon. But this amounts to circular reasoning or "persuasion." 5. Thus, there is no rational way to judge between competing paradigms. Kuhn's claim that there is no (objective or paradigm-independent) perspective of epistemic values that can be appealed to in order to judge competing paradigms establishes his subsequent claim that there is no rational or objective basis for paradigm choice. IS This argument motivates Kuhn's further claim that paradigm choice inevitably involves sociological factors such as persuasion or conversion. Given the problem of incommensurable values in science, Kuhn maintains that the social consensus of the current scientific community (and especially their favoured epistemic values) plays a decisive role in determining a paradigm shift within a scientific tradition. Kuhn's argument, I think, establishes that "paradigm choice can never be unequivocally settled by logic and experiment alone" (1996, p. 94). Paradigm choice is rational, according to Kuhn, to the extent that the relevant community of scientists are capable of arguing on the basis of shared values; however, these shared values themselves are ultimately a matter of subjective preference (which reflect and are relative to a particular social-historical context). Corresponding to this picture of paradigm choice is a discontinuous picture of science implied by Kuhn's philosophical-historical thesis of the incommensurability of meaning. 16 Kuhn claims that a revolution in science necessarily involves the abolishment of the previous paradigm. I? As such, scientific revolutions represent discontinuities in science and Friedman maintains that science is not a cumulative process. Many commentators have argued against this characterization of science. In the context of physics, e.g., some commentators have pointed out that Newtonian physics is derivable from relativistic dynamics as a special case. Kuhn's counter to this claim is that the older Newtonian paradigm cannot be derived at all because the concepts in relativistic dynamics do not mean the same thing as those in the Newtonian paradigm. Here, Kuhn appeals to his controversial incommensurability of meaning thesis in claiming that there is no common framework for both paradigms. Kuhn writes: Newtonian mass is conserved; Einsteinian is convertible with energy. Only at low relative velocities may the two be measured in the same way, and even then they must not be conceived to be the same.... Unless we change the definitions

588

Dialogue

of the variables ... the statements we have derived are not Newtonian ... at least not in any sense of "derive" now generally recognized. (1996, p. 102)

Kuhn's resulting picture of science is a discontinuous one in which changes in paradigms are accompanied by the abolishment of previous paradigms. It is characteristic of such scientific revolutions that there is a corresponding shift in the epistemic values presupposed by the paradigm (which playa large role in the triumph of a paradigm). 5.2 Friedman's Argumentfor Scientific Rationality Friedman takes exception to Kuhn's picture of theory change, which rules out any notion of "truly universal human rationality" for the sciences. Friedman remarks that "the only notion of scientific rationality we have left is a relativized, or sociological one according to which all there ultimately is to scientific rationality ... is the otherwise arbitrary commitment of some particular social community ... to one particular paradigm ... rather than another" (p. 48). Moreover, Friedman argues persuasively that Kuhn's subsequent attempts to defend scientific rationality cannot be judged as successful (pp. 50-53).18 In light of this consequence, Friedman endeavours to provide his own argument for scientific rationality in the face of Kuhnian relativism. In what follows, I explicate and evaluate Friedman's argument against Kuhn. I argue that Friedman's argument for scientific rationality cannot be judged as successful because it fails to address Kuhn's first argument against scientific rationality premised on the incommensurability of values thesis. In a somewhat unexpected manoeuver, Friedman appeals to the Frankfurt school in arguing for scientific rationality, viz., the Habermasian distinction between instrumental and communicative rationality (pp. 54ff.). Instrumental rationality, which is subjective or personal, refers to a capacity to engage in effective means-ends reasoning (where an agreedupon goal is already given). By contrast, communicative rationality, which is intersubjective, refers to a capacity to engage in deliberative argumentative reasoning (where no agreed-upon goal is given) aimed at bringing about consensus. Against Kuhn's picture of science that suggests that communicative rationality is limited to normal science, i.e., rationality within a paradigm, Friedman argues that "there is also an important sense in which [the scientific enterprise] aims for, and successfully achieves, agreement or consensus across different paradigms" (p. 58; emphasis in original). To support this idea of interparadigm rationality, Friedman again appeals to the history of physics. The history of mathematical physics, argues Friedman, shows us that (I) earlier constitutive frameworks are exhibited as limiting cases, holding approximately under certain precisely defined conditions, and (2) the concepts and principles of succeeding constitutive frameworks evolve continuously, by a series of

Reason in Science

589

natural transformations, from previous constitutive frameworks (pp. 5866,95-101). Taken together, Friedman maintains that "we can thus view the evolution of succeeding paradigms ... as a convergent series ... in which we successively refine our constitutive principles in the direction of even greater generality and adequacy" (p. 63).19 For instance, in the transition from Newtonian mechanics to special relativity, the constitutive principles of mechanics (viz., Euclidean geometry and the law of inertia) are retained, but in moving to the more general four-dimensional spatialtemporal structure, special relativity yields the Newtonian Euclidean spatial geometry as a limiting case. Granting Friedman this much, it does not follow that these paradigms, Le., Newtonian and special relativity, both equally share a constitutive framework in the sense needed to secure the rationality of theory change denied by Kuhn. Friedman admits this much when he writes, [T]his kind of relationship of successive approximation between paradigms, and the resulting notion of inter-paradigm agreement, is highly anachronistic or "Whiggish," in that it is constructed wholly from the point of view of the successor paradigm and, in truly revolutionary cases, uses concepts and principles that simply do not exist from the point of view of the preceding paradigm. (p. 59; emphasis in the original)

As Friedman observes, what is really needed to combat Kuhn's scepticism on theory change is an explanation of how a new paradigm can be chosen on a rational basis, i.e., a rational basis for prospective (interparadigm) continuity must be illustrated (pp. 99-101). Friedman argues for a rational basis for prospective continuities between present and future paradigms with the notion of a philosophical "meta-framework" proffered in his book (pp. 66-68,93-103). According to Friedman, prospective interparadigm continuity is successfully achieved in the sciences insofar as new philosophical meta-frameworks (which arise in dynamical interaction with older meta-frameworks, older constitutive frameworks, and developments in empirical science) function to motivate and sustain continuous transformations and modifications of previous constitutive frameworks (pp. 65-66, 99-103). Although Friedman is not entirely clear on what he means to argue in this brief discussion, the basic point (contra Kuhn) is that paradigm changes within a particular scientific tradition are continuous and hence attain interparadigm (communicative) rationality. To support this point, Friedman argues that succeeding paradigms in a scientific revolution are more similar to different stages of development within a linguistic tradition than disparate languages in separate cultures (p. 100). Relating this analogy to Friedman's earlier claims about interparadigm retrospective communicative rationality, which I have granted for the sake of argument, Friedman's further

590 Dialogue

claim is that new paradigms come about as deliberate modifications of earlier paradigms against the backdrop of a common framework of problems, concepts, and goals. In this way, a constitutive framework, which gives rise to standards of communicative rationality, can be expanded or transformed into a new constitutive framework that was not previously possible. 2o In this sense, we can see how a new paradigm can become a defensible ("rational or reasonable") option from the perspective of a current paradigm. Granting this point to Friedman regarding interparadigm prospective communicative rationality, we have the picture that (1) a constitutive framework contains the previous constitutive framework as a limiting case; (2) the new constitutive principles evolve continuously from the older constitutive principles by a series of natural transformations; and (3) this process of continual transformations will be motivated and sustained by an appropriate new philosophical meta-framework (pp. 66, 99-101). This sense of communicative rationality, viz., a continuity or convergence between paradigm shifts, Friedman thinks, captures a sense of scientific rationality that can be defended in the face of Kuhnian relativism. But does this sense of "rationality" defeat the relativism associated with Kuhn's picture of theory change or theory choice? I think not. The fundamental difficulty with Friedman's picture of scientific rationality is that it fails to address the "historical relativism" implied by Kuhn's argument on the incommensurability of values. 21 Even if we grant the sense of rationality that Friedman presents, this sense of communicative rationality only has persuasive force against Kuhn's argument that different paradigms are incommensurable with respect to meaning, not the argument that different paradigms are incommensurable with respect to values. Friedman's failure to squarely address this latter issue, I think, highlights the correctness (and perhaps triviality) of Kuhn's argument against scientific rationality. What Friedman does with Kuhn's analysis on incommensurable values is grant that there is no paradigm-independent perspective of values or standards that can be appealed to in jUdging competing paradigms, arguing that these values can be continually negotiated on the level of meta-frameworks. Friedman wants to call this negotiation process "rational" insofar as the transition between paradigms is both modified from older paradigms and motivated by an "appropriate" new philosophical meta-framework; however, what is missing from this analysis is an argument that establishes the connection between scientific continuity and rationality. What is not shown by Friedman's analysis, in particular, is why the consensus regarding values endorsed by later metaframeworks should be viewed as progressions towards rationality rather than historical contingencies. Friedman's failure to clearly address this point on values, renders his argument for a "universal scientific rationality for the sciences" optimistic at best (see n. 19). While the notion of communicative rationality, I think, provides the right sort of initial response

Reason in Science

591

to Kuhn's problematic, Friedman's analysis, as it stands, requires further articulation in order to persuasively overcome Kuhn's "relativistic or sociological" conception of scientific rationality. 6. Conclusion

Despite some weaker points in the more ambitious portions of the book, Dynamics of Reason is an admirable and provocative defence of neoKantianism in the context of contemporary philosophy of science. In my view, the most cogent and interesting argument presented in Friedman's book is its diagnosis of why philosophical naturalism, despite its intuitive appeal, cannot provide an adequate perspective on the sciences. If naturalism cannot provide a satisfactory picture of the nature of scientific knowledge in physics, traditionally conceived as the paradigmatic domain of scientific knowledge in philosophy of science, then naturalism cannot serve as a satisfactory perspective for the sciences. As indicated in this article, however, Friedman would do well (with regards to his argument against naturalism and for his own neo-Kantian view) by expanding his perspective to other domains of science (e.g., biology), where Quine's perspective, arguably, provides a better account of theory change. Despite Friedman's best intentions, his argument for scientific rationality, in the face of Kuhnian relativism, cannot be judged as successful. Friedman's failure to address Kuhn's analysis on the incommensurability of scientific values, I think, renders his subsequent response to be entirely unsatisfactory. Friedman's failure in this regard indicates the immense challenge facing philosophers of science on the problem of epistemological relativism. Although many readers will remain unmoved by Friedman's suggestion for a neo-Kantian philosophy of science, in my opinion his perspective merits attention, especially among philosophers of mathematical physics. In presenting and articulating his position, Friedman displays his gifts and impressive breadth of knowledge as a historian, especially in the domains of history of twentieth-century analytic philosophy and the exact sciences. In the face of "anti-positivist" movements in contemporary philosophy of science, Friedman's perspective is especially refreshing in its attempt to retain some lost insights of the positivists, while avoiding simplistic characterizations. In his creative assimilation the positivists' worldview with Kuhn's-two views typically thought to be at oddsFriedman makes a strong case for re-examining the problem of coordination in the context of physical theory. Ultimately, the nature of Friedman's proposal for a neo-Kantian perspective can only be taken as a Carnapian external assertion, i.e., as a practical suggestion for philosophers of science. As I have indicated, I think that Friedman presents a persuasive argument for the potential fruits of such a perspective and his proposal deserves further consideration. 22

592

Dialogue

Notes

2

3

4

5

6

In his revisionist writings on logical positivism, Friedman (1997, 1999) argues that the true innovation advanced by positivists such as Carnap, Reichenbach, and Schlick was not a radical form of empiricism (the "received view"), but a relativized conception of the a priori. For English translations and commentary of the Schlick-Reichenbach correspondence, see Coffa (1991, pp. 201-204) and Howard (1994, pp. 56-63). Schlick initiates the correspondence on November 26, 1920, in response to Reichenbach's note in his book: "It is characteristic of Schlick's psychologizing method that he believes to have refuted by many proofs the correct part of Kant's theory, namely, the constitutive significance of the coordinating principles" (1965, p. 116, n.27). Here, Reichenbach refers to Schlick (1985). In the context of contemporary philosophy of science, Carnap's distinction between internal and external questions arises most often in the realism-antirealism debates. Arthur Fine (1984), for example, in forwarding his "Natural Ontological Attitude" expresses his essential agreement with Carnap when he states that realists are fooling themselves in thinking that they can justifiably adopt a stance outside science (i.e., outside the language of scientific theories) and affirm a metaphysical reality that science captures. Howard Stein views his attempted reconciliation of realism and instrumentalism as a rehearsal of Carnap's distinction of internal and external questions, and in a passing remark he attributes similar considerations to be motivating Hilary Putnam's internal realism (1989, p. 51). Putnam's (1981) rejection of "Metaphysical Realism" (and its associated "God's eye view") and endorsement of "internal realism" follows Camap's suggestion that the notion of Truth can only be understood relative to a particular linguistic framework or "conceptual scheme." Camap's conventionalism is nicely captured in the closing lines of his article: "Let us grant to those who work in any special field of investigations the freedom to use any form of expression which seems to be useful to them .... Let us be cautious in making assertions and critical in examining them, but tolerant in permitting linguistic forms" (1983, p. 257; emphasis in original). In a 1991 interview, Kuhn describes himself as a "post-Darwinian Kantian" arguing that the mind can only order experiences by presupposing some sort of a priori paradigm (see Kuhn, cited in Horgan 1996, p. 44). Whereas Kant views such a priori paradigms (or categories) as fixed, Kuhn, like Reichenbach, views such paradigms as constantly evolving. Friedman articulates this conception of physical knowledge more precisely with a picture of physical theories that involves three asymmetrically functioning parts (pp. 79-80). First, there is a mathematical part that describes some spatio-temporal framework, e.g., Euclidean space, four-dimensional Minowski space-time, or semi-Riemmannian space-time manifolds. Second, there is an empirical part that applies such mathematical structures to formulate empirical laws to describe some concrete empirical phenomena, e.g., the law of universal gravitation, Maxwell's equations, or Einstein's equations for the

Reason in Science 593

7

8

9

10

11 12

13

gravitational field. Third, there is a mechanical part that functions to coordinate or set up a correspondence between the mathematical and empirical part, e.g., the Newtonian laws of motion, the light principle, or the principle of equivalence. While the second part is empirical, Friedman claims that the first and third parts are constitutive a priori, i.e., the mathematical and mechanical parts function to make possible empirical laws. Friedman does defend Carnap's way of drawing the analytic-synthetic distinction insofar as he argues, successfully I think, that Quine's attack on the analytic-synthetic distinction premised on Duhemian underdetermination considerations does not undermine, in the least, the relativized notion of a priori defended by Carnap (pp. 37-41). Also see n. 10 below. It is fair to grant to Friedman that he has provided a historical argument in his analysis of the development of theories of physics for drawing a distinction between constitutive a priori and empirical principles (pp. 35-41, 75-82). This argument, however, cannot be taken as an independent argument for the existence of constitutive principles since it is already framed within a neo-Kantian perspective (and it is circular in this sense). As such, Friedman's claim for the existence of constitutive a priori principles hinges on whether the reader accepts his neo-Kantian interpretation of physics. As a broad criticism on this issue, it could be objected that Friedman has provided an interesting interpretation of the history of physics; however, it is only one of many possible interpretations. It is worth mentioning in this connection that Quine's (1969, 1980) attack on Carnap, which is presented against Carnap's (1967) epistemological project in the Aufbau (originally published in 1928), only gains its force by completely distorting the nature of Carnap's project as afoundationalist project addressing the empiricist problem of gaining certain knowledge the external world (see Friedman 1999, pp. 89-94, 116-24, and Richardson 1998, chap. I). Friedman rightly notes that Quine is extremely misleading when he equates analyticity with unrevisability (see p. 33, n.38), since Carnap's own philosophy of linguistic frameworks is premised on Duhemian holism and the idea that both analytic and synthetic principles can be revised in the development of empirical science (see Carnap 1937, §82). Carnap squarely addresses this issue in his reply to Quine (see Schilpp 1963, pp. 921-22). For an extended discussion on the Carnap-Quine debates on analyticity, see Creath (1990). Friedman provides a brief response to this possible objection on pp. 39-41. As previously mentioned, however, a severe limitation of Friedman's analysis lies in its exclusive attention to the domain of mathematical physics at the exclusion of all other sciences. A defender of naturalism could object to Friedman's general conclusion (viz., that his neo-Kantian perspective accounts for science better than naturalism) on the ground that holism explains other sciences (e.g., biology) better than Friedman's neo-Kantian account. As two concrete examples of philosophical discourse informing science, Friedman appeals to (1) Newton's philosophical encounters with Descartes and

594

14

15

16

17

18

19

20

Dialogue

Leibniz (on topics such as the nature of space, time, matter, force, interaction, and divinity) informing the formulation of his universal theory of gravitation, and (2) Einstein's philosophical encounters with Helmholtz and Poincare (on the foundations of geometry) informing his "geometrical" notion of gravitation in general relativity (pp. 44-45, 107-17). In response to accusations that he had portrayed science as irrational, Kuhn subsequently clarified his position as the claim that science isarational (Kuhn, cited in Horgan 1996, p. 42). It is important to note that Kuhn's conclusion that there is no rational basis for paradigm choice implicitly equates "rationality" with "objectivity" (i.e., intersubjectivity). This second thesis, which is (rightfully) attacked by philosophers of science, is the thesis Friedman also targets. As I will argue subsequently, however, Kuhn's first philosophical argument is sufficient to establish his claim that there is no purely rational standpoint in paradigm choice. I will also argue that although Friedman is successful in refuting Kuhn's second argument, he fails to sufficiently address Kuhn's first argument on the problem of values in science. Kuhn defines scientific revolutions (somewhat circularly given his purposes) as "those non-cumulative developmental episodes in which an older paradigm is . replaced in whole or in part by an incompatible new one" (1996, p. 92; emphasis added). Friedman argues that Kuhn's historical analysis of scientific revolutions (viz., from Newtonian to Einsteinian theory) is incorrect insofar as it neglects an important sense in which the old and new paradigms are compatible. Kuhn (1983, 1993) argues that science is progressive and rational insofar as subsequent paradigms can solve more puzzles than earlier paradigms (c( Laudan 1977). This view, however, presupposes that there is a stable set of epistemic values that are definitive of science, an assumption that seems to be undermined by Kuhn's (1977) claim that both the application of, and relative weights attached to, such values do not remain stable over time. Along these lines, Friedman objects that Kuhn only secures a subjective sense of rationality, rather than a truly objective sense (pp. 50-53, 93). Friedman suggests that this picture of convergence can be fruitfully assimilated with Kant's regulative use of Reason (pp. 64-65). Although science can only be conceived as an ongoing process of revision, Friedman suggests that we necessarily need to conceive of our present constitutive principles, which represent one stage in a convergent series, as an approximation to a "final. ideal community of inquiry . .. that has achieved a universal. trans-historical communicative rationality on the basis of thefully general and adequate constitutive principles reached in the ideal limit of scientific progress" (p. 64; emphasis added). For a commentary on this passage, see Richardson 2002, pp. 261-69. Friedman emphasizes the importance of the history of scientific philosophy as a factor to help explain scientific rationality and understand the ongoing dialectic of scientific knowledge (p. 44; also see Disalle 2002). For instance, in the case

Reason in Science

595

of the transition from special to general relativity, Friedman argues that the rational shift in constitutive frameworks can be illustrated with reference to how Einstein engaged in the ideas of both Helmholtz and Poincare (pp. 101-103). 21 Popper, I think, accurately dubbed Kuhn's view as a species of "historical relativism" (1970, p. 55). Kuhn likens scientific development to biological development as a "unidirectional and irreversible process" (1996, p. 206). This "evolutionary" view of science implies that all scientific knowledge is relative to the puzzle-solving efficacy of a historically situated paradigm. Scientific knowledge and scientific values, on this view (contra Friedman), are not approaching some ideal, but are simply solving puzzles in particular socialhistorical environments. Kuhn concedes that even if this picture of scientific knowledge is relativistic, he "cannot see that the relativist loses anything needed to account for the nature and development of the sciences" (1996, p.207). 22 I am grateful to Philip Hanson and Steven Davis for earlier discussions on such issues at Simon Fraser University. This article benefited from discussion with my colleagues in the Committee on Conceptual and Historical Studies of Science and the Fishbein Center for the History of Science and Medicine (who participated in a discussion seminar of Friedman's book), as well as with Michael Friedman (who clarified and defended the ideas of his book in a guest lecture at the University of Chicago in March 2002). I would also like to thank two anonymous referees from this journal for useful suggestions. Finally, I would like to acknowledge the Social Sciences and Humanities Research Council of Canada for research support.

References Carnap, Rudolf 1937 The Logical Syntax of Language. Translated by A. Smeaton. London: Kegan Paul. Originally published as Logische Syntax der Sprache. Wien: Springer, 1934. 1967 The Logical Structure of the World. Translated by R. A. George. Berkeley, CA: University of California Press. Originally published as Der logische Aufbau der Welt. Berlin: Weltkreis, 1928. 1983 . "Empiricism, Semantics, and Ontology." Reprinted (with revisions) in Philosophy of Mathematics: Selected Readings. 2nd ed. Edited by P. Benacerraf and H. Putnam. Cambridge: Cambridge University Press, pp. 241-57. Originally published in Revue Internationale de Philosophie, II (1950): 20-40. Coffa, 1. Alberto 1991 The Semantic Traditionfrom Kant to Carnap: To the Vienna Station Edited by L. Wessels. Cambridge: Cambridge University Press. Creath, Richard, ed. 1990 Dear Carnap. Dear Van: The Quine-Carnap Correspondence and Related Work. Berkeley, CA: University of California Press.

596

Dialogue

Disalle, Robert 2002 "Reconsidering Kant, Friedman, Logical Positivism, and the Exact Sciences." Philosophy of Science, 69: 191-211. Fine, Arthur 1984 "The Natural Ontological Attitude." In Scientific Realism. Edited by 1. Leplin. Berkeley, CA: University of California Press, pp. 83107. Friedman, Michael 1983 Foundations of Space-Time Theories: Relativistic Physics and the Philosophy of Science. Princeton, NJ: Princeton University Press. 1992 Kant and the Exact Sciences. Cambridge, MA: Harvard University Press. 1997 "Philosophical Naturalism." Proceedings and Addresses of the American Philosophical Association, 71: 7-21. 1999 Reconsidering Logical Positivism. Cambridge: Cambridge University Press. 2001 Dynamics of Reason: The 1999 Kant Lectures at Stanford University. Stanford, CA: CSLI Publishers. Horgan, John 1996 The End of Science: Facing the Limits of Knowledge in the Twilight of the Scientific Age. New York: Broadway. Howard, Don 1994 "Einstein, Kant, and the Origins of Logical Empiricism." In Logic, Language, and the Structure of Scientific Theories: Proceedings of the Carnap-Reichenbach Centennial, University of Konstanz, 21-24 May 1991. Edited by W. Salmon and G. Wolters. Pittsburgh, PA: University of Pittsburgh Press, pp. 45-105. Kuhn, Thomas S. 1977 "Objectivity, Value Judgment, and Theory Choice." In The Essential Tension: Selected Studies in Scientific Tradition and Change Chicago, IL: University of Chicago Press, pp. 320-39. 1983 "Rationality and Theory Choice." Journal of Philosophy, 80: 56370. 1993 "Afterwords." In World Changes: Thomas Kuhn and the Nature of Science. Edited by P. Horwich. Cambridge, MA: MIT Press, pp. 311-41. 1996 The Structure of Scientific Revolutions. 3rd ed. Chicago, IL: University of Chicago Press. Originally published in 1962. Laudan, Larry 1977 Progress and Its Problems: Towards a Theory of Scientific Growth Berkeley, CA: University of California Press.

Reason in Science

597

Popper, Karl R. 1970 "Normal Science and Its Dangers." In Criticism and the Growth of Knowledge. Edited by I. Lakatos and A. Musgrave. Cambridge: Cambridge University Press, pp. 51-58. Putnam, Hilary 1981 Reason, Truth, and History. Cambridge: Cambridge University Press. Pyle, Andrew, ed. 1999 Key Philosophers in Conversation: The Cogito Interviews. London: Routledge. Quine, Willard V. 1969 "Epistemology Naturalized." In Ontological Relativity and Other Essays. New York: Columbia University Press, pp. 69-90. 1970 Philosophy of Logic. Englewood Cliffs, NJ: Prentice-Hall. 1980 "Two Dogmas of Empiricism." Reprinted (with revisions) in From a Logical Point of View: Nine Logico-Philosophical Essays. 2nd rev. ed. Cambridge, MA: University of Harvard Press, pp. 20-46. Originally published in Philosophical Review, 60 (1951): 20-43. Reichenbach, Hans 1958 The Philosophy of Space and Time. Translated by M. Reichenbach and 1. Freund. New York: Dover. Originally published as Philosophie der Raum-Zeit-Lehre. Berlin: de Gruyter, 1928. 1965 The Theory of Relativity and A Priori Knowledge. Translated by M. Reichenbach. Berkeley, CA: University of California Press. Originally published as Relativitiitstheorie und Erkenntnis Apriori. Berlin: Springer, 1920. 1969 Axiomatization of the Theory of Relativity. Translated by M. Reichenbach. Berkeley, CA: University of California Press. Originally published as Axiomatik der relativistischen Raum-ZeitLehre. Braunschweig: Vieweg, 1924. Reisch, George A. 1991 "Did Kuhn Kill Logical Empiricism?" Philosophy of Science, 58: 264-77. Richardson, Alan W. 1998 Carnap's Construction of the World: The Aufbau and the Emergence of Logical Empiricism. Cambridge: Cambridge University Press. 2002 "Narrating the History of Reason Itself: Friedman, Kuhn, and a Constitutive A Priori for the Twenty-First Century." Perspectives on Science, 10: 253-74. Schilpp, Paul A., ed. 1963 The Philosophy of Rudolf Carnap. The Library of Living Philosophers, Vol. 11. La Salle, IL: Open Court.

598

Dialogue

Schlick, Moritz 1979 "Critical or Empiricist Interpretation of Modern Physics?" Translated by P. Heath. Reprinted in Moritz Schlick: Philosophical Papers, Vol. I. Edited by H. L. Mulder and B. F. B. van de VeldeSchlick. Dordrecht: Reidel, pp. 322-34. Originally published as "Kritizistische oder empiristische Deutung der neuen Physik?" Kant-Studien, 26 (1921): 96-111. 1985 General Theory of Knowledge. Translated by A. Blumberg. La Salle, IL: Open Court. Originally published as Allgemeine Erkellntnislehre. Berlin: Naturwissenschaftliche, 1918. Stein, Howard 1989 "Yes, but ... : Some Skeptical Remarks on Realism and AntiRealism." Dialectica, 43: 47-65.